© T. Silou et al., Published by EDP Sciences, 2017

1 Introduction

The palm oil tree is endemic to the Congo basin in the Gulf of Guinea. Africa was the principal producer of palm oil in the 1960s, but since 2000 it has been largely overtaken by Asia, in particular by countries such as Indonesia (18.3 million tons) and Malaysia (17.4 million tons), which accounted for two thirds of the world's palm oil production (41 MT) in 2007–2008. China, India, Pakistan, Bangladesh and Egypt are the main palm oil purchasers (Dronne and Forslund, 2009). The consumption of palm oil rose from 11 kg to 24.5 kg per person per year between 1976 and 2006 (Corley, cited by Rival, 2013).

The palm oil tree is exceptionally productive, with an average yield of 3.5 t of oil per hectare. This production reaches 7 t/ha in certain plantations in Asia and 10 t/ha in experimental stations. Covering 7% of the crop areas occupied by oil seeds, the palm oil tree produces 30% of all vegetable oils; palm oil presents the lowest manufacturing costs for vegetable oils, 20% lower than for oils from soya. Up to 80% of palm oil is used in foodstuffs (salad oil, frying oil, margarines, fat for baking, pastry-making and other food preparations), about 19% in oil chemistry (cosmetics, soap, lubricants and greases, candles, etc.) and 1% as biomass fuel (biodiesel) (Rival, 2013).

Rival (2013) reports that out of the 21 million hectares of primary forest that disappeared in Indonesia between 1990 and 2005, only 3 million were used to create palm plantations, contrary to general belief. Environmentally friendly palm oil cropping is possible on a reasonable scale.

In Cameroon, palm oil meets 80% of edible oil needs, with a deficit of 20,000 t; in Congo-Brazzaville, more than 80% of the local palm oil consumed comes from domestic production (Silou, 2012). Likewise in Gabon. This local production mobilizes a significant portion of the actors in the agrifood sector, particularly women. However, the processes used are arduous and the quality of the oil obtained remains substandard.

The local populations of the Congo basin have developed a technique of oil extraction, consisting in the manual mixing of crushed palm nuts with hot water. Oil is collected by gravity after cooling. This process uses a mixer instead of a press. A pilot unit, in the course of optimization, including an ecological palm oil tree plantation (1 ha) and using a 300 L mixer has been set up in Congo Brazzaville. This paper, which presents the first results of this work, is meant to help small producers in the search for an alternative process, based on local knowledge, for a better production of red palm oil.

2 Materials and methods

2.1 Plant material

2.1.1 Experimental palm oil plantation

For teaching purposes an experimental palm oil plantation was implemented at Loukoko, by the Higher School of Technology “Les Cataractes” (Rural Campus of Loukoko (RCL)) in 1997 with the improved pre-germinated seeds of the tenera variety purchased by the Research Centre and originating from palm oil trees at Pobè in Benin.

This plantation was set up, with a spacing of 10 m × 10 m, in a valley on clay-sand soil on the Plateau des Cataractes (Congo-Brazzaville) exposed to the lower Congo climate, and covering about 1 ha (95 palm trees).

2.1.2 Raw materials of the oil extraction: Palm nuts

Three varieties of palm nuts are often found in the Congo basin (Fig. 1): the dura variety, with a thick hull, thin pulp and poor yield, the pisfera variety with a thin hull, thick pulp, generally sterile, and the hybrid variety tenera, with a thin hull, thick pulp and high oil yield, a crossbreed of dura (male) and pisifera (female).

To characterize nuts, the length (main axis) and the width (small axis) of the fruits were measured using a calliper, and the masses of nut, pulp and seed were obtained by weighing with a balance.

Fig. 1 Palm oil nuts (varieties: dura, pisifera and tenera).

2.2 Traditional process of palm oil extraction

2.2.1 Domestic process

In the Congo basin, housewives transform palm nut into mwambe (emulsion of water in oil) for the preparation of local dishes.

