The Optimization and Mathematical Modeling of Quality Attributes of Parboiled Rice Using a Response Surface Method

1College of Engineering, Nanjing Agricultural University, Nanjing, Jiangsu 210031, China 2College of Engineering, China Agricultural University, Beijing 100083, China 3Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA 4Department of Structure and Environmental Engineering, University of Agriculture, Faisalabad, Pakistan 5Nanjing Research Institute for Agricultural Mechanization, Ministry of Agriculture, Nanjing 210014, China


Introduction
Rice is a staple food consumed by more than half of the world's population.The world's annual production of rice in 2014 was estimated at 744.4 million tons.The Asian continent plays a vital role in rice production, with rice production in this region recorded as 673.6 million tons in 2014, accounting for 90% of the world's overall rice paddy production.Researchers are working to find the most efficient way to process rice to preserve its nutritional value and to determine the best germination conditions.In general, parboiling is a hydrothermal process used on rough rice and includes various treatments to improve rice quality.Parboiling consists of a series of processing stages including the soaking, steaming, and drying of paddy rice grains.The parboiling process has resulted in increased milling yield and enhanced nutritional value and resistance to spoilage [1,2].
The major advantages of the parboiling method include rice starch gelatinization and hardening of the rice kernels, which reduces endosperm germination and breakage losses during milling [3].Researchers have used a variety of techniques to parboil rice.Unnikrishnan and Bhattacharya [4] described pressure parboiling as low moisture parboiling in which the rice paddy is partially soaked or simply washed with water and then steamed at high pressure to gelatinize the rice starch.Walter et al. [5] used conventional and laboratory steaming methods in which rough rice was placed at 1.5 kg/cm 2 for 25 minutes, or, in the laboratory method, rough rice was steamed at ambient pressure for 20 minutes using an autoclave.The effectiveness of the steaming process depends on the steam conditions (either saturated in an open system or superheated in a closed system), steam pressure, and steaming time [6,7].Another study revealed that, during the parboiling process, the steaming time has an influence on the physical characteristics of both the Japonica and Indica rice varieties.Moreover, steaming at 1 kg/cm 2 pressure for 15-30 minutes and steaming at atmospheric pressure for 1-2 minutes improved the cooking qualities, milling qualities, and physical characteristics [8].The HRY should be increased after successful parboiling [9].Bello et al. [10] reported that as steaming time was increased 45-75 min, the hardness of individual rice kernels was found to increase from 13.1 to 25.3 N. Islam et al. [11] revealed that average hardness increased with increasing steaming time (60 min at 90 ∘ C and 10 min at 100 ∘ C).The effect of the parboiling process on the color value has been reported by many researchers [1,12,13].The color value usually increases during parboiling while lightness decreases; the magnitude of the decrease in lightness increases with the severity of the parboiling process [14].
The response surface method (RSM) is an excellent approach used by many researchers to optimize the various factors in their research studies.Sharma et al. [15] reviewed the combined effects of various components on biomass growth using RSM.Response surface method is a unique combination of mathematical and statistical approaches to evaluate the optimal conditions by reducing the number of experiments between factors and responses [16].Taghinezhad and Brenner [17] optimized the effect of two independent variables (soaking temperature and steaming time) on the quality characteristics of parboiled rice.Danbaba et al. [18] used RSM to optimize parboiling conditions (soaking temperature, steaming time, and drying time) to maximize HRY.
The scope of the study covers the optimization and mathematical modeling of rice quality properties during parboiling process by considering three independent variables (soaking temperature, soaking time, and steaming time).To the best of our knowledge, no optimization study has focused on predicting the combined effect of three factors (soaking temperature, soaking time, and steaming time on HRY, hardness, cooking time, lightness, and color) during parboiling process.Therefore, the current study was conducted to fill this research gap.This study was carried out to investigate (1) the optimum condition for rice quality parameters affected by process indicators (soaking temperature, soaking time, and steaming time) during parboiling process; (2) the prediction of parboiled rice quality indicators by mathematical modeling affected by three independent variables.

