Reservoir Material-Balance Analysis
Use pressure and cumulative production to test an in-place-volume hypothesis. This is a tank-level
screening method, not a spatial reservoir simulator or a reserves classification.
When to Use
- Estimate or reconcile gas initially in place (GIIP), original oil in place (OOIP), or total
surface-component inventories from pressure and production history.
- Compare a prior volumetric estimate with a pressure-production material balance.
- Test depletion-drive, gas-cap, known-water-influx, or known-compressibility cases.
- Separate formations or pressure compartments before matching a field total.
For compositional gas-condensate depletion, produced composition, or wellstream forecasts, use
neqsim-reservoir-history-match-forecast. Use both workflows only when their fluid and volume bases
are explicitly reconciled.
Inputs
- Dated static/average reservoir pressures, including test type, depth, unit, datum, source and
correction provenance.
- Formation-allocated cumulative production and injection for oil, gas and water on one declared
surface-volume basis.
- Initial and pressure-dependent PVT, initial pressure and temperature, and relevant rock/fluid
compressibilities.
- Supported formation/tank boundaries, pressure communication evidence, and gas-cap/aquifer inputs
where applicable.
- Prior in-place estimates and their date, range, volumetric/PVT basis and allocation assumptions.
Outputs
- In-place volumes and units by tank and supported model case.
- Pressure/volume residuals, fit diagnostics and identifiability warnings.
- Cumulative production, injection, aquifer influx and drive contributions on a consistent basis.
- Sensitivity range, comparison with the prior estimate, data gaps and discriminating next evidence.
Required Evidence
Before fitting, assemble one time-aligned history per candidate tank:
- Static or otherwise justified average reservoir pressure with date, wellbore, test type, gauge
depth, pressure unit, datum, source, quality and correction method.
- Cumulative oil, gas and water produced and injected, on one declared surface-volume basis. Keep
the formation allocation and its uncertainty; a field total is not automatically a formation
history.
- PVT at initial conditions and at each pressure observation: pressure, Bo, Bg, Bw, Rs and Rv as
applicable, with units, standard conditions and separator basis.
- Initial pressure, temperature, formation/water compressibility and saturation assumptions.
- Formation and pressure-compartment mapping, well completion intervals, shut-in/stabilization
status, and evidence for or against pressure communication.
- The previous in-place estimate, estimate date, range, and the pore-volume, saturation, PVT and
allocation basis used to calculate it.
Normalize pressures to absolute pressure and a common datum before matching. Preserve the raw
measurement and the correction alongside every normalized value. Do not mix flowing pressures with
static observations or treat a wellhead pressure as average tank pressure without a documented
conversion.
Select the Model
| Reservoir/fluid case |
NeqSim route |
Important boundary |
| Dry or lean gas, closed tank, negligible water drive |
MaterialBalanceHistoryMatch.fitPzLine |
This is a p/z versus cumulative-production line fit. Check the pressure and Z basis, residuals and outliers; sigma clipping does not validate the data. |
| Conventional black-oil depletion or gas-cap drive |
OilMaterialBalance |
Build and inspect withdrawal and expansion terms before fitDepletionDrive or fitGasCapDrive. |
| Conventional black-oil water drive |
OilMaterialBalance.fitWaterDrive |
Requires an independently specified cumulative aquifer-influx history. It does not infer that history from pressure alone. |
| Initially oil or gas, with both solution gas Rs and vaporized oil Rv |
GeneralizedMaterialBalance with BlackOilPVTTable.Record |
Use one fixed surface basis and net cumulative produced-minus-injected components. The fitted residual is reservoir-volume balance, not pressure error. |
| Compositional gas condensate with changing produced composition |
neqsim-reservoir-history-match-forecast |
Use a tuned EOS and composition/yield observations; do not substitute a black-oil fit when liquid dropout and composition change matter. |
OilMaterialBalance uses the Havlena-Odeh balance
F = N(Eo + m Eg + Efw) + We Bw. Its depletion-drive and gas-cap regressions estimate OOIP
(surface oil volume); the water-drive regression takes known We. GeneralizedMaterialBalance
supports initially-oil, initially-gas, specified-gas-cap and joint oil/gas fits. It rejects
non-identifiable two-component fits by a scaled condition-number gate, but that gate is not a full
uncertainty interval.
Engineering Method
- Freeze the basis. Record standard conditions, surface-volume units, pressure datum, initial
state, PVT source, rate allocation, and the date through which each history is final.
- Choose tank boundaries. Split formations or compartments only when the pressure and
completion evidence supports distinct inventories. Keep alternative tank maps when the evidence
is ambiguous.
- Check closure and inventory. Reconcile allocated field and well totals, net production versus
injection, and formation-level allocation. A fitted inventory must not be smaller than the net
component withdrawal it is required to contain.