The nuts cooked with water are crushed in a mortar (or pot); the crushed nuts are mixed manually with hot water to give an emulsion of water in oil (Fig. 2), called mwambe or mosaka in Congo Brazzaville. This intermediate product of the process is used to prepare vegetable sauces (Bandzouzi, 2010); the emulsion is heated on a low fire, and the oil separates on cooling; the supernatant oil is collected with a suitable instrument such as a ladle.

Fig. 2 Mwambe: emulsion of water in oil obtained from crushed nuts mixed manually with hot water (Franceville, Gabon).

2.2.2 Small-scale process

The small palm oil producers have adapted the domestic process at craft industry scale (100–200 kg of palm nuts), working in 200 L reclaimed barrels.

The harvested bunches are stored for 3–4 days to let the nuts soften, which facilitates the separation of the nuts and the extraction of the oil.

The nuts are steamed for 2–3 h with 50 L of water to sterilize them (sterilizer step). This steaming also allows easy separation of the pulp and seed.

The oil is extracted from the crushed pulp after a series of mixing steps in hot water (100 L in all at about 60 °C).

The mixture obtained is gently heated in a clarifier until all yellow foam has disappeared. The mixture is then left to settle and the supernatant dark red oil is skimmed off the aqueous phase.

Sterilization is carried out in a 200 L barrel (200 kg of nuts). The mixer (Fig. 3a) is made of a half-barrel equipped with a hardwood axle on which iron mixing bars are set at right angles. The clarification (defoaming) and the decantation take place in half-barrels.

Fig. 3 Artisanal mixers: (a) mixer used by local producers and (b) improved mixer, Rural Campus of Loukoko (RCL).

2.2.3 Improved small-scale process

The Higher School of Technology “Les Cataractes” improved the small-scale process by introducing (i) equipment to make the process of extraction of palm oil less arduous, and (ii) procedures and good practices to guarantee the quantity and quality of the final product.

The sterilizer, the mixer (Fig. 3b) and the decanter of capacity 300 L each, are made of black plate (mild steel); the sterilizer and the decanter are designed to be heated with wood or gas.

The extraction process chain, in a motorized version, will be made of food quality stainless steel at end of the optimization process.

2.2.4 Modelling of the improved small-scale process: Principal and interaction effects of the factors

The variables indexed here, which influence the extraction yield, are: quantity of nuts, volume and temperature of the mixing water, time of clarification (decantation) and extraction process.

The dependence of the response of palm oil extraction Y to three unspecified variables X₁, X₂ and X₃ is expressed mathematically by: with Y the response; F the function of the response and X₁, X₂ and X₃ variables or factors.

The experimentation thus consists in highlighting the effects of these factors on the response.

The factorial design on two levels as developed by Davies (1954) seems sufficiently well-suited to solving this kind of problem, and has the advantage of calling on only very elementary mathematical knowledge (Ortigosa, 1993).

For this design, the mathematical function connecting the response to the factors is a first degree polynomial for each factor, with interaction effects of factors: The factors being centred and reduced, the coefficients have simple meanings:

b₀ average of the response, b_i principal effects of the factors, b_ij effects of the interactions of order 2, and b_ijk effects of the interactions of order 3.

The experimental design and the matrix of experiments are shown in Tables 1 and 2, according to whether the real or reduced/centred variables are considered.

2.3 Products obtained

In two independent consumer surveys conducted in 2009 at Brazzaville (Congo) by Degbia (Degbia, 2010) and in 2013 by Pamba (Pamba-Boundena, 2012) at Franceville, Lambaréné, Oyem (Gabon), samples of oil resulting from the domestic and small-scale processes were collected for a summary evaluation (as an indication) of the quality of the oils consumed in the Congo basin, particularly in the localities concerned.

Samples from the pilot unit (RCL) during the crop seasons 2014–2015 and 2015–2016 underwent physical and chemical analysis (acid values (AV), peroxide values (PV), carotenoid and polyphenol contents, FAs).

