Drying kinetics of baru flours as function of temperature

Cinética de secagem de farinha da amêndoa de baru em função da temperatura R E S U M O Vários tipos de sementes têm sido introduzidos em formulações na indústria de alimentos, graças ao grande potencial que as proteínas vegetais apresentam. O baru, fruto disseminado no bioma cerrado, apresenta em sua castanha um alto valor nutricional. O presente trabalho teve como objetivo determinar e analisar a cinética de secagem da farinha integral e desengordurada da amêndoa do baru em diferentes temperaturas. A farinha resultante da moagem das amêndoas foi desengordurada por éter de petróleo. As secagens foram realizadas nas temperaturas de 40, 50 e 60 °C. Os dados experimentais foram ajustados aos modelos matemáticos de Page, Henderson e Pabis, Midilli & Kucuk, Thompson e Aproximação da Difusão. Os resultados demonstraram notável efeito da temperatura do ar na cinética de secagem da farinha integral e desengordurada da amêndoa do baru. A farinha da amêndoa integral apresentou perdas de umidade mais lentas do que a farinha desengordurada. Segundo os parâmetros estatísticos de análises, os modelos de Midilli & Kucuk e Page foram os que obtiveram os melhores ajustes dos dados experimentais. Os valores de difusividade efetiva encontrados variaram de 8,02 × 10–10 a 19,90 × 10–10 m2 s-1 e para a energia de ativação foram de 22,39 e 39,37 KJ mol-1 para a amêndoa integral e desengordurada, respectivamente.


Introduction
The Cerrado (Brazilian savanna) is known as the second largest biome in South America.It has a typically hot and semi-humid climate in which there are several species of trees flourishing in the native flora of the region (Sousa et al., 2011;Santos et al., 2016).In this context, the fruits of native plants from the Brazilian Cerrado, such as baru (Dipteryx alata Vog.), have been standing out because they have a nutritional potential with a great sensorial and economic appeal.In addition, baru is used as raw material for the formulation of new food products.Baru seed is also named as chestnut or almond.It is rich in lipids and proteins and it is usually processed and commercialized fresh, roasted or in the form of flour, generating income for several regional communities that live in the Cerrado area, being also very valued by the international market (Takemoto et al., 2001;Rocha & Santiago, 2009;Fernandes et al., 2010).
Taking into account the high lipid contents, baru almond defatting brings about an increase in the protein content, which can be used for the production of several products since the proteins contribute for increasing the nutritional and functional value and the technological properties of the food system (Wang et al., 1999;Ribeiro & Seravalli, 2007).
Another method that enables the concentration of these components is drying.This technique is a complex process in which heat and the mass transfer happen concurrently, reducing moisture and leading to a substantial reduction in mass and volume of the final product.Other benefits associated with food drying are the increase in product lifetime, easiness in transportation and commercialization (Fellows, 2009;Vega-Gálvez et al., 2010;Bettega et al., 2014).
The drying kinetics provides a physical view of the drying process and its principle is based on building a set of mathematical equations which are able to characterize properly the moisture loss as a function of time, in an accurate and simple way, being able to describe the drying process better (Barati & Esfahani, 2011;Rosa et al., 2015).
Until now, only a few papers about the drying kinetics of baru almond have been found in the literature (Teixeira et al., 2015); however, no study was found on the drying kinetics of whole and defatted baru almond flours, emphasizing its defatting.In this context, this paper aimed to determine and analyze the drying kinetics of whole and defatted baru almond flours at different temperatures.