Materials and Methods
This study used a popular local Indica rice variety in China, Rong Youhua Zhan (RYZ), with an amylose content of 20.4% and an initial moisture content of 13 ± 1% w.b.The RYZ rice was collected from the Rice Research Institute of Guangdong Academy of Agricultural Sciences, China.After proper cleaning and shade drying at safe moisture content, paddies were packed in 1 kg zip lock polyethylene bags and stored at 5 ± 1 ∘ C in a refrigerator (BCD-232TDek, Hisense, China) for further experimentation [19].

Parboiling Process and Brown Rice
Preparation.Hot soaking was done using an electronically controlled water bath (DK-600A, China), adding 200 g of rice sample to 1000 ml of deionized water, for a rice to water ratio of 1 : 5.The samples were soaked according to the five levels presented in Table 1, with stirring at 10-minute intervals.Extra water was discarded, the samples were tempered for 30 min, and the open ends of the beakers were covered to prevent rapid cooling.
The soaked samples were treated at a working pressure of 1 kg/cm 2 (121 ∘ C) for different durations as shown in Table 1 using an autoclave (LS-35HD, China).After steaming, the pressure was released and the rice was tempered for 20-30 min [20].The moisture content of the Indica samples was found to be within the range of 20.34-32.11%.However, to avoid excessive fissuring, the rice samples were shade dried at room temperature (25-30 ∘ C, 60 ± 5% RH) for 2 to 3 days to reduce the moisture content to 13 ± 1% [21,22].The moisture content of the rice samples was measured by an oven drying method at 105 ± 1 ∘ C for 24 h (w.b.).After that, the samples were coded according to the treatment applied and stored for moisture content equilibration and hardness stabilization.

Head Rice Yield.
To measure percentage HRY, each 200 g sample of the whole paddy was dehusked using the laboratory satake rubber roller type rice husker (THU35C, China).The rice was polished with an abrasive type polisher for 90 seconds using satake CBS300AS, Japan [23].Prior to HRY measurement, the broken grains were separated from the whole paddy and cleaned.The HRY was measured thrice and expressed as a percentage of the whole grains with respect to the paddy [10].

Hardness.
The hardness was measured in terms of bending force.Hardness was measured by the three-point breakage mechanism using TMS-Pro texture analyzer, USA.The load of 500 N was applied to each grain at the 5% accuracy level to measure the force (N) using the texture lab procomputer program.As shown in Figure 1, the rice kernels were placed horizontally on the base plate with a clearance of 3.0 mm and a flat loading head of 1.0 mm thickness was used with a deformation rate of 0.5 mm/s.Fifty healthy whole grains were randomly selected for hardness testing from each sample, and the average values were reported as the final hardness [2].

Cooking Time.
The cooking time was determined by the method proposed by Juliano [24].Ten grams of whole kernels was weighed using an electronic balance (JM-A20002, China), mixed with deionized water, and then cooked for 20 minutes in vigorously preheated boiling water.Afterwards, ten rice kernels were randomly selected and pressed between two glass plates to identify translucent kernels.Samples were taken after every two-minute interval.This procedure was repeated until 90% of the whole kernels were found to be translucent, and this time was recorded as the cooking time [25].

