Reality Checks for Well Time and Cost Forecasts

Abstract

Management Much has been written to explain the origins of the credit crisis that has engulfed the international financial system. At the heart of each explanation is the failure to quantify commercial risks. The companies at the centre of the crisis were often long-established, considered profitable, and were endowed with great resources to make mathematical models of their activities. Despite this, it is safe to conclude that many of their models were wrong. Risk is a feature of financial transactions and drilling wells that no amount of engineering will completely remove. From the crisis in banking it is obvious that the skills required to build a credible model of activities inclusive of risk are in short supply. It would be convenient to assume that because the energy industry employs lots of numerate people it is good at quantifying uncertainty; unfortunately, many of the time and cost models used for well planning can be easily disproven. The reasons for this go beyond the tools available to the human bias toward optimism. There is a tendency to treat each well project as unique and disregard the fact that past performances will not only inform us about averages but also about uncertainty. A Suitable Benchmark The budget for new well projects will often first be defined with little detailed engineering information. There are perfectly valid reasons for this; out of a basket of many prospects an operator is unlikely to develop every one. Early budgetary estimates play a part in the selection of the best prospects to drill. Irrespective of the method used to define a budget figure, and irrespective of the maturity of the estimate, some guidance on confidence should accompany the result. In common with any other forecasting discipline, well cost estimation never is and never will be an exact science. The very best that can be done is to describe the probability of a particular scenario and to use the results from similar previous wells to back this up. If a model of well project duration cannot be substantiated with empirical evidence it must be treated with extreme caution. The inputs to the model matter much less in this respect than the output; it is surprisingly easy to make a poor probabilistic model from good information. In many mature drilling provinces such as the UK, public data can be used to provide the justification necessary for models of project duration which in turn can be used to model cost.