4D/5D Analysis — Construction Schedule and Costing from the Federated Model DB

Foundation: Enterprise · LTU A-House · DATA_MODEL

What was once PRIMAVERA/ORACLE's domain, is now ours.

What We Have

The Federated Model DB (_extracted.db) drives both 4D scheduling and 5D costing in 2 seconds from a single SQLite file. All IFC disciplines live in this one DB. No IFC file open, no geometry iterator, no RAM spike.

Dimension	Script	Input	Output
4D Schedule	scripts/schedule_generator.py	elements_meta + rel_aggregates	construction_schedule table + Excel
5D Costing	scripts/simple_qto_extract.py	elements_meta + elements_rtree	simple_qto table + Excel

Proven Results (LTU A-House — 125,997 elements, 8 disciplines)

5D Costing — Executive Summary with Material + Labour + Equipment per discipline. Grand Total: RM 67M (CIDB 2024 standard rates). 133 QTO line items. Generated in ~2 seconds.

5D Costing — RM 67M across 8 disciplines. Cost Breakdown pie + Cost Components bar.

4D Schedule — 224 construction tasks, precedence-linked by phase. Substructure → Superstructure → MEP Rough-in → Architecture → MEP Final → Finishes.

4D Schedule — 224 tasks. Task Distribution by Phase + Task Count by Discipline.

4D Dashboard — S-Curve, resource workload, milestones — all from one query on the DB.

BIM 5D Analytics Dashboard — Phase Duration, Resource Workload, S-Curve, Milestones.

Current Limitation: Unknown Phase

64% of LTU elements (80,788) are generic IFC2x3 classes (IfcFlowSegment, IfcFlowFitting, IfcFlowController, etc.) not mapped in CONSTRUCTION_SEQUENCE_RULES. These land as "Unknown" phase: 8,703 days / 17,407 man-days — distorting the S-Curve and dashboards. Fix: map all IFC classes in scripts/schedule_generator.py. See prompts/5D_nD_backend_readiness.md for the full class list.

---

Industry Value — What QS Firms Pay For

No commercial tool produces discipline-level 4D+5D analytics from a Federated Model DB in 2 seconds at zero licence cost. Here is what the industry wants and where we stand:

Tier 1 — We Have It Now

Capability	Industry tool	Cost	Our pipeline
IFC → priced BOQ	CostX, Cubicost	$10-50K/seat/yr	simple_qto_extract.py — 2 sec
Discipline-level costing	Manual in Excel	Weeks of QS time	8 sub-disciplines auto-tagged
Construction schedule from BIM	Synchro, Navisworks	$5-15K/seat/yr	schedule_generator.py — 2 sec
S-Curve generation	Primavera, MS Project	$3-10K/seat/yr	Built into Excel export
Phase precedence	Manual CPM network	Days of planner time	Auto from IFC class rules
Multi-discipline MEP breakdown	Not available	—	VENT/HVAC/HEAT/PLB/SAN via --disc-map

Tier 2 — One Query Away

These capabilities need only a JOIN or a new rate table on the existing DB:

Capability	What it needs	Who pays for it
Cash flow forecast	S-curve × monthly cost = spend profile	Banks, project financiers
Tender comparison	Same QTO, swap rate table per bidder	Main contractors, QS firms
Carbon costing (6D)	material_name → embodied carbon factor	ESG compliance, green building certification
Lifecycle costing	ifc_class + material_name → 30-year replacement rates	Asset owners, FM companies
Change order impact	Re-run on modified DB, diff vs baseline	Claims consultants, project managers

Tier 3 — Needs Additional Data Input

Capability	What it needs	Who pays for it
Earned Value (EVM)	Actual progress % per element (site photos / IoT / manual)	PMOs, project controls
Monte Carlo cost risk	Rate variance distributions per element type	Risk consultants, insurance
Claim substantiation	Time-linked cost evidence + GUID traceability	Dispute resolution, arbitration
Resource levelling	Crew availability constraints + calendar	Construction planners

The Holy Grail

A bankable project finance model generated in 2 seconds from the Federated Model DB.

Banks release progress payments based on: 1. What was planned (4D schedule) — we have it 2. What it should cost (5D BOQ) — we have it 3. What was actually done (earned value) — needs progress input 4. Cash flow projection — one JOIN away

Items 1+2+4 from the extracted DB = 90% of the way to a bankable model. Item 3 is the real-world input that no software can auto-generate, but the DB has the GUID-level structure to receive it.

---

Architecture

IFC files ──→ extractIFCtoDB.py ──→ _extracted.db
                                        │
                    ┌───────────────────┼───────────────────┐
                    │                   │                   │
              4D Schedule         5D Costing          Bonsai Viewer
              (schedule_         (simple_qto_        (bbox preview +
               generator.py)     extract.py)          full mesh)
                    │                   │                   │
              Excel + Gantt      Excel + Charts       3D viewport
              S-Curve            BOQ + Rates          125K elements
              Milestones         Discipline cost      smooth nav

The extracted DB is the single source of truth. 4D and 5D are queries on it. The viewer renders it. All from the same SQLite file.

---

Next Steps

1. Fix Unknown phase — map all IFC classes in schedule_generator.py (zero Unknown)
2. Editable Excel — Java backend generates .xlsx with formulas (=qty × unit_rate) so QS can adjust rates and totals recalculate. Apache POI.
3. Cash flow forecast — JOIN S-curve with monthly cost buckets
4. REST endpoints — /api/boq/{building}/excel, /api/schedule/{building}/excel
5. Storey normalisation — LTU's 19 storey names → unified construction sequence

See prompts/5D_nD_backend_readiness.md for the full task spec.