BP’s Crystal Ball templates for well risking

Hugh Williamson explains how BP risks it drilling decisions with a standard Crystal Ball template.

In a Decisioneering webcast last month, BP’s Hugh Williamson showed how BP’s drillers use Crystal Ball to optimize time and cost estimates, using Monte Carlo analysis (MCA) as a risk management tool. BP is currently drilling 1,000 wells per year at a cost of $5bn (30% of Capex). The answer to risk management is a well costing template and workflow, BP’s Single Well Estimator (SWE), a.k.a. ‘Monte Carlo for the masses’ is a Microsoft Excel/Crystal Ball workbook developed by Williamson. Williamson insists on the importance of using a standard template rather than letting users develop their own. Even though engineers like to do their own thing, ‘it is best to rein them in!’


Evaluation starts by finding suitable ‘analog wells’ for comparison. These are wells in similar circumstances, not necessarily close to the target well. BP tries to avoid automated estimation. The key is ‘conversation’ around possibilities. Automation stifles conversation.


Cost/risk estimation starts by prioritizing top sensitivities in a ‘risk register’ table. All significant risks are included in the probabilistic cost estimate. Monte Carlo is not ‘random’ but focuses the mind on what is important in analysis. The P10/90 estimates of well costs represent the range of likely cost. BP always asks are these predictions sound or ‘aspirational’? Deliverables from the SWE include a drilling and completion ‘uncertainty statement’ with a time and cost summary along with MC plots of estimates. Supplying MC to the masses implies continuous development, training and support. BP’s technical management now expects to see, understand and criticize such estimates. Asked if BP applies MC to its business as a whole Williamson stated ‘We are starting to do this but things get complex with options, and there are non technical problems—decision makers may not want MC results or they may not find them useful. Opinions differ, but I think that all-inclusive MC is the best way of understanding the big picture. Specialists like me are trying to push this forward.’

This article originally appeared in Oil IT Journal 2006 Issue # 6.

For more information or to comment on this topic email here.