**3. Results**

#### *3.1. Drying Curves and Drying Rate Curves*

Figure 3 shows the change in moisture ratio of lemongrass at different temperatures, from the initial moisture content of 6.40 g water/g dry matter to the constant moisture content. It took 310, 290, 260 and 200 min to completely dry lemongrass at 50, 55, 60 and 65 ◦C, respectively. The results indicate that higher air-drying temperature resulted in greater slope of the curve and shorter drying time. The explanation for this could be two-fold. First, since moisture removal of the material occurs in parallel with moisture diffusion from center to surface material and from the surface to environment, higher air-drying temperature reduces relative humidity on the surface material, which in turn promotes surface evaporation in the drying process [23]. Second, increased air-drying temperature also leads to an improved temperature gradient and surface evaporation rate, accelerating moisture diffusion from the center to the surface. These results are consistent with another study, where the decreased drying time was attributed to increases of the air-drying temperature [24].

**Figure 3.** Drying curves of lemongrass slices at different temperature.

Figure 4 shows the drying rate curves of lemongrass slices at different temperatures. Evidently, higher drying temperature was associated with increased drying rate. Since higher temperature induces more heat transfer to the sample, which in turn leads to increased moisture diffusion to the inside and the outside of the materials, moisture removal was accelerated at higher air-drying temperatures. On the other hand, the drying process pattern described in Figure 4 only exhibited two periods—the initial and falling-rate periods. This is different from a standard drying process consisting of three periods: (i) initial period, (ii) constant-rate period, and (iii) falling-rate period. To be specific, the period starting from the initial moisture content to the moisture content of 5 g water/g dry matter coincided with the initial drying period and the period with moisture of higher than 5 g water/g dry matter represented the falling-rate period. The absence of the constant drying rate could be explained by the thin-layer arrangemen<sup>t</sup> and the high flow of the drying agent, which quickly accelerates evaporation and circumvents the saturation state of the material. Recent studies have also shown that the constant rate period was absent in the drying processes of fruits and vegetables, since this period often occurs very quickly [25]. At very low moisture content, of less than 0.2 g water/g dry matter, differences between drying rates of the four temperature levels were indistinguishable. This could be mainly due to the lack of water after the removal of free moisture, leading to the diversion of thermal energy into the breaking of bonds instead of heating water molecules. However, since the remaining water molecules were strongly bond to the cellulose fibers, increasing drying temperature from 50 to 65 ◦C was inadequate to induce a noticeable change in drying rate for lemongrass materials with low moisture.

**Figure 4.** Drying rate curves of lemongrass slices at different temperatures.

#### *3.2. Mathematical Models of Drying Curves*

Seven common thin-layer drying models (Table 1) were fitted to assess the suitability of experimental data and the results of nonlinear regression analysis are presented in Table 2. The most suitable model to describe lemongrass drying is the model giving highest coefficient of determination (R2), lowest Root Mean Square Error (RMSE), and lowest chi-square (*x*2). From the results of Table 2, all models show that the R<sup>2</sup> values range from 0.93582 to 0.99983 and the RMSE values range from 0.00362 to 0.06887 and *x*2 range from 1.50 × 10−<sup>5</sup> to 5.08 × 10−5. Therefore, any of these models of thin layer drying can be used to estimate the change in moisture content over time [26].


**Table 1.** Mathematical models used to predict the moisture ratios values [25].


**Table 2.** Result of statistical analyses on the modeling of moisture content and drying time.

Based on the results of the statistical analysis and the estimated coefficients of the mathematical models shown in Table 2, the Weibull model achieved the highest values of R2, averaging at 0.99813, the lowest values of *x*2, averaging at 0.000166, and lowest RMSE, averaging at 0.010683. This indicates that the model Weibull is the best model to estimate moisture ratio of lemongrass compared to the other models, followed by the Midilli model in which R2, *x*2, and RMSE averaged at 0.99802, 0.000169, and 0.011065 respectively. Since the Weibull model is rarely included in comparison studies of drying kinetics, the present results are in line with results of Onwude et al. (2016), demonstrating that approximately 24% of the literature sources supports the Midilli model; and with Simha et al. (2016), who affirmed the suitability of the Midilli model in microwave drying of *Cymbopogon citratus* [12,25]. From Table 2, it can also be seen that the predictive power of the Midilli model is weaker than that of the Weibull model in terms of R2, RMSE, and *x*2. At each air-drying temperature, the Weibull model exhibited higher R2, and lower RMSE and *x*2 in comparison with the other fitted models. In addition, the Weibull model estimates showed that the drying constant k0 increased when the temperature increased from 50 to 65 ◦C. This result is consistent with previous studies, where the Weibull model was suggested to be able to well-describe the drying kinetics of fruits and various vegetables such as garlic, quinces, and persimmon [25].

#### *3.3. Estimation of the Effective Moisture Diffusivity*

Figure 5 shows the approximated linear relationship between ln(MR) and drying time (t) at different drying temperatures of 50, 55, 60 and 65 ◦C. Linear regression was used to calculate the effective moisture diffusion coefficient. The effective moisture diffusion coefficient and the R<sup>2</sup> correlation coefficient correspond to each temperature and were calculated by linear regression of experimental value. ln(MR) (dimensionless) with drying time (t) (second) are presented in Table 3.

**Figure 5.** The change of ln(MR) by time (seconds) of sliced lemongrass.

**Table 3.** Deffat different drying temperatures.


As shown in Table 3, the Deff value ranged from 7.64089 × 10−<sup>11</sup> m2/s to 1.47784 × 10−<sup>10</sup> m2/s. This result is within the normal value range of Deff in typical food drying processes, which is from 10−<sup>12</sup> to 10−<sup>6</sup> m2/s [14]. At temperatures of 50 ◦C, 55 ◦C, 60 ◦C, and 65 ◦C, the Deff values were 7.64089 × <sup>10</sup>−11, 1.02741 × <sup>10</sup>−10, 1.16917 × <sup>10</sup>−10, and 1.47784 × 10−<sup>10</sup> m2/s respectively. This indicates that the effective diffusivity coefficient increased proportionally with temperature. At 65 ◦C, moisture content of citronella peaked, since higher temperature quickens evaporation of water molecules on the surface of the lemongrass material.

#### *3.4. Estimation of Activation Energy*

The graph of the change in Deff value (RTa−1) is displayed in Figure 6. Using exponential regression, the activation energy of Ea of sliced lemongrass was determined to be 38.34 kJ/mol. This result is consistent with the normal range of 33.21 to 39.03 kJ/mol in the drying process of basil leaf [25]. To compare, for activation energy of fruits and vegetables, more than 90% of the activation energy values that were found in previous studies ranged between 14.42 and 43.26 kJ/mol, and 8% of the values were in the range 78.93 to 130.61 kJ/mol [25]. The present result of the activation energy for lemongrass moisture evaporation is relatively high. As a result, the separation of moisture from lemongrass could be difficult, suggesting that, in order to completely dry the lemongrass material, either drying duration or drying temperature should be set at high levels.

**Figure 6.** The relationship between change of Deff and RTa−1.
