**6. Uniflow vs. loop scavenging**

In literature it is quite hard to find an objective comparison between a uniflow and a loop design under real operating conditions, since it is very expensive and time-consuming to build two different prototypes complying with the same constraints and targets, and devel‐ oped with the same degree of technical sophistication.

As an example, in [19] a comparison was presented between a uniflow and a loop design hav‐ ing the same bore (80 mm) and compression ratio (19). Unfortunately, the uniflow engine fea‐ tured an external blower and a stroke/bore ratio of 1.23, while the loop design was characterized by crankcase scavenging and bore/stroke ratio 0.875. In such a different condi‐ tions, the outcome of the study, i.e. the superiority of the uniflow design, is quite questionable.

In a theoretical study presented in [13] and [14], a uniflow and a loop design were devel‐ oped on the same starting base, a commercial aircraft engine, named WAM 100, whose fea‐ tures are listed in table 2.

The new designs were developed trying to maintain as much as possible of the original en‐ gine: therefore, bore, stroke, number of cylinders, air metering system, et cetera are the same, while the rated power is set at a higher value (150 HP), thanks to the introduction of a specifically developed combustion system featuring direct injection and a Common Rail sys‐ tem. Since the Uniflow scavenging was already optimized in the original engine, most of the attention was paid to the loop version. Here, a ports design as the one visible in figure 3 was adopted, and optimized via CFD-3D simulations.


**Table 2.** Main features of the WAM 100 engine, assumed as a starting base for the CFD study presented in [13] and [14].

**Figure 5.** Computational mesh of a Uniflow design analyzed by Hori [18], showing the layout of the inlet ports

In literature it is quite hard to find an objective comparison between a uniflow and a loop design under real operating conditions, since it is very expensive and time-consuming to build two different prototypes complying with the same constraints and targets, and devel‐

As an example, in [19] a comparison was presented between a uniflow and a loop design hav‐ ing the same bore (80 mm) and compression ratio (19). Unfortunately, the uniflow engine fea‐ tured an external blower and a stroke/bore ratio of 1.23, while the loop design was characterized by crankcase scavenging and bore/stroke ratio 0.875. In such a different condi‐ tions, the outcome of the study, i.e. the superiority of the uniflow design, is quite questionable.

In a theoretical study presented in [13] and [14], a uniflow and a loop design were devel‐ oped on the same starting base, a commercial aircraft engine, named WAM 100, whose fea‐

The new designs were developed trying to maintain as much as possible of the original en‐ gine: therefore, bore, stroke, number of cylinders, air metering system, et cetera are the same, while the rated power is set at a higher value (150 HP), thanks to the introduction of a specifically developed combustion system featuring direct injection and a Common Rail sys‐ tem. Since the Uniflow scavenging was already optimized in the original engine, most of the

**6. Uniflow vs. loop scavenging**

160 Advances in Internal Combustion Engines and Fuel Technologies

tures are listed in table 2.

oped with the same degree of technical sophistication.

A comparison between the scavenging parameters calculated under real engine operating conditions (2000, 2500 and 3000 rpm, full load) is presented in figure 6. Figure 7 presents a pictorial view of the fresh charge concentration on a plane passing through the cylinder axis, at different crank angle. Engine speed is 2500 rpm, full load.

The scavenging parameters are defined as follows. The Trapping Efficiency (TE) is the ratio of the mass of fresh air retained within the cylinder to the mass of fresh air delivered; the Scavenging Efficiency (SE) is the ratio of the mass of fresh charge retained to the total cylin‐ der mass (fresh+exhaust); the Exhaust Gas Purity is the mass fraction of fresh charge in the exhaust flow leaving the cylinder; finally, the reference mass is calculated considering the average delivery density and the total displaced volume.

Analyzing figures 6 and 7, it is observed that operating conditions affect Uniflow scaveng‐ ing very slightly, while the influence is more evident on Loop. It should be considered that these conditions are defined not only by speed, but also by the pressure traces forced at both the inlet and the outlet boundaries, which are obviously different from case to case for rep‐ resenting real engine operations. The lower data scattering of the Uniflow design may be mainly explained by the more regular pressure traces.

