Glossary of Terms
Analogous: In reference to oil/gas wells, means they share characteristics, namely a common reservoir and similar design. (Analogous well data is required to produce type curves for PUDs.)
Bayesian network: A probabilistic graphical model (a type of statistical model) that represents a set of random variables and their conditional dependencies via a directed acyclic graph (DAG)
BZ Machine™: The BetaZi engine, program, built algorithm.
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
Calibration: A test of the model to assure its accuracy and validate results produced. BetaZi uses Q-Q plotting.
Engineered risk: The risk profile of a financial instrument composed of assets that have been carefully chosen or hedged to match the objectives of a particular investor
Field Series Expansion Machine: A type of Open Universe Bayesian Network model which solves physical problems using a distribution on the number of terms in an expansion representing the solution to a physical partial differential equation combined with statistical models for the deviation from physics
Forecast: Estimate of future well production volumes over a given period of time.
Physio-Stats™: Physio-statistical analysis 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
pScore™: the average percentile value of the production of a well measured against a type curve, which provides an accurate representation of how a particular well stacks up against members of a population of similar wells.
P-value: The probability under a specified statistical model that a statistical summary of the data (for example, the sample mean difference between two compared groups) would be equal to or more extreme than its observed value. BetaZi P-values (also known as percentile values) designate the following:
p10 -- Upper bound. Actual production is expected to exceed this bound 10% of the time
p50 -- Median. Production should be above or below this bound an equal fraction of the time
p90 -- Lower bound. Actual production is expected to exceed this bound 90% of the time
PDNP: Proved Developed Non-Producing Production -- wells are shut in but scheduled to be reopened
PDP: Proved, Probable or Possible Developed Production -- wells are currently producing
PUD: Proved UnDeveloped -- undrilled wells for which other, similar wells exist which can be used to forecast production
Quantile-Quantile (Q-Q) plot: A calibration of the accuracy of the model results vs. the data that was analyzed; generated through blind testing of heldback well data. Q-Q plots are the standard in statistical analysis, though only BetaZi can and does generate them in the O&G forecasting space.
Roll-Up: Statistical aggregation of future production volumes for a series of wells. May be PDP, PDNP and PUD or all production types combined.
Samples: Synthetic well histories generated to explain the input data, typically then computed into percentiles.
Statistical Type Curve: Statistical representation of future probable production from a completed PUD, drawn from analogous wells.