Engineering Deliverables and Handover

NeqSim can compile process and engineering results into a coordinated, machine-readable package. The purpose is to make values, identities, cases, methods, dependencies, findings, and review state consistent across deliverables. A generated package is review-ready evidence, not automatic approval or fitness for construction.

Choose the correct output

Required output	Entry point
Quick Proteus-compatible P&ID or pyDEXPI visualization	`DexpiXmlWriter` or `DexpiDiagramBridge`
Native DEXPI 2.0 Plant or Process exchange	`Dexpi20XmlWriter` or `Dexpi20ProcessModelWriter`
DEXPI plus coordinated cases, registers, calculations, and validation	`EngineeringDeliverableCompiler`
Coordinated multi-area packages and root manifest	`ProcessModelEngineeringSimulator.Result.compile(...)`
Review-required generated P&ID proposal	`PidEngineeringPackageExporter`
Project information handover staging	`Cfihos20HandoverExporter` applied to the canonical engineering graph
Integrity-protected portable model revision	`NeqSimModelPackage` and `ModelPackageValidator`
Fail-closed qualification and readiness backlog	`EngineeringProductionReadinessAssessment` and `EngineeringQualificationPlan`

For DEXPI profile selection and qualification, start with the DEXPI Engineering Guide.

Package layers

Identity and provenance

The canonical EngineeringGraph provides stable identities and relationships for projects, equipment, lines, instruments, calculations, requirements, evidence, documents, and approvals. It also provides the dependency basis for revision diff and impact analysis.

Principal artifacts include:

engineering-model.json;
engineering-connectivity.json;
engineering-calculation-dag.json; and
engineering-revision-diff.json or revision-impact reports when a baseline is supplied.

The compiler also writes neqsim-model-package.json. NeqSimModelPackage records asset/model identity, revision, baseline revision, lifecycle state, qualification state, dependencies, software versions, artifact inventory, and SHA-256 hashes. ModelPackageValidator detects missing or altered files. Integrity is not engineering approval.

Cases and calculations

The package retains controlled case definitions, per-case results, governing values, calculation methods, units, constraints, warnings, and readiness:

engineering-design-case-matrix.json;
design-case-envelope.json;
engineering-calculations.json;
engineering-numerical-health.json; and
engineering-discipline-orchestration.json, engineering-production-readiness.json, engineering-qualification-plan.json, and other discipline calculation artifacts.

See Design Cases and Governing Envelopes for case construction and interpretation.

Registers and datasheets

Coordinated outputs may include:

equipment, line, valve, instrument, I/O, alarm, trip, and relief registers;
equipment and PSV datasheets;
process design basis and utility summary;
materials-selection and preliminary mechanical reports;
shutdown narratives and flare/blowdown reports; and
unresolved engineering actions and evidence registers.

Every calculated field should retain its unit, governing case, method or source calculation, and readiness state. The package also includes an engineering-approval-ledger.json and external-evidence register. These preserve supplied decisions and evidence state; they do not manufacture approval.

DEXPI exchange

The package can contain native DEXPI 2.0 Plant and Process XML plus Proteus and pyDEXPI representations, validation reports, internal structural round-trip evidence, and deterministic diagram layout. These representations share graph identity but serve different receiving tools and information-model purposes.

CFIHOS staging

Cfihos20HandoverExporter maps canonical graph identities and properties through a project-controlled CFIHOS 2.0 Core or Extended RDL mapping. It generates deterministic staging tables, a gap register, file digests, and a fail-closed assessment.

The staging CSVs are not official bulk-loader templates. The Principal still owns contractual completeness, transformation to required exchange templates, named target-system validation, and final information acceptance.

Recommended issue workflow

Freeze inputs. Record project ID, revision, process-model fingerprint, case basis, standards, methods, and controlled evidence.
Run engineering. Require converged cases, a closed design loop, satisfied constraints, and explicit findings.
Compile once. Generate all coordinated artifacts from the same designed process copy and canonical graph.
Validate the package. Check schemas, identities, references, units, fingerprints, manifests, and cross-artifact consistency.
Complete external qualification. Attach vendor, independent benchmark, named-CAE, safety-lifecycle, and authority evidence required by the issue purpose.
Review and approve. Record accountable discipline decisions against stable graph subjects; do not overwrite previous decisions.
Transform and load. Map to Principal requirements, load into the named target system, and retain receipts, rejections, and reconciliation evidence.
Issue the controlled revision. Preserve package manifest and hashes with the released files.

