digilib@itb.ac.id +62 812 2508 8800

Coal Bed Methane (CBM) is one of the contributors to the natural gas supply, but the CBM reservoir can be barely produced at an economic rate by primary recovery. One of the methods to enhance CBM production is the injection CO2 into a coal seam (CO2-ECBM), an economic analysis plays a crucial role in identifying the feasibility of CO2-ECBM application. A model that can predict the CO2-ECBM performance by the economic indicator will greatly useful in the project's initial feasibility study. This study focused on constructing the proxy model to predict the CO2-ECBM performance for 10 years using the Profitability Index (PI) as the main objective. Several parameters being used to construct the proxy model which consists of technical and economical aspects. Of 23 parameters, 1104 experiment models have been generated by Latin Hypercube Sampling(LHS) method. Quality control was applied to improve model quality. The experiment models are used for sensitivity analysis using CMG-CMOST to construct the predictive proxy model of CO2-ECBM performance based on PI. The model was constructed by regress the experiment models in CMG-CMOST. The polynomial quadratic regression develops the best proxy model which gives the R-squared 0.91. the proxy model composed 11 of 23 parameters that were chosen based on the magnitude of the effect from each parameter in the tornado chart from Sobol analysis. The predictive proxy model also gives a quite good statistical error, which 0.149 in average maximum absolute errors. After the proxy model has been constructed, a model validation will be carried out. The PI result from the proxy model is compared to the value from the CMG-GEM reservoir simulator using another CBM data. The result is quite good, the PI from the proxy model gives value 0.614 and 0.571 from CMG-GEM which is give 7.5% percentage error. Based on the model quality and model validation, it is expected that the predictive proxy model is quite accurate to predict the PI from a CO2-ECBM project.