2.4 Determination of the chemical characteristics (AV, PV)

The acid values (ml KOH/g) and the peroxide values (mEq. O₂/kg) were measured using IUPAC methods, (1979).

2.5 CG and HPLC analysis

2.5.1 Determination of the fatty acid (FA) composition

The AV (ml KOH/g) and the PV (mEq. O₂/kg) were measured using IUPAC methods, (1979). The composition of the oils was determined by gas chromatography using a gas chromatograph (type HP 5890) equipped with a non-polar column (HP 5M, 30 m long, 0.25 mm inside diameter and 0.2 μm thick) and with a flame ionization detector (FID) in the following experimental conditions:

• Carrier gas helium with constant flow of 1 ml/mm;

• Oven temperature ramp 50–280 °C, at 5 °C/min;

• Injector temperature 250 °C;

• Detector temperature 280 °C;

• Quantity injected 1 μl.

The FAs were identified by their respective retention times, and quantified by their peak areas, in the presence of an internal standard at a known concentration (C17:0).

2.5.2 Determination of triacylglycerol (TAG) composition

The TAG profile was obtained by high performance liquid chromatography (Héron et al., 2006). The chromatographic system consisted of a model LC-6A pump (Shimadzu, Kyoto, Japan), a model 7125 injector valve with a 10 μl loop (Rheodyne, Cotati, CA, USA) and a model Sedex 75 light-scattering detector (Sedere, Alfortville, France). The nebulizing gas was air at 3.5 bars, the evaporation temperature was 37 °C and the gain (PM) 11. The column temperature was controlled using an Igloo-CIL column oven (Cluzeau, Sainte-Foy-la-Grande, France) set at 30 °C. An evaporative light-scattering detector (Sedere Sedex 75) was used. AGs were separated at 20 °C using a Kromacil C18 (5 μm), 250 mm × 4.6 mm column (Part Number KR5C18-25QS mfg #806610) and were eluted from the column with an acetonitrile/dichloromethane mixture (70/30–40/60) (v/v/v). Chromatograms were recorded using Azur (v 3.0), data acquisition software (Datalys, Saint-Martin-d'Heres, France).

2.6 Evaluation of carotenoid content by UV visible spectrometry

Carotenes were assayed on a spectrophotometer with a 1% oil solution in chloroform using the external calibration curves for beta-carotene (standard) plotted by Davis et al. (2007/2008) at 460, 490 and 520 nm. The linear regression equations at these various wavelengths are:

• to 460 nm: % carotenoid = 4.57 Abs (R² = 0.989);

• to 490 nm: % carotenoid = 5.60 Abs (R² = 0.989);

• to 520 nm: % carotenoid = 25.5 Abs (R² = 0.989).

2.7 Assay of total phenolics by the Folin–Ciocalteu method

Total phenolics were assayed by the Folin–Ciocalteu method (Singleton et al., 1999). The external calibration graph uses gallic acid between 0 and 500 mg/L in a methanol–water mixture as solvent (0, 50, 100, 250, 500 mg/L).

2.8 Statistical processing

The descriptive statistics and graphs were run on Microsoft Excel 8.0, the multivariate statistics (Principal component analysis (PCA)) on XLSTAT (www.xlstat.com) and the modelling of the extraction on NEMRODW (lprai@nemrodw.com).

3 Results and discussion

3.1 Characterization of palm nuts and oil

3.1.1 Fruits of the tenera, dura and pisifera varieties (RCL)

Of the four African varieties of palm nuts identified by Adriens in 1951 (macrocarpa, dura, tenera and pisifera), three occur to ranging extents in the Congo basin, and two are of particular economic interest: dura and tenera.

The pisifera variety, considered as sterile, is very seldom found in the wild; about 10 fruits were studied, as shown in Table 3.

