The Impact of Errors in Measurements of Coalbed Methane Reservoir Properties on Well Production Forecasts

Abstract

SPE Members Abstract This paper presents the results from two simulation parametric studies of coalbed methane well forecasting techniques. Phase I qualitatively identifies the relative impact of key reservoir properties, and rank orders them. Phase II uses a Monte Carlo simulation approach to quantify the degree of confidence associated with coalbed methane well forecasts for two sets of distributions of key reservoir properties. Results indicate that the range of error associated with the best currently-available reservoir property measurement techniques significantly impedes the engineer's ability to accurately forecast production with confidence, and points to the need for improved measurement accuracy for key reservoir properties. Background Similar to conventional oil and gas reservoirs, it is important for reservoir engineers to generate production forecasts for coalbed methane wells to evaluate the value of the in-place reserves, design field facilities, and optimize development practices. The process of forecasting coalbed methane well production, however, has proven to be relatively difficult compared the process of forecasting production for conventional gas wells due to the relatively complex nature of the methane storage and flow mechanisms in coals. The two techniques that have been used for predicting coalbed methane well production are (1) decline curves, and (2) reservoir simulation. Several recent papers have illustrated the use of decline curve techniques as applied to producing coalbed methane wells. Although there is little theoretical or historical basis for applying decline curves to coalbed methane wells, these authors have shown that, once producing coalbed methane wells exhibit a sustained declining production trend, decline curves can adequately match the production decline history. Presumably, decline curves can be used to predict future well production for these wells. The major drawback of using decline curves for forecasting coalbed methane well production is that a relatively long producing history (more than 2 years) is required to establish the appropriate decline rate. It has also been shown that the decline rate of wells within a field varies significantly. Therefore, it is extremely difficult to use decline curves to accurately forecast production for offset acreage, infill well locations, or producing wells that have not yet exhibited a declining production trend. Reservoir simulators are the best available tool for predicting long-term production of coalbed methane wells. Coalbed methane reservoir simulators are the only tool available to reservoir engineers that can correctly account for gas desorption, methane diffusion characteristics, relative permeability effects, well to well interference effects, and well operating procedures simultaneously. Simulators are also able to account for the dynamics of permeability and porosity variations over time that may occur as a result of the large, inherent compressibility of coal. The main disadvantage of using reservoir simulators to forecast coalbed methane well production is that a relatively large amount of data must be assembled to run the simulator. Table 1 is a listing of the properties required for simulating coalbed methane well production, and the common sources of these data. Because the data used to construct a simulation dataset are obtained from a variety of sources, and because many of the measurements are from core tests (which may need to be modified for use as input into a simulator due to scaling effects), there is obvious concern over the accuracy of a production forecast generated using these data. P. 517^