ABSTRACT

 Bio gas is a renewable energy comprising of a mixture of gases (methane and carbon dioxide) produced from organic matter, mostly waste, by a mixture of bacteria anaerobically. It can be burnt to generate heat, electricity or be used as vehicle fuel. The residual (digestate) can be used as fertilizer. This study was carried out to optimize the biogas yield from three independent variables (K=3) namely some organic matter and cow dung as feed stock, pH and retention time designed according to Box-Wilson ( 1951) experimental design matrix. The variables which were varied, each into five equal spaced levels; feedstock (6.00, 8.00, 10.00, 12.00, and 14.0) liters; pH (2, 4, 6, 8, and IO) and retention time (7, 14, 21, 28, and 35) days. The feedstock was fermented inside a locally fabricated bioreactor for 35 days. A total of 2l experimental runs were generated and subjected to response surface data analysis using MINITAB (version 11.21). A Central Composite Rotatable Response Surface Design (CCRRSD) was employed to study the linear, quadratic and cross product effects of the three variables on the biogas yield, temperature and biogas gas components. Three dimensional response surface contour figures were plotted to visualize the effects of process variables on the responses with MATLAB (version R2007b). Results revealed that at variable combinations of 10 liters, 6 pH and 35days respectively for feedstock volume, pH and retention time, there were maximum biogas yield (4.420 liters), methane (2.990 liters), CO2 (0.950 liters), hydrogen sulphide (0.040 liters) and temperature (33C). Minimum values of biogas yield (0.950 liters), methane (0.610 liters), CO> (0.310 liters), hydrogen sulphide (0.0 IO liters) and temperature (28C) were obtained at variable combinations of IO liters, 6 pH. 7 days respectively for feedstock volume, pH and retention time. Temperature obtained was at mesophilic range. The model also revealed that all the independent variables were significantly (p<0.05) linearly, quadratic and interactively related to their responses and the model gave good fit for the data based on the correlation coefficient (0,6203-0.8222) obtained.