Reservoir Layer
Applicable to
- Dry-gas and slightly wet-gas reservoirs (CGR < 5 STB/MMscf)
- Volumetric reservoirs (or with explicit aquifer term, future extension)
- Single-phase gas above the dewpoint
- Multi-tank systems with inter-layer transmissibility
- Black-oil reservoirs — generalised Schilthuis material balance with
solution-gas drive, Corey relative permeability and a producing GOR from
the mobility ratio. Standing and Glasø correlations supply the PVT.
Not applicable to
- Volatile-oil systems — the black-oil PVT treatment does not hold near the
critical point; needs a compositional model
- Oil PVT verified by behaviour, not against a published table. Bubble-point
capping, monotonic Rs, the Bo peak and its continuity,
and the viscosity trend are all checked; a systematic offset in the
correlations would not be caught by those checks
- Strong water influx without aquifer model (Carter-Tracy / Fetkovich)
- Retrograde condensate below dewpoint — needs compositional MBE
- CBM with desorption — needs Langmuir term
- Reservoirs with significant pressure-dependent permeability
Well Layer
Applicable to
- Vertical gas wells with constant tubing geometry
- Pseudo-steady-state radial flow (after transient effects die out)
- Single-phase gas in the wellbore
Not applicable to
- Horizontal wells — needs different inflow model (Joshi, Babu-Odeh)
- Fractured wells in transient flow — needs trilinear/finite-conductivity model
- Liquid loading — the Coleman/Turner critical rate is computed and reported
per well, but a loaded well is flagged rather than having its rate reduced
- Multi-phase wellbore flow (significant liquid holdup)
Surface Network
Applicable to
- Gas gathering systems (single-phase dry gas)
- Horizontal pipes (negligible elevation change)
- Tree topology (each station has one downstream destination)
- Steady-state pressure drop at each timestep
- Manifolds — a station that feeds another is a junction whose pressure is
derived from the trunk at the total rate passing through, so wells sharing
that line interfere with one another
- Compression — adiabatic power with equal-ratio staging, at the station or
on individual wells. A machine has finite installed power, so the suction
it can hold rises with throughput; the setpoint is not always reached
Not applicable to
- Looped/redundant gathering networks — needs Newton-Raphson with continuity
- Significant elevation profiles in pipelines — needs Beggs-Brill or vertical correction
- Multi-phase pipe flow (slug flow, liquid accumulation) — flowlines are
solved as dry gas by Weymouth
- Choke modeling
- Line pack — the surface network is steady-state, so pipeline storage
during a rate change is not represented
Numerical Choices
| Parameter | Default | Rationale |
|---|
| Timestep | 30.4 days | Monthly accounting matches typical operator reporting cadence |
| m(p) table | 200 intervals | Simpson's rule converges to <0.01% by 100 intervals; 200 for margin |
| MBE bisection | 40 iterations | Converges to <0.05 psi for any reasonable bracket |
| Nodal inner loop | 20 iterations | Damped fixed-point typically converges in 5-10 |
| Outer network loop | 6 iterations | Sufficient for most cases; convergence checked on rate change |
| Convergence tolerance | 1% of mean rate | Tighter is academic; physical uncertainty dominates |
Coordinate Conventions
- Depth measured along the wellbore from rotary table (or kelly bushing) downward. For vertical wells, this equals true vertical depth.
- Pipe length is the actual flow path length, not horizontal distance.
- Temperature uses Rankine throughout the engine; UI inputs are converted internally.
- Pressure is absolute (psia), not gauge.
Calibration Discipline
The "matched" status only means the engine could fit the data — it does not validate the underlying model. Always:
- Inspect the R² and visually overlay history vs match before trusting
- Cross-check calibrated G against volumetric estimates
- Be wary of unrealistically negative skin or unphysical OGIP
- Use the upload feature with field-verified data, not interpolated values
This is a forecasting tool, not a simulator. Numerical reservoir simulators handle geometry, heterogeneity, and multi-phase physics that IPRS deliberately omits in exchange for speed and transparency. Use IPRS for system-level decisions, screening studies, and what-if analysis; use a finite-difference simulator when grid-level accuracy is required.