Estimating Arsenic Concentration in Compost Production Using ANN Model

Abstract

Arsenic concentration is one of the most critical concerns in compost production. Measuring arsenic concentration is time consuming and expensive process for compost companies. Also, compost factories need to test the arsenic concentration in their product continuously. Developing a practical model can help compost factory managers to make a faster decision with minimum cost. This study has been designed to predict arsenic concentration in compost, based on input materials and outside temperature, using an Artificial Neural Network model (ANN) in Christchurch, New Zealand. After investigating several ANN model structures a modular ANN model was selected with minimum error margin. The final ANN model developed was based on monthly input of kerbside collections, food wastes, river wastes, and average monthly air temperature for the last eight years. Comparing observed and predicted data indicated that the ANN model could predict  arsenic concentration in different conditions, which is accounted for 94% of the variance for training and 97% of the variance for validation data. However, it should be mentioned each model would be unique best on inputs and weather condition in each city or factory.