Lightness and Color
Measurement.The precision color reader (HP-200, China) was used for the color measurement.The instrument was first calibrated with white and black plates provided by the manufacturer.The results were   expressed in terms of  * ,  * , and  * values, where the  * value represents brightness from black to white, a positive " * " value represents redness while a negative " * " value represents greenness, and a positive " * " value represents yellowness while a negative " * " value represents blueness.Six measurements were performed on each sample by shaking the sample gently to avoid error during the measurement, and average values were determined for further analysis.The color value () of parboiled rice was determined using the following [11,12,26]: 2.6.Experimental Design.RSM combined with central composite design (CCD) was used to investigate the effect of three factors on the response surface in the area of study.A CCD with three factors at five levels was investigated [27,28].The factors were selected based on reports in the literature that the factors with the maximum effect on HRY are soaking temperature, soaking time, and steaming time [17].In the current study with the three independent variables of soaking temperature, soaking time, and steaming time, a total of 20 experiments were required, calculated as 2  + 2 + 6, where  is the total number of factors.
The responses could then be related to the factors by linear or quadratic models, enabling optimization.A quadratic model that also includes a linear model is shown in the following [28]: The CCD was executed using the statistical software Design-Expert5 8.0.6 (Stat-Ease, Inc., USA).The fourteen experiments were investigated with a combination of six replicates to estimate the pure error.The run numbers and experimental conditions arranged by CCD with coded and actual values are shown in Table 1.The Design-Expert program was used to process the data using (2), and analysis of variance (ANOVA) was used to find the interactions between independent variables and responses.The fitness of the model was expressed by the coefficient of determination  2 , and its significance was also determined by an -test.

Results and Discussion
3.1.CCD and Model Development.The process of paddy parboiling consists of three major steps, soaking, steaming, and drying of rough rice before milling.The benefits of the parboiling process are generally to obtain maximum HRY, increase nutritional value, and achieve resistance to spoilage by insects.Drying is a major factor in the paddy parboiling process, but the current study focused on the three factors of soaking temperature, soaking time, and steaming time.These three factors were chosen as independent variables, and their effects on the responses of HRY, hardness, cooking time, lightness, and color were determined.The actual responses and predicted values of each run are shown in Table 2.
The regression equations in coded form with the correlation coefficient  2 , adjusted  2 , and each coefficient of ( 2) are presented in Table 3.In the current study, a quadratic model suitably described the responses, and the  2 values were acceptable.However, the hardness, cooking time, and color were positively affected by steaming time while lightness was negatively affected, all of which were highly significant at ( < 0.01), but steaming time did not affect the HRY.The soaking temperature and time mainly exerted significant effects ( < 0.01) on HRY, hardness, and cooking time, with no effects on lightness and color.Examining the interactions between factors, the interactions of soaking temperature with steaming time and with soaking-steaming time negatively affected the HRY and positively affected lightness.Furthermore, lightness was positively affected by the soaking temperature-time combination, and no other significant interaction effects were observed.

Head Rice Yield.
The maximum HRY of 73.88% was obtained on run four when the soaking temperature, soaking time, and steaming time were 70 ∘ C, 150 min, and 5 min, respectively, and the minimum HRY of 60.82% was attained on the first run when these factors were 40 ∘ C, 90 min, and 5 min, respectively, as shown in Table 2. Software analysis showed that the optimized values for soaking temperature, soaking time, and steaming time were 69.88 ∘ C, 150 min, and 6.73 min, respectively, resulting in a predicted HRY of 73.43%.The RSM generated regression equation for HRY% is presented in Table 3.
HRY% was obtained as a function of soaking temperature ( ∘ C), soaking time (min), and steaming time (min) (Table 3).Furthermore, the significance of the developed quadratic equation was cross-checked by an -test.The analysis of variance (ANOVA) for the quadratic model indicated that model was significant (prob >  0.0001 less than -value 29.22).The goodness of fit was evaluated by the coefficient of determination  2 , which ranges from 0 to 1 and should be more than 80% for a well-fitted regression model [29].For HRY,  2 and adjusted  2 are 0.96 and 0.93, indicating the appropriateness of the modeled regression equation for HRY if these variables are combined [17].The nonsignificant lack of fit indicates that the model provides a good fit; the lack of fit -value was 0.56 for HRY.Moreover, the current model was appropriate to capture the design space.
The response surface curves for HRY obtained from Design-Expert software are shown in Figure 2. The interactions among three independent variables were studied to maximize the percentage HRY.Figure 2(a) shows the 3D representation of the interaction between soaking temperature and soaking time.HRY was significantly improved by increasing soaking temperature and soaking time while the steaming time was at its center point.
The effect of soaking temperature and steaming time is depicted in Figure 2(b).The following interaction showed that, initially, HRY increased with increasing soaking temperature and steaming time up to a certain point; after this point, HRY decreased with gradually increasing steaming time, as previously expected.The experimental results for HRY in the current study were in agreement with [22].
The steaming time is the most important factor in the parboiling process.The percentage HRY was found to be minimal at the lowest soaking-steaming time combination (60.82%) as shown in Figure 2(c).As soaking time increased, HRY also increased up to a particular steaming time, after which HRY was observed to decrease with a further increase in steaming time.This result was expected because long steaming time has been reported to reduce HRY [6].In addition, at longer steaming times, rice grains absorbed excessive water, and husks were also observed to split.Furthermore, the parboiling process, in particular the steaming time, would also affect the gelatinization of starch.If the degree of gelatinization is high, then deformed grains would be obtained during parboiling, and deformed grains would be lost along with the exuded part of the endosperm [11,30].According to Marshall et al. [31], it is not recommended to parboil paddies for a longer steaming period to increase the HRY.Similar results were found in the current study, with a maximum HRY (73.88%) obtained at a lower steaming time (5 min).

