Meeting the gas demand-supply imbalance is crucial in addressing energy requirements for the future. An increasing percentage of the supply of gas will be sourced from unconventional gas reservoirs, for instance, tight sands/shales and coal bed methane reservoirs. However, while unconventional resources contribute an ever increasing fraction of the total gas supply, an integrated analysis and characterization of these reservoirs is often difficult owing to uncertainties associated with various parameters in reservoir and geologic description. Moreover, the high cost of data acquisition limits the amount of information available to build reservoir models.
The main goal of this paper is to provide a methodology for integrated assessment of the petrophysical, geologic and reservoir parameter uncertainties to provide decision makers a rational basis for managing uncertainties leading to improved decisions and resource management. The workflow involves identifying key parameters influencing uncertainties in recovery predictions through an Experimental Design/Response Surface methodology. The need to do a global sensitivity analysis motivates the workflow presented in our study.
The relevant parameters are used to build an emulator or an analog of the reservoir which is then used to rapidly provide performance predictions, taking into account the uncertainties associated with the parameters. The entire framework is applied to a case study to illustrate the cost and time benefits of such an approach. The significant contribution of such an analysis is identifying the effects on reservoir performance due to uncertainties associated with characterizing the reservoir leading to an informed and rational decision making process with significant cost savings and a reduction in computational effort.
Unconventional gas reservoirs are playing an ever increasing role towards satisfying current and future energy demands. The renewed interest in developing these resources in the past few years has led to a better understanding of the dynamics of fluid flow and has enabled us to incorporate detailed and resource-specific flow physics into our reservoir simulation models leading to better performance predictions and analysis. The selection of a development strategy depends on the ability to characterize the reservoir adequately and rapidly; however for newer fields and recently developed resources, inputs to simulation studies are often uncertain. During the nascent stages of a project, various sensitivity studies are performed to quantify the impact of uncertainty in input parameters on various performance indicators, like net present value or cumulative gas recovery. The goal of these studies is to identify the parameters to which the performance indicators are most sensitive.
Classical sensitivity studies involve perturbing the input parameters, one at a time, from a reference value and computing the corresponding change to the model output. Due to the highly non-linear relationship between model output and input parameters, these sensitivity studies are often valid only in the vicinity of the reference value. Moreover, such approaches may be prohibitively time-consuming and computationally expensive due to the large number of simulation runs required. This has led to an increased interest in experimental design and response surface methodology (ED/RSM) which is an efficient technique to perform global sensitivity analyses for investigating input-output relationships. In previous studies, ED/RSM techniques have been used for performance prediction, uncertainty modeling, sensitivity studies and history matching (Cullick et al. 2006;
Damsleth et al. 1992; Dejean and Blanc 1999; Manceau et al. 2001; Friedmann 2003; Friedmann and Li 2005; White et al. 2001; Landa and Guyaguler 2003; Friedmann and Li 2006).