Response Surface Method as Experimental Design and Regression Modeling

2. Fundamentals and State of Knowledge

2.6 Response Surface Method as Experimental Design and Regression Modeling

It is assumed that in the optimum region, the curvature of the response surface can be approximated by a function of the process variables and their squares together with second order cross-products. Equation 2.22 expresses the general form of a response surface of the response variable Y as a function of “n” independent process variables from X1 to Xn.

Y = B0 + 

 n

i 1BiXi + 

 n

i 1 2 i iiX

B + 

 n

j i

j

i, 1BijXiXj (2.22) In order to obtain the equation 2.22, many kinds of experimental design can be applied, such as Box-Wilson central composite design, full factorial design, Box-Benhken design. Among them, Box-Wilson central composite design, commonly called a ‘central composite design’ is one of the most popular designs associated with RSM. This design includes three parts as shown in Figure 2.15:

- A two-level full or fractional factorial design

- A group of axial (or star) points in which each factor is varied to high and low levels with all other factors held constant

- Centerpoints

Figure 2.15: Schematic diagram of a central composite design

The configuration of axial points leads to variations. If the distance from the center of the design space to a factorial point is ±1 unit for each factor, the distance from the center of the design space to a axial point is ±, the precise value was chosen to ensure rotability. The axial points can also be set at the center of each face of the factorial space, so = ±1. In this case, the design is called Central Composite Face Centered Design (CCFCD). 3 levels for each factor are required for this variety. CCFCD provides relatively high quality predictions over the entire design space without using points outside the original factor range. Therefore, the number of experiments can be reduced comparing to other methods. However, it gives poor precision for estimating pure quadratic coefficients.

3. BSupercritical Fluid Extraction of Palm Oil

Palm oil is well known all over the world because of its high quality. This oil is commonly produced by extracting the crude oil from the digested fruit mash [54]. However, a significant quantity of tocochromanols and carotenoids remains in the residue from palm oil production by traditional screw pressing. The objective of the following study was to investigate an alternative environmentally friendly extraction method, which can better recover these valuable minor compounds. The work included a study of extraction of palm mesocarp by SCCO2 and subcritical propane at different pressure, temperature and flow rate. Total oil yield and co-extracted water were investigated in the course of extraction. Tocochromanols and carotenoids were evaluated not only in the extraction oil, but also in the residual fiber. Modeling of extraction process was also performed for a further up-scaling.

3.1 BMaterials and Methods 3.1.1 Materials

Palmitic acid (>99%) and squalane (GC grade) were supplied by Merck (Germany).

Monopalmitin (99%), dipalmitin (99%) and tetradecane (99%) were from Sigma (USA). Pyridine (99.8%), hexane (>95%), acetone (>99.8%), acetonitril (HPLC grade) and N-methyl-N- trimethylsilyl-trifluoracetamide were purchased from Fluka (Switzerland), Lab-Scan (Ireland), Riedel-de Họen (Germany), Prolabo (France), and Macherey-Nagel (Germany), respectively.

Palm fruit (Elaeis guineensis) was kindly supported from Carotech Company (Malaysia).

The fruits were separated into skin, mesocarp and kernel. The yellow part of the mesocarp is investigated. The average particle size of the pulp ready for supercritical extraction was about 1x2x6 (mm x mm x mm).

3.1.2 Equipment and experimental procedure

A standardized low cost supercritical extraction system, developed at the Institute for Thermal and Separation Processes – at the Hamburg University of Technology (TUHH), was used. The simplified flow sheet is shown as Figure 1. Fluids from the reservoir tank were pumped by a Maximator pump (max. 600 bar) to the 100mL steel extractor cell, which was loaded with 14.5 g of palm mesocarp (fixed bed) for each run. The system was monitored at the investigated pressure (200 - 400 bar) and temperature (35 - 65°C) with the flow rate from 14 to 56 (kg/h) of

gas per kg of sample. The extracts were collected continuously in 10mL glass vials, used as sample collectors at atmospheric pressure.

