Weyburn CO2 Miscible Flood Conceptual Design and Risk Assessment

For the three years leading up to the end of 1995, a multi-disciplinary project team evaluated the business opportunity of injecting carbon dioxide into a portion of the Weyburn Unit to improve recovery. As with any project of this magnitude, a comprehensive analysis of the opportunities and risks in the project had to be clearly defined and evaluated. A risk assessment using a Monte Carlo simulation approach was undertaken to combine the technical with the non-technical issues associated with the project in order to define the full cycle opportunities and risks. The risk assessment process was used to optimize the project configuration, to focus the team on key issues, and to find ways to mitigate the risks inherent in the project. Key uncertainties impacting the range in expected project returns appeared in the areas of commodity prices, reservoir performance, costs and fiscal terms. The risk assessment process and conceptual development plan, incorporating studies and analysis completed up to the end of 1995, is the subject of this paper. FIGURE 1: Weyburn unit outline. An analogy to the Weyburn field can be drawn from the Midale Unit, situated just east of the Weyburn field. Shell Canada Ltd., the operator of the Unit, implemented a CO2 injection pilot in 1984 which was completed in 1988. Due to encouraging results from the pilot(2), Shell implemented a “CO2 demonstration project,” an eight pattern flood in 1991 which was designed to improve oil recovery from the area and provide technical and economic data to further assess full scale development potential. This flood is still in progress. Input data forming the basis of this assessment incorporated some of the Midale information as well as opinions from internal experts, industry consultants and other analogous CO2 floods in the Permian Basin(3). Traditional Evaluation Process Traditional engineering, design, and estimating methodology sets out a logical step-by-step process which leads through the project life cycle. Such a process uses single point estimates for project outcomes, capital cost, operating cost, plant performance and revenue streams. Inherent in the analysis is a set of assumptions developed by individual members of the project team. These assumptions are often buried in the analysis and are not communicated internally between members of the project team and the decision makers. This process tends to focus on one project strategy and often fails to consider the impacts of alternative environments on project outcomes. The single point result of the analysis is seldom right and surprises are encountered during project execution. More importantly, this approach fails to identify project improvements which could provide significant increases to the project’s value, and the resulting analysis fails to clarify project control priorities to mitigate the negative impact of unplanned events. Risk and decision analysis techniques can be used to improve the quality of key project planning decisions in the formative stages of new project development. Early utilization of the process provides the team with a tool for ongoing decision making as the project moves forward. Risk Analysis Process Risk analysis is the centerpiece of a risk management process (Figure 2). Risk analysis is used to quantify uncertainties in a project by placing it in an uncertain environment and calculating the likelihood of various outcomes within that environment. The process(4) should support the management of risk in more significant ways: 1. Provides a significant benefit in developing insight and understanding about the impact of risk on the project outcomes. This generally leads to a change in direction in the project either in the form of design changes or some other project plan revisions. 2. Provides for the development and review of contingency and risk mitigation plans. Given certain changes in the project environment, the ability of the analysis to test “what-if” scenarios prepares the project team for the appropriate response before the fact. 3. Forms the basis for a dynamic risk monitoring system over the life of the project. Analysis of the risks as the project proceeds provides an ideal mechanism to forecast the future end result of the project. A rigorous process is employed to conduct risk and decision analysis. Figure 3 illustrates the five step process, from framing the problem through to implementation. 2 Journal of Canadian Petroleum Technology FIGURE 2: Risk analysis is the centerpiece of a risk management process. FIGURE 3: Five steps in a risk and decision analysis process. Step 1—Frame the Problem The first step in framing the problem is to define the preferred project strategy and its alternatives. The strategy table charts the alternative paths through the decision options. The project team then identifies all sources of risk to the project in a brainstorming session. These risks are graphically charted to demonstrate the impacts on the project outcome using an influence diagram. The Weyburn project’s influence diagram is illustrated in its simplest form in Figure 4. In this influence diagram, hexagons are key project decisions, ovals represent project risks, arrows represent the flow of the risks and interactions, while the rectangles represent key calculations for the major project components. The octagons are the measurement criteria for project outcomes(5). Step 2—Develop the Analysis Base Modeling the project environment is an important second step. As discussed by