**4. Experimental part**

#### **4.1 Oil characterization**

Oil characterization before proceeding with the standardization of the raw material is a very important issue. Some properties remain in fact unchanged from the starting material to the finished biodiesel, or they are anyway predetermined. It is so important to check that the values of such chemical and physical oil properties are in range with those required by the standard regulations (see Table 3).

The experimental procedures to get the values of such properties are also standardized and are indicated in the regulations. The following are parameters for starting oil that can affect the quality of the final biodiesel.

Sulfur and Phosphorous content:

High sulphur fuels cause greater engine wear and in particular shorten the life of the catalyst. Biodiesel derived from soybean oil, as well as from pure rapeseed oil, is known to contain virtually no sulphur (Radich, 2004; Zhiyuan et al., 2008).

The phosphorus content of the vegetable oil depends mainly on the grade of refined oil and arises mainly from phospholipids within the starting material. Measurement of the SO2 from sulphur is accomplished by ultraviolet fluorescence (ASTM D5453, 2002), whereas the analytical method to determine phosphorous requires an Inductively Coupled Plasma Atomic Emission Spectrometry (ASTM International, 2002).

Linoleic acid methyl ester and polyunsaturated methyl esters

Soy, sunflower, cottonseed and maize oils contain a high proportion of linoleic fatty acids, so affecting the properties of the derived ester with a low melting point and cetane number. Quantitative determination of linoleic acid methyl ester is accomplished by gas chromatography with the use of an internal standard after the substrate has been transesterificated and allows also the quantification of the other acid methyl esters (Environment Australia, 2003).

A typical fatty acid methyl esters composition of soybean oil and other feedstock oils is given in Tab. 1, paragraph 2.

Iodine Value

328 Recent Trends for Enhancing the Diversity and Quality of Soybean Products

diameter (Ǻ) 300 240 230 170 235 220 n.d.

Acidity (meq H+/g) 4.7 5.4 5.0 2.2 0.43 2.55 1.00

temperature (K) 393 423 403 413 393 463 403

The main advantage represented by the use of these catalysts lies in the possibility of adopting very mild reaction conditions. In particular, working at temperatures lower than the methanol boiling point (64,7°C), FFA esterification can be performed without overpressure. In this way no expensive and complex plants are required, making this technology adaptable also for little biodiesel manufacturers. Another interesting aspect of these catalysts is their small deactivation even after long operating periods. In fact, if no particular critical conditions are present in the system during the process (e.g. mechanical fragmentation of the catalyst (Pirola et al., 2010) or presence of metallic ions as Fe3+ in the starting TG (Tesser et al., 2010)), no remarkable diminution of the catalytic performance is

Different types of reactors exploiting these ion-exchange resins have been proposed for FFA esterification (Santacesaria et al., 2007; Pirola et al., 2010). The most studied system is a

The main drawback of the slurry system lies in the fragmentation of the catalyst's particles due to their collision one against the other and against the inner reactor's walls (Pirola et al.,

Alternatives to the slurry reactor are the PFR (Plug Flow) reactor, the Carberry-type reactor,

Oil characterization before proceeding with the standardization of the raw material is a very important issue. Some properties remain in fact unchanged from the starting material to the finished biodiesel, or they are anyway predetermined. It is so important to check that the values of such chemical and physical oil properties are in range with those required by the

The experimental procedures to get the values of such properties are also standardized and are indicated in the regulations. The following are parameters for starting oil that can affect

High sulphur fuels cause greater engine wear and in particular shorten the life of the catalyst. Biodiesel derived from soybean oil, as well as from pure rapeseed oil, is known to

Table 2. Characteristics of some ion-exchange resins (Amberlyst® - Dow Chemical).

(cc/g) 0.40 0.20 0.20 0.15 0.15 0.20 n.d.

Average pore

Total pore volume

Max. operating

[a]: Nitrogen BET; [b]: Dry weight

observed for several operating hours.

the chromatographic reactor or spray tower loop reactor.

contain virtually no sulphur (Radich, 2004; Zhiyuan et al., 2008).

slurry configuration reactor.

**4. Experimental part 4.1 Oil characterization** 

standard regulations (see Table 3).

the quality of the final biodiesel. Sulfur and Phosphorous content:

2010).

Catalyst A15d A36d A39w A40w A46w A70w D5081 Surface area (m2/g) 53 33 32 33 75 36 n.d.

> The iodine value (IV) is an index of the number of double bonds in biodiesel, and therefore is a parameter that quantifies the degree of unsaturation of biodiesel. Both EN and ASTM standard methods measure the IV by addition of an iodine/chlorine reagent.

