Internal Service Layer
These modules are contributor implementation owners. Process-engineer code
uses OpenPinch.PinchProblem or OpenPinch.PinchWorkspace.
The service layer is the boundary between validated input data and the
prepared/solved in-memory model. It is the right integration surface when you
want more control than PinchProblem
provides but do not want to invoke individual low-level algorithms directly.
Layering
The service stack is designed in three steps:
validate or receive typed request data
prepare the inputs into a
Zonehierarchydispatch direct, indirect, HPR, exergy, cogeneration, or area/cost targeting
Use Cases
Use the service layer when you need to:
develop repository applications that need more control than the supported request/response boundary
prepare a zone hierarchy once and run multiple advanced studies against it
inspect the prepared model before solving
mutate a live prepared model with process components before rerunning targets
bypass file handling entirely and work with typed inputs
apply exergy or cogeneration as post-processing on already solved targets
Main Service Surface
Deterministic engineering analyses and specialist calculations.
Preparation Entry Point
The preparation stage is the key boundary between external inputs and the
internal model. It validates configuration choices, builds the zone tree,
applies dt_cont multipliers, instantiates process and utility streams, and
produces the Zone object consumed by the solver stack.
Period-valued inputs remain period-aware after preparation, but period selection does
not happen inside prepare_problem(...). Instead, the selected period is
applied later through the targeting-service args dictionaries or the higher
level problem.target.*(..., period_id=...) wrappers.
- OpenPinch.application._problem.input.construction.prepare_problem(streams=None, utilities=None, options=None, project_name='Site', zone_tree=None)[source]
Build the top-level zone hierarchy for analysis.
- Parameters:
streams (List[StreamSchema] | None)
utilities (List[UtilitySchema] | None)
options (Dict[str, Any] | None)
project_name (str)
zone_tree (ZoneTreeSchema)
- Return type:
Heat Exchanger Network Synthesis Entry
Heat exchanger network synthesis is problem-rooted. User code should normally
enter through PinchProblem.design:
problem.design.enhanced_heat_exchanger_network(quality_tier=2)
problem.design.open_hens()
problem.design.heat_exchanger_network()
problem.design.multiperiod_heat_exchanger_network()
problem.design.network_evolution((existing_network,))
The internal service entry point owns method dispatch and final result caching. It dispatches to the same direct services exposed by the design accessor. Each named method selects one workflow without a method-name string.
Service entry and dispatch for heat exchanger network synthesis.
- OpenPinch.analysis.heat_exchanger_networks.service.heat_exchanger_network_evolution_method_service(problem, *, initial_networks=None, options=None, workspace_variant=None, executor=None)[source]
Run only seeded network evolution and update the problem cache.
- Parameters:
problem (PinchProblem)
initial_networks (HeatExchangerNetwork | Sequence[HeatExchangerNetwork] | None)
options (dict[str, Any] | None)
workspace_variant (str | None)
executor (SynthesisExecutor | None)
- Return type:
HeatExchangerNetworkSynthesisResult
- OpenPinch.analysis.heat_exchanger_networks.service.heat_exchanger_network_open_hens_method_service(problem, *, options=None, workspace_variant=None, executor=None)[source]
Run the original tier-1 OpenHENS PDM -> TDM -> EVM sequence.
- Parameters:
problem (PinchProblem)
options (dict[str, Any] | None)
workspace_variant (str | None)
executor (SynthesisExecutor | None)
- Return type:
HeatExchangerNetworkSynthesisResult
- OpenPinch.analysis.heat_exchanger_networks.service.heat_exchanger_network_pinch_design_method_service(problem, *, options=None, workspace_variant=None, executor=None)[source]
Run only the pinch design method and update the problem cache.
- Parameters:
problem (PinchProblem)
options (dict[str, Any] | None)
workspace_variant (str | None)
executor (SynthesisExecutor | None)
- Return type:
HeatExchangerNetworkSynthesisResult
- OpenPinch.analysis.heat_exchanger_networks.service.heat_exchanger_network_synthesis_service(problem, *, method=None, initial_networks=None, options=None, workspace_variant=None, executor=None)[source]
Dispatch one HEN design method and update the problem cache.
- Parameters:
problem (PinchProblem)
method (HeatExchangerNetworkDesignMethod | str | None)
initial_networks (HeatExchangerNetwork | Sequence[HeatExchangerNetwork] | None)
options (dict[str, Any] | None)
workspace_variant (str | None)
executor (SynthesisExecutor | None)
- Return type:
HeatExchangerNetworkSynthesisResult
- OpenPinch.analysis.heat_exchanger_networks.service.heat_exchanger_network_thermal_derivative_method_service(problem, *, initial_networks=None, options=None, workspace_variant=None, executor=None)[source]
Run only seeded TDM and update the problem cache.
