Introduction

Foreword

Because d4SU is not yet a production environment for specific identified administrative processes, the presented features are actually illustrations of capabilities rather that a set of product features.

Features specific to the platform rely on Python tools such as IfcOpenShell-python, pandas DataFrames and Series.

However, other features rely explicitely on standard SQL and other constructs that are programming language agnostic. The code is in Python but could as well be in Typescript, Rust, Java, C#, ...

Unless explicitly specified only IFC2X3, IFC4X* are supported. IFC5 is still under development.

Key Functional Features

The following schema shows key expected features/capabilities

block-beta
    columns 5
    blockArrowId1<["</br>1. Import a model</br>from a BIM/IFC file</br>&nbsp;"]>(right)
    blockArrowId2<["2. Filter the model and</br> keep only essential</br> elements for the </br>use case at hand"]>(right)
    space:3
    space:1
    blockArrowId3<["&nbsp;</br>3. Store the ifcJSON in</br>the Database</br>&nbsp;"]>(right)
    blockArrowId4<["4. Extract the data</br>for target</br> Spatial Unit(s) /</br>Spatial Zone(s)"]>(right)
    blockArrowId5<["&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;"]>(down)
    space:2
    blockArrowId12<["</br>9. Store an ifcJSON<br/> for the bespoke model</br>&nbsp;"]>(right)
    blockArrowId10<["&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;"]>(right)
    blockArrowId6<["5. Enrich / Complement</br>the data for the</br>use case at hand"]>(y)
    blockArrowId7<["6. Create a semantically</br> rich 3D representation</br>that can be explored</br> with an IFC Viewer"]>(right)
    blockArrowId8<["</br>7. Import a bespoke</br>non BIM/IFC model</br>&nbsp;"]>(right)
    blockArrowId9<["</br>8. Store the model</br>into the database</br>&nbsp;"]>(up)
    space:2

This diagram provide a high level story board of what we want to show/validate as capabilities

Import a BIM/IFC from an external source and transform the IFC into an ifcJSON
Filter the ifcJSON while retaining only the elements that are relevant for the use case(s) at hand. In some cases, the Administrative Authority may have detailed what is expected in the BIM/IFC. This is the case for the Digital Building Permit of Dubai where detailed templates and instructions are provided for Archicad, Bentley OpenBuilding and Revit so that the exported BIM/IFC is fully adequate for the use of the Administrative Authority. However, this might be difficult to generalize and we start with the assumption that a single BIM/IFC will potentially be used across distinct Administrative Authorities and need to be filtered depending on the use case(s) at hand.
Store the ifcJSON into the database (PostgreSQL)
Extract the data from the database for a target 'Spatial Unit' (a Building, a Storey, an Apartment, Common part(s),...), while leveraging the power of SQL, possibly complemented by other capabilities when needed.
Enrich/Complement the data for the use case at hand. This means, create new groups (such as Zones, Spatial Zones or simply Groups) an create/update PropertySets. This is where external applications will highly interact with d4SU platform.
Read data from the database and rebuild an ifcJSON with the relevant data and convert it to an IFC file so that standard IFC viewer can be used to navigate through the data (not just a 3D shape)
Import a bespoke non BIM/IFC model (typically from a legacy system). The import will transform the legacy model into 'objects', 'representations', 'relationships' and 'propertySets'. We assume that it cannot be expected that every input will be BIM/IFC from day one or that the past can be ignored. Therefore, other models must be accounted for.
Store the bespoke model into the database
Create and Store an ifcJSON for the bespoke model. From there it will behave as a standard BIM model, although not suited for use by the AEC insdustry to 'construct' the represented building or part thereof. We will give an example of how simple it is to build an IFC from a model using IfcOpenShell. But practically, creating a meaningful IFC for a bespoke model will always require a development to accomodate the specifics of the model.

Key Technical Features

The following schema depicts key technical features.

flowchart LR
    Ifc
    Custom["Custom Model"]
    ifcJSON
    Other["CityJSON, LDB, glTF,..."]
    DB[("Database")]
    API <--> IfcOpenShell
    API <--> SQL
    API <--> ifcJSON
    API <--> Graph
    Graph <--> DB
    Custom -- convert --> Ifc
    Custom -- is stored --> DB
    SQL <--> DB
    IfcOpenShell --> Ifc
    Ifc <-- convert --> ifcJSON
    Ifc --> Other
    Other --> Ifc
    ifcJSON <-- store | retrieve --> DB
    style API fill:#148F77
    style DB fill:#148F77

Technically speaking, FastAPI and Celery provide the glue to bring together the capabilities provided by the BIM/IFC ecosystem in order to cover the needs of d4SU.