> Soybean oil is reported to have an IV ranging from about 117 to 143 (Knothe, 1997), having quite the same unsaturation level of sunflower oil.

Cold Filter Plugging Point

The cold filter plugging point (CFPP) is the temperature at which wax crystals precipitate out of the fuel and plug equipment filters. At temperatures above this point, the fuel should give trouble free flow. These limits are to be decided by each EU member state according to its climate conditions, whereas the US ASTM D 6751 does not set any limit.

The test requires that the sample is cooled and, at intervals of fixed temperature, is drawn through a standard filter so determining the temperature at which the fuel is no longer filterable within a specified time limit.

The CFPP of soybean oil is reported to be around -5°C (Georgianni et al., 2007; Ramos et al., 2009), i.e. accomplishing only a part of the EU members countries (Meher et al., 2006).

Cetane Number

The cetane number (CN) measures the readiness of a fuel to auto-ignite when injected into the engine. It is also an indication of the smoothness of combustion. The CN of biodiesel depends on the distribution of fatty acids in the original oil. The CN determination is accomplished with the use a diesel engine called *Cooperative Fuel Research* (CFR) engine, under standard test conditions. The CFPP of soybean oil is reported to be higher than 50 (Ramos et al., 2009), so matching in the most cases the limit required by both EN and ASTM biodiesel standards.

#### **4.2 Oil standardization: the esterification reaction**

As already remarked in paragraph 3, pre-esterification of FFA in oils assumes great importance to obtain a feedstock suitable to be processed in the transesterification reaction.

In the recent years the authors have deepened the study of the pre-esterification process investigating the effect of the use of different kinds of oils, different types of reactors and catalysts and different operating conditions.

In the following paragraphs, the most relevant aspects of the experimental work and the results obtained by the authors for what concerns the pre-esterification process are reported.

Soybean Oil De-Acidification as a First Step Towards Biodiesel Production 331

A remarkable aspect of the proposed process is represented by the mild operative

Each single reaction has been carried out for six hours withdrawing samples from the reactor at pre-established times and analysing them through titration with KOH 0.1 M. The percentage of FFA content per weight was calculated as otherwise reported (Marchetti &

Unless otherwise specified, all the esterification experiments have been conducted using a slurry reactor as the one represented in Fig. 2a. A slurry reactor is the simplest type of catalytic reactor, in which the catalyst is suspended in the mass of the regents thanks to the agitation. In Fig. 2b a typical kinetic curve for the esterification reaction performed with

conditions, i.e. low temperature (between 303 and 338 K) and atmospheric pressure.

Fig. 2. a) Scheme of the slurry reactor; b) Example of kinetic curve of crude soybean oil with initial acidity=5% (wt.): FFA conversion (%) vs. time, slurry reactor, T=338K, catalyst: Amberlyst® 46, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10.

Much attention has been paid by the authors to the use of acid ion exchange resins. Amberlysts ®, i.e. a commercial product by Dow Advanced Materials, and D5081, a catalyst at the laboratory development stage by Purolite® have been successfully applied in this reaction. The main features of the employed catalysts are reported in Tab. 2 and described in

In Fig. 3 the results from the esterification reaction of different starting oils are shown. From the graph it can be noted that the lowest final acidity values are obtained with the refined materials, in spite of their initial acidities are the highest due to the addition of pure oleic acid. Refined oils are undoubtedly more easily processable with the esterification in comparison to crude oils, probably due to their lower viscosity which does not result in

This result has been confirmed by the addition of rapeseed oil, less viscous, to the waste cooking oil in different ratios: increasing the ratio of rapeseed oil to waste oil, the FFA

The differences in the acidic composition seem not to affect the yield of the reaction; in fact, similar values of FFA conversions are obtained for both the soybean oil and the animal fat,

**FFA conv (%)**

0 50 100 150 200 250 300 350 400

**Time (mins)**

**4.2.1 General reaction conditions** 

Errazu, 2007, Pirola et al. 2010).

a b

**4.2.2 Effect of the use of different kinds of oil** 

conversion after 6 hours increases.

limitations to the mass transfer of the reagents towards catalysts.

soybean oil is displayed.

paragraph 3.


Table 3. Standard specifications for biodiesel (automotive fuels).

In the following table (Table 4) the IV obtained by the authors using the standard procedure are listed for different kinds of not refined feedstock.


Table 4. Iodine values of some potential feedstock for biodiesel production.

