Kriging Mathematical Modelling of a Biogas Production from Anaerobic Co-digestion of Napier Glass Silage, Cassava Pulp, Chicken Manure and Cow Dung

Abstract

In this paper, a Kriging model was applied to determine the relationship of a biogas production system. The operating system used various raw materials mixed and was not fed in constant proportion depending on the season of the raw materials. This causes the relationship of the system to be non-linear and highly uncertain. The Kriging model was chosen because it is highly efficient for describing nonlinear relationships and is an interpretable model that was trained to approximate the predictions of a black-box model. Therefore, the data that affects a biogas production factor from the automation system was used. Napier grass silage, cassava pulp, chicken manure, cow dung and the biogas production temperature were variables to build the Kriging model. As a result, the most significant variables affecting biogas production model and methane content model was cassava pulp and the least effect was cow dung. The models were calibrated with actual operating data. As a result, the biogas production prediction model and methane content prediction model had an RMSE (root mean square error) of 1.42% and 1.25% respectively and had an r² (correlation coefficient) of 0.9981 and 0.9973 respectively. Therefore, the models can predict. Moreover, it can also represent the behavior of biogas production systems for optimization to control the biogas production system effectively in the future.