Non-linear Regression Models for Predicting Biogas Yields from Selected Bio-wastes

Abstract

The benefits of biogas as alternative energy to other fossil fuel sources, due to its renewability, environmentally friendly nature, health benefits, etc., cannot be overemphasized. There are numerous models for predicting biogas production rate from bio-materials, including the modified Gompertz equation. These models are primarily dependent on specific biomass parameters. When any of these parameters, like the slurry volume, changes, another round of experiments must be conducted and curve fitted before biogas yield predictions can be made. This could be time-consuming and costly. Using experimentally published data, simple empirical models can be developed for predicting biogas yields over a range of input parameters. This will eliminate the need for always performing experiments before biogas yield predictions can be made. In light of this, scarce literature provides explicit models for predicting biogas yield over a range of parameters based on published data. This study developed non-linear regression models using published data on parameters that affect biogas yields, like the slurry volume, carbon-to-nitrogen ratio, temperature, total solids, volatile solids, hydraulic retention time, and pH. The data covered seven readily available bio-wastes, including cow dung, cow dung with plant waste, cow dung with poultry dung, poultry dung with grass, pig dung, and plant wastes. On validation of the models, the results showed that the models had a relatively low standard error of estimates, Akaike information criterion, Schwarz criterion, and Hannan-Quinn information criterion. Furthermore, the coefficients of determination, R2, were between 94.62 and 98.93%. The percentage average absolute deviation (% AAD) for each model was less than 7 %. The non-linear models were found to adequately predict the biogas yields within the limits of the available data set.