Hardness.
Hardness is one of the most important rice quality characteristics during the parboiling process because milling breakage decreases with increasing hardness.The rice obtained by the parboiling process is harder than unprocessed rice [32].Figures 3(a), 3(b), and 3(c) depict the relationships between hardness and process variables.Hardness increased with the severity of the parboiling process [33].Figure 3(a) shows a linear relationship between hardness and the soaking temperature-time combination.Figures 3(b) and 3(c) shows that the hardness of parboiled rice increased with increasing steaming time, as the moisture absorbed by the rice grains during steaming causes starch melting and a decrease in Δ at longer steaming durations [34].In addition, increased severity of parboiling increases starch gelatinization, leading to low-quality rice [35].The bending force ranged from 25.24 to 32.62 N, comparable to the findings described by [2,10].The RSM determined the regression equation for hardness in coded units shown in Table 3.

Cooking Time.
The relationship between cooking time and the independent variables was captured in a quadratic model ( 2 = 0.90 and adjusted  2 = 0.82).The cooking time was influenced by all three factors, but soaking temperature and soaking time had significant effects while the effect of steaming time was not significant.The maximum (35 min) cooking time was observed for run eight (Table 2), with the minimum (23 min) observed for the first run.Figure 4(a) depicts the soaking time and soaking temperature interaction, showing that decreasing these two factors reduced the cooking time of parboiled rice.The relationship between cooking time and the independent variables is presented in Table 3.

Lightness.
Lightness is an important quality indicator for the rice parboiling process; market value is directly related to lightness as consumers dislike dark colored rice [36].
The effects of soaking temperature-time and steaming time are shown in Figure 5. Lightness decreased as the severity of the parboiling process increased.The steaming duration had a significant effect on the lightness of parboiled rice as shown in Figures 5(b) and 5(c); the combination of soaking time-temperature and steaming time reduced the lightness values.The darkest value (77.2) was observed for run fourteen (Table 2) when soaking and steaming times were 120 and 18.41 min, respectively.The maximum brightness (85.5) was observed at the lowest parboiling condition when the soaking temperature-time and steaming time were 40 ∘ C, 90 min, and 5 min, respectively.The regression equation for lightness determined by the RSM is shown in Table 3.

