BIM-based concrete printing

Kay Smarsly1, Patricia Peralta1, Daniel Luckey1, Sebastian Heine2, and Horst-Michael Ludwig2

1 Bauhaus University Weimar, Chair of Computing in Civil Engineering, Weimar, Germany 2 Bauhaus University Weimar, F.A. Finger Institute for Building Material Engineering,

Weimar, Germany kay.smarsly@uni-weimar.de

Abstract. Extrusion-based additive manufacturing (AM), or three-dimensional (3D) printing, has matured into a set of advanced methods to automate the con-struction of large-scale concrete structures, while minimizing cost and material waste. However, current AM data models are inadequate for 3D concrete print-ing due to insufficient incorporation of information on the relationships be-tween process, material, and geometry, which may cause redundancy, infor-mation loss, and inconsistencies. Aiming at improving AM data modeling for concrete printing, this paper proposes a metamodeling approach for AM of con-crete structures, referred to as “printing information modeling”, which takes ad-vantage of building information modeling (BIM). As will be shown in this pa-per, the BIM-based printing information model, serving as a metamodel, incor-porates the digital data triplet of process, material, and geometry parameters to generate computer numerical control (CNC) commands that may readily be used for concrete printing. A validation test is performed, which instantiates the printing information model, using a BIM model, for generating CNC com-mands, enabling optimal digital data exchange from BIM models to concrete printers. As a result of this study, it is demonstrated that printing information modeling adequately defines the information required for AM of concrete struc-tures using a BIM-based approach, showing promising potential to improve cur-rent AM data modeling efforts.

Keywords: Additive manufacturing, concrete printing, printing information modeling (PIM), metamodeling, building information modeling (BIM).

1 Introduction

Driven by recent Industry 4.0 developments, the architecture, engineering, and con-struction (AEC) industry is experiencing an era of transformation and ongoing auto-mation (Tay, et al., 2018). Recent approaches towards automating construction meth-ods are centered on extrusion-based additive manufacturing (AM), or three-dimensional (3D) printing, of large-scale concrete structures (Bos, et al., 2016; Bus-well, et al., 2007). ASTM International, formerly known as American Society for Testing and Materials, defines AM as a process of generating objects from 3D model data by joining materials, usually layer upon layer (ASTM, 2012). Research to date

2

has focused on developing viable AM hardware and materials to achieve concrete structures with certain degrees of geometrical precision, while reducing cost and waste (Buswell, et al., 2018). In an attempt to improve geometrical precision and process control, sensor technologies in AM of concrete structures have been presented by Wolfs, et al. (2017), illuminating the advantages of real-time monitoring to adjust process parameters during printing. However, the full potential of deploying sensing technologies, which has become common in other field within the AEC industry (Dragos & Smarsly, 2016; Hartmann, et al., 2011), has not yet been fully exploited in AM.

In recent years, efforts have been undertaken towards AM data modeling for con-crete printing, where converting 3D model data into printed components is derived in the same way as in conventional AM (Peralta, et al., 2019). In conventional AM, data modeling is mainly based on Standard Tessellation Language (STL) and G-code, i.e. ISO 6983-1 (Bonnard, et al., 2018). However, current data models are not suitable for concrete printing of components of high load-bearing capacity and geometric preci-sion. Shortcomings regarding AM data modeling for concrete printing include a lack of knowledge on how processes, materials, and geometries interrelate in AM data modeling, causing redundancy, information loss, and inconsistencies. Thus, current data modeling approaches limit concrete printing to an inconsistent and unreliable process, where expertise is required to determine ideal printing process settings, tool-paths, and printing strategies, taking into account variations in material properties of fresh concrete over time. The main challenge in AM data modeling, therefore, is to understand the relationships between AM processes, materials, and geometries, i.e. the digital data triplet.

To compensate for the shortcomings and to optimize AM data modeling for con-crete printing, a metamodel for AM of concrete structures that takes advantage of building information modeling (BIM) is proposed in this paper. The BIM-based ap-proach towards concrete printing enables AM data modeling and digital data ex-change without breaks in the digital thread, incorporating real-time monitoring of material properties and adjustments of process parameters, based on previous studies (Theiler & Smarsly, 2018). The metamodel represents a step necessary for standardiz-ing the digital data triplet into a unified printing information model for concrete print-ing in civil engineering.