#### **4.2.1 General reaction conditions**

330 Recent Trends for Enhancing the Diversity and Quality of Soybean Products

EN ISO 12185

preEN ISO 20884

EN14109

Specification Units limits Method Min Max Ester content % (m/m) 96.5 EN 14103 Density 15°C kg/m3 860 900 EN ISO 3675

Viscosity 40°C mm2/s 3.50 5.00 EN ISO 3104 Sulphur mg/kg - 10.0 preEN ISO 20846

(10% dist.residue) % (m/m) - 0.30 EN ISO 10370 Cetane number 51.0 EN ISO 5165 Sulphated ash % (m/m) - 0.02 ISO 3987 Water mg/kg - 500 EN ISO 12937

Total contamination mg/kg - 24 EN 12662 Cu corrosion max - EN ISO 2160

110°C h (hours) 6.0 EN 14112 Acid value mg KOH/g - 0.5 EN 14104 Iodine value gr I2/100 gr - 120 EN 14111 Linoleic acid ME % (m/m) - 12.0 EN 14103 Methanol % (m/m) - 0.20 EN 14110 Monoglyceride % (m/m) - 0.80 EN 14105 Diglyceride % (m/m) - 0.20 EN 14105 Triglyceride % (m/m) - 0.20 EN 14105 Free glycerol % (m/m) - 0.02 EN 14105 Total glycerol % (m/m) - 0.25 EN 14105 GpI metals (Na+K) mg/kg - 5.0 EN 14108

(Ca+Mg) mg/kg - 5.0 EN14538 Phosphorous mg/kg - 5.0 EN 14538

In the following table (Table 4) the IV obtained by the authors using the standard procedure

Brassica juncea (Indian mustard) 111 Brassica napus (Rapeseed) 115 Cartamus tinctorius (Safflower) 109 Heliantus annus (Sanflower) 143 Nicotiana tabacum (Tobacco) 137 Waste Cooking Oil 54.0

Table 4. Iodine values of some potential feedstock for biodiesel production.

Oilseed Iodine Value (g I2/100 g fat)

Table 3. Standard specifications for biodiesel (automotive fuels).

are listed for different kinds of not refined feedstock.

Carbon residue

Oxidation stability,

Gp II metals

A remarkable aspect of the proposed process is represented by the mild operative conditions, i.e. low temperature (between 303 and 338 K) and atmospheric pressure.

Each single reaction has been carried out for six hours withdrawing samples from the reactor at pre-established times and analysing them through titration with KOH 0.1 M. The percentage of FFA content per weight was calculated as otherwise reported (Marchetti & Errazu, 2007, Pirola et al. 2010).

Unless otherwise specified, all the esterification experiments have been conducted using a slurry reactor as the one represented in Fig. 2a. A slurry reactor is the simplest type of catalytic reactor, in which the catalyst is suspended in the mass of the regents thanks to the agitation. In Fig. 2b a typical kinetic curve for the esterification reaction performed with soybean oil is displayed.

Fig. 2. a) Scheme of the slurry reactor; b) Example of kinetic curve of crude soybean oil with initial acidity=5% (wt.): FFA conversion (%) vs. time, slurry reactor, T=338K, catalyst: Amberlyst® 46, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10.

Much attention has been paid by the authors to the use of acid ion exchange resins. Amberlysts ®, i.e. a commercial product by Dow Advanced Materials, and D5081, a catalyst at the laboratory development stage by Purolite® have been successfully applied in this reaction. The main features of the employed catalysts are reported in Tab. 2 and described in paragraph 3.

#### **4.2.2 Effect of the use of different kinds of oil**

In Fig. 3 the results from the esterification reaction of different starting oils are shown.

From the graph it can be noted that the lowest final acidity values are obtained with the refined materials, in spite of their initial acidities are the highest due to the addition of pure oleic acid. Refined oils are undoubtedly more easily processable with the esterification in comparison to crude oils, probably due to their lower viscosity which does not result in limitations to the mass transfer of the reagents towards catalysts.

This result has been confirmed by the addition of rapeseed oil, less viscous, to the waste cooking oil in different ratios: increasing the ratio of rapeseed oil to waste oil, the FFA conversion after 6 hours increases.

The differences in the acidic composition seem not to affect the yield of the reaction; in fact, similar values of FFA conversions are obtained for both the soybean oil and the animal fat,

Soybean Oil De-Acidification as a First Step Towards Biodiesel Production 333

A 36 A 46 A 70 A 39 A 40 A 15

338 K, 10% wt of catalyst vs. fat 323 K, 10% of catalyst vs. fat

A70 shows the best performance in all the operative conditions. For this reason, it was further tested to evaluate its catalytic activity in milder operating conditions, i.e. lower temperatures and lower catalyst/fat ratio. The results thus obtained are displayed in Figs. 5a and b, showing that the activity of the catalyst decreases as the reaction temperature or its concentration decrease. However it is worth remarking that even at room temperature. Catalysts A46 and D5081, have been compared at different temperatures and catalyst's