- Parameters:
problem (PinchProblem)
initial_networks (HeatExchangerNetwork | Sequence[HeatExchangerNetwork] | None)
options (dict[str, Any] | None)
workspace_variant (str | None)
executor (SynthesisExecutor | None)
- Return type:
HeatExchangerNetworkSynthesisResult
The HEN synthesis package is intentionally method-oriented:
targetingcontains method-specific orchestration.executioncontains settings, task builders, executor contracts, and fallback policy.resultscontains result assembly and seed lookup.reportingcontains ranking, verification, and export helpers.solvercontains optional-dependency checks, array adapters, backend calls, and network extraction.modelscontains the equation/unit model layer for pinch design and stagewise models.
Network Grid Diagrams
The presentation owner constructs a grid diagram from the selected heat exchanger network:
from OpenPinch.presentation.network_grid.service import build_grid_diagram
design = problem.results.design
period_id = design.network.period_ids[0]
diagram = build_grid_diagram(design.network, period_id=period_id)
The standalone service remains available for batch rendering one or more
HeatExchangerNetwork
objects, for example when displaying several ranked candidates.
Service entrypoint for heat exchanger network grid diagrams.
- OpenPinch.presentation.network_grid.service.build_grid_diagram(networks, *, index=None, period_id=None, temperature_scaled=False)[source]
Return OpenHENS-style grid diagrams for one or more networks.
- Parameters:
networks (HeatExchangerNetwork | Sequence[HeatExchangerNetwork])
index (int | None)
period_id (str | None)
temperature_scaled (bool)
- Return type:
HeatExchangerNetworkGridDiagram | tuple[HeatExchangerNetworkGridDiagram, …]
Network Controllability
Solved heat exchanger networks can also be screened for steady-state controllability. The service treats process-stream outlet temperatures as controlled outputs and practical bypass or utility-flow adjustments as manipulated variables, then scores the resulting duty-normalised interaction matrix.
For multiperiod networks, pass period_id to both diagram and
controllability services. OpenPinch does not silently select period zero.
design = problem.results.design
period_id = design.network.period_ids[0]
assessment = design.network.quantify_controllability(period_id=period_id)
assessment.score
assessment.components.rank
The score is a screening metric rather than a dynamic closed-loop simulation. It is intended for comparing candidate HEN topologies and identifying networks with weak actuator coverage, poor pairing, low thermal margin, or insufficient control redundancy.
Steady-state controllability analysis for heat exchanger networks.
- OpenPinch.analysis.heat_exchanger_networks.controllability.quantify_heat_exchanger_network_controllability(network, *, period_id=None, active_only=True, include_utility_actuators=True, minimum_interaction=1e-09, minimum_approach_temperature=5.0, desired_redundancy=2, rank_tolerance=None, condition_warning_threshold=25.0)[source]
Return a 0-1 controllability assessment for a solved HeatExchangerNetworkLabel.
The service builds a steady-state interaction matrix from available network data. Rows are process-stream outlet temperatures, columns are practical manipulated variables: recovery bypass fractions and utility flow rates. Entries are duty-normalised thermal authority values, which provide a deterministic controllability proxy when no dynamic HEN model is available.
- Parameters:
network (HeatExchangerNetwork)
period_id (str | None)
active_only (bool)
include_utility_actuators (bool)
minimum_interaction (float)
minimum_approach_temperature (float)
desired_redundancy (int)
rank_tolerance (float | None)
condition_warning_threshold (float)
- Return type:
HeatExchangerNetworkControllabilityResult
Typical Preparation and Solve Pattern
from OpenPinch.contracts.input import TargetInput
from OpenPinch.analysis import (
data_preprocessing_service,
direct_heat_integration_service,
indirect_heat_integration_service,
)
source_data = {"streams": [...], "utilities": [...]}
input_data = TargetInput.model_validate(source_data)
zone = data_preprocessing_service(input_data, project_name="Example")
direct_heat_integration_service(zone, {"period_id": "peak"})
indirect_heat_integration_service(zone, {"period_id": "peak"})
Each targeting service mutates the prepared zone in place, records the requested period metadata on the zone, and adds or refreshes the corresponding target model.
The exergy service follows a slightly different contract from the base thermal targeting services: it enriches an already existing compatible target for the requested period instead of re-solving direct or indirect targeting internally.
Application Facade
The supported orchestration facade is documented in PinchProblem and PinchWorkspace.
Choosing Between Interfaces
Use
OpenPinch.PinchProblemfor supported application integration.Use
problem.components.add_process_mvr(...)when the study needs direct gas/vapour MVR stream replacement before ordinary target reruns.Use concrete analysis owners for repository development or advanced research.
The service layer is also the best place to look when you are trying to understand how the narrative workflow maps onto the actual analysis pipeline.