Physics-Based™ Predictive Analytics
for Oil & Gas Production Forecasting
The Science of 21st Century Forecasting with Physio-Stats™
Oil and Gas is the biggest industry in the world, and one of the most technologically advanced. Yet it is an open secret that most production forecasts are still being made by petroleum engineers, essentially by hand, by fitting old-fashioned curves to data. This practice of hand-crafting introduces unnecessary risk and subjectivity into forecasts, which can lead to heavily biased results. By contrast, BetaZi forecasting technology is fully automatic and fast, delivering analysis results of oil and gas wells and fields (as well as water and geothermal reservoirs) that are verifiable, repeatable, and far more detailed and comprehensive than anything else currently available.
Physio-Stats™:
Physio-statistical analysis ("physio-stats™") is the combining of physical laws with the statistical behavior of physical systems as learned from big data in order to do predictive analytics and estimate uncertainty with repeatable and testable results
Predictive Analytics:
The science of using past facts to understand the present and predict the future
Calibrated Uncertainty:
When the bounds or error bars on individual forecasts have been validated by computing how well on average they hold up against actual data
There are essentially two ways to think about a problem scientifically. One is in terms of physical laws, which are usually equations. The other is in terms of statistics, which is essentially behavior (looking at large numbers of systems and seeing how those systems tend to behave). Physio-statistical analysis ("Physio-Stats™") starts with physical laws and learns from the study of large datasets how actual systems tend to deviate from what would be predicted by physics alone. Using that hybrid system for predictive analytics, BetaZi's analytical model ("the BZ Machine") can look at a data record and not be confused by all the deviations from physics; when forecasting scenarios are then pushed forward into the future, there's a guarantee that the resulting prediction is something that still obeys the laws of physics and actually could happen. The power of good physical statistics (BetaZi Physio-Stats™) is in creating a system of checks and balances. On the one hand you have knowledge of the behavior of the system (which is learned), but also embedded physical laws so that what the model outputs are realistic scenarios.
The BZ Machine works both physics and experience into its solutions, then, using a combination of fluid flow equations and statistical models calibrated using thousands of actual wells. It does not get confused by messy, aggregated production data complicated by multiple shut-ins, stimulations and missing records. It requires no human intervention, which means that no human bias is introduced. It is also fast: tens of thousands of wells can be run overnight.
The service provided by BetaZi is unique at this time in bringing patented state-of-the-art predictive analytics to oil and gas production forecasting. Predictive Analytics is the science of using past facts to understand the present and predict the future. Using monthly oil, gas and water production volumes, the BZ Machine automatically forecasts future hydrocarbon production by finding a million possible scenarios (samples) which explain the data. Those samples are projected forward so that the model is able to return accurate statistics along with its forecasts. BetaZi forecasts can be used to compute future rates and volumes, production targets, Estimated Ultimate Recovery (EUR) and Present Value (PV), as well as many other derived quantities such as equivalent interest rates and time to payout. BetaZi predictions can also be used to improve reserve estimates.
Every BetaZi forecast includes calibrated uncertainty: solid bounds which delineate both the upside potential and the risk of an asset. The bounds have been calibrated by testing on hundreds of thousands of wells using metrics designed by a team of experts in computer science, statistical inference and petroleum engineering. In calibration studies, actual well production exceeds our low bound (called a p90) 90% of the time. Actual production tends to exceed the upper bound (p10) slightly more than 10% of the time, because engineers are often able to stimulate, treat and otherwise coax extra production out of wells – which is what engineers should be doing instead of manually setting forecast curves.
Calibrated uncertainty is a powerful measure of the performance of wells. It is also what makes BetaZi a disruptive technology. With it comes the ability to accurately assess and engineer risk. Right now, oil and gas wells are considered risky assets, but they are no more unsure than other assets such as mortgages and technology stocks. In fact, they should be less so because they are underlain by physical commodities (oil and gas power the world) and production is fundamentally governed by physics, not the ability of debtors to pay or the whims of consumers. But when production forecasts are unreliable and/or biased, the investor who relies on these figures does not know what is actually going to come out of the ground. Not only is he or she then subject to loss through price volatility, but expectations of future return and payout could be inordinately high or low. With BetaZi analysis and classification technology, risk can be computed, calibrated, understood and exploited.
To download a 12-page paper on BetaZi Science, request here.