Goodyear et al.,(6) the risk model is designed to quantify the impact of change in value or probability of the uncertain variables upon the expected value of each alternative, or upon the components which make up that value—i.e., capital cost, operating cost, productivity, schedule, etc. Step 3—Evaluate the Risks The risk evaluation step is the key to success of the risk analysis process. Because the quality of the analysis is dependent on the experts’ judgment, it is important that these people are selected based on their credibility with the decision makers. A series of assessment meetings are held to assign a range of values to each risk variable. Interview techniques are designed to avoid biases in the experts’ judgment. Often, expert opinion from outside the project brings an independent alternative view on the important variables. The experts are asked to provide both probabilities of occurrence and the quantification on the impact of each variable on the project results. The experts also provide a description of the environment that would lead to the extremes. Sharing these “stories” provides an important communication function among the experts. Documentation of the assessment and the insights behind the extreme values is important when considering ways and means to mitigate the risks. Step 4—Interpret the results The interpretation step involves running the analysis model to determine the key risk contributors and to compare decision and strategy alternatives on a “risked value” basis. At this stage the initial results are fed back to the experts. This may require focus upon and re-examination of the most critical variables. This step is repeated until the project team and experts are satisfied that the model reflects a realistic picture of the project. Step 5—Compare Alternatives The analysis model is used to produce a risk adjusted picture of the project. A set of graphical tools are then used to describe the uncertainty in the project. Cumulative probability curves can be used to compare the alternatives. Probabilities of achieving required results can be read directly from these curves. Figure 5 shows an example of a comparison of two hypothetical options. The significant upside opportunity for Option B would not be apparent in a single point analysis. In addition, the degree of uncertainty can be measured by examining the slope of the distribution. A vertical line indicates Special Edition 1999, Volume 38, No. 13 3 FIGURE 4: Weyburn risk analysis model structure. FIGURE 5: Probability distribution showing risk and opportunities. no uncertainty, while a horizontal line indicates extreme uncertainty. Tornado diagrams (Figure 6) demonstrate the impact of key risk variables contributing to uncertainty in the project. Each risk variable is, in turn, set at the assessed P10 and P90 range while all other variables are set at their expected values. The model is then run to determine the range in outcomes for each variable range. The results are then ranked in descending order with variables contributing the most uncertainty appearing at the top of the diagram. Weyburn Risk Analysis Model The Weyburn risk analysis model was constructed using an integrated suite of technologies and software packages. A Microsoft Excel spreadsheet application and @Risk, an add-on package to Excel, were used to model the project. MS Excel(7) is an application used by most members of the project team. With this tool, the algorithms could be investigated by the team, thus reducing the “black box” syndrome. This adds to the credibility of the model and acceptance by the project team. Due to the complicated nature of “rolling-out” the individual patterns in the EOR flood, a set of functions were written in “C” programming language and saved in a dynamic link library (DLL). These functions are called by the spreadsheet as required. @Risk(8) is a separate software package which facilitates a Monte Carlo analysis. In a simulation, a number of outcomes are calculated and recorded. Each of these outcomes is a product of an iteration, where each variable is randomly assigned a value from its discrete probability distribution. The outcomes can then be analysed to determine the probability of a specific result. The model is designed to function in three modes. The base case mode has all the risk variables set at their base values. The variables have no impact on the calculation and the single point estimate for the project is calculated. In the sensitivity mode, each variable is set at its expected value. The calculated project outcome approximates the result of the Monte Carlo analysis. The sensitivity of each variable can be tested by changing its value and observing the change in the project outcome. In the probability mode, the analysis uses a Monte Carlo simulation to calculate the range of possible project outcomes in order to measure the project’s performance in a number of different environments. Base Case Assumptions The term “base case” is used to describe results calculated using all of the base estimates for schedule, production, capital and operating costs. The base case contains a number of assumptions which reflect the current project design and can be tested by the risk analysis model. The following assumptions were embedded in the Weyburn base case. • CO2 pipeline capacity – 4,000 Tne/d • Minimum CO2 contract volume – 3,000 Tne/d • Recompression facilities – central or satellite • Recompression facilities – process 4,000 Tne/d • Initial field injection capacity – 6,000 Tne/d • Venting of excess CO2 – minimized