Crystal Ball 2005 Oil & Gas User Group

The Excel plug-in offers Monte Carlo simulation to engineers and economists. BP uses Crystal Ball for ‘drillability’ assessments and all major decisions taken by Ecopetrol are supported by the tool.

Decisioneering’s Crystal Ball (CB) is a Microsoft Excel plug-in that offers Monte-Carlo modeling of spreadsheet data. Monte Carlo (MC) techniques offer a simple way of propagating errors (or ‘risk’) through an economic or scientific model. One enthusiastic CB user is William Standifird from geopressure specialists Knowledge Systems (KS). KS develops compaction, reservoir and depositional history models—all of which require calibration and statistical modeling. Uncertainty is omnipresent which is why KS uses MC modeling on ‘unstable’ parameters. These are used to evaluate prospect viability including seal efficiency and drilling feasibility – some planned well paths are technically impossible to drill. BP now does regular ‘drillability’ assessments – important in a $100 million well! KS offers two modeling approaches: hard coded ‘perfect’ solutions, using bespoke algorithms where CB is too slow and ‘imperfect’ and more flexible solutions, leveraging CB as a ‘fit for purpose’ uncertainty evaluation technology.


Oscar Bravo Mendoza stated that in Ecopetrol, which has 150 CB licenses, ‘no major decisions are taken without CB.’ Ecopetrol measures risk to be in a better negotiating position with respect to banks and joint venture partners. Despite perceptions, Columbia is not a high risk country as witnessed by its score in the Emerging Markets Bond Index Global. Colombia is not in the same league as Chavez’ Venezuela! Plots of project complexity against uncertainty or risk likelihood against risk impact are used and shown to insurance companies to negotiate rate reductions. MC modeling can be combined with decision trees to show a project’s upside potential.


Pat Leach* (Decision Strategies) presented a case history of FPSO design which determined how much NPV would be lost if the facility’s water handling capacity was too low. Leach used Murtha’s stochastic production profile generator to model water breakthrough. While some cases showed ‘significant’ production loss, others deferred production. In the end a compromise was reached with a compact water processing installation that left room for extra capacity to be added if needed.


Steve Hoye’s presentation enumerated the many pitfalls of the spreadsheet as a modeling tool. As a Decisioneering trainer Hoye has seen it all and has a few horror stories to relate on spreadsheet worst practices. In general it is the spreadsheet rather than the stochastics that gets the average user into trouble. Spreadsheets get unmanageable. Hoye recommends using color and space as aids to spreadsheet readability. Complex stuff needs breaking down into separate worksheets or workbooks. Range names should be used rather than cell references. Astonishingly, some 40% of CB trainees don’t know about this best practice. Other tips include protecting your data cells, using templates and sharing assumptions across the business to avoid multiple hypotheses. Finally, don’t put reports in models, use Latin Hypercube sampling, avoid the ‘killer formula’ and avoid macros if possible as they reduce transparency.

Vose Consulting

For those of you who may be daunted by doing your own CB analysis, Decisioneering, in association with Vose Consulting, is ready to help. Vose now offers a range of services including model building and audits, risk analysis and on-site Crystal Ball training.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.