**Review of the New Combustion Technologies in Modern Gas Turbines**

M. Khosravy el\_Hossaini

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54403

#### **1. Introduction**

The combustion chamber is the most critical part of a gas turbine. The chamber had to be designed so that the combustion process to sustain itself in a continuous manner and the temperature of the products is sufficiently below the maximum working temperature in the turbine. In the conventional industrial gas turbine combustion systems, the combustion chamber can be divided into two areas: the primary zone and the secondary zone. The pri‐ mary zone is where the majority of the fuel combustion takes place. The fuel must be mixed with the correct amount of air so that a stoichiometric mixture is present. In the secondary zone, unburned air is mixed with the combustion products to cool the mixture before it en‐ ters the turbine. In some design, there is an intermediate zone where help secondary zone to eliminate the dissociation products and burn-out soot.

The majority of the combustors are developed base on diffusion flames as they are very stable and fuel flexibility option. In a diffusion flame, there will be always stoichiometric regions re‐ gardless of overall stoichiometry. The main disadvantage of diffusion-type combustor is the emission as high temperature of the primary zone produced larger than 70 ppm NOx in burn‐ ing natural gas and more than 100 ppm for liquid fuel [1]. Several techniques have been tried in order to reduce the amount of NOx produced in conventional combustors. In general, it is difficult to reduce NOx emissions while maintaining a high combustion efficiency as there is a tradeoff between NOx production and CO/UHC production.

In some recent installations, the premixed type of combustion has been selected to reduce NOx emissions bellow 10 ppm. Apart from the flame type change, there are some method such as "wet diffusion combustion", FGR1 and SCR2 . In an example of wet combustion, a nuz‐ zle through which steam is injected is provided in the vicinity of the fuel injector. The level of NOx emission is controlled by the amount of steam. However, there is a limit on the increas‐

© 2013 el\_Hossaini; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 el\_Hossaini; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

ing the steam flow rate as cause corresponding considerable CO emission. Furthermore, pre‐ paring pure steam in the required injection condition increases operational costs. Nowadays, wet combustion rarely applies due to water consumption and the penalty of reduced efficien‐ cy. Post Combustion treatments such as SCR are those which convert NOx compounds to ni‐ trogen or absorb them from flue gas. These methods are relatively inexpensive to install but does not achieve NOx removal levels better than modern gas turbine combustor.

**2.2. Flame stabilization**

for a pressure drop3

30,000 and stoichiometric mixture [2].

the swirl number which is defined as:

3 This pressure drop is named the cold loss.

After the fuel has been injected into the air flow, the flow will enter the flame region. It does this with quite a high velocity, so to make sure the flame isn't blown away; suitable flame stabilization techniques must be applied. First, the high velocity of flow will be responsible

air has a high velocity. So combustion chambers benefit from diffusers to slow down the air flow. There are two normal kinds of flame stabilizers: bluff-body flame holders and swirlers. The shape of the bluff–body flame holder affects the flow stability characteristics through the influence on the size and shape of the wake region. Since the flame stabilization depends on size of the zone of recirculation behind the bluff–body, different geometries such as trian‐ gular, rectangular, circular and more complex shapes are being use. One of the basic prob‐ lem of bluff-body flame holders is a considerable effect on pressure loss. Figure 2 shows a

**Figure 2.** High speed images of the circular cylinder (top), square cylinder (middle) and V gutter (bottom) at Re =

Flow reversal can be applied in the primary zone. The best way to reverse the flow is to swirl it through using swirlers. The two most important types of swirlers are axial and radi‐ al. The advantage of flow reversal is that the flow speed varies a lot. So there will be a point at which the airflow velocity matches the flame speed where a flame could be stabilized. The degree of swirl in the flow is quantified by the dimensionless parameter, *Sn* known as

high speed image of three flame holders in atmospheric condition.

. Secondly, the flame in the combustion chamber cannot survive if the

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

147

In this chapter, a short introduction of combustion process and then a description of some new pioneer combustor have been presented. As gas turbine manufacturers are looking for continuous operation or stable combustion, satisfactory emission level, minimum pressure loss and durability or life. Hence, the advanced combustor might include all of these criteria, so some of them are selected to discuss in details.

#### **2. The combustion process**

#### **2.1. Type of combustion chamber**

The diffusion and premixed flame are two main type of combustion, which are using in gas turbines. Apart from type of flame, there are two kind of combustor design, annular and tubular. The annular type mostly recommended in the propulsion of aircraft when small cross section and low weight are important parameters. Can or tubular combustors are cheaper and several of them can be adjusted for an industrial engine identically. Although there are different types of combustors, but generally, all combustion chambers have a dif‐ fuser, a casing, a liner, a fuel injector and a cooling arrangement. An entire common layout is visualized in figure 1.

**Figure 1.** The layout of the combustion chamber.

<sup>1</sup> Flue Gas Recirculation

<sup>2</sup> Selective Catalytic Reduction

#### **2.2. Flame stabilization**

ing the steam flow rate as cause corresponding considerable CO emission. Furthermore, pre‐ paring pure steam in the required injection condition increases operational costs. Nowadays, wet combustion rarely applies due to water consumption and the penalty of reduced efficien‐ cy. Post Combustion treatments such as SCR are those which convert NOx compounds to ni‐ trogen or absorb them from flue gas. These methods are relatively inexpensive to install but

In this chapter, a short introduction of combustion process and then a description of some new pioneer combustor have been presented. As gas turbine manufacturers are looking for continuous operation or stable combustion, satisfactory emission level, minimum pressure loss and durability or life. Hence, the advanced combustor might include all of these criteria,

The diffusion and premixed flame are two main type of combustion, which are using in gas turbines. Apart from type of flame, there are two kind of combustor design, annular and tubular. The annular type mostly recommended in the propulsion of aircraft when small cross section and low weight are important parameters. Can or tubular combustors are cheaper and several of them can be adjusted for an industrial engine identically. Although there are different types of combustors, but generally, all combustion chambers have a dif‐ fuser, a casing, a liner, a fuel injector and a cooling arrangement. An entire common layout

does not achieve NOx removal levels better than modern gas turbine combustor.

so some of them are selected to discuss in details.

**2. The combustion process**

146 Progress in Gas Turbine Performance

**2.1. Type of combustion chamber**

is visualized in figure 1.

**Figure 1.** The layout of the combustion chamber.

1 Flue Gas Recirculation 2 Selective Catalytic Reduction After the fuel has been injected into the air flow, the flow will enter the flame region. It does this with quite a high velocity, so to make sure the flame isn't blown away; suitable flame stabilization techniques must be applied. First, the high velocity of flow will be responsible for a pressure drop3 . Secondly, the flame in the combustion chamber cannot survive if the air has a high velocity. So combustion chambers benefit from diffusers to slow down the air flow. There are two normal kinds of flame stabilizers: bluff-body flame holders and swirlers.

The shape of the bluff–body flame holder affects the flow stability characteristics through the influence on the size and shape of the wake region. Since the flame stabilization depends on size of the zone of recirculation behind the bluff–body, different geometries such as trian‐ gular, rectangular, circular and more complex shapes are being use. One of the basic prob‐ lem of bluff-body flame holders is a considerable effect on pressure loss. Figure 2 shows a high speed image of three flame holders in atmospheric condition.

**Figure 2.** High speed images of the circular cylinder (top), square cylinder (middle) and V gutter (bottom) at Re = 30,000 and stoichiometric mixture [2].

Flow reversal can be applied in the primary zone. The best way to reverse the flow is to swirl it through using swirlers. The two most important types of swirlers are axial and radi‐ al. The advantage of flow reversal is that the flow speed varies a lot. So there will be a point at which the airflow velocity matches the flame speed where a flame could be stabilized. The degree of swirl in the flow is quantified by the dimensionless parameter, *Sn* known as the swirl number which is defined as:

<sup>3</sup> This pressure drop is named the cold loss.

$$S\mathfrak{m} = \frac{G\_{\theta}}{G\_{x}r} \tag{1}$$

then premixed flame will propagate into the unburned reactants. If fuel and air mix at the same time and same place as they react, the diffusion or non-premixed combustion will ap‐ pear. Partially premixed combustion systems are premixed flames with non-uniform fuel-

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

149

Gas turbines' manufacturers traditionally tend to use diffusion flame where fuel mixes with air by turbulent diffusion and the flame front stabilized in the locus of the stoichiometric mixture. The temperature of reactant is as high as 2000 ○C, so the acceptable temperature at the combustor walls and turbine blades would be provide by diluted air. Although the nonpremixed mixture in gas turbine combustors shows more stability in operation than pre‐ mixed mixtures, but their shortcoming is high level of nitrogen oxide emission. Two most common ways of emission reduction are water injection and catalytic converter. However, the former technique is not capable of reducing NOx to the expected level at many sites,

The idea of Dry Low NOx (DLN) systems proposed base on lean premixed combustion to reduce flame temperature by a non-stoichiometric mixture. Premixed systems can be operat‐ ed at a much lower equivalence ratio such that the flame temperature and thermal NOx pro‐ duction throughout the system are decreased comparing with a diffusion system. The disadvantage of premixed systems is flame stability, especially at low equivalence ratios. Al‐ so, there is a tendency for the flame to flashback. Indeed, the current challenge of GT's de‐

while SCR adds complexity and expense to any project.

**Figure 4.** Operating range of premixed flames [5].

oxidizer mixtures.

Where:

$$\begin{aligned} & \text{Where:}\\ & G\_{\partial} = \bigcap\_{0}^{\stackrel{\circ}{\ \mu}} (\rho \mu w + \overline{\rho u^{'} w^{'}}) r^{2} dr\\ & \stackrel{\circ}{\underset{\circ}{\ \mu}} (\rho \mu^{2} + \overline{\rho u^{'}}^{2} + (p - p\_{\Leftrightarrow})) r dr\\ & \stackrel{\circ}{\underset{\circ}{\ \mu}} \end{aligned}$$

As this equation requires velocity and pressure profile of fluid, researchers proposed vari‐ ous expressions for calculating the swirl number. Indeed, the swirl number is a non-dimen‐ sional number representing the ratio of axial flux of angular momentum to the axial flux of axial momentum times the equivalent nozzle radius [3]. Tangential entry, guided vanes and direct rotation are three principal methods for generating swirl flow.

**Figure 3.** Photo of 60° flat guided vane swirler [4].

#### **2.3. Type of flame**

Most of the literatures divide combustible mixture into three categories as premixed, nonpremixed and partially premixed combustion. If fuel and oxidizer are mixed prior ignition, then premixed flame will propagate into the unburned reactants. If fuel and air mix at the same time and same place as they react, the diffusion or non-premixed combustion will ap‐ pear. Partially premixed combustion systems are premixed flames with non-uniform fueloxidizer mixtures.

Gas turbines' manufacturers traditionally tend to use diffusion flame where fuel mixes with air by turbulent diffusion and the flame front stabilized in the locus of the stoichiometric mixture. The temperature of reactant is as high as 2000 ○C, so the acceptable temperature at the combustor walls and turbine blades would be provide by diluted air. Although the nonpremixed mixture in gas turbine combustors shows more stability in operation than pre‐ mixed mixtures, but their shortcoming is high level of nitrogen oxide emission. Two most common ways of emission reduction are water injection and catalytic converter. However, the former technique is not capable of reducing NOx to the expected level at many sites, while SCR adds complexity and expense to any project.

**Figure 4.** Operating range of premixed flames [5].

*x*

As this equation requires velocity and pressure profile of fluid, researchers proposed vari‐ ous expressions for calculating the swirl number. Indeed, the swirl number is a non-dimen‐ sional number representing the ratio of axial flux of angular momentum to the axial flux of axial momentum times the equivalent nozzle radius [3]. Tangential entry, guided vanes and

Most of the literatures divide combustible mixture into three categories as premixed, nonpremixed and partially premixed combustion. If fuel and oxidizer are mixed prior ignition,

direct rotation are three principal methods for generating swirl flow.

= (1)

*<sup>G</sup> Sn G r* q

Where:

*∞*

*∞*

(*ρuw* + *ρu* ′

148 Progress in Gas Turbine Performance

(*ρ<sup>u</sup>* <sup>2</sup> <sup>+</sup> *<sup>ρ</sup><sup>u</sup>* <sup>2</sup> ¯

*w* ¯′

)*r* <sup>2</sup> *dr*

´ + (*<sup>p</sup>* <sup>−</sup> *<sup>p</sup>∞*))*rdr*

**Figure 3.** Photo of 60° flat guided vane swirler [4].

**2.3. Type of flame**

*<sup>G</sup><sup>θ</sup>* <sup>=</sup> *∫* 0

*Gx* <sup>=</sup> *∫* 0

> The idea of Dry Low NOx (DLN) systems proposed base on lean premixed combustion to reduce flame temperature by a non-stoichiometric mixture. Premixed systems can be operat‐ ed at a much lower equivalence ratio such that the flame temperature and thermal NOx pro‐ duction throughout the system are decreased comparing with a diffusion system. The disadvantage of premixed systems is flame stability, especially at low equivalence ratios. Al‐ so, there is a tendency for the flame to flashback. Indeed, the current challenge of GT's de‐

velopers is proposing a fuel flexible combustor for a stable combustion in all engine loads. The narrow range of fuel/air mixtures between the production of excessive NOx and exces‐ sive CO is illustrated in figure 4. NOx reduces by lowering flame temperature in a leaner mixture but CO, and unburned hydrocarbons (UHC) would increase contradictorily.

By increasing combustion residence time (volume) and preventing local quenching, CO and UHC will dissociate to CO2 and the other products. CO burns away more slowly than the other radicals, so to obtain very low level emission such as 10 ppm; it requires over 4 ms. As shown in figure 5, below 1100 ○C the CO reaction becomes too slow to effectively remove the CO in an improved combustion chamber. The residence time usually does not change much on part-load because the normalized flow approximately remains constant with a var‐ iable loading.

$$NF = \frac{\dot{m}\sqrt{T}}{P} \tag{2}$$

tend the capability of dry low emission gas turbine technologies to low heat value fuels pro‐ duced by gasification of biomass and H2 enriched fuels [8-10]. Significant quantity of hydrogen in fuel has the benefit of high calorific value, but the disadvantage of high flame speed and very fast chemical times. To classify gas turbine's fuels, a common way is to split them between gas and liquid fuels, and within the gaseous fuels, to split by their calorific

**Typical composition Lower Heating Value kJ/Nm3 Typical specific fuels**

Blast furnace gas (BFG), Air blown IGCC, Biomass gasification

http://dx.doi.org/10.5772/54403

151

Coke oven gas, Corex gas

gas

Residual oils, Bio-liquids

5,500-11,200 (150-300) Refinery gas, Petrochemical gas,

Review of the New Combustion Technologies in Modern Gas Turbines

11,200-30,000 Weak natural gas, Landfill gas,

30,000-45,000 Natural gas Liquefied natural

45,000-190,000 Liquid petroleum gas (butane,

< 11,200 (< 300)

C Hydrogen power xHy = 0-40%

C propane) Refinery off-gas xHy > 10%

**Liquid fuels** CxHy, with x > 6 32,000-45,000 Diesel oil, Naphtha Crude oils,

Next-generation gas turbines will operate at higher pressure ratios and hotter turbine inlet temperatures conditions that will tend to increase nitrogen oxide emissions. To conform to future air quality requirements, lower-emitting combustion technology will be required. In this section, a number of new combustion systems have been introduced where some of

The trapped vortex combustor (TVC) may be considered as a promising technology for both pollutant emissions and pressure drop reduction. TVC is based on mixing hot combustion products and reactants at a high rate by a cavity stabilization concept. The trapped vortex

them could be found in the market, and the others are under development.

value as shown in table 1.

H2 < 10%

CH4 < 10% N2+CO > 40%

H2 > 50%

CH4 < 60%

N2+CO2 = 30-50% H2 = 10-50%

CH4 = 90%

CxHy = 5% Inert = 5%

CH4 and higher hydrocarbons

**3. New combustion systems for gas turbines**

**Ultra/Low LHV gaseous fuels**

**High hydrogen gaseous fuels**

**Medium LHV gaseous fuels**

**Natural gas**

**High LHV gaseous fuels**

**Table 1.** Classification of fuels [11].

**3.1. Trapped vortex combustion (TVC)**

Where *m*˙ is the mass flow, *T* is combustion bulk temperature and *P* is combustor pressure. This will set a lower limit for the length of the primary zone in a DLN combustion system.

**Figure 5.** Calculated reaction time to achieve a CO concentration of 10 ppm in a commercial gas turbine exhaust [6].

#### **2.4. Fuel**

One of the features of heavy-duty gas turbines is a wide fuel capability. They can operate with vast series of commercial and process by-product fuels such as natural gas, petroleum distillates, gasified coal or biomass, gas condensates, alcohols, ash-forming fuels. In a review article, Molière offered essential aspects of fuel/machine interactions in thermodynamic per‐ formance, combustion and gaseous emission [7]. To sequester and store the CO2 of fossil fuel, some new research projects aim to assess the combustion performances of alternative fuels for clean and efficient energy production by gas turbines. Another objective is to ex‐ tend the capability of dry low emission gas turbine technologies to low heat value fuels pro‐ duced by gasification of biomass and H2 enriched fuels [8-10]. Significant quantity of hydrogen in fuel has the benefit of high calorific value, but the disadvantage of high flame speed and very fast chemical times. To classify gas turbine's fuels, a common way is to split them between gas and liquid fuels, and within the gaseous fuels, to split by their calorific value as shown in table 1.


**Table 1.** Classification of fuels [11].

velopers is proposing a fuel flexible combustor for a stable combustion in all engine loads. The narrow range of fuel/air mixtures between the production of excessive NOx and exces‐ sive CO is illustrated in figure 4. NOx reduces by lowering flame temperature in a leaner

By increasing combustion residence time (volume) and preventing local quenching, CO and UHC will dissociate to CO2 and the other products. CO burns away more slowly than the other radicals, so to obtain very low level emission such as 10 ppm; it requires over 4 ms. As shown in figure 5, below 1100 ○C the CO reaction becomes too slow to effectively remove the CO in an improved combustion chamber. The residence time usually does not change much on part-load because the normalized flow approximately remains constant with a var‐

mixture but CO, and unburned hydrocarbons (UHC) would increase contradictorily.

*m T NF*

Where *m*˙ is the mass flow, *T* is combustion bulk temperature and *P* is combustor pressure. This will set a lower limit for the length of the primary zone in a DLN combustion system.

**Figure 5.** Calculated reaction time to achieve a CO concentration of 10 ppm in a commercial gas turbine exhaust [6].

One of the features of heavy-duty gas turbines is a wide fuel capability. They can operate with vast series of commercial and process by-product fuels such as natural gas, petroleum distillates, gasified coal or biomass, gas condensates, alcohols, ash-forming fuels. In a review article, Molière offered essential aspects of fuel/machine interactions in thermodynamic per‐ formance, combustion and gaseous emission [7]. To sequester and store the CO2 of fossil fuel, some new research projects aim to assess the combustion performances of alternative fuels for clean and efficient energy production by gas turbines. Another objective is to ex‐

*<sup>P</sup>* <sup>=</sup> & (2)

iable loading.

150 Progress in Gas Turbine Performance

**2.4. Fuel**

#### **3. New combustion systems for gas turbines**

Next-generation gas turbines will operate at higher pressure ratios and hotter turbine inlet temperatures conditions that will tend to increase nitrogen oxide emissions. To conform to future air quality requirements, lower-emitting combustion technology will be required. In this section, a number of new combustion systems have been introduced where some of them could be found in the market, and the others are under development.

#### **3.1. Trapped vortex combustion (TVC)**

The trapped vortex combustor (TVC) may be considered as a promising technology for both pollutant emissions and pressure drop reduction. TVC is based on mixing hot combustion products and reactants at a high rate by a cavity stabilization concept. The trapped vortex combustion concept has been under investigation since the early 1990's. The earlier studies of TVC have been concentrated on liquid fuel applications for aircraft combustors [12].

a number of experiments and numerical simulations have been performed to enhance the stability of reacting flow inside trapped vortex. Xing et al. experimentally investigated lean blow-out of several combustors and the performance of slight temperature-raise in a single trapped vortex [20, 21]. In an experimental laboratory research, Bucher et al. proposed a new

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

153

Lean direct injection (LDI) and rich-burn/quick-quench/lean-burn (RQL) are two of the prominent low-emissions concepts for gas turbines. LDI operates the primary combustion region lean, hence, adequate flame stabilization has to be ensured; RQL is rich in the pri‐ mary zone with a transition to lean combustion by rapid mixing with secondary air down‐ stream. Hence, both concepts avoid stoichiometric combustion as much as possible, but flame stabilization and combustion in the main heat release region are entirely different. Relative to aviation engines, the need for reliability and safety has led to a focus on LDI of liquid fuels [23]. However, RQL combustor technology is of growing interest for stationary gas turbines due to the attributes of more effectively processing of fuels with complex com‐ position. The concept of RQL was proposed in 1980 as a significant effort for reducing NOx

It is known that the primary zone of a gas turbine combustor operates most effectively with rich mixture ratios so, a "rich-burn" condition in the primary zone enhances the stability of the combustion reaction by producing and sustaining a high concentration of energetic hy‐ drogen and hydrocarbon radical species. Secondly, rich burn conditions minimize the pro‐ duction of nitrogen oxides due to the relative low temperatures and low population of oxygen containing intermediate species. Critical factors of a RQL that need to be considered are careful tailoring of rich and lean equivalence ratios and very fast cooling rates. So the combustion regime shifts rapidly from rich to lean without going through the high NOx route as shown in figure 7. The drawback of this technology is increased hardware and com‐

The mixing of the injected air takes the reaction to the lean-burn zone and rapidly reduces their temperature as well. On the other hand, the temperature must be high enough to burn CO and UHC. Thus, the equivalence ratio for the lean-burn zone must be carefully selected to satisfy all emissions requirements. Typically the equivalence ratio of fuel-rich primary

Turbulent jet in a cross-flow is an important characteristic of RQL; so many researches have been conducted to improve it. The mixing limitation in a design of RQL/TVC com‐ bustion system addressed by Straub et al. [26]. Coaxial swirling air discussed experimen‐ tally by Cozzi and Coghe [27]. Furthermore, an experimental study of the effects of elevated pressure and temperature on jet mixing and emissions in an RQL reported by Jermakian et al. [28]. Fuel flexible combustion with RQL system is an interest of turbine manufacturer. GE reported results of a RQL test stand in their integrated gasification combined cycle (IGCC) power plants program [29, 30]. The test of Siemens-Westinghouse Multi-Annular Swirl Burner (MASB) was successfully performed at the University of

zone is 1.2 to 1.6 and lean-burn combustion occurs between 0.5 and 0.7 [25].

design for lean-premixed trapped vortex combustor [22].

**3.2. Rich burn, quick- mix, lean burn (RQL)**

emission [24].

plexity of the system.

The trapped vortex technology offers several advantages as gas turbines burner:


Flame stability is achieved through the use of recirculation zones to provide a continuous ignition source which facilitates the mixing of hot combustion products with the incoming fuel and air mixture [13]. Turbulence occurring in a TVC combustion chamber is "trapped" within a cavity where reactants are injected and efficiently mixed. Since part of the combus‐ tion occurs within the recirculation zone, a "typically" flameless regime can be achieved, while a trapped turbulent vortex may provide significant pressure drop reduction [14]. Be‐ sides this, TVC is having the capability of operating as a staged combustor if the fuel is in‐ jected into both the cavities and the main airflow. Generally, staged combustion systems are having the potential of achieving about 10 to 40% reduction in NOx emissions [15]. It can also be operated as a rich-burn, quick-quench lean-burn (RQL) combustor when all of the fuel is injected into the cavities [16].

**Figure 6.** Trapped vortex combustor schematic.

An experiment in NASA with water injected TVC demonstrated a reduction in NOx by a factor three in a natural gas fueled and up to two in a liquid JP-8 fueled over a range in wa‐ ter/fuel and fuel/air ratios [17]. Replacement of natural gas fuel with syngas and hydrogen fuels has been studied numerically by Ghenai et al. [18]. The effects of secondary air jet mo‐ mentum on cavity flow structure of TVC have been studied recently by Kumar and Mishra [19]. Although the actual stabilization mechanism facilitated by the TVC is relatively simple, a number of experiments and numerical simulations have been performed to enhance the stability of reacting flow inside trapped vortex. Xing et al. experimentally investigated lean blow-out of several combustors and the performance of slight temperature-raise in a single trapped vortex [20, 21]. In an experimental laboratory research, Bucher et al. proposed a new design for lean-premixed trapped vortex combustor [22].

#### **3.2. Rich burn, quick- mix, lean burn (RQL)**

combustion concept has been under investigation since the early 1990's. The earlier studies of TVC have been concentrated on liquid fuel applications for aircraft combustors [12].

**•** It is possible to operate at high excess air premixed regime, given the ability to support

**•** NOx emissions reach extremely low levels without dilution or post-combustion treat‐

Flame stability is achieved through the use of recirculation zones to provide a continuous ignition source which facilitates the mixing of hot combustion products with the incoming fuel and air mixture [13]. Turbulence occurring in a TVC combustion chamber is "trapped" within a cavity where reactants are injected and efficiently mixed. Since part of the combus‐ tion occurs within the recirculation zone, a "typically" flameless regime can be achieved, while a trapped turbulent vortex may provide significant pressure drop reduction [14]. Be‐ sides this, TVC is having the capability of operating as a staged combustor if the fuel is in‐ jected into both the cavities and the main airflow. Generally, staged combustion systems are having the potential of achieving about 10 to 40% reduction in NOx emissions [15]. It can also be operated as a rich-burn, quick-quench lean-burn (RQL) combustor when all of the

An experiment in NASA with water injected TVC demonstrated a reduction in NOx by a factor three in a natural gas fueled and up to two in a liquid JP-8 fueled over a range in wa‐ ter/fuel and fuel/air ratios [17]. Replacement of natural gas fuel with syngas and hydrogen fuels has been studied numerically by Ghenai et al. [18]. The effects of secondary air jet mo‐ mentum on cavity flow structure of TVC have been studied recently by Kumar and Mishra [19]. Although the actual stabilization mechanism facilitated by the TVC is relatively simple,

The trapped vortex technology offers several advantages as gas turbines burner:

**•** It is possible to burn a variety of fuels with medium and low calorific value.

**•** Produces the extension of the flammability limits and improves flame stability.

high-speed injections, which avoids flashback.

fuel is injected into the cavities [16].

**Figure 6.** Trapped vortex combustor schematic.

ments.

152 Progress in Gas Turbine Performance

Lean direct injection (LDI) and rich-burn/quick-quench/lean-burn (RQL) are two of the prominent low-emissions concepts for gas turbines. LDI operates the primary combustion region lean, hence, adequate flame stabilization has to be ensured; RQL is rich in the pri‐ mary zone with a transition to lean combustion by rapid mixing with secondary air down‐ stream. Hence, both concepts avoid stoichiometric combustion as much as possible, but flame stabilization and combustion in the main heat release region are entirely different. Relative to aviation engines, the need for reliability and safety has led to a focus on LDI of liquid fuels [23]. However, RQL combustor technology is of growing interest for stationary gas turbines due to the attributes of more effectively processing of fuels with complex com‐ position. The concept of RQL was proposed in 1980 as a significant effort for reducing NOx emission [24].

It is known that the primary zone of a gas turbine combustor operates most effectively with rich mixture ratios so, a "rich-burn" condition in the primary zone enhances the stability of the combustion reaction by producing and sustaining a high concentration of energetic hy‐ drogen and hydrocarbon radical species. Secondly, rich burn conditions minimize the pro‐ duction of nitrogen oxides due to the relative low temperatures and low population of oxygen containing intermediate species. Critical factors of a RQL that need to be considered are careful tailoring of rich and lean equivalence ratios and very fast cooling rates. So the combustion regime shifts rapidly from rich to lean without going through the high NOx route as shown in figure 7. The drawback of this technology is increased hardware and com‐ plexity of the system.

The mixing of the injected air takes the reaction to the lean-burn zone and rapidly reduces their temperature as well. On the other hand, the temperature must be high enough to burn CO and UHC. Thus, the equivalence ratio for the lean-burn zone must be carefully selected to satisfy all emissions requirements. Typically the equivalence ratio of fuel-rich primary zone is 1.2 to 1.6 and lean-burn combustion occurs between 0.5 and 0.7 [25].

Turbulent jet in a cross-flow is an important characteristic of RQL; so many researches have been conducted to improve it. The mixing limitation in a design of RQL/TVC com‐ bustion system addressed by Straub et al. [26]. Coaxial swirling air discussed experimen‐ tally by Cozzi and Coghe [27]. Furthermore, an experimental study of the effects of elevated pressure and temperature on jet mixing and emissions in an RQL reported by Jermakian et al. [28]. Fuel flexible combustion with RQL system is an interest of turbine manufacturer. GE reported results of a RQL test stand in their integrated gasification combined cycle (IGCC) power plants program [29, 30]. The test of Siemens-Westinghouse Multi-Annular Swirl Burner (MASB) was successfully performed at the University of Tennessee Space Institute in Tullahoma [31]. Others, such as references [32-35] utilize CFD to investigate the performance of RQL combustor.

**Figure 8.** COSTAIR combustion concept [38].

**3.4. Mild combustion**

HiTAC5

, HiCOT6

5 High Temperature Air Combustion 6 High-temperature Combustion Technology 7 Moderate or Intense Low-oxygen Dilution

9 Colorless Distributed Combustion

8 FLameless OXidation

, MILD7

Staged combustion can occur in either a radial or axial pattern, but in either case the goal is to design each stage to optimize particular performance aspects. The main advantages or

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

155

Heat recirculating combustion was clearly described by Weinberg as a concept for improv‐ ing the thermal efficiency [40]. In 1989, a surprising phenomenon was observed during ex‐ periments with a self-recuperative burner. At furnace temperatures of 1000°C and about 650°C air preheated temperature; no flame could be seen, but the fuel was completely burnt. Furthermore, the CO and NOx emissions from the furnace were considerably low [41]. Dif‐ ferent combustion zones against rate of dilution and oxygen content is shown in figure 9. In flameless combustion, the oxidation of fuel occurs with a very limited oxygen supply at a very high temperature. Spontaneous ignition occurs and progresses with no visible or audi‐ ble signs of the flames usually associated with burning. The chemical reaction zone is quite diffuse, and this leads to almost uniform heat release and a smooth temperature profile. All these factors could result in a much more efficient process as well as reducing emissions.

