Train a model on CSV data, evaluate unseen wells, and investigate operating constraints with an explainable advisor.
Three ways to lift a well.
Change the well. Tune the equipment. Compare the response.
Each curve changes only the selected operating control; all other inputs remain fixed. Rates are liquid volume at model conditions, with formation volume factor assumed 1.
Model equations & boundaries
Inflow: Pwf = Pr − Q / PI. Outflow: Pwh + ρgD / 10⁶ + kQ² (MPa). Constant density, linear PI, vertical well and steady flow are assumed.
Gas lift: illustrative gas fraction α = 0.65I / (I + 10,000); effective density = ρ(1 − α); added friction = 2 × 10⁻⁹I² MPa. I is standard m³/d. This deliberately simplified response is not a validated multiphase or gas-expansion correlation.
ESP: H = N × 12 × [(f/50)² − (Q/400)²] m within the reference curve. N is stage count. Generic 50 Hz reference curve, not a manufacturer's pump. Power = ρgQH / η with Q in m³/s and η = 0.65. Operating rate solves inflow versus outflow minus pump boost.
Sucker-rod: Q = πd²S × SPM × 1440 × fillage / 4, capped by reservoir supply at an assumed 1 MPa minimum intake pressure. Downhole stroke is specified directly. Power uses required hydraulic boost / 0.65. Fluid load excludes rod weight, inertia, friction and stress.
Educational comparison only. No gas-lift valve design, compressor power, ESP gas handling, viscosity corrections, motor limits, rod dynamics or field calibration. A larger predicted rate does not establish the best lift method. Pump setting depth equals reservoir depth. No cost ranking is implied.
Suggested experiments & references
- Sweep gas injection and observe the competition between density reduction and friction.
- Change ESP frequency or stage count and inspect the production response.
- Reduce reservoir pressure and increase rod-pump speed to observe the supply cap.
SLB: Artificial lift · KSB: Pump affinity laws
These references support the operating principles; all numerical equipment curves and gas-lift coefficients here are illustrative assumptions.