Material and Methods
The experiments were performed in the Laboratory of Engineering and Agroindustrial Processing (LEPA) located at the university campus Deputado Estadual Renê Babour (Mato Grosso State University -UNEMAT), in Barra do Bugres, MT, Brazil.The raw material used was baru almond purchased at the local market from Mato Grosso Southwest region.The almonds were manually selected considering their physical integrity.
Almond milling was performed in a hammer-type food processor (Vieira MCO260) with a granulometric sieve of 0.7 mm.Then, the flour obtained was defatted based on the methodology of Boatright & Hettiarachchy (1995), adapted by substituting the hexane solvent by petroleum ether, because it is an organic solvent and volatilizes completely when exposed to ambient temperature without leaving residues.
The defatted flour and whole flour of baru chestnut were submitted to a washing process with distilled water aiming to remove the non-proteinaceous soluble fractions.Subsequently, a proteinaceous isolate was obtained from the baru almond, adapted from Carvalho et al. (2009).
Drying processes were performed under controlled air temperature conditions of 40, 50 and 60 ºC, in triplicate.The samples were divided into 20-g portions and were uniformly placed in Petri dishes and then in a forced ventilation oven (Quimis Q314M242).During all the drying process, the samples were weighed periodically on an electronic scale (Bioprecisa FA2104n, 0.1 mg precision and four decimal places) until they reached a constant weight.
Moisture content at 105 °C and total lipids analyses were performed according to the Adolfo Lutz Institute (IAL, 2008) protocols.
The drying curves were obtained by converting the water loss data into the dimensionless parameter of moisture ratio (RU), Eq. 1 was used.
RU -Moisture ratio of the product, dimensionless; t -Drying time, h; K -Drying coefficients; a, b, n -Constants of the models

Model designation
Model Page (Page, 1949) RU = e −Kt n (2) Henderson & Pabis (Henderson & Pabis, 1961) Thompson (Thompson et al., 1968) RU = e (−a−(a 2 +(4 .b.t) 0.5 ).(2 .b)) (5) Approximation of diffusion (Ertekin & Yaldiz, 2001) Different mathematical models were used to describe the drying rate of the process.Aiming at obtaining information about the drying kinetics of the baru almond whole and defatted flours, the curves of the moisture ratio as a function of time, were constructed for different drying air temperatures.
The drying curves of the baru almond whole and defatted flours were shown through five mathematical models (Table 1) fitted by non-linear regression using the statistical program XLSTAT (Addinsoft, 2016).
The models were fitted through non-linear regression analyses by using the Quasi-Newton method.The degree (1) of fit of each model took into account the magnitude of the determination coefficient (R²) and the estimated average error (SE).D o -pre-exponential factor, m 2 s -1 ; E a -activation energy, J mol -1 ; R -universal gas constant, 8.314 J mol -1 K -1 ; and, T -absolute temperature, K.

Results and Discussion
The curves shown in Figure 1 2010) (from 3.20 to 4.00%).After washing with distilled water, the moisture was around 68% for the whole flour and 70% for the defatted flour.
Analyzing the drying curves in Figure 1 (C, D, E), there is an evident difference between both flours at the three evaluated temperatures, where the defatted flour reached equilibrium in less time compared with the whole flour.This effect was attributed to higher lipid content in the whole flour (45.55% dry basis) comparing to the defatted flour (4.97% dry basis); therefore, it became more difficult for the water to break the hydrophobic barrier formed by the lipids, which increased the drying time.These results are in accordance with those found by Cyprian et al. (2015) for the Capelin (Mallotus villossus) drying, where the moisture loss was slower for the samples with higher lipid contents.A.
In this way, the lipid content works as a limiting factor during the drying process, acting as a physical barrier to heat transfer, which is responsible for water evaporation as well as its diffusivity from the interior to the surface of the food.
Using the dimensionless moisture data from Figure 1 (A, B), it is possible to fit them with the mathematical models shown in Table 1, as well as determine the coefficient of determination (R²), estimated average error (SE) and verify which model is the best to represent adequately the drying process of the baru almond samples.
Table 2 shows the parameters of the mathematical models fitted to the experimental data of whole and defatted almond flours through non-linear regression at the three temperatures, as well as their coefficients of determination (R 2 ) and estimated average error (SE).
It is possible to identify that, for the analyzed models, the estimated average error (SE) of the moisture ratio, which describes the value of the standard deviation for the estimate, has relatively low values.It is also possible to observe that high determination coefficients (R²) were obtained, higher than 90%, indicating a successful representation of the drying process in the studied conditions (Table 2).
As noted in Table 2, the value of the drying constant k increased with the temperature rise in almost all samples, which occurs because higher temperatures result in higher drying rates, reaching the equilibrium content in a shorter process time.These results were also observed by Corrêa et al. (2010) with coffee drying.
All models fitted well to the experimental data, mainly the Midilli & Kucuk model for the whole flour and Page model for the defatted flour, since both had R² values closer to 100% of the curve fitting and lower SE value for the samples.
The Figure 2 shows a graphical representation of the mathematical models which fitted best to the data for both types of samples at the three temperatures.
Based on the parameters found through the best data fits with the mathematical models, an analytical procedure was done and it was possible to represent graphically the moisture variation rate in relation to time in both raw materials, shown in Figure 3.
The curves in Figure 3 (A, C, E) describe the moisture loss rate in relation to time, highlighting a meaningful difference in the drying kinetics of the whole flour compared to the defatted flour.The comparison of moisture loss between both raw materials is represented in Figure 3 (B, D, F).This difference is related to the lipid content of the sample, since the whole flour showed a lipid content of 45.55% (dry basis) and the defatted flour of 4.97% (dry basis).These aspects are in accordance with the experiment done by Cyprian et al. (2015).
Using the Fick's law equation (Eq.8) for products with a flat plate geometric shape, the values of the effective diffusivity were calculated from the experimental data.The effective diffusion coefficient values increased when the drying air temperature increased, which demonstrates a reduction of the internal resistances to the drying processes.They were 11.90 x 10 -10 and 8.02 x 10 -10 (40 ºC), 15.30 x 10 -10 and 12.90 x 10 -10 (50 ºC); and 19.90 x 10 -10 and 19.50 x 10 -10 (60 ºC) for WBF and DBF, respectively.12.70 to 110 KJ mol -1 .In general, it is possible to say that the higher the temperature, the faster the activation energy will be overcome and, consequently, the food begins to lose its moisture more quickly.

