Prepare yourself for the AI for Reservoir Performance & Production Forecasting course
Develop the judgement to assess whether an AI-supported production forecast is useful for a reservoir management decision. This AI for Reservoir Performance & Production Forecasting training course connects data quality, model performance and physical plausibility with the purpose and horizon of a forecast. Participants compare machine-learning and hybrid methods with decline-curve benchmarks, examining where each approach provides useful evidence and where changing operating conditions limit confidence.
The course examines how uncertainty, injection changes and development scenarios affect the interpretation of forecast results. Participants learn to challenge implausible predictions, recognise extrapolation limits and communicate the assumptions behind a forecast. They also define how forecasts should be reviewed, monitored and updated within surveillance routines. These capabilities support decisions about adopting a model, assigning accountable ownership and presenting results to technical teams and asset decision-makers.
Highlights of this AI for Reservoir Performance & Production Forecasting training course includes:
- Production and surveillance data readiness, including data leakage
- Decline-curve benchmarks and machine-learning forecast comparison
- Time-aware validation, backtesting and operational error measures
- Physics-informed modelling, uncertainty ranges and scenario analysis
- Forecast assurance, monitoring, ownership and decision support
The learning journey
Course Summary Outline
Day 1
Reservoir Data and Decision Framing
Day 2
Machine Learning for Production Forecasting
Day 3
Physics-Informed Reservoir Analytics
Day 4
Uncertainty, Scenarios and Model Assurance
Day 5
Deployment and Decision Support
Is This Course Right for You?
The AI for Reservoir Performance & Production Forecasting training course is designed for professionals who prepare, evaluate or use production forecasts in reservoir surveillance, field development and asset decisions.
- Reservoir Engineers
- Petroleum Engineers
- Production Engineers
- Production Technologists
- Subsurface Data Scientists
- Reservoir Surveillance Engineers
- Reserves and Resources Evaluators
- Field Development Engineers
- Subsurface Managers
- Upstream Digital Leads
Course Summary
- AI forecasting use cases and decision requirements
- Forecast targets, horizons and operating constraints
- Feature engineering for time-series and well data
- Regression and tree-ensemble forecasting methods
- Material balance and forecast constraints
- Flow regimes and reservoir behaviour
- Capacitance-resistance models for reservoir response
- Ensembles and quantile forecasting methods
- P10/P50/P90 interpretation and probability conventions
- Forecasting pipelines from prototype to operational use
- Integration with surveillance and reservoir management
Want the full outline?
Key Learning Outcomes and Objectives?
The AI for Reservoir Performance & Production Forecasting training course equips participants to judge how data, model choice and engineering assumptions affect a production forecast. Participants learn to compare methods using consistent baselines, validate predictions over time and assess their physical plausibility. They also develop the ability to interpret uncertainty, explain model limitations and define the ownership, review and monitoring arrangements needed to use forecasts in reservoir management decisions.
- Select AI forecasting use cases with clear targets, horizons and constraints.
- Assess production and surveillance data for quality and modelling suitability.
- Compare decline-curve, machine-learning and hybrid approaches using consistent evaluation criteria.
- Validate forecasts over time and check them against reservoir physics.
- Interpret uncertainty ranges and explain scenario assumptions and limitations.
- Define a forecasting workflow with review, monitoring and retraining responsibilities.
The Excentra Learning Approach
Participants use prepared notebooks and a public field dataset to run, adjust and interpret forecasting models with guidance. They compare outputs with decline-curve benchmarks and examine how data choices, operating conditions and validation methods influence the result.
Discussion of conventional and unconventional assets connects the modelling work with reservoir management decisions. Participants develop a forecast supported by engineering checks, uncertainty analysis and an implementation roadmap that defines ownership, monitoring and review requirements.
Six capabilities, and how far this course takes each one
Data Intelligence
Intelligent Monitoring
Workflow & Agent Design
Responsible Implementation
Business Transformation
Ready to take the next step?
Reserve your place, ask us a question, or arrange private training for your team.
Frequently Asked Questions
The course is designed for reservoir, petroleum and production engineers, subsurface analysts, surveillance specialists and reserves evaluators. It is also relevant to technical managers who review forecasting methods or approve their operational use. A working knowledge of reservoir and production engineering helps participants interpret the models and assess their engineering relevance.
You will learn to define a forecasting problem, prepare production data and compare decline-curve, machine-learning and physics-informed approaches. The course covers time-aware validation, uncertainty ranges and scenario testing, alongside the ownership and monitoring arrangements needed for operational use. You will interpret P10/P50/P90 estimates using a stated probability convention and explain the assumptions behind the forecast.
AI-supported production forecasting uses machine-learning models to estimate future production from information such as historical rates, pressures, well tests and operating conditions. Participants assess its value by comparing predictions with engineering baselines on data that represents future use. Forecast suitability also depends on physical plausibility, uncertainty and the decision the forecast is intended to support.
The course evaluates AI alongside decline-curve analysis and reservoir simulation. Decline curves provide a forecasting benchmark, while simulation represents reservoir behaviour under defined assumptions. Participants examine where machine learning can add value, such as screening multiple wells or approximating simulation outputs, and where the evidence favours an established engineering method.
Physics-informed machine learning incorporates engineering knowledge, such as material balance or flow relationships, into model design, training or evaluation. The course also examines hybrid approaches, including capacitance-resistance models and simulation proxies. Participants assess whether these methods produce plausible forecasts and where their assumptions limit use under unfamiliar operating conditions.
Prepared notebooks allow participants to run and adjust models with guidance. Familiarity with spreadsheets, basic scripting or data analysis is helpful, alongside an understanding of reservoir and production concepts. The focus is on interpreting results, identifying weaknesses and judging whether a forecast is suitable for its intended decision.
Participants use time-aware data splits and backtesting to assess forecasts against later observations. They compare errors with decline-curve benchmarks, review uncertainty ranges and check physical plausibility. They also examine data leakage, where information unavailable at the forecast date incorrectly enters model development, and define how performance should be monitored as new data arrives.
Yes. Mature-field scenarios examine how production history, injection changes and operating events affect forecast reliability. Unconventional scenarios consider well-level variability, type curves and forecasting across larger well populations. Participants assess how the available data and reservoir behaviour influence model choice, validation and the interpretation of results.