A new Scan to BIM to FEM procedure: Integration of 3D modelling techniques
The main aim of this thesis, by the author Ameer Emad Saud Hajjat, is to explore the state-of-the-art of the Scan to FEM procedures and apply them to an important monument located in Guimarães, Portugal, the São Torcato Church.
The Scan-to-BIM-to-FEM procedures have been spreading interest in the last few years due to the new innovative technologies of the remote sensors techniques, i.e. laser scanning and digital photogrammetry that made it possible to detect the geometry of historic buildings.
The literature review explores the state-of-art of the different scan-to-FEM procedures and the parametric 3-dimensional modeling approaches through other researchers’ findings, gives a critical overview of the field’s existing knowledge, and highlights the gap of state of the art.
Heritage buildings are usually characterized by complex (non-parametric) and irregular geometries that make the traditional modeling techniques of the exact shape of the building inaccurate and a time-consuming process.
REMOTE SENSOR TECHNIQUES
The development of different surveying technologies offers great potential in obtaining detailed measurements of existing buildings to generate 3D models that may be adopted for several purposes, including structural simulations.
Thanks to state-of-the-art technologies in architectural surveying, it has become possible to define the exact measurements of any building in a short time. However, choosing the appropriate survey method depends on different aspects, such as:
- The scale of the building,
- Its geometrical complexity,
- Its accessibility,
- The budget and time available.
1 – PHOTOGRAMMETRY
Photogrammetry is defined as “the science, and art, of determining the size and shape of objects as a consequence of analyzing images recorded on film or electronic media”.
Within the context of heritage preservation, the photogrammetry technique has long been used for collecting three-dimensional information of cultural heritage objects together with texture information.
Concerning the scale of the captured object, the photogrammetry procedures can be summarized as follows:
- Image Capture;
- Image Matching;
- Dense Point Cloud Generation;
Click here if you want to know more about SAHC advanced masters
2 – LASER SCANNING
Laser scanning is an active, fast, and automatic acquisition technique using laser light for measuring, without any contact and in a dense regular pattern, 3D coordinates of points on surfaces. Laser scanners emit and receive their own electromagnetic radiation rather than relying on reflected ambient or artificial light as in (passive) photography.
Within the cultural heritage context, laser scanning techniques have greatly impacted because of its simplicity, speed, and accuracy. Laser scanners acquire historic constructions by detecting thousands of points in a short period. The collected data can deliver digital 2D drawings or 3D models useful for various cultural heritage preservation applications.
Laser scanners can operate following three different principles:
- Pulse (time-of-flight; ToF);
APPLICATION OF REMOTE SENSORS ON HISTORIC MASONRY STRUCTURES
Table 3 provides a summary of the main features that were compared of both photogrammetry and laser scanning techniques within what has been found through other researcher’s works and publications.
CASE STUDY: SÃO TORCATO CHURCH
History of São Torcato Church
São Torcato village is located on the left side of the Selho River, seven kilometers north-east of Guimarães, in the northern part of Portugal.
Sao Torcato church is characterized by different construction technologies that can be observed with various styles and materials used in the current building. The building was first designed by the Portuguese architect Luis Ignacio Barros Lima, who started the building with the construction of the apse in 1825 resulting in a shaped temple that was concluded in 1855.
Since the laying of the first stone, the São Torcato church continued under the supervision of several architects over nearly two centuries, until the church was completed in 2008.
The church shows a Latin cross longitudinal plan with a symmetrical layout. The design of the church combines several architectural styles, like Classic, Gothic, Renaissance, and Romantic, that have been expressed in the church’s interior and exterior facades. This “hybrid” style is also called “Neo-Manuelino” in Português.
- The central nave has a length of nearly 58m and 11m width, which ends with a transept and an apse.
- Above the main entrance, in the opposite direction to the apse, the choir looks over the holy space.
- The transept intersects the nave and has a length of 37m and a width of 11.5m.
- The loads are lifted by bearing columns on each limb that carry a parallel vault connected to semicircular arches. A wooden truss roof above them protects the vaults.
- The crossing between the transept and longitudinal nave is covered with a dome that sits on an octagonal tambour supported by four semi-circular arches. The facade is a splayed portal with a central rose window, at the top, and a balustrade bounding a gallery that is accessible from the gable roof.
- Two spired towers having a rectangular plan of 7.5m × 6.3m and with a total height of 58m symmetrically enclose the facade.
- The towers have an inner stone staircase that runs along the walls up to the level of the bells, present only in the western tower
The church has demonstrated the use of mainly two different materials that can be observed in different locations along the building as follows:
- Regular granite masonry blocks with thin mortar joints for the towers and nave walls.
- Reinforced concrete walls covered with granite veneer at the apse and the main altar.
Structural Retrofitting for Past Interventions
The church has been under different inspection works and interventions to investigate the structural problems that appeared over time:
- Ramos et al. (2013) performed dynamic structural health monitoring activities in the church. The case study has shown significant structural problems due to soil settlements that develop visible cracks on the main and lateral facades.