1-Gas cylinder

2-One-way-valve 3-Condenser 4-Plunger pump

5-Back pressure regulator 6-Extraction vessel 7-Heat exchanger 8-Metering valve

9-Sample collector 10-Gas flow-meter 11-Pressure gauge 12-Thermometer

Figure. 3.1: Flow-sheet of the extraction unit

3.1.3 Analytical method

a) High Performance Liquid Chromatography (HPLC)

A HPLC system from Gynkotek with RF 1002 Fluorescent detector was used for the analysis. Tocopherols and tocotrienols in the oil samples were separated on a LiChrosorb Diol, 5

m, 250 x 4.6 mm column (Chrompack N° 612834). The mobile phase was hexane (96%) and butyl-methyl-ether (4%) at a flow rate of 1300 L/min. Injection volume was 20 L. External standard curves were used to determine tocochromanols content in the oil samples (Appendix A1).

b) Gas Chromatography (GC)

Free fatty acids (FFAs), monoacylglycerols (MAGs) and diacylglycerols (DAGs) were determined by GC analysis (Hewlett Packard HP 5890A capillary gas chromatograph) with integrator (HP3396 Series II). The stationary phase was a J & W Scientific fused silica (DB-5ht) column (30mì0.25mm i.d. with 0.1 àm coating). Carrier gas was nitrogen (2 L/min). The oven

PI

4

8

3

6

7 9

PI TI

1

PI

5

F 2 10

2

11 11 12 11

temperature was: 120°C, 2 min constant; 10 °C/min to 220 °C; 5 °C/min to 360°C; 360 °C, 10 min constant. Injection volume was 1àl with a split ratio of 1:20. For a better peak recording, sample compounds were silylated with N-methyl-N-trimethylsilyl-trifluoracetamide (MSTFA).

BDetermination of the response coefficients by internal standard

4.0% (w/v) solution of tetradecane in pyridine was used as internal standard. The mixtures of reference substances to internal standard (IS) were injected at various ratios for GC analysis. The response coefficient kX of the substance X to IS is determined by equation 3.1.

IS X X IS

X A

A W

k W  (3.1)

Where:

X: palmitic acid, mono- , and dipalmitin,

AX and WX : peak area and weight (mg) of the reference substance X, AIS and WIS : peak area and weight (mg) of the IS.

The response factor of palmitic acid, mono-, and dipalmitin were used to calculate the whole group of FFAs, MAGs and DAGs respectively, because they are dominant in palm oil.

(Appendix A2)

BApplication to calculate chemical composition of sample

Content mZ of FFAs, MAGs, DAGs in a sample was determined by the following equation:

1 100

(%)   

sample IS IS

Z Z

Z W

W A

A

m k (3.2)

Where:

Z : FFAs, MAGs, DAGs,

mZ: percent (w/w) of mass of component Z in sample, kZ : response factor of component Z,

AZ and AIS : peak area of the component Z and internal standard in sample, WIS and Wsample: mass of internal standard and sample (mg).

c) BSoxhlet extraction

A Soxhlet extraction was used to extract the total oil of the original palm mesocarps and the residual fibers from the supercritical CO2 extraction. Hexane was used as the extraction solvent. The extraction time was 8 hours.

d) BSpectrometer

BUV-Vis Spectroscopy (spectrometer UV-120-02 from Shimadzu) was used to determine the content of carotenoids in the analyzed samples. For a measurement, an amount of 10 to 20 mg of oil sample is diluted with 2 mL mixture of acetone and hexane (30:70 by Vol. %). The absorbance (ABS) was recorded at the wavelength of 450 nm and compared with the standard curve determined with a series of samples with a known amount of -carotene.

e) Karl - Fischer titration

BDead-Stop Titrator TR 52 (Fa. Schott, Hofeim) was used to determine the water concentration from the supercritical extracts. The titrating agents were Hydranal-Composite 5 (Riedel-de Họen). To obtain the water content, first the titer number was determined regularly by titrating known amounts of distilled water.

T O H

V

Titer  m 2 (3.3)

Bwhere:

O

mH

2 B: amount of water in mg,

BVT: volume the titrant in mL.

BThe titrating solutions had a water equivalent of approximately 5 mg H2O/mL. Water content (%) in the samples was finally dertermined as:

P T

m Titer O V

H 100

2 (%)



  (3.4) where:

mP: amount of sample in mg.

3.2 Results and Discussion

3.2.1 BCharacteristics of palm mesocarp

The composition of palm mesocarp varies with the size and the age of every palm fruit.

Table 3.1 presents the average composition of the palm mesocarp used as the material input for supercritical extraction in this study. The results show that palm mesocarp is a good source to extract oil and valuable minor compounds like carotenoids or tocochromanols. However, the free fatty acid content of this palm fruit is rather high. This can be attributed to the enzymatic hydrolysis of the oil under the influence of an endogenous lipase in the fruits [133].