Flameless combustion is defined where the reactants exceed self-ignition temperature as well as entrain enough inert combustion products to reduce the final reaction temperature [42]. In the other word, the essence of this technology is that fuel is oxidized in an environ‐ ment that contains a substantial amount of inert (flue) gases and some, typically not more than 3–5%, oxygen. Several different expressions are used to identify similar though such as

temperature by preheating systems such as regenerators. HiCOT commonly belongs to the

and CDC9

. HiTAC refers to increase the air

combustion, FLOX8

major drawbacks of each type have been discussed by Lefebvre [25].

**Figure 7.** Rich-Burn, Quick-Mix, Lean-Burn combustor.

#### **3.3. Staged air combustion**

The COSTAIR4 combustion concept uses continuously staged air and internal recirculation within the combustion chamber to obtain a stable combustion with low NOX and CO emis‐ sions. Research work on staged combustors started in the early 1970s under of the Energy Efficient Engine (E3 ) Program in the USA [36] and now widely used in industrial engines burning gaseous fuels, in both axial and radial configurations. The aero-derived GE LM6000 and CFM56-5B as well as RR211 DLE industrial engine employ staged combustion of pre‐ mixed gaseous fuel/air mixtures. Recently, a research project proposed a COSTAIR burner system optimized for low calorific gases within a micro gas turbine [37].

The principle of staged air combustion is illustrated in Figure 8. It consists of a coaxial tube; the combustion air flows through the inner tube and the fuel through the outer cylinder ring. The combustion air is continually distributed throughout the combustion chamber by an air distributor with numerous openings on its contour, and fuel enters by several jets ar‐ ranged around the air distributor.

The COSTAIR burner has the advantages of operating in full diffusion mode or in partially premixed mode. The heat is released more uniformly throughout the combustion chamber also the recirculated gas absorb some of the heat of combustion. It capable to work stable at cold combustor walls as well as high air ratio. Experimental measurements show that this combustion system allows clean exhaust. For instance, in an experimental research project of European Commission [39], NOx emission values was in the range of 2-4 ppm at an air ratio of 2.5 over different loading. Furthermore, the corresponding CO emission was less than 7 ppm.

<sup>4</sup> COntinuous STaged Air

**Figure 8.** COSTAIR combustion concept [38].

Staged combustion can occur in either a radial or axial pattern, but in either case the goal is to design each stage to optimize particular performance aspects. The main advantages or major drawbacks of each type have been discussed by Lefebvre [25].

#### **3.4. Mild combustion**

Tennessee Space Institute in Tullahoma [31]. Others, such as references [32-35] utilize

combustion concept uses continuously staged air and internal recirculation

) Program in the USA [36] and now widely used in industrial engines

within the combustion chamber to obtain a stable combustion with low NOX and CO emis‐ sions. Research work on staged combustors started in the early 1970s under of the Energy

burning gaseous fuels, in both axial and radial configurations. The aero-derived GE LM6000 and CFM56-5B as well as RR211 DLE industrial engine employ staged combustion of pre‐ mixed gaseous fuel/air mixtures. Recently, a research project proposed a COSTAIR burner

The principle of staged air combustion is illustrated in Figure 8. It consists of a coaxial tube; the combustion air flows through the inner tube and the fuel through the outer cylinder ring. The combustion air is continually distributed throughout the combustion chamber by an air distributor with numerous openings on its contour, and fuel enters by several jets ar‐

The COSTAIR burner has the advantages of operating in full diffusion mode or in partially premixed mode. The heat is released more uniformly throughout the combustion chamber also the recirculated gas absorb some of the heat of combustion. It capable to work stable at cold combustor walls as well as high air ratio. Experimental measurements show that this combustion system allows clean exhaust. For instance, in an experimental research project of European Commission [39], NOx emission values was in the range of 2-4 ppm at an air ratio of 2.5 over different loading. Furthermore, the corresponding CO emission was less than 7

system optimized for low calorific gases within a micro gas turbine [37].

CFD to investigate the performance of RQL combustor.

**Figure 7.** Rich-Burn, Quick-Mix, Lean-Burn combustor.

**3.3. Staged air combustion**

154 Progress in Gas Turbine Performance

ranged around the air distributor.

The COSTAIR4

ppm.

4 COntinuous STaged Air

Efficient Engine (E3

Heat recirculating combustion was clearly described by Weinberg as a concept for improv‐ ing the thermal efficiency [40]. In 1989, a surprising phenomenon was observed during ex‐ periments with a self-recuperative burner. At furnace temperatures of 1000°C and about 650°C air preheated temperature; no flame could be seen, but the fuel was completely burnt. Furthermore, the CO and NOx emissions from the furnace were considerably low [41]. Dif‐ ferent combustion zones against rate of dilution and oxygen content is shown in figure 9. In flameless combustion, the oxidation of fuel occurs with a very limited oxygen supply at a very high temperature. Spontaneous ignition occurs and progresses with no visible or audi‐ ble signs of the flames usually associated with burning. The chemical reaction zone is quite diffuse, and this leads to almost uniform heat release and a smooth temperature profile. All these factors could result in a much more efficient process as well as reducing emissions.

Flameless combustion is defined where the reactants exceed self-ignition temperature as well as entrain enough inert combustion products to reduce the final reaction temperature [42]. In the other word, the essence of this technology is that fuel is oxidized in an environ‐ ment that contains a substantial amount of inert (flue) gases and some, typically not more than 3–5%, oxygen. Several different expressions are used to identify similar though such as HiTAC5 , HiCOT6 , MILD7 combustion, FLOX8 and CDC9 . HiTAC refers to increase the air temperature by preheating systems such as regenerators. HiCOT commonly belongs to the

<sup>5</sup> High Temperature Air Combustion

<sup>6</sup> High-temperature Combustion Technology

<sup>7</sup> Moderate or Intense Low-oxygen Dilution

<sup>8</sup> FLameless OXidation

<sup>9</sup> Colorless Distributed Combustion

wider sense, which exploits high-temperature reactants; therefore, it is not limited to air. A combustion process is named FLOX or MILD when the inlet temperature of the main reac‐ tant flow is higher than mixture autoignition temperature and the maximum allowable tem‐ perature increase during combustion is lower than mixture autoignition temperature, due to dilution [42]. The common key feature to achieve reactions in CDC mode (non-premixed conditions) is the separation and controlled mixing of higher momentum air jet and the low‐ er momentum fuel jet, large amount of gas recirculation and higher turbulent mixing rates to achieve spontaneous ignition of the fuel to provide distributed combustion reactions [43]. Figure 10 schematically shows a comparison between conventional burner and flameless combustion.

**•** Highly transparent flame,

**•** Low acoustic oscillation and

**•** Low NOx and CO emissions.

**Figure 10.** Flame (left) and flameless (right) firing.

plication in a number of publications [43, 50-55].

**3.5. Surface stabilized combustion**

MW/m3

MW/m3

In spite of a number of activities for industrial furnaces, the application of flameless com‐ bustion in the gas-turbine combustion system is in the preliminary phase [44]. The results from techno-economic analysis of Wang et al. showed that the COSTAIR and FLOX cases had technical and economic advantages over SCR [45]. Luckerath, R., et al., investigated flameless combustion in forward flow configuration in elevated pressure up to 20atm for ap‐ plication to gas turbine combustors [44, 46]. In a novel design of Levy et al. that named FLOXCOM, flameless concept has been proposed for gas turbines by establishing large re‐ circulation zone in the combustion chamber [47, 48]. Lammel et al. developed a FLOX com‐ bustion at high power density and achieved low NOx and CO levels [49]. The concept of colorless distributed combustion has been demonstrated by Gupta et al. for gas turbine ap‐

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

157

One specification of gas turbine combustor is higher thermal intensity range (at least 5

concept, except a technology named NanoSTAR from Alzeta Corporation. Alzeta reported



**Figure 9.** Different combustion regimes [64].

To recap, the main characteristics of flameless oxidation combustion are:


<sup>10</sup> A dimensionless number, equal to the ratio of the turbulence time scale to the time it takes chemical reaction.

**•** Highly transparent flame,

wider sense, which exploits high-temperature reactants; therefore, it is not limited to air. A combustion process is named FLOX or MILD when the inlet temperature of the main reac‐ tant flow is higher than mixture autoignition temperature and the maximum allowable tem‐ perature increase during combustion is lower than mixture autoignition temperature, due to dilution [42]. The common key feature to achieve reactions in CDC mode (non-premixed conditions) is the separation and controlled mixing of higher momentum air jet and the low‐ er momentum fuel jet, large amount of gas recirculation and higher turbulent mixing rates to achieve spontaneous ignition of the fuel to provide distributed combustion reactions [43]. Figure 10 schematically shows a comparison between conventional burner and flameless

combustion.

156 Progress in Gas Turbine Performance

**Figure 9.** Different combustion regimes [64].

**•** Low Damköhler number (Da10),

**•** Reduce temperature peaks,

**•** Low stable adiabatic flame temperature,

**•** Reduced oxygen concentration at the reactance,

To recap, the main characteristics of flameless oxidation combustion are:

**•** Recirculation of combustion products at high temperature (normally > 1000 ○C),

10 A dimensionless number, equal to the ratio of the turbulence time scale to the time it takes chemical reaction.


**Figure 10.** Flame (left) and flameless (right) firing.

In spite of a number of activities for industrial furnaces, the application of flameless com‐ bustion in the gas-turbine combustion system is in the preliminary phase [44]. The results from techno-economic analysis of Wang et al. showed that the COSTAIR and FLOX cases had technical and economic advantages over SCR [45]. Luckerath, R., et al., investigated flameless combustion in forward flow configuration in elevated pressure up to 20atm for ap‐ plication to gas turbine combustors [44, 46]. In a novel design of Levy et al. that named FLOXCOM, flameless concept has been proposed for gas turbines by establishing large re‐ circulation zone in the combustion chamber [47, 48]. Lammel et al. developed a FLOX com‐ bustion at high power density and achieved low NOx and CO levels [49]. The concept of colorless distributed combustion has been demonstrated by Gupta et al. for gas turbine ap‐ plication in a number of publications [43, 50-55].

#### **3.5. Surface stabilized combustion**

One specification of gas turbine combustor is higher thermal intensity range (at least 5 MW/m3 -atm) than industrial furnaces which operate at thermal intensity of less than 1 MW/m3 -atm. Therefore, designs of gas turbine's combustors are based on turbulent flow concept, except a technology named NanoSTAR from Alzeta Corporation. Alzeta reported the proof-of-concept of high thermal intensity laminar surface stabilized flame by using a porous metal-fiber mat since 2001 [56-58]. Lean premixed combustion technology is limited by the apparition of combustion instabilities, which induce high pressure fluctuations, which can produce turbine damage, flame extinction, and CO emissions [59]. However, full scale test of NanoSTAR demonstrated low emissions performance, robust ignition and ex‐ tended turndown ratio [60]. In particular, the following characteristics form the key specifi‐ cations of NanoSTAR for distributed power generation gas turbine combustors [61]:

The operation of this type of surface stabilized combustion is characterized by the schematic in Figure 12(left), which shows premixed fuel and air passing through the metal fiber mat in two distinct zones. Premixed fuel comes through the low conductivity porous and burns in narrow zones, A, as it leaves the surface. Under lean conditions this will manifest as very short laminar flamelets, but under rich conditions the surface combustion will become a dif‐ fusion dominated reaction stabilized just over a millimeter above the metal matrix, which proceeds without visible flame and heats the outer surface of the mat to incandescence. Sec‐ ondly, adjacent to these radiant zones, the porous plate is perforated to allow a high flow of the premixed fuel and air. This flow forms a high intensity flame, B, stabilized by the radiant zones so, it is possible to achieve very high fluxes of energy, up to 2MMBtu/hr/ft² [63]. A picture of an atmospheric burner in operation clearly shows the technology in action (right

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

159

The specific perforation arrangement and pattern control the size and shape of the laminar flamelets. The perforated zones operate at flow velocities of up to 10 times the laminar flame speed producing a factor of ten stretch of the flame surface and resulting in a large laminar flamelets. The alternating arrangement of laminar blue flames and surface combustion, al‐ lows high firing rates to be achieved before flame liftoff occurs, with the surface combustion stabilizing the long laminar flames by providing a pool of hot combustion radicals at the

A review of technologies for reducing NOx emissions as well as increasing thermal efficien‐ cy and improving combustion stability has been reported here. Trade-offs when installing low NOx burners in gas turbines include the potential for decreased flame stability, reduced

of figure 12).

flame edges.

**4. Conclusion**

**Figure 12.** Surface stabilized burner pad firing at atmospheric conditions.


A single prototype burner Porous burner which sized to fit inside an annular combustion liner (about 2.5 inches in diameter by 7 inches in length) is shown in figure 11 with its ar‐ rangement in a typical combustor.

**Figure 11.** NanoSTAR burner and its arrangement in a canted combustion system [62].

The operation of this type of surface stabilized combustion is characterized by the schematic in Figure 12(left), which shows premixed fuel and air passing through the metal fiber mat in two distinct zones. Premixed fuel comes through the low conductivity porous and burns in narrow zones, A, as it leaves the surface. Under lean conditions this will manifest as very short laminar flamelets, but under rich conditions the surface combustion will become a dif‐ fusion dominated reaction stabilized just over a millimeter above the metal matrix, which proceeds without visible flame and heats the outer surface of the mat to incandescence. Sec‐ ondly, adjacent to these radiant zones, the porous plate is perforated to allow a high flow of the premixed fuel and air. This flow forms a high intensity flame, B, stabilized by the radiant zones so, it is possible to achieve very high fluxes of energy, up to 2MMBtu/hr/ft² [63]. A picture of an atmospheric burner in operation clearly shows the technology in action (right of figure 12).

**Figure 12.** Surface stabilized burner pad firing at atmospheric conditions.

The specific perforation arrangement and pattern control the size and shape of the laminar flamelets. The perforated zones operate at flow velocities of up to 10 times the laminar flame speed producing a factor of ten stretch of the flame surface and resulting in a large laminar flamelets. The alternating arrangement of laminar blue flames and surface combustion, al‐ lows high firing rates to be achieved before flame liftoff occurs, with the surface combustion stabilizing the long laminar flames by providing a pool of hot combustion radicals at the flame edges.

#### **4. Conclusion**

the proof-of-concept of high thermal intensity laminar surface stabilized flame by using a porous metal-fiber mat since 2001 [56-58]. Lean premixed combustion technology is limited by the apparition of combustion instabilities, which induce high pressure fluctuations, which can produce turbine damage, flame extinction, and CO emissions [59]. However, full scale test of NanoSTAR demonstrated low emissions performance, robust ignition and ex‐ tended turndown ratio [60]. In particular, the following characteristics form the key specifi‐

**•** Turbine Rotor Inlet Temperatures (TRIT) over 2200°F (valid for the Mercury 50, although

A single prototype burner Porous burner which sized to fit inside an annular combustion liner (about 2.5 inches in diameter by 7 inches in length) is shown in figure 11 with its ar‐

cations of NanoSTAR for distributed power generation gas turbine combustors [61]:

**•** Total combustor pressure drop limited to 2-4% of the system pressure.

**•** Operation at combustion air preheat temperatures up to 1150°F.

**•** Volumetric firing rates approaching 2 MMBtu/hr/atm/ft³.

**•** Operation with axial combustors or external can combustors.

**•** Expected component lifetimes of 30,000 hours for industrial turbines.

**Figure 11.** NanoSTAR burner and its arrangement in a canted combustion system [62].

**•** The combustor fuel is limited to natural gas.

158 Progress in Gas Turbine Performance

Allison has operated combustors at 2600°F).

rangement in a typical combustor.

A review of technologies for reducing NOx emissions as well as increasing thermal efficien‐ cy and improving combustion stability has been reported here. Trade-offs when installing low NOx burners in gas turbines include the potential for decreased flame stability, reduced operating range and more strict fuel quality specifications. In the other word, although, the turbine inlet temperature is the major factor determining the overall efficiency of the gas tur‐ bine but higher inlet temperatures will result in larger NOx emissions. So the essential re‐ quirement of new combustor design is a trade-off between low NOx and improved efficiency.

[10] Juste GL (2006) Hydrogen Injection as Additional Fuel in Gas Turbine Combustor. Evaluation of Effects. International Journal of Hydrogen Energy. 31: 2112-2121. [11] Jones R, Goldmeer J and Monetti B (2011) Addressing Gas Turbine Fuel Flexibility.

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

161

[12] Haynes J, Janssen J, Russell C and Huffman M (2006) Advanced Combustion Sys‐ tems for Next Generation Gas Turbines. United States. Dept. of Energy, Washington,

[13] Sturgess GJ and Hsu KY (1998) Combustion Characteristics of a Trapped Vortex

[14] Bruno C and Losurdo M (2007) The Trapped Vortex Combustor: An Advanced Com‐ bustion Technology for Aerospace and Gas Turbine Applications. In: Syred N and Khalatov A, Syred N and Khalatov A, editors. Advanced Combustion and Aerother‐

[15] Mishra DP (2008) Fundamentals of Combustion, Prentice-Hall Of India Pvt. Limited. [16] Acharya S, Mancilla PC and Chakka P (2001) Performance of a Trapped Vortex Spray

[17] Hendricks RC, Shouse DT and Roquemore WM (2005) Water Injected Turbomachi‐

[18] Ghenai C, Zbeeb K and Janajreh I (2012) Combustion of Alternative Fuels in Vortex

[19] Ezhil Kumar PK and Mishra DP (2011) Numerical Simulation of Cavity Flow Struc‐ ture in an Axisymmetric Trapped Vortex Combustor. Aerospace Science and Tech‐

[20] Xing F, Wang P, Zhang S, Zou J, Zheng Y, Zhang R and Fan W (2012) Experiment and Simulation Study on Lean Blow-out of Trapped Vortex Combustor with Various

[21] Xing F, Zhang S, Wang P and Fan W (2010) Experimental Investigation of a Single Trapped-Vortex Combustor with a Slight Temperature Raise. Aerospace Science and

[22] Bucher J, Edmonds RG, Steele RC, Kendrick DW, Chenevert BC and Malte PC (2003) The Development of a Lean-Premixed Trapped Vortex Combustor. ASME Turbo Ex‐

[23] Dunn-Rankin D (2008) Lean Combustion: Technology and Control, Academic Press,

[24] Mosier SA and Pierce RM (1980) Advanced Combustion Systems for Stationary Gas Turbine Engines. Volume I. Review and Preliminary Evaluation. Final Report De‐

Combustor. ASME International Gas Turbine Conference. ASME.

Trapped Combustor. Energy Conversion and Management. In press

Aspect Ratios. Aerospace Science and Technology. 18: 48-55.

cember 1975-September 1976. pp Medium: X; Size: Pages: 49.

Combustor. Applied vehicle technology panel symposium.

mal Technologies. Springer Netherlands, pp 365-384.

nery. NASA, Glenn Research Center

nology. In Press

Technology. 14: 520-525.

USA.

po 2003 Power for Land, Sea, and Air.

GE Energy

D.C.; Oak Ridge, Tenn.

#### **Author details**

M. Khosravy el\_Hossaini

Research Institute of Petroleum Industry, Iran

#### **References**


[10] Juste GL (2006) Hydrogen Injection as Additional Fuel in Gas Turbine Combustor. Evaluation of Effects. International Journal of Hydrogen Energy. 31: 2112-2121.

operating range and more strict fuel quality specifications. In the other word, although, the turbine inlet temperature is the major factor determining the overall efficiency of the gas tur‐ bine but higher inlet temperatures will result in larger NOx emissions. So the essential re‐ quirement of new combustor design is a trade-off between low NOx and improved

[1] Chambers A and Trottier S (2007) Technologies for Reducing Nox Emissions from Gas-Fired Stationary Combustion Sources. Alberta Research Council, Edmonton,

[2] Kiel B, Garwick LK, Gord JR, Miller J, Lynch A, Hill R and Phillips S (2007) A De‐ tailed Investigation of Bluff Body Stabilized Flames. 45th AIAA Aerospace Sciences

[4] Jaafar MNM, Jusoff K, Osman MS and Ishak MSA (2011) Combustor Aerodynamic Using Radial Swirler. International Journal of the Physical Sciences. 6: 3091 - 3098.

[5] Moore MJ (1997) Nox Emission Control in Gas Turbines for Combined Cycle Gas Turbine Plant. Proceedings of the Institution of Mechanical Engineers, Part A: Jour‐

[6] Kajita S and Dalla Betta R (2003) Achieving Ultra Low Emissions in a Commercial 1.4 Mw Gas Turbine Utilizing Catalytic Combustion. Catalysis Today. 83: 279-288.

[7] Molière M (2000) Stationary Gas Turbines and Primary Energies: A Review of Fuel Influence on Energy and Combustion Performances. International Journal of Ther‐

[8] Gupta KK, Rehman A and Sarviya RM (2010) Bio-Fuels for the Gas Turbine: A Re‐

[9] Gökalp I and Lebas E (2004) Alternative Fuels for Industrial Gas Turbines (Aftur).

view. Renewable and Sustainable Energy Reviews. 14: 2946-2955.

[3] Gupta AK, Lilley DG and Syred N (1984) Swirl Flows, Abacus Press.

efficiency.

**Author details**

**References**

Canada

Meeting and Exhibit.

nal of Power and Energy. 211: 43-52.

Applied Thermal Engineering. 24: 1655-1663.

mal Sciences. 39: 141-172.

M. Khosravy el\_Hossaini

160 Progress in Gas Turbine Performance

Research Institute of Petroleum Industry, Iran


[25] Lefebvre AH and Ballal DR (2010) Gas Turbine Combustion: Alternative Fuels and Emissions, Taylor & Francis.

[38] Al-Halbouni A, Flamme M, Giese A, Scherer V, Michalski B and Wünning JG (2004) New Burner Systems with High Fuel Flexibility for Gas Turbines. 2nd International

Review of the New Combustion Technologies in Modern Gas Turbines

http://dx.doi.org/10.5772/54403

163

[39] Flamme M (2004) New Combustion Systems for Gas Turbines (Ngt). Applied Ther‐

[40] WEINBERG F (1996) Heat-Recirculating Burners : Principles and Some Recent Devel‐

[42] Cavaliere A and de Joannon M (2004) Mild Combustion. Progress in Energy and

[43] Arghode VK, Gupta AK and Bryden KM (2012) High Intensity Colorless Distributed Combustion for Ultra Low Emissions and Enhanced Performance. Applied Energy.

[44] Li P, Mi J, Dally B, Wang F, Wang L, Liu Z, Chen S and Zheng C (2011) Progress and Recent Trend in Mild Combustion. SCIENCE CHINA Technological Sciences. 54:

[45] Wang YD, Huang Y, McIlveen-Wright D, McMullan J, Hewitt N, Eames P and Re‐ zvani S (2006) A Techno-Economic Analysis of the Application of Continuous Stag‐ ed-Combustion and Flameless Oxidation to the Combustor Design in Gas Turbines.

[46] Luckerath R, Meier W and Aigner M (2008) Flox Combustion at High Pressure with Different Fuel Compositions. Journal of Engineering for Gas Turbines and Power.

[47] Costa M, Melo M, Sousa J and Levy Y (2009) Experimental Investigation of a Novel Combustor Model for Gas Turbines. Journal of Propulsion and Power 25: 609-617.

[48] Levy Y, Sherbaum V and Arfi P (2004) Basic Thermodynamics of Floxcom, the Low-Nox Gas Turbines Adiabatic Combustor. Applied Thermal Engineering. 24:

[49] Lammel O, Schutz H, Schmitz G, Luckerath R, Stohr M, Noll B, Aigner M, Hase M and Krebs W (2010) Flox Combustion at High Power Density and High Flame Tem‐

[50] Arghode VK and Gupta AK (2011) Development of High Intensity Cdc Combustor

[51] Khalil AEE and Gupta AK (2011) Distributed Swirl Combustion for Gas Turbine Ap‐

[52] Arghode VK and Gupta AK (2010) Effect of Flow Field for Colorless Distributed Combustion (Cdc) for Gas Turbine Combustion. Applied Energy. 87: 1631-1640.

peratures. Journal of Engineering for Gas Turbines and Power. 132: 121503.

for Gas Turbine Engines. Applied Energy. 88: 963-973.

plication. Applied Energy. 88: 4898-4907.

Conference on Industrial Gas Turbine Technologies.

opments. Combustion Science and Technology. 121: 3-22.

[41] Wünning J (2005) Flameless Oxidation. 6th HiTACG Symposium.

mal Engineering. 24: 1551-1559.

Combustion Science. 30: 329-366.

Fuel Processing Technology. 87: 727-736.

92: 822-830.

255-269.

130: 011505.

1593-1605.


[25] Lefebvre AH and Ballal DR (2010) Gas Turbine Combustion: Alternative Fuels and

[26] Straub DL, Casleton KH, Lewis RE, Sidwell TG, Maloney DJ and Richards GA (2005) Assessment of Rich-Burn, Quick-Mix, Lean-Burn Trapped Vortex Combustor for Sta‐ tionary Gas Turbines. Journal of engineering for gas turbines and power. 127: 36-41.

[27] Cozzi F and Coghe A (2012) Effect of Air Staging on a Coaxial Swirled Natural Gas

[28] Jermakian V, McDonell VG and Samuelsen GS (2012) Experimental Study of the Ef‐ fects of Elevated Pressure and Temperature on Jet Mixing and Emissions in an Rql Combustor for Stable, Efficient and Low Emissions Gas Turbine Applications. Ad‐

[29] Feitelberg AS, Jackson MR, Lacey MA, Manning KS and Ritter AM (1996) Design and Performance of a Low Btu Fuel Rich-Quench-Lean Gas Turbine Combustor. Ad‐ vanced coal-fired power systems review meeting. USA DOE, Morgantown Energy

[30] Feitelberg AS and Lacey MA (1998) The Ge Rich-Quench-Lean Gas Turbine Combus‐ tor. Journal of Engineering for Gas Turbines and Power, Transactions of the ASME.

[31] Brushwood J (1999) Syngas Combustor for Fluidized Bed Applications 15th Annual

[32] Howe GW, Li Z, Shih TI-P and Nguyen HL (1991) Simulation of Mixing in the Quick Quench Region of a Richburn-Quick Quench Mix-Lean Burn Combustor. 29th Aero‐

[33] Cline MC, Micklow GJ, Yang SL and Nguyen HL (1992) Numerical Analysis of the

[34] Talpallikar MV, Smith CE, Lai MC and Holdeman JD (1992) Cfd Analysis of Jet Mix‐ ing in Low Nox Flametube Combustors. Journal of Engineering for Gas Turbines and

[35] Blomeyer M, Krautkremer B, Hennecke DK and Doerr T (1999) Mixing Zone Optimi‐ zation of a Rich-Burn/Quick-Mix/Lean-Burn Combustor. Journal of Propulsion and

[36] Wulff A and Hourmouziadis J (1997) Technology Review of Aeroengine Pollutant

[37] Leicher J, Giese A, Görner K, Scherer V and Schulzke T (2011) Developing a Burner System for Low Calorific Gases in Micro Gas Turbines: An Application for Small Scale Decentralized Heat and Power Generation International Gas Union Research

Emissions. Aerospace Science and Technology. 1: 557-572.

vanced Power and Energy Program, University of California, Irvine

Flame. Experimental Thermal and Fluid Science. In press

Emissions, Taylor & Francis.

162 Progress in Gas Turbine Performance

Technology Center.

Fluidized Bed Conference.

space Sci Meeting. AIAA.

Power. 114: 416-424.

Power 15: 288-303.

Conference.

Flow Fields in a Rql Gas Turbine Combustor.

120: 502-508.


[53] Arghode VK, Khalil AEE and Gupta AK (2012) Fuel Dilution and Liquid Fuel Opera‐ tional Effects on Ultra-High Thermal Intensity Distributed Combustor. Applied En‐ ergy. 95: 132-138.

**Chapter 7**

**Provisional chapter**

**Experimental Investigation of the Influence of**

**FLOX©-Based Micro Gas Turbine Combustor**

**Experimental Investigation of the Influence of**

**Combustor Cooling on the Characteristics of a**

**Based Micro Gas Turbine Combustor**

Jan Zanger, Monz Thomas and Aigner Manfred

efficiency which is recently at ≈ 30% in natural gas operation.

innovative combustion concepts are needed.

Additional information is available at the end of the chapter

Zanger Jan, Monz Thomas and Aigner Manfred

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54405

**1. Introduction**

**Combustor Cooling on the Characteristics of a FLOX©-**

Decentralised combined heat and power generation (CHP) offers a higly efficient and sustainable way for domestic and industrial energy supply. In contrast to electrical power generation by large scale power plants in the MW-range the waste heat of the electical power generation of Micro-CHP-Units can be used by the customer without extensive grid losses. Using the combined heat and power concept overall power plant efficiencies of up to 90 % for sub MW-range CHP-units are possible [1]. In recent CHP plants conventional piston gas engines are mostly used since these systems show a good electric efficiency paired with moderate investment costs. On the other hand cycles based on micro gas turbine (MGT) systems have the potential to play an important role in decentralised power generation. In small plants for distributed power generation the flexible application of different gaseous fuels (natural gas qualities, bio fuels and low calorific gases) is an important factor. Furthermore, the national standards for exhaust gas emission levels need to be met not only at the time of installation but also after years of operation. Here, compared to piston engines MGT systems have advantages regarding fuel flexibility, maintenance costs and exhaust gas emissions [2]. This gives the possibility to avoid the installation of a cost-intensive exhaust gas treatment. Due to higher exhaust gas temperatures MGTs are more suitable for the generation of process heat and cooling. Furthermore, MGTs can be operated in a wider range of fuel gas calorific value and they are less sensitive to the fuel gas composition. Beside the advantages MGT systems need to be optimised in terms of electric

In order to increase fuel flexibility, electrical efficiency, product life time and reliability of micro gas turbine systems while meeting today's and future exhaust gas emission levels, further development of the combustion systems needs to be done. To meet these tasks

Modern combustion systems for MGTs are mainly based on swirl-stabilised lean premixed concepts which promise low levels of exhaust gas pollutants. Here, the central recirculation

> ©2012 Jan et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Zanger et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,

© 2013 Zanger et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

distribution, and reproduction in any medium, provided the original work is properly cited.


