A subsea tie-back study asks a small number of questions in a fixed order, and each answer constrains the next. This page walks that order and points at the class that answers each step.

The order matters

  1. What pressure can a closed-in well impose? That sets the flowline design pressure, which sets the wall thickness, which is part of the cooldown thermal mass. Start here.
  2. What size line? The erosional limit gives a floor on diameter.
  3. What inlet pressure does the host’s arrival pressure demand? The reservoir has to supply it, or the concept does not work.
  4. What wall thickness? DNV-ST-F101, against the design pressure from step 1.
  5. How much insulation? This depends on the wall from step 4 — that coupling is the one most often missed.

1. Shut-in wellhead pressure

SubseaWell.calculateShutInWellheadPressure integrates a static column from the reservoir datum to the seabed, re-flashing at each step so the density follows pressure and temperature. A single-density estimate over-predicts the drop badly for a gas column, because the gas is much lighter near the top.

SubseaWell well = new SubseaWell("P-1", stream);
well.setReservoirPressure(345.0);
well.setReservoirTemperature(90.0);
well.setTrueVerticalDepth(3100.0);
well.setWaterDepth(261.0);

double shutIn = well.calculateShutInWellheadPressure(reservoirFluid, 40.0, 60);

There is no friction term: a shut-in well is not flowing. The result is the settled-out pressure, which is the correct basis for a containment design.

2. Flowline size

FlowlineSizeSelector applies API RP 14E,

\[v_e = \frac{1.22\,c}{\sqrt{\rho}}\]

with $v_e$ in m/s, $\rho$ in kg/m³ and $c = 100$ for continuous service.

Evaluate it at the condition where the mixture is least dense — normally the arrival, not the inlet. Sizing on inlet density under-sizes the line.

FlowlineSizeSelector selector = new FlowlineSizeSelector()
    .setBasisFromFluid(fluidAtArrivalConditions, 16.13)
    .setWallThickness(0.0159);
Map<String, Object> selected = selector.select();

select() returns the smallest standard size that passes, or null when nothing does. Every candidate is in getCandidates() with its velocity and its percentage of the erosional limit.

This is a velocity screen only. A size that passes still has to deliver the required arrival pressure and pass DNV-ST-F101.

3. Hydraulics on a surveyed route

Use the two-fluid model for a wellstream

PipeBeggsAndBrills is a correlation fitted on small-diameter air-water laboratory loops at liquid fractions around 1–2 % and above. A wet-gas tie-back runs well below that, where the two-phase friction multiplier is extrapolated and over-predicts the pressure drop substantially. Prefer TwoFluidPipe, which resolves the phases separately.

Always check the solver’s convergence status before using a two-fluid profile:

pipe.run();
if (!pipe.isSteadyStateConverged()) {
  throw new IllegalStateException("Unconverged two-fluid profile");
}

MultiphaseFlowIntegrator defaults to HydraulicModel.TWO_FLUID and rejects non-converged profiles. Such a result has isFeasible() == false and an explanatory getInfeasibilityReason(); it must not be used as evidence of a feasible tie-back. Non-finite arrival pressure, temperature, velocity ratio or screening limits are also rejected. Set setHydraulicModel(HydraulicModel.BEGGS_BRILL) for a fast correlation check on a liquid-dominated line inside the correlation’s calibration range.

Before calculating screening velocities, the integrator refreshes outlet volume and mass density after the pipe has set its outlet flow. This prevents invalidated property caches from producing infinite velocities and a NaN erosional ratio.

sizePipeline(inlet, minimumArrivalPressureBara, maximumVelocityRatio) returns the smallest standard diameter that passes the screening constraints, including equality at the specified velocity-ratio limit. If no candidate passes, it throws IllegalStateException rather than recommending the largest failed candidate. The configured diameter is restored after success, exhaustion or an exception. Callers must handle sizing failure explicitly; this is a behavior change from the previous fallback. The velocity-ratio limit must be finite and positive, and the minimum arrival pressure must be finite and positive in bara.

Iteration count alone is not a validity gate: a short line can converge on the first refinement sweep. The solver’s pressure-floor, time-budget and convergence status determine whether its profile is usable.

Feeding it a real seabed

RouteProfile takes survey data once and hands each pipe model what it wants. Depths are positive downwards as surveyed; elevations are negative downwards as the models expect, and the conversion happens in one place.

