Compressor off-design coordinates and train candidate assessment

NeqSim already provides compressor map backends, MW-map interpolation, capacity constraints, operating-point snapshots, CompressorOptimizationHelper, ProductionOptimizer, CompressorAntiSurgeApplication and minimum-speed pressure/recycle coordination. Use these models to solve the process before comparing candidate operating settings.

Public map queries

CompressorChart.polytropicEfficiency(flow,speed) now delegates to the selected backend’s actual efficiency calculation (percent) instead of returning 100. checkSurge1(flow,head) uses an active explicit head-based surge curve. checkSurge2(flow,speed) and checkStoneWall(flow,speed) compare with digitized minimum/maximum flow endpoints at the nearest map speed. Equality is not classified as outside the boundary; it is not an independent adequate surge-control margin either.

Missing/nonfinite boundary evidence raises an exception rather than returning an unconditional false. Invalid coordinates are rejected. The digitized endpoints are screening bounds; only OEM data identifies a physical surge/choke boundary. Speed-based query methods can use a nearest curve outside the digitized speed range: inspect speed bounds separately. The lowercase getStonewallFlowAtSpeed dispatches to getStoneWallFlowAtSpeed, so the MW interpolation backend can supply the appropriate maximum-flow bound.

Acoustic similarity and flow basis

CompressorMapSimilarity explicitly transforms coordinates at fixed geometry and matched inlet Mach number/flow coefficient. Construct it with reference/actual sound speed in m/s and reference/actual density in kg/m3, or use fromFluids(reference,actual), which flashes owned clones, verifies single-phase gas and reads EOS isentropic sound speed and density. The original fluids and compressor chart remain unchanged.

With $r_a=a_{ref}/a_{act}$ and $r_\rho=\rho_{ref}/\rho_{act}$:

\[Q_{ref}=Q_{act}r_a,\qquad N_{ref}=N_{act}r_a\] \[H_{ref}=H_{act}r_a^2,\qquad \dot m_{ref}=\dot m_{act}r_a r_\rho\]

$Q$ is actual suction volume in m3/hr, $N$ is rpm, $H$ is kJ/kg and $\dot m$ is kg/s. Normal/standard volume is a different quantity and cannot be passed as actual suction flow. Inverse volume/mass transforms support a reference stonewall screening estimate. For the same ideal gas at unchanged temperature, doubling pressure preserves sound speed and actual-volume coordinates while doubling density and achievable actual mass flow. Temperature and composition also change sound speed; pressure can change real-gas sound speed through the EOS. These are similarity coordinates, not proof that efficiency, surge or choke curves remain invariant under arbitrary composition/real-gas changes.

The existing CompressorCurveCorrections.calculateSonicVelocity uses the approximate $\sqrt{k Z R T/M}$ expression. It is not the exact real-gas isentropic derivative used by fromFluids. The new API does not alter the legacy empirical correction helpers or claim ASME PTC-10 conformance. Reynolds number, Mach range, polytropic path and OEM applicability still require qualification.

Whole-train candidate comparison

CompressionTrainAssessment reads already-converged compressor bodies and a caller-identified net export stream. It sums each body’s actual shaft power once, checks the aggregate power budget, stage operating-point/pressure/capacity evidence, wet suction, required map evidence, digitized flow/speed range and caller-selected minimum surge/stonewall margins. It also checks actual net export pressure and positive single-phase gas export.

Specific shaft energy is:

\[e_{shaft}=\frac{\sum_i P_{shaft,i}}{\dot m_{net,export}}\]

Power is kW and net export is kg/s, giving kJ/kg. Serial stage flows are not summed as production. Gross recycled gas is not substituted for net export. Choose a consistent net stream after recycle takeoff. The API cannot infer topology, convergence, hidden constraint providers or stale model state; these are responsibilities of the enclosing process solve. A chartless thermodynamic case can be assessed only when requireMaps=false; that choice cannot qualify surge/choke protection or a speed limit.

lowestSpecificEnergy selects only feasible candidates and returns null if none passes. A low-power candidate that misses pressure or violates a driver/equipment limit cannot win. The result is a detached snapshot suitable for bounded cooler-temperature, HP2-speed, pressure and recycle/load-sharing sweeps. It does not install a continuous load-sharing controller or run an automatic optimization.

Define separate assessments for separate shaft-power-budget groups. This initial aggregation does not validate common-shaft speed/torque, motor efficiency, cooler/coolant power, separator capacity, antiwindup, valve response, train-trip dynamics or controller authority. Evaluate these with the existing enclosing models and the acceptance cases in control/load-sharing #4146, train optimization #4147 and off-design maps #4148.

Validation evidence

CompressorChartQueryTest exercises two map backends, missing/invalid evidence, equality and MW-interpolated endpoints. CompressorMapSimilarityTest checks analytical identity/round-trip, ideal temperature/pressure invariants, EOS composition/T/P trends and rejection of liquid states. CompressionTrainAssessmentTest solves a synthetic methane/ethane two-stage train, compares 300 K and 340 K interstage cooling on the same 10 kg/s net export basis, checks summed power and specific-energy units, and rejects inadequate pressure, power/capacity limits, missing map evidence, duplicates and missing/zero export. This qualifies software screening and physical invariants, not off-design OEM maps or closed-loop dynamic performance.

See compressor controls, compressor curves and program roadmap #4149. Primary sources for the physical/control basis are NASA compressible mass-flow equations and CCC antisurge/control coordination.