Christian Bos, Senior E&P Reserves and Decision and Risk Expert with TNO-NITG, gave a presentation in December on “Why the E&P Industry Is Waiting for a New Generation of Petroleum Business Engineers” to the Netherlands young professionals group. He said there are conflicts between models that describe the Earth as accurately as possible and models tailored to E&P business decisions. The first category tends to be highly detailed, whereas the latter category generally is more probabilistic, based on integrated data, and contains less detail. The main focus of the presentation addressed determining the best approach to take in building models that support E&P business decisions, and how the bias in decisions that has characterized the industry in the past can be prevented.
Looking at stock market returns, it is clear that the E&P industry has underperformed other industries in North America, he said. Field production forecasts tend to be optimistic and operational expenses are generally underestimated, indicating a bias in how the industry is run and indicating that the industry does not learn from its past mistakes. The question now is, What is the origin of this bias? Is it our way of work; is our common practice the best practice; is it our mindset; is it the nature of business? And, finally, what is the solution?
The E&P business historically has been weak at forecasting. However, at the same time, few initiatives have been deployed to improve industrywide learning as has been done in other sectors (e.g., weather forecasting). Because of the nature of the work, a lot of different specialists are involved in creating models, each with their own language, focal points, and workflows. These models tend to be highly detailed to capture all aspects of the subsurface. Although a lot of statistics are involved, the models rarely are truly probabilistic. Any model should optimally support E&P business decisions. This means that the initial model should accurately integrate data and should be tailored for risk and uncertainty management. From a technical point of view, the model should contain enough detail to support data integration and statistics. During the business process, the model is gradually improved on the basis of the decision requirements. To improve the record of the industry, a paradigm change is necessary: Information needs to be captured to support business decisions, not for fancy models, Bos said.