The Development of Models to Predict the Characterization and Treatment Feasibility for Fruit and Vegetable Wash-water
Large quantities of fresh water are used for washing and processing fruits and vegetables. Post-harvest processing includes washing of soils from root vegetables to washing tree fruit for sanitization, in addition to the cutting and peeling of fresh-cut fruits and vegetables. Variabilities from one facility to another also exist due to differences in washing and processing units for the different fruits and vegetables. As such, the generated wastewater or wash-waters vary significantly in terms of water quality, making it challenging to treat. Compounding the problem is the lack of tools for determining wash-water/wastewater quality parameters and treatment feasibility for disposal or reuse applications. The research herein hypothesized and proved that wash-water treatment decision tools (models derived using power-rank, multiple linear regression (MLR) and artificial neural network techniques (ANN)) can be developed for wash-waters for the fruit and vegetable industry. This overall goal is achieved by first characterizing wash-waters followed by bench-scale treatment to target solids, effective reduction of solids is key. Level of water quality parameters varied significantly, for example, SS ranged from 43 – 140 mg/L for tree fruit, 182 – 12,730 mg/L for root crops, 30 – 215 mg/L for leafy greens, and 290 – 650 mg/L for mixed wash-waters. On-site membrane bioreactor (MBR) achieved a greater than 90 % reduction in all water quality parameters, while biological and settling treatments varied from 40 – 90 % reduction, depending on water quality parameter. Bench‐scale treatments selected for testing included settling, coagulation and flocculation with settling, centrifuge, dissolved air flotation, electrocoagulation, screening, and hydrocyclone. The developed decision matrices summarize the removal efficiency of the tested treatments. The Power-Rank models were created to easily determine raw wash-water quality parameters. The final objective consisted of taking all treatment data and further processing it to develop treatment effluent water quality prediction tools by using multiple linear regression (MLR) and artificial neural network (ANN) analysis. A cost-benefit analysis was also incorporated to determine treatment and operational costs in terms of dollar per cubic meter of water treated ($/m3). These findings and tools are useful to growers/producers in determining treatment options that did not previously exist.