A professional BIM-to-cost pipeline that extracts data from Revit models (2015–2026), classifies elements with AI, deco…
A professional BIM-to-cost pipeline that extracts data from Revit models (2015–2026), classifies elements with AI, decomposes them into construction works, and generates detailed cost estimates using the open-source DDC CWICR database. Produces HTML reports and Excel exports with full resource breakdown. Who's it for BIM Managers** automating quantity takeoff and cost estimation Cost Engineers** integrating 5D workflows into design pipelines Construction Companies** standardizing estimates from Revit models General Contractors** doing rapid budget checks during design MEP Engineers** pricing mechanical/electrical/plumbing systems Developers** building custom BIM-to-cost integrations What it does Extracts BIM data from Revit model via converter (RvtExporter) Classifies building vs non-building elements using AI Detects project type (Residential/Commercial/Industrial) Generates construction phases and assigns element types Decomposes each BIM type into detailed work items Searches DDC CWICR vector database for matching rates Calculates costs with unit mapping and resource breakdown Validates work completeness and checks for gaps Generates professional HTML report + Excel file How it works ┌─────────────────────────────────────────────────────────────────────────────┐ │ REVIT MODEL (.rvt) │ │ Revit 2015–2026 supported │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 1: CONVERSION │ │ RvtExporter.exe → Excel with BIM element schedules │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 2: DATA LOADING & CLASSIFICATION │ │ • Filter 3D View elements only │ │ • AI analyzes headers → aggregation rules (sum/mean/last) │ │ • AI classifies building vs non-building elements │ │ • Hard exclude: Grids, Levels, Annotations, Views, etc. │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 3: PROJECT ANALYSIS (Stages 0–3) │ │ STAGE 0: Collect filtered BIM data │ │ STAGE 1: AI detects project type │ │ STAGE 2: AI generates construction phases │ │ STAGE 3: AI assigns element types to phases │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 4: WORK DECOMPOSITION (Stage 4) │ │ Loop through each BIM type: │ │ • AI decomposes type into work items │ │ • Example: Window → Demolition, Installation, Sealing, Hardware │ │ • Prepares search queries for pricing │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 5: PRICING & CALCULATION (Stages 5–7) │ │ STAGE 5: Vector search in Qdrant (text-embedding-3-large, 3072 dim) │ │ STAGE 6: Map BIM units → Rate units (m² → 100 m²) │ │ STAGE 7: Calculate costs (Qty × Unit Price) │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 6: VALIDATION & AGGREGATION │ │ STAGE 7.5: AI validates work completeness │ │ STAGE 8: Aggregate costs by phases │ └─────────────────────────────────────────────────────────────────────────────┘ ↓ ┌─────────────────────────────────────────────────────────────────────────────┐ │ BLOCK 7: REPORT GENERATION (Stage 9) │ │ • Professional HTML report with expandable rows │ │ • Excel-compatible XLS file │ │ • Auto-opens in browser │ └─────────────────────────────────────────────────────────────────────────────┘ Pipeline Stages | Stage | Name | Description | |-------|------|-------------| | 0 | Collect | Gather filtered BIM data | | 1 | Project Type | AI detects Residential/Commercial/Industrial | | 2 | Phases | AI generates construction phases | | 3 | Assignment | AI assigns element types to phases | | 4 | Decomposition | AI breaks types into work items | | 5 | Vector Search | Query Qdrant for pricing rates | | 6 | Unit Mapping | Convert BIM units to rate units | | 7 | Calculation | Compute costs (Qty × Price) | | 7.5 | Validation | AI checks completeness, finds gaps | | 8 | Aggregation | Sum costs by phases | | 9 | Reports | Generate HTML + XLS outputs | Prerequisites | Component | Requirement | |-----------|-------------| | n8n | v1.30+ with Execute Command node | | Revit Exporter | RvtExporter.exe (provided separately) | | OpenAI API | For embeddings + LLM tasks | | Qdrant | Vector DB with DDC CWICR collections | | DDC CWICR Data | GitHub | | Windows | For Revit converter execution | Setup 1. Configure File Paths In Setup - Define file paths node: { "path_to_converter": "C:\\path\\to\\RvtExporter.exe", "project_file": "C:\\path\\to\\your_project.rvt", "group_by": "Type Name", "language_code": "DE" } 2. Select Language & Region | Code | Language | City | Currency | |------|----------|------|----------| | AR | Arabic | Dubai | AED | | ZH | Chinese | Shanghai | CNY | | DE | German | Berlin | EUR | | EN | English | Toronto | CAD | | ES | Spanish | Barcelona | EUR | | FR | French | Paris | EUR | | HI | Hindi | Mumbai | INR | | PT | Portuguese | São Paulo | BRL | | RU | Russian | St. Petersburg | RUB | 3. Configure AI Model Connect your preferred LLM in the model nodes: | Provider | Model | Notes | |----------|-------|-------| | OpenAI | GPT-4o | Default, recommended | | Anthropic | Claude Opus 4 | High quality | | Google | Gemini 2.5 Pro | Good for large contexts | | xAI | Grok 4 | Fast inference | | DeepSeek | DeepSeek Chat | Cost-effective | | OpenRouter | Various | Multi-model access | 4. Set Up Qdrant Ensure DDC CWICR collections are loaded: DE_BERLIN_workitems_costs_resources_EMBEDDINGS_3072_DDC_CWICR ENG_TORONTO_workitems_costs_resources_EMBEDDINGS_3072_DDC_CWICR RU_STPETERSBURG_workitems_costs_resources_EMBEDDINGS_3072_DDC_CWICR ... 5. Configure OpenAI Credentials Set up OpenAI API credential for: Embeddings (text-embedding-3-large, 3072 dimensions) LLM calls (if using OpenAI as primary model) Features | Feature | Description | |---------|-------------| | 🏗️ Revit Integration | Direct extraction from .rvt files (2015–2026) | | 🤖 Multi-LLM Support | OpenAI, Claude, Gemini, Grok, DeepSeek | | 🔍 Smart Classification | AI separates building from non-building elements | | 📊 Work Decomposition | Breaks BIM types into detailed work items | | 🎯 Vector Search | Semantic matching via Qdrant + OpenAI embeddings | | 🧮 Unit Mapping | Automatic conversion (m² → 100 m², pcs → sets) | | ✅ AI Validation | Checks for missing works and duplications | | 📈 Phase Aggregation | Costs grouped by construction phases | | 📄 HTML Report | Professional report with quality indicators | | 📑 Excel Export | XLS file with formulas and links | | 🌍 9 Languages | Full localization + regional pricing | Hard Exclude Categories The pipeline automatically excludes non-physical elements: Levels, Grids, Reference Planes Annotations, Dimensions, Text Notes Tags, Views, Sheets, Schedules Legends, Viewports, Section Boxes Scope Boxes, Match Lines Model Groups, Detail Groups Entourage (RPC people, cars, plants) Example Output Input: Residential building Revit model (45 element types) Processing: Project type detected: Residential Multi-Family Phases generated: Foundations → Structure → Envelope → MEP → Finishes Types assigned: 45 types → 5 phases Works decomposed: 45 types → 280 work items Rates found: 245/280 (87.5%) Output Files: project_2024-12-08.html → Professional HTML report project_2024-12-08.xls → Excel with full breakdown HTML Report Features: KPI summary (total cost, items, phases) Expandable phase sections Quality indicators (● green/yellow/red) Resource breakdown per work item Clickable rate codes Responsive design Output Structure 📊 Cost Estimate: Residential Building ├── 📁 Phase 1: Foundations │ ├── Foundation walls — 125.5 m³ — €12,450 │ ├── Concrete footings — 45.2 m³ — €8,340 │ └── Waterproofing — 280 m² — €4,200 ├── 📁 Phase 2: Structure │ ├── Concrete columns — 18 pcs — €9,720 │ ├── Floor slabs — 450 m² — €67,500 │ └── Stairs — 3 flights — €8,100 ├── 📁 Phase 3: Envelope │ ├── Exterior walls — 680 m² — €95,200 │ ├── Windows — 42 pcs — €25,200 │ └── Roof system — 225 m² — €33,750 └── 💰 TOTAL: €485,240 Notes & Tips First run:** Conversion takes 1–3 minutes depending on model size Cached conversion:** Subsequent runs skip conversion if Excel exists Testing mode:** Limit to 10 types for faster debugging Rate accuracy:** Depends on DDC CWICR coverage for your region Custom phases:** AI adapts phases based on project type Missing rates:** Flagged with red indicator in report Extending the Pipeline Add custom rates:** Extend Qdrant collection with your pricing Chain to PM tools:** Connect to OpenProject, Monday, Asana Email reports:** Add email node after report generation Cloud storage:** Upload to Google Drive, OneDrive, S3 Webhook trigger:** Replace manual trigger for API access Categories AI · Data Transformation · Document Ops · Files & Storage Tags bim, revit, cost-estimation, 5d-bim, 4d-bim, qdrant, vector-search, openai, construction, quantity-takeoff, html-report, multilingual Author DataDrivenConstruction.io https://DataDrivenConstruction.io info@datadrivenconstruction.io Consulting & Training We help AEC firms implement: BIM-to-cost automation pipelines 4D/5D integration workflows Custom Revit data extractors AI-powered estimation systems Vector database deployment for construction data Contact us to adapt this pipeline to your Revit templates and regional pricing. Resources DDC CWICR Database:** GitHub Qdrant Documentation:** qdrant.tech/documentation OpenAI Embeddings:** platform.openai.com n8n Execute Command:** docs.n8n.io ⭐ Star us on GitHub! github.com/datadrivenconstruction/DDC-CWICR. An n8n automation workflow template by Artem Boiko.
Artem Boiko
web
free
Others in the same category, ranked by how often they are opened.
