Derive well flow rates from choke and vertical lift performance equations using PVT-based fluid properties.
Oil, Gas & Energy
Virtual Flow Metering with Physics-Based and Data-Driven Models
Design virtual flow meters that blend choke hydraulics and multiphase correlations with machine learning to estimate oil, gas and water rates between physical well tests.
🗓️Duration5 training days
📚Content4 modules · 8 sessions
🎓On completionAccredited attendance certificate
About the programme
Course Overview
Wellhead multiphase flow meters are costly to install and maintain, so most producing wells are measured only periodically by test separator or portable multiphase meter while day-to-day rates are inferred by a virtual flow meter. This course examines how those inferred rates are actually produced: mechanistic choke and vertical lift performance equations built on PVT fluid characterisation, correlations such as Beggs and Brill for pressure drop, and data-driven regression and neural network models trained on historian tags, plus the hybrid grey-box designs that combine both. Participants work through calibration against well test data, bias correction, uncertainty banding, and the drift diagnostics that trigger recalibration when choke wear or watercut changes shift model accuracy. Sessions move from single-well modelling to field-wide allocation, covering the governance needed when virtual rates feed fiscal or partner allocation statements. Teaching combines annotated calculation walkthroughs, a hands-on model-build exercise using sample production data, and review of validation reports. Participants finish able to specify a virtual flow metering programme, judge model performance against test data, and write the validation and uncertainty documentation an allocation committee will accept.
Expected Learning Outcomes
Build a data-driven regression or neural network model to estimate multiphase rates from historian sensor tags.
Combine mechanistic and data-driven components into a hybrid grey-box virtual flow meter.
Calibrate virtual flow meter outputs against test separator and multiphase meter measurements.
Quantify uncertainty bands around virtual flow meter estimates for allocation and reporting purposes.
Diagnose model drift caused by choke wear, watercut change or sensor bias and trigger recalibration.
Draft the validation and governance documentation required to support fiscal and partner allocation statements.
Who Should Attend
Production and reservoir engineers responsible for well surveillance and allocation metering.
Instrumentation and control engineers who maintain SCADA and historian data feeds for producing wells.
Data scientists assigned to build or support machine learning models for upstream production data.
Production allocation specialists who prepare fiscal and partner statements from measured and inferred rates.
Field and asset engineers on subsea or unmanned platforms where physical multiphase metering is impractical.
Digital oilfield and production optimisation teams evaluating virtual metering technology for a producing asset.
Course Modules
Select any module to see its sessions and points.
01Multiphase Flow Fundamentals and Physics-Based Modelling
2 sessions · 8 points
Session 1Fluid Behaviour and Choke Performance Equations
- Characterise reservoir fluid behaviour with PVT data, including bubble point, gas-oil ratio and formation volume factor.
- Apply critical and sub-critical choke flow equations to relate wellhead pressure drop to flow rate.
- Model vertical lift performance and inflow performance relationships to establish a well's operating point.
- Select and apply mechanistic multiphase correlations, including Beggs and Brill, for pipe and riser pressure loss.
Session 2Building the Mechanistic Virtual Flow Meter
- Assemble a steady-state mechanistic model linking reservoir, wellbore and choke into a single flow estimate.
- Identify the sensor inputs a mechanistic model needs from wellhead pressure, temperature and choke position tags.
- Test model sensitivity to PVT assumptions and identify which parameters dominate estimation error.
- Document mechanistic model assumptions and limitations for later validation against well test data.
02Data-Driven and Hybrid Modelling Techniques
2 sessions · 8 points
Session 1Machine Learning Models for Rate Estimation
- Engineer features from historian time series, including pressure, temperature, choke position and vibration tags.
- Train regression, random forest or neural network models to estimate multiphase rates from sensor features.
- Evaluate data-driven model accuracy against held-out well test data using appropriate error metrics.
- Recognise the extrapolation risk of data-driven models when operating conditions move outside the training range.
Session 2Hybrid Grey-Box Model Design
- Combine a mechanistic backbone with a data-driven correction term to build a grey-box virtual flow meter.
- Use data-driven residual modelling to compensate for mechanistic model bias without discarding physical structure.
- Balance model complexity against maintainability when choosing between physics-based, data-driven and hybrid designs.
- Select a modelling architecture appropriate to well type, instrumentation and available historical data.
03Calibration, Uncertainty and Drift Management
2 sessions · 8 points
Session 1Calibration against Well Test Data
- Design a well test schedule that supplies sufficient calibration points across the expected operating envelope.
- Reconcile virtual flow meter output against test separator and multiphase meter readings to derive bias corrections.
- Quantify estimation uncertainty and express it as a confidence band around the reported rate.
- Apply measurement standards guidance relevant to allocation metering when documenting calibration evidence.
Session 2Detecting and Managing Model Drift
- Set drift detection thresholds that flag when virtual flow meter error exceeds an acceptable limit.
- Distinguish drift caused by choke wear or scale build-up from drift caused by changing watercut or gas-oil ratio.
- Define a recalibration workflow that updates model parameters without interrupting production reporting.
- Build a monitoring dashboard that tracks virtual flow meter performance against test data over time.
04Deployment, Allocation and Governance
2 sessions · 8 points
Session 1Integrating Virtual Flow Meters into Allocation
- Distinguish fiscal metering requirements from allocation metering requirements in a commingled production system.
- Apply virtual flow meter output within an allocation methodology shared by multiple working-interest partners.
- Explain how allocation factors are adjusted when virtual flow meter estimates disagree with periodic well tests.
- Coordinate with partners and regulators on the metering philosophy documented in a measurement management plan.
Session 2Programme Governance and Continuous Improvement
- Write a validation report that presents model performance, uncertainty and calibration history to an allocation committee.
- Define roles and responsibilities for maintaining virtual flow meter models across engineering and data teams.
- Plan a phased rollout of virtual flow metering across a field, prioritising wells by value of information.
- Establish a review cycle that reassesses model choice as well behaviour and instrumentation change over field life.
What the participant receives
4 course modules
A structured syllabus
8 training sessions
across 5 days
32 detailed points
Applied, detailed content
Accredited attendance certificate
On completing the programme
Complete your registration
We will contact you within one business day to confirm.
Ready to start?
Reserve your seat and start building the skill.