The remainder of this paper is organized as follows. First, the metamodeling ap-proach, referred as printing information modeling (PIM), is shown. The PIM model defines input parameters and relationships within the AM digital data triplet to gener-ate computer numerical control (CNC) code for providing hardware control of con-crete printers. Second, using the PIM model as a formal basis for concrete printing, a validation test is devised using BIM models that are defined according to the Industry Foundation Classes (IFC) standard as input sources to generate CNC code, enabling an optimal digital thread from BIM models to concrete printers. Finally, the paper concludes with a summary and an outlook on potential future research efforts.

3

2 Printing information modeling for concrete printing

The printing information modeling approach outlines a metamodel, which comprises the main input parameters of the AM digital data triplet, describing the relationship between the parameters to semantically represent specifications for concrete printing jobs. Following a unified modeling language (UML) modeling approach, classes are defined to coherently group the input parameters of the digital data triplet. As a result, breaks in the digital thread are reduced, while optimizing digital data exchange for concrete printing. As a result, the PIM model represents an understandable and instan-tiable metamodel that may be implemented into technology-independent semantic descriptions of buildings, such as the IFC standard. In the following subsections, the main input parameters of the digital data triplet, the relationships, and the PIM model for concrete printing are introduced.

2.1 Main parameters and relationships

According to the digital data triplet, the main parameters relevant to the PIM model presented in the following subsection are subdivided into process parameters, material parameters, and geometry parameters. When describing the parameters, elements directly reflected in the PIM model are printed in italics.

Process parameters. Most concrete printers have adopted an extrusion-based AM method, referred to as “fused deposition modeling (FDM)”. Main elements of printing hardware related to data modeling in FDM printing are pump system, printhead, and motion-controlled printing system. The pump system is devised to transport concrete from the pump system to the printhead without segregation and bleeding. Relation-ships regarding pump system parameters have been identified, e.g., in studies of Bos, et al. (2016) and Paul, et al. (2018a). Pump system parameters include pump speed and pump pressure, which is associated with pump speed. The printhead allows the concrete to be printed at target locations with target speed under target angles, and it is composed of several parts, including a nozzle at its end. Printhead parameters in-clude printhead speed, printhead position and orientation, extruder speed, and nozzle shape. Relationships regarding printhead parameters have been reported, e.g., by Salet, et al. (2017), Paul, et al. (2018a), Lim, et al. (2012), and Wolfs, et al. (2017) for printhead speed, printhead orientation, nozzle shape, and printhead height above the printing surface. The geometry of the concrete filament extruded from the printhead influences the toolpath definition and the buildability of structural components to be printed. Monitoring the geometry of the filament provides control of the geometric precision during the printing process. The motion-controlled printing system process-es the CNC code, which defines toolpaths, controls the multiple axes, and monitors and adjusts the printhead performance based on sensing technologies. Influences of toolpaths definitions and sensing technologies on the properties of the concrete layers have been described, e.g., in studies of Salet, et al. (2017) and Wolfs, et al. (2017).

4

Material parameters. Concrete printing involves concrete as main printing material and reinforcement, which may be incorporated before, during, or after the printing process (Asperone, et al., 2018; Buswell, et al., 2018). The composition of concrete, defined in the mix design, determines whether a concrete mix satisfies printing pro-cess requirements and material specifications. Setting reactions of concrete show strong interactions with the process parameters and with the printing strategy applied for manufacturing. Relationships between material properties of fresh state concrete (e.g. green strength and layer interval time) and process parameters, together with the component size, are relevant to the way the printing process should be conducted to ensure extrudability, buildability, and layer adhesion, as identified in recent studies (Bos, et al., 2016; Paul, et al., 2018b; Salet, et al., 2017; Wolfs, et al., 2017).

Geometry parameters. Definitions of dimensions, points, and geometric objects are necessary to describe 3D geometries of the structural components to be printed. The definitions may advantageously be inherited from digital 3D models, such as BIM models. Dimensional parameters are defined to ensure geometric precision. Point parameters are used to define vectors and geometries. Geometric object parameters, which define component geometries, depend on the geometric representation used in the digital 3D models. For example, Correa (2016) has demonstrated that IFC-compliant geometric representations, such as boundary representation and tessellated representation, are adaptable to 3D printing. In addition, geometric parameters related to the AM processes should be considered together with the geometrical descriptions, such as contour lines resulting from slicing.

