AIDC·AI.IO
AutoToolContact

ENGINE ARCHITECTURE

Five layers, one model, every view a consumer.

Schema → sizing → layout → validation → evidence. Each layer reads the engine output from the layer above and feeds the layer below. There is no separate spreadsheet model sitting beside the engine.

  1. L1 · SCHEMAPROFILE-AWARE

    Capability-axis schema & catalog

    One generic schema describes the design. Future-vendor and site-specific behavior plugs in as optional profiles rather than as hard-coded modes.

    Inputs

    • Load basis
    • MVA basis
    • PUE basis
    • Rack profiles
    • EPOD / RMU ring groups
    • Generator basis
    • Cooling topology

    Outputs

    • Profile-aware model
    • Vendor RFI placeholders
    • Validation source pointers
  2. L2 · SIZINGPROFILE-AWARE

    Sizing & electrical engine

    The single engine feeds every consumer — totals, panels, Sankey, SLD, exports, validators, and the configurator UI all read the same output. No reconciliation between views.

    Inputs

    • Capability model
    • Hall load profile
    • MV topology constraints

    Outputs

    • Site totals
    • Hall panels
    • Sankey
    • Single-line view
    • Excel / CSV exports
    • Validator feed
  3. L3 · LAYOUTPROFILE-AWARE

    Layout engine — Hall · Tech · Plant · EX-Plant

    Layout is generated against the same engine model. EPOD blocks, RMU rooms, UPS, generator yards, and substations are placed against measured site geometry, not generic templates.

    Inputs

    • Engine model
    • Site geometry
    • Vendor footprint catalog

    Outputs

    • Rack & EPOD placement
    • Mech EPOD
    • UPS / battery rooms
    • Generator yards
    • Substation footprint
  4. L4 · VALIDATIONPROFILE-AWARE

    Validator rule families

    Electrical, layout, cooling, safety, and data-quality rules fire on every change. Uncertain vendor data becomes an RFI until confirmed, so the design's evidentiary basis stays traceable.

    Inputs

    • Engine model + layout
    • Vendor data / RFI ledger

    Outputs

    • Findings
    • Open RFIs
    • Blocking issues
  5. L5 · EVIDENCEPROFILE-AWARE

    External validation / export ledger

    External validators are the audience. The engine produces structured handoff bundles intended to feed them, and we surface them as evidence workflows rather than claiming finished integrations.

    Inputs

    • Engine output + RFI ledger

    Outputs

    • pandapower bundle
    • pypowsybl / SLD inputs
    • EnergyPlus / OpenStudio cases
    • OpenFOAM domain
    • Ecodial / SIMARIS / DOC / xSpider records

NEXT

Look at engine output, then talk basis.

Open private demoSchedule a basis review