It is also important to notice that the scavenging parameters under real operating condi‐ tions can be quite different from the ones expected when performing a steady characteri‐ zation. First of all, the mass flow rates entering and leaving the cylinder are all but constant (in a properly tuned exhaust system, the flow must change its direction after transfer port closing, to reduce the loss of fresh charge); furthermore, the density of the charge entering the cylinder changes along the cycle, as well as the pressure drop across the ports. Among the dynamic effects, a very good help can be found in the 3-cylinder lay-out. In fact, the pressure trace in the exhaust manifold is made up of a sequence of three pulses, one for each combustion, distributed at a distance of 120°: therefore, before exhaust port/valve closing, the cylinder outflow is blocked by the pulse generated by a neighboring cylinder. This is particularly helpful in the Loop engine, when there is a long delay between exhaust and inlet port closing.

served in the Loop case. This difference is expected to have a big influence on combustion: while the swirl angular momentum decays very slowly, supporting turbulence around TDC and later, the tumble vortex is destroyed well before the start of combustion. Furthermore, the turbulent kinetic energy field at TDC depends more on the momentum transferred from fuel injection than on the in-cylinder flow patterns. Therefore, combustion in loop scav‐ enged engines is much less sensitive to the scavenging patterns, and it must be optimized

Advances in The Design of Two-Stroke, High Speed, Compression Ignition Engines

http://dx.doi.org/10.5772/54204

163

**LOOP SCAVENGING UNIFLOW SCAVENGING** 

**Figure 6:** Trapping efficiency, Purity and Scavenging efficiency plotted as a function of Delivery Ratio. Values calculated at three different operating conditions on the optimized Loop and Uniflow configurations (engine speed: 2000, 2500 and 3000 rev/'), reference [14]

**Figure 6.** Trapping efficiency, Purity and Scavenging efficiency plotted as a function of Delivery Ratio. Values calculat‐ ed at three different operating conditions on the optimized Loop and Uniflow configurations (engine speed: 2000,

according to new concepts.

2500 and 3000 rev/'), reference [14]

As generally expected, scavenging is more efficient in the Uniflow design: here, the proc‐ ess can be approximated to a perfect displacement for a DR up to 0.6; after that, some fresh charge leaves the cylinder mixed with exhaust (see the purity graph). This mixing occurs when the stream of fresh charge climbing along the liner wall reaches the cylin‐ der top, as typical for uniflow engines: figure 8 shows this process clearly. For values of DR higher than 0.6 some air is lost through the valves, but TE remains very high be‐ cause of the charge stratification within the cylinder: the air concentration in the head re‐ gion is always lower than in the other parts of the cylinder. As a result, at the maximum values of DR (1.1), TE is well beyond 80%. The SE graph of figure 6 indicates that, even at the maximum DR of 1.1, about a 20% of residuals remains within the cylinder. The presence of swirl affects SE, increasing the mixing between fresh charge and residuals in the cylinder bulk volume. This negative effect can be balanced by a higher degree of boost, which reduces the amount of burned gas by increasing DR.

Loop scavenging is reasonably good: the flow patterns remains very close to those of a per‐ fect displacement up to a DR of 0.5, while for higher delivery rates the situation is inter‐ mediate between a perfect mixing and a perfect scavenging. TE plots are consistent with Purity trends: the drop of retaining capability corresponds to the presence of fresh charge in the exhaust outflow. Scavenging features seem to improve a little bit as engine speed de‐ creases, but this effect may not be related only to speed, as already discussed. An advantage of Loop on Uniflow can be observed in the SE plots: Loop seems to better sweep the residu‐ als from the cylinder, at any DR value. This effect is ascribed to the lower permeability of Uniflow, in particular of exhaust valves in comparison to ports: it is well known that a pis‐ ton controlled port yields a much larger average flow area than a valve of about similar di‐ mensions. As a consequence, in the Uniflow cylinder the residuals leaves at a slower pace, and a larger quantity remains trapped for each DR. However, the better scavenging efficien‐ cy of Loop in comparison to Uniflow is not a general result, but it strictly depends on the specific geometric details and on the valve actuation law.

From the pictorial view of figure 8, it may be observed how different is the in-cylinder flow field between Uniflow and Loop, after BDC. While in Uniflow the swirl ratio can be adjust‐ ed varying the tangential inclination of the transfer ports, a strong tumble is always ob‐ served in the Loop case. This difference is expected to have a big influence on combustion: while the swirl angular momentum decays very slowly, supporting turbulence around TDC and later, the tumble vortex is destroyed well before the start of combustion. Furthermore, the turbulent kinetic energy field at TDC depends more on the momentum transferred from fuel injection than on the in-cylinder flow patterns. Therefore, combustion in loop scav‐ enged engines is much less sensitive to the scavenging patterns, and it must be optimized according to new concepts.