- Build competing hypotheses. Include closed depletion, supported aquifer, gas-cap, and
compressibility cases only where their inputs are evidenced. Hold uncertain allocation and
pressure corrections as explicit sensitivities, not hidden fit parameters.
- Fit and diagnose. Report residuals by date, tank and observation source; retained/rejected
pressure points; condition number or other identifiability warnings; bound hits; and sensitivity
to tank split, PVT, allocation, compressibility and aquifer assumptions.
- Compare with the prior estimate. Compare on the same initial date and surface basis. Reconcile
differences back to measurable inputs and present ranges for each defensible model structure.
- Separate in-place from recoverable volumes. Do not call a material-balance inventory a
recoverable reserve or a forecast without a supported recovery and deliverability model.
Python Usage Pattern
This is a basis-selection and result-interpretation workflow, not a generic runnable script. Use
capability discovery and verified NeqSim API signatures for the selected Java model. Keep data
preparation, unit normalization and plotting in a task notebook only when needed; validate the
calculation independently before reporting an in-place estimate.
Validation Checklist
- Pressure observations are static/average values normalized to absolute pressure and a common
datum with raw values retained.
- Production and injection are formation-allocated and reconciled on one surface-volume basis.
- The fitted inventory satisfies net component withdrawal and the selected PVT/model assumptions.
- Fit residuals, identifiability, bounds and sensitivity are reported; pressure fit alone is not
treated as proof of a unique volume.
- The prior estimate comparison uses the same initial date and volume basis.
Common Mistakes
- Inferring a lower in-place volume from pressure observations alone.
- Treating flowing pressure as average tank pressure or ignoring datum corrections.
- Fitting independent aquifers per formation and labeling them one shared aquifer.
- Comparing field-total production with a formation-level balance without validated allocation.
- Treating a material-balance inventory as recoverable reserves or a production forecast.
Aquifer and Shared-Support Gate
Pressure decline by itself cannot distinguish a larger tank with weaker support from a smaller tank
with stronger support. Use measured aquifer pressure, water production/injection, pressure
communication, aquifer geometry and compressibility to constrain support. Fetkovich or
Van Everdingen-Hurst aquifer calculations are forward scenarios when their physical inputs are
specified; they do not establish aquifer size from pressure history alone.
For multiple formations that may share an aquifer, do not fit each formation with an independent
aquifer and present the results as one common-aquifer model. A shared support hypothesis needs one
consistent coupled balance and a documented allocation of influx to the tanks. If the selected
workflow cannot solve that coupled structure, report the alternative cases and the limitation, or
route the question to a reservoir simulator for a coupled evaluation.
Output Interpretation
For every tank and case, report:
- Model type, tank boundaries and evidence tier for each input.
- Estimated in-place volume and units; fitted gas-cap ratio when applicable.
- Pressure/volume residuals and fit diagnostics, including identifiability or condition warnings.
- Cumulative production, injection, aquifer influx and drive contributions on the same basis.
- Sensitivity range and comparison with the previous estimate.
- Data gaps, unsupported assumptions, and the observation or test that would best discriminate
between competing hypotheses.
Limitations
Stop Conditions
- No unambiguous initial PVT or surface-volume basis.
- Pressure points cannot be reconciled to a common datum, or are mainly flowing/non-stabilized.
- Formation-level production allocation is missing and materially changes the inferred volume.
- An aquifer or gas-cap case is being fitted without an independent constraint.
- A joint oil/gas fit is ill-conditioned, violates component inventory, or depends on a bound.
- Competing tank/support hypotheses fit within the observation uncertainty but imply different
in-place volumes.
In these cases report the limitation and the next required evidence. Do not silently substitute a
default drive mechanism or report a single best-fit volume as a conclusion.
References
- Havlena, D. and Odeh, A.S. (1963, 1964), material-balance analysis of volumetric and water-drive
reservoirs.
- Walsh, M.P. (1995), generalized material balance for oil and gas reservoirs,
doi:10.2118/95-01-07.
- Craft, B.C. and Hawkins, M.F., Applied Petroleum Reservoir Engineering.