**Provisional chapter**

#### **Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©- Based Micro Gas Turbine Combustor Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor**

Jan Zanger, Monz Thomas and Aigner Manfred Zanger Jan, Monz Thomas and Aigner Manfred

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54405

#### **1. Introduction**

[53] Arghode VK, Khalil AEE and Gupta AK (2012) Fuel Dilution and Liquid Fuel Opera‐ tional Effects on Ultra-High Thermal Intensity Distributed Combustor. Applied En‐

[54] Khalil AEE, Arghode VK, Gupta AK and Lee SC (2012) Low Calorific Value Fuelled Distributed Combustion with Swirl for Gas Turbine Applications. Applied Energy.

[55] Khalil AEE and Gupta AK (2011) Swirling Distributed Combustion for Clean Energy

[56] Greenberg SJ, McDougald NK and Arellano LO (2004) Full-Scale Demonstration of Surface-Stabilized Fuel Injectors for Sub-Three Ppm Nox Emissions. ASME Confer‐

[57] Greenberg SJ, McDougald NK, Weakley CK, Kendall RM and Arellano LO (2003) Surface-Stabilized Fuel Injectors with Sub-Three Ppm Nox Emissions for a 5.5 Mw Gas Turbine Engine. International Gas Turbine and Aeroengine Congress and Exhi‐

[58] Weakley CK, Greenberg SJ, Kendall RM, McDougald NK and Arellano LO (2002) De‐ velopment of Surface-Stabilized Fuel Injectors with Sub-Three Ppm Nox Emissions. International Joint Power Generation Conference. American Society of Mechanical

[59] Cabot G, Vauchelles D, Taupin B and Boukhalfa A (2004) Experimental Study of Lean Premixed Turbulent Combustion in a Scale Gas Turbine Chamber. Experimen‐

[60] Mcdougald NK (2005) Development and Demonstration of an Ultra Low Nox Com‐ bustor for Gas Turbines. USA DOE, Office of Energy Efficiency and Renewable Ener‐

[61] Arellano LO, Bhattacharya AK, Smith KO, Greenberg SJ and McDougald NK (2006) Development and Demonstration of Engine-Ready Surface-Stabilized Combustion

[62] Arellano L, Smith KO, California Energy Commission. Public Interest Energy R and Solar Turbines I (2008) Catalytic Combustor-Fired Industrial Gas Turbine Pier Final

[63] Clark H, Sullivan JD, California Energy Commission. Public Interest Energy R, Cali‐ fornia Energy Commission. Energy Innovations Small Grant P and Alzeta C (2001) Improved Operational Turndown of an Ultra-Low Emission Gas Turbine Combus‐

[64] Arvind G. Rao and Yeshayahou Levy, "A New Combustion Methodology for Low Emission Gas Turbine Engines", 8th HiTACG conference, July 5-8 2010, Poznan.

System. ASME Turbo Expo 2006: Power for Land, Sea, and Air.

tor, California Energy Commission, Sacramento, Calif.

Project Report, California Energy Commission, [Sacramento, Calif.].

Conversion in Gas Turbine Applications. Applied Energy. 88: 3685-3693.

ergy. 95: 132-138.

ence Proceedings. 2004: 393-401.

bition. American Society of Mechanical Engineers.

tal Thermal and Fluid Science. 28: 683-690.

gy, Washington, D.C; Oak Ridge, Tenn.

98: 69-78.

164 Progress in Gas Turbine Performance

Engineers.

Decentralised combined heat and power generation (CHP) offers a higly efficient and sustainable way for domestic and industrial energy supply. In contrast to electrical power generation by large scale power plants in the MW-range the waste heat of the electical power generation of Micro-CHP-Units can be used by the customer without extensive grid losses. Using the combined heat and power concept overall power plant efficiencies of up to 90 % for sub MW-range CHP-units are possible [1]. In recent CHP plants conventional piston gas engines are mostly used since these systems show a good electric efficiency paired with moderate investment costs. On the other hand cycles based on micro gas turbine (MGT) systems have the potential to play an important role in decentralised power generation. In small plants for distributed power generation the flexible application of different gaseous fuels (natural gas qualities, bio fuels and low calorific gases) is an important factor. Furthermore, the national standards for exhaust gas emission levels need to be met not only at the time of installation but also after years of operation. Here, compared to piston engines MGT systems have advantages regarding fuel flexibility, maintenance costs and exhaust gas emissions [2]. This gives the possibility to avoid the installation of a cost-intensive exhaust gas treatment. Due to higher exhaust gas temperatures MGTs are more suitable for the generation of process heat and cooling. Furthermore, MGTs can be operated in a wider range of fuel gas calorific value and they are less sensitive to the fuel gas composition. Beside the advantages MGT systems need to be optimised in terms of electric efficiency which is recently at ≈ 30% in natural gas operation.

In order to increase fuel flexibility, electrical efficiency, product life time and reliability of micro gas turbine systems while meeting today's and future exhaust gas emission levels, further development of the combustion systems needs to be done. To meet these tasks innovative combustion concepts are needed.

Modern combustion systems for MGTs are mainly based on swirl-stabilised lean premixed concepts which promise low levels of exhaust gas pollutants. Here, the central recirculation

©2012 Jan et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative

zone induced by the swirl of the income air flow serves to stabilise the flame of the main stage generating compact flames. However, these combustion systems tend to a liability to thermo-acoustic instabilities and coherent flow structures like PVC, especially at lean premixed conditions [3, 4] which can cause high amplitude pressure oscillations. This can lead to serious damage of components in the combustion system as well as the turbo engine itself. In addition, oscillating pressure and heat release zones as well as local flame extinction, caused by coherent flow structures can have a huge impact on the production of combustion emissions. Moreover, swirl-stabilised combustion is limited in using different fuel gas compositions regarding flame flashback incidences [5] and reliable operation range. In particular, fuel gas compositions with high hydrogen fractions limit the application of swirl-stabilised combustion concepts. Studies have shown the potential of the Flameless Oxidation (FLOX©)[6] based jet-stabilised combustion concept to achieve both low exhaust gas emission levels and reduced risk towards thermo-acoustic instabilities in combination with high fuel flexibility. As discussed by Hambdi et al. [7] the general idea of this concept is also known as MILD combustion [8], colourless distributed combustion [9] or high temperature air combustion (HiTAC) [10] to name a few. The main characteristic which all those similar concepts share is the use of high temperature process air and a high dilution of the fresh gas mixture by recirculated flue gases. In particular the FLOX©-concept is characterised by non-swirled technically premixed high impulse jets penetrating a combustion chamber in a circular arrangement. These jets drive a strong inner recirculation resulting in an effective mixing process of hot exhaust gases and the fresh incoming fuel/air mixture. This enhances the flame stabilisation but also reduces the chemical reaction rates by a strong dilution. Hence, the reaction zone is stretched over a larger volume compared to swirl-stabilised combustion concepts. This volumetric reaction region exhibits an almost homogeneously distributed temperature profile inside the combustion chamber close to the adiabatic flame temperature of the global equivalence ratio Φ promising low NOx emission levels [11]. Due to the high momentum jets and therefore, the absence of low velocity zones of the income air mixture, the combustion concept has a high resistance to flashback incidents even at highly premixed conditions [12] and high hydrogen fractions [13].

considerable amount of cool compressor air bypasses the combustion chamber and remixes with the hot exhaust gases at the combustion chamber exit. Since the FLOX© combustion at high air numbers exhibits volumetric reaction regions, the combustion requires more room compared to swirl-stabilised combustion concepts. The low production of harmful emissions as well as the high flame stability of the FLOX©-regime would be negatively influenced by the injection of cold dilution air into extensively expanded volumetric reaction zones. Therefore, combustion system design parameters like length of the combustion chamber and position, shape and pattern of dilution holes have to be considered for a final combustor development. In order to improve flame stability, lifetime, operating range and exhaust gas emissions of the combustion system in a commercial Turbec T100 MGT, a FLOX©-based jet-stabilised combustor was designed for natural gas utilising the advantages of this concept. The recent work covers an experimental study of two combustor configurations differing in the combustor front plate cooling. The paper presents the influence of the combustor cooling air on flame characteristics, lean blow off (LBO) limits and exhaust gas emissions. Flame characteristics are analysed, using measurements of the OH∗-chemiluminescence (OH∗-CL) signal at selected power loads and air numbers. From these images the height above burner, the dispersion of OH∗-CL signal and its homogeneity are derived. These quantaties are discussed with respect to the effects of cooling air, thermal power load and air number on

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

167

http://dx.doi.org/10.5772/54405

In Figure 1 and 2 both FLOX-based combustor configurations are shown which were used in this study. The reference combustor without any combustor front plate cooling is displayed in Figure 1. In order to increase the combustor lifetime an additional impingement front plate cooling is implemented into the second configuration to decrease the temperature at the location of the highest thermal loads. Both combustor designs consist of 20 fuel and air nozzles in a circular arrangement. Natural gas is injected concentrically into the air nozzles, which are oriented co-axially with respect to the combustion chamber. Both air and fuel gas flow are injected without any swirl. Partial premixing of fuel and air is achieved by the special design of the injecting system. Vortical structures generated by the air injection system provide a macroscopic mixing of fuel and air. Turbulence produced by the fuel injection system provides mixing on a microscopic scale. For the impingement cooling system a small part of the overall air mass flow is fed into a cooling air plenum and passes a perforated plate. This perforated plate produces small cooling air jets impinging the combustor front plate from the back side. After the cooling air has hit the combustor front plate it is led radially through channels and injected into the combustion chamber through 20 holes. The cooling air holes are positioned in between the FLOX© combustor air nozzles on a slightly larger diameter. In addition to the impingement cooling the front plate of the

In order to run parametric studies of the combustion performance independent from the MGT load point limitations, the combustor was implemented into an atmospheric test rig.

cooled design is coated with a zirconium oxide thermal protection layer.

the combustor performance.

**2. Experimental setup**

**2.1. Combustor design**

**2.2. Test rig description**

Typical temperature and velocity fields of a FLOX©-based combustor are exemplarily reported by Schütz et al. [14]. Combustion stability, limits of the flameless regime as well as a comparison between experimental and numerical results obtained by Large-Eddy Simulation is reported by Duwig et al. [15]. Lückerath et al. [12] compared OH-PLIF and OH\*-chemiluminescence images of a jet-stabilised burner for thermal powers up to 475kW at elevated pressure. Major species concentrations, velocity and temperature fields as well as reaction regions were reported by Lammel et al. [16] for a generic single nozzle setup using particle image velocimetry, laser raman spectroscopy as well as OH-PLIF measurements.

Beside the discussed advantages the implementation of a FLOX©-based combustion concept to a MGT system poses some challenges and tasks. First of all, the quality of the air/fuel premixture has a significant influence on the flame characteristics and emission levels. Therefore, the combustor has to be carefully designed to generate an optimised air/fuel exit profile. Moreover, since MGT systems need to have competitive prices compared to piston engines, the turbo components usually exhibit a most simple design. Therefore, the turbine blades are not cooled internally resulting in much lower turbine inlet temperature limits compared to industrial gas turbines. In order to adjust the turbine inlet temperature profile a considerable amount of cool compressor air bypasses the combustion chamber and remixes with the hot exhaust gases at the combustion chamber exit. Since the FLOX© combustion at high air numbers exhibits volumetric reaction regions, the combustion requires more room compared to swirl-stabilised combustion concepts. The low production of harmful emissions as well as the high flame stability of the FLOX©-regime would be negatively influenced by the injection of cold dilution air into extensively expanded volumetric reaction zones. Therefore, combustion system design parameters like length of the combustion chamber and position, shape and pattern of dilution holes have to be considered for a final combustor development.

In order to improve flame stability, lifetime, operating range and exhaust gas emissions of the combustion system in a commercial Turbec T100 MGT, a FLOX©-based jet-stabilised combustor was designed for natural gas utilising the advantages of this concept. The recent work covers an experimental study of two combustor configurations differing in the combustor front plate cooling. The paper presents the influence of the combustor cooling air on flame characteristics, lean blow off (LBO) limits and exhaust gas emissions. Flame characteristics are analysed, using measurements of the OH∗-chemiluminescence (OH∗-CL) signal at selected power loads and air numbers. From these images the height above burner, the dispersion of OH∗-CL signal and its homogeneity are derived. These quantaties are discussed with respect to the effects of cooling air, thermal power load and air number on the combustor performance.

#### **2. Experimental setup**

2 Gas Turbine

hydrogen fractions [13].

zone induced by the swirl of the income air flow serves to stabilise the flame of the main stage generating compact flames. However, these combustion systems tend to a liability to thermo-acoustic instabilities and coherent flow structures like PVC, especially at lean premixed conditions [3, 4] which can cause high amplitude pressure oscillations. This can lead to serious damage of components in the combustion system as well as the turbo engine itself. In addition, oscillating pressure and heat release zones as well as local flame extinction, caused by coherent flow structures can have a huge impact on the production of combustion emissions. Moreover, swirl-stabilised combustion is limited in using different fuel gas compositions regarding flame flashback incidences [5] and reliable operation range. In particular, fuel gas compositions with high hydrogen fractions limit the application of swirl-stabilised combustion concepts. Studies have shown the potential of the Flameless Oxidation (FLOX©)[6] based jet-stabilised combustion concept to achieve both low exhaust gas emission levels and reduced risk towards thermo-acoustic instabilities in combination with high fuel flexibility. As discussed by Hambdi et al. [7] the general idea of this concept is also known as MILD combustion [8], colourless distributed combustion [9] or high temperature air combustion (HiTAC) [10] to name a few. The main characteristic which all those similar concepts share is the use of high temperature process air and a high dilution of the fresh gas mixture by recirculated flue gases. In particular the FLOX©-concept is characterised by non-swirled technically premixed high impulse jets penetrating a combustion chamber in a circular arrangement. These jets drive a strong inner recirculation resulting in an effective mixing process of hot exhaust gases and the fresh incoming fuel/air mixture. This enhances the flame stabilisation but also reduces the chemical reaction rates by a strong dilution. Hence, the reaction zone is stretched over a larger volume compared to swirl-stabilised combustion concepts. This volumetric reaction region exhibits an almost homogeneously distributed temperature profile inside the combustion chamber close to the adiabatic flame temperature of the global equivalence ratio Φ promising low NOx emission levels [11]. Due to the high momentum jets and therefore, the absence of low velocity zones of the income air mixture, the combustion concept has a high resistance to flashback incidents even at highly premixed conditions [12] and high

Typical temperature and velocity fields of a FLOX©-based combustor are exemplarily reported by Schütz et al. [14]. Combustion stability, limits of the flameless regime as well as a comparison between experimental and numerical results obtained by Large-Eddy Simulation is reported by Duwig et al. [15]. Lückerath et al. [12] compared OH-PLIF and OH\*-chemiluminescence images of a jet-stabilised burner for thermal powers up to 475kW at elevated pressure. Major species concentrations, velocity and temperature fields as well as reaction regions were reported by Lammel et al. [16] for a generic single nozzle setup using particle image velocimetry, laser raman spectroscopy as well as OH-PLIF measurements. Beside the discussed advantages the implementation of a FLOX©-based combustion concept to a MGT system poses some challenges and tasks. First of all, the quality of the air/fuel premixture has a significant influence on the flame characteristics and emission levels. Therefore, the combustor has to be carefully designed to generate an optimised air/fuel exit profile. Moreover, since MGT systems need to have competitive prices compared to piston engines, the turbo components usually exhibit a most simple design. Therefore, the turbine blades are not cooled internally resulting in much lower turbine inlet temperature limits compared to industrial gas turbines. In order to adjust the turbine inlet temperature profile a

#### **2.1. Combustor design**

In Figure 1 and 2 both FLOX-based combustor configurations are shown which were used in this study. The reference combustor without any combustor front plate cooling is displayed in Figure 1. In order to increase the combustor lifetime an additional impingement front plate cooling is implemented into the second configuration to decrease the temperature at the location of the highest thermal loads. Both combustor designs consist of 20 fuel and air nozzles in a circular arrangement. Natural gas is injected concentrically into the air nozzles, which are oriented co-axially with respect to the combustion chamber. Both air and fuel gas flow are injected without any swirl. Partial premixing of fuel and air is achieved by the special design of the injecting system. Vortical structures generated by the air injection system provide a macroscopic mixing of fuel and air. Turbulence produced by the fuel injection system provides mixing on a microscopic scale. For the impingement cooling system a small part of the overall air mass flow is fed into a cooling air plenum and passes a perforated plate. This perforated plate produces small cooling air jets impinging the combustor front plate from the back side. After the cooling air has hit the combustor front plate it is led radially through channels and injected into the combustion chamber through 20 holes. The cooling air holes are positioned in between the FLOX© combustor air nozzles on a slightly larger diameter. In addition to the impingement cooling the front plate of the cooled design is coated with a zirconium oxide thermal protection layer.

#### **2.2. Test rig description**

In order to run parametric studies of the combustion performance independent from the MGT load point limitations, the combustor was implemented into an atmospheric test rig.

Air Nozzle Fuel Nozzle

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

169

http://dx.doi.org/10.5772/54405

Perforated Plate

Cooling Air Plenum

Process Air

Impingement Cooled

Cooling Air Exit Holes

**Figure 2.** Impingement Cooled Combustor.

**Figure 3.** Atmospheric Combustor Test Rig

Dc

Front Plate

**Figure 1.** Uncooled Combustor.

The experimental setup shown in Figure 3 comprises the air and fuel supply, the combustor and an optically accessible, hexagonal combustion chamber comprising six quartz glass windows. The hexagonal cross-section was chosen as a trade-off between good optical accessibility and the analogy to the circular cross-section of the original MGT liner. A circular fuel plenum which is situated under the air plenum but not shown in Figure 3 ensures an equal supply of all 20 fuel gas nozzles. In order to emulate the combustor inlet conditions of the MGT the air can be preheated electrically up to 925K by five 15kW "Leister" air heater units. The complete air supply system is decoupled acoustically from the test bench by a perforated plate located at the air inlet. When entering the air plenum the air flow is directed via a baffle in a way that a 180°deflection at the combustor inlet of the original MGT is reproduced. This was found to be essential to generate a specific premixing profile in the combustor nozzle and hence is important for flame characteristics and flame stabilisation. After passing the baffle a small part of the air flow enters the cooling plenum and the major part is fed into the air nozzles of the combustor, where the premixing with the fuel takes place. In this study the combustor is operated with natural gas (LHV = 47.01 *M J kg* , AFR*stoech* = 16.2). On the top of the combustion chamber an exhaust gas duct is flanged.

<sup>168</sup> Progress in Gas Turbine Performance Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor http://dx.doi.org/10.5772/54405 169

**Figure 2.** Impingement Cooled Combustor.

4 Gas Turbine

Air Nozzle Fuel Nozzle

Dc

**Figure 1.** Uncooled Combustor.

Process Air

The experimental setup shown in Figure 3 comprises the air and fuel supply, the combustor and an optically accessible, hexagonal combustion chamber comprising six quartz glass windows. The hexagonal cross-section was chosen as a trade-off between good optical accessibility and the analogy to the circular cross-section of the original MGT liner. A circular fuel plenum which is situated under the air plenum but not shown in Figure 3 ensures an equal supply of all 20 fuel gas nozzles. In order to emulate the combustor inlet conditions of the MGT the air can be preheated electrically up to 925K by five 15kW "Leister" air heater units. The complete air supply system is decoupled acoustically from the test bench by a perforated plate located at the air inlet. When entering the air plenum the air flow is directed via a baffle in a way that a 180°deflection at the combustor inlet of the original MGT is reproduced. This was found to be essential to generate a specific premixing profile in the combustor nozzle and hence is important for flame characteristics and flame stabilisation. After passing the baffle a small part of the air flow enters the cooling plenum and the major part is fed into the air nozzles of the combustor, where the premixing with the fuel takes

place. In this study the combustor is operated with natural gas (LHV = 47.01 *M J*

= 16.2). On the top of the combustion chamber an exhaust gas duct is flanged.

*kg* , AFR*stoech*

**Figure 3.** Atmospheric Combustor Test Rig

#### **2.3. Instrumentation**

To analyse all relevant process parameters inside the system the test rig is equipped with a detailed instrumentation. The data acquisition at a frequency of 2 Hz is realised by "Delphin" modules. For temperature measurements a total number of 27 thermocouples (type N, precision class 2) are installed. The arithmetic average of the temperature *T*<sup>1</sup> and *T*<sup>2</sup> which are situated in the air flow at the combustor inlet defines the combustor preheat temperature *TV* with an uncertainty of ±0.85% of the actual value. The rig furthermore comprises 2 total and 13 static pressure transducers read out by pressure scanners "Netscanner Model 9116" and "Model 9032" by Esterline Pressure Systems. All pressures can be optained with a manufacturer's accuracy of ±4 mbar. Combustor pressure loss is determined by the static pressure *p*<sup>1</sup> measured short before the combustor inlet (see figure 3) and the ambient pressure. Applying a suitable calibration, the mass flow through the cooling system is calculated as a function of the pressure loss between the static pressures *pcool*,01 and *pcool*,02 situated in front of and behind the perforated plate. The fuel mass flow is controlled by a "Bronkhorst Cori-Flow" coriolis mass flow controller with a manufacturer's accuracy of ±0.5% of the actual value and the air mass flow is regulated by a "Bronkhorst EL-Flow" thermal mass flow controller with a manufacturer's accuracy of ±0.8%. As indicated in figure 3 a radially traversable suck-up exhaust gas probe is mounted inside the exhaust gas duct. The probe is equipped with a coaxial air cooling keeping the probe tip at a constant temperature of 120°C to achieve a sufficient quenching of the measured exhaust gas. This ensures defined measuring conditions. The sucked-up exhaust gases are directed via heated hoses to an "ABB" exhaust gas analysing system. The flue gas species O2, CO, CO2, NO, NO2 and unburned hydrocarbons (UHC) are measured by a magnetomechanical analyser "Magnos106", a infrared analyser "Uras14", a UV photometer "Limas11 HW" and a flame ionisation detector "MultiFID14". The species O2, CO and CO2 are measured in a dry environment, whereas all other species are detected in wet conditions. The measurements of the species shown in this study have manufacturer's accuracies as indicated in Table 1.

As indicated in Figure 4 the detection volume covers four air nozzles on each side of the combustor. Due to assembly restrictions the nozzles on opposing sides are arranged with a

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

The measurements shown in this section were carried out at steady-state combustion conditions, whereupon every single load point is time-averaged over 5 min at an acquisition rate of 2Hz. In order to analyse the operating range of the combustor configurations the overall air number *λoverall* and the thermal power were varied at a constant combustor inlet air preheat temperature *Tv*. In this study the normalised thermal power *Qth*,*<sup>n</sup>* with respect to MGT full load conditions is presented. The overall air number is defined as the reciprocal of the fuel equivalence ratio Φ calculated with the overall air and fuel mass flows. All mass flows of the MGT load points were scaled to atmospheric conditions keeping the velocity

> **Parameter Range Unit** *Qth*,*<sup>n</sup>* 35 - 100 % *λoverall* 1.8 - LBO - *Tv* 580 °C

Figure 5 visualises the operating range of the uncooled combustor configuration displaying the normalised axial jet velocity *unozzle* at the exit of a single air nozzle as a function of *Qth*,*n*.

*pc*,*in* · *Anozzle*

with the combustor inlet temperature *Tc*,*in*, the static combustor inlet pressure *pc*,*in*, the planar-averaged specific gas constant of the air/fuel mixture *Rmix*, the nozzle cross-section

/*unozzle*,*max* (1)

Turbine Combustor

171

http://dx.doi.org/10.5772/54405

*unozzle* <sup>=</sup> *<sup>m</sup>*˙ *nozzle* · *Rmix* · *Tc*,*in*

fields constant. Table 2 shows the matrix of the parametric study.

small misalignment with respect to the line of sight.

**Figure 4.** Line of Sight of the OH\*-CL measurements

**3. Experimental results**

**Table 2.** Ranges of the Measuring Matrix.

**3.1. Operating range**

*Unozzle* is defined as


**Table 1.** Ranges and Corresponding Accuracies of the Measured Exhaust Gas Species.

OH\* chemiluminescence measurements were used to study the shape, location and homogeneity of the heat release zone. The electronically excited OH\* radical is formed by chemical reactions in the reaction zone, predominately via *CH* <sup>+</sup> *<sup>O</sup>*<sup>2</sup> <sup>→</sup> *CO* <sup>+</sup> *OH*<sup>∗</sup> [17]. Since its lifetime is very short, the emitted OH\*-CL signal originates only from within the reaction region. Therefore, the OH\*-CL signal is a very good marker for the location and dimension of the reaction zone. However, this technique is a line-of-sight method giving only spatially integrated information in the combustion chamber depth. The OH\*-CL emissions were imaged using a "LaVision FlameStar 2" intensified CCD camera in combination with a "Halle" 64mm, f/2 UV lens and a UV interference filter (*λ* = 312 ± 20nm). All OH\*-CL data was time-averaged over a time series of 200 instantaneous images acquired with a repetition rate of 3.6Hz. Due to the substantial difference of the OH\*-CL signal over the complete load point range the gate width was varied between 18 and 600 *µs* at maximum gain factor. As indicated in Figure 4 the detection volume covers four air nozzles on each side of the combustor. Due to assembly restrictions the nozzles on opposing sides are arranged with a small misalignment with respect to the line of sight.

**Figure 4.** Line of Sight of the OH\*-CL measurements

#### **3. Experimental results**

6 Gas Turbine

**2.3. Instrumentation**

To analyse all relevant process parameters inside the system the test rig is equipped with a detailed instrumentation. The data acquisition at a frequency of 2 Hz is realised by "Delphin" modules. For temperature measurements a total number of 27 thermocouples (type N, precision class 2) are installed. The arithmetic average of the temperature *T*<sup>1</sup> and *T*<sup>2</sup> which are situated in the air flow at the combustor inlet defines the combustor preheat temperature *TV* with an uncertainty of ±0.85% of the actual value. The rig furthermore comprises 2 total and 13 static pressure transducers read out by pressure scanners "Netscanner Model 9116" and "Model 9032" by Esterline Pressure Systems. All pressures can be optained with a manufacturer's accuracy of ±4 mbar. Combustor pressure loss is determined by the static pressure *p*<sup>1</sup> measured short before the combustor inlet (see figure 3) and the ambient pressure. Applying a suitable calibration, the mass flow through the cooling system is calculated as a function of the pressure loss between the static pressures *pcool*,01 and *pcool*,02 situated in front of and behind the perforated plate. The fuel mass flow is controlled by a "Bronkhorst Cori-Flow" coriolis mass flow controller with a manufacturer's accuracy of ±0.5% of the actual value and the air mass flow is regulated by a "Bronkhorst EL-Flow" thermal mass flow controller with a manufacturer's accuracy of ±0.8%. As indicated in figure 3 a radially traversable suck-up exhaust gas probe is mounted inside the exhaust gas duct. The probe is equipped with a coaxial air cooling keeping the probe tip at a constant temperature of 120°C to achieve a sufficient quenching of the measured exhaust gas. This ensures defined measuring conditions. The sucked-up exhaust gases are directed via heated hoses to an "ABB" exhaust gas analysing system. The flue gas species O2, CO, CO2, NO, NO2 and unburned hydrocarbons (UHC) are measured by a magnetomechanical analyser "Magnos106", a infrared analyser "Uras14", a UV photometer "Limas11 HW" and a flame ionisation detector "MultiFID14". The species O2, CO and CO2 are measured in a dry environment, whereas all other species are detected in wet conditions. The measurements of the species shown in this study have manufacturer's accuracies as indicated in Table 1.

> CO NO<sup>x</sup> UHC O<sup>2</sup> **[ppm] [ppm] [ppm] [Vol-%]**

Range 1 0-8 0-24 0-9 0-25 Accuracy 1 0.1 0.5 0.1 0.25 Range 2 8-80 24-238 9-90 Accuracy 2 1 5 1

OH\* chemiluminescence measurements were used to study the shape, location and homogeneity of the heat release zone. The electronically excited OH\* radical is formed by chemical reactions in the reaction zone, predominately via *CH* <sup>+</sup> *<sup>O</sup>*<sup>2</sup> <sup>→</sup> *CO* <sup>+</sup> *OH*<sup>∗</sup> [17]. Since its lifetime is very short, the emitted OH\*-CL signal originates only from within the reaction region. Therefore, the OH\*-CL signal is a very good marker for the location and dimension of the reaction zone. However, this technique is a line-of-sight method giving only spatially integrated information in the combustion chamber depth. The OH\*-CL emissions were imaged using a "LaVision FlameStar 2" intensified CCD camera in combination with a "Halle" 64mm, f/2 UV lens and a UV interference filter (*λ* = 312 ± 20nm). All OH\*-CL data was time-averaged over a time series of 200 instantaneous images acquired with a repetition rate of 3.6Hz. Due to the substantial difference of the OH\*-CL signal over the complete load point range the gate width was varied between 18 and 600 *µs* at maximum gain factor.

**Table 1.** Ranges and Corresponding Accuracies of the Measured Exhaust Gas Species.

The measurements shown in this section were carried out at steady-state combustion conditions, whereupon every single load point is time-averaged over 5 min at an acquisition rate of 2Hz. In order to analyse the operating range of the combustor configurations the overall air number *λoverall* and the thermal power were varied at a constant combustor inlet air preheat temperature *Tv*. In this study the normalised thermal power *Qth*,*<sup>n</sup>* with respect to MGT full load conditions is presented. The overall air number is defined as the reciprocal of the fuel equivalence ratio Φ calculated with the overall air and fuel mass flows. All mass flows of the MGT load points were scaled to atmospheric conditions keeping the velocity fields constant. Table 2 shows the matrix of the parametric study.