In addition, the bell towers exhibited leaning movement behavior, with failure mechanisms and structural cracks at the nave arches along with vertical deformations.
- Sánchez-Aparicio et al. (2014) identified the cracks on the main facade, arches, and vaults using laser scanning and photogrammetry techniques. In addition, the author represented the possible structural failure mechanisms of the towers and the main facade to the moderate-severe structural damage inspected.
- Masciotta et al. (2017) carried out an extensive experimental campaign to the case study, including visual inspections, geometric survey, monitoring and control, and damage diagnosis.
Click here if you want to know more about SAHC advanced masters
SCAN-TO-BIM-TO-FEM: FRAMEWORK AND APPLICATION
The proposed workflow is a parametric modeling approach based on flow-based programming and generative algorithms.
Generative programming is a programming paradigm that uses automated source code creation through generic frames, prototypes, classes, aspects, templates, and code generators to enhance programmer productivity.
Generative Design can be used as a design method in architectural modeling where objects and forms are generated using rules and algorithms integrated into computational tools and scripting platforms such as Rhinoceros, Grasshopper, Processing, and others.
Visual Programming also refers to some Generative Algorithms software that uses graphic illustration to make more sense to humans than typical text-based programming languages
Generative algorithm modeling raises significant importance in generating 3D models for complex architectural geometries. Although this procedure has been widely used for new constructions mainly, this study introduces a novel approach by applying a Non-Uniform Rational Basis-Splines (NURBS) based parametric generative modelling for historic constructions.
Theoretical framework: Scan-to-FEM for historic constructions
The first step of the procedure consists of a 3D survey of the historical building.
The present work will only focus on transforming the discrete numerical model, e.g. point cloud, into 3D objects and elements.
The proposed approach may be applied to case studies, presenting similar elements repeated several times and following rational rules in the final architectural layout. The research questions behind the current work are:
Can a structure be modeled using a parametric approach?
Is the error generated reasonable?
Heritage buildings are characterized by uniqueness. However, some heritage buildings present modules repetition following rational rules, a typical characteristic of parametric objects.
However, the parametric approach has shown difficulties in applying for extreme non-regular geometries, where there are no identical or similar architectural elements. In this case, the model may be generated following:
- Manual modeling approaches;
- Blending other 3D modeling strategies (polygonal modeling).
Description of the proposed procedure
After analyzing the structure and identifying similar geometries, it is possible within this methodology to create a set of sub-components, which are the basic architectural elements, i.e. columns, bases, etc.
Those elements can be modeled separately considering the required level of detail, and then they can be gathered and located within the model based on the architectural and geometrical configuration.
The level of detail of the 3D model depends on several aspects, e.g. time, costs, etc. The gathering of the sub-components then creates the components that can be intended as the architectural module that the designer adopted.
After the sub-components and components generation, the structure can be assembled by locating the objects in their positions by performing the correct interconnections between architectural and structural elements. The procedure explained before is displayed within the red box in the proposed workflow.
At this stage, the sub-components component does not contain any information about the damage state, crack pattern, quality of the masonry, etc. In order to include such information, many approaches can be theoretically implemented:
- Drawing a polyline that envelops the damaged structural part.
- Another methodology, mainly devoted to visualization purposes, uses the texture mapping technique by associating the geometrical elements with a texture arising from a photogrammetric survey.
Finally, the generated NURBS model can be turned into a BIM model by importing it into the commercial package Autodesk RevitThe final result is a complete HBIM solution for parametric 3D modeling that exploits point clouds for simple and complex shapes, including the damage parameterization and the capability to be used for structural analysis purposes (e.g. FEM analysis).
This methodology allows performing structural analyses without redrawing a new model used only for FEM purposes, as the obtained BIM model can be converted to a FEM model.
In order to consider architectural features and decorations, the user can import further entities as non-structural components to the BIM model, e.g. by using other numerical modeling approaches.
Application Scan-to-BIM-to-FEM Methodology to the São Torcato Church
Based on the visual inspection of the interior and exterior of the Basilica, it was possible to observe similar geometries and construction elements repeated along with the church geometry. Furthermore, the church shows a symmetric layout and architectural elements repeated in several locations.
Therefore, it is definitely convenient to model the Sao Torcato Church by following a parametric approach. To this end, a point cloud previously acquired by means of a 3D laser scanner has been adopted.
In order to validate such an approach, the point clouds of similar elements have been geometrically compared using CloudCompare open-source software, which allows the detection of the divergence between two similar geometries having the same topological characteristics. The comparison computes:
- The mean distances,
- Standard deviation;
- Minimum and maximum values of the two entities using the cloud-to-cloud tool.
Before computing the distances, fundamental operations should be applied, such as:
- Point cloud cleaning;
Once the computation is done, the software displays the results by hiding the reference cloud and showing the divergence map of the other point cloud.