As can be seen from the graphs, catalyst D5081 shows better results than A46 in milder operative conditions. This can be easily explained by the higher number of acid sites located on its surface (compared with Tab. 2). In particular, the use of a ratio of 10% of catalyst D5081 vs. oil allows reaching the maximum conversion in 2 hours. The outcome of this study suggested that a fixed amount of acid active sites per gram of FFA was required to reach the maximum of conversion in 4 hours. Based on the experimental data, this amount

To verify this hypothesis, different batches of sunflower oil with different initial acidity were prepared and then de-acidified by loading a quantity of A46 corresponding to 1,2 meq of H+ per gram of FFA. The obtained results are shown in Fig. 7 and confirm the hypothesis set out above. In fact, a complete conversion, corresponding to a FFA concentration lower

Fig. 4. FFA conversion (%) after 6 h. Comparison of Amberlysts in different operative conditions, initial acidity=5%, slurry reactor, T=338K, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with FFA content 0.5% per weight, i.e. suitable for industrial

loadings. The results of this study are summarized in the Figs. 6 a and b.

than 0,5%, is reached after 4 hours regardless of the initial FFA amount.

applications.

338 K, 5% wt of catalyst vs. fat

was found to be equal to 1,2 meq of H+.

**FFA conv (%)**

in spite of their different acidic compositions. Indicative compositions of some oils used in the experimentation are given in Tab.1

Fig. 3. Acidity removed by esterification (6 h) and residual acidity of different oils used as raw material: slurry reactor, T=338K, catalyst: Amberlyst® 46 weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10; \*commercial, refined fats with the addition of pure oleic acid.

#### **4.2.3 Comparison among different catalysts at different loadings and temperatures**

A comparison among the different kinds of Amberlyst at different temperatures has been performed by the authors in a recently published paper dealing with the de-acidification of animal fat (Bianchi et al., 2010). In Fig. 4 the results of this study are summarized.

Under the applied conditions, all the catalysts perform quite well in the esterification reaction, with the exception of A40. Its unsatisfactory performance can be explained taking into account its lower specific surface area and a lower acid site concentration if compared to other Amberlysts. Being these two parameters directly connected to catalytic activity, their simultaneous deficiency is clearly the cause of the unsatisfactory performance.

The catalytic performances of the sample A46 appear to be remarkable, in spite of the low concentration of active acid sites. This result can be explained considering the particular configuration of the catalytic particles, where acid sites are located only on its surface, thus being immediately and easily available for the reaction.

in spite of their different acidic compositions. Indicative compositions of some oils used in

Residual Acidity Acidity removed by esterification

the experimentation are given in Tab.1

5.76

4.66

0.52

0

oleic acid.

1

2

3

4

FFA (% wt.)

5

6

7

0.31 0.23

0.74


0.52 0.48

Fig. 3. Acidity removed by esterification (6 h) and residual acidity of different oils used as raw material: slurry reactor, T=338K, catalyst: Amberlyst® 46 weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10; \*commercial, refined fats with the addition of pure

**4.2.3 Comparison among different catalysts at different loadings and temperatures**  A comparison among the different kinds of Amberlyst at different temperatures has been performed by the authors in a recently published paper dealing with the de-acidification of

Under the applied conditions, all the catalysts perform quite well in the esterification reaction, with the exception of A40. Its unsatisfactory performance can be explained taking into account its lower specific surface area and a lower acid site concentration if compared to other Amberlysts. Being these two parameters directly connected to catalytic activity,

The catalytic performances of the sample A46 appear to be remarkable, in spite of the low concentration of active acid sites. This result can be explained considering the particular configuration of the catalytic particles, where acid sites are located only on its surface, thus

animal fat (Bianchi et al., 2010). In Fig. 4 the results of this study are summarized.

their simultaneous deficiency is clearly the cause of the unsatisfactory performance.

being immediately and easily available for the reaction.


1.75 1.68

0.27 0.35 0.25


4.14

2.17


0.88


2.30

2.96

0.52 0.37

2.20 2.20

Fig. 4. FFA conversion (%) after 6 h. Comparison of Amberlysts in different operative conditions, initial acidity=5%, slurry reactor, T=338K, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with FFA content 0.5% per weight, i.e. suitable for industrial applications.

A70 shows the best performance in all the operative conditions. For this reason, it was further tested to evaluate its catalytic activity in milder operating conditions, i.e. lower temperatures and lower catalyst/fat ratio. The results thus obtained are displayed in Figs. 5a and b, showing that the activity of the catalyst decreases as the reaction temperature or its concentration decrease. However it is worth remarking that even at room temperature.