**Table 2.** Ranges of the Measuring Matrix.

#### **3.1. Operating range**

Figure 5 visualises the operating range of the uncooled combustor configuration displaying the normalised axial jet velocity *unozzle* at the exit of a single air nozzle as a function of *Qth*,*n*. *Unozzle* is defined as

$$
\mu\_{nozzle} = \frac{\textit{\dot{m}\_{nozzle}} \cdot \textit{R}\_{\textit{mix}} \cdot T\_{c,in}}{p\_{c,in} \cdot A\_{nozzle}} / \mu\_{nozzle,\text{max}}\tag{1}
$$

with the combustor inlet temperature *Tc*,*in*, the static combustor inlet pressure *pc*,*in*, the planar-averaged specific gas constant of the air/fuel mixture *Rmix*, the nozzle cross-section area *Anozzle* and the averaged mass flow of a single nozzle *m*˙ *nozzle*. The whole measuring field of the operating range was scanned by adjusting a constant *Qth*,*<sup>n</sup>* and increasing the *<sup>λ</sup>overall* up to lean blow-off (LBO) conditions. In this study the lean blow-off is defined as the point where the flame actually extinguishes. The filled points symbolise the feasible operating points whereas the blank points emblematise the points at which LBO occured. All illustrated load points represent an individual measurement which means that no information about the reproducibility of the LBO limit can be given. Regarding the points around LBO it is visible that the feasible *<sup>λ</sup>overall* slopes with increasing thermal power. The LBO at *Qth*,*<sup>n</sup>* = 35% is with *<sup>λ</sup>overall* = 3.26 substantially higher compared to *Qth*,*<sup>n</sup>* = 100% with *<sup>λ</sup>overall* = 3.01. Moreover, figure 5 indicates that higher jet velocities can be realised with increasing *Qth*,*<sup>n</sup>* before LBO occurs. The axial jet velocity *unozzle* at LBO significantly varries from 58% at *Qth*,*<sup>n</sup>* = 35% to approximately 140% at *Qth*,*<sup>n</sup>* = 100%. This effect was also observed and discussed by Vaz et al. [18] for a similar system.

In Figure 6 the comparison between the LBO conditions of the uncooled combustor and the cooled configuration is presented as a function of *Qth*,*n*. Due to mass flow limitations the LBO of the cooled design could only be measured for *Qth*/*Qth*,*max* <sup>≤</sup> 85%, therefore, only this range is visualised. For the cooled design the overall air number *λoverall* of the combustor is shown as well as the local air number *λnozzle* of a single nozzle which is reduced by the cooling air mass flow. However, for the uncooled design the local and the overall air numbers are equal due to the lack of cooling air. The *λoverall* at LBO of the cooled configuration follows the general sloping trend of the uncooled design for increasing *Qth*,*<sup>n</sup>* but exhibits a steeper gradient and with *<sup>λ</sup>overall* = 3.44 at *Qth*,*<sup>n</sup>* = 35% higher values at low thermal loads. This means using a cooled combustor configuration that for *Qth*/*Qth*,*max* <sup>≤</sup> 52% more overall air can be fed into the combustion chamber before LBO occurs. This behaviour is advantageous if the amount of combustion air is to be maximised. However, for *Qth*/*Qth*,*max* <sup>≥</sup> 70% the difference between both designs is negligible. Furthermore, the graph of the cooled design exhibits a kink at *Qth*,*<sup>n</sup>* = 70%. This behaviour is reproducible but in order to give a conclusive explanation further investigation need to be carried out. Regarding *λnozzle* the cooled configuration exhibits a much lower LBO limit for the whole operating range, whereas the difference between the designs increases with rising thermal power. This indicates that the fraction of cooling air, which interacts with the combustion

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

173

http://dx.doi.org/10.5772/54405

The carbon monoxide (CO) exhaust gas emission profiles of the uncooled combustor configuration are shown in figure 7 as a function of *λoverall* for all thermal powers, whereas CO is normalised to 15% oxygen concentration. All curves exhibit a similar U-shape with a distinct minimum. Regarding the graph at *Qth*,*<sup>n</sup>* = 100% the CO concentrations decrease from 50 ppm at *λoverall* = 1.8 to 24 ppm at *λoverall* = 2.42 followed by a rise up to 58 ppm at *λoverall* = 2.8. In the left hand branch of the CO curves the trend of the measured concentrations follow the trend of the equilibrium conditions which decrease for higher air numbers [19]. The right hand branch, however, is dominated by non-equilibrium effects [20]. Here, the residence time of the flue gases inside the combustion chamber before reaching the exhaust gas probe is insufficiently long to achieve equilibrium state resulting in higher measured CO concentrations. Moreover, due to the expansion of the reaction zones at LBO-near overall air numbers, decribed in section 3.3, the exhaust gas probe is for these conditions located inside the reaction zone resulting in incomplete combustion process at the position of measurement. Regarding the CO profiles at different thermal powers a destinct layered arrangement is observable showing higher CO concentrations for increasing *Qth*,*n*. Here again, the influence of the residence time is visible since at constant *λoverall* higher thermal powers exhibit higher

In figure 8 the comparison between the CO profiles of the uncooled combustor and the cooled configuration is shown as a function of *<sup>λ</sup>nozzle*. For clarity reasons only the profiles for *Qth*,*<sup>n</sup>* = 53% and 100% are visualised exemplarily but it should be mentioned that the trend of the displayed curves applies to the other thermal powers as well. The CO profiles of both combustor designs show similar U-shaped trends. But it is clearly visible that the profiles of the cooled configuration are shifted horizontally to lower air numbers. The magnitude of the shift is thereby dependent on thermal power. For *Qth*,*<sup>n</sup>* = 100% the shift seems to be approximately constant for all *<sup>λ</sup>nozzle*, however, for *Qth*,*<sup>n</sup>* = 53% the CO profiles fit well at

process, changes with *Qth*,*n*.

**3.2. Exhaust gas emissions**

fuel and air mass flows reducing the overall residence time.

**Figure 5.** Operating Range of Uncooled Combustor Configuration.

**Figure 6.** Air Number at LBO of the Cooled and Uncooled Combustor Configurations.

In Figure 6 the comparison between the LBO conditions of the uncooled combustor and the cooled configuration is presented as a function of *Qth*,*n*. Due to mass flow limitations the LBO of the cooled design could only be measured for *Qth*/*Qth*,*max* <sup>≤</sup> 85%, therefore, only this range is visualised. For the cooled design the overall air number *λoverall* of the combustor is shown as well as the local air number *λnozzle* of a single nozzle which is reduced by the cooling air mass flow. However, for the uncooled design the local and the overall air numbers are equal due to the lack of cooling air. The *λoverall* at LBO of the cooled configuration follows the general sloping trend of the uncooled design for increasing *Qth*,*<sup>n</sup>* but exhibits a steeper gradient and with *<sup>λ</sup>overall* = 3.44 at *Qth*,*<sup>n</sup>* = 35% higher values at low thermal loads. This means using a cooled combustor configuration that for *Qth*/*Qth*,*max* <sup>≤</sup> 52% more overall air can be fed into the combustion chamber before LBO occurs. This behaviour is advantageous if the amount of combustion air is to be maximised. However, for *Qth*/*Qth*,*max* <sup>≥</sup> 70% the difference between both designs is negligible. Furthermore, the graph of the cooled design exhibits a kink at *Qth*,*<sup>n</sup>* = 70%. This behaviour is reproducible but in order to give a conclusive explanation further investigation need to be carried out. Regarding *λnozzle* the cooled configuration exhibits a much lower LBO limit for the whole operating range, whereas the difference between the designs increases with rising thermal power. This indicates that the fraction of cooling air, which interacts with the combustion process, changes with *Qth*,*n*.

#### **3.2. Exhaust gas emissions**

8 Gas Turbine

al. [18] for a similar system.

**Figure 5.** Operating Range of Uncooled Combustor Configuration.

**Figure 6.** Air Number at LBO of the Cooled and Uncooled Combustor Configurations.

area *Anozzle* and the averaged mass flow of a single nozzle *m*˙ *nozzle*. The whole measuring field of the operating range was scanned by adjusting a constant *Qth*,*<sup>n</sup>* and increasing the *<sup>λ</sup>overall* up to lean blow-off (LBO) conditions. In this study the lean blow-off is defined as the point where the flame actually extinguishes. The filled points symbolise the feasible operating points whereas the blank points emblematise the points at which LBO occured. All illustrated load points represent an individual measurement which means that no information about the reproducibility of the LBO limit can be given. Regarding the points around LBO it is visible that the feasible *<sup>λ</sup>overall* slopes with increasing thermal power. The LBO at *Qth*,*<sup>n</sup>* = 35% is with *<sup>λ</sup>overall* = 3.26 substantially higher compared to *Qth*,*<sup>n</sup>* = 100% with *<sup>λ</sup>overall* = 3.01. Moreover, figure 5 indicates that higher jet velocities can be realised with increasing *Qth*,*<sup>n</sup>* before LBO occurs. The axial jet velocity *unozzle* at LBO significantly varries from 58% at *Qth*,*<sup>n</sup>* = 35% to approximately 140% at *Qth*,*<sup>n</sup>* = 100%. This effect was also observed and discussed by Vaz et

> The carbon monoxide (CO) exhaust gas emission profiles of the uncooled combustor configuration are shown in figure 7 as a function of *λoverall* for all thermal powers, whereas CO is normalised to 15% oxygen concentration. All curves exhibit a similar U-shape with a distinct minimum. Regarding the graph at *Qth*,*<sup>n</sup>* = 100% the CO concentrations decrease from 50 ppm at *λoverall* = 1.8 to 24 ppm at *λoverall* = 2.42 followed by a rise up to 58 ppm at *λoverall* = 2.8. In the left hand branch of the CO curves the trend of the measured concentrations follow the trend of the equilibrium conditions which decrease for higher air numbers [19]. The right hand branch, however, is dominated by non-equilibrium effects [20]. Here, the residence time of the flue gases inside the combustion chamber before reaching the exhaust gas probe is insufficiently long to achieve equilibrium state resulting in higher measured CO concentrations. Moreover, due to the expansion of the reaction zones at LBO-near overall air numbers, decribed in section 3.3, the exhaust gas probe is for these conditions located inside the reaction zone resulting in incomplete combustion process at the position of measurement. Regarding the CO profiles at different thermal powers a destinct layered arrangement is observable showing higher CO concentrations for increasing *Qth*,*n*. Here again, the influence of the residence time is visible since at constant *λoverall* higher thermal powers exhibit higher fuel and air mass flows reducing the overall residence time.

> In figure 8 the comparison between the CO profiles of the uncooled combustor and the cooled configuration is shown as a function of *<sup>λ</sup>nozzle*. For clarity reasons only the profiles for *Qth*,*<sup>n</sup>* = 53% and 100% are visualised exemplarily but it should be mentioned that the trend of the displayed curves applies to the other thermal powers as well. The CO profiles of both combustor designs show similar U-shaped trends. But it is clearly visible that the profiles of the cooled configuration are shifted horizontally to lower air numbers. The magnitude of the shift is thereby dependent on thermal power. For *Qth*,*<sup>n</sup>* = 100% the shift seems to be approximately constant for all *<sup>λ</sup>nozzle*, however, for *Qth*,*<sup>n</sup>* = 53% the CO profiles fit well at

**Figure 7.** CO Emissions of the Uncooled Combustor Configuration.

low air numbers but differ at high *λnozzle*. Due to a high sensitivity to the cooling air mass flow's error the maximum uncertainty of *λnozzle* after propagation of error is approximately ±5% for the cooled configuration. However, the uncooled design exhibits an uncertainty better than 1% due to the lack of cooling air. Nevertheless, since the observed shift of the CO profiles shows a distinct systematic trend, it is believed to represent a physical effect.

**Figure 9.** CO Emissions of the Cooled and Uncooled Combustor Designs as a Function of *λoverall* .

combustor designs.

*λnozzle*(mod).

In order to approximate the amount of cooling air which interacts with the reaction region, the *λnozzle* of the cooled configuration is modified in a way that the CO profiles of both combustor designs fit well as shown in figure 10. This new air number is called *λnozzle*(mod). For the cooled design it is approximately *λnozzle*(mod) ≈ *λnozzle* + 0.1 �= *const*. and for the uncooled configuration it equals *λnozzle*. This modified air number, which is based on the CO emissions, is an auxiliary quantity for comparing both combustor configurations. Since the visualisation as a function of *λnozzle*(mod) neutralises the shift of air number between both designs, the comparison of the magnitudes of certain quantities are facilitated. Therefore, the modified air number is used in the following sections to characterise the difference of both

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

175

http://dx.doi.org/10.5772/54405

**Figure 10.** CO Emissions of the Cooled and Uncooled Combustor Configurations as a Function of the Modified Air Number

The nitrogen oxide (NOx) emissions at 15% *O*<sup>2</sup> of both combustor designs are visualised in figure 11 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 100%. All NOx profiles show a similar exponential decreasing characteristics and similar magnitudes for rising modified

When comparing the CO emissions of both combustor designs with respect to *λoverall*, see figure 9, the difference between both configurations is even more pronounced but shifted to the opposite direction compared to the display as a function of *λnozzle*. In the case that the CO profiles of the cooled configuration were equal to the uncooled design for *λoverall* this would suggest that all cooling air participates in the reaction process. On the other hand equal CO profiles for *λnozzle* would mean that no cooling air enters the reaction region. Therefore, the observed behaviour indicates that only a part of the cooling air participates in the primary reaction zone.

**Figure 8.** CO Emissions of the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*.

<sup>174</sup> Progress in Gas Turbine Performance Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor http://dx.doi.org/10.5772/54405 175

**Figure 9.** CO Emissions of the Cooled and Uncooled Combustor Designs as a Function of *λoverall* .

10 Gas Turbine

reaction zone.

**Figure 7.** CO Emissions of the Uncooled Combustor Configuration.

low air numbers but differ at high *λnozzle*. Due to a high sensitivity to the cooling air mass flow's error the maximum uncertainty of *λnozzle* after propagation of error is approximately ±5% for the cooled configuration. However, the uncooled design exhibits an uncertainty better than 1% due to the lack of cooling air. Nevertheless, since the observed shift of the CO profiles shows a distinct systematic trend, it is believed to represent a physical effect. When comparing the CO emissions of both combustor designs with respect to *λoverall*, see figure 9, the difference between both configurations is even more pronounced but shifted to the opposite direction compared to the display as a function of *λnozzle*. In the case that the CO profiles of the cooled configuration were equal to the uncooled design for *λoverall* this would suggest that all cooling air participates in the reaction process. On the other hand equal CO profiles for *λnozzle* would mean that no cooling air enters the reaction region. Therefore, the observed behaviour indicates that only a part of the cooling air participates in the primary

**Figure 8.** CO Emissions of the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*.

In order to approximate the amount of cooling air which interacts with the reaction region, the *λnozzle* of the cooled configuration is modified in a way that the CO profiles of both combustor designs fit well as shown in figure 10. This new air number is called *λnozzle*(mod). For the cooled design it is approximately *λnozzle*(mod) ≈ *λnozzle* + 0.1 �= *const*. and for the uncooled configuration it equals *λnozzle*. This modified air number, which is based on the CO emissions, is an auxiliary quantity for comparing both combustor configurations. Since the visualisation as a function of *λnozzle*(mod) neutralises the shift of air number between both designs, the comparison of the magnitudes of certain quantities are facilitated. Therefore, the modified air number is used in the following sections to characterise the difference of both combustor designs.

**Figure 10.** CO Emissions of the Cooled and Uncooled Combustor Configurations as a Function of the Modified Air Number *λnozzle*(mod).

The nitrogen oxide (NOx) emissions at 15% *O*<sup>2</sup> of both combustor designs are visualised in figure 11 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 100%. All NOx profiles show a similar exponential decreasing characteristics and similar magnitudes for rising modified nozzle air numbers. For *Qth*,*<sup>n</sup>* = 100% the NOx emissions reduce from 18ppm at *<sup>λ</sup>nozzle*(mod) = 1.75 down to 2 ppm at *λnozzle*(mod) = 2.8 for both combustors. Since the measurements were conducted at lean atmospheric conditions using natural gas, the major NOx formation mechanism is the thermal dominated Zeldovich mechanism. Therefore, the trend of the NOx profiles reflects the exponential decrease of the thermal NOx formation with falling flame temperature and rising air number, respectively. With respect to the profiles at different thermal powers, a very low dependence on *Qth*,*<sup>n</sup>* can be observed. For the uncooled design the curves of all thermal powers match very well, however, for the cooled configuration the NOx profiles at *Qth*,*n*< 100% exhibit slightly higher magnitudes at low *<sup>λ</sup>nozzle*(mod) compared to full load conditions. Regarding the fact that the NOx emissions of both combustor designs fit very well as a function of *λnozzle*(mod), the significance of this modified nozzle air number for comparing the configurations is backed up. This affirms the assumption that only a part of the cooling air interacts with the reaction region.

are characterised by compact shape and distinctly separated flames. For air numbers between 2.2 and 2.4 the flame length stays approximately constant, whereas the height above burner (HAB) increases with rising *λoverall*. Furthermore, for 2.2 ≤ *λoverall* ≤ 2.8 the reaction zones continuously merge into each other in horizontal direction evolving from separated flames to a single reaction region. For air numbers above 2.6 the reaction zone spreads in all spatial directions. Simultaneously, the HAB declines whereas the length of the reaction zone grows substantially. At *λoverall* = 3.1, which is the last operating point before LBO occured, the reaction zone is distributed over almost the whole combustion chamber volume. Moreover, regarding the scaling factors of the images the signal intensity decreases significantly with rising air numbers leading to a blueish visible flame of very low luminosity for *λoverall* ≥ 2.8.

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

177

http://dx.doi.org/10.5772/54405

**Figure 12.** Time-averaged OH\*-CL Images of the Uncooled Combustor at *Qth*,*<sup>n</sup>* = 69% as a Function of *λoverall* .

reaction regions.

In order to quantify and compare the phenomena observed in the time-averaged OH\*-CL images, characteristic parameters are derived from the data. For the subsequent analysis only signals above 50% of the image maximum intensity are taken into account. This definition has been found as the most appropriate method to cover the image pixels corresponding visually to the flame. In the following these pixels are called reaction or flame region. With this definition the HAB equals the axial distance between the combustor front plate and the horizontally averaged lower flame boundary. The Dispersion of the OH\*-signal is defined as the area of the flame region divided by the overall area of OH\*-measurement. Therefore, the dispersion is a marker for the relative reaction volume, but not for its homogeneity. The horizontal distribution of the reaction regions is evaluated by the Relative Horizontal Inhomogeneity *dIFlame*/*dx*. This parameter is calculated by vertically averaging the horizontal spatial intensity gradients inside the reaction regions and normalising it to the flame average intensity of each image. This definition ensures that *dIFlame*/*dx* is comparable for different signal intensity levels and flame shapes. The Relative Horizontal Inhomogeneity is a marker for the discreetness of the flames in horizontal direction declining for merging

Figure 13 visualises the Dispersion of OH\*-signal of the uncooled combustor design for all thermal powers as a function of *<sup>λ</sup>overall*. Regarding *Qth*,*<sup>n</sup>* = 100% the dispersion stays approximately constant at 25% between *λoverall* = 1.8 and 2.15. For *λoverall* ≥ 2.4 the magnitude increases significantly up to 73%. This means that with rising Dispersion of OH\*-signal the occupied volume of the reaction region inside the combustion chamber increases substantially until the flame is distributed over almost the whole volume. This

The emission levels of unburned hydrocarbons (UHC), not shown in this section, are under the detection threshold for both combustors and all thermal powers apart from air numbers close to the LBO. Here, a rapid increase of UHCs is detectable due to incomplete combustion close to the blow-off limit.

**Figure 11.** NOx Emissions of the Cooled and Uncooled Combustor Configurations as a Function of the Modified Air Number *λnozzle*(mod).

#### **3.3. Flame shape and location**

The flame characteristics regarding shape, location and homogeneity are discussed for both combustor designs in this section using time-averaged OH\*-CL images. Figure 12 shows a series of OH\*-CL images for the uncooled combustor configuration at *Qth*,*<sup>n</sup>* = 69% as a function of *λoverall*. All images are scaled between zero and their maximum signal intensity. The corresponding scaling factors are indicated in the upper left corner. The points of origin of abscissa and ordinate mark the centre axis of the combustor and the combustor front plate, respectively. The dimensions are normalised with respect to the combustor diameter. Moreover, the azimuthal positions of the air nozzles located at the window in front of the camera are indicated at the bottom. Regarding the images at low overall air numbers discrete reaction zones around the entering fresh gas jets can be observed. Here, the reaction zones 12 Gas Turbine

nozzle air numbers. For *Qth*,*<sup>n</sup>* = 100% the NOx emissions reduce from 18ppm at *<sup>λ</sup>nozzle*(mod) = 1.75 down to 2 ppm at *λnozzle*(mod) = 2.8 for both combustors. Since the measurements were conducted at lean atmospheric conditions using natural gas, the major NOx formation mechanism is the thermal dominated Zeldovich mechanism. Therefore, the trend of the NOx profiles reflects the exponential decrease of the thermal NOx formation with falling flame temperature and rising air number, respectively. With respect to the profiles at different thermal powers, a very low dependence on *Qth*,*<sup>n</sup>* can be observed. For the uncooled design the curves of all thermal powers match very well, however, for the cooled configuration the NOx profiles at *Qth*,*n*< 100% exhibit slightly higher magnitudes at low *<sup>λ</sup>nozzle*(mod) compared to full load conditions. Regarding the fact that the NOx emissions of both combustor designs fit very well as a function of *λnozzle*(mod), the significance of this modified nozzle air number for comparing the configurations is backed up. This affirms the assumption that only a part

The emission levels of unburned hydrocarbons (UHC), not shown in this section, are under the detection threshold for both combustors and all thermal powers apart from air numbers close to the LBO. Here, a rapid increase of UHCs is detectable due to incomplete combustion

**Figure 11.** NOx Emissions of the Cooled and Uncooled Combustor Configurations as a Function of the Modified Air Number

The flame characteristics regarding shape, location and homogeneity are discussed for both combustor designs in this section using time-averaged OH\*-CL images. Figure 12 shows a series of OH\*-CL images for the uncooled combustor configuration at *Qth*,*<sup>n</sup>* = 69% as a function of *λoverall*. All images are scaled between zero and their maximum signal intensity. The corresponding scaling factors are indicated in the upper left corner. The points of origin of abscissa and ordinate mark the centre axis of the combustor and the combustor front plate, respectively. The dimensions are normalised with respect to the combustor diameter. Moreover, the azimuthal positions of the air nozzles located at the window in front of the camera are indicated at the bottom. Regarding the images at low overall air numbers discrete reaction zones around the entering fresh gas jets can be observed. Here, the reaction zones

of the cooling air interacts with the reaction region.

close to the blow-off limit.

*λnozzle*(mod).

**3.3. Flame shape and location**

are characterised by compact shape and distinctly separated flames. For air numbers between 2.2 and 2.4 the flame length stays approximately constant, whereas the height above burner (HAB) increases with rising *λoverall*. Furthermore, for 2.2 ≤ *λoverall* ≤ 2.8 the reaction zones continuously merge into each other in horizontal direction evolving from separated flames to a single reaction region. For air numbers above 2.6 the reaction zone spreads in all spatial directions. Simultaneously, the HAB declines whereas the length of the reaction zone grows substantially. At *λoverall* = 3.1, which is the last operating point before LBO occured, the reaction zone is distributed over almost the whole combustion chamber volume. Moreover, regarding the scaling factors of the images the signal intensity decreases significantly with rising air numbers leading to a blueish visible flame of very low luminosity for *λoverall* ≥ 2.8.

**Figure 12.** Time-averaged OH\*-CL Images of the Uncooled Combustor at *Qth*,*<sup>n</sup>* = 69% as a Function of *λoverall* .

In order to quantify and compare the phenomena observed in the time-averaged OH\*-CL images, characteristic parameters are derived from the data. For the subsequent analysis only signals above 50% of the image maximum intensity are taken into account. This definition has been found as the most appropriate method to cover the image pixels corresponding visually to the flame. In the following these pixels are called reaction or flame region. With this definition the HAB equals the axial distance between the combustor front plate and the horizontally averaged lower flame boundary. The Dispersion of the OH\*-signal is defined as the area of the flame region divided by the overall area of OH\*-measurement. Therefore, the dispersion is a marker for the relative reaction volume, but not for its homogeneity. The horizontal distribution of the reaction regions is evaluated by the Relative Horizontal Inhomogeneity *dIFlame*/*dx*. This parameter is calculated by vertically averaging the horizontal spatial intensity gradients inside the reaction regions and normalising it to the flame average intensity of each image. This definition ensures that *dIFlame*/*dx* is comparable for different signal intensity levels and flame shapes. The Relative Horizontal Inhomogeneity is a marker for the discreetness of the flames in horizontal direction declining for merging reaction regions.

Figure 13 visualises the Dispersion of OH\*-signal of the uncooled combustor design for all thermal powers as a function of *<sup>λ</sup>overall*. Regarding *Qth*,*<sup>n</sup>* = 100% the dispersion stays approximately constant at 25% between *λoverall* = 1.8 and 2.15. For *λoverall* ≥ 2.4 the magnitude increases significantly up to 73%. This means that with rising Dispersion of OH\*-signal the occupied volume of the reaction region inside the combustion chamber increases substantially until the flame is distributed over almost the whole volume. This indicates a decrease of the Damköhler number which means that the chemical time scale increases in relation to the fluid dynamic time scale [21, 22]. An important influence factor for this behaviour is the exhaust gas recirculation rate which is enhanced by higher jet velocities and higher air numbers [11], respectively. Increasing recirculation serves to enhance the dilution of the fresh gas jets by hot flue gases [23] reducing the chemical reaction rates [24]. Simultaneously, rising jet velocity decreases the fluid dynamic time scale. Thus, both effects lead to a declining Damköhler number. On the other hand by increasing the air number the premixing quality of the air/fuel jets is altered as well. Since these quantities cannot be separated in the recent study, the major influence factor cannot be determined.

The general trend of the Dispersion of OH\*-signal of the uncooled combustor design is similar for all thermal powers. However, the magnitude of the region of constant dispersion at low air numbers decreases for declining power from 25% at *Qth*,*<sup>n</sup>* = 100% down to 14% at *Qth*,*<sup>n</sup>* = 35%. The point of ascending dispersion is located between *<sup>λ</sup>overall* = 2.4 and 2.6 for all thermal powers but no distinct shift of this point can be observed for the uncooled configuration. In terms of reducing temperature peaks and simultaneously reducing NOx emissions, the rapid rise of the Dispersion of OH\*-signal is desireable. However, for the application inside a MGT combustion system the available combustion chamber length limits the feasible volumetric expansion of the reaction region as discussed in section 1. At LBO-near conditions the flame region is expanded substantially so that the reaction process even continue after leaving the exhaust gas duct of the atmospheric test rig. Since operating pressure as well as wall heat loss also have a distinct effect on the flame characteristics, a final selection of the maximum feasible operating point can only be made on the base of a measurement campaign in the MGT test rig.

**Figure 14.** Dispersion of OH\*-signal for the Cooled Combustor Configuration.

The height above burner of the cooled combustor design is visualised in figure 15 for all thermal powers as a function of *λoverall*. This quantity describes the axial distance between the lower flame boundary and the combustor front plate. In the following graphs the HAB is presented normalised with respect to the combustor radius. Regarding the profile at *Qth*,*<sup>n</sup>* = 69% the HAB rises from 42% at *λoverall* = 1.8 up to 68% at *λoverall* = 2.6. At this air number the flame reaches its maximum lift-off height for *Qth*,*<sup>n</sup>* = 69%. With increasing air numbers beyond this point the HAB declines significantly exhibiting at LBO-near conditions with 23% its lowest magnitude. All HAB profiles at different thermal powers show a similar trend as well as a similar maximum lift-off height. However, for decreasing thermal power the profiles are shifted to higher overall air numbers exhibiting lower magnitudes at low air numbers. As described above, the recirculation rate intensifies at rising thermal power. Owing to higher jet velocities and enhanced fresh gas dilution by recirculated flue gases, the lift-off height increases with rising thermal power at low overall air numbers. Furthermore, the maxima of the HAB profiles are directly related to the points of ascending Dispersion of OH\*-signal in figure 14. Due to the shift from discrete reaction zones into a volumetric combustion the reaction region expands to all spatial directions explaining the descent of the HAB. For high *λoverall* this effect of expanding reaction regions outbalances the increase of lift-off height caused by higher jet velocities at rising thermal powers which dominates at low air numbers. The HAB profiles of both combustor designs are visualised in figure 16 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 100%. The graph shows that the profiles of the presented thermal powers fit very well for both combustor configurations. It should be mentioned that this behaviour is consistent with all measured thermal powers. This behaviour demonstrates that the HAB is not affected by the cooling air when both combustors are compared as a function of *λnozzle*(mod). However, *λnozzle*(mod) is based on a shift of *λnozzle* to higher air numbers for the cooled combustor configuration. Therefore, identical profiles of the cooled and uncooled designs illustrate that the cooling air shifts the flame properties to lower nozzle air numbers compared to the uncooled design. In addition, this indicates that the chosen definition of the modified nozzle air number *λnozzle*(mod), which is based on the shift of the CO profiles, is in good agreement with the flame characteristics as well. This affirms the

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

179

http://dx.doi.org/10.5772/54405

assumption that only a part of the cooling air participates in the reaction process.

**Figure 13.** Dispersion of OH\*-signal for the Uncooled Combustor Configuration.

Figure 14 presents the Dispersion of OH\*-signal of the cooled combustor configuration for all thermal powers as a function of *λoverall*. In contrast to the uncooled design, shown in figure 13, the cooled configuration exhibits with a range between 15% and 20% a similar magnitude of the region of constant dispersion at low air numbers for all thermal powers. Moreover, the point of ascending dispersion shifts with decreasing thermal power towards higher overall air numbers. This means that for a constant *λoverall* ≥ 2.6 higher thermal powers show a higher dispersion of the OH\*-signal occupying a significantly larger combustion chamber volume.