It is the smallest of the three varieties, with an average length of 31 ± 4 mm, width 19 ± 2 mm, and mass 7.1 ± 1.1 g compared with tenera (length 34 ± 3 mm, width 20 ± 2 mm, mass: 8.3 ± 1.1 g) and dura (length 32 ± 5 mm, width 24 ± 3 mm, mass 12.1 ± 2.7 g). With a sphericity index of 0.6, the fruits of pisifera resemble the fruits of tenera more than they do those of dura (Tab. 3).

The improved tenera variety, generally used for industrial plantations, is now spreading in rural area, particularly in co-operative structures and family businesses.

The fruits of a given bunch of tenera are very similar, as reflected in their low coefficients of variation in length, width, and mass of nuts (Tab. 3). PCA, built on five variables describing nuts, does not permit any classification; fruits are fairly evenly sized in any given bunch (Fig. 4a). The index of sphericity close to 0.5 indicates quasi-spherical globular fruits (Tab. 3).

The tenera variety is the one of most interest for the production of palm oil (oil of pulp).

The dura variety is endemic, and used by the local populations in the domestic production of palm oil. The seed of dura, which represents on average 72% of the fruit mass (Tab. 3), is of commercial value; its kernel also yields significant quantities of oil (Tab. 4).

The oil contents of dura pulp and kernel, from the various localities and according to the AFNOR method (1981) (Tab. 4), highlight very high and very narrowly variable contents (coefficient of variation (CV) = 8%) for pulp (mean: 82.4%) and high and slightly more variable contents (CV = 10%) for kernel (mean: 48.9%). The dura variety is thus of interest for its pulp (palm oil) and its kernel (kernel oil).

Fig. 4 Principal component analysis (PCA), built with 5 variables: repartition of individuals on the first principal plan F1F2: (a) tenera variety and (b) dura variety.

3.1.2 Palm oils

The 42 samples collected in Congo (Brazzaville (BZV)), Loukoko (LOUK)) and in Gabon (Franceville (FCV), Lambaréné (LAMB), Oyem) during surveys were analysed for an indicative evaluation of their qualities (Tab. 5).

The oils studied seem to vary broadly in status, with a high variability of the two indices considered, namely acid value and peroxide value.

The AV of palm oils from Gabon, collected at markets or in households, or extracted in traditional units, vary over a fairly narrow range (7–10 units), with an average of 7.7 ± 1.4. These values are common for crude oils produced locally, but exceed those recommended by the standards.

The situation is completely different in Congo, where the AV of the samples extracted with the equipment and process at the RCL range from 3 to 10, with an average of 6.1 ± 1.8. These values are in line with those obtained in Gabon, and those reported in the literature, for crude palm oils (Tan et al., 2009). By contrast, samples collected at the market and in the households and which were stored for ranging times present very high AV between 6 and 35, with an average of 28.7 ± 1.8.

These values highlight a deterioration of the oil quality, due to hydrolysis of the TAGs during storage, probably because of the presence of water, enzymes and impurities.

However, these oils, collected from 14 households in Brazzaville and resulting from small-scale production, starting from tenera and/or dura varieties, have an extremely stable FA composition. Table 6 summarizes the composition of these oils, with 16 FAs identified on the samples studied, of which more than two thirds in trace amounts. Four FAs make up almost all the oil (98%).

They are oils with a low palmitic acid content (P, on average 38.0 ± 2.9%), rich in oleic acid (O, on average 38.8 ± 5.0%) and present a very favourable content of linoleic acid (L, on average 13.9 ± 5.5%).

This stable FA profile contrasts with the very high variability and high levels of the AV.

Pulp and kernel of Eleais guineensis give completely different oils. On the other hand, the same parts (pulp or kernel) of dura and tenera varieties have similar FA profiles (Tab. 7).

Table 7 gives the composition of FA of palm and kernel oils are discriminated by the value of: Relative to this indicator, the oils from pulp of tenera and dura are nearly the same (0.9 and 1.0); those of seeds contain more saturated FAs than those of pulp (2.7 and 4.4). For seeds, the oil of the dura variety contains more saturated FAs than oil of the tenera variety. Theoretically, the proportion of solids, in oils that are in principle semi-solid, should increase in the order: dura pulp oil ≈ tenera pulp oil < tenera seed oil < dura seed oil.