On the left we have the API and on the right we have the Database. Between them, different representations are activated to leverage the capabilities of IfcOpenShell (including IfcPatch, IfcConvert), IfcJSON (two-way conversion between IFC(STEP Model & file) and IfcJSON (a JSON Model & file)), SQL and possibly CYPHER. Also, different conversion tools can be enacted to produce other representations: e.g.:

COLLADA or glTF for Google Earth, Cesium, ... (IfcConvert)
CityJSON for GIS applications (IfcConvert, IfcCityJson)
LBD (Linked Building Data) with IFCtoLBD

Complex sequences of transformations can be executed, with the help of IfcPatch to:

Extract elements of the model, for instance all standards elements [Wall, Slabs, Beams, Columns, Windows, Doors, Spaces, Stairs]
Split in Storeys / Levels
Migrate a version, e.g from Ifc 2x3 to Ifc 4
Purge Data
Tesselate elements (simplify geometry rendering)

Link to other Capabilities (examples)

As more functionalities are made available, they could be leveraged by the platform. E.g., IfcLCA for Life Cycle Analysis or bim2sim for Heating, Ventilation Air-conditioning (HVAC), Building Performance Simulation (BPS) or Computational Fluid Dynamics (CFD)

Code structure

The code is availble on GitHub under /src with a specific structure

directory	use
/web	API layer
/service	Service layer
/model	Model layer; provides the models and the tables definitions
/data	Data layer; takes care of all that pertains to CRUD on postgreSQL or to read/write of files through PyArrow
/local_module	Modules from other sources (with small adjustments)
/long_bg_task	Base for Celery (with celery.py and task.py)
/long_bg_task/Task_modules	Long running background tasks (executed by a Celery Worker). By 'long running', it is meant 'with a runtime measured in hundreds of milliseconds or in seconds'

Ifc Files

Ifc files used for tests and explorations come from reference sources such as:

The following files have been extensively used:

Duplex_A_20110907_optimized.ifc; 'Autodesk Revit Architecture 2011 - 1.0','20100326_1700 (Solibri IFC Optimizer)',''); FILE_SCHEMA(('IFC2X3'));
IFC Schependomlaan.ifc; with BREP geometry and 2ndLevel IfcRelSpaceBoundary. IFC file generated by Graphisoft ArchiCAD-64 18.0.0 NED FULL Windows version (IFC2x3 add-on version: 6000 NED FULL)
Duplex_M_20111024_ROOMS_AND_SPACES.ifc; with BREP Geometry; 'Autodesk Revit MEP 2011 - 1.0','20100903_2115(x64)','');FILE_SCHEMA(('IFC2X3'));
ARK_NordicLCA_Housing_Concrete_BuildingPermit_Revit.ifc; Autodesk Revit 24.0.5.432 (ENU) - IFC 24.0.5.432','');FILE_SCHEMA(('IFC4'))
AC20-FZK-Haus.ifc; an IFC demo model with an extensive use of IFC modeling capabilities (e.g. use of product types with representationMaps instead of direct product representations).

Duplex_A_20110907_optimized.ifc

The file Duplex_A_20110907_optimized.ifc provides a model with many features. The model contains as usual a project with a building and several storeys. Also, as usual, there are walls, slabs, beams, doors, windows, stairs, spaces and furnishing elements. Spaces are delimited by 'IfcRelSpaceBoundary' '1stLevel' relationships that provide space delimiters such as 'walls', 'slabs', 'doors', 'windows'. Also, there are many 'IfcRelConnectsElements' that documents the connections between walls and have some side effects that must be handled.

View of Duplex_A_20110907_optimized.ifc with OpenIFCViewer

View of Duplex_A_20110907_optimized.ifc with That Open

View of Duplex_A_20110907_optimized.ifc with BIMCollab zoom

View of Duplex_A_20110907_optimized.ifc with Solibri Anywhere

IFC Schependomlaan.ifc

IFC Schependomlaan.ifc is a typically rich (hence complex) IFC file. Among its many features it contains a SpaceBoundary2ndLevelView. Spaces are delimited by 'IfcRelSpaceBoundary' '2ndLevel' relationships that are suited for energy analysis. There are however some anomalies in the file. There are 10 apartments in the building, some have doors (an not windows) in the space boundary and some have windows (and not doors) in the space boundary.

Residential Building from Nordic Sustainable Construction BIM4LCA Files

BIM4LCA - Residential Building is a recent set of IFC files with different variants published 09/2024. According to the publishing site, "In all design phases of the project, the disciplines carried out the design work solely by modelling. As a result, several model versions were produced. All the models are available to download... They are free for anyone to use under the Creative Commons license (CC BY-SA 4.0)". Spaces are delimited by 'IfcRelSpaceBoundary' '2ndLevel' relationships that are - expectedly - suited for energy analysis.