1---2name: neqsim-reservoir-material-balance3calculation_basis: "neqsim-java"4version: "0.1.0"5description: "Screen gas and oil reservoir in-place volumes with pressure-production material balance, including gas p/z, Havlena-Odeh, black-oil and volatile-oil fits, aquifer-support scenarios, uncertainty and identifiability checks. USE WHEN: an MBAL model, STOIIP/OOIP/GIIP estimate, pressure history match, lower-in-place-volume claim, gas-cap or water-drive evaluation is needed."6last_verified: "2026-10-31"7requires:8 python_packages: [numpy, pandas, jneqsim]9 java_packages: [neqsim]10 env: []11 network: []12metadata:13 source: https://skillmd.com/skills/equinor/neqsim-reservoir-material-balance14---1516# Reservoir Material-Balance Analysis1718Use pressure and cumulative production to test an in-place-volume hypothesis. This is a tank-level19screening method, not a spatial reservoir simulator or a reserves classification.2021## When to Use2223- Estimate or reconcile gas initially in place (GIIP), original oil in place (OOIP), or total24 surface-component inventories from pressure and production history.25- Compare a prior volumetric estimate with a pressure-production material balance.26- Test depletion-drive, gas-cap, known-water-influx, or known-compressibility cases.27- Separate formations or pressure compartments before matching a field total.2829For compositional gas-condensate depletion, produced composition, or wellstream forecasts, use30`neqsim-reservoir-history-match-forecast`. Use both workflows only when their fluid and volume bases31are explicitly reconciled.3233## Inputs3435- Dated static/average reservoir pressures, including test type, depth, unit, datum, source and36 correction provenance.37- Formation-allocated cumulative production and injection for oil, gas and water on one declared38 surface-volume basis.39- Initial and pressure-dependent PVT, initial pressure and temperature, and relevant rock/fluid40 compressibilities.41- Supported formation/tank boundaries, pressure communication evidence, and gas-cap/aquifer inputs42 where applicable.43- Prior in-place estimates and their date, range, volumetric/PVT basis and allocation assumptions.4445## Outputs4647- In-place volumes and units by tank and supported model case.48- Pressure/volume residuals, fit diagnostics and identifiability warnings.49- Cumulative production, injection, aquifer influx and drive contributions on a consistent basis.50- Sensitivity range, comparison with the prior estimate, data gaps and discriminating next evidence.5152## Required Evidence5354Before fitting, assemble one time-aligned history per candidate tank:5556- Static or otherwise justified average reservoir pressure with date, wellbore, test type, gauge57 depth, pressure unit, datum, source, quality and correction method.58- Cumulative oil, gas and water produced and injected, on one declared surface-volume basis. Keep59 the formation allocation and its uncertainty; a field total is not automatically a formation60 history.61- PVT at initial conditions and at each pressure observation: pressure, Bo, Bg, Bw, Rs and Rv as62 applicable, with units, standard conditions and separator basis.63- Initial pressure, temperature, formation/water compressibility and saturation assumptions.64- Formation and pressure-compartment mapping, well completion intervals, shut-in/stabilization65 status, and evidence for or against pressure communication.66- The previous in-place estimate, estimate date, range, and the pore-volume, saturation, PVT and67 allocation basis used to calculate it.6869Normalize pressures to absolute pressure and a common datum before matching. Preserve the raw70measurement and the correction alongside every normalized value. Do not mix flowing pressures with71static observations or treat a wellhead pressure as average tank pressure without a documented72conversion.7374## Select the Model7576| Reservoir/fluid case | NeqSim route | Important boundary |77|---|---|---|78| Dry or lean gas, closed tank, negligible water drive | `MaterialBalanceHistoryMatch.fitPzLine` | This is a p/z versus cumulative-production line fit. Check the pressure and Z basis, residuals and outliers; sigma clipping does not validate the data. |79| Conventional black-oil depletion or gas-cap drive | `OilMaterialBalance` | Build and inspect withdrawal and expansion terms before `fitDepletionDrive` or `fitGasCapDrive`. |80| Conventional black-oil water drive | `OilMaterialBalance.fitWaterDrive` | Requires an independently specified cumulative aquifer-influx history. It does not infer that history from pressure alone. |81| Initially oil or gas, with both solution gas Rs and vaporized oil Rv | `GeneralizedMaterialBalance` with `BlackOilPVTTable.Record` | Use one fixed surface basis and net cumulative produced-minus-injected components. The fitted residual is reservoir-volume balance, not pressure error. |82| Compositional gas condensate with changing produced composition | `neqsim-reservoir-history-match-forecast` | Use a tuned EOS and composition/yield observations; do not substitute a black-oil fit when liquid dropout and composition change matter. |8384`OilMaterialBalance` uses the Havlena-Odeh balance85`F = N(Eo + m Eg + Efw) + We Bw`. Its depletion-drive and gas-cap regressions estimate OOIP86(surface oil volume); the water-drive regression takes known `We`. `GeneralizedMaterialBalance`87supports initially-oil, initially-gas, specified-gas-cap and joint oil/gas fits. It rejects88non-identifiable two-component fits by a scaled condition-number gate, but that gate is not a full89uncertainty interval.9091## Engineering Method92931. **Freeze the basis.