<sup>178</sup> Progress in Gas Turbine Performance Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor http://dx.doi.org/10.5772/54405 179

**Figure 14.** Dispersion of OH\*-signal for the Cooled Combustor Configuration.

14 Gas Turbine

volume.

indicates a decrease of the Damköhler number which means that the chemical time scale increases in relation to the fluid dynamic time scale [21, 22]. An important influence factor for this behaviour is the exhaust gas recirculation rate which is enhanced by higher jet velocities and higher air numbers [11], respectively. Increasing recirculation serves to enhance the dilution of the fresh gas jets by hot flue gases [23] reducing the chemical reaction rates [24]. Simultaneously, rising jet velocity decreases the fluid dynamic time scale. Thus, both effects lead to a declining Damköhler number. On the other hand by increasing the air number the premixing quality of the air/fuel jets is altered as well. Since these quantities cannot be

The general trend of the Dispersion of OH\*-signal of the uncooled combustor design is similar for all thermal powers. However, the magnitude of the region of constant dispersion at low air numbers decreases for declining power from 25% at *Qth*,*<sup>n</sup>* = 100% down to 14% at *Qth*,*<sup>n</sup>* = 35%. The point of ascending dispersion is located between *<sup>λ</sup>overall* = 2.4 and 2.6 for all thermal powers but no distinct shift of this point can be observed for the uncooled configuration. In terms of reducing temperature peaks and simultaneously reducing NOx emissions, the rapid rise of the Dispersion of OH\*-signal is desireable. However, for the application inside a MGT combustion system the available combustion chamber length limits the feasible volumetric expansion of the reaction region as discussed in section 1. At LBO-near conditions the flame region is expanded substantially so that the reaction process even continue after leaving the exhaust gas duct of the atmospheric test rig. Since operating pressure as well as wall heat loss also have a distinct effect on the flame characteristics, a final selection of the maximum feasible operating point can only be made on the base of a

separated in the recent study, the major influence factor cannot be determined.

measurement campaign in the MGT test rig.

**Figure 13.** Dispersion of OH\*-signal for the Uncooled Combustor Configuration.

Figure 14 presents the Dispersion of OH\*-signal of the cooled combustor configuration for all thermal powers as a function of *λoverall*. In contrast to the uncooled design, shown in figure 13, the cooled configuration exhibits with a range between 15% and 20% a similar magnitude of the region of constant dispersion at low air numbers for all thermal powers. Moreover, the point of ascending dispersion shifts with decreasing thermal power towards higher overall air numbers. This means that for a constant *λoverall* ≥ 2.6 higher thermal powers show a higher dispersion of the OH\*-signal occupying a significantly larger combustion chamber The height above burner of the cooled combustor design is visualised in figure 15 for all thermal powers as a function of *λoverall*. This quantity describes the axial distance between the lower flame boundary and the combustor front plate. In the following graphs the HAB is presented normalised with respect to the combustor radius. Regarding the profile at *Qth*,*<sup>n</sup>* = 69% the HAB rises from 42% at *λoverall* = 1.8 up to 68% at *λoverall* = 2.6. At this air number the flame reaches its maximum lift-off height for *Qth*,*<sup>n</sup>* = 69%. With increasing air numbers beyond this point the HAB declines significantly exhibiting at LBO-near conditions with 23% its lowest magnitude. All HAB profiles at different thermal powers show a similar trend as well as a similar maximum lift-off height. However, for decreasing thermal power the profiles are shifted to higher overall air numbers exhibiting lower magnitudes at low air numbers. As described above, the recirculation rate intensifies at rising thermal power. Owing to higher jet velocities and enhanced fresh gas dilution by recirculated flue gases, the lift-off height increases with rising thermal power at low overall air numbers. Furthermore, the maxima of the HAB profiles are directly related to the points of ascending Dispersion of OH\*-signal in figure 14. Due to the shift from discrete reaction zones into a volumetric combustion the reaction region expands to all spatial directions explaining the descent of the HAB. For high *λoverall* this effect of expanding reaction regions outbalances the increase of lift-off height caused by higher jet velocities at rising thermal powers which dominates at low air numbers.

The HAB profiles of both combustor designs are visualised in figure 16 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 100%. The graph shows that the profiles of the presented thermal powers fit very well for both combustor configurations. It should be mentioned that this behaviour is consistent with all measured thermal powers. This behaviour demonstrates that the HAB is not affected by the cooling air when both combustors are compared as a function of *λnozzle*(mod). However, *λnozzle*(mod) is based on a shift of *λnozzle* to higher air numbers for the cooled combustor configuration. Therefore, identical profiles of the cooled and uncooled designs illustrate that the cooling air shifts the flame properties to lower nozzle air numbers compared to the uncooled design. In addition, this indicates that the chosen definition of the modified nozzle air number *λnozzle*(mod), which is based on the shift of the CO profiles, is in good agreement with the flame characteristics as well. This affirms the assumption that only a part of the cooling air participates in the reaction process.

**Figure 15.** HAB for the Cooled Combustor Configuration as a Function of *λoverall* .

**Figure 17.** Relative Horizontal Inhomogeneity for the Uncooled Combustor Design as a Function of *λoverall* .

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

181

http://dx.doi.org/10.5772/54405

**Figure 18.** Relative Horizontal Inhomogeneity for the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*(mod). The comparison of *dIFlame*/*dx* between the uncooled and the cooled combustor configuration is presented in figure 18 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 84%. These thermal powers are chosen exemplarily since the occuring effects are well pronounced for these profiles but it should be mentioned that the discussed behaviour applies to all thermal powers. For rising *λnozzle*(mod) the curves of both designs show a similar declining trend of *dIFlame*/*dx* at all *Qth*,*n*. For higher air numbers the profiles of the cooled and uncooled configurations fit very well in magnitude at a constant thermal power converging to a similar lower threshold of approximately 0.5 %/dPixel. However, at low air numbers the uncooled design distinctly exhibits higher magnitudes of the Relative Horizontal Inhomogeneity. The difference of both designs decreases with rising thermal power as well as rising air number. This means that the difference in *dIFlame*/*dx* decreases with increasing axial jet velocity *unozzle*. As shown in figure 2 the cooling air penetrates the combustion chamber through small exit holes located between the air nozzles. Since the split between cooling air and process air is almost constant over the whole operating range, the cooling jet velocity scales with the axial jet velocity *unozzle* and air number, respectively. Figure 18 indicates that for

**Figure 16.** HAB for the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*(mod).

Figure 17 visualises the Relative Horizontal Inhomogeneity *dIFlame*/*dx* for the uncooled combustor configuration for all thermal powers as a function of *λoverall*. This quantity describes the discreetness of the reaction zones in horizontal direction. The profile at *Qth*,*<sup>n</sup>* = 100% exhibits an inhomogeneity of 1.4 %/dPixel at *λoverall* = 1.8 converging exponentially with increasing air numbers towards a lower threshold of approximately 0.5 %/dPixel which is reached at *λoverall* ≥ 2.6. This means that for lower air numbers separated flames around the penetrating fresh gas jets exist, which merge together for rising air numbers resulting in a horizontally distributed reaction region for high *λoverall*. The profiles of all thermal powers exhibit a lower threshold of the same magnitude which is reached at LBO-near conditions. The general trend of the profiles is similar for 53% <sup>≤</sup> *Qth*,*<sup>n</sup>* <sup>≤</sup> 100%. Only the profile at *Qth*,*<sup>n</sup>* = 35% differs from the exponential declining trend. However, at low overall air numbers the profiles' magnitudes are staggered in thermal power reaching higher values for low *Qth*,*n*. This signifies that higher thermal powers exhibit better horizontally distributed flames which is due to an enhanced recirculation rate, better premixing quality at high flow rates as well as an enhanced interjet mixing rate.

<sup>180</sup> Progress in Gas Turbine Performance Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas Turbine Combustor http://dx.doi.org/10.5772/54405 181

16 Gas Turbine

**Figure 15.** HAB for the Cooled Combustor Configuration as a Function of *λoverall* .

**Figure 16.** HAB for the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*(mod).

as an enhanced interjet mixing rate.

Figure 17 visualises the Relative Horizontal Inhomogeneity *dIFlame*/*dx* for the uncooled combustor configuration for all thermal powers as a function of *λoverall*. This quantity describes the discreetness of the reaction zones in horizontal direction. The profile at *Qth*,*<sup>n</sup>* = 100% exhibits an inhomogeneity of 1.4 %/dPixel at *λoverall* = 1.8 converging exponentially with increasing air numbers towards a lower threshold of approximately 0.5 %/dPixel which is reached at *λoverall* ≥ 2.6. This means that for lower air numbers separated flames around the penetrating fresh gas jets exist, which merge together for rising air numbers resulting in a horizontally distributed reaction region for high *λoverall*. The profiles of all thermal powers exhibit a lower threshold of the same magnitude which is reached at LBO-near conditions. The general trend of the profiles is similar for 53% <sup>≤</sup> *Qth*,*<sup>n</sup>* <sup>≤</sup> 100%. Only the profile at *Qth*,*<sup>n</sup>* = 35% differs from the exponential declining trend. However, at low overall air numbers the profiles' magnitudes are staggered in thermal power reaching higher values for low *Qth*,*n*. This signifies that higher thermal powers exhibit better horizontally distributed flames which is due to an enhanced recirculation rate, better premixing quality at high flow rates as well

**Figure 17.** Relative Horizontal Inhomogeneity for the Uncooled Combustor Design as a Function of *λoverall* .

**Figure 18.** Relative Horizontal Inhomogeneity for the Cooled and Uncooled Combustor Designs as a Function of *λnozzle*(mod).

The comparison of *dIFlame*/*dx* between the uncooled and the cooled combustor configuration is presented in figure 18 as a function of *<sup>λ</sup>nozzle*(mod) for *Qth*,*<sup>n</sup>* = 53% and 84%. These thermal powers are chosen exemplarily since the occuring effects are well pronounced for these profiles but it should be mentioned that the discussed behaviour applies to all thermal powers. For rising *λnozzle*(mod) the curves of both designs show a similar declining trend of *dIFlame*/*dx* at all *Qth*,*n*. For higher air numbers the profiles of the cooled and uncooled configurations fit very well in magnitude at a constant thermal power converging to a similar lower threshold of approximately 0.5 %/dPixel. However, at low air numbers the uncooled design distinctly exhibits higher magnitudes of the Relative Horizontal Inhomogeneity. The difference of both designs decreases with rising thermal power as well as rising air number. This means that the difference in *dIFlame*/*dx* decreases with increasing axial jet velocity *unozzle*. As shown in figure 2 the cooling air penetrates the combustion chamber through small exit holes located between the air nozzles. Since the split between cooling air and process air is almost constant over the whole operating range, the cooling jet velocity scales with the axial jet velocity *unozzle* and air number, respectively. Figure 18 indicates that for lower jet velocities the penetrating cooling air serves to broaden the discreet reaction regions horizontally and therefore to homogenise their distribution. However, the homogenising of the flame regions for the cooled combustor configuration seems to have no effect on exhaust gas emissions at low air numbers as shown in figure 10 and 11.

[3] Fritsche D, Füri M, Boulouchos K (2007) An Experimental Investigation of Thermoacoustic Instabilities in a Premixed Swirl-stabilized Flame, Combustion and

Experimental Investigation of the Influence of Combustor Cooling on the Characteristics of a FLOX©-Based Micro Gas

Turbine Combustor

183

http://dx.doi.org/10.5772/54405

[4] Huang Y, Yang V (2009) Dynamics and Stability of Lean-premixed Swirl-stabilized

[5] Nauert A, Peterson P, Linne M, Dreizler A (2007) Experimental Analysis of Flashback in Lean Premixed Swirling Flames: Conditions Close to Flashback, Exp Fluids 43, pp.

[6] Wünning, J.A. and Wünning, J.G.(1997) Flameless Oxidation to Reduce Thermal

[7] Hamdi M, Benticha H, Sassi M (2012) Fundamentals and Simulation of MILD Combustion, Thermal Power Plants, Dr. Mohammad Rasul (Ed.), ISBN:

[8] Weber R, Smarta J P, vd Kamp W (2005) On the (MILD) Combustion of Gaseous, Liquid and Solid Fuels in High Temperature Preheated Air, Proceedings of the Combustion

[9] Arghode V K, Gupta A K (2010) Effect of Flow Field for Colorless Distributed Combustion (CDC) for Gas Turbine Combustion, Applied Energy 87(5), pp. 1631-1640.

[10] Tsuji H, Gupta A K, Hasegawa T, Katsuki M, Kishimoto K, Morita M (2003) High

[11] Li G, Gutmark E J, Stankovic D, Overman N, Cornwell M, Fuchs L, Vladimir M (2006) Experimental Study of Flameless Combustion in Gas Turbine Combustors, Proceedings

[12] Lückerath R, Meier W, Aigner M (2007) FLOX© Combustion at High Pressure with Different Fuel Compositions, Proceedings of GT2007 ASME Turbo Expo 2007,

[13] Lammel O, Schütz H, Schmitz G, Lückerath R, Stöhr M, Noll B, Aigner M (2010) FLOX© Combustion at High Power Density and High Flame Temperatures, Proceedings of

[14] Schütz H, Lückerath R, Kretschmer T, Noll B, Aigner M (2006) Analysis of the Pollutant Formation in the FLOX© Combustion, Proceedings of GT2006 ASME Turbo Expo 2006,

[15] Duwig C, Stankovic D, Fuchs L, Li G, Gutmark E (2008) Experimental and Numerical Study of Flameless Combustion in a Model Gas Turbine Combustor, Combustion

of 44th AIAA Aerospace Sciences Meeting and Exhibit, AIAA 2006-546.

Combustion, Progress in Energy and Combustion Science 35, pp. 293-364.

NO-Formation, Prog. Energy ComlnaL Sci.23, pp. 81-94.

Temperature Air Combustion, CRC Press, Florida.

GT2010 ASME Turbo Expo 2010, GT2010-23385.

Science and Technology 180, pp. 279-295.

978-953-307-952-3, InTech. pp. 43-64.

Institute, Vol. 30, pp. 2623-2629

GT2007-27337.

GT2006-91041.

Flame 151, pp. 29-36.

89-100.

#### **4. Conclusion**

A FLOX©-based micro gas turbine combustor was introduced. The presented experimental study compared an impingement cooled combustor configuration to an uncooled design for natural gas. The influence of selected operating conditions on shape, location and homogeneity of the reaction zones was analysed under atmospheric conditions using time-averaged OH\*-chemiluminescence images. Furthermore, the dependencies of jet velocity and combustor front plate cooling on LBO limits were discussed. Exhaust gas emissions were presented and a definition of a modified nozzle air number was derived from comparing the CO profiles of the cooled and uncooled design. With the help of this parameter the differences of both designs were analysed.

Regarding the parameters derived from OH\*-chemiluminescence images a distinct increase of Dispersion of OH\*-signal was observed for rising air numbers leading to a volumetric reaction region at LBO-near conditions. Simultaneously, the detachedness and the horizontal inhomogeneity of the reaction regions reduced substantially. The influence of the cooling air was observed to generate a shift of all emission and flame profiles to lower nozzle air numbers. However, it was discussed that only a part of the cooling air interacts with the reaction region whereas the rest of the cooling air passes the combustion chamber without participating in the combustion process. Moreover, it was shown that with the cooled combustor design higher overall air numbers can be realised at low thermal powers which is advantageous if the amount of combustion air is to be maximised.

#### **Acknowledgements**

The financial support of this work by the EnBW Energie Baden-Württemberg AG and the German Federal Ministry of Economics and Technology is gratefully acknowledged. Furthermore, the authors would like to thank Marco Graf for his support.

#### **Author details**

Zanger Jan⋆, Monz Thomas and Aigner Manfred

<sup>⋆</sup> Address all correspondence to: jan.zanger@dlr.de

German Aerospace Center, Institute of Combustion Technology, Stuttgart, Germany

#### **References**


[3] Fritsche D, Füri M, Boulouchos K (2007) An Experimental Investigation of Thermoacoustic Instabilities in a Premixed Swirl-stabilized Flame, Combustion and Flame 151, pp. 29-36.

18 Gas Turbine

**4. Conclusion**

**Acknowledgements**

**Author details**

**References**

lower jet velocities the penetrating cooling air serves to broaden the discreet reaction regions horizontally and therefore to homogenise their distribution. However, the homogenising of the flame regions for the cooled combustor configuration seems to have no effect on exhaust

A FLOX©-based micro gas turbine combustor was introduced. The presented experimental study compared an impingement cooled combustor configuration to an uncooled design for natural gas. The influence of selected operating conditions on shape, location and homogeneity of the reaction zones was analysed under atmospheric conditions using time-averaged OH\*-chemiluminescence images. Furthermore, the dependencies of jet velocity and combustor front plate cooling on LBO limits were discussed. Exhaust gas emissions were presented and a definition of a modified nozzle air number was derived from comparing the CO profiles of the cooled and uncooled design. With the help of this

Regarding the parameters derived from OH\*-chemiluminescence images a distinct increase of Dispersion of OH\*-signal was observed for rising air numbers leading to a volumetric reaction region at LBO-near conditions. Simultaneously, the detachedness and the horizontal inhomogeneity of the reaction regions reduced substantially. The influence of the cooling air was observed to generate a shift of all emission and flame profiles to lower nozzle air numbers. However, it was discussed that only a part of the cooling air interacts with the reaction region whereas the rest of the cooling air passes the combustion chamber without participating in the combustion process. Moreover, it was shown that with the cooled combustor design higher overall air numbers can be realised at low thermal powers which is

The financial support of this work by the EnBW Energie Baden-Württemberg AG and the German Federal Ministry of Economics and Technology is gratefully acknowledged.

gas emissions at low air numbers as shown in figure 10 and 11.

parameter the differences of both designs were analysed.

advantageous if the amount of combustion air is to be maximised.

Zanger Jan⋆, Monz Thomas and Aigner Manfred <sup>⋆</sup> Address all correspondence to: jan.zanger@dlr.de

Conversion and Management 42, pp. 115-4.

and Power, Applied Thermal Engineering 20, pp. 1421-1429.

Furthermore, the authors would like to thank Marco Graf for his support.

German Aerospace Center, Institute of Combustion Technology, Stuttgart, Germany

[1] Bhatt M S (2001) Mapping of General Combined Heat and Power Systems, Energy

[2] Pilavachi P A (2000) Power Generation with Gas Turbine Systems and Combined Heat


[16] Lammel O, Stöhr M, Kutne P, Dem C, Meier W, Aigner M (2011) Experimental Analysis of Confined Jet Flames by Laser Measurement Techniques, Proceedings of GT2011 ASME Turbo Expo 2011, GT2011-45111.

**Section 3**

**Fault Detection in Systems and Materials**


**Fault Detection in Systems and Materials**

20 Gas Turbine

pp. 131-150.

184 Progress in Gas Turbine Performance

[16] Lammel O, Stöhr M, Kutne P, Dem C, Meier W, Aigner M (2011) Experimental Analysis of Confined Jet Flames by Laser Measurement Techniques, Proceedings of GT2011

[17] Dandy D S, Vosen S R (1992) Numerical and Experimental Studies of Hydroxyl Radical Chemiluminescence in Methane-Air Flames, Combustion Science and Technology 82,

[18] Vaz D C, Buiktenen J P, Borges A R J, Spliethoff H (2004) On the Stability Range of a Cylindrical Combustor for Operation in the FLOX Regime, Proceedings of GT2004

[19] Lefevbre A H, Ballal D R (2010) Gas Turbine Combustion: Alternative Fuels and

[21] Borghi R (1985) On the Structure and Morphology of Turbulent Premixed Flames,

[22] Sadanandan R, Lückerath R, Meier W, Wahl C (2011) Flame Characteristics and Emissions in Flameless Combustion Under Gas Turbine Relevant Conditions, Journal

[23] Levy Y, Rao G A, Sherbaum V (2007) Preliminary Analysis of a New Methodology for Flameless Combustion in Gas Turbine Combustors, Proceedings of GT2007 ASME

[24] Li P, Mi J, Dally B B, Wang F, Wang L, Liu Z, Chen S, Zheng C (2010) Progress and recent trend in MILD combustion, Science China Technological Sciences 54(2), pp. 255-269.

Recent Advances in Aerospace Sciences, Plenum Press, New Yorck.

ASME Turbo Expo 2011, GT2011-45111.

ASME Turbo Expo 2004, GT2004-53790.

Emissions – 3rd ed., CRC Press, Boca Raton.

of Propulsion and Power 27(5), pp. 970-980.

Turbo Expo 2007, GT2007-27766.

[20] Joos F (2006) Technische Verbrennung, Springer, Berlin.

**Chapter 8**

**Engine Condition Monitoring and Diagnostics**

Any engine exhibits the effects of wear and tear over time. Several mechanisms cause the deg‐ radation and potential failures of gas turbines such as dirt build-up, fouling, erosion, oxida‐ tion, corrosion, foreign object damage, worn bearings, worn seals, excessive blade tip clearances, burned or warped turbine vanes or blades, partially or wholly missing blades or vanes, plugged fuel nozzles, cracked and warped combustors, or a cracked rotor disc or blade.

Fouling is caused by liquid or solid particles accumulated to airfoils and annulus surfaces. Deposits consist of varying amounts of moisture, oil, soot, water-soluble constituents, in‐ soluble dirt, and corrosion products of the compressor blades material whish are held to‐ gether by moisture and oil. The result is a build-up of material that causes increased surface roughness and to some degree changes the shape of the airfoil. Hot corrosion is the loss or deterioration of material from flow path components caused by chemical reactions between the component and certain contaminants, such as salts (for example sodium and potassium), mineral acids or reactive gases (such as hydrogen sulfide or sulfur oxides). Corrosion is caused by noxious fumes or ash-forming substances present in the fuel such as aluminum, calcium, iron, nickel, potassium, sodium, silicon, magnesium. Corrosion increases surface roughness and causes pitting. Erosion is the abrasive removal of material from the flow path by hard or incompressible particles impinging on flow surfaces. Damage may also be caused by foreign objects striking the flow path components (Figure. 1a). Foreign Object Damage (FOD) is defined as material (nuts, bolts, ice, birds, etc.) ingested into the engine from out‐ side the engine envelope. Domestic Object Damage (DOD) is defined as objects from any

Different causes and mechanisms of performance deterioration of jet engines are reviewed in [1]. Degradation in both land and aero gas turbines is also reviewed by Kurz and Brun

> © 2013 Stamatis; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,

© 2013 Stamatis; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

distribution, and reproduction in any medium, provided the original work is properly cited.

Anastassios G. Stamatis

http://dx.doi.org/10.5772/54409

other part of the engine itself.

**1. Introduction**

Additional information is available at the end of the chapter

## **Engine Condition Monitoring and Diagnostics**

Anastassios G. Stamatis

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/54409

#### **1. Introduction**

Any engine exhibits the effects of wear and tear over time. Several mechanisms cause the deg‐ radation and potential failures of gas turbines such as dirt build-up, fouling, erosion, oxida‐ tion, corrosion, foreign object damage, worn bearings, worn seals, excessive blade tip clearances, burned or warped turbine vanes or blades, partially or wholly missing blades or vanes, plugged fuel nozzles, cracked and warped combustors, or a cracked rotor disc or blade.

Fouling is caused by liquid or solid particles accumulated to airfoils and annulus surfaces. Deposits consist of varying amounts of moisture, oil, soot, water-soluble constituents, in‐ soluble dirt, and corrosion products of the compressor blades material whish are held to‐ gether by moisture and oil. The result is a build-up of material that causes increased surface roughness and to some degree changes the shape of the airfoil. Hot corrosion is the loss or deterioration of material from flow path components caused by chemical reactions between the component and certain contaminants, such as salts (for example sodium and potassium), mineral acids or reactive gases (such as hydrogen sulfide or sulfur oxides). Corrosion is caused by noxious fumes or ash-forming substances present in the fuel such as aluminum, calcium, iron, nickel, potassium, sodium, silicon, magnesium. Corrosion increases surface roughness and causes pitting. Erosion is the abrasive removal of material from the flow path by hard or incompressible particles impinging on flow surfaces. Damage may also be caused by foreign objects striking the flow path components (Figure. 1a). Foreign Object Damage (FOD) is defined as material (nuts, bolts, ice, birds, etc.) ingested into the engine from out‐ side the engine envelope. Domestic Object Damage (DOD) is defined as objects from any other part of the engine itself.

Different causes and mechanisms of performance deterioration of jet engines are reviewed in [1]. Degradation in both land and aero gas turbines is also reviewed by Kurz and Brun

© 2013 Stamatis; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Stamatis; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

[2], who pointed out differences in mechanisms for the two types. Industrial gas turbine de‐ terioration has been discussed by Diakunchak [3].

faults will become evident as vibration increases or by a change in lubrication oil tempera‐ ture. However, some serious faults can be detected only through gas path analysis. The gas

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

189

Diagnosis of a mechanical condition is the ability to infer about the condition of parts of the engine, without dismantling the engine or getting direct access to these parts, but only from observations of information coming to the engine exterior. The field of engineering science covering the techniques for achieving a diagnosis is called diagnostics. The aim of diagnos‐ tics is to detect the presence and identify the kind of faults appearing in a engine. Diagnos‐ tics does not require that the engine is either stopped or disassembled. Information is gathered while the engine is in operation. This is vital for engines in the process industry or energy production, as they must run without interruption for long time intervals. Detection of an incipient failure in a jet engine leads to taking action necessary to prevent a catastroph‐

In order to establish the possibility of diagnosing engine condition a correspondence of this condition to the values of the measured quantities should be known. In general terms, this correspondence is intrinsically established through the physical laws governing the opera‐ tion of the machine. The behavior of any relevant physical quantity is linked through these laws to the detailed geometry of the machine and the kind of phenomena taking place in it. If we consider a machine using a fluid as a working medium, the variation of the flow quan‐ tities at one particular location in the machine is determined, via the laws of fluid mechan‐ ics, from the geometry of the solid boundaries and the physical properties of the fluid. A change in geometry will then reflect on the values of the flow quantities and could be calcu‐ lated by application of the relevant physical laws. If suitable quantities are measured, they reflect changes in geometry or material and can therefore be used to indicate the presence of a fault. It is obvious that according to the change occurring in an operating machine, differ‐ ent quantities will be influenced. For example, the operation of rotating components is al‐ ways linked to the exertion of periodic forces, with a frequency which is usually a multiple of the frequency of rotation. In this respect, the quantities characterizing a vibration are suit‐ able for diagnostic purposes. On the other hand, severe corrosion, as it changes turbine air‐

Many techniques for inferring engine status or change in engine condition have been proposed and/or applied to various engine configurations with varying success. Some of them (e.g. Vi‐ bration monitoring, Trending Analysis) are parts of computer-controlled data-acquisition sys‐ tems that permit the on-line acquisition and reduction of a very large amount of performance information. While fault detection or general deterioration could be based on immediate obser‐ vation of reduced measurable quantities, such observation is not, generally, adequate. It should also be noted that a change in any measured parameter does not necessarily indicate a particular independent parameter fault. For example, a change in compressor discharge pres‐ sure (CDP) does not necessarily indicate a dirty compressor. The change could also be due to a combined compressor and turbine fault or to a turbine fault alone. In order to have access to the variables, which possess diagnostic information (such as component efficiencies) modeling of an engine is essential. Thermodynamic (Gas Path) analysis methods employ engine models to

path, in its simplest form, consists of the compressors, combustor, and turbines.

ic failure which might follow.

foil geometry, is detectable through gas path analysis.

**Figure 1.** (a) FOD effects, (b) Turbine nozzles with deposits.

Three major effects determine the performance deterioration of the gas turbine compressor due to fouling: Increased tip clearances, changes in airfoil geometry, and changes in airfoil surface quality. In compressors, erosion increases tip clearance, shortens blade chords, in‐ creases pressure surface roughness, blunts the leading edge, and sharpens the trailing edge. Turbine blade oxidation, corrosion and erosion are normally longtime processes with mate‐ rial losses occurring slowly over a period of time. However, damage resulting from impact by a foreign object is usually sudden. Impact damage to the turbine blades and vanes will result in parameter changes similar to severe erosion or corrosion. Corrosion, erosion, oxida‐ tion or impact damage increases the area size of the turbine nozzle. When crude oil is burned in the GT the hot end is subjected to additional harmful deposits, including salt de‐ posits originating in the inlet or from fuel additives. As hot combustion products pass through the first stage nozzle, they experience a drop in static temperature and some ashes may be deposited on the nozzle blades decreasing the nozzle area (Figure 1b). The combus‐ tion system is not likely to be the direct cause for performance deterioration. The combus‐ tion efficiency will usually not decrease, except for severe cases of combustor distress. However, plugged nozzles and/or combustor and transition piece failures will always result in distorted exhaust gas temperature patterns. This is a result of the swirl effect through the turbine from the combustor to the exhaust gas temperature-measuring plane. Distortion in the temperature pattern or temperature profile not only affects combustor performance but can have a far reaching impact as local temperature peaks can damage the turbine section.

All the above causes and effects may be considered as faults. Generally speaking, fault is a condition of a machine linked to a change of the form of its parts and of its way of operation, from what the machine was originally designed for and was achieved during its initial oper‐ ation. In this respect a fault manifests itself by a change of geometrical characteristics or/and integrity of the material of parts of an engine. Change in geometry is inevitably linked to common experience faults, as for example when a part is broken, or deformed. Typical in‐ tegrity fault is the occurrence of cracks inside the material, which are not associated to any geometrical change but can nevertheless result into catastrophic consequences. Some of the faults will become evident as vibration increases or by a change in lubrication oil tempera‐ ture. However, some serious faults can be detected only through gas path analysis. The gas path, in its simplest form, consists of the compressors, combustor, and turbines.

[2], who pointed out differences in mechanisms for the two types. Industrial gas turbine de‐

(a) (b)

Three major effects determine the performance deterioration of the gas turbine compressor due to fouling: Increased tip clearances, changes in airfoil geometry, and changes in airfoil surface quality. In compressors, erosion increases tip clearance, shortens blade chords, in‐ creases pressure surface roughness, blunts the leading edge, and sharpens the trailing edge. Turbine blade oxidation, corrosion and erosion are normally longtime processes with mate‐ rial losses occurring slowly over a period of time. However, damage resulting from impact by a foreign object is usually sudden. Impact damage to the turbine blades and vanes will result in parameter changes similar to severe erosion or corrosion. Corrosion, erosion, oxida‐ tion or impact damage increases the area size of the turbine nozzle. When crude oil is burned in the GT the hot end is subjected to additional harmful deposits, including salt de‐ posits originating in the inlet or from fuel additives. As hot combustion products pass through the first stage nozzle, they experience a drop in static temperature and some ashes may be deposited on the nozzle blades decreasing the nozzle area (Figure 1b). The combus‐ tion system is not likely to be the direct cause for performance deterioration. The combus‐ tion efficiency will usually not decrease, except for severe cases of combustor distress. However, plugged nozzles and/or combustor and transition piece failures will always result in distorted exhaust gas temperature patterns. This is a result of the swirl effect through the turbine from the combustor to the exhaust gas temperature-measuring plane. Distortion in the temperature pattern or temperature profile not only affects combustor performance but can have a far reaching impact as local temperature peaks can damage the turbine section.