2.2 Printing information model

As described earlier, several studies have documented relationships between the pa-rameters of the data triplet. Herein, the studies are taken as basis to develop the PIM model, shown in the UML class diagram in Figure 1. The PIM model describes the printing information necessary for concrete printing according to the digital data tri-plet with three main components materialized in the UML class diagram in terms of the abstract classes,

ProcessInformation, MaterialInformation, and GeometryInformation.

Printing process data and monitoring data are included in the ProcessInformation abstract class, categorized into the PumpSystemData, PrintheadData, FilamentData, ToolpathData, and ControlData subclasses. The data of the material employed in the printing process, such as concrete and reinforcement, is described in the MateriaIn-formation abstract class with the subclasses ReinforcementData and ConcreteData, the latter including the subclasses MaterialSpecification, MixDesign, and Materi-alProperty: Material specifications define a concrete mix design, which, in turn, has a set of material properties that may be monitored. Finally, geometry data that may be

5

inherited from BIM models is defined in the GeometryInformation abstract class, categorized into the Dimension, Point, GeometricObject and ContourLine subclasses, to achieve a complete geometrical description.

Fig. 1. Extract of the PIM model.

3 A building information modeling approach towards concrete printing

Building information modeling is a promising means to provide complete information relevant to AM of concrete structures, incorporating the data triplet in a single data model. As shown in previous studies, with the IFC schema being extended to describe sensor systems (Theiler & Smarsly, 2018) or cyber-physical systems (Fitz, et al., 2019), incorporating control data for monitoring and adjustments of concrete printing jobs becomes possible on a standardized BIM basis. A BIM approach that implements the PIM model as a formal basis for concrete printing is presented in this section. The geometry parameters and several (but not all) material parameters required for AM are already specified in common BIM models compliant to the IFC schema. In this study, by incorporating all parameters and parameter relationships defined in the PIM model into a BIM model, printing strategies and toolpaths are defined and, then, commands that provide hardware control for concrete printing are generated as CNC code. Clearly, IFC schema extensions are needed to incorporate the process parame-ters and the material parameters that are not yet defined in the IFC schema, which is, however, not within the scope of this paper. For validating the BIM-based approach towards concrete printing, a software tool, the PIM test application, is designed and

6

implemented that extracts information related to the geometry and material parame-ters from IFC-compliant BIM models. The PIM application generates CNC com-mands from the geometry and material parameters, while process parameters are add-ed manually as user-defined parameters. The PIM application and the validation test are described in the following subsections, complemented by a summary and a dis-cussion of the validation test results.

3.1 Design and implementation of a PIM test application

To generate CNC code, the PIM application, developed in Java language, instantiates PIM models for concrete printing jobs based on process parameters entered manually as well as material and geometry parameters extracted from BIM models, respective-ly. In other words, the input data for the PIM application is the data stemming (i) from BIM models and (ii) from user-defined inputs (e.g. printhead and pump parameters), while the output is CNC code for hardware control of concrete printers. Furthermore, an algorithm for calculating slicing (to obtain contour line data) and an algorithm for calculating filling (to obtain toolpaths) is implemented into the PIM application. The slicing algorithm solves the plane intersection problem originating from slicing geo-metric objects into layers. The filling algorithm implements a raster-scanning ap-proach to determine the infill pattern for each layer of the geometric object and de-fines toolpaths, using a user-defined printing strategy (i.e. unidirectional or bidirec-tional) and deploying filament data to increase geometric precision. The CNC code is generated in terms of G-code files, incorporating specifications of printhead, pump, material, and toolpaths. So called “PIM factors” are applied to modify the printhead and pump parameters, aiming to compensate for variations in the material properties. The software architecture of the PIM application, with main packages and classes, is shown in Figure 2.

Fig. 2. PIM test application for validating the BIM-based approach.

