System

New Tyne Crossing Transport Appraisal System, second subsystem decomposition session. The {{entity:Traffic Microsimulation Subsystem}} was decomposed in sessions 263-264. This session tackles the {{entity:Transport Demand Modelling Subsystem}} — the foundational layer whose trip matrices feed microsimulation, economic appraisal, and environmental assessment. Project now has 55 requirements across 6 documents, with 2 of 7 subsystems decomposed.

Decomposition

The {{entity:Transport Demand Modelling Subsystem}} decomposes into seven components reflecting standard UK WebTAG 4-stage modelling practice with variable demand:

1. {{entity:Trip End Forecasting Module}} {{hex:40A53B58}} — NTEM/TEMPro constrained trip generation with local authority growth adjustments for three forecast horizons
2. {{entity:Trip Distribution Engine}} {{hex:50B73308}} — doubly-constrained gravity model with purpose-specific deterrence functions calibrated to NTS North East data
3. {{entity:Mode-Destination Choice Model}} {{hex:40B43208}} — incremental hierarchical logit implementing TAG Unit M2 variable demand, calibrated to observed cross-river mode shares
4. {{entity:Highway Assignment Engine}} {{hex:41F73308}} — Wardrop user equilibrium on a 3,000+ link network with BPR curves calibrated to A19, A1058, and Tyne Tunnel approach journey times
5. {{entity:Public Transport Skim Generator}} {{hex:40A53308}} — generalised cost skims covering Metro, bus, and rail using Nexus/BODS timetable feeds
6. {{entity:Matrix Estimation Processor}} {{hex:51B73308}} — ME2 entropy-maximising estimation against 120+ count sites with trip length distribution preservation
7. {{entity:Demand-Supply Convergence Controller}} {{hex:40B73A08}} — MSA outer loop managing demand-assignment iteration to simultaneous convergence

The key architectural choice ({{sys:ARC-DECISIONS-002}}) was separating matrix estimation from distribution and isolating the convergence controller from assignment. ME2 must be independently auditable for DfT review, and the convergence controller needs to experiment with damping strategies because the three-way route competition between Tyne Bridge, Tyne Tunnel, and the proposed new crossing creates oscillation-prone assignment dynamics.

flowchart TB
  DAQ["Data Acquisition Subsystem"]
  TEF["Trip End Forecasting"]
  TDE["Trip Distribution Engine"]
  MEP["Matrix Estimation Processor"]
  HAE["Highway Assignment Engine"]
  PTSG["PT Skim Generator"]
  MDC["Mode-Destination Choice"]
  DSCC["Convergence Controller"]
  TMS["Traffic Microsimulation"]
  EAE["Economic Appraisal Engine"]

  DAQ -->|NTEM, census, planning data| TEF
  DAQ -->|Traffic count data| MEP
  TEF -->|Trip ends by zone| TDE
  TDE -->|Prior OD matrices| MEP
  MEP -->|Calibrated matrices| HAE
  HAE -->|Highway cost skims| MDC
  PTSG -->|PT cost skims| MDC
  MDC -->|Updated demand matrices| DSCC
  DSCC -->|Re-assignment trigger| HAE
  HAE -->|Link flows and cordon matrices| TMS
  MDC -->|Demand responses and user benefits| EAE

The diagram shows the iterative feedback loop at the heart of the demand model: assignment produces cost skims, the choice model produces updated demand, and the convergence controller manages the cycle until equilibrium. External interfaces (amber) connect to the microsimulation and economic appraisal subsystems.

Analysis

The {{entity:Highway Assignment Engine}} {{hex:41F73308}} shares its hex pattern with the {{entity:Simulation Execution Manager}} from microsimulation — both are active computational engines that process data iteratively to convergence. Cross-domain analogs included the warehouse {{entity:Put-Away Assignment Engine}} and naval {{entity:Track Management Subsystem}}, reflecting the common pattern of optimisation-under-constraints.

The Tyne screenline validation requirement ({{sub:SUB-REQS-014}}) is arguably the single most critical calibration check in the entire appraisal system. If the model cannot reproduce existing cross-river demand splits between the five current Tyne crossings, no amount of downstream microsimulation or economic modelling will produce credible results for a new crossing.

Lint produced 2 high-severity ontological mismatches (transport appraisal system and DMRB LA 104 flagged as non-physical despite requirements referencing physical constraints) — both acknowledged as false positives since the system is software and DMRB is a standard. 2 previously acknowledged lint findings unchanged.

Requirements

Created 8 subsystem requirements ({{sub:SUB-REQS-007}} to {{sub:SUB-REQS-014}}), 6 interface requirements ({{ifc:IFC-DEFS-004}} to {{ifc:IFC-DEFS-009}}), and 6 verification entries (VER-METHODS-004 to VER-METHODS-009). All IFC requirements have corresponding VER entries — 100% interface verification coverage for this subsystem.

Key requirements by criticality: {{sub:SUB-REQS-014}} (Tyne screenline validation, GEH < 5.0 per crossing), {{sub:SUB-REQS-012}} (simultaneous demand-supply convergence < 0.1%), and {{ifc:IFC-DEFS-005}} (separated time/distance/toll cost skim components for income-disaggregated values of time). Two cross-subsystem interfaces defined: {{ifc:IFC-DEFS-008}} (15-minute cordon matrices to microsimulation) and {{ifc:IFC-DEFS-009}} (demand responses to economic appraisal with rule-of-half data).

All subsystem requirements trace to parent SYS requirements. All IFC requirements trace to both SYS parents and VER entries. 48 trace links total across the project.

Next

Five subsystems remain undecomposed: Economic Appraisal Engine, Environmental Assessment Subsystem, Geospatial Analysis Platform, Data Acquisition and Management Subsystem, and Appraisal Reporting Subsystem. The Economic Appraisal Engine should be next — it consumes outputs from both demand modelling and microsimulation, and its WebTAG BCR calculations are the primary decision metric for the scheme. The Environmental Assessment Subsystem is also high priority given DMRB LA 104 requirements already in STK/SYS.