Effect of moisture content on selected physical properties of shea kernel of varying slice thickness

Bulk density, kernel density, particle porosity and shrinkage of sheanut krnel slices were determined as a function of moisture content and slice thickness in order to establish preliminary design parameters for drying in an indirect solar dryer. Solid density increased linearly while bulk density decreased nonlinearly during drying as the moisture content decreased from 150 to 5 %d.b. The porosities of the kernels ranged from 60.12 -67.91, 57.05-64.87 and 56.88-62.49%, respectively, for the 5, 10 and 15 mm slices. The volumetric shrinkage of the shea slices reduced with decrease in moisture content for all the particle sizes evaluated in this study. Among the shrinkage models tested, the empirical quadratic model gave the best fit of the experimental data. It is proposed that the mechanism of shrinkage in sheanut kernel slices is linked to the removal of water that leads to structural collapse.


INTRODUCTION
The shea tree has long been reported as important livelihood tree (Maranz et al., 2003).This is because in addition to the medicinal value its leaves and roots has, the tree produces fruits with high oil content (45-50%) usually referred to as shea butter.In the processing of shea fruits to obtain butter, drying of the kernels after depulping and dehusking has been described as one of the key steps indispensable in the production process (Womeni et al., 2004).This is because sheanut kernels at harvest contain about 45-60% moisture content on a wet mass bases and are therefore prone to deterioration if not properly preserved.The kernel size varies from one locality to the other.For example, it measures 28.3 by 20.6 mm with respect to length and diameter in Chad (Mohagir, 2010), 36.63 by 27.93 mm in Nigeria (Olajide et al., 2000).In Cameroon, it measures averagely 45 by 30 mm (Bup Nde et al., 2012).Sheanut kernel is therefore one of the biggest oilseed in the world.Traditional drying of the whole kernel is carried out either under the sun or in solar dryers.This may take up to 20 days which therefore prolongs the processing time of these kernels.Drying experiments of whole kernels carried out in an electric dryer at 45°C reduced drying times to about 12 days but allowed the development of a surface coat which hindered uniform drying of the kernels (Kapseu et al., 2007).Drying for more than 12 days may equally lead to the development of moulds which contain lipases that catalyse hydrolyses thereby leading to rancidity in the resultant oil (Kapseu et al., 2007).To reduce this drying time it is advisable to dry these kernel as thin slices in solar dryers which can be cheap and easily affordable to local African women, the main processors of shea fruits to butter.The understanding and modelling of the drying behaviour of sheanut kernel slices and the development of a solar *Corresponding author.E-mail: bupdiv@yahoo.fr.dryer for sheanut kernel slices requires the mastery of its physical properties such as densities, porosities and shrinkage.Although some of these physical properties have been reported in the literature for shea fruits, nuts and kernels (Olajide et al., 2000;Aviara et al., 2005;Bup Nde et al., 2012) those of sheanut kernels slices remain very scarce.
An important aspect of the study of shrinkage involves modelling so that they can be incorporated into drying models to study the effect of shrinkage on the drying behaviour of the product.Many of such models have been reported in the literature and one needs to make a judicious choice of which one best suits the product under study.The aims of this work were therefore to: 1. Determine the bulk density, solid density, porosity and volumetric shrinkage of sheanut kernel slices as a function of moisture content and slice thickness and 2. Model the volumetric shrinkage phenomenon in sheanut kernel slices.

Sampling and sample treatment
Sheanuts obtained from Tchabal Ngaoundere were cooked in water using a laboratory oven at 80°C for 120 min (Bup Nde et al., 2012).They were then withdrawn from water, allowed to cool on a laboratory bench overnight, cracked and the kernels were cut into 5, 10 and 15 mm thick slabs using a Tommy Slicer (model Siemens, Erlangen, Germany).The sheanut kernel slices were then dried in an indirect solar dryer (drying air temperature 45 ± 5°C; airspeed 1.4 m/s, air realative humidity 55 ± 5%) for predefined periods of time and the resulting moisture contents was measured by the oven method.Results of the physical properties were then expressed as a function of moisture content.

Determination of bulk density of the kernels
The bulk density was determined using the AOAC (1980) method.This involved the filling of a 500 ml cylinder having an inner diameter of 49.54 mm with sheanut kernel slices to a height of 15 cm and weighing the contents.The bulk density b in kg/m 3 was given by: [1] where Vb is the bulk volume.These experiments were carried out for the 5, 10 and 15 mm thick slices.Each experiment was replicated four times at each particle size.