Catalysts A46 and D5081, have been compared at different temperatures and catalyst's loadings. The results of this study are summarized in the Figs. 6 a and b.

As can be seen from the graphs, catalyst D5081 shows better results than A46 in milder operative conditions. This can be easily explained by the higher number of acid sites located on its surface (compared with Tab. 2). In particular, the use of a ratio of 10% of catalyst D5081 vs. oil allows reaching the maximum conversion in 2 hours. The outcome of this study suggested that a fixed amount of acid active sites per gram of FFA was required to reach the maximum of conversion in 4 hours. Based on the experimental data, this amount was found to be equal to 1,2 meq of H+.

To verify this hypothesis, different batches of sunflower oil with different initial acidity were prepared and then de-acidified by loading a quantity of A46 corresponding to 1,2 meq of H+ per gram of FFA. The obtained results are shown in Fig. 7 and confirm the hypothesis set out above. In fact, a complete conversion, corresponding to a FFA concentration lower than 0,5%, is reached after 4 hours regardless of the initial FFA amount.

Soybean Oil De-Acidification as a First Step Towards Biodiesel Production 335

Fig. 7. FFA content (%) vs. reaction time: kinetic curves of the FFA esterification in

weight ratio alcohol/oil= 16:100, T = 338K.

**Crude Palm oil as feedstock and catalyst washed with methanol after each run**

further details, (Pirola et. al., 2010).

**0**

catalyst/oil=1:10.

**20**

**40**

**60**

**FFA conv (%)**

**80**

**100**

sunflower oil with different initial acidities using a fixed catalyst/FFA ratio, slurry reactor,

conversions, measured at the end of each of the 6–hour reactions, are reported in Fig. 8 as a function of the run number. At the end of the recycles, a decrease of activity of about 25% was observed, to be probably ascribed to some fragmentation of catalyst's particles. For

> **y = -0.0045x2 + 0.0539x + 86.007 R² = 0.9392**

**0 10 20 30 40 50 60 70 80 90 100**

Fig. 8. FFA conversion % after 6 h during 90 successive runs with the same Amberlyst 46®

The recycle of the use of catalyst was also performed for D5081 in the FFA esterification of rapeseed oil. The obtained results are outlined in the following graphs (Figs. 9a and b).

sample. Slurry reactor, T= 338 K, weight ratio alcohol/oil= 16:100, weight ratio

**run number**

**Palm oil as feedstock** 

**Soybean oil as feedstock** 

Fig. 5. a) FFA conversion (%) vs reaction time in presence of A70 for different reaction temperatures, initial acidity=5%, slurry reactor, 10 % per weight (wt) of catalyst vs. fat, weight ratio alcohol/oil= 16:100; b) FFA conversion (%) vs. time in presence of A70, slurry reactor, weight ratio alcohol/oil= 16:100, T = 338K with different amounts of the catalysts. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with a FFA content 0.5% per weight, i.e. suitable for industrial applications.

Fig. 6. a) FFA conversion (%) vs reaction time for different amounts of catalysts A46 and D5081, rapeseed oil with initial acidity=5%, slurry reactor, weight ratio alcohol/oil= 16:100, T=338K; b) FFA conversion (%) vs. time at different temperatures, slurry reactor, 10 % per weight (wt) of catalyst vs. fat, weight ratio alcohol/oil= 16:100, T = 338K. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with a FFA content 0.5% per weight, i.e. suitable for industrial applications.

#### **4.2.4 Study of catalysts' lifetime**

A crucial parameter for the industrial application is the catalyst lifetime; this parameter has been evaluated by the authors in a recent work (Pirola et al., 2010) by performing ninety consecutive batch de-acidification runs, each lasting 6 hours, were conducted using crude palm oil or soybean oil as a feedstock and Amberlyst® 46 as a catalyst. The final FFA

b

b

0 100 200 300 400

**Time (mins) 10% of catalyst vs. fat 5% of catalyst vs. fat 2,5% of catalyst vs. fat 1,25% of catalyst vs. fat**

0 60 120 180 240 300 360 420

**D5081 338K D5081 318K**

**A46 338K A46 318K**

**D5081 338 K, no pretreatment**

**Time (mins)**

**FFA conv (%)**

**FFA conv (%)**

Fig. 6. a) FFA conversion (%) vs reaction time for different amounts of catalysts A46 and D5081, rapeseed oil with initial acidity=5%, slurry reactor, weight ratio alcohol/oil= 16:100, T=338K; b) FFA conversion (%) vs. time at different temperatures, slurry reactor, 10 % per weight (wt) of catalyst vs. fat, weight ratio alcohol/oil= 16:100, T = 338K. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with a FFA content