Table 3.1: Composition of palm fruits used in this work.

Component Concentration Water 20%

Total oil 45%

Carotenoids 450 ppm

Tocochromanols 800 ppm

Monoacylglycerols 3%

Diacylglycerols 7%

Free fatty acids 15%

Triaclyglycerols 75%

3.2.2 BEffect of process parameters a) BEffect of pressure

BAs a result of changing fluid density and solvent power, the solubility of oil in a dense fluid varies with pressure [1]. Therefore, total palm oil yields are significantly affected by the pressure applied. Following the conditions described in Figure 3.2, increasing pressure from 200 to 400 bar increased the total oil yield from 35 to 47% after 120 minutes of extraction; while at the first 60 minutes the difference of yield was much larger. In general, it is observed that the extraction yield increases with higher pressure at any temperature.

0 10 20 30 40 50

0 20 40 60 80 100 120 140

Extraction Time (min)

Extracted oil / original oil (%)

400 bar 300 bar 200 bar

Figure 3.2: Extraction of mesocarp with SCCO2 at 45°C, 14 (kg/h)/kg

b) BEffect of temperature

Temperature has a great effect on an extraction process. At a high pressure (300 or 400 bar), extraction yield increases with higher temperature (Figure 3.3). This phenomenon is explained by the fact that the effect on palm oil vapor pressure is more pronounced than the effect of decreasing solubility when changing extraction temperature [1]. At lower pressure (200 bar), for the first 60 minutes of extraction time, lower temperature is more favorable for the extraction.

However, when the available oil near the surface is depleted, mass transfer of oil from the center of the fruit pulp to the outer surface is the main factor to determine the extraction rate of the supercritical extraction. Consequently, higher mass transfer at 55°C gives a better extraction yield after 60 minutes of extraction.

c) BEffect of flow rate

The flow rate affects the total oil recovery and the mass transfer inside the palm pulp. In Figure 3.4, it is observed that an increase in flow rate will increase the amount of the collected oil, therefore shorten the extraction time. However, the balance between additional oil recovery and extra cost should be considered in the economical point of view. Brunner [1] pointed out that a high solvent ratio causes enhanced operating costs and higher capital costs, because the equipment is more expensive due to its larger size.

0 10 20 30 40 50 60

0 20 40 60 80 100 120 140

Extraction time (min)

Extracted oil / original oil (%)

400 bar, 55°C 300 bar, 55°C 200 bar, 55°C 400 bar, 35°C 300 bar, 35°C 200 bar, 35°C

Figure 3.3: Extraction of mesocarp with SCCO2 at 14 (kg/h)/kg

0 10 20 30 40 50 60 70 80 90

0 30 60 90 120 150

Extraction Time (min)

Extracted oil / original oil (%)

400 bar, 35 (kg/h)/kg 400 bar, 14(kg/h)/kg 200 bar, 14 (kg/h)/kg 200 bar, 56 (kg/h)/kg

Figure 3.4: Extraction of mesocarp with SCCO2 at 55°C.

3.2.3 BExtraction with different fluids a) Total amount of extract

From data presented in the previous section, it can be concluded that the extraction was not sufficient for oil recovery at 200 bar, even with a very high flow rate. Consequently, SCCO2

at 400 or 300 bar with a higher flow rate of 35 kg/kg/h was further investigated and compared with subcritical propane at 50 bar at different extraction temperatures. The results show that within these conditions, palm oil could be recovered up to 60 – 70% after 60 minutes of extraction or 80-90% after 120 minutes (Figure 3.5).

0 20 40 60 80 100

0 30 60 90 120 150

Extraction Time (min)

Extracted oil / original oil (%)

Propane 50 bar, 55°C Propane 50 bar, 45°C SCCO2 400 bar, 65°C SCCO2 400 bar, 55°C SCCO2 300 bar, 65°C

Figure 3.5: Extraction of mesocarp with SCCO2 and subcritical propane at 35 (kg/h)/kg.

Furthermore, it can be concluded that the temperature effect is more pronounced in case of the extraction with SCCO2 than with subcritical propane. On the other hand, free oil was more soluble in subcritical propane than in SCCO2. However, the extraction with SCCO2 gave better total oil yields after 45 minutes than those with subcritical propane when the available oil near the surface was depleted. The difference in oil recovery can be attributed to the changing of the palm’s pulp structure during the process. Residues from the extraction show that some parts of the mesocarp were divided into small pieces in case of the extraction with SCCO2, but not with subcritical propane (Figures 3.6 & 3.7). Kim [134] and Zheng [135] reported that SCCO2 can affect the cellulose structure by increasing the accessible surface area of the cellulosic substrates.