However, the pattern is more complex in practice: the physical state of the oil is also determined by the quantity of saturated FAs present, and its extraction conditions, in particular residual water and suspended solid impurities. Drying and filtration/centrifuging are then required.

3.2 Modelling of the small-scale extraction of the palm oil

3.2.1 Comparison of the small-scale method and the improved small-scale method (RCL)

The process of palm oil extraction used at RCL seeks to improve the traditional process used by the local producers of palm in Plateau des Cataractes (Congo) and largely used throughout the Congo basin.

The performance of the two processes was evaluated using a 2³ full factorial design (Tabs. 1 and 2). Table 8 shows the responses of the process: palm oil extraction yields (y₁) and palm oil AV (y₂).

The first degree model with interactions led to the following expression: with an average response of 42.8%.

The statistics for the coefficients (giving R² = 0.985 and R²A = 0.897) validate the model selected a posteriori.

Only the mass of nuts has a negative effect on response; the method effect, which has a positive value, compensates exactly for the mass effects. The temperature during the mixing process presents the strongest effect in absolute value and is positive (+6.250).

The effects of second order interaction are all negative with the highest absolute value for the masses/temperature interaction (−5.250).

The third order interaction has a weakly positive effect (+0.750).

Examination of the second order interactions from two different full factorial plans (yield = f (method, quantity of nuts, temperature of mixing water) and yield = f (method, quantity of mixing water, temperature of mixing water) highlights the superiority of method B (small-scale improved RCL) over method A (small-scale “local producer”); in all the cases, the extraction leads to a higher yield for B than for A. Increasing (i) nut quantity or mixing water quantity decreases extraction yield, while increasing mixing water temperature increase the yield (Fig. 5).

The average of the responses, the principal and interaction effects of AV (I_A) are given in the following expression: The average of four AV, provided by the model, indicates a good quality of crude oil (freshly extracted). This result validates the process studied a posteriori.

The effects are more marked on I_A (i.e. quality) that on the yield (quantity).

Domestic extraction on very small quantity of fresh palm nuts (<10 kg) leads to a very good yield and good quality palm oil. The quantity and the quality fall drastically on a small scale (≥100 kg).

The twofold challenge is (i) to make the process less arduous with larger quantities of treated nuts (by greatly reducing effort) and (ii) to solve the problems arising from the conditioning and storage of the oil, which are essentially responsible for the poor quality of crude oil produced in the Congo basin.

Fig. 5 Graphic representation of the second order interactions effects (method/quantity of nuts (a), method/quantity of mixing water (b), and method/temperature of mixing water method (c) of palm oil extraction at RCL (Congo).

3.2.2 Modelling of the improved small-scale method (RCL)

The data used and the responses of the 2³ full factorial design are summarized in Table 9.

In a first degree model, the effects of factors are given by the following polynomial for the yield: The average oil extraction yield is about 40%, all principal effects of factors on the response are positive, all second order interaction effects are negative, and the single third order interaction acts positively.

There is a considerable margin for a quantitative improvement of the process to approach 50% yield, obtained by the small producers on very small quantity of nuts.

The model used for acid value gives the following expression: The process gives an oil with an average acid value of 6.17. The quantity and the temperature of the mixing water have a positive effect on this index, while the duration of the decantation process has a negative effect. The positive effects are overall twice higher, in absolute values, than the negative ones.

There are also good prospects for qualitative improvement of the process with the extraction equipment in the course of optimization.

3.3 Quantitative and qualitative evaluation of the production of palm oil by the improved small-scale process (RCL)

3.3.1 Quantitative evaluation

The RCL crop season covers the period from September (year n−1) to May (year n). Table 10 gives the results of the market years 2014–2015 and 2015–2016. Figure 6 recapitulates the evolution of this production in a graph.