All the above causes and effects may be considered as faults. Generally speaking, fault is a condition of a machine linked to a change of the form of its parts and of its way of operation, from what the machine was originally designed for and was achieved during its initial oper‐ ation. In this respect a fault manifests itself by a change of geometrical characteristics or/and integrity of the material of parts of an engine. Change in geometry is inevitably linked to common experience faults, as for example when a part is broken, or deformed. Typical in‐ tegrity fault is the occurrence of cracks inside the material, which are not associated to any geometrical change but can nevertheless result into catastrophic consequences. Some of the

terioration has been discussed by Diakunchak [3].

188 Progress in Gas Turbine Performance

**Figure 1.** (a) FOD effects, (b) Turbine nozzles with deposits.

Diagnosis of a mechanical condition is the ability to infer about the condition of parts of the engine, without dismantling the engine or getting direct access to these parts, but only from observations of information coming to the engine exterior. The field of engineering science covering the techniques for achieving a diagnosis is called diagnostics. The aim of diagnos‐ tics is to detect the presence and identify the kind of faults appearing in a engine. Diagnos‐ tics does not require that the engine is either stopped or disassembled. Information is gathered while the engine is in operation. This is vital for engines in the process industry or energy production, as they must run without interruption for long time intervals. Detection of an incipient failure in a jet engine leads to taking action necessary to prevent a catastroph‐ ic failure which might follow.

In order to establish the possibility of diagnosing engine condition a correspondence of this condition to the values of the measured quantities should be known. In general terms, this correspondence is intrinsically established through the physical laws governing the opera‐ tion of the machine. The behavior of any relevant physical quantity is linked through these laws to the detailed geometry of the machine and the kind of phenomena taking place in it. If we consider a machine using a fluid as a working medium, the variation of the flow quan‐ tities at one particular location in the machine is determined, via the laws of fluid mechan‐ ics, from the geometry of the solid boundaries and the physical properties of the fluid. A change in geometry will then reflect on the values of the flow quantities and could be calcu‐ lated by application of the relevant physical laws. If suitable quantities are measured, they reflect changes in geometry or material and can therefore be used to indicate the presence of a fault. It is obvious that according to the change occurring in an operating machine, differ‐ ent quantities will be influenced. For example, the operation of rotating components is al‐ ways linked to the exertion of periodic forces, with a frequency which is usually a multiple of the frequency of rotation. In this respect, the quantities characterizing a vibration are suit‐ able for diagnostic purposes. On the other hand, severe corrosion, as it changes turbine air‐ foil geometry, is detectable through gas path analysis.

Many techniques for inferring engine status or change in engine condition have been proposed and/or applied to various engine configurations with varying success. Some of them (e.g. Vi‐ bration monitoring, Trending Analysis) are parts of computer-controlled data-acquisition sys‐ tems that permit the on-line acquisition and reduction of a very large amount of performance information. While fault detection or general deterioration could be based on immediate obser‐ vation of reduced measurable quantities, such observation is not, generally, adequate. It should also be noted that a change in any measured parameter does not necessarily indicate a particular independent parameter fault. For example, a change in compressor discharge pres‐ sure (CDP) does not necessarily indicate a dirty compressor. The change could also be due to a combined compressor and turbine fault or to a turbine fault alone. In order to have access to the variables, which possess diagnostic information (such as component efficiencies) modeling of an engine is essential. Thermodynamic (Gas Path) analysis methods employ engine models to process measurement data, in order to diagnose changes in component performance which may be linked to degradation, aging, or incipient failure.

#### **2. Gas path analysis**

An engine may be viewed as a system, whose operating point is defined by means of a set of variables, denoted as **u**. The operation of each component follows predictable thermody‐ namic laws. Therefore, each component will behave in a predictable manner when operating under a given set of conditions. The health condition of its components is assumed to be represented through the values of a set of appropriate ''health'' parameters such as efficien‐ cies and flow capacities, contained in a vector **f**. The system is observed through measured variables, such as speeds, pressures, temperatures, contained in a vector **y**. When the engine operates at a certain operating point measured quantities are produced for given values of health parameters. The operating engine establishes a relationship between these parame‐ ters, which can be expressed though a functional relation:

$$\mathbf{y} = \mathbf{F}(\mathbf{u}, \mathbf{f}) \tag{1}$$

Many variants of Gas Path Analysis based diagnosis with different features and complexity have been developed and reported in the open literature. Extensive reviews of existing

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

191

The data used can be taken in steady state or transient operation. The model could be a physical one representing the aerothermodynamic processes taking place in the engine com‐ ponents and the mechanical coupling between them or a black box mathematical model re‐ lating data with health parameters. The diagnostic decision making procedure may be a conventional pattern recognition technique applied to health parameter space or an artificial

Accordingly the proposed methods are classified on the basis of the kind of the comprising elements as: Steady state or Transient, Physical or Mathematical, Conventional or Artificial

In linear gas path analysis, the health parameters are represented as the unknown ''deltas'' of component performance parameters (typically efficiency and mass flow capacity). They are related to known measurement ''deltas'' through relations produced by linearization of the general nonlinear thermodynamic relations, assuming small deviations. [6].The classical linear approach is formulated as follows: For a given operating point u the measurement values depend only on the health condition of engine components. After linearization and taking into account measurement uncertainty (by adding a noise vector **v** with zero mean

where Δ is called delta and represents percentage deviation from a reference value(when the engine is in intact condition) and C the well-known influence coefficient matrix. Estimation

**Δy = C Δf v** × + (2)

ˆ *<sup>T</sup>* -<sup>1</sup> ×× × **-1 Δf = S C R Δy** (3)

and known covariance R), the typical GPA equations take the form:

of health parameters is obtained from the relations

Generally speaking any GPA method at least consists of the following elements:

**•** A data processing model relating measured data with health parameters

methods provided by Li [4], and Marinai et al. [5].

**•** A diagnostic decision making procedure.

intelligence based expert system.

**3. Physical models based GPA**

intelligence method.

**3.1. Linear methods**

**•** Measured data

A computer model materializing this relation can reproduce the values of any thermody‐ namic quantity measured along the engine gas path. It is interesting to note that by assign‐ ing appropriate values to the components of vector **f**, the effect of engine component faults or deterioration on measured quantities can be reproduced.

The problem of diagnostics (Figure 2) is to seek a solution to the inverse problem, namely to determine the values of the estimated health parameters *f* **^** from a given set of measure‐ ments using a diagnostic method (*DM*). Particular faults can then be detected if deviations of health parameters from the reference state are observed.

**Figure 2.** Gas Path Analysis diagnostics formulation.

Many variants of Gas Path Analysis based diagnosis with different features and complexity have been developed and reported in the open literature. Extensive reviews of existing methods provided by Li [4], and Marinai et al. [5].

Generally speaking any GPA method at least consists of the following elements:

**•** Measured data

process measurement data, in order to diagnose changes in component performance which

An engine may be viewed as a system, whose operating point is defined by means of a set of variables, denoted as **u**. The operation of each component follows predictable thermody‐ namic laws. Therefore, each component will behave in a predictable manner when operating under a given set of conditions. The health condition of its components is assumed to be represented through the values of a set of appropriate ''health'' parameters such as efficien‐ cies and flow capacities, contained in a vector **f**. The system is observed through measured variables, such as speeds, pressures, temperatures, contained in a vector **y**. When the engine operates at a certain operating point measured quantities are produced for given values of health parameters. The operating engine establishes a relationship between these parame‐

A computer model materializing this relation can reproduce the values of any thermody‐ namic quantity measured along the engine gas path. It is interesting to note that by assign‐ ing appropriate values to the components of vector **f**, the effect of engine component faults

The problem of diagnostics (Figure 2) is to seek a solution to the inverse problem, namely to

ments using a diagnostic method (*DM*). Particular faults can then be detected if deviations

**y F(u,f)** = (1)

from a given set of measure‐

**^**

may be linked to degradation, aging, or incipient failure.

ters, which can be expressed though a functional relation:

or deterioration on measured quantities can be reproduced.

determine the values of the estimated health parameters *f*

of health parameters from the reference state are observed.

**Figure 2.** Gas Path Analysis diagnostics formulation.

**2. Gas path analysis**

190 Progress in Gas Turbine Performance


The data used can be taken in steady state or transient operation. The model could be a physical one representing the aerothermodynamic processes taking place in the engine com‐ ponents and the mechanical coupling between them or a black box mathematical model re‐ lating data with health parameters. The diagnostic decision making procedure may be a conventional pattern recognition technique applied to health parameter space or an artificial intelligence based expert system.

Accordingly the proposed methods are classified on the basis of the kind of the comprising elements as: Steady state or Transient, Physical or Mathematical, Conventional or Artificial intelligence method.

#### **3. Physical models based GPA**

#### **3.1. Linear methods**

In linear gas path analysis, the health parameters are represented as the unknown ''deltas'' of component performance parameters (typically efficiency and mass flow capacity). They are related to known measurement ''deltas'' through relations produced by linearization of the general nonlinear thermodynamic relations, assuming small deviations. [6].The classical linear approach is formulated as follows: For a given operating point u the measurement values depend only on the health condition of engine components. After linearization and taking into account measurement uncertainty (by adding a noise vector **v** with zero mean and known covariance R), the typical GPA equations take the form:

$$
\Delta \mathbf{y} = \mathbf{C} \cdot \Delta \mathbf{f} + \mathbf{v} \tag{2}
$$

where Δ is called delta and represents percentage deviation from a reference value(when the engine is in intact condition) and C the well-known influence coefficient matrix. Estimation of health parameters is obtained from the relations

$$
\Delta \hat{\mathbf{f}} = \mathbf{S}^{\mathbf{1}} \cdot \mathbf{C}^{\mathbf{T}} \cdot \mathbf{R}^{-1} \cdot \Delta \mathbf{y} \tag{3}
$$

$$\mathbf{S} = \mathbf{M}^{-1} + \mathbf{C}^{T} \cdot \mathbf{R}^{-1} \cdot \mathbf{C} \tag{4}$$

( ) NOP 1


The so called information matrix P is crucial in the sense that its condition determines the diagnostic effectiveness. The condition of the matrix is represented by its condition number. Investigations concerning effects of both the number of operating points used and the 'dis‐ tance' of the operating points on information matrices have been reported ([14]-[15]). Addi‐ tional details on assessing identifiability in multipoint gas turbine estimation problems are given in [15]. Although all the works implementing the multipoint approach agree that the idea more or less improves the diagnostic effectiveness, there are also results (see [16]), indi‐ cating that the theoretically attainable multi-point improvements are difficult to realize in

In order to understand the reasons for potential problems concerning diagnosis using a mul‐ tipoint approach it is necessary to examine the underlying assumptions of the method. The main assumption of the method is that the 'deltas' concerning the health parameters remain constant with regard to change in operating conditions. This assumption is obviously true for some parameters (for example the parameter expressing the effective turbine area or the area of non-variable nozzle jet engine), but there are indications that for other parameters this is a week assumption. Several works ([3], [17]), have provided evidence that when dete‐ rioration is present, the deviations of parameters such as flow compressor capacity and effi‐ ciency change with the operating point. In fact different working-point means different aerodynamic conditions and, in this sense, efficiencies and flow capacities deltas can signifi‐ cantly vary with the operating condition. The resulting diagnosis risk is not only to impre‐ cisely calculate the engine new state after some deterioration but even more to indicate as

Recently a new variant of GPA method named Artificial Multi Operating Point Analysis (AMOPA) has been proposed [18]. The new method uses existing sensor information pro‐ duced when artificial operating points are used close to an initial operating point by using different parameters for each operating point definition. Therefore the assumption that the 'deltas' of the health parameters remain constant is reasonable. The method proved to be ca‐

In nonlinear methods, the full thermodynamic equations are treated directly without simpli‐ fication. An example of such a method, the method of adaptive modeling introduced by Sta‐ matis et al. [19], uses component maps ''modification factors'' as health parameters and solves for them through an optimization procedure applied to a function based on differen‐ ces of the predicted and measured values. Variants of the nonlinear GPA have been pro‐ posed (see [20-22]), the main differences being the objective function formulation as well as the method used for the optimization. The more general objective function (OF) to be mini‐

pable of both isolating and identifying the fault in individual components.

**P CR C** = ×× å (9)

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

193

*T kk k*

1

=

*k*

practical engine applications.

**3.2. Nonlinear methods**

mized was proposed in ref. [23]:

responsible for the fault the wrong component(s).

where M represents known statistics for the deviation of health parameters.

Although the formulation for classical GPA has proven to be successful for practical purpos‐ es and existing commercial systems ([7, 8]) are based on it, identifiability problems exist due to limited instrumentation. Sufficient engine health assessment requires at least the estima‐ tion of the parameters associated with the main engine components. Considering an existing engine, a typical situation is characterized by the fact that the number of available sensors is smaller than the number of parameters to be calculated. Accordingly, all the initially imple‐ mented methods were compelled to adopt various assumptions. Most of the methods use a priori information about the statistics of the calculated parameters introducing thus bias in the estimation. In that case, inversion of matrix S is only possible when it is dominated by M. The main drawback is the effect discussed by Doel [9]. The algorithm tends to ''smear'' the fault over many components.

#### *3.1.1. Multi operating point GPA*

GPA Multi Operating Point Analysis (MOPA) methods have been developed trying to ex‐ ploit information provided by the existing sensors when different operating points are con‐ sidered. The origin for the multi operating point analysis (MOPA) methods was the Discrete Operating point GPA, introduced in [10].The method, based on information given by exist‐ ing sensors when different operating points are considered, improved significantly the diag‐ nostic effectiveness. The implementation of the method was an extension of the classical linear gas path analysis. MOPA methods though do not use a priori statistics for the param‐ eters rely on the questionable assumption of non-varying health parameters. Other research groups applied the same principle for the nonlinear case, [11-13].

The linear implementation for the MOPA approach using NOP operating points is given by Eqs. (5)-(9).

$$
\Delta \mathbf{y}\_k = \mathbf{C}\_k \cdot \Delta \mathbf{f} \qquad k = 1, \text{NOP} \tag{5}
$$

$$\mathbf{C}\_{k} = \begin{bmatrix} c\_{\langle \boldsymbol{\eta}\_{\langle \boldsymbol{\eta}\_{\boldsymbol{\eta}} \boldsymbol{k} \rangle} \end{bmatrix}} \tag{6}$$

$$\mathcal{L}\_{ij,k} = \left(\partial \Delta y\_i \;/\,\partial \Delta \xi\_j\right)\_k \tag{7}$$

$$\hat{\mathbf{A}}\hat{\mathbf{f}} = \mathbf{P}^{-1} \cdot \sum\_{k=1}^{\mathrm{NOP}} \left( \mathbf{C}\_{k}^{T} \cdot \mathbf{R}\_{k}^{-1} \cdot \Delta \mathbf{y}\_{k} \right) \tag{8}$$

#### Engine Condition Monitoring and Diagnostics http://dx.doi.org/10.5772/54409 193

$$\mathbf{P} = \sum\_{k=1}^{\mathrm{NOP}} \left( \mathbf{C}\_k^T \cdot \mathbf{R}\_k^{-1} \cdot \mathbf{C}\_k \right) \tag{9}$$

The so called information matrix P is crucial in the sense that its condition determines the diagnostic effectiveness. The condition of the matrix is represented by its condition number. Investigations concerning effects of both the number of operating points used and the 'dis‐ tance' of the operating points on information matrices have been reported ([14]-[15]). Addi‐ tional details on assessing identifiability in multipoint gas turbine estimation problems are given in [15]. Although all the works implementing the multipoint approach agree that the idea more or less improves the diagnostic effectiveness, there are also results (see [16]), indi‐ cating that the theoretically attainable multi-point improvements are difficult to realize in practical engine applications.

In order to understand the reasons for potential problems concerning diagnosis using a mul‐ tipoint approach it is necessary to examine the underlying assumptions of the method. The main assumption of the method is that the 'deltas' concerning the health parameters remain constant with regard to change in operating conditions. This assumption is obviously true for some parameters (for example the parameter expressing the effective turbine area or the area of non-variable nozzle jet engine), but there are indications that for other parameters this is a week assumption. Several works ([3], [17]), have provided evidence that when dete‐ rioration is present, the deviations of parameters such as flow compressor capacity and effi‐ ciency change with the operating point. In fact different working-point means different aerodynamic conditions and, in this sense, efficiencies and flow capacities deltas can signifi‐ cantly vary with the operating condition. The resulting diagnosis risk is not only to impre‐ cisely calculate the engine new state after some deterioration but even more to indicate as responsible for the fault the wrong component(s).

Recently a new variant of GPA method named Artificial Multi Operating Point Analysis (AMOPA) has been proposed [18]. The new method uses existing sensor information pro‐ duced when artificial operating points are used close to an initial operating point by using different parameters for each operating point definition. Therefore the assumption that the 'deltas' of the health parameters remain constant is reasonable. The method proved to be ca‐ pable of both isolating and identifying the fault in individual components.

#### **3.2. Nonlinear methods**


where M represents known statistics for the deviation of health parameters.

the fault over many components.

192 Progress in Gas Turbine Performance

*3.1.1. Multi operating point GPA*

Eqs. (5)-(9).

Although the formulation for classical GPA has proven to be successful for practical purpos‐ es and existing commercial systems ([7, 8]) are based on it, identifiability problems exist due to limited instrumentation. Sufficient engine health assessment requires at least the estima‐ tion of the parameters associated with the main engine components. Considering an existing engine, a typical situation is characterized by the fact that the number of available sensors is smaller than the number of parameters to be calculated. Accordingly, all the initially imple‐ mented methods were compelled to adopt various assumptions. Most of the methods use a priori information about the statistics of the calculated parameters introducing thus bias in the estimation. In that case, inversion of matrix S is only possible when it is dominated by M. The main drawback is the effect discussed by Doel [9]. The algorithm tends to ''smear''

GPA Multi Operating Point Analysis (MOPA) methods have been developed trying to ex‐ ploit information provided by the existing sensors when different operating points are con‐ sidered. The origin for the multi operating point analysis (MOPA) methods was the Discrete Operating point GPA, introduced in [10].The method, based on information given by exist‐ ing sensors when different operating points are considered, improved significantly the diag‐ nostic effectiveness. The implementation of the method was an extension of the classical linear gas path analysis. MOPA methods though do not use a priori statistics for the param‐ eters rely on the questionable assumption of non-varying health parameters. Other research

The linear implementation for the MOPA approach using NOP operating points is given by

*k ij k*, é ù *<sup>c</sup>*

( ) NOP 1 1 1 ˆ *<sup>T</sup>*

*k*


*kk k*

1,NOP *k k* **Δy C f** = ×D = *k* (5)

ë û **C =** (6)

*ij k i j* , ( / )*<sup>k</sup> c yf* = ¶D ¶D (7)

**Δf P C R Δy** =× ×× å (8)

groups applied the same principle for the nonlinear case, [11-13].

In nonlinear methods, the full thermodynamic equations are treated directly without simpli‐ fication. An example of such a method, the method of adaptive modeling introduced by Sta‐ matis et al. [19], uses component maps ''modification factors'' as health parameters and solves for them through an optimization procedure applied to a function based on differen‐ ces of the predicted and measured values. Variants of the nonlinear GPA have been pro‐ posed (see [20-22]), the main differences being the objective function formulation as well as the method used for the optimization. The more general objective function (OF) to be mini‐ mized was proposed in ref. [23]:

$$\text{OF} = \sum\_{i=1}^{m} \left[ \frac{y\_i^{\text{calc}} (\mathbf{f}) - y\_i}{y\_i \sigma\_{Y\_i}} \right]^2 + \mathbf{C}\_A \cdot \sum\_{j=1}^{n} \left| \frac{f\_j - f\_j'}{f\_j'' \sigma\_{f\_j}} \right| + \mathbf{C}\_S \cdot \sum\_{j=1}^{n} \left[ \frac{f\_j - f\_j'}{f\_j'' \sigma\_{f\_j}} \right]^2 \tag{10}$$

The methodology for diagnosing single component faults using the above procedure is based on the following reasoning. Since measurement data are noisy, the estimations based on a single data set differ from the actual values due to noise propagation. They can be im‐ proved when more than one measurement data sets are available. In such a case a solution is

thus available. The mean value and standard deviation of the percentage change from refer‐

components that are faulty, with the aid of a parameter, which we call diagnostic index. We define as diagnostic index the ratio of the absolute mean value to the standard deviation for

*j*

Health parameters exhibiting small deviations from reference state or parameters with large standard deviations (large uncertainty on derived estimations) will have small values for di‐ agnostic index. On the other hand, health parameters with large mean value or small stand‐ ard deviation (small uncertainty on derived estimations) will present large values for diagnostic index. It is thus expected that the health parameters, which deviate due to fault occurrence will be those with the largest value of the diagnostic index. Thus we identify as faulty the component containing the parameter with the largest diagnostic index. This stage

After the detection of a faulty component, a more accurate estimation of fault magnitude can be performed. The optimization problem is solved again by keeping as unknowns only the health parameters of the component found faulty. CA, CS are zeroed, to avoid biases im‐ posed by the corresponding terms. (Note that with much fewer unknowns a unique solution can be derived by minimizing differences only from measurements namely the first term of the objective function eq (10).) After performing a series of estimations with this formulation from the available data sets, the average values of the obtained parameters are kept as the

Nonlinear GPA methods have proved accurate and robust provided that appropriate meas‐ ured variables and estimated health parameters have been selected. This is not a trivial

When application of a GPA technique is envisaged on an engine, the existence of certain re‐ strictions is recognized. Considering an existing engine, there is always a given set of availa‐ ble measurements. Addition of instrumentation can be difficult or even impossible. It is therefore important to have the possibility to adopt a convenient formulation of a method, so that an optimal use of the existing measurements is achieved. On the other hand, when a

*f*

s

D

*j*

*DI*

*j*

= (11)

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

*f*

are then calculated. A criterion then is proposed for isolating the parameters of the

becomes

195

obtained for each individual data set. A series of values for each health parameter *f*<sup>j</sup>

ence *Δf<sup>j</sup>*

each estimated health parameter.

is called fault localization.

estimations for the fault magnitude.

problem as explained in the following.

**3.3. Sensors and health parameters selection**

where n and m the dimensionalities of **f** and **y** correspondingly. The first term express the fact that the health parameters under estimation **f** must be such that the values of measured quantities **y** are reproduced as accurately as possible. The second and third terms ensure that the values of health parameters cannot be significant different from their reference, a fact resulting from experience. It is the addition of these terms that allows the derivation of a solution for **f**, even when a smaller number of measurements is available. All *deltas* are weighted by the inverse of the standard deviation of the corresponding quantity. Weight factors *CA, CS* are also included, for the possibility to change the relative importance of the two groups of terms. The reference values **f***<sup>r</sup>* of the health parameters can be chosen to repre‐ sent a 'best' guess of the values to be determined. From studies in estimation theory, it has been found that it is useful to include in the objective function a term of sum of absolute values, since this term may improve the numerical behavior of the estimation procedure by increasing its robustness (see [24]).

The way of determining the vector **f** for minimization of this function can take advantage of the physical characteristics of the problem to be solved. For example the fact that deviation of component efficiencies should not be positive could be formulated as a constraint in the optimization. In the case of slow deterioration tracking, the reference values can be chosen to vary slowly with time while a filtering procedure can be applied, taking advantage of the regular variation of component deviations, as described in [25]. For the case of individual component faults the fault usually affects one or two neighboring components.

All these features should be taken into account when formulating the diagnostic algorithm. The solution is obtained with the interaction of a non-linear engine performance model and an optimisation algorithm, as shown in figure 3.

**Figure 3.** Schematic representation of nonlinear diagnostic procedure

The methodology for diagnosing single component faults using the above procedure is based on the following reasoning. Since measurement data are noisy, the estimations based on a single data set differ from the actual values due to noise propagation. They can be im‐ proved when more than one measurement data sets are available. In such a case a solution is obtained for each individual data set. A series of values for each health parameter *f*<sup>j</sup> becomes thus available. The mean value and standard deviation of the percentage change from refer‐ ence *Δf<sup>j</sup>* are then calculated. A criterion then is proposed for isolating the parameters of the components that are faulty, with the aid of a parameter, which we call diagnostic index. We define as diagnostic index the ratio of the absolute mean value to the standard deviation for each estimated health parameter.

$$DI\_{\dot{f}} = \frac{\boxed{\Delta f\_{\dot{f}}}}{\sigma\_{f\_{\dot{f}}}} \tag{11}$$

Health parameters exhibiting small deviations from reference state or parameters with large standard deviations (large uncertainty on derived estimations) will have small values for di‐ agnostic index. On the other hand, health parameters with large mean value or small stand‐ ard deviation (small uncertainty on derived estimations) will present large values for diagnostic index. It is thus expected that the health parameters, which deviate due to fault occurrence will be those with the largest value of the diagnostic index. Thus we identify as faulty the component containing the parameter with the largest diagnostic index. This stage is called fault localization.

After the detection of a faulty component, a more accurate estimation of fault magnitude can be performed. The optimization problem is solved again by keeping as unknowns only the health parameters of the component found faulty. CA, CS are zeroed, to avoid biases im‐ posed by the corresponding terms. (Note that with much fewer unknowns a unique solution can be derived by minimizing differences only from measurements namely the first term of the objective function eq (10).) After performing a series of estimations with this formulation from the available data sets, the average values of the obtained parameters are kept as the estimations for the fault magnitude.

Nonlinear GPA methods have proved accurate and robust provided that appropriate meas‐ ured variables and estimated health parameters have been selected. This is not a trivial problem as explained in the following.

#### **3.3. Sensors and health parameters selection**

2 2

ss

å åå **<sup>f</sup>** (10)

*i jj*

where n and m the dimensionalities of **f** and **y** correspondingly. The first term express the fact that the health parameters under estimation **f** must be such that the values of measured quantities **y** are reproduced as accurately as possible. The second and third terms ensure that the values of health parameters cannot be significant different from their reference, a fact resulting from experience. It is the addition of these terms that allows the derivation of a solution for **f**, even when a smaller number of measurements is available. All *deltas* are weighted by the inverse of the standard deviation of the corresponding quantity. Weight factors *CA, CS* are also included, for the possibility to change the relative importance of the two groups of terms. The reference values **f***<sup>r</sup>* of the health parameters can be chosen to repre‐ sent a 'best' guess of the values to be determined. From studies in estimation theory, it has been found that it is useful to include in the objective function a term of sum of absolute values, since this term may improve the numerical behavior of the estimation procedure by

The way of determining the vector **f** for minimization of this function can take advantage of the physical characteristics of the problem to be solved. For example the fact that deviation of component efficiencies should not be positive could be formulated as a constraint in the optimization. In the case of slow deterioration tracking, the reference values can be chosen to vary slowly with time while a filtering procedure can be applied, taking advantage of the regular variation of component deviations, as described in [25]. For the case of individual

All these features should be taken into account when formulating the diagnostic algorithm. The solution is obtained with the interaction of a non-linear engine performance model and

component faults the fault usually affects one or two neighboring components.

1 11

*r r m nn calc j j j j i i A S r r i jj i Y j f j f y y f -f f -f C C* = == *y ff*

é ù é ù - ê ú ê ú <sup>=</sup> + × +× ê ú ê ú ë û ê ú ë û

( ) OF

194 Progress in Gas Turbine Performance

increasing its robustness (see [24]).

an optimisation algorithm, as shown in figure 3.

**Figure 3.** Schematic representation of nonlinear diagnostic procedure

s

When application of a GPA technique is envisaged on an engine, the existence of certain re‐ strictions is recognized. Considering an existing engine, there is always a given set of availa‐ ble measurements. Addition of instrumentation can be difficult or even impossible. It is therefore important to have the possibility to adopt a convenient formulation of a method, so that an optimal use of the existing measurements is achieved. On the other hand, when a new engine is designed, or when an intervention to instrument an engine is performed, it is desirable to define an optimum combination of sensors to be installed.

The problems that may be faced in such a situation can be summarized as follows: When a given set of measured quantities is provided, what is the optimum set of health parameters? The particular problem is to define the best possible parameters for a given measurement possibility. This is a problem faced usually by the engine user, who has very few or no pos‐ sibilities of intervening and adding measurements on the engine. When the decision for in‐ strumenting an engine has to be taken, both the manufacturer and the user are faced with the inverse problem: (a) The user wants to know the optimum set of measuring instruments to be added in order to provide enough information for a required level of resolution. (b) The manufacturer wants to decide which instruments will accompany the engine, in order to ensure a good capability of in-service monitoring.