Determination of solid density of sheanut kernel slices
To determine the true or solid density of the shea kernel slices, an analytical balance (model Scout Pro SPU402, OHAUS, USA) adapted for this purpose was used.The balance was set to the specific gravity mode.A spring was attached to the balance from which the sample tied to a string of negligible weight was hung.The weight of the sample was taken in air.The sample was then immersed into a beaker of water placed on the balance and the new weight was taken.The specific gravity of the sample was then determined from the relation (as indicated in the User's Manual of the balance): This was then converted to solid density given that the density of water at 22°C is 1 g/cm 3 .At each moisture content, five kernel slices were used for each determination.The experiment was replicated thrice at each moisture content.These studies were carried out on 5, 10 and 15 mm thick sheanut kernel slices.

Determination of porosity
The porosity was calculated from the solid and bulk densities using the relationship given by Pabis et al. (1988).The porosity, , was given by the equation:

Determination of volumetric shrinkage of sheanut kernel slices
Volumetric shrinkage ratio was determined by measuring the volume of the kernels at each moisture content and defining the ratio V/V0 as volumetric shrinkage ratio.V0 is the corresponding volume of the kernels at the initial moisture content X0, V is the corresponding volume of the kernels at moisture content X.Both empirical and fundamental shrinkage models with and without porosity terms were tested and some selected to describe the shrinkage behaviour in sheanut kernel slices.The first group of these models included empirical linear (Equation 3) and quadratic models (Equation 4): The second group included the Vacarezza (1975) (Equation 5) and Suzuki et al. (1976) (Equation 6) models which are fundamental models without porosity: and Figure 1.Influence of moisture content on the particle density of sheanut kernel slices.
The third group was the modified Perez and Calvedo model (Mayor and Sereno, 2004) (Equation 7) which included a porosity term, that is: a, b, c, d, e are constants of the empirical models, qi are parameters of fundamental models (variable), Xe is the equilibrium moisture content and рe is the equilibrium density, р0 is density at X0, рs and рw are the densities of the dry solid and water respectively.q5 is a constant when the equilibrium density (рe) and the equilibrium moisture content (Xe) are known.ε and ε0 are the porosities at moisture contents X and X0.

Validation of models
The criteria for evaluating the reliability of the simulations were the correlation coefficients and/or the standard relative error of deviation observed on the moisture content between the experimental and theoretical results.The standard relative error (SRE) of deviation of theoretical from experimental results was determined from Equation 8. [8] where Yexp and Ymod are the values obtained from experiments and from the model respectively.P is the number of points at which measurements were carried out.

Solid density of sheanut kernel slices
The solid density increased linearly during the drying period as the moisture content decreased from 150 to 5% d.b. (Figure 1).The values ranged from 1.086-1.160,1.098-1.131and 1.086-1.163g/cm 3 for the 5, 10 and 15 mm slices, respectively, in the moisture range studied.Solid density is simulated with linear equations with regression coefficients of 0.956, 0.951 and 0.963 for the 5, 10 and 15 mm thick kernels, respectively.The increase in solid density with decrease in moisture content could be due to the fact that, during the drying process, the kernels lost water and became more compact and were therefore capable of packing more regularly and closely in the measuring cylinder.This regular and closed packing could have been acompanied by a more rapid decrease in volume as compared to its mass and led to increased solid density which obviously increased with decrease in moisture content.The solid density for all the slices at all moisture contents was greater than unity, implying that the slices are heavier than water.This property can be very useful in the design of cleaning and separation equipments for the slices.No clear relation was established between solid density and particle size as the solid density was higher for the 10 mm thick slices at higher moisture content (150-100% d.b.) as compared to the 5 and 15 mm thick slices.The trends were reversed as the moisture content decreased below 100% d.b.The variation of solid density with moisture content was  modelled with a linear equation with the regression coefficients greater than 0.95% and the SRE less than 5%.

Bulk density of sheanut kernel slices
Figure 2 shows the influence of moisture content on the bulk density of sheanut kernels slices.The bulk density of the sheanut kernel slices decreased non-linearly as the moisture content decreased from 150 to 5% d.b.In the final stages of drying, the bulk density slightly increased.
The non-uniform decrease of bulk density with moisture content could be due to the fact that the bulk volume and mass of the kernels might not have changed uniformly in the course of drying.The increase in bulk density in the final stages of drying could be attributed to the fact that, at that stage, the mass of the product became constant due to the limitation of the movement of tightly bound water as observed in the drying of sheanut kernel slices (Kapseu et al., 2007).However, its volume continued to decrease due to the unbalanced pressure between the inner portion of the product and the outer environment thereby leading to an increase in the mass to volume ratio (bulk density).The bulk density decreased from 0.435 to 0.374, 0.472 to 0.399 and 0.468 to 0.429 g/cm 3 for the 5, 10 and 15 mm slices, respectively, in the moisture range studied.From Figure 2, it was equally observed that the bulk density was generally dependent on the particle size, increasing as the particle size increased.