This figure makes it possible to date the beginning and end of the crop season, and also to highlight that about half of the oil produced during a market year was produced within 1 month.

The near-constant oil extraction yield (CV = 7.2% for the marketing year 2014–2015 and CV = 3% for 2015–2016) shows an excellent control of the process by the workers in charge of the extraction at the RCL. With an extraction yield of 49% in 2014–2015 and 56% in 2015–2016, the RCL extraction process using a mixer is much better than most of the processes used for palm oil extraction.

Fig. 6 Nut and red palm oil productions of tenera variety from RCL (Congo) (Crop seasons: 2014–2015 and 2015–2016).

3.3.2 Fine chemical characterization of the palm oil from RCL

A detailed study of the glyceride and unsaponifiable fractions of RCL palm oil gives the results shown in Tables 11 and 12.

The palm oil studied consists of nearly 91.6% of triglycerides, 6.7% of diglycerides and 1.7% of free fatty acids FFAs). Two classes of the identified TAGs constitute 88.6% of the total (T50 and T52), the remainder counts two others (T48: 3.8% and T54: 7.5%). It is a palmito-oleo-linoleic type oil with %P ≥ %O>%L (POL), and a cumulated content of more than 90%. Stearic, myristic, linoleic, arachidic and palmitoleic acids are minor components (Tab. 11).

This profile is consistent with that reported in the literature for Malaysia (Tan et al., 2009 and cited references) and Ivory Coast (Mondé et al., 2008).

Red palm oil has a high carotenoid content (610 mg/kg). The beta sitosterol is the most abundant sterol (69.7%) for a total sterol content of 44.1 mg/100 g and tocopherols together with tocotrienols (tocols) is present at 76.1 mg/100 g with γ-tocotrienol expressed this value into mg/100 g (46.0%) as a major constituent (Tab. 12).

3.3.3 Qualitative evaluation of the oils produced at RCL (2014–2016)

• (i)

  Physical and chemical characteristics

Crop seasons 2014–2015 and 2015–2016 used the improved small-scale process (RCL), and the oil obtained was systematically analysed at the exit of the oil mill (48 h at most for 2015–2016, and 1–2 months maximum for 2014–2015 for the physical and chemical characteristics). Five samples from Franceville produced by local producers (Palm Plantation MVOUMA) were also examined.

The AV measured range from 2 to 12, with an average roughly between 4 and 10 for the great majority of samples (Fig. 7).

These studies also establish that after a strong increase (more than 100%) in the first week following the extraction, these values fall back to the starting level. This could explain why the indices measured at 48 h after the extraction are of the same order of magnitude as those measured several months after extraction.

The values obtained are higher than those of WHO standards; they can be improved by minor innovations in the process (e.g. quality of extraction water, or drying and purification of oil by centrifugation).

The presence of free water and suspended impurities, which support the hydrolysis of the TAGs, explains the increase in AV in the first 2 weeks after extraction.

In a study from Cameroon, Pondo Pondo (2010) notes, here also, a very marked increase in the IP during the first month following the extraction, and a fall in this value with the decomposition of the peroxide formed, returning to starting values beyond 2 months.

• (ii)

  Composition in FAs and structure of the triacyglycerols (TAGs)

The FA composition of the palm oils extracted from the tenera variety (RCL) closely resemble that of tenera and dura oils produced by the small producers and purchased on the market of Brazzaville (Tabs. 6 and 13, Fig. 8)

Palmitic and oleic acids alternately take first and second place in the samples studied, and stearic and linoleic acids follow in third and fourth positions. These four FAs constitute more than 95% of total FAs.

Palm oil is representative of oils and fats frequently found in the Congo basin, and are palmito-oleo-linoleic type oils (POL).

With nearly 50% of saturated FAs, Dubois et al. (2008) classify palm oil in the saturated FA group and under the palmitic acid sub-group, while Salas et al. (2009) classify them among oils rich in palmitic acid.