Validate before issue

Gate	Minimum check
Structural	Every JSON artifact matches its supported schema version and URI
Referential	Graph and register references resolve; controlled tags are unique
Units	Values use the controlled unit vocabulary and retain units across artifacts
Case consistency	Governing case IDs resolve to successful case results
Calculation consistency	Results point to methods, inputs, subjects, and dependencies
DEXPI	Selected profiles validate; internal references and identities survive structural round trip
Manifest	Artifact inventory, revision, graph fingerprint, and file digests agree
Model package	`neqsim-model-package.json` inventories every governed artifact and all hashes verify
Production readiness	Failed gates and the exact qualification backlog remain visible; no absent gate is inferred as passed
Multi-area package	Every area package, shared dependency, executed/reused-area decision, and root manifest validates
Review state	Warnings, blockers, actions, approvals, and revalidation requirements remain visible

Readers should reject an unknown major schema version instead of silently treating it as the current contract.

Approval and readiness states

State	What it establishes
Calculated	A declared method produced a result
Structurally valid	Schemas, identities, references, units, and manifests are coherent
Internally qualified	Required internal benchmarks and structural round trips passed
Controlled-pilot qualified	The configured facility slice passed its nine explicit topology, case, design, safety, standards, and evidence gates
Qualified for FEED support	Every configured production-readiness gate passed for the declared support purpose
Review-ready	Evidence, assumptions, findings, and actions are organized for accountable review
Approved	An authorized role accepted the controlled subject and evidence for a declared purpose
Accepted into target system	The Principal validated the transformed data in the named system
Fit for construction	Detailed design, vendor, independent, safety, regulatory, and construction gates are closed

These states are not interchangeable. Successful compilation establishes neither external qualification nor approval.

Manage revisions

When a model, case, method, design variable, or evidence item changes:

create or derive a deterministic ModelChangeEvent for the new governed revision;
compare the new canonical graph with the controlled baseline;
use GeneralizedImpactAnalyzer to identify added, removed, modified, and downstream-impacted objects;
mark dependent calculations, documents, validations, and approvals for rework;
retain unchanged evidence only when its dependencies and applicability remain valid; and
issue the impact report and updated manifest with the new revision.

For a multi-area ProcessModel, ProcessModelEngineeringSimulator.runIncremental(...) reruns changed and connected areas and records which unaffected baseline areas were reused.

An approval affected by revision impact becomes REVALIDATION_REQUIRED; the earlier decision remains in the ledger instead of being silently replaced.

DEXPI and CFIHOS handover boundary

DEXPI provides	CFIHOS handover adds
Process/plant information-model exchange	Project information-requirement and RDL mapping context
Equipment, piping, ports, streams, instruments, safeguards, and diagrams	Controlled tag/equipment classes, property/UOM identifiers, document types, and relationships
Schema and supported-profile conformance evidence	Mapping approval, gap register, staging manifest, and target-system transformation boundary

Use both when the project requires machine-readable engineering exchange and controlled information handover. Neither replaces discipline review, document control, vendor data, target-system acceptance, or construction authority.

Consumer checklist

Confirm project, area, revision, and package fingerprint.
Confirm supported schema major versions before reading values.
Check validation reports and unresolved actions before using registers or datasheets.
Resolve every governing case and calculation reference.
Confirm the exact DEXPI profile and receiving-tool qualification evidence.
Confirm CFIHOS RDL delivery, mapping revision, approval, and target transformation when applicable.
Do not infer approval from a successful calculation, export, schema validation, or file load.
Retain the package manifest and hashes with any downstream transformation.

Common anti-patterns

Issuing a drawing separately from the calculations and registers that produced its values.
Copying values between artifacts without preserving units, governing cases, or method identity.
Treating DEXPI schema validation as named-tool or engineering approval.
Treating neutral CFIHOS staging CSVs as the contracted target-system load format.
Deleting warnings or unresolved actions to make the package appear complete.
Reusing approvals after a dependency changed without impact analysis.
Declaring fitnessForConstruction=true from preliminary or screening calculations.