** Record standard conditions, surface-volume units, pressure datum, initial94 state, PVT source, rate allocation, and the date through which each history is final.952. **Choose tank boundaries.** Split formations or compartments only when the pressure and96 completion evidence supports distinct inventories. Keep alternative tank maps when the evidence97 is ambiguous.983. **Check closure and inventory.** Reconcile allocated field and well totals, net production versus99 injection, and formation-level allocation. A fitted inventory must not be smaller than the net100 component withdrawal it is required to contain.1014. **Build competing hypotheses.** Include closed depletion, supported aquifer, gas-cap, and102 compressibility cases only where their inputs are evidenced. Hold uncertain allocation and103 pressure corrections as explicit sensitivities, not hidden fit parameters.1045. **Fit and diagnose.** Report residuals by date, tank and observation source; retained/rejected105 pressure points; condition number or other identifiability warnings; bound hits; and sensitivity106 to tank split, PVT, allocation, compressibility and aquifer assumptions.1076. **Compare with the prior estimate.** Compare on the same initial date and surface basis. Reconcile108 differences back to measurable inputs and present ranges for each defensible model structure.1097. **Separate in-place from recoverable volumes.** Do not call a material-balance inventory a110 recoverable reserve or a forecast without a supported recovery and deliverability model.111112## Python Usage Pattern113114This is a basis-selection and result-interpretation workflow, not a generic runnable script. Use115capability discovery and verified NeqSim API signatures for the selected Java model. Keep data116preparation, unit normalization and plotting in a task notebook only when needed; validate the117calculation independently before reporting an in-place estimate.118119## Validation Checklist120121- Pressure observations are static/average values normalized to absolute pressure and a common122 datum with raw values retained.123- Production and injection are formation-allocated and reconciled on one surface-volume basis.124- The fitted inventory satisfies net component withdrawal and the selected PVT/model assumptions.125- Fit residuals, identifiability, bounds and sensitivity are reported; pressure fit alone is not126 treated as proof of a unique volume.127- The prior estimate comparison uses the same initial date and volume basis.128129## Common Mistakes130131- Inferring a lower in-place volume from pressure observations alone.132- Treating flowing pressure as average tank pressure or ignoring datum corrections.133- Fitting independent aquifers per formation and labeling them one shared aquifer.134- Comparing field-total production with a formation-level balance without validated allocation.135- Treating a material-balance inventory as recoverable reserves or a production forecast.136137## Aquifer and Shared-Support Gate138139Pressure decline by itself cannot distinguish a larger tank with weaker support from a smaller tank140with stronger support. Use measured aquifer pressure, water production/injection, pressure141communication, aquifer geometry and compressibility to constrain support. Fetkovich or142Van Everdingen-Hurst aquifer calculations are forward scenarios when their physical inputs are143specified; they do not establish aquifer size from pressure history alone.144145For multiple formations that may share an aquifer, do not fit each formation with an independent146aquifer and present the results as one common-aquifer model. A shared support hypothesis needs one147consistent coupled balance and a documented allocation of influx to the tanks. If the selected148workflow cannot solve that coupled structure, report the alternative cases and the limitation, or149route the question to a reservoir simulator for a coupled evaluation.150151## Output Interpretation152153For every tank and case, report:154155- Model type, tank boundaries and evidence tier for each input.156- Estimated in-place volume and units; fitted gas-cap ratio when applicable.157- Pressure/volume residuals and fit diagnostics, including identifiability or condition warnings.158- Cumulative production, injection, aquifer influx and drive contributions on the same basis.159- Sensitivity range and comparison with the previous estimate.160- Data gaps, unsupported assumptions, and the observation or test that would best discriminate161 between competing hypotheses.162163## Limitations164165### Stop Conditions166167- No unambiguous initial PVT or surface-volume basis.168- Pressure points cannot be reconciled to a common datum, or are mainly flowing/non-stabilized.169- Formation-level production allocation is missing and materially changes the inferred volume.170- An aquifer or gas-cap case is being fitted without an independent constraint.171- A joint oil/gas fit is ill-conditioned, violates component inventory, or depends on a bound.172- Competing tank/support hypotheses fit within the observation uncertainty but imply different173 in-place volumes.174175In these cases report the limitation and the next required evidence. Do not silently substitute a176default drive mechanism or report a single best-fit volume as a conclusion.177178## References179180- Havlena, D. and Odeh, A.S. (1963, 1964), material-balance analysis of volumetric and water-drive181 reservoirs.182- Walsh, M.P. (1995), generalized material balance for oil and gas reservoirs,183 doi:10.2118/95-01-07.184- Craft, B.C. and Hawkins, M.F., *Applied Petroleum Reservoir Engineering*.