RouteProfile route = RouteProfile.fromDepths(kpMetres, depthMetres)
    .withRiser(216.0, 25.0)
    .resample(60);
route.applyTo(twoFluidPipe);

getElevationProfile() deliberately has one more entry than getSectionLengths(), which is what TwoFluidPipe.setElevationProfile requires. getLowPointKp() returns the local minima — the terrain-slug traps, and the first place to look when a line is filling with liquid.

applyTo(PipeBeggsAndBrills) collapses the route to a single averaged segment and warns when the maximum inclination exceeds 5°, because that averaging discards the undulation that drives hold-up.

Solving for the inlet pressure

A host imposes an arrival pressure; the question is what inlet the reservoir must supply. Both pipe models march forward from a fixed inlet, so this used to be a hand-rolled bisection in every study.

pipe.setCalculationMode(PipeBeggsAndBrills.CalculationMode.CALCULATE_INLET_PRESSURE);
pipe.setOutletPressure(70.0, "bara");
pipe.run();
double requiredInlet = pipe.getSolvedInletPressure();

A trial that throws — typically because the pressure ran negative part way along — is treated as “inlet too low” rather than propagating, so the bracket closes from below. If the line cannot deliver the target at that rate at any inlet, the run fails with that stated explicitly.

The inlet-pressure bracket also searches below the specified arrival pressure. A downhill liquid line can gain pressure from hydrostatic head, so its required inlet pressure can be lower than the host pressure. The solve must meet the configured arrival-pressure tolerance before publishing getSolvedInletPressure(). If the iteration budget is exhausted, it throws IllegalStateException, restores the original inlet pressure, and leaves the solved-pressure getter as NaN.

4. Wall thickness

DnvStF101PipelineDesignCalculator returns every limit state with its utilisation and a PASS/FAIL status. Sweep the wall and take the thinnest that passes all of them:

DnvStF101PipelineAssessment assessment = DnvStF101PipelineDesignCalculator.calculate(input);
for (DnvStF101LimitStateCheck check : assessment.getChecks()) {
  // check.getLimitState(), check.getUtilization(), check.getStatus()
}

The governing check is usually system-test-pressure containment at high design pressure, and propagation buckling at low design pressure in deep water.

5. Insulation — and why it depends on step 4

This is the coupling that gets missed. The steel wall is part of the cooldown thermal mass. A decision that thins the wall — typically installing HIPPS so the line need not be rated for full shut-in — removes stored heat and lengthens the insulation needed to hold the same no-touch time.

A worked case: rating a tie-back for full shut-in required a 22.2 mm wall, and 75 mm of insulation comfortably exceeded an 8 h no-touch target. Protecting it instead and rating it for the flowing pressure dropped the wall to 14.3 mm — and the same 75 mm then gave only 7.9 h. The steel saving was real; so was the extra insulation. Neither analysis alone would have shown it.

TiebackThermalDesign sweeps both together:

TiebackThermalDesign design = new TiebackThermalDesign(wetFluid)
    .setInternalDiameter(0.1873)
    .setWallThicknesses(new double[] {0.0143, 0.0222})
    .setInsulationThicknesses(new double[] {0.050, 0.075, 0.100})
    .setSeabedTemperature(6.0)
    .setOperatingTemperature(35.0)
    .setRequiredNoTouchTime(8.0);
design.calculate();
double insulationForThinWall = design.getRequiredInsulationForWall(14.3);

The fluid must carry water

A fluid with no water cannot form hydrates, so SurfCooldownAnalyzer reports NO_HYDRATE_RISK with an unbounded no-touch time. That is correct for a deliberately dry gas — and it is the same answer a wet line gives when its fluid file simply has no water component. The verdict alone cannot tell the two apart.

So check it explicitly, and in a design workflow make it a hard gate:

analyzer.isWaterPresent();          // false for a dry gas AND for a wrong file
analyzer.getWaterMoleFraction();
analyzer.setRequireWater(true);     // throw rather than return an unbounded time

TiebackThermalDesign sets setRequireWater(true) by default, because in a tie-back design a fluid without water is almost always the wrong file.

Add water through NeqSim’s own path, which sets the water binary interaction parameters and enables the VLLE flash — not a bare addComponent:

SystemInterface wet = EclipseFluidReadWrite.read(file, true);        // kij 0.5
EclipseFluidReadWrite.addWaterToFluid(existingFluid, 0.5);           // in place

Do not call setMixingRule after EclipseFluidReadWrite.read: read installs the mixing rule and the file’s BIC block, and setMixingRule reloads database kij, silently discarding any regressed interaction parameters the characterisation carried.

Checklist