Since the other 50% of palm oil is composed of unsaturated FAs, we previously suggested distinguishing between a mixed-fat group, a significant source at the same time of saturated FAs, mono-unsaturated FAs and poly-unsaturated FAs, alongside the groups that are sources of only one FA or one family of FAs: saturated, mono-unsaturated or polyunsaturated (Silou, 2014).

Such a classification allows a better evaluation of the nutritional utility of the palm oil, and similar oils in the Congo basin, e.g. Raphia laurentii, R. sese, Dacryodes edulis, Canarium schweinfurthii and Luffa cylindrica.

This utility should take into consideration the presence of a very high content of oleic and linoleic acids, unsaturated FAs, as a positive nutritional contribution (Dubois et al., 2008).

When using TAGs as classification elements of fats, we distinguish external (sn1,3) from internal (sn2) bonding positions of FAs on glycerol: this distinction casts a new light on the status of palmitic acid in the palm oil, and in other palmito-oleo-linoleic type oils.

As reported by Lecerf (2013), palm oil has only 11% of its palmitic acid in position 2, and 87% of the FAs in position 2 are unsaturated (O and L), so that most of the palmitic acid is in position 1–3.

Palmitic acid can increase LDL cholesterol (“bad” cholesterol), particularly when it is in the internal position on glycerol.

Stereospecific analyses of the POL type oils establish that palmitic acid is fixed preferentially on the 1–3 positions of glycerol, and oleic and linoleic acids, preferentially in position 2, with an advantage for linoleic acid, according to Litchfield on the bonding priority of FAs on TAGs (Litchfield, 1972).

Palmitic acid occupies 1–3 positions in oils of raphia (Silou et al., 2012) and safou (Kinkéla and Bézard, 1993; Silou and Kama-Niamayoua, 1999, Kinkéla and Silou, 2004). Oleic and linoleic acids preferentially occupy the internal position, with a slight advantage for linoleic acid, which is di-unsaturated.

The stability of the palm oil FA profile leads to a stable TAG profile, around a standard composition presented in Table 14. Two TAGs, PPO and POO, constitute 72% of the total TAGs. Five other TAGs are in the list, with a cumulated contribution of 23% and an individual contribution in the range 4–7%: POL, PPL, (PPP + SOO), PSO. Only PPP is formed by three saturated FAs (P), carrying palmitic acid in the internal position of glycerol. The other TAGs containing at least one unsaturated FA have this position preferentially occupied by this acid.

• (iii)

  Carotenoid content

Carotenoid contents were determined on 1% oil in chloroform at 460 nm using an external calibration curve of beta-carotene over an absorbance range from 0 to 3. The curve due to Davis et al., 2008/2009: % carotenoid = 4.67 × absorbance (r² = 99) gives a carotenoid content in the range 400–1000 μg/ml, consistent with those in the literature, which generally range between 500 and 700 ppm (Swoboda, 1985; Tan, 1989), and between 800 and 3575 ppm for the most powerfully improved hybrids at certain research centres (Mondé et al., 2008).

Variations in carotenoid contents of the oils studied are shown in Figure 9. The carotenoid content values of the main samples range from 500 to 1000 μg/g, attesting the good quality of the oils studied.

• (iv)

  Polyphenol content

Polyphenol contents were determined according to the method of Folin-Ciocalteu (Singleton et al., 1999) using an external calibration curve plotted with gallic acid for concentrations ranging from 0 to 600 ppm and yielding the equation:

Results obtained on oils from Congo and Gabon are reported in Figure 10. The samples studied, which are rich in polyphenol compounds, present good anti-oxidant properties (4–14 mg/g).

These values are close to those obtained for oils of colza (5–7 mg/g), soya (2–7 mg/g) and sunflower (4–9 mg/g) by pressing followed by solvent extraction (Régis et al., 2016).

