Summary
Well-developed methodology exists for handling uncertainty for a single reservoir. However, development of multiple fields presents a significant challenge when uncertainty in a large number of variables, such as gas in place and liquid yield, occur in each reservoir. Some of the challenges stem from our need to forecast the system behavior involving a coupled reservoir/wellbore/surface (CRWS) network for the entire spectrum of variables so that facilities can be designed for the range of fluid composition and throughput. Of course, assessing well count and sequencing well drills are some of the important objectives.
This paper describes probabilistic production forecasting with a compositional CRWS network model for nine reservoirs involved in delivering gas supply to a liquefied natural gas (LNG) plant in Nigeria. Our main objective was to use an economic indicator to select the optimal design of two main pipelines, each transporting 200 and 300 MMscf/D from the two production platforms, located 15 and 5 km, respectively, from the processing platform.
Rate and cumulative profiles showed that sustained deliverability of gas could be realized for approximately 11 years before the decline occurred in high-permeability reservoirs. In other words, uncertainty in gas in place did not surface during the plateau period, only during the decline period lasting another 5 years after the first 11. In contrast, the liquid rates exhibited a large uncertainty band throughout, a direct manifestation of the condensate yield issue. The uncertainty band among each of the 12 components aided facilities design. Differences in net present value (NPV) and discounted profitability index (DPI) were used as discriminators for discerning optimal pipe size from the standpoint of project economics.
Introduction
In recent years, probabilistic forecasting has gained popularity and has become the preferred approach when assessing the value of a project, given the uncertainty of many input variables. Uncertainties arise because both static and dynamic variables are ascertained from rather small volumetric samples of a reservoir and subsequent key variables are estimated from interpretations. Systematic approaches have emerged to account for uncertainty of both static and dynamic variables involving statistical approaches. These methods have been detailed elsewhere (Damsleth et al. 1992; Friedmann et al. 2003; Kabir et al. 2004) for a single reservoir. However, very few studies exist in which production is sought from multiple reservoirs with uncertainty associated with each one of them. Cullick et al. (2004) and Narayanan et al. (2003) have presented case studies of production forecasting under uncertainty for multiple fields. In their studies, flow-simulation tools were integrated with economic evaluation tools and the Monte Carlo (MC) algorithm. Optimization was sought for an objective function (NPV, for instance) honoring various constraints.
The objective of this study was to investigate the impact of uncertainty in input variables on the production forecast for systems consisting of multiple gas/condensate reservoirs, honoring wellbore constraints. We studied multiple reservoirs with multiple wells producing independently. The complexity arises because of the interactions through the common flowline system. The wellbore model was coupled with the reservoir model to honor wellbore constraints. The surface network interfaced with disparate wells through producing rules or constraints. Some of the producing rules included production upper limits to avoid erosional velocity and meeting CO2 production constraints because blending of various streams occurs. In this study, the types of uncertainty considered are in-place volume, condensate yield, capital costs, and operating costs.
We segmented this study into two phases. In Phase 1, we used an analytic simulator to generate the pressure and production forecasts for dry-gas reservoirs, coupled with a simple economic model but without the surface network. The intrinsic idea was to establish well count with a simplistic approach on a spreadsheet. In Phase 2, a CRWS model allowed us to discern the pipe diameter of two main trunk lines transporting gas/condensate fluids by use of incremental economics.