Porosity of sheanut kernel slices
The porosity of the kernel slices increased non-uniformly with moisture content in the range 150-5% d.b. (Figure 3).The porosities of the kernels ranged from 60.12 -67.91, 57.05-64.87 and 56.88-62.49%,respectively, for the 5, 10 and 15 mm slices.The porosities decreased with an increase in kernel size at all moisture contents.The slight decrease in porosity observed at the final stages of drying was properly due to the decrease in bulk volume at that stage as explained for bulk density above.The variation of porosity with moisture content of sheanut kernel slices observed in this work were different from trends reported by Aviara et al. (2005) for sheanuts in which porosity of the nuts increased with an increase in moisture content to a maximum value and then decreased sharply afterwards.

Volumetric shrinkage of sheanut kernel slices
The volumetric shrinkage of the shea slices reduced with decrease in moisture content for all the particle sizes evaluated in this study.It is clearly observed from Figure 4 that shrinkage of the kernel slices decreased nonlinearly as the moisture content of the kernels decreased.The higher decrease in volume observed at higher moisture contents could be linked to the high elastic behaviour of the material still completely in its rubbery state.Under these conditions, they were sufficiently elastic to shrink into the space left by the evaporated water.This elasticity reduced as more and more water was evaporated from the product.The material became more rigid and led to a reduction in the rate at which the product shrunk.This suggested that shrinkage was probably caused by the volume of water removed.As earlier mentioned, five models were tested to fit the data for shrinkage (Vacarezza, 1975;Suzuki et al., 1976); empirical linear and quadratic models and the modified  Perez and Calvedo model (Mayor and Sereno, 2004).Of these models, the empirical quadratic model gave the best fit of the experimental data as indicated by its very high R 2 and very low SRE (Table 1).This model was therefore used to describe the shrinkage of the sheanut kernel slices.content in the investigated moisture content range indicated that, during the drying process shrinkage is not essentially uniform.This assertion was buttressed by the fact that more than 60% of the samples were visually cracked after the drying experiments.Mayor and Sereno (2004) stated that surface cracking occurs during drying when shrinkage is not uniform leading to the formation of unbalanced stresses and failure of the material.A uniform shrinkage of the kernel slices could be linked to the fact that phase transition (change from a rubbery to the glassy state) in the material through out the drying process may have been highly limited or negligible.That is, if the material remained mostly in the rubbery state through out the drying process, then shrinkage almost entirely compensated for moisture lost and the volume of the material decreased linearly with moisture content.However, if the material passed from a rubbery to the glassy state (phase transition), the mobility of the solid matrix reduced and the rate and extent of shrinkage decreased significantly leading to deviations from linearity.This last phenomenon was observed for the sheanut kernel slices investigated in this work.Deviations of shrinkage from linearity especially at low moisture contents have been reported for garlic, potato and sweet potato (Ratti, 1994) and for apple and potato (Wang and Brennan, 1995).So in this work, it is proposed that the mechanism of shrinkage in sheanut kernel slices is linked to the removal of water that leads to structural collapse.Water removal becomes highly limited in the later periods of drying.This culminates into deviations from linearity which could be attributed to phase transitions.Further work is needed on the glass transition temperatures of sheanut kernel slices to verify this assertion.

Conclusion
This work reported some of the physical properties of sheanut kernel slices as a function of moisture content.Apart from the solid density whose variation was linear, bulk density and porosity varied non lineraly with mois-ture content and differed from those of whole large kernels in the mass range 21-28 g.Shrinkage was modelled by an empirical quadratic equation and it is proposed that the mechanism in sheanut kernel slices is linked to the removal of water that leads to structural collapse.The results of these physical properties can be employed in modelling drying processes for an eventual design of a dryer for sheanut slices.

Figure 2 .
Figure 2. Influence of moisture content on the bulk density of sheanut kernel slices.

Figure 3 .
Figure 3. Influence of moisture content on the porosity of sheanut kernel slices.

Figure 4 .
Figure 4. Influence of moisture content on the volumetric shrinkage of sheanut kernel slices.

Table 1 .
The modified Perez and Calvedo model gave very low R 2 and high SRE values indicating that it could not be used to describe shrinkage in sheanut kernel slices.The values of its constants are therefore not reported.The failure of the modified Perez and Calvado model could be attributed to the fact that, the porosity considered in this study was external porosity (voidage) instead of intrinsic porosity of the kernels.The non linear variation of volumetric shrinkage with moisture Model constants and R 2 and SRE values for equations used in modelling shrinkage of sheanut kernel slices.