b.; W0 is the

initial moisture content, g 100 g−1, d.b.; n is the constant drying rate, s−1; t is the drying time, s; tcr is the critical time, transition from first to second drying period, s; Wc is the critical moisture content, g 100 g−1, d.b.; k is the coefficient of the decreasing drying rate, s−1. For long drying times and very thin samples (<0.005 m), only first-order Fick's law (Equation (2)) was considered for the effective diffusion coefficient, considering product geometry as a flat plate, according to Baroni and Hubinger (1998). equation(2) Def=K4L2Π2 Where Def is the effective diffusion coefficient, m2 s−1; L is the material thickness, m; and k is the drying rate coefficient, s−1. Tenofovir in vivo The degree of fit of the model considered GDC-0449 concentration the magnitude of the coefficient of determination (R2), the magnitude of the average relative error (P) and the standard deviation of the estimate

(SE). The average relative error and standard deviation of the estimate for each model were calculated according to Equations (3) and (4), respectively. equation(3) P=100n∑r=1n|Y−Y0|Y equation(4) SE=∑r=1n(Y−Y0)2GLRWhere Y is the experimentally observed value, Y0 is the value calculated by the model, n is the number of experimental observations, and GLR is the number of degrees of freedom of the model. For the coefficients obtained by fitting of the model (W0, n, Wc) and the diffusion coefficient, regression analyses were carried out at 5% probability of error

using the response surface method, with non-significance of the lack of fit as criterion so as to obtain the best relationship between the parameters measured with the contents of yam starch and glycerol at each temperature. Values of R², P and SE of the models adjusted to the 11 treatments and 5 drying temperatures are presented in Table 2. The coefficient of determination (R2) ranged from 99.91 to 99.99, always very close to 100%. The coefficient of determination alone MycoClean Mycoplasma Removal Kit is not a good criterion for the selection of nonlinear models; therefore, the values of the average estimated error (SE) and average relative error (P) were considered ( Madamba, Driscoll, & Buckle, 1996). Standard deviation of the estimate ranged from 0.167 to 0.958, where values lower than 1 indicate good fit to the model. Average relative errors encountered for all models at the temperatures evaluated were less than 10%. The P values indicate the deviation of observed values in relation to the curve estimated by the model ( Kashani-Nejad, Mortazavi, & Safekordi, 2005). Values lower than 10% are recommended for selecting models ( Mohapatra & Rao, 2005); therefore, the fitted model proved to be adequate for the observed data. These moisture content data were plotted with respect to time (Fig. 1 obtained for treatment 1 and fit to the proposed model).