Process Optimization of Fluidized Bed Drying for Water Spinach: Evaluating the Effect of Blanching Through RSM and ANN Models

Abstract

The quality of the dried leafy vegetables, such as water spinach (Ipomoea aquatica), has been found to be significantly affected by the drying process in terms of moisture content and retention of important nutrients, namely vitamin C and β-carotene. There is great potential for fluidized bed drying to be applied for leafy vegetables in optimizing process parameters for maximum nutrient retention since it has not been researched. This work investigated the effect of temperature, drying time, and bed thickness on the nutritional quality of blanched and unblanched water spinach samples. In the present study, the fluidized bed drying process has been designed and optimized using a Central Composite Design (CCD) and Response Surface Methodology (RSM). For this study, both RSM and artificial neural network (ANN) predictive models are developed for further comparison. Using a multiobjective desirability function, the best-optimized response was given from the experimental model for the responses of moisture content, vitamin C, and β-carotene retention. Appropriate statistical metrics are applied, for example, AARD (Absolute Average Relative Deviation), MRD (Mean Relative Deviation), MSE (Mean Squared Error), and R ² (Coefficient of Determination), which helped in model comparison during the study. It was observed from the experiment that all the response variables are significantly affected by drying temperature, time, and bed thickness. Variation of bed thickness in the blanched samples affected > 16% in moisture content attainment compared to unblanched samples, and vitamin C content exhibited a variation of more than 25% due to changes in bed thickness for blanched samples on the contrary. RSM has shown a better performance than ANN in its precision and prediction power. The optimized drying conditions came out to be 60°C as the drying temperature, 7.19 min as the drying time, and 5.12 cm as the bed thickness, which resulted in 2.95% of moisture content, 5.99 mg/100 g vitamin C, and 139.16 μg/g of β-carotene. The close alignment between predicted and experimental values confirms the suitability of the optimized conditions for industrial-scale drying of leafy vegetables.

Keywords: artificial neural network (ANN); fluidized bed drying; optimization; predictive modeling; response surface methodology (RSM); water spinach.

FIGURE 1

Diagram of the overall research approach.

FIGURE 2

Schematic diagram of the lab‐scale fluidized bed dryer (FBD); original diagram by the authors, information based on (Akhtaruzzaman et al. 2021).

FIGURE 3

Configuration of the neural network used in the current research.

FIGURE 4

Predicted versus actual values plots produced by Design expert; (a) Moisture content, (b) amount of Vit. C, and (c) amount of ß‐carotene for unblanched samples.

FIGURE 5

Predicted versus actual values plots produced by Design expert; (a) Moisture content, (b) amount of Vit. C, and (c) amount of ß‐carotene for blanched samples.

FIGURE 6

Response surface plot produced by Design expert; (a) Moisture content (b) amount of Vit. C, and (c) amount of ß‐carotene for unblanched samples.

FIGURE 7

Response surface plot produced by Design expert; (a) Moisture content (b) amount of Vit. C, and (c) amount of ß‐carotene for blanched samples.

FIGURE 8

Input factors, responses, and combined optimized desirability values.