7

3.2 Validation test

The validation test is performed for an L-shaped concrete wall of 80 mm height, 150 mm thickness and a base of 800 mm × 600 mm (length × width), designed as an IFC-compliant BIM model. The validation test is conducted on a gantry concrete printer, shown in Figure 3, and a custom-designed concrete mix with a design strength of 30 MPa. The gantry concrete printer has a capacity to print 6000 cm3 of concrete, a nozzle of 150 mm × 40 mm (length × width), and a printing area of 2000 mm × 1000 mm (length × width). The geometry and material parameters are extracted from the IFC file of the BIM model and used as input parameters to the PIM application. Regarding the process parameters, a printing speed of 100 mm/s, a pumping speed of 22 mm/s for a pump pressure of 1 MPa, a layer height of 40 mm and a unidirectional printing strategy are specified manually.

Fig. 3. Gantry concrete printer used for the validation test.

The PIM application slices the L-shaped wall, defined in the BIM model, into layers and generates a toolpath for each layer (i.e., two layers here). Specifically, the PIM application generates the CNC code for each layer, taking into account the material properties of the custom-designed concrete mix, such as slower printing speed and higher pump pressure when printing the first layer. The CNC code generated by the PIM application and the visualization of the toolpath corresponding to the CNC code are shown in Figure 4. The CNC code has a start command block to set initial process parameters, a positioning command block to set the start position of the printhead, a printing command block that controls the extrusion of the concrete printer, and a homing command block to finalize printing. The extrusion of concrete by the pump system is controlled with commands to turn the pump on and off. Finally, the CNC code generated by the PIM application is transferred to the motion-controlled printing system of the gantry concrete printer, and the wall is printed.

8

a) CNC code generated by the PIM application.

b) Visualization of the toolpath provided by the PIM application.

Fig. 4. Output of the PIM application.

3.3 Results and discussion

The validation test demonstrates the feasibility of the BIM-based approach towards concrete printing, using the PIM model as formal basis. From the IFC file of the BIM model, geometry and material parameters have successfully been processed. Addi-tional material parameters have manually been specified within the PIM application for the custom-designed concrete mix, such as an open time of 90 minutes, a setting time of one minute, and a maximum layer interval time of 5 minutes. While printing, which is shown in Figure 5, the pump speed and the printhead speed are adjusted from layer to layer to compensate for the changes in the material properties, caused by a loss of moisture of the concrete over time. However, it must be noted that the algo-rithms of the PIM application are not suitable for AM data modelling of structural components with complex geometries. The slicing algorithm implemented in the BIM

9

application is limited to the IFC entity IfcSolidModel defined as IfcSweptAreaSolid for generating contour lines. Furthermore, the scanning distance of the filling algo-rithm is defined by a filament width of 150 mm, resulting in a single infill line for the L-shaped wall. Also, it has been observed that the scanning-raster approach may gen-erate faulty toolpaths for components with more complex shapes.

Fig. 5. BIM-based concrete printing.

4 Summary and conclusions

A BIM-based metamodeling approach, referred to as “printing information model-ing”, has been developed as a first step towards standardizing the digital data triplet for additive manufacturing of concrete structures. Validating the PIM model in a test application, a BIM model has been used as input source to generate CNC code for a gantry concrete printer. It has been demonstrated that the PIM model, as a metamodel, semantically describes the AM digital data triplet, facilitating instantiable models for BIM-based concrete printing. Although a successful validation test can be reported, there exists room for improvements. For example, further studies are required to op-timize the filling algorithm for calculating toolpaths and to further advance digital models that implement the PIM model to generate CNC code for BIM-based concrete printing able to incorporate sensor data for process monitoring and real-time printing adjustments. Another direction for future work, which deserves being explored, is the extension of the IFC schema with process parameters and (further) material parame-ters, enabling additive manufacturing of concrete structures with IFC-compliant BIM models as a sole formal basis.

10

Acknowledgments

Financial support of the German Research Foundation (DFG) through grant SM 281/7-1 is gratefully acknowledged. Major parts of this work have been conducted in the “Structural Health Monitoring Laboratory”, sponsored by the European Union through the European Fund for Regional Development (EFRD) and the Thuringian Ministry for Economic Affairs, Science and Digital Society (TMWWDG) under grant 2016 FGI 0009. Any opinions, findings, conclusions, or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of DFG, EFRD, or TMWWDG.

References

1. Asperone, D., Menna, C., Bos, F., Salet, T. A., Mata-Falcón, J., & Kaufmann, W. (2018). Rethinking reinforcement for digital fabrication with concrete. Cement and Concrete Re-search, 112(2018), 111-121.