They are higher than that of the olive oils extracted by traditional processes in Tunisia (0.2 mg/g, Mirouane et al.), and much lower than that obtained on improved palm plantations in an experimental research station (64–130 mg/g) by Mondé et al. (2008).

Fig. 7 Variation of acid values (AV; ml KOH/g) and peroxide values (PV, mEq. O2/kg) for palm oils samples of tenera variety from Rural Campus of Loukoko (Congo) and oil Palm Plantation Mvouma (Franceville Gabon).

Fig. 8 Variation of fatty acid Composition (%) of palm oils of tenera variety from the Congo (RCL crop season: 2014–2015) and Gabon (Plantation Mvouma).

Fig. 9 Variation of caroteinoïd contents in samples of tenera variety from RCL (Congo) and Palm Plantation Mvouma (Gabon).

Fig. 10 Variation of polyphenol contents in samples of tenera variety from RCL (Congo) and Palm Plantation Mvouma (Gabon).

4 Conclusion

Red palm oil, extracted by the traditional method, is very widely consumed in the countries of the Congo basin particularly in Cameroon, the Congo, DR Congo and Gabon. This traditional production process is arduous, and the product quality is poor compared with standards (low yield of 25% with the dura variety, high acid and PV, low carotenoid content, etc.). Substantial improvements made to the process during this work on the equipment (powerful mixer, steam sterilizer, modern clarifier and modern decanter) and on the process (modelling) permit gains in both quantity (yield around 50%) and quality (low acid value of around 4; very low peroxide index of around 6; high carotenoids (600–1200 ppm) and polyphenols (4–15 mg/g)). Other improvements are in progress, including (i) the use of food-quality stainless steel, (ii) motorization of the mixer, and (ii) the introduction of a centrifugal filter for oil drying, should further improve the quantitative and qualitative gains already obtained at mid-course.

Acknowledgements

The authors thank Dr. Z. Mouloungui (Research Director, INRA/INP-ENSIACET Toulouse, France) and his team, for the detailed analysis of palm oil from the RCL (Congo Brazzaville), and Dr L.C. Obame-Engonga and Dr. S.A. Ondo Azi, from the University of Sciences and Technology of Masuku (USTM), Franceville (Gabon) for their important contribution in the part relating to polyphenols.

References

All Tables

All Figures

	Fig. 1 Palm oil nuts (varieties: dura, pisifera and tenera).
In the text

	Fig. 2 Mwambe: emulsion of water in oil obtained from crushed nuts mixed manually with hot water (Franceville, Gabon).
In the text

	Fig. 3 Artisanal mixers: (a) mixer used by local producers and (b) improved mixer, Rural Campus of Loukoko (RCL).
In the text

	Fig. 4 Principal component analysis (PCA), built with 5 variables: repartition of individuals on the first principal plan F1F2: (a) tenera variety and (b) dura variety.
In the text

	Fig. 5 Graphic representation of the second order interactions effects (method/quantity of nuts (a), method/quantity of mixing water (b), and method/temperature of mixing water method (c) of palm oil extraction at RCL (Congo).
In the text

	Fig. 6 Nut and red palm oil productions of tenera variety from RCL (Congo) (Crop seasons: 2014–2015 and 2015–2016).
In the text

	Fig. 7 Variation of acid values (AV; ml KOH/g) and peroxide values (PV, mEq. O2/kg) for palm oils samples of tenera variety from Rural Campus of Loukoko (Congo) and oil Palm Plantation Mvouma (Franceville Gabon).
In the text

	Fig. 8 Variation of fatty acid Composition (%) of palm oils of tenera variety from the Congo (RCL crop season: 2014–2015) and Gabon (Plantation Mvouma).
In the text

	Fig. 9 Variation of caroteinoïd contents in samples of tenera variety from RCL (Congo) and Palm Plantation Mvouma (Gabon).
In the text

	Fig. 10 Variation of polyphenol contents in samples of tenera variety from RCL (Congo) and Palm Plantation Mvouma (Gabon).
In the text