A systematic study for methods of choice of measurements and parameters in a way opti‐ mal as to diagnostic effectiveness was first presented by Stamatis et al. [26]. They introduced criteria for optimal measurement or health parameter selection. We present here the pro‐ posed method for measurement selection. Let **f**(r) be the baseline diagnostic vector corre‐ sponding to a healthy engine (typically **f**(*r*) = **I** ), and **f**(*j*) the diagnostic vector resulting when the *j*th element of / deviates from the baseline value (f(r)) by a percentage amount *h*<sup>j</sup> .

$$\mathbf{f}^{(j)} = \mathbf{f}^{(r)} \cdot + h\_{\mathbf{j}} \cdot \mathbf{e}^{(j)} \quad \mathbf{j} = 1, \ldots, m \tag{12}$$

hj is a small constant (0.001 <*hj* <0.01). Then, from Eq. (3) we have

$$\mathbf{Y}^{(r)} = \mathbf{F}\left(\mathbf{f}^{(r)}\right) \tag{13}$$

So, the problem of selecting the appropriate measurements is expressed mathematically as follows: For a given set of health condition parameters, we must select as measured parame‐

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

197

In later years more works have appeared, approaching the problem from different points of

An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by the way biological nervous systems, such as the brain, process information. The key ele‐ ment of this paradigm is the novel structure of the information processing system. It is com‐ posed of a large number of highly interconnected processing elements (neurons) working together to solve specific problems. ANNs, like people, learn by example. Learning in bio‐ logical systems involves adjustments to the synaptic connections that exist between the neu‐ rons. This is true of ANNs as well. Two tasks which can be performed by neural nets, which are relevant to the procedure of monitoring and diagnostics of a gas turbine, are: modeling

A typical use of a model is to produce reference values for quantities which are monitored. It can be also used for other purposes such as generation of influence coefficient matrices, and sensitivity analyses. ANNs are known to be able to model non-linear systems and there‐ fore can be used for gas turbine performance modeling. A first advantage offered by model‐ ing engine performance through ANN is the much shorter computational time required, once the net is trained and verified, in comparison to any full scale aerothermodynamic model. The latter involves the solution of a set of non-linear equations, which is achieved through iterative schemes, resulting in a number of arithmetic operations significantly larger than those performed by an ANN. A further advantage is related to the possibility of adapt‐ ing to a particular engine, if data is available. A well-known fact is that for a model to be accurately representing the operation of an engine, it has to be adapted to the particular en‐ gine (as discussed, for example, in [19]). A model using ANN provides inherently this possi‐ bility, through the way it is being set up. The existence of a learning phase, (called "training" in the ANN terminology) allows the adaptation to a particular engine, if enough data is

The second area of possible application, detection and identification of faults, comes from one of the most powerful capabilities of ANN, namely the capability of identifying and clas‐ sifying patterns. Any method of fault detection and identification uses a set of changes in the values of some parameters, to detect and identify a component malfunction. The task of assigning such sets of changes to machine status is one very much suited to ANN. Neural networks, with their remarkable ability to derive meaning from complicated or imprecise data can be used to extract patterns and detect trends that are too complex to be noticed by

ters these parameters giving on the norm of Eq. (16) the m greater values.

the performance of a gas turbine and detection and classification of faults.

**4. Artificial intelligence GPA methods**

view, [14, 27].

available.

either humans or other computer techniques.

**4.1. Neural networks**

$$\mathbf{Y}^{(j)} = \mathbf{F}\left(\mathbf{f}^{(j)}\right) = \mathbf{F}\left(\mathbf{f}^{(r)} + h\_{\rangle} \cdot \mathbf{e}^{(j)}\right) \tag{14}$$

The sensitivity of each dependent parameter on each individual health index is evaluated as

$$
\Delta \mathbf{Y}\_k^{(j)} = \left( \mathbf{Y}\_k^{(j)} - \mathbf{Y}\_k^{(r)} \right) / \mathbf{Y}\_k^{(r)} \qquad \qquad k = \mathbf{1}, \ldots, n \tag{15}
$$

We also define an overall sensitivity measure for each parameter with the norm

$$\Delta SY\_k = \left[\frac{1}{m} \cdot \sum\_{j=1}^{m} \left(\Delta Y\_k^{(j)}\right)^2\right]^{1/2} \qquad k = 1, \ldots, n \tag{16}$$

So, the problem of selecting the appropriate measurements is expressed mathematically as follows: For a given set of health condition parameters, we must select as measured parame‐ ters these parameters giving on the norm of Eq. (16) the m greater values.

In later years more works have appeared, approaching the problem from different points of view, [14, 27].

#### **4. Artificial intelligence GPA methods**

#### **4.1. Neural networks**

new engine is designed, or when an intervention to instrument an engine is performed, it is

The problems that may be faced in such a situation can be summarized as follows: When a given set of measured quantities is provided, what is the optimum set of health parameters? The particular problem is to define the best possible parameters for a given measurement possibility. This is a problem faced usually by the engine user, who has very few or no pos‐ sibilities of intervening and adding measurements on the engine. When the decision for in‐ strumenting an engine has to be taken, both the manufacturer and the user are faced with the inverse problem: (a) The user wants to know the optimum set of measuring instruments to be added in order to provide enough information for a required level of resolution. (b) The manufacturer wants to decide which instruments will accompany the engine, in order

A systematic study for methods of choice of measurements and parameters in a way opti‐ mal as to diagnostic effectiveness was first presented by Stamatis et al. [26]. They introduced criteria for optimal measurement or health parameter selection. We present here the pro‐ posed method for measurement selection. Let **f**(r) be the baseline diagnostic vector corre‐

= **I** ), and **f**(*j*)

<0.01). Then, from Eq. (3) we have

*<sup>j</sup>* **f =f e** ×+ × = *h jm* (12)

*<sup>j</sup>* **Y =F f F f e** = +× *h* (14)

( ) ( ) ( ) ( ) ( ) / 1,..., *j jr r Y YY Y k n k kk k* D- = **<sup>=</sup>** (15)

**<sup>=</sup>** å (16)

( ) ( ) ( ) *r r* **Y =F f** (13)

the *j*th element of / deviates from the baseline value (f(r)) by a percentage amount *h*<sup>j</sup>

( ) ( ) ( ) 1,..., *jr j*

( ) ( ) ( ) ( ) ( ) ( ) *jj r j*

We also define an overall sensitivity measure for each parameter with the norm

( ) ( )

*<sup>m</sup> <sup>j</sup>*

é ù

ë û

1

*k k j*

*m* <sup>=</sup>

The sensitivity of each dependent parameter on each individual health index is evaluated as

1/2 2

1 1,...,

*SY Y k n*

ê ú × D <sup>=</sup> ê ú

the diagnostic vector resulting when

.

desirable to define an optimum combination of sensors to be installed.

to ensure a good capability of in-service monitoring.

sponding to a healthy engine (typically **f**(*r*)

hj is a small constant (0.001 <*hj*

196 Progress in Gas Turbine Performance

An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by the way biological nervous systems, such as the brain, process information. The key ele‐ ment of this paradigm is the novel structure of the information processing system. It is com‐ posed of a large number of highly interconnected processing elements (neurons) working together to solve specific problems. ANNs, like people, learn by example. Learning in bio‐ logical systems involves adjustments to the synaptic connections that exist between the neu‐ rons. This is true of ANNs as well. Two tasks which can be performed by neural nets, which are relevant to the procedure of monitoring and diagnostics of a gas turbine, are: modeling the performance of a gas turbine and detection and classification of faults.

A typical use of a model is to produce reference values for quantities which are monitored. It can be also used for other purposes such as generation of influence coefficient matrices, and sensitivity analyses. ANNs are known to be able to model non-linear systems and there‐ fore can be used for gas turbine performance modeling. A first advantage offered by model‐ ing engine performance through ANN is the much shorter computational time required, once the net is trained and verified, in comparison to any full scale aerothermodynamic model. The latter involves the solution of a set of non-linear equations, which is achieved through iterative schemes, resulting in a number of arithmetic operations significantly larger than those performed by an ANN. A further advantage is related to the possibility of adapt‐ ing to a particular engine, if data is available. A well-known fact is that for a model to be accurately representing the operation of an engine, it has to be adapted to the particular en‐ gine (as discussed, for example, in [19]). A model using ANN provides inherently this possi‐ bility, through the way it is being set up. The existence of a learning phase, (called "training" in the ANN terminology) allows the adaptation to a particular engine, if enough data is available.

The second area of possible application, detection and identification of faults, comes from one of the most powerful capabilities of ANN, namely the capability of identifying and clas‐ sifying patterns. Any method of fault detection and identification uses a set of changes in the values of some parameters, to detect and identify a component malfunction. The task of assigning such sets of changes to machine status is one very much suited to ANN. Neural networks, with their remarkable ability to derive meaning from complicated or imprecise data can be used to extract patterns and detect trends that are too complex to be noticed by either humans or other computer techniques.

There are various neural network models. Among all different neural networks, the backpropagation and the probabilistic neural nets are the architectures, which have mostly been investigated for gas turbine diagnostics. The majority of the researchers refer to performance diagnostics [28, 29], while fewer refer to sensor fault detection and isolation. Kanelopoulos et al. [30] studied the performance of back-propagation (BP) neural nets for both sensor and actual engine component faults for a single shaft industrial gas turbine. The BP neural net‐ works, however, have two main limitations: (1) difficulty of determining the network struc‐ ture and the number of nodes; (2) slow convergence of the training process.

( )

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

(18)

199


*)*

2

, and *P(x)* a normalization factor representing the 'a pri‐

Sk

Xm

Xm-1

Xm-2

X3

**. .**

X2

X1

S2

**. . . .**

S1

s

*<sup>i</sup> <sup>n</sup> (i) T i*

<sup>å</sup> **x-x x x**

*j j i*

× ×× ë û

is the j-th pattern of the training set of patterns that 'belong' to class i, |*Si*

the number of the training patterns that 'belong' to class i, *σ<sup>i</sup>* is a smoothing parameter, *P(Si*

ori' probability of pattern *x*, which is constant assuming mutually exclusive classes, covering

é ù - - <sup>=</sup> <sup>×</sup> ê ú

2 1 ( )( ) ( ) ( |) exp <sup>2</sup> ( ) (2 )

*<sup>i</sup> <sup>m</sup> <sup>m</sup> <sup>j</sup> <sup>i</sup> i i*

 s=

*P S*

p

*P S P S*

**x**

**x**

is the 'a priori' probability of class Si

a1

a2

**. . . .**

an

**Figure 4.** The general structure of the Probabilistic Neural Network.

an-1

where, *xj*

(*i*)

all possible situations.

Probabilistic Neural Networks (PNN), exhibit certain advantages that make them attractive, a significant one being that their particular structure does not require a training procedure, needed for other types of neural networks. The training information is produced during the network set-up and is then embedded in its structure. PNN `training' can thus be consid‐ ered to be much faster than for other types of network, such as back-propagation. Addition‐ ally, PNNs perform a probabilistic rather than a deterministic diagnosis, something closer to physical reality.

#### *4.1.1. Probabilistic Neural Networks (PNN)*

The Probabilistic Neural Network (PNN) is a multi-layer feed forward network. The learn‐ ing procedure of this network is a supervised learning procedure. During the learning pro‐ cedure the PNN classifies the training patterns to classes (represented by the output nodes). When an unknown pattern is presented to the PNN, the network estimates the probability that this pattern belongs to each class. The procedure followed and the network itself is briefly described in the following:

Let us suppose that, for training the PNN, we use the group of the m, n-dimensional, train‐ ing patterns:

$$\mathbf{x}\_{\mathbf{j}} = \begin{pmatrix} a\_{1\mathbf{j}\prime} & a\_{2\mathbf{j}\prime} & \dots & a\_{nj} \end{pmatrix} \mathbf{j} = \mathbf{1}, \dots \mathbf{m} \tag{17}$$

The graph of the resulting network is shown in figure 4. The PPN consists of three layers. The n nodes of the first layer represent the n-dimensional input. The m nodes of the second layer (hidden layer) represent the training patterns, while each one of the k nodes of the third (output) layer represents a class to which a pattern can be classified into.

Every node of the input layer of the PPN is linked to every node of the hidden layer. Each node of the hidden layer (representing a training pattern) is linked only to the node of the output layer that represents the class where the training pattern 'belongs'.

When a pattern **x**∈m is given as an input to the network, the output is the probability density functions*: P(Si |* **x***),* i=1,…,k.

If we assume that the probability density functions, P(**x**|Si ), are Gaussian, we have:

#### Engine Condition Monitoring and Diagnostics http://dx.doi.org/10.5772/54409 199

$$P(\mathbf{S}\_i \mid \mathbf{x}) = \frac{P(\mathbf{S}\_i)}{P(\mathbf{x}) \cdot (2\pi)^{\frac{m\mathcal{N}}{2}} \cdot \sigma\_i^m \cdot \left| \mathbb{S}\_i \right|} \cdot \sum\_{j=1}^{n\_i} \exp\left[\frac{-(\mathbf{x} \cdot \mathbf{x}\_j^{(i)})^T (\mathbf{x} - \mathbf{x}\_j^{(i)})}{2\sigma\_i^2} \right] \tag{18}$$

where, *xj* (*i*) is the j-th pattern of the training set of patterns that 'belong' to class i, |*Si* |=n*<sup>i</sup>* is the number of the training patterns that 'belong' to class i, *σ<sup>i</sup>* is a smoothing parameter, *P(Si )* is the 'a priori' probability of class Si , and *P(x)* a normalization factor representing the 'a pri‐ ori' probability of pattern *x*, which is constant assuming mutually exclusive classes, covering all possible situations.

There are various neural network models. Among all different neural networks, the backpropagation and the probabilistic neural nets are the architectures, which have mostly been investigated for gas turbine diagnostics. The majority of the researchers refer to performance diagnostics [28, 29], while fewer refer to sensor fault detection and isolation. Kanelopoulos et al. [30] studied the performance of back-propagation (BP) neural nets for both sensor and actual engine component faults for a single shaft industrial gas turbine. The BP neural net‐ works, however, have two main limitations: (1) difficulty of determining the network struc‐

Probabilistic Neural Networks (PNN), exhibit certain advantages that make them attractive, a significant one being that their particular structure does not require a training procedure, needed for other types of neural networks. The training information is produced during the network set-up and is then embedded in its structure. PNN `training' can thus be consid‐ ered to be much faster than for other types of network, such as back-propagation. Addition‐ ally, PNNs perform a probabilistic rather than a deterministic diagnosis, something closer to

The Probabilistic Neural Network (PNN) is a multi-layer feed forward network. The learn‐ ing procedure of this network is a supervised learning procedure. During the learning pro‐ cedure the PNN classifies the training patterns to classes (represented by the output nodes). When an unknown pattern is presented to the PNN, the network estimates the probability that this pattern belongs to each class. The procedure followed and the network itself is

Let us suppose that, for training the PNN, we use the group of the m, n-dimensional, train‐

The graph of the resulting network is shown in figure 4. The PPN consists of three layers. The n nodes of the first layer represent the n-dimensional input. The m nodes of the second layer (hidden layer) represent the training patterns, while each one of the k nodes of the

Every node of the input layer of the PPN is linked to every node of the hidden layer. Each node of the hidden layer (representing a training pattern) is linked only to the node of the

When a pattern **x**∈m is given as an input to the network, the output is the probability density

third (output) layer represents a class to which a pattern can be classified into.

output layer that represents the class where the training pattern 'belongs'.

If we assume that the probability density functions, P(**x**|Si

**x***j jj* = {*aa a* 1 2 , , .... , j 1, m *nj*} = ¼ (17)

), are Gaussian, we have:

ture and the number of nodes; (2) slow convergence of the training process.

physical reality.

198 Progress in Gas Turbine Performance

ing patterns:

*4.1.1. Probabilistic Neural Networks (PNN)*

briefly described in the following:

functions*: P(Si |* **x***),* i=1,…,k.

**Figure 4.** The general structure of the Probabilistic Neural Network.

For example if it is considered that the 'a priori' probability is equal for all classes,

$$P(S\_i) = \frac{1}{k}, \text{ i } = 1, \dots, \text{k} \tag{19}$$

functions in belief networks theory. An expert system dealing with uncertainty and proved

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

201

BBN is a probabilistic expert system, graphically represented by a set of 'nodes' and a set of 'links' connecting them. The topological features of a BBN that must be fully specified in or‐ der the network to be complete are the following: Nodes express the parameters of the rep‐ resented domain. In figure 5 an example of a belief network referred to a gas turbine is presented. This network has four nodes expressing the parameters of the engine taken into account. These are: the 'efficiency factor of the high pressure compressor' (n(HPC)), the 'effi‐ ciency factor of the high pressure turbine' (n(HPT)), the 'pressure ratio' (πc) and the 'turbine

> Node: n(HPC) Node: n(HPT) States: Normal States: Normal

> Node: π<sup>c</sup> Node: TIT States: Normal States: Normal

Not Normal Not Normal

Not Normal Not Normal

Each node has two or more discrete states, expressing all the different states of the parame‐ ter they refer to. For instance, in the network of figure 5, node TIT has two states: 'normal temperature' and 'not normal temperature'. In each case, the set of states of a node must be exhaustive and mutually exclusive. In other words, any possible condition of a parameter expressed by a node in a BBN is represented by one and only one state of this node. Links among the nodes express the 'rules' of interdependence that hold among them. For example, the link from node n(HPC), on the network of figure 5, to node πc expresses the fact that the condition (state) of node n(HPC) affects directly the condition of node πc. The absence of a link between two nodes doesn't mean that these two nodes are independent, but expresses the fact that the condition (state) of the one doesn't directly affect the condition of the other.

Each node has a Conditional Probability Table (CPT), expressing the probability each state of the node to occur, when the state of each other node, ending up directly to it (called 'pa‐ rent node'), is known. In case that, a node has no other nodes ending up directly to it (called a 'root node'), the CPT of this node express the 'a priori' probability each state of this node to occur. In Table 1 an example of how the CPTs, of the nodes of the network of figure 5,

to be very efficient in fault diagnosis is described below.

*4.2.1. Bayesian Belief Network (BBN)*

inlet temperature' (TIT).

could be, is shown.

**Figure 5.** An example of a belief network of a gas turbine.

During the training of the PNN, we provide the training patterns and the classes they be‐ long to. From this information the number of nodes of each layer, as well as the links of the network with the related weights, are specified.

The weight of the link from node j1 of the input layer to node j2 of the hidden layer is:

$$w\_{/1,/2}^{(1)} = a\_{/1/2} \tag{20}$$

while, the weight of the link from node Xi of the hidden layer to node Sj of the hidden layer, is:

$$w\_{X\_i, S\_j}^{(2)} = \frac{1}{2 \cdot \sigma\_j^2} \tag{21}$$

where, *σ<sup>j</sup>* is the smoothing parameter of class j, represented by node Sj of the output layer of the network. During the testing of the network, the probability density functions for each class are calculated, using equation (18).

Comparative and parametric investigations of the diagnostic ability of PNN on turbofan en‐ gines have been carried out in [31]. The work has also provided some general information about PNN diagnostic ability. The use of probabilistic neural networks for sensor fault de‐ tection and estimation of the sensor bias has been demonstrated in [32]. The technique pro‐ posed was shown to provide a powerful sensor validation tool, for cases where a rather limited number of measuring sensors is available, such as when data from an engine onboard an aircraft are available.

#### **4.2. Expert systems**

In contrast to neural networks, which learn knowledge by training on observed data with known inputs and outputs, Expert systems(ES) utilize domain expert knowledge in a com‐ puter program with an automated inference engine to perform reasoning for problem solv‐ ing. Three main reasoning methods for ES used in the area of engine diagnostics are rulebased reasoning, case-based reasoning and model-based reasoning. In condition monitoring practice, knowledge from domain specific experts is usually inexact and reasoning on knowledge is often imprecise. Therefore, measures of the uncertainties in knowledge and reasoning are required for ES to provide more robust problem solving. Commonly used un‐ certainty measures are probability, fuzzy member functions in fuzzy logic theory and belief functions in belief networks theory. An expert system dealing with uncertainty and proved to be very efficient in fault diagnosis is described below.

#### *4.2.1. Bayesian Belief Network (BBN)*

For example if it is considered that the 'a priori' probability is equal for all classes,

network with the related weights, are specified.

200 Progress in Gas Turbine Performance

while, the weight of the link from node Xi

class are calculated, using equation (18).

board an aircraft are available.

**4.2. Expert systems**

where, *σ<sup>j</sup>*

<sup>1</sup> ( ) , i 1, ,k *<sup>i</sup> P S*

The weight of the link from node j1 of the input layer to node j2 of the hidden layer is:

(2)

*w*

, 2 1

*j*

is the smoothing parameter of class j, represented by node Sj of the output layer of

s

the network. During the testing of the network, the probability density functions for each

Comparative and parametric investigations of the diagnostic ability of PNN on turbofan en‐ gines have been carried out in [31]. The work has also provided some general information about PNN diagnostic ability. The use of probabilistic neural networks for sensor fault de‐ tection and estimation of the sensor bias has been demonstrated in [32]. The technique pro‐ posed was shown to provide a powerful sensor validation tool, for cases where a rather limited number of measuring sensors is available, such as when data from an engine on-

In contrast to neural networks, which learn knowledge by training on observed data with known inputs and outputs, Expert systems(ES) utilize domain expert knowledge in a com‐ puter program with an automated inference engine to perform reasoning for problem solv‐ ing. Three main reasoning methods for ES used in the area of engine diagnostics are rulebased reasoning, case-based reasoning and model-based reasoning. In condition monitoring practice, knowledge from domain specific experts is usually inexact and reasoning on knowledge is often imprecise. Therefore, measures of the uncertainties in knowledge and reasoning are required for ES to provide more robust problem solving. Commonly used un‐ certainty measures are probability, fuzzy member functions in fuzzy logic theory and belief

2 *X Si j*

During the training of the PNN, we provide the training patterns and the classes they be‐ long to. From this information the number of nodes of each layer, as well as the links of the

*<sup>k</sup>* = =¼ (19)

(1) *w a j j jj* 1, 2 1 2 <sup>=</sup> (20)

<sup>=</sup> <sup>×</sup> (21)

of the hidden layer, is:

of the hidden layer to node Sj

BBN is a probabilistic expert system, graphically represented by a set of 'nodes' and a set of 'links' connecting them. The topological features of a BBN that must be fully specified in or‐ der the network to be complete are the following: Nodes express the parameters of the rep‐ resented domain. In figure 5 an example of a belief network referred to a gas turbine is presented. This network has four nodes expressing the parameters of the engine taken into account. These are: the 'efficiency factor of the high pressure compressor' (n(HPC)), the 'effi‐ ciency factor of the high pressure turbine' (n(HPT)), the 'pressure ratio' (πc) and the 'turbine inlet temperature' (TIT).

**Figure 5.** An example of a belief network of a gas turbine.

Each node has two or more discrete states, expressing all the different states of the parame‐ ter they refer to. For instance, in the network of figure 5, node TIT has two states: 'normal temperature' and 'not normal temperature'. In each case, the set of states of a node must be exhaustive and mutually exclusive. In other words, any possible condition of a parameter expressed by a node in a BBN is represented by one and only one state of this node. Links among the nodes express the 'rules' of interdependence that hold among them. For example, the link from node n(HPC), on the network of figure 5, to node πc expresses the fact that the condition (state) of node n(HPC) affects directly the condition of node πc. The absence of a link between two nodes doesn't mean that these two nodes are independent, but expresses the fact that the condition (state) of the one doesn't directly affect the condition of the other.

Each node has a Conditional Probability Table (CPT), expressing the probability each state of the node to occur, when the state of each other node, ending up directly to it (called 'pa‐ rent node'), is known. In case that, a node has no other nodes ending up directly to it (called a 'root node'), the CPT of this node express the 'a priori' probability each state of this node to occur. In Table 1 an example of how the CPTs, of the nodes of the network of figure 5, could be, is shown.


that combines a thermodynamic model of the engine under examination and a BBN, con‐ structed by use of statistical data of the engine. Palmer [34], presented a statistically also

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

203

The first attempt to propose a general procedure of building a BBN for diagnostic purposes, has been presented by Romessis et al. [35]. The objective of the investigation was to reveal a possible way of setting up such a network with aid of an engine performance model. The way of building diagnostic BBNs, allowing implementation into any type of engine, and the disengagement of the BBN from hard to find statistical data, were two elements that made the work interesting and promising. The effectiveness of the proposed diagnostic method was examined on benchmark fault case scenarios, in a typical modern turbofan engine of civil aviation. The diagnosis was based on the observation of fewer measurements (7) than the considered fault parameters (11). Inference with BBN showed that such a network is very reliable, since in the 96% of the cases where a fault was detected, it was detected cor‐

A more efficient method even in fault cases with smaller health parameters' deviations was proposed in [36]. The improvement was due to the way the BBN is constructed: probabilistic relationships among variables are more accurately represented. The effectiveness of the pro‐ posed method has been demonstrated by its strong diagnostic ability with various fault sce‐ narios and cases at several operating conditions, including coverage of an operational

Despite research in various methods for engine fault diagnostics, there is still no method which can effectively address all issues. One way to approach the problem is to try and off‐ set the limitations of one technique with the strength of the other. Hybrid models have at‐

An integrated fault diagnostics model for identifying shifts in component performance and sensor faults using Genetic Algorithm and Artificial Neural Network was presented in [37]. The diagnostics model operates in two distinct stages. The first stage uses response surfaces for computing objective functions to increase the exploration potential of the search space while easing the computational burden. The second stage uses concept of a hybrid diagnostics model in which a nested neural network is used with genetic algorithm to form a hybrid diagnostics model. The nested neural network functions as a pre-processor or filter to reduce the number of fault classes to be explored by the genetic algorithm based diagnostics model. The hybrid mod‐ el improves the accuracy, reliability and consistency of the results obtained. In addition signifi‐ cant improvements in the total run time have also been observed. Ecstase [38], presents an example of the use of fuzzy logic combined with influence coefficients applied to engine testcell data to diagnose gas-path related performance faults. The diagnostic process to identify module level engine performance faults has been validated using eight examples from realworld test-cell data. Many combinations of faults were examined in an attempt to explain the

constructed BBN for fault detection of the CF6 family of engines.

rectly. Only a 4% of the cases were attributed to a wrong fault.

**5. Hybrid and fusion information techniques**

envelope of a typical flight.

tempted to bridge this gap.


Once a BBN is constructed, inference can be realized any time evidence is available. Infer‐ ence is the procedure where the probabilities of each state of each node of the network are updated each time that evidence is available. 'Evidence' is the knowledge of the state of one or more nodes of the network.

Bayesian Belief Networks have some features that make them very attractive in the field of diagnosis of faults in gas turbines. The most important of these features are: BBN allow probabilistic diagnosis; it is more realistic to make diagnosis expressing the belief (probabili‐ ty) of whether an event occurred or not, than expressing a deterministic answer. Mathemati‐ cal relationships among the variables of a network are not required in order to form a BBN. Only the way that these variables affect each other is required. This is very helpful since such mathematical relationships may be unknown. Modern approximate algorithms for in‐ ference with BBN are able, nowadays, to answer queries, once 'evidence' is provided, within few seconds, even for complicated networks, performing with adequate accuracy. Each node of a BBN can be an 'evidence' as well as a 'query'. There is no restriction to the number of 'query' or 'evidence' nodes. Therefore, there is no limitation on how many or which are the 'evidence' nodes in order to estimate the probabilities of all the other nodes of a net‐ work. It allows also the inclusion of information of different nature and from different sour‐ ces for diagnostics.

Such networks have been employed in the field of gas turbine diagnostics by few research‐ ers. Breese et al. [33], presented a method for detecting specific faults on large gas turbines that combines a thermodynamic model of the engine under examination and a BBN, con‐ structed by use of statistical data of the engine. Palmer [34], presented a statistically also constructed BBN for fault detection of the CF6 family of engines.

The first attempt to propose a general procedure of building a BBN for diagnostic purposes, has been presented by Romessis et al. [35]. The objective of the investigation was to reveal a possible way of setting up such a network with aid of an engine performance model. The way of building diagnostic BBNs, allowing implementation into any type of engine, and the disengagement of the BBN from hard to find statistical data, were two elements that made the work interesting and promising. The effectiveness of the proposed diagnostic method was examined on benchmark fault case scenarios, in a typical modern turbofan engine of civil aviation. The diagnosis was based on the observation of fewer measurements (7) than the considered fault parameters (11). Inference with BBN showed that such a network is very reliable, since in the 96% of the cases where a fault was detected, it was detected cor‐ rectly. Only a 4% of the cases were attributed to a wrong fault.

A more efficient method even in fault cases with smaller health parameters' deviations was proposed in [36]. The improvement was due to the way the BBN is constructed: probabilistic relationships among variables are more accurately represented. The effectiveness of the pro‐ posed method has been demonstrated by its strong diagnostic ability with various fault sce‐ narios and cases at several operating conditions, including coverage of an operational envelope of a typical flight.

#### **5. Hybrid and fusion information techniques**

**Table 1.** An example of the CPTs of the nodes of the network of figure 5.

or more nodes of the network.

202 Progress in Gas Turbine Performance

ces for diagnostics.

Once a BBN is constructed, inference can be realized any time evidence is available. Infer‐ ence is the procedure where the probabilities of each state of each node of the network are updated each time that evidence is available. 'Evidence' is the knowledge of the state of one

Bayesian Belief Networks have some features that make them very attractive in the field of diagnosis of faults in gas turbines. The most important of these features are: BBN allow probabilistic diagnosis; it is more realistic to make diagnosis expressing the belief (probabili‐ ty) of whether an event occurred or not, than expressing a deterministic answer. Mathemati‐ cal relationships among the variables of a network are not required in order to form a BBN. Only the way that these variables affect each other is required. This is very helpful since such mathematical relationships may be unknown. Modern approximate algorithms for in‐ ference with BBN are able, nowadays, to answer queries, once 'evidence' is provided, within few seconds, even for complicated networks, performing with adequate accuracy. Each node of a BBN can be an 'evidence' as well as a 'query'. There is no restriction to the number of 'query' or 'evidence' nodes. Therefore, there is no limitation on how many or which are the 'evidence' nodes in order to estimate the probabilities of all the other nodes of a net‐ work. It allows also the inclusion of information of different nature and from different sour‐

Such networks have been employed in the field of gas turbine diagnostics by few research‐ ers. Breese et al. [33], presented a method for detecting specific faults on large gas turbines Despite research in various methods for engine fault diagnostics, there is still no method which can effectively address all issues. One way to approach the problem is to try and off‐ set the limitations of one technique with the strength of the other. Hybrid models have at‐ tempted to bridge this gap.

An integrated fault diagnostics model for identifying shifts in component performance and sensor faults using Genetic Algorithm and Artificial Neural Network was presented in [37]. The diagnostics model operates in two distinct stages. The first stage uses response surfaces for computing objective functions to increase the exploration potential of the search space while easing the computational burden. The second stage uses concept of a hybrid diagnostics model in which a nested neural network is used with genetic algorithm to form a hybrid diagnostics model. The nested neural network functions as a pre-processor or filter to reduce the number of fault classes to be explored by the genetic algorithm based diagnostics model. The hybrid mod‐ el improves the accuracy, reliability and consistency of the results obtained. In addition signifi‐ cant improvements in the total run time have also been observed. Ecstase [38], presents an example of the use of fuzzy logic combined with influence coefficients applied to engine testcell data to diagnose gas-path related performance faults. The diagnostic process to identify module level engine performance faults has been validated using eight examples from realworld test-cell data. Many combinations of faults were examined in an attempt to explain the performance degradation observed in the engine under- going repair. This aspect of the proc‐ ess enabled the status of 17 faults to be determined, despite only five engine parameters being used. The method correctly identified the faults for all except for one fault which had a very small degradation effect on the engine performance.

deriving a final diagnostic decision. Then, in the second step, a new diagnostic problem is formulated and a final set of faulty health parameters are defined with higher confidence. In the proposed method the non-linear gas path analysis is the core diagnostic method, while information provided by vibration measurements trends is used to narrow the domain of unknown health parameters and lead to a well-defined solution. Finally a comprehensive

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

205

Although many diagnostic methods have been proposed and some of them have been tested in real engines only few are known to be incorporated in ECMD integrated systems. An in‐ dustrial monitoring and diagnostic system must comply with several requirements. For

**•** Be as automated as possible and integrated namely performing all actions from data col‐

**•** Have an as wide as possible coverage of detectable faults. Additionally, it should allow additions of other newly discovered faults, which have not been included in the initial

**•** Have prognostic capabilities concerning future maintenance and repair actions. This helps in ensuring that long lead-time spares are available and that outages be minimized.

