feat: add PressureRatioCompressor and CommonPressureRatioSolver by olelod · Pull Request #1483 · equinor/ecalc

olelod · 2026-04-14T12:34:15Z

What

Introduces a ratio-based solving path for compressor trains that operate at a fixed pressure ratio — no shaft speed modelling. This is the domain foundation for the upcoming SIMPLIFIED_SERIAL YAML type (follow-up PR).

Why

The existing solver (CommonSpeedWithPressureControlSolver, formerly OutletPressureSolver) searches for a shaft speed to meet a target outlet pressure. Simplified compressor trains (using GENERIC_FROM_INPUT or GENERIC_FROM_DESIGN_POINT charts) don't model shaft speed — they distribute the total pressure ratio equally
across stages. This PR adds domain support for that approach.

Changes

New domain entities:

PressureRatioCompressor — a ProcessUnit that evaluates at a given pressure ratio via propagate_stream_at_pressure_ratio(). Only accepts generic chart types.
CommonPressureRatioSolver — distributes equal pressure ratio (geometric mean) across all PressureRatioCompressor stages, applies minimum recirculation per stage (individual anti-surge). Exposes get_max_standard_rate() for stonewall-based feasibility.

Solver hierarchy refactored:

New ProcessSystemSolver ABC with find_solution(pressure_constraint, inlet_stream) → Solution
OutletPressureSolver → renamed to CommonSpeedWithPressureControlSolver (speed-based path)
CommonPressureRatioSolver (ratio-based path, no speed search)

FeasibilitySolver split:

RunnerBasedFeasibility — uses runner + stonewall limits (speed-based path)
RatioBasedFeasibility — uses per-stage stonewall limits from chart data directly

Cross-validation guards:

CommonPressureRatioSolver rejects speed-based Compressor units
ProcessSystemRunner rejects PressureRatioCompressor units

Both raise DomainValidationException at construction time.

Tests

test_pressure_ratio_compressor.py — unit tests for the new process unit
test_common_pressure_ratio_solver.py — solver tests with 2-stage systems
Existing solver tests updated for the rename

tj098895 · 2026-04-15T10:27:49Z

Can this be rather solved by:

Not use a shaft (and hence no shaft / speed strategy)
Use a "PressureRatioStrategy" across compressors instead?

Can we then reuse the existing compressor etc and separate solving from simulation better?

But at the same time, this compressor is a "synthetic generic compressor", so might be relevant to keep that as a separate thing anyways somehow? Not sure how this aligns with reality and what is the use of something "unreal"?

olelod · 2026-04-15T10:49:07Z

Can this be rather solved by:

Not use a shaft (and hence no shaft / speed strategy)

Use a "PressureRatioStrategy" across compressors instead?

Can we then reuse the existing compressor etc and separate solving from simulation better?

But at the same time, this compressor is a "synthetic generic compressor", so might be relevant to keep that as a separate thing anyways somehow? Not sure how this aligns with reality and what is the use of something "unreal"?

We of course could. But would not that mean having to deal with shaft: Shaft | None, speed: float | None etc?

The generic chart is fundamentally not a real compressor — it's a synthetic stand-in used when you don't have actual vendor curves. You give it an efficiency and it tells you how much work it does for a given inlet stream and pressure ratio. There's no shaft speed involved because the machine isn't modelled at that level of detail. We don't
take the work performed and the rate put in, to impose a speed (interpolate it in the compressor chart) - and find three different speeds for three generic compressors in a train. We simply don't care about speed at all - only the work performed by the generic compressor.

The typical use case is early-phase studies or concept screening — you want a rough energy estimate for a compression stage before you have a vendor quote or a real performance curve. You're not pretending to know the machine; you're explicitly saying you don't.

tj098895 · 2026-04-15T10:59:12Z

+                    inner_inlet = current_inlet
+
+                outlet_with_recirc = unit.propagate_stream_at_pressure_ratio(
+                    inner_inlet, pressure_ratio=pressure_ratio_per_stage


can pressure ratio be a config that is being set instead, calculated by the compressor pressure ratio solver, in order to always use the propagate_stream method? I see now however, that the compressor is a bit different wrt. behaviour, so probably a good idea to separate it from a "normal compressor process impl"

Maybe PressureRatioCompressor should not inherit from ProcessUnit at all? Make it something separate?

will such a system use ASV/recirc loop and other chokes, or not relevant?

There is of course individual recirculation to surge line (if the rate is too low), but this is just using min/max_rate_given_head on the compressor chart - no need for recirc ProcessUnits.

and upstream/downstream choke is not relevant. multiple streams and pressures should not be relevant for simplified trains.

frodehk · 2026-04-16T08:50:16Z

+            max=max(0.0, max_sm3_per_day - inlet_sm3_per_day),
+        )
+
+    def propagate_stream_at_pressure_ratio(self, inlet_stream: FluidStream, pressure_ratio: float) -> FluidStream:


Suggestion: Have a setter for pressure_ratio, and use propagate_stream() to fulfill the ProcessUnit contract. Other process units are already doing this, e.g. Choke (set_pressure_change).

