Automating Code Compliance in AEC

This article looks at why the process of getting building approvals from regulatory agencies remains very much a 2D process, with the building drawings still being checked manually for code compliance, when much of it could have been automated if a BIM model was being used. It explores the various automated code-checking efforts that have been initiated so far like CORENET e-PlanCheck, SMARTcodes, and AutoCodes and their current status, as well as the technologies that are needed to actually automate the code-checking process.


5 thoughts on “Automating Code Compliance in AEC

  1. I am very sorry to learn that the CORENET code checking effort has not (yet) been successful and has been dropped. I wonder if you can get more information regarding the reasons for this. There are many possibilities, but it might be possible to learn from their experience so that future efforts can succeed.

  2. Thanks for your thorough and insightful article. The major barrier to automating code compliance was and remains the content of the models. In order to check whether a corridor meets some minimum width requirement, for example, the space itself must be defined, the space function must be identified as a corridor, and the check needs to be performed along the full length of the corridor, through turns and doors. The geometry is not enough – if the semantic information of the design subject and intent is not explicitly made available, the code check cannot be performed.

    While BIM is of course a major improvement over any kind of 2D representation, the concepts required for most code checking are still absent from most models. There are two reasons for this:
    a) BIM modelers are not concerned with making their models explicit in every way. Why make the extra effort to explicitly model a ‘space’ object when modeling four walls is sufficient for all of your immediate needs?
    b) Different design codes and building codes in different jurisdictions define their rules with different concepts and enumerated values, so that even if one could persuade modelers to model in a particular way, they would need to tailor their work carefully for every different check.

    People can quite easily infer the meaning and design intent of a building from 2D drawings if they have the appropriate professional background and training, and thus the task of code checking is still in the hands of people. Despite the resources devoted to the problem described in the article, code-checking has not yet been successfully automated.

    If one could shift the onus of generating explicit information from the BIM modelers to the receiving applications, in this case the code-checking applications, then it becomes possible to envisage coed-checking systems that can receive generic BIM models and ‘semantically enrich’ them as a pre-processing stage before checking them. This is the approach we have adopted in the SEEBIM research (Semantic Enrichment Engine for BIM models) – see:

    The idea is to embed the domain knowledge of a person in an expert system for pre-processing. A trained code-checker, for example, can look at a BIM model and see the corridors or any other semantic constructs by virtue of their understanding of the spatial topology, materials, object types, and any other cues that are present, and add the ‘corridor’ space object with its function as a preparation for code-checking. In general, their process of inference can be expressed as a set of ‘if-then’ rules that can be evaluated against a model in order to add new facts. This has been shown to work in the domains of precast concrete modeling, quantity take-off and cost estimation for precast concrete, and a new project has just begun to apply the approach to the compilation of highway bridge models (see

    While the SEEBIM approach is still in the research stage, it may offer an alternative to the current need for rigorous preparation of BIM models that is currently necessary for any kind of code-checking, and make automated code-checking possible in the future. A major advantage is that it may be better aligned with the business interests of the BIM software vendors, in that the cost of development is shifted to the importing application and away from the exporting application.

  3. Hi,
    It made a very interesting read.We are an IT company from Chennai,India.We have developed a Building plan scrutiny (code check) system for our local bodies which is operational for about 10 years now.The submission drawings are 2D and the software extracts the relevant CAD data and compares it with the bye-laws of the city councils and generates a report on compliance and thus helps speed up the compliance check process.
    We are still sticking to 2D for this process primarily because 3D BIM models are yet to catch up and the cost of 3D BIM models are still very high ( limited trained resources compared to 2D draftsmen).We would be glad to give a live online demo to anyone interested.

  4. Good to hear about the work being done on automating code-compliance, both in academia and in the industry. I will plan a follow-up article and explore these initiatives in more detail. I also hope to find out what happened to CORENET! It seemed so promising.

Comments are closed.