It can be observed how the compared structural elements present a low divergence. This ensures that the structural elements are almost identical and can be modeled parametrically.
Sub-components and components definition
The subsequent stage of the proposed procedure consists of the identification of sub-components and components.
- The base;
- Rectangular column;
- Cylinder column;
- And the arch shown above has been modeled separately and afterward has been assembled to create half of the complete geometry.
As the element is symmetrical, it is possible to mirror the generated object to obtain the complete component without repeating the same process to model the other half.
RhinoGrasshopper software was chosen to do the parametric NURBS modeling for all of the structural and architectural elements.
By using this approach, it is possible to modify the basic geometrical parameters by using Number Sliders in an interactive, smooth, and simple way (1,7,8), to control the X, Y, and Z coordinates, the origin of the geometry, and the necessary dimensions.
The same procedure was repeated for the other subcomponents. It is worth mentioning that the resulting structural elements have been placed in their correct positions based on each other’s coordinates by linking them to the same sliders. This method ensures accurate interconnections between the elements and later helps to parameterize the full component using fundamental parameter sliders.
Once the parameterized definition of the first component is obtained, a proper Python script, including the repetition rules identified by means of the geometrical survey, has been adopted to locate the components.
The process becomes faster as the other recognized components share the same geometrical characteristics and contain the same sub-components. Furthermore, most of them have been adapted by changing the dimension parameters without the need to redraw them again, which made this approach efficient in generating the full geometry of the Sao Torcato Church within a short time.
After generating the full geometry of the case study, the inspected damages were parameterized using the polyline method. The full geometry of the church is shown below.
As reported in Figure 41, the full NURBS-based geometry generated using Rhino + Grasshopper software has been imported to the Autodesk Revit software as a final choice. This can be done by using the usual export/import processes.
However, the proposed methodology used a novel technique by means of Rhino.Inside.Revit plugin (www.rhino3d.com/inside/revit/beta/), which allows the user to real-time connect a parametric geometry from Grasshopper to Revit.
It should be noted that BIM methodology does not represent the architectural elements in terms of geometry only, but also contributes to:
- Heritage management;
- Conservation status monitoring;
- Preventive maintenance;
- Conservation and restoration planning;
- Analysis of intervention options;
- Structural simulation;
- Disaster preparedness.
It is also possible to add the non-structural parts and decorations to the BIM model by using triangular mesh generation algorithms derived directly from the point cloud. This method can be done by using Cloud Compare software employing the Poisson Surface Reconstruction.
FE MODEL: GEOMETRY AND CALIBRATION
Model Calibration: Non retrofitted configuration
The next figure represents the FE model imported in ABAQUS CAE. The mesh discretization is achieved using tetrahedron (Delaunay) FE’s due to its adaptability to more complex geometries.
Model Calibration: Retrofitted configuration
The following retrofitting interventions were modeled:
1. Anchoring design;
2. Micro-pile design;
3. Crack injections.
In this dissertation, the Generative Programming paradigm was adopted for the Scan-to-FEM purpose of historic masonry buildings. The proposed procedure was applied to São Torcato Church.
Once the acquisition of the point cloud, the geometry is analyzed semantically to identify repetitions of modules, symmetries, etc.
An abacus of components and sub-components was created. Such a strategy allows the users to use the components several times and also in different projects.
Afterward, the model generation passed through implementing the rationale rules that defined the original layout of the case study. Such a stage was also implemented using a GHPython script. Thus, the input data were all the previously defined components, whereas the outputs were them, in turn, rotated, moved, and copied in order to match the point cloud model. The most challenging task was the translation of the original architectural layout in a set of coding scripts.
This dissertation has also covered the gap in the literature regarding the automatic importing of geometry and material properties into FE software. A proper Python Script was created to perform the link between Grasshopper and Abaqus CAE.
Such a code can automatically compile a script that can then be run into Abaqus to assemble the FE model. This code has as input data:
- The output geometry of the components;
- The elastic properties of each component (which is defined taking into account the damage state of each structural part by means of the application of penalty values);
- The coordinates that define the position of the constraints;
- The computational mesh size that the user wants to apply for each entity the loads.
Finally, the proposed procedure demonstrated its capability to minimize the modeling time and, thus, the costs. Thanks to the point cloud arising from the laser scanning technique, it was possible to model the geometry with remarkable accuracy in a reasonable time.
The model calibration was performed. Two configurations were investigated: before and after executing the retrofitting intervention (2014-2015).
- The criterion adopted to calibrate the numerical model was based on the frequencies only.
- As initial values for the calibration procedure, the optimized mechanical parameters reported in Sánchez-Aparicio et al. (2014) were adopted, which refer to the non-retrofitted configuration.
- The numerical results are in good agreement with those experimentally obtained by Masciotta et al. (2017) for both configurations.
- Future development of the proposed tool would include segmentation algorithms, e.g. based on Artificial Intelligence.
- In addition, such a futuristic view could include the automatic re-position of components and subcomponents based on the data arising from the digital survey.