Modelling Bio-Methane Production In Ruminant Livestock Farming

- Methane from ruminants is a principal contributor to greenhouse gases. Consequently, sustainable mitigation strategies for enteric emission are in high demand. This study is aimed at modeling bio-methane production process in the bio-digesters using a combination of correspondent feed additives for methane emissions reduction. In this study, each of the experimental animals were exposed to 5 different feeding trials and theirdung were collected after the feeding trials. A set up of 5 units of 50litres biogas digesters to cater for the digestion of 4 animal waste substrate and 1 control sample was used as experimental facility for biogas generation and collection.Biogas yield was measured at the end of 14 days. Bio-gas samples collected from each bio reactor was analyzed using the 263-50 Gas chromatograph and the result was displayed by the aid of D2500 Gas Chromato-Integrator. A first order linear model was developed using XL STAT Software, 2021 premium version for the prediction of methane emission from different animal feed additives. Comparison was carried out between predicted and observed bio-methane emission values for the different feed additives. Performance of the model was evaluated using model evaluation metrics in order to determine the consistency of the predicted values with observed values. Specific analyses were performed to validate the model outputs against measured data. Predicted values were paired against measured values using mean square error (MSE), root mean square error (RMSE) and mean absolute percentage error (MAPE). In addition, a measure of goodness of fit known as coefficient of determination (R 2 ) was used to determine the closeness of the predicted values to the measured values. The results indicate that all the developed first order linear models adequately fit the measured data set with goodness of fits greater than 95% (R 2 > 0.95). Test Unit 1 model explained 99.11% of the measured values; Test Unit 2 model explained 98.96% of the measured values; Test Unit 3 model explained 98.19% of the measured values; Test Unit 4 model explained 97.73% of the measured values; while Test Unit 5 model explained 97.93% of the measured values. Modeling error metrics shows that Test Unit 1 has that lowest mean prediction errors; this is followed by Test Units 2 and 5. This suggests that the first order linear model accurately predicts that naturally occurring of methane emission in the control group; while the feed additives in the other Test Units are variables that influence the potential of the first order linear models to accurately predict methane emission in the substrate degradation or utilization. This study shows that the derived first order linear model significantly predicted methane emission in substrate degradation and therefore can be used to forecast methane production from animal feeds. Result further revealed that all the developed first order linear models significantly predicted the methane emission with probability p-values < 0.05, 95% confidence interval/ level-(CI) and coefficient of determination (R 2 )> 0.95. Therefore, mixed additives could be used as effective anti-methanogenic compounds to efficiently reduce enteric methane production.