Conclusions
1.The increase in the drying temperature led to higher water removal rates of the product.
2. The flour of whole almond showed slower moisture loss than the defatted flour and the lipid content acted as a limiting factor during the drying process.
3. The Midilli & Kucuk model had the best results for the whole flour and Page model had the best fits for the defatted flour.
4. The defatting contributed to the reduction of the drying time of the samples, being the most indicated process to be executed industrially.
pre -moisture ratio predicted by the model; RU exp -experimental moisture ratio; and, N -number of observations made during the experiment.The values of average effective moisture diffusivity were determined by analytical solution of Fick's law for liquid water diffusion in on solid, taking into account the conditions of the material in question.The activation energy (Ea) was obtained from the dependence of effective diffusivity (D ef ) on temperature, analyzed by an Arrhenius-type equation, Eq. 8.
(A, B) indicate the effect caused by the increase in air temperature through the drying kinetics, facilitating the energy transfer in the form of heat to the samples, which consequently increases the moisture removal rate of the product.This trend can be usually observed in drying experiments.These results are in accordance with other studies, such as Andrade et al. (2006), who worked with drying kinetics in bean seeds; Costa et al. (2011) with crambe seeds; Santos et al. (2013) with urucum flour; and Teixeira et al. (2015) with drying of whole baru almonds.The initial moisture values found for the whole and defatted baru almond flours were 3.22 and 3.51%, which are within the range found by Vera et al. (2009) from 2.93 to 5.07%, Lima et al. (2010) (3.23%) and Fernandes et al. (

Figure 1 .
Figure 1.Drying curves of whole (A) and defatted (B) baru almond flours at 40, 50, 60 °C and a comparison of the two drying curves under temperatures of 40 (C), 50 (D) and 60 ºC (E) WBF -Whole baru flour; DBF -Defatted baru flour; K -Drying coefficients; A, b, n -Constants of the models; R 2 -Magnitude of the coefficient of determination; SE -Estimated average error;

Figure 4 .
Figure 4. Graphic representation for the effective diffusivity (D ef x 10 -10 ) as a function of the drying air temperature (A) and Arrhenius representation for the effective diffusion coefficient (B) for both baru flours

Table 1 .
Mathematical models used to describe the drying process

Table 2 .
Mathematical models fitted to the experimental data at drying temperatures of 40, 50 and 60 °C for whole and defatted baru almond flours