Therefore, it can be recommended from the results that subcritical propane is suitable for fractionating crude palm oil, while CO2 at high pressure can be used to extract oil directly from the fruit pulp. A maximum extraction yield of 92% was obtained after 120 minutes with SCCO2

at 400 bar, 65°C and a flow rate 35 (kg/kg/h). Lau obtained 77.3% oil yield at 300 bar and 80°C [64].

BFigure 3.6: Residues from the extraction with propane at 50 bar, 65°C, 35 (kg/h)/kg.

Figure 3.7: Residues from the extraction with SCCO2 at 400 bar, 45°C, 35 (kg/h)/kg.

b) Solubility of palm oil in subcritical propane and SCCO2

The loading of the solvent during the extraction is presented in Figure 3.8. This value is commonly considered as apparent solubility. This value was calculated from the extraction when the palm oil is easily accessible throughout the fixed bed (at constant extraction rate). In the case of palm oil, SCCO2 has a loading capacity from 0.8 to 2.6%, depending on the extraction condition; while subcritical propane could reach an oil loading up to 4.5%.

0 10 20 30 40 50

30 40 50 60 70

Temperature (°C)

Loading of solvent (g oil/ kg gas)

Figure 3.8: Palm oil loading capacity of subcritical propane and SCCO2.

c) BCo-extracted water

Water is always found in natural plant products. The studied palm mesocarp contained ca.

20% of water. Among the investigated fluids, SCCO2 has the ability to dissolve a small amount of water [136, 137]. Therefore, knowledge about co-extracted water with palm oil in SCCO2 is required. Figures 3.9 & 3.10 show the amount of water and oil co-extracted by SCCO2 at 300 and 400 bar at different temperatures. In the extraction pressure and temperature ranges of the experiments, the extracted oil had only ca. 2 - 4% water content. That value is rather small. Crude palm oil, extracted from palm fruit by pressing, contains more than 10% water with impurities consisting of vegetable matter [54]. As a consequence, the process for removing water from palm oil can be reduced.

Propane 50 bar

SCCO2 400 bar

SCCO2 300 bar

SCCO2 200 bar

0 1 2 3 4 5 6

0 30 60 90 120 150

Extraction Time (min)

Amount of extracted oil (g)

0 0.1 0.2 0.3 0.4 0.5 0.6

Amount of co-extracted water (g)

extracted oil at 65°C extracted oil at 55°C extracted oil at 45°C extracted water at 65°C extracted water at 55°C extracted water at 45°C

Figure 3.9: Extraction of mesocarp with SCCO2 at 300 bar, 35 (kg/h)/kg.

0 1 2 3 4 5 6 7

0 30 60 90 120 150

Extraction Time (min)

Amount of extracted oil (g)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Amount of co-extracted water( g)

extracted oil at 65°C extracted oil at 55°C extracted oil at 45°C extracted water at 65°C extracted water at 55°C extracted water at 45°C

Figure 3.10: Extraction of mesocarp with SCCO2 at 400 bar, 35 (kg/h)/h.

d) BExtraction of tocochromanols and carotenoids

Tocochromanols and carotenoids are interesting valuable minor components in supercritical fluid extraction [62, 72]. It was shown that the amount of carotene extracted from the residue from mechanical processing of palm oil is not high enough to allow an economic industrial size SCCO2 extraction [70]. Therefore, efficient recovery of these compounds during oil extraction is considered as alternative. Concentrations of tocochromanols and carotenoids in the extracted oil depend on their solubilities and those of the other compounds at the same time.

Figure 3.11 shows how the concentration of carotenoids changes with extraction time using SCCO2. Different fractions at different extraction time (20-30 minutes, 45-60 minutes, 60-90 minutes, and 90-120 minutes) were analyzed. It is observed that at a low pressure (200 bar), the carotenoids content increased with extraction time. This agrees with the results reported in the extraction of crude palm oil (CPO) at a similar condition [60]. Therefore, it was also suggested that palm oil can be separated into different carotenoids fractions at the pressure of 200 bar, although carotenoids are not well soluble in SCCO2. In contrast, at a higher pressure (300 or 400 bar), the carotenoids concentration remains constant. The fractionated oil can reach the same carotenoids value as the commercial crude palm oil.

