CAD-to-GIS & BIM-to-GIS Automation

What problem does this solve?

Every project handover arrives as drawings and models, and every one of them is different. One consultant delivers DWGs on a site grid with the origin shifted, another delivers them in a projected coordinate system but with the utilities on layers named after the person who drew them. A Revit model has the information in shared parameters, an IFC export has it in property sets, and the room numbers in both disagree with the ones in the facilities system. Somebody in the GIS team then spends weeks per project turning this into feature classes.

Because it is manual, it is inconsistent. The same layer is interpreted one way in March and another way in September; polygons that were not closed in CAD become gaps in the GIS; text labels that carried the diameter or the material are lost because there was no field to put them in. Six months later nobody can say whether the geodatabase reflects the as-built or the design, and the asset register that depends on it inherits the doubt.

The deeper issue is that CAD and BIM describe geometry for construction while GIS describes features for operation. A wall in Revit is not an asset; a pipe drawn as two parallel lines in CAD is not a network edge. Any automation that ignores this produces a technically successful import of content that no operational workflow can use — which is why so many BIM-to-GIS efforts stall after the first pilot building.

How the solution works

Treat conversion as a pipeline with a written specification, not as an import. The pipeline georeferences the source using control points or a known transformation, maps CAD layers, blocks and attributes — or Revit categories and parameters, or IFC entities and property sets — onto target feature classes, fields and domains, validates geometry and attributes against the geodatabase rules, and loads the result into a staging area before anything touches production. Records that fail validation go to a rejects queue with the drawing name and object handle so they can be fixed at source rather than quietly dropped.

The tooling depends on the source. ArcGIS Pro reads DWG/DXF directly and supports Revit files through a BIM file workspace, which covers a large share of building work with no extra licence. ArcPy and Python toolboxes automate the repetitive parts and schedule them. Where sources are messy, IFC-based, or need substantial attribute logic, ArcGIS Data Interoperability or FME provides the transformation engine. Swedish Technology builds and hands over the pipeline, the mapping specification and the validation reports, so the next handover is a scheduled run rather than a project.

1. 1
   Input Handover package: DWG/DXF drawings, Revit models, IFC exports, survey control points, and the project's coordinate and layering standard if one exists.
2. 2
   Capture Files land in a watched folder or a document system. The pipeline registers each package, records its checksum and source project, and extracts the drawing or model structure for inspection.
3. 3
   Processing Georeferencing to the corporate projected CRS; geometry repair (close polygons, remove duplicates, snap within tolerance, dissolve construction linework); mapping layers, blocks, categories and property sets to feature classes, fields and coded domains; deriving centrelines or footprints where the source is drawn as construction geometry.
4. 4
   Integration Validated features are written into a staging geodatabase, compared against existing production content, and loaded through a versioned edit or an append with editor tracking. Asset keys from SAP, Maximo or the asset register are attached where a match exists.
5. 5
   Action Reviewers see the as-built comparison — added, changed, removed, unmatched — and approve or reject the load. Rejected objects go back to the consultant with drawing name, layer and object handle.
6. 6
   Reporting Each run produces a validation and comparison report: counts by feature class, rejection reasons, georeferencing residuals, and a summary of what changed in production. Reports are archived with the handover package as the audit trail.

Reference architecture

Five parts. The mapping specification and the rejects queue matter more than the transformation engine — those are what make the pipeline survivable when a new consultant delivers something unexpected.

Deployment options: Everything runs inside the customer network: a Windows server with ArcGIS Pro or ArcGIS Server for the scheduled runs, the staging and production geodatabases, and file storage for handover packages. No content needs to leave the environment, which suits classified site drawings and defence-adjacent projects.

Key capabilities

Integrations

Converted data is only useful once it is connected to the systems that operate the asset.

Industry use cases

Implementation approach

1. 1
   Sample assessment (1–2 weeks) Take three to five representative handovers and measure what is actually convertible: coordinate handling, layer discipline, attribute completeness, geometry quality. This sets realistic expectations before any build.
2. 2
   Target model confirmation Confirm the destination feature classes, fields, domains, topology and attribute rules. Where the target model does not exist yet, designing it is a prerequisite step.
3. 3
   Mapping specification Write the layer-to-feature-class, block-to-field and value-to-domain tables with the GIS data owner, including defaults, units and what to do when a value is missing.
4. 4
   Pipeline build Implement georeferencing, geometry repair, mapping, validation and staging load; choose ArcPy, Data Interoperability or FME per source type; build the rejects queue and reporting.
5. 5
   As-built comparison and review workflow Implement the comparison against production content and the reviewer approval step, including how a rejected handover is returned to the submitter.
6. 6
   Pilot on live handovers (4–6 weeks) Run real submissions through the pipeline in parallel with the manual process; measure automation rate, rejection reasons and reviewer time.
7. 7
   Handover standard and checker Publish the delivery standard for consultants and give them a checking tool, then measure the drop in rejections over the following submissions.
8. 8
   Hand-over and support Source code, mapping specification, runbook, scheduling configuration and training for the GIS team; support agreement covering ArcGIS Pro and FME version upgrades.

Security & deployment

Handover packages often contain sensitive site information — utility routes, security systems, floor plans of government buildings — so the pipeline is designed to keep them inside the customer network. Intake storage, transformation server, staging and production geodatabases all sit on-premise or in a UAE-region private cloud, and cloud-based conversion services are excluded where classification requires it; we have run the pipeline in fully air-gapped environments with offline installers. Access to intake folders and staging data follows the same directory groups as the production geodatabase, service accounts are least-privilege, and every package is recorded with checksum, submitter and reviewer so a change in production can always be traced back to the drawing that caused it.

Limitations & prerequisites

• Automation cannot invent information that is not in the drawing. If diameter, material or asset ID exist only as free text or not at all, the pipeline can flag the gap but not fill it.
• Poorly disciplined CAD — exploded blocks, everything on layer 0, unclosed polygons, drawings on an unknown local grid — reduces the automatable share sharply; a published delivery standard is what raises it, not more code.
• BIM models are built for construction, not operation. Deciding which Revit or IFC objects become GIS assets is a data-modelling exercise that must be done by the asset owner, and it cannot be skipped.
• IFC support varies by exporter and by schema version; the same building exported from two tools can produce different property sets, so mappings are usually per-source rather than universal.
• ArcGIS Data Interoperability and FME involve additional licensing; where budget rules them out, ArcPy plus ArcGIS Pro covers CAD and Revit but with more custom code for IFC.
• Expect a manual residue. A realistic target is a high automated share with a reviewed exception list, not a fully hands-off process — and the review step is what keeps the geodatabase trustworthy.
• Georeferencing accuracy is limited by the control information supplied; without control points or a documented transformation, alignment is an assumption and should be reported as one.

Tooling options for CAD/BIM to GIS conversion

Choose by source complexity and by what already exists in your licence portfolio.

Most organisations combine them: ArcPy for the scheduled CAD path and the reporting, and Data Interoperability or FME for the IFC and awkward-source cases where writing the logic in code would be slower than building a workbench.

FAQ

Related topics & solutions

• ArcGIS Python Toolbox & ArcPy Automation
• ArcGIS Pro Add-in Development
• ArcGIS – SAP Integration (PM/EAM, S/4HANA)
• ArcGIS – IBM Maximo Integration
• ArcGIS Experience Builder Custom Widgets
• RASM ÔÇô Digital Twin
• Facility Management
• Document Management & Correspondence System