2. ASTM. (2012). F2792. 2012 Standard terminology for additive manufacturing technolo-gies. West Conshohocken, PA, USA: ASTM International.

3. Bonnard, R., Hascoët, J.-Y., Mognol, P., & Stroud, I. (2018). STEP-NC digital thread for additive manufacturing: data model, implementation and validation. International Journal of Computer Integrated Manufacturing, 31(11), 1141-1160.

4. Bos, F., Wolfs, R., Ahmed, Z., & Salet, T. (2016). Additive manufacturing of concrete in construction: Potentials and challenges of 3D concrete printing. Virtual and Physical Pro-totyping, 11(3), 209-225.

5. Buswell, R. A., Leal de Silva, W. R., Jones, S. Z., & Dirrenberger, J. (2018). 3D printing using concrete extrusion: A roadmap for research. Cement and Concrete Research, 112(2018), 37-49.

6. Buswell, R. A., Soar, R. C., Gibb, A. G., & Thorpe, A. (2007). Freeform construction: Mega-scale rapid manufacturing for construction. Automation in Construction, 16(2), 224-231.

7. Correa, F. R. (2016). Robot-Oriented Design for Production in the context of Building In-formation Modeling. In: Proceedings of the 33rd International Symposium on Automation and Robotics in Construction. Auburn, AL, USA, July 21, 2016.

8. Dragos, K. & Smarsly, K. (2016). Distributed adaptive diagnosis of sensor faults using structural response data. Smart Materials and Structures, 25(10), 105019.

9. Fitz, T., Theiler, M., & Smarsly, K. (2019). A metamodel for cyber-physical systems. Ad-vanced Engineering Informatics, 41(2019), 100930.

10. Hartmann, D., Smarsly, K., Law, K. H. (2011). Coupling sensor-based structural health monitoring with finite element model updating for probabilistic lifetime estimation of wind energy converter structures. In: Proceedings of the 8th International Workshop on Struc-tural Health Monitoring. Stanford, CA, USA, September 13, 2011.

11. Lim, S., Buswell, R., Le, T., Austin, S., Gibb, A., & Thorpe, T. (2012). Developments in construction-scale additive manufacturing processes. Automation in Construction, 21(2012), 262-268.

12. Paul, S., Tay, Y. D., Panda, B., & Tan, M. (2018a). Fresh and hardened properties of 3D printable cementitious materials for buildings and construction. Archives of Civil and Me-chanical Engineering, 18(1), 311-319.

11

13. Paul, S., van Zijl, G., Tan, M., & Gibson, I. (2018b). A review of 3D concrete printing sys-tems and material properties: Current status and future research prospects. Rapid Prototyp-ing Journal, 24(4), 784-798.

14. Peralta, P., Mthunzi, E. M., Heine, S., Ludwig, H.-M. & Smarsly, K., 2019. A metamodel for 3D concrete printing. In: Proceedings of the Sixteenth International Conference on Civ-il, Structural & Environmental Engineering Computing. Riva del Garda, Italy, September 16, 2019.

15. Salet, T. A., Bos, F. P., Wolfs, R. J., & Ahmed, Z. Y. (2017). 3D concrete printing – a structural engineering perspective. In: Proceedings of the 2017 fib Symposium – High Tech Concrete: Where Technology and Engineering Meet. Maastricht, NL, June 12, 2017.

16. Tay, S. I., Lee, T.C., Hamid, N. Z. A., & Ahmad, A. N. A. (2018). An Overview of Indus-try 4.0: Definition, Components, and Government Initiatives. Journal of Advanced Re-search in Dynamical and Control Systems, 10(14), 1379-1387.

17. Theiler, M., & Smarsly, K. (2018). IFC Monitor – An IFC extension for modeling struc-tural health monitoring systems. Advanced Engineering Informatics, 37(2018), 54-65.

18. Wolfs, R. M., Bos, F. P., van Strien, E. F., & Salet, T. M. (2017). A real-time height meas-urement and feedback system for 3D concrete printing. In: Proceedings of the 2017 fib Symposium – High Tech Concrete: Where Technology and Engineering Meet. Maastricht, NL, June 12, 2017.