The concentration of carotenoids and tocochromanols for the extraction with propane at 50 bar at different extraction times are presented in Figure 3.12. It was observed that these compounds were extracted at a constant concentration together with the oil during the extraction process.

0 200 400 600 800 1000

30 60 90 120

Extraction tim e (min)

Carotenoids concentration (ppm)

200 bar 300 bar 400 bar

Figure 3.11: Carotenoids as extracted with SCCO2 at 45°C and 14 (kg/h)/kg.

0 200 400 600 800 1000 1200

0 20 40 60 80 100 120 140

Extraction Time (min)

Concentration (ppm)

tocochromanols, 65°C carotenoids, 65°C tocochromanols, 55°C carotenoids, 55°C

Figure 3.12: Concentration of tocochromanols and carotenoids at different extraction times extracted with propane at 50 bar, 35 (kg/h)/kg.

The results above, together with Figure 3.13, show that using SCCO2 or subcritical propane can both co-extract the valuable minor compounds with concentrations in the same range as a normal commercial plant processing palm oil. This agrees with the conclusion from Lau for SCCO2 extraction of dried palm fruits [64]. However, the residual oil contains much less minor compounds than in a screw pressing process.

0 200 400 600 800 1000 1200

SCCO2, 400 bar Propane, 50 bar

Concentration (ppm)

carotenoids in residue oil carotenoids in extracted oil tocochromanols in residue oil tocochromanols in extracted oil

Figure 3.13: Concentration of tocochromanols and carotenoids 55°C, 35 (kg/h)/kg.

oil residue the

in X component of

ion concentrat

oil extracted the

in X component of

ion concentrat X

K _ _ _ _ _ _ _

_ _

_ _ _ _

) _

( 

As shown in Figure 3.13, there is a different behavior in tocochromanols and carotenoids extraction with different fluids. Besides, the composition of palm mesocarp varies with the size and the age of palm fruits. Thus, to objectively evaluate the efficiency of recovery of tocochromanols and carotenoids by using different supercritical extraction techniques at different sample collecting time, the relative comparison of the concentration of these compounds in the extracted oil and in the residue oil was used (Table 3.2). Enrichment factor K of component X is defined as following equation:

(3.5) As a result, the extraction technique with a higher value of K(X) gives a better potential to recover component X.

Table 3.2: Enrichment factors of tocochromanols and carotenoids with different extraction techniques.

Technique Temp. (°C) K (carotenoids) K (tocochromanols)

SCCO2 300 bar 45°C 0.79 0.67

55°C 1.21 0.41

65°C 0.79 0.41

SCCO2 400 bar 45°C 1.19 0.54

55°C 0.98 0.62

65°C 0.90 0.86

Propane 50 bar 45°C 1.24 0.79

55°C 1.14 1.13

65°C 1.27 1.03

Screw pressing

calculated after (Choo, 1996)

ca. 0.12 ca. 0.27

The results in Table 3.2 show that, using supercritical CO2 or subcritical propane, much more tocochromanols and carotenoids can be recovered than with using traditional screw pressing. Recovery of tocochromanols by propane extraction is better than by CO2 extraction, while recovery of carotenoids is nearly the same.

As mentioned in the literature review, the extractions of carotenoids and tocochromanols from fiber residues were not high enough for a possible industrial application. Supercritical fluid extraction of palm oil, simultaneously recovering its high valuable minor components from the palm fruits could open a new way for an alternative extraction method.

3.2.4 Mathematical modeling of the extraction a) BSovova model

The pseudo steady state model of Sovova [12] is applied to describe the modeling of the palm oil extraction process (Figure 3.14).

0 10 20 30 40 50

0 20 40 60 80 100 120 140

Extraction time (min)

Total yield (%)

Figure 3.14: Comparison of experimental data with calculated curves at 35 (kg/h)/kg.

(♦) 300 bar, 45°C; (+) 300 bar, 55°C (▲) 300 bar, 65°C; (■) 400 bar, 65°C; (---) Sovova model.

It can be observed that the experimental data are fitted well with the Sovova model. Palm oil extraction with SCCO2 can be divided into 3 stages with different functions. In the first stage (Constant Extraction Rate, CER), where the oil is easily accessible throughout the fixed bed, the extraction curve is a straight line through the origin. The Falling Extraction Rate period (FER) is