An Integrated Production Modelling process is presented that is used to automate CSG (Coal Seam Gas) production forecasts from reservoir to sales. This involved incorporating type curves representing future reservoir performance into a hydraulic model of the surface network. One of the key challenges was to convert time-dependent decline curve forecasts into time and pressure-dependent reservoir-well models suitable for Integrated Production Modelling process. Results are compared to other methods with reduced simulation time, improved accuracy and scenario analysis.
This innovative approach used a combination of techniques to incorporate type curve based reservoir models into two large integrated production models. The first model consisted of 250 wells and a single compressor station and the second model consisted of 500 wells and multiple compressor stations. Different techniques were applied for wells that were pre or post dewatering and incorporated into a surface network modelling tool. This allowed production forecasts to be generated automatically for the single compressor station model and semi-automated for the multiple compressor station model whereby the system was solved to meet demand or to maximise production taking into account constraints.
Type curve data was converted to a ‘tight gas’ reservoir model which showed a good match for wells post dewatering but not for wells that were still inclining. A combination of proprietary scripting in a network simulation program and macro code in Excel were used to handle system constraints optimisation. The time taken to run each scenario reduced significantly as the restriction moved from a human to the amount of computing power. Accuracy and repeatability of results also improved due to models being setup and solved in a consistent manner, thereby removing discrepancies associated with manually driven models. This allowed for more scenarios to be modelled in the time allocated to the production forecasting processes allowing for improved analysis and decision making.
Previously type curved based reservoir models could not be solved by a surface network modelling tool automatically without human intervention. This was not possible due to the size and complexity of the surface network and because of the lack of time and pressure-dependent reservoir-well models suitable for Integrated Production Modelling process. The process outlined in this paper overcame this challenge.