A crucial parameter for the industrial application is the catalyst lifetime; this parameter has been evaluated by the authors in a recent work (Pirola et al., 2010) by performing ninety consecutive batch de-acidification runs, each lasting 6 hours, were conducted using crude palm oil or soybean oil as a feedstock and Amberlyst® 46 as a catalyst. The final FFA

Fig. 5. a) FFA conversion (%) vs reaction time in presence of A70 for different reaction temperatures, initial acidity=5%, slurry reactor, 10 % per weight (wt) of catalyst vs. fat, weight ratio alcohol/oil= 16:100; b) FFA conversion (%) vs. time in presence of A70, slurry reactor, weight ratio alcohol/oil= 16:100, T = 338K with different amounts of the catalysts. The dotted line represents the value of FFA conversion necessary to obtain a feedstock with

**FFA conv (%)**

a

**FFA conv (%)**

0 100 200 300 400

**338 K 323 K 313 K 303 K**

a FFA content 0.5% per weight, i.e. suitable for industrial applications.

**Time (mins)**

0 60 120 180 240 300 360 420

**D5081 10% of catalyst vs. oil D5081 8% of catalyst vs. oil D5081 6% of catalyst vs. oil A46 10% of catalyst vs. oil A46 8% of catalyst vs. oil A46 6% of catalyst vs. oil**

**a b**

Catalyst's loading: mcat/moil

**Time (mins)**

0.5% per weight, i.e. suitable for industrial applications.

**4.2.4 Study of catalysts' lifetime** 

Fig. 7. FFA content (%) vs. reaction time: kinetic curves of the FFA esterification in sunflower oil with different initial acidities using a fixed catalyst/FFA ratio, slurry reactor, weight ratio alcohol/oil= 16:100, T = 338K.

conversions, measured at the end of each of the 6–hour reactions, are reported in Fig. 8 as a function of the run number. At the end of the recycles, a decrease of activity of about 25% was observed, to be probably ascribed to some fragmentation of catalyst's particles. For further details, (Pirola et. al., 2010).

Fig. 8. FFA conversion % after 6 h during 90 successive runs with the same Amberlyst 46® sample. Slurry reactor, T= 338 K, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10.

The recycle of the use of catalyst was also performed for D5081 in the FFA esterification of rapeseed oil. The obtained results are outlined in the following graphs (Figs. 9a and b).

Soybean Oil De-Acidification as a First Step Towards Biodiesel Production 337

(b) (b) (c)

Fig. 10. Continuous or semi-continuous experimental set-up: (a) feeding vessel, (b) mechanic

Fig. 11. Semi-continuous experiments with 7 g of catalyst; FFA conversion % vs. time using crude palm oil (o) or soybean oil (). Pump flow = 10 mL min-1; T = 338 K, molar ratio

The FFA conversion increases with time using both oils, but it is higher using the crude palm oil (FFA conversion after 700 min: about 90% and 70% for palm and soybean oil, respectively). This result can be explained considering the difference between the two different raw materials, which concerns both the composition of the substrate (different unsaturation levels between the two oils) and the composition of the FFA (only oleic acid for soybean oil and a mixture of C12-C18 acids for crude palm oil). These differences obviously affect both the lifetime of the oil/methanol emulsion and the diffusional aspects along the

stirrer, (c) pump, (d) mixing chamber, (e) catalytic packed bed reactor.

(d)

(e)

(a)

FFA/alcohol = 1: 6, weight ratio catalyst/ oil = 1: 10.

catalytic bed.

Fig. 9. FFA conversion % vs. reaction time of a) recycles 2 to 7 and b) 8 to 15, rapeseed oil, initial acidity=5% (oleic acid), slurry reactor, T = 338 K, weight ratio alcohol/oil= 16:100, weight ratio catalyst/oil=1:10.

In the initial stage (use 2÷7) the resin does not results in well defined kinetic curves. The reason of slight diminution of FFA conversion with time is probably ascribable to catalyst's settling in the system as it has to adapt to the ambient of reaction before giving a stable performance. From the 8th recycle of catalyst's use on, the curves of FFA conversion overlap: the conversion reached after pre-established times is the same for different runs with the use of the same batch of catalyst.

#### **4.2.5 Reactors**

The experimental results discussed in the previous paragraph suggested that a packed-bed reactor, where the catalyst particles are immobilized inside it, could eliminate the mechanical stress of the catalyst particles typical of a slurry reactor. On the other hand, a packed-bed reactor makes the contact between the organic phase (oil/FFA mixture) and methanol less effective. For this reason, in the employed experimental setup a mixing chamber was located just before the catalytic reactor.