Alternatively, split the ProcessUnit interface? e.g. RatioBasedUnit

Introduce a ratio-based solving path for compressor trains that operate at a fixed pressure ratio (no shaft speed, the legacy simplified train). The total pressure ratio is distributed equally across stages (geometric mean), with individual anti-surge recirculation computed analytically. New domain entities: - PressureRatioCompressor: ProcessUnit that evaluates at a given pressure ratio via propagate_stream_at_pressure_ratio(). Validates that only generic chart types are used (GENERIC_FROM_INPUT or GENERIC_FROM_DESIGN_POINT). - CommonPressureRatioSolver: distributes equal pressure ratio across all PressureRatioCompressor stages, applies minimum recirculation per stage. Implements ProcessSystemSolver ABC. Exposes get_max_standard_rate() for stonewall-based feasibility. Solver hierarchy refactored: - ProcessSystemSolver ABC: find_solution(pressure_constraint, inlet_stream) - CommonSpeedWithPressureControlSolver (renamed from OutletPressureSolver): speed-based path with choke/ASV pressure control - CommonPressureRatioSolver: ratio-based path, no speed search FeasibilitySolver split into two implementations: - RunnerBasedFeasibility: uses runner + stonewall limits (speed-based path) - RatioBasedFeasibility: uses per-stage stonewall limits from chart data

PressureRatioCompressor no longer inherits ProcessUnit (StreamPropagator). It now implements the new PressureRatioUnit ABC which only contracts the ratio-based evaluation methods. - Add PressureRatioUnit ABC to process_system/ - PressureRatioCompressor: swap base class, remove NotImplementedError propagate_stream() - CommonPressureRatioSolver: isinstance checks on PressureRatioUnit, drop Compressor guard - ProcessSystemRunner: drop PressureRatioCompressor guard (now enforced by type system) - Remove test_propagate_stream_raises_not_implemented (behaviour is gone by design)

…o, CommonSpeedWithPressureControlSolver → OutletPressureSolverSpeed

olelod · 2026-05-03T06:44:47Z

replaced by #1527

olelod force-pushed the feature/pressure-ratio-domain branch 2 times, most recently from 865110e to fb0b0b0 Compare April 15, 2026 08:16

tj098895 reviewed Apr 15, 2026

View reviewed changes

frodehk reviewed Apr 16, 2026

View reviewed changes

Comment thread src/libecalc/domain/process/process_solver/common_pressure_ratio_solver.py Outdated

frodehk reviewed Apr 16, 2026

View reviewed changes

Comment thread src/libecalc/domain/process/process_solver/common_speed_with_pressure_control_solver.py Outdated

olelod force-pushed the feature/pressure-ratio-domain branch from fb0b0b0 to 4573d90 Compare April 17, 2026 11:48

olelod added 3 commits April 22, 2026 09:19

refactor: rename CommonPressureRatioSolver → OutletPressureSolverRati…

1c30aaf

…o, CommonSpeedWithPressureControlSolver → OutletPressureSolverSpeed

olelod force-pushed the feature/pressure-ratio-domain branch from 4573d90 to 1c30aaf Compare April 22, 2026 07:23

olelod mentioned this pull request May 3, 2026

feat: multi-shaft solver for independently-shafted compressor trains #1527

Merged

olelod closed this May 3, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add PressureRatioCompressor and CommonPressureRatioSolver#1483

feat: add PressureRatioCompressor and CommonPressureRatioSolver#1483
olelod wants to merge 3 commits intomainfrom
feature/pressure-ratio-domain

olelod commented Apr 14, 2026

Uh oh!

tj098895 commented Apr 15, 2026 •

edited

Loading

Uh oh!

olelod commented Apr 15, 2026 •

edited

Loading

Uh oh!

tj098895 Apr 15, 2026

Uh oh!

olelod Apr 15, 2026

Uh oh!

tj098895 Apr 15, 2026

Uh oh!

olelod Apr 15, 2026

Uh oh!

olelod Apr 15, 2026

Uh oh!

frodehk Apr 16, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

olelod commented May 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

olelod commented Apr 14, 2026

Uh oh!

tj098895 commented Apr 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

olelod commented Apr 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tj098895 Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

olelod Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

tj098895 Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

olelod Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

olelod Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

frodehk Apr 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

olelod commented May 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

tj098895 commented Apr 15, 2026 •

edited

Loading

olelod commented Apr 15, 2026 •

edited

Loading

frodehk Apr 16, 2026 •

edited

Loading