It is also important to notice that the scavenging parameters under real operating condi‐ tions can be quite different from the ones expected when performing a steady characteri‐ zation. First of all, the mass flow rates entering and leaving the cylinder are all but constant (in a properly tuned exhaust system, the flow must change its direction after transfer port closing, to reduce the loss of fresh charge); furthermore, the density of the charge entering the cylinder changes along the cycle, as well as the pressure drop across the ports. Among the dynamic effects, a very good help can be found in the 3-cylinder lay-out. In fact, the pressure trace in the exhaust manifold is made up of a sequence of three pulses, one for each combustion, distributed at a distance of 120°: therefore, before exhaust port/valve closing, the cylinder outflow is blocked by the pulse generated by a neighboring cylinder. This is particularly helpful in the Loop engine, when there is a

As generally expected, scavenging is more efficient in the Uniflow design: here, the proc‐ ess can be approximated to a perfect displacement for a DR up to 0.6; after that, some fresh charge leaves the cylinder mixed with exhaust (see the purity graph). This mixing occurs when the stream of fresh charge climbing along the liner wall reaches the cylin‐ der top, as typical for uniflow engines: figure 8 shows this process clearly. For values of DR higher than 0.6 some air is lost through the valves, but TE remains very high be‐ cause of the charge stratification within the cylinder: the air concentration in the head re‐ gion is always lower than in the other parts of the cylinder. As a result, at the maximum values of DR (1.1), TE is well beyond 80%. The SE graph of figure 6 indicates that, even at the maximum DR of 1.1, about a 20% of residuals remains within the cylinder. The presence of swirl affects SE, increasing the mixing between fresh charge and residuals in the cylinder bulk volume. This negative effect can be balanced by a higher degree of

Loop scavenging is reasonably good: the flow patterns remains very close to those of a per‐ fect displacement up to a DR of 0.5, while for higher delivery rates the situation is inter‐ mediate between a perfect mixing and a perfect scavenging. TE plots are consistent with Purity trends: the drop of retaining capability corresponds to the presence of fresh charge in the exhaust outflow. Scavenging features seem to improve a little bit as engine speed de‐ creases, but this effect may not be related only to speed, as already discussed. An advantage of Loop on Uniflow can be observed in the SE plots: Loop seems to better sweep the residu‐ als from the cylinder, at any DR value. This effect is ascribed to the lower permeability of Uniflow, in particular of exhaust valves in comparison to ports: it is well known that a pis‐ ton controlled port yields a much larger average flow area than a valve of about similar di‐ mensions. As a consequence, in the Uniflow cylinder the residuals leaves at a slower pace, and a larger quantity remains trapped for each DR. However, the better scavenging efficien‐ cy of Loop in comparison to Uniflow is not a general result, but it strictly depends on the

From the pictorial view of figure 8, it may be observed how different is the in-cylinder flow field between Uniflow and Loop, after BDC. While in Uniflow the swirl ratio can be adjust‐ ed varying the tangential inclination of the transfer ports, a strong tumble is always ob‐

long delay between exhaust and inlet port closing.

162 Advances in Internal Combustion Engines and Fuel Technologies

boost, which reduces the amount of burned gas by increasing DR.

specific geometric details and on the valve actuation law.

**Figure 6:** Trapping efficiency, Purity and Scavenging efficiency plotted as a function of Delivery Ratio. Values calculated at three different operating conditions on the optimized Loop and Uniflow configurations (engine speed: 2000, 2500 and 3000 rev/'), reference [14] **Figure 6.** Trapping efficiency, Purity and Scavenging efficiency plotted as a function of Delivery Ratio. Values calculat‐ ed at three different operating conditions on the optimized Loop and Uniflow configurations (engine speed: 2000, 2500 and 3000 rev/'), reference [14]

**Loop Scavenging Uniflow Scavenging** 

**TPO +10 CAD**

**TPO + 20 CAD**

**BDC**

**TPC -10**

**EVC -5**

**Figure 8.** Velocity vectors plotted on a plane passing through the cylinder axis at different crank angles. Comparison

Advances in The Design of Two-Stroke, High Speed, Compression Ignition Engines

http://dx.doi.org/10.5772/54204

165

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**TPO +10 CAD**

**TPO + 20 CAD**

**BDC**

**TPC -10**

**EPC -5**

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**Velocity [cm/s]** 

**Velocity [cm/s]** 

between Loop and Uniflow designs at 2500 rpm, full load [14]

**Velocity [cm/s]** 

**Figure 7.** Fresh charge concentration plotted on a plane passing through the cylinder axis at different crank angles. Comparison between Loop and Uniflow at 2500 rpm, full load [14]

**Loop Scavenging Uniflow Scavenging** 

**TPO +10 CAD**

**TPO + 20 CAD**

**BDC**

**TPC -10**

**EVC -5**

**Figure 7.** Fresh charge concentration plotted on a plane passing through the cylinder axis at different crank angles.

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**TPO +10 CAD**

**TPO + 20 CAD**

**BDC**

**TPC -10**

**EPC -5**

164 Advances in Internal Combustion Engines and Fuel Technologies

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

**Fresh Charge Concentration** 

Comparison between Loop and Uniflow at 2500 rpm, full load [14]

**Fresh Charge Concentration** 

**Figure 8.** Velocity vectors plotted on a plane passing through the cylinder axis at different crank angles. Comparison between Loop and Uniflow designs at 2500 rpm, full load [14]