The reaction in both continuous and semicontinuous modes was conducted using the experimental setup shown in Fig. 10.

The methanol/oil mixture is taken from the vessel (a), where it is continuously mechanically stirred (b), and then it is admitted into the mixing chamber (d) by a pump (c). This chamber (0.2 L) is located just before the catalytic reactor in order to obtain the maximum contact between oil and methanol (not fully soluble) inside the catalytic bed (e). The catalytic reactor (0.5 L) contains a packed bed of Amberlyst 46® (7 g). The pump flow is maintained at 10 mL min-1, so obtaining a contact time in the catalytic bed equal to 1 min.

In the semi-continuous experiments the reaction stream, leaving the catalytic reactor (e) was returned to the vessel (a); in continuous experiments the reaction stream from the catalytic reactor (e) was continuously discharged from the system.

In Fig. 11, the results obtained for both crude palm oil and soybean oil are reported using the experimental setup shown in Fig. 11 as a semicontinuous reactor.

Fig. 9. FFA conversion % vs. reaction time of a) recycles 2 to 7 and b) 8 to 15, rapeseed oil, initial acidity=5% (oleic acid), slurry reactor, T = 338 K, weight ratio alcohol/oil= 16:100,

In the initial stage (use 2÷7) the resin does not results in well defined kinetic curves. The reason of slight diminution of FFA conversion with time is probably ascribable to catalyst's settling in the system as it has to adapt to the ambient of reaction before giving a stable performance. From the 8th recycle of catalyst's use on, the curves of FFA conversion overlap: the conversion reached after pre-established times is the same for different runs with the use

The experimental results discussed in the previous paragraph suggested that a packed-bed reactor, where the catalyst particles are immobilized inside it, could eliminate the mechanical stress of the catalyst particles typical of a slurry reactor. On the other hand, a packed-bed reactor makes the contact between the organic phase (oil/FFA mixture) and methanol less effective. For this reason, in the employed experimental setup a mixing

The reaction in both continuous and semicontinuous modes was conducted using the

The methanol/oil mixture is taken from the vessel (a), where it is continuously mechanically stirred (b), and then it is admitted into the mixing chamber (d) by a pump (c). This chamber (0.2 L) is located just before the catalytic reactor in order to obtain the maximum contact between oil and methanol (not fully soluble) inside the catalytic bed (e). The catalytic reactor (0.5 L) contains a packed bed of Amberlyst 46® (7 g). The pump flow is maintained at 10 mL

In the semi-continuous experiments the reaction stream, leaving the catalytic reactor (e) was returned to the vessel (a); in continuous experiments the reaction stream from the catalytic

In Fig. 11, the results obtained for both crude palm oil and soybean oil are reported using

weight ratio catalyst/oil=1:10.

of the same batch of catalyst.

chamber was located just before the catalytic reactor.

min-1, so obtaining a contact time in the catalytic bed equal to 1 min.

the experimental setup shown in Fig. 11 as a semicontinuous reactor.

reactor (e) was continuously discharged from the system.

experimental setup shown in Fig. 10.

**4.2.5 Reactors** 

Fig. 10. Continuous or semi-continuous experimental set-up: (a) feeding vessel, (b) mechanic stirrer, (c) pump, (d) mixing chamber, (e) catalytic packed bed reactor.

Fig. 11. Semi-continuous experiments with 7 g of catalyst; FFA conversion % vs. time using crude palm oil (o) or soybean oil (). Pump flow = 10 mL min-1; T = 338 K, molar ratio FFA/alcohol = 1: 6, weight ratio catalyst/ oil = 1: 10.

The FFA conversion increases with time using both oils, but it is higher using the crude palm oil (FFA conversion after 700 min: about 90% and 70% for palm and soybean oil, respectively). This result can be explained considering the difference between the two different raw materials, which concerns both the composition of the substrate (different unsaturation levels between the two oils) and the composition of the FFA (only oleic acid for soybean oil and a mixture of C12-C18 acids for crude palm oil). These differences obviously affect both the lifetime of the oil/methanol emulsion and the diffusional aspects along the catalytic bed.

