Geometrical verification of as-built BIM models by deviation analysis

Within the framework of model creation it has to be ensured whether the existing as-built documents correspond to the actual condition of the as-built buildings or whether there are deviations. 3D models for renovation projects are often generated on the basis of 2D drawings and therefore represent the status quo in the drawings, depending on the objective of the project and planned activities a 3D model corresponding with the exact as-built situation is required. The most accurate way to get the 3D model matching with the as-built situation is by the use of a laser scan point cloud. Doing laser scans of the inner and outer surfaces of the building leads to an exact representation of the building in form of a point cloud. The point cloud is to be further processed and afterwards used to compare it with the already existing 3D model; results are measurable deviations, which provides information on whether the model needs adjustments or not.

The following process map gives an overview of the specific process step “deviation analysis”.

Figure 1: Process map Deviation Analysis

The 3D BIM model of the existing building is prepared based on model requirements, with the possibility of generating the model from existing 2D drawings. The point cloud derived from laser scans are imported and purged, thereby removing all unnecessary points. Several point clouds are aligned by rotating and adjusting the height in order to compile the combined scans to one point cloud (called “manually registration”). Simultaneously, a 3D mesh is extracted from the existing 3D model and the mesh and point cloud are analyzed for deviations. This manual alignment of both will be used to determine the deviation of the 3D BIM model from the as-built situation. This allows to decide if the 3D BIM model can be used for further use cases.

Demonstration site:

Deviation analysis of the existing 3D model against the available point cloud has been carried out on the Dutch Demo site located in Warmond. The 3D model was generated based on 2D drawings, while in parallel, a laser scan of the surrounding and outer shell of the building had been done.

The first step carried out was to create a 3D mesh of the 3D model generated from 2D drawings. 

Figure 2: Creation of 3D mesh

The laser scans on site contain all collected information as individual point clouds. In order to create a single point cloud, the several scans have been combined, by the following steps:

Import : The laser scans are imported to Leica Cyclone, thereby the point clouds are compiled as unregistered.

Purge scans: The unregistered point clouds are checked upon which the mirrored and unnecessary points (for instance, wrong measurements due to mirroring effects of windows or flat surfaces like puddles) from point clouds are removed, thereby purging the point clouds.

Figure 3: Purging of point cloud

Alignment: The two separate purged scans are chosen, rotated and aligned in order to optimize adjustment. The result obtained is a registered and aligned scanned model.

Figure 4: Aligning of point cloud

Evaluation of deviation analysis:  Upon manual alignment, the results are optimized by the computer to reach the best possible results, which are then reviewed in order to check if the alignment is accurate enough. The errors and their weights are checked and the end result of this stage is a combined scan of the point cloud registration.

Figure 5: Analyzing and optimizing results

The results from deviation analysis performed on the Dutch demonstration site yielded the results shown in the image below. The blue color indicates that the point cloud differs with the inner direction of the building shell, while red indicates deviation in the point cloud to the outer surrounding.

Figure 6: Evaluation of the model overlaid with the point cloud

 

Figure 7: Data exchange between activities within Deviation Analysis 

Exchange requirements identified for this use-case are indicated as input and output for every activity performed in the execution of the use case:

• BEP and modeling guidelines form the basis of definition of model requirements which further leads to model preparation.
• A point cloud is purged while simultaneously, a 3D mesh is extracted of the existing 3D model, both of which are outputs from individual activities and forms the input for deviation analysis to be performed.
• The coordination model is the output of the deviation analysis performed which is examined for allowable tolerances.