Opinion

Fundamentals of Point Cloud to BIM

 

Building information modeling has become a prerequisite for large scale construction projects. It is not just the new building construction projects that benefit from BIM. Today several as-built projects for heritage buildings, hospitals, airports, educational institutions and universities, high rise apartments and office buildings are using BIM for documentation, to enable renovation, retrofitting, reconstruction and spatial management and for efficient O&M.

The best way to manage as-built projects is to laser scan the building or facility, and use this captured data to develop BIM ready models. This process is widely known as Scan to BIM, however; practically a lot transpires between ‘Scan’ and ‘BIM’. While the process of capturing an accurate point cloud model via laser survey is by no means an easy task, the real challenge lies in taking the point cloud model as reference and developing a parametric 3D BIM model.

  • Handling the Point Cloud Model

When laser surveys are done for huge buildings and heritage sites, the captured survey data is massive; it may consist of around 300 to 500 scans. Such huge files need hi-end processors and equally massive storage space. Having licensed software tools like Autodesk Revit is also one of the prerequisites.

Although software tools might be capable of importing entire point clouds, it is difficult to handle these within the software ecosystem. Hence modelers need to extract sections and model them individually. Experts may then use techniques like point snapping and accurately retrace the point cloud regions with 3D elements such as nurbs, polygons and surfaces. This process is known as surface reconstruction.

  • Why using the right software tool for Point Cloud to BIM conversion is important?

Several software tools used for BIM modeling do not accept point cloud data. Now in this case point cloud data is viewed separately, and 2D CAD drawings, ortho images and planar projections are used as reference to build 3D BIM model. These scanned images may also be setup in a photo like panoramic view and used as reference. However, now the entire process of laser scanning and deriving point clouds becomes pointless, as the inputs are reduced to mere 2D images and displays. This also increases the number and level of engineering assumptions to be made, and also affects the overall accuracy of the BIM model.

  • Complete automation of Point Cloud to BIM conversion is not viable, and still needs expert intervention. Why?

It would be great if we just had to scan a building to get a BIM ready model. However convenient, this is not a viable option. Why? Because; artificial intelligence has not yet developed to the extent that it can make engineering assumptions on its own.

When real objects i.e. the building and objects within the building are scanned, they exhibit more characteristics and details than what needs to be included in a BIM model. Additionally, an occupied or non occupied building might have objects such as furniture, plants, curtains etc that pose as an obstruction in the process of capturing complete building data that is required. For example if a chair or a couch is placed against a window, the scanner cannot capture that part of the window which is hidden behind this piece of furniture. In such cases, the modeler’s expertise and experience comes into the picture.

Expert’s intervention cannot be undermined because, they understand the requirements and application of BIM in a construction project, and hence the model, accordingly. This is done by eliminating the unwanted characteristics and details, and modeling the missing parts due to interferences, using engineering assumptions.

An as-built structure may have walls that have buckled or are out of alignment, and point cloud captures the as-built with all these imperfections and these deviations are corrected during the point cloud processing stage. Also for retrofit and refurbishment projects, the differences between the ideal geometric primitives and actual conditions are recorded. Repair work and retrofitting is engineered accordingly.

The best approach is to make assumptions during modeling phase, and also use standard Revit families wherever it is feasible.

BIM and Point Cloud to BIM, have become a norm, and are imbibed into the processes for construction and widely used technology for all major as-built projects.

Thanks for reading!

Please enjoy a limited number of articles over the next 30 days.

For total access log in to your The BIM Hub account. Or register now, it's free.

Register Sign in

Published
Source: www.truecadd.com/news/fundamentals-of-point-cloud-to-bim