Soybean Oil De-Acidification as a First Step Towards Biodiesel Production 339

Being the transesterification an equilibrium reaction, it was performed in two steps, removing the formed glycerine after the first step. The adopted conditions were the

1st step: weight ratio methanol/oil=20:100, weight ratio MeONa/oil=1:100, 233 K,

2nd step: weight ratio methanol/oil=5:100, weight ratio MeONa/oil=0.5:100, 233 K,

The total ester content is a measure of the completeness of the transesterification reaction. Many are the factors that affect ester yield in the transesterification reaction: molar ratios of glycerides to alcohol, type of catalyst(s) used, reaction conditions, water content, FFA

The European preEN14214 biodiesel standard sets a minimum limit for ester content of >96.5% mass, whereas the US ASTM D 6751 biodiesel standard does not set a specification

Mono- and di-glycerides as well as tri-glycerides can remain in the final product in small quantities. Most are generally reacted or concentrated in the glycerine phase and separated

The analyses of methyl esters and unreacted mono-, di- and triglycerides are accomplished

The detailed requirements for biodiesel according to both EN 14214 and US ASTM D 6751

In order to develop a process simulation of the FFA esterification, able to predict the

The considered reaction system turns out to be a highly non-ideal system, being formed by a mixture of oil, methylester, methanol, FFA and water. The interactions among these molecules are absolutely not ideal, in fact they are only partially soluble and a two phase

Indeed, the activity coefficients are used not only for the phase and chemical equilibria calculations, but also for the kinetic expressions. Modified UNIFAC model was used adopting the parameters available in literature and published by Gmehling et al. (2002).

FFA + methanol methylester + water

Fig. 13. FFA esterification reaction and hydrolysis (indirect reaction) considered in the

A pseudohomogeneous model was used for describing the kinetic behaviour of the reaction

The oil is considered as non-reacting solvent, being present in large quantity.

(Pöpken et al., 2000). The adopted model is displayed in the following equation:

reaction progress, a thermodynamic and kinetic analysis was performed.

system is formed if the quantity of methanol is greater than 6-8 wt%.

The considered reaction is the following (Fig.13).

following:

1,5 h

1 h

for ester content.

from the ester.

**5. Simulation** 

through gas chromatography.

**5.1 Thermodynamic aspects** 

**5.2 Reaction kinetics** 

process simulation.

are listed in paragraph 1.

concentration, etc. (Environment Australia, 2003).

To improve the stability of the methanol/oil emulsion, the authors substituted the classical mixing chamber with an emulsificator based on five co-axial rotating ring gears which are able to break the biphasic mixture into very tiny drops. Using this device and starting from two entirely separated liquid phases (oil and methanol) it was possible to obtain a much more stable emulsion. In Fig. 12 a comparison between FFA conversion obtained with the classical mixing chamber and the emulsificator is reported: the better results reached using the emulsificator are evident.

Fig. 12. Semi-continuous experiment: FFA conversion % vs. time using crude palm oil as feedstock and having before the catalytic bed: classical mechanic stirring reactor () or emulsificator at 500 rpm (▲). Pump flow = 10 mL min-1; T = 338 K, molar ratio FFA: alcohol = 1: 6, weight ratio catalyst: oil = 1: 10.

The methanol/oil emulsion lifetime without the emulsificator device, measured in the same way at the exit of the mixing chamber, is of about 15 seconds. Referring to our system, the contact time between the catalyst and the reactant is 1 min for 7 g of catalyst (pump flow = 10 mL min-1): for this reason without the emulsificator device a significant part of the catalytic bed (about 75%) does not work.

#### **4.3 Oil transformation: the transesterification reaction and biodiesel characterization**

The transesterification reaction has been performed by the authors on the raw materials deacidified with the esterification process described in the previous paragraph.

Sodium methoxide (MeONa) was employed as catalyst. MeONa is known to be the most active catalyst for triglycerides transesterification reaction, but it requires the total absence of water (Schuchardt, 1996). For this reason, the unreacted methanol and the reaction water were evaporated from the de-acidified oils before processing them with the transesterification reaction.

The employed experimental setup was the same as displayed in Fig. 2.

Being the transesterification an equilibrium reaction, it was performed in two steps, removing the formed glycerine after the first step. The adopted conditions were the following:


The total ester content is a measure of the completeness of the transesterification reaction. Many are the factors that affect ester yield in the transesterification reaction: molar ratios of glycerides to alcohol, type of catalyst(s) used, reaction conditions, water content, FFA concentration, etc. (Environment Australia, 2003).

The European preEN14214 biodiesel standard sets a minimum limit for ester content of >96.5% mass, whereas the US ASTM D 6751 biodiesel standard does not set a specification for ester content.

Mono- and di-glycerides as well as tri-glycerides can remain in the final product in small quantities. Most are generally reacted or concentrated in the glycerine phase and separated from the ester.

The analyses of methyl esters and unreacted mono-, di- and triglycerides are accomplished through gas chromatography.

The detailed requirements for biodiesel according to both EN 14214 and US ASTM D 6751 are listed in paragraph 1.