**•** Derive information with high confidence. In this respect, derivation of the same conclu‐

**•** Employ as few instruments as possible. The instrumentation should be kept as simple as

**•** Be modular and flexible with open circuit architecture in order to be adapted to operator's

**•** Be very user friendly, so that it can be used by non-specialized personnel, while its output

In order to materialize a monitoring system, which possesses these features, the procedures,

**ii.** Data evaluation in order to discard unreliable readings and possibly detect sensor

**iii.** Data processing using appropriate techniques in order to derive diagnostic in‐

**•** Be "robust", namely not very susceptible to noise or faulty input information.

presentation of different fusion possibilities offered is given in [47].

**6. ECMD integrated systems**

such a system to be effective it should:

repertory of the system.

needs.

lection to derivation of diagnostic decisions.

sion by different methods is a very useful feature.

possible and include the minimum number of instruments.

is clear enough to need very little or no interpretation.

which should be implemented, are as follows:

**i.** Measurement data acquisition.

faults.

formation.

A diagnostic method consisting of a combination of Kalman filters and Bayesian Belief Net‐ work (BBN) is presented in [39]. A soft-constrained Kalman filter uses a priori information derived by a BBN at each time step, to derive estimations of the unknown health parame‐ ters. The resulting algorithm has improved identification capability in comparison to the stand-alone Kalman filter. Besides the improvements in accuracy and stability, this kind of method allows information or sensor fusion, which is a very important field of research for future works. The key advantage of combining methods is that it replaces the problem of comparing classification techniques to regression techniques by the problem of choosing which information they can share. Romessis et al. [40] proposed a statistical processing of the diagnostic conclusions provided by a least-square based gas path diagnostic method, in order to improve diagnosis. In a similar attempt (see [41]) a combinatorial approach (statisti‐ cal evaluation of least squares estimations) combined with fuzzy logic rules to calculate fault probabilities. The possibility of creating a mixed fault classification that incorporates both model-based and data driven fault classes was investigated in [42]. Such a classification combines a common diagnosis with a higher diagnostic accuracy for the data-driven classes. The performed analysis has revealed no limitations for realizing a principle of the mixed classification in real monitoring systems.

Information Fusion is the integration of data or information from multiple sources, to achieve improved accuracy and more specific inferences than can be obtained from the use of a single sensor alone. It is generally believed that an ensemble of methods improves diagnostic accura‐ cy when compared to individual methods. In [43] several fusion architectures and classifiers were evaluated. Fusing classifiers that are performing very well had little positive effect. How‐ ever, it was shown that fusing marginal classifiers can increase the diagnostic performance substantially, while reducing their variability. Enhanced fault localization using probabilistic fusion with gas path analysis algorithms is referred in [44],while a fusion technique allowing the merge of conclusions provided by diagnostic methods that act independently for the detec‐ tion of gas turbine faults is described in [45]. The proposed technique adopts the principles of Dempster-Schafer theory for the fusion of two diagnostic methods namely a Bayesian Belief Networks (BBN) and a Probabilistic Neural Networks (PNN). The technique has been applied for the detection of thermodynamic as well as mechanical faults on gas turbines. In all cases, the effectiveness of the proposed fusion technique demonstrated that the merge of diagnostic in‐ formation from different sources leads to better and safer diagnosis.

A fusion method that utilizes performance data and vibration measurements for gas turbine component fault identification is presented in [46]. The proposed method operates during the diagnostic processing of available data (process level) and adopts the principles of cer‐ tainty factors theory. Both performance and vibration measurements are analyzed separate‐ ly, in a first step, and their results are transformed into a common form of probabilities. These forms are interwoven, in order to derive a set of possible faulty components prior to deriving a final diagnostic decision. Then, in the second step, a new diagnostic problem is formulated and a final set of faulty health parameters are defined with higher confidence. In the proposed method the non-linear gas path analysis is the core diagnostic method, while information provided by vibration measurements trends is used to narrow the domain of unknown health parameters and lead to a well-defined solution. Finally a comprehensive presentation of different fusion possibilities offered is given in [47].

#### **6. ECMD integrated systems**

performance degradation observed in the engine under- going repair. This aspect of the proc‐ ess enabled the status of 17 faults to be determined, despite only five engine parameters being used. The method correctly identified the faults for all except for one fault which had a very

A diagnostic method consisting of a combination of Kalman filters and Bayesian Belief Net‐ work (BBN) is presented in [39]. A soft-constrained Kalman filter uses a priori information derived by a BBN at each time step, to derive estimations of the unknown health parame‐ ters. The resulting algorithm has improved identification capability in comparison to the stand-alone Kalman filter. Besides the improvements in accuracy and stability, this kind of method allows information or sensor fusion, which is a very important field of research for future works. The key advantage of combining methods is that it replaces the problem of comparing classification techniques to regression techniques by the problem of choosing which information they can share. Romessis et al. [40] proposed a statistical processing of the diagnostic conclusions provided by a least-square based gas path diagnostic method, in order to improve diagnosis. In a similar attempt (see [41]) a combinatorial approach (statisti‐ cal evaluation of least squares estimations) combined with fuzzy logic rules to calculate fault probabilities. The possibility of creating a mixed fault classification that incorporates both model-based and data driven fault classes was investigated in [42]. Such a classification combines a common diagnosis with a higher diagnostic accuracy for the data-driven classes. The performed analysis has revealed no limitations for realizing a principle of the mixed

Information Fusion is the integration of data or information from multiple sources, to achieve improved accuracy and more specific inferences than can be obtained from the use of a single sensor alone. It is generally believed that an ensemble of methods improves diagnostic accura‐ cy when compared to individual methods. In [43] several fusion architectures and classifiers were evaluated. Fusing classifiers that are performing very well had little positive effect. How‐ ever, it was shown that fusing marginal classifiers can increase the diagnostic performance substantially, while reducing their variability. Enhanced fault localization using probabilistic fusion with gas path analysis algorithms is referred in [44],while a fusion technique allowing the merge of conclusions provided by diagnostic methods that act independently for the detec‐ tion of gas turbine faults is described in [45]. The proposed technique adopts the principles of Dempster-Schafer theory for the fusion of two diagnostic methods namely a Bayesian Belief Networks (BBN) and a Probabilistic Neural Networks (PNN). The technique has been applied for the detection of thermodynamic as well as mechanical faults on gas turbines. In all cases, the effectiveness of the proposed fusion technique demonstrated that the merge of diagnostic in‐

A fusion method that utilizes performance data and vibration measurements for gas turbine component fault identification is presented in [46]. The proposed method operates during the diagnostic processing of available data (process level) and adopts the principles of cer‐ tainty factors theory. Both performance and vibration measurements are analyzed separate‐ ly, in a first step, and their results are transformed into a common form of probabilities. These forms are interwoven, in order to derive a set of possible faulty components prior to

formation from different sources leads to better and safer diagnosis.

small degradation effect on the engine performance.

204 Progress in Gas Turbine Performance

classification in real monitoring systems.

Although many diagnostic methods have been proposed and some of them have been tested in real engines only few are known to be incorporated in ECMD integrated systems. An in‐ dustrial monitoring and diagnostic system must comply with several requirements. For such a system to be effective it should:


In order to materialize a monitoring system, which possesses these features, the procedures, which should be implemented, are as follows:


**iv.** Diagnostic inference in order to decide what is the nature, the location and the se‐ verity of a malfunction present, if any.

tionary blades. An easy and reliable way of identification of the malfunction of the turbine is provided by the method of adaptive modeling. The technique has been applied to test data from this turbine and it gave a clear picture of the problem. Comparison of health parame‐ ters deviation obtained from data from the initial condition of the engine and after the pres‐ ence of the problem was detected is shown in figure 6b. It is clearly shown that the swallowing capacity of both turbines has been significantly reduced, as factor f3 shows a re‐ duction of more than 1.5% and f5 more than 3%. The reduction in f1 (of ~ 0.8%) indicates

(a) (b)

deviation, for a gas turbine, which has suffered severe turbine fouling, caused by fuel additives.

**7. Conclusions**

area of ECMD.

of efficient validation approaches.


**Figure 6.** (a) Display of a user friendly monitoring software for an industrial gas turbine. (b) Health Indices Percentage

In this chapter, we have attempted to present basic principles of the engine condition moni‐ toring and diagnostics (ECMD) subject. It would be impossible to cover in few pages all the aspects of ECMD. Thousands of papers have been published and a vast amount of knowl‐ edge has been accumulated. Even extensive reviews cannot mention all the proposed meth‐ ods. In this respect we presented selective methods representative of three main steps of an ECMD approach, namely data acquisition, data processing and diagnostic decision-making, with emphasis on the last two steps. Few recently developed hybrid, data and method fu‐ sion techniques have also been briefly discussed. The structure of an integrated ECMD sys‐ tem incorporating different diagnostic technics and already in operation is also presented.

The following conclusions are the outcome of over twenty five years of experience in the

**•** The main problems with respect to the industry adoption of advanced technics are the fol‐ lowing: a) lack of data due to no data collection and/or data storage at all; b) lack of effi‐ cient communication between method developers and maintenance practitioners; c) lack

**%**

Initial Condition

Fouled Turbine

df1 df2 df3 df4 df5 df6

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

207

that the compressor has also suffered some deterioration.

**v.** Data management in order to keep historical data records for long term monitor‐ ing, without storing too much unnecessary information.

Such an integrated system and experience gained from its implementation on an operating industrial gas turbine has been presented in [48]. The main functions of this system material‐ izing the procedures mentioned above are as follows:

**Data Acquisition and Management:** Data are acquired from a number of different measur‐ ing instruments, for slowly or fast varying quantities. The obtained measurements are being on-line validated and then organized in a database. The system also gives the possibility to play back measurements database in order to recreate real time operation. Additional fea‐ tures of the developed data acquisition feature are its flexibility and its capability to easily meet the requirements of any particular implementation.

**Performance Analysis:** The acquired thermodynamic measurements are being on-line proc‐ essed using the adaptive modeling method [19]). Thus, at any given operating conditions, the overall engine performances and individual components health indices are being evalu‐ ated. The method can also be used off-line for the analysis of previously recorded data.

**EGT Monitoring:** The hot section, being the most critical area of the engine, is receiving par‐ ticular attention, through exhaust gas temperature profile monitoring ([49]. This monitoring provides indication of possible burner malfunctions or thermocouple faults. Off- line analy‐ sis of historic data stored in measurements database can also be performed.

**Vibration Monitoring:** The means of identifying mechanical faults are provided by this function of the system. For data from vibration sensors, the following diagnostic features are extracted and assessed: a) overall vibration level, b) power spectra (on-line frequency analy‐ sis), and c) spectral signatures [50]). Finally, as the other monitoring modules, it offers the possibility of off-line line analysis of historic data stored in measurements database.

These functions are performed by the system continuously, while the engine is in operation. Their implementation provides adequate diagnostic information about engine condition. This information is being further assessed using a rule based inference engine that provides an engine condition assessment. Thus, the user is being informed in real time about the en‐ gine's condition and performance. The main interface of the system implemented on a PC is shown in figure 6a. It comprises an axial cut-out of the monitored engine and gives the most critical information about the engine condition. The system offers the possibility to perform a more detailed analysis by activating the previously described functions through the but‐ tons on the menu at the upper right hand corner.

An example of system effectiveness in diagnosing is the following. A twin-shaft industrial gas turbine with 21 MW nominal output, used for electricity production in a power station, is considered. The turbine suffered from the formation of deposits on gas generator and power turbine blades, very soon after it was put on operation (see figure 1a). A remedy ac‐ tion taken by the manufacturer was a small re-staggering (opening) of power turbine sta‐ tionary blades. An easy and reliable way of identification of the malfunction of the turbine is provided by the method of adaptive modeling. The technique has been applied to test data from this turbine and it gave a clear picture of the problem. Comparison of health parame‐ ters deviation obtained from data from the initial condition of the engine and after the pres‐ ence of the problem was detected is shown in figure 6b. It is clearly shown that the swallowing capacity of both turbines has been significantly reduced, as factor f3 shows a re‐ duction of more than 1.5% and f5 more than 3%. The reduction in f1 (of ~ 0.8%) indicates that the compressor has also suffered some deterioration.

**Figure 6.** (a) Display of a user friendly monitoring software for an industrial gas turbine. (b) Health Indices Percentage deviation, for a gas turbine, which has suffered severe turbine fouling, caused by fuel additives.

#### **7. Conclusions**

**iv.** Diagnostic inference in order to decide what is the nature, the location and the se‐

**v.** Data management in order to keep historical data records for long term monitor‐

Such an integrated system and experience gained from its implementation on an operating industrial gas turbine has been presented in [48]. The main functions of this system material‐

**Data Acquisition and Management:** Data are acquired from a number of different measur‐ ing instruments, for slowly or fast varying quantities. The obtained measurements are being on-line validated and then organized in a database. The system also gives the possibility to play back measurements database in order to recreate real time operation. Additional fea‐ tures of the developed data acquisition feature are its flexibility and its capability to easily

**Performance Analysis:** The acquired thermodynamic measurements are being on-line proc‐ essed using the adaptive modeling method [19]). Thus, at any given operating conditions, the overall engine performances and individual components health indices are being evalu‐ ated. The method can also be used off-line for the analysis of previously recorded data.

**EGT Monitoring:** The hot section, being the most critical area of the engine, is receiving par‐ ticular attention, through exhaust gas temperature profile monitoring ([49]. This monitoring provides indication of possible burner malfunctions or thermocouple faults. Off- line analy‐

**Vibration Monitoring:** The means of identifying mechanical faults are provided by this function of the system. For data from vibration sensors, the following diagnostic features are extracted and assessed: a) overall vibration level, b) power spectra (on-line frequency analy‐ sis), and c) spectral signatures [50]). Finally, as the other monitoring modules, it offers the

These functions are performed by the system continuously, while the engine is in operation. Their implementation provides adequate diagnostic information about engine condition. This information is being further assessed using a rule based inference engine that provides an engine condition assessment. Thus, the user is being informed in real time about the en‐ gine's condition and performance. The main interface of the system implemented on a PC is shown in figure 6a. It comprises an axial cut-out of the monitored engine and gives the most critical information about the engine condition. The system offers the possibility to perform a more detailed analysis by activating the previously described functions through the but‐

An example of system effectiveness in diagnosing is the following. A twin-shaft industrial gas turbine with 21 MW nominal output, used for electricity production in a power station, is considered. The turbine suffered from the formation of deposits on gas generator and power turbine blades, very soon after it was put on operation (see figure 1a). A remedy ac‐ tion taken by the manufacturer was a small re-staggering (opening) of power turbine sta‐

possibility of off-line line analysis of historic data stored in measurements database.

sis of historic data stored in measurements database can also be performed.

ing, without storing too much unnecessary information.

verity of a malfunction present, if any.

206 Progress in Gas Turbine Performance

izing the procedures mentioned above are as follows:

meet the requirements of any particular implementation.

tons on the menu at the upper right hand corner.

In this chapter, we have attempted to present basic principles of the engine condition moni‐ toring and diagnostics (ECMD) subject. It would be impossible to cover in few pages all the aspects of ECMD. Thousands of papers have been published and a vast amount of knowl‐ edge has been accumulated. Even extensive reviews cannot mention all the proposed meth‐ ods. In this respect we presented selective methods representative of three main steps of an ECMD approach, namely data acquisition, data processing and diagnostic decision-making, with emphasis on the last two steps. Few recently developed hybrid, data and method fu‐ sion techniques have also been briefly discussed. The structure of an integrated ECMD sys‐ tem incorporating different diagnostic technics and already in operation is also presented.

The following conclusions are the outcome of over twenty five years of experience in the area of ECMD.

**•** The main problems with respect to the industry adoption of advanced technics are the fol‐ lowing: a) lack of data due to no data collection and/or data storage at all; b) lack of effi‐ cient communication between method developers and maintenance practitioners; c) lack of efficient validation approaches.

**•** Both physics based and data-driven models show benefits and drawbacks. From the deci‐ sion making point of view both traditional and artificial intelligence techniques are used, although it seems that hybrid approaches are more promising.

[4] Li YG. Performance-analysis-based gas turbine diagnostics: A review. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy.

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

209

[5] Marinai L, Probert D, Singh R. Prospects for aero gas-turbine diagnostics: A review.

[6] Urban LA, Volponi AJ. Mathematical methods of relative engine performance diag‐ nostics. SAE 1992 transactions, vol. 101, journal of aerospace, technical paper 922048.

[7] Barwell MJ. Compass - Ground Based Engine Monitoring Program for General Ap‐

[8] Doel DL. TEMPER - a gas-path analysis tool for commercial jet engines. Journal of

[9] Doel DL. An assessment of weighted-least-squares-based gas path analysis. J Eng

[10] Stamatis A, Papailiou KD. Discrete operating condition gas path analysis. AGARD-CP-448, Engine Condition Monitoring - Technology and Experience. 1988.

[11] Gronstedt TV. A multi point gas path analysis tool for gas turbine engines with a moderate level of instrumentation. 2001.XV ISABE, Bangalore, India, Sept. 3-7. [12] Gulati A, Taylor D, Singh R. Multiple operating point analysis using genetic algo‐ rithm optimization for gas turbine diagnostics. 2001. XV ISABE, Bangalore, India,

[13] Pinelli M, Spina PR, Venturini M. Optimized Operating Point Selection for Gas Tur‐ bine Health State Analysis by Using a Multi-Point Technique. ASME Turbo Expo 2003, collocated with the 2003 International Joint Power Generation Conference

[14] Mathioudakis K, Kamboukos P. Assessment of the effectiveness of gas path diagnos‐ tic schemes. Journal of Engineering for Gas Turbines and Power. 2006;128(1):57-63.

[15] Gronstedt T. Identifiability in multi-point gas turbine parameter estimation prob‐ lems. ASME Turbo Expo 2002: Power for Land, Sea, and Air (GT2002) June 3–6, 2002,

[16] Henriksson M, Borguet S, Leonard O, Gronstedt T. On inverse problems in turbine

[17] Aker GF, Saravanamuttoo HIH. Predicting gas turbine degradation due to compres‐ sor fouling using computer simulation techniques. ASME paper 88-GT-206.

[18] Stamatis AG. Evaluation of gas path analysis methods for gas turbine diagnosis.

[19] Stamatis A, Mathioudakis K, Papailiou KD. Adaptive simulation of gas turbine per‐ formance. Journal of Engineering for Gas Turbines and Power. 1990;112(2):168-75.

Amsterdam, The Netherlands, Paper no. GT2002-30020.

engine parameter estimation.. ASME Paper no. GT2007-27756

Journal of Mechanical Science and Technology. 2011;25(2):469-77.

(GT2003) June 16–19, 2003, Atlanta, Georgia, USA, Paper no. GT2003-38191.

2002;216(5):363-77.

Sept. 3-7

Applied Energy. 2004;79(1):109-26.

plication. 1987. SAE technical paper 871734.

Engineering for Gas Turbines and Power. 1994;116(1):82-9.

Gas Turbines Power, Trans ASME. 1994;116:365-73.


The following research directions are required for the next generation of ECMD systems: Enhancement of ECMD systems to collect accurate information, especially fault event infor‐ mation. This information would be very useful for model building and model validation as well. Advanced models and methods for utilization of the transient data diagnostic informa‐ tion as well as detailed higher order models for deterioration mechanisms and faults reliable simulation should be developed. Accurate prognostic models development is also necessa‐ ry. Finally, there is a need for establishment of efficient validation approaches through benchmark test cases to compare the merits and the drawbacks of different modeling and algorithmic approaches.

#### **Author details**

Anastassios G. Stamatis

Address all correspondence to: tastamat@uth.gr

Mechanical Engineering Department, Polytechnic School, University of Thessaly, Volos, Greece

#### **References**


[4] Li YG. Performance-analysis-based gas turbine diagnostics: A review. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy. 2002;216(5):363-77.

**•** Both physics based and data-driven models show benefits and drawbacks. From the deci‐ sion making point of view both traditional and artificial intelligence techniques are used,

**•** The value of vibration monitoring and other sources data in refining gas-path monitoring results has been recognized. The approach of combining different monitoring results, i.e.,

**•** Usage and life monitoring for fatigue critical or life-limited parts are increasingly impor‐

**•** Collaboration of ECMD research groups is necessary in order to produce integrated plat‐ forms for enhancing an ECMD system since each research group has its own specialty

The following research directions are required for the next generation of ECMD systems: Enhancement of ECMD systems to collect accurate information, especially fault event infor‐ mation. This information would be very useful for model building and model validation as well. Advanced models and methods for utilization of the transient data diagnostic informa‐ tion as well as detailed higher order models for deterioration mechanisms and faults reliable simulation should be developed. Accurate prognostic models development is also necessa‐ ry. Finally, there is a need for establishment of efficient validation approaches through benchmark test cases to compare the merits and the drawbacks of different modeling and

Mechanical Engineering Department, Polytechnic School, University of Thessaly, Volos,

[1] Zaita AV, Buley G, Karlsons G. Performance deterioration modeling in aircraft gas turbine engines. Journal of Engineering for Gas Turbines and Power. 1998;120(2):

[2] Kurz R, Brun K. Degradation in gas turbine systems. Journal of Engineering for Gas

[3] Diakunchak IS. Performance Deterioration in Industrial Gas-Turbines. Journal of En‐

although it seems that hybrid approaches are more promising.

data fusion, is becoming an active area of research.

tant

and focus in the area.

208 Progress in Gas Turbine Performance

algorithmic approaches.

Anastassios G. Stamatis

Address all correspondence to: tastamat@uth.gr

Turbines and Power. 2001;123(1):70-7.

gineering for Gas Turbines and Power. 1992;114(2):161-8.

**Author details**

Greece

**References**

344-9.


[20] Li YG, Korakianitis T. Nonlinear weighted-least-squares estimation approach for gas-turbine diagnostic applications. Journal of Propulsion and Power. 2011;27(2): 337-45.

[34] Palmer CA. Combining Bayesian belief networks with gas path analysis for test cell

Engine Condition Monitoring and Diagnostics

http://dx.doi.org/10.5772/54409

211

[35] Romessis A, Stamatis A, Mathioudakis K. Setting up a belief network for turbofan di‐

[36] Romessis C, Mathioudakis K. Bayesian network approach for gas path fault diagno‐ sis. Journal of Engineering for Gas Turbines and Power. 2006;128(1):64-72.

[37] Sampath S, Singh R. An integrated fault diagnostics model using genetic algorithm and neural networks. Journal of Engineering for Gas Turbines and Power,2006;128,

[38] Eustace RW. A real-world application of fuzzy logic and influence coefficients for gas turbine performance diagnostics. Journal of Engineering for Gas Turbines and Pow‐

[39] Dewallef P, Romessis C, Leonard O, Mathioudakis K. Combining classification tech‐ niques with Kalman filters for aircraft engine diagnostics. Journal of Engineering for

[40] Romessis C, Kamboukos P, Mathioudakis K. The use of probabilistic reasoning to im‐ prove least squares based gas path diagnostics. Journal of Engineering for Gas Tur‐

[41] Lipowsky H, Staudacher S, Nagy D, Bauer M. Gas turbine fault diagnostics using a fusion of least squares estimations and fuzzy logic rules. ASME Turbo Expo 2008: Power for Land, Sea, and Air (GT2008)June 9–13, 2008, Berlin, Germany. ASME Pa‐

[42] Loboda I, Yepifanov S. A mixed data-driven and model based fault classification for gas turbine diagnosis. Proceedings of ASME Turbo Expo 2010: International Techni‐ cal Congress, 8p., Scotland, UK, June 14-18, Glasgow, ASME Paper No.

[43] Donat W, Choi K, An W, Singh S, Pattipati K. Data visualization, data reduction and classifier fusion for intelligent fault diagnosis in gas turbine engines. Journal of Engi‐

[44] Kyriazis A, Mathioudakis K. Enhanced fault localization using probabilistic fusion with gas path analysis algorithms. Journal of Engineering for Gas Turbines and Pow‐

[45] Romessis C, Kyriazis A, Mathioudakis K. Fusion of gas turbines diagnostic inference - The dempster-schafer approach. Proceedings of IGTI/ASME Turbo Expo 2007, 9p.,

[46] Kyriazis A, Tsalavoutas A, Mathioudakis K, Bauer M, Johanssen O. Gas turbine fault identification by fusing vibration trending and gas path analysis. ASME Turbo Expo

Canada, May 14-17, 2007, Montreal, ASME Paper GT2007-27043.

agnosis with the aid of an engine performance model. ISABE-2001-1032.

diagnostics and overhaul. ASME paper 98-GT-168.

Gas Turbines and Power. 2006;128(2):281-7.

neering for Gas Turbines and Power. 2008;130(4).

bines and Power. 2007;129(4):970-6.

per no. GT2008-50190

GT2010-23075.

er. 2009;131(5).

(1):49-56.

er. 2008;130(6).


[34] Palmer CA. Combining Bayesian belief networks with gas path analysis for test cell diagnostics and overhaul. ASME paper 98-GT-168.

[20] Li YG, Korakianitis T. Nonlinear weighted-least-squares estimation approach for gas-turbine diagnostic applications. Journal of Propulsion and Power. 2011;27(2):

[21] Ogaji S, Sampath S, Singh R, Probert D. Novel approach for improving power-plant availability using advanced engine diagnostics. Applied Energy. 2002;72(1):389-407.

[22] Zedda M, Singh R. Gas turbine engine and sensor fault diagnosis using optimization

[23] Mathioudakis K, Kamboukos P, Stamatis A. Gas turbine component fault detection from a limited number of measurements. Proceedings of the Institution of Mechani‐

[24] Grodent M, Navez A. Engine Physical Diagnosis Using a Robust Parameter Estima‐ tion Method., 37th AIAA/ASME /SAE/ASEE Joint Propulsion Conference and Exhib‐

[25] Mathioudakis K, Kamboukos P, Stamatis A. Turbofan performance deterioration tracking using nonlinear models and optimization techniques. Journal of Turboma‐

[26] Stamatis A, Mathioudakis K, Papailiou K. Optimal measurement and health index selection for gas turbine performance status and fault diagnosis. Journal of Engineer‐

[27] Ogaji SOT, Sampath S, Singh R, Probert SD. Parameter selection for diagnosing a gas-

[28] Eustace R, Merrington G. Fault diagnosis of fleet engines using neural networks. Fault Diagnosis of Fleet Engines Using Neural Networks. ISABE paper, ISABE

[29] Volponi AJ, DePold H, Ganguli R, Daguang C. The use of kalman filter and neural network methodologies in gas turbine performance diagnostics: A comparative

study. Journal of Engineering for Gas Turbines and Power. 2003;125(4):917-24.

gas turbine performance diagnostics. ASME paper 97-GT-35.

Gas Turbines and Power. 2003;125(3):634-41.

performance in gas turbines. ASME paper 92-GT-399.

[30] Kanelopoulos K, Stamatis A, Mathioudakis K. Incorporating neural networks into

[31] Romessis C, Stamatis A, Mathioudakis K. A parametric investigation of the diagnos‐ tic ability of probabilistic neural networks on turbofan engines. ASME paper 2001-

[32] Romesis C, Mathioudakis K. Setting up of a probabilistic neural network for sensor fault detection including operation with component faults. Journal of Engineering for

[33] Breese JS, Gay R, Quentin GH. Automated decision-analytic diagnosis of thermal

turbine's performance-deterioration. Applied Energy. 2002;73(1):25-46.

techniques. Journal of Propulsion and Power. 2002;18(5):1019-25.

it, 8-11 July 2001, Salt Lake City, Utah, paper AIAA-2001-3768.

ing for Gas Turbines and Power. 1992;114(2):209-16.

cal Engineers, Part A: Journal of Power and Energy. 2004;218(8):609-18.

337-45.

210 Progress in Gas Turbine Performance

chinery. 2002;124(4):580-7.

95-7085.

GT-0011.


2009: Power for Land, Sea, and Air (GT2009),June 8–12, 2009, Orlando, Florida, USA, ASME Paper no. GT2009-59942

**Chapter 9**

**System Safety of Gas Turbines: Hierarchical Fuzzy**

Reliability, safety and durability represent important properties of modern aircraft, which is

The reason of the main hazard for aircraft are both random and determined negative influ‐ ences rendering the controlled object during its use. Faults, failures, disturbances, noises, in‐ fluences of environment and control errors represent the objectively existing stream of

Statistically, in the recent years the majority of aircraft incidents are connected with the hu‐ man factor and late fault detection in plane systems. In this regard, requirements to flight safety which demand development of new methods and algorithms of control-and-condi‐ tion monitoring/ diagnostic for complex objects raise every year. The analysis of modern gas turbine engines has shown that most faults appears in the engine itself and its FADEC

The percentage of faults for FADEC depends on the achieved values for no-failure operation

During the development of FADEC, it is necessary to adhere to the principles and methods guaranteeing safety and reliability of aircraft in use to guarantee proper responses in all

Full information on its work is necessary for complete control of a condition of the engine:

**1.** Reliable detection of a fault cause providing decision-making on a technical condition

© 2013 Kulikov et al.; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

© 2013 Kulikov et al.; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

G. G. Kulikov, V. Yu. Arkov and A.I. Abdulnagimov

Additional information is available at the end of the chapter

**Markov Modelling**

http://dx.doi.org/10.5772/54443

necessary for its effective in-service use.

random negative influences on the object.

(40-75% for FADEC, Figure 1).

range of negative influences.

of gas turbines;

indicators of the engine and FADEC.

**1. Introduction**