3.6.
Color.The parboiling process profoundly affects the color of the processed rice. 2 and adjusted  2 for the equation describing color were found to be 0.94 and 0.89, respectively, and the lack of fit was also nonsignificant, indicating that the current quadratic model was found to be suitable to explain the relationship.Figures 6(a), 6(b), and 6(c) show the interaction effects of the soaking temperaturetime and steaming time combination on the color value.The color value ranged from 11.80 to 14.56, with the darkest color rice obtained at the highest steaming time (18.41 min), which can also be seen from Figures 6(b) and 6(c).The soaking temperature slightly affected the rice color, but steaming time greatly affected the rice color, as longer duration soaking and steaming led the rice bran pigments to penetrate into the endosperm [13,37].The reaction of released sugar with an amino acid of grain during parboiling could lead to discoloration of rice [38].The color finding in the current study revealed that greater severity of parboiling led to the development of dark colored rice during parboiling.The regression equation generated by the Design-Expert for color is presented in Table 3.   13.43 + 0.077

Conclusion
The RSM was used to jointly optimize the set of responses.The optimal values were obtained for responses by solving the regression equations of independent variables using Design-Expert software.In this study, the optimized combinations of soaking temperature, soaking time, and steaming time were determined primarily to improve HRY and other quality attributes.HRY is the most important rice quality characteristic.Therefore, in the Design-Expert software, HRY was maximized while other quality parameters were maintained within the appropriate ranges.color is 12.24, with a composite desirability of 0.9658.Variation in the factors studied here highly influenced the HRY and other quality traits.Increased soaking temperature and soaking time above the optimum condition resulted in lower HRY as explained in the earlier section.Similarly, increased steaming time affected color and lightness and increased hardness and cooking time due to the severe gelatinization of starch.The optimal condition maximizes HRY and maintains other quality properties within acceptable ranges.The RSM combined with the CCD was successfully applied to the optimization of rice quality indicators (HRY, hardness, cooking time, lightness, and color) during the parboiling process.The  2 values for all responses were greater than 80%, and the lack of fit test was also observed  to be nonsignificant for all response variables.Rice parboiled under the following conditions (soaking temperature = 69.88∘ C, soaking time = 150 min, and steaming time = 6.73 min) is predicted to exhibit optimal HRY and other quality indicators.The results showed that the RSM could be used to obtain the optimal HRY and other quality indicators of parboiled rice.The current experimental results could be used by the rice processing industry to significantly improve rice quality.

Additional Points
Practical Application.The primary purpose of parboiling process is to restore rice nutritional value for the quality point of view.The cost of parboiling process depends on a range of parboiling process indicators (soaking temperature, soaking time, and steaming time); if these variables are high then cost will be increased.Thus, there is a need to find optimal condition of these variables to decrease the process  cost and improve plant turnover for better rice quality.For this purpose, a mathematical model was used to predict the quality properties (HRY, hardness, cooking time, lightness, and color) by parboiling process indicators.The parboiling industry could use the findings of a current study to get the desired quality of parboiled rice.Therefore, optimization of parboiling process is economically essential for the rice industry.The manuscript will be helpful for researchers working particularly on commercializing the parboiled rice.

AFigure 2 :
Figure 2: Effect of soaking temperature, soaking time, and steaming time on percentage HRY.

Figure 3 :
Figure 3: Interactional effect of three variables on hardness of rice kernels.

Figure 4 :
Figure 4: The cooking time affected by soaking temperature, soaking time, and steaming time.

Figure 5 :
Figure 5: Effect of soaking temperature, soaking time, and steaming time on lightness.

ADesignB
: s o a k in g te m p e r a tu re C : s t e a m i n g t i m e Factor coding: actual X1 = A: soaking temperature X2 = C: steaming time Actual factor B: soaking time = 120.00: s o a k in g ti m e C : s t e a m i n g t i m e Factor coding: actual X1 = B: soaking time X2 = C: steaming time Actual factor A: soaking temperature = 55.

Figure 6 :
Figure 6: Effect of soaking temperature, soaking time, and steaming time on color (-value).

Table 1 :
Design of CCD for independent variables.

Table 2 :
Observed and predicted values for HRY, hardness, cooking time, lightness, and color of rice.

Table 3 :
Regression equations in coded form with  2 Based on RSM data analysis, the predicted optimum soaking temperature, soaking time, and steaming time were 69.88 ∘ C, 150 min, and 6.73 min.Under these conditions, the predicted HRY is 73.43%, hardness is 29.95 N, cooking time is 32.14 min, lightness is 83.03, and