4D Design and Simulation Technologies and Process Design Patterns

TSINGHUA SCIENCE AND TECHNOLOGYISSN 1007-0214 29/67 pp179-184Volume13,Number S1,October20084D Design and Simulation Technologies and Process Design Patterns to Support Lean Construction MethodsManfred Breit, Manfred Vogel, Fritz Häubi**, Fabian Märki, Micheal Raps i4Ds, University of Applied Sciences Northwestern Switzerland, Steinackerstrasse 5, CH-5210 Windisch, Switzerland Abstract: The objective of this ongoing joint research program is to determine how 3D/4D modeling, simula-tion and visualization of Products (buildings), Organizations and Processes (POP) can support lean con-struction. Initial findings suggest that Process Design Pattern may have the potential to intuitively supportICT based lean construction. We initiated a “Process Archeology” in order to reveal the requirements fortools that can support the planning, simulation and control of lean construction methods. First findings showthat existing tools provide only limited support and therefore, we started to develop new methodologies andtechnologies to overcome these shortcomings. Through the introduction of Process Design Patterns, we in-tent to establish process thinking in the interdisciplinary POP design. Optimized construction processes maybe synthesized with semi-automatic methods by applying Process Design Patterns on building structures.By providing process templates that integrate problem solution and expert knowledge, Process Design Pat-terns may have the potential to ensure high quality process models.Key words:4D design; process and organization modeling and simulation; process design patterns;lean construction methodsIntroductionUnder the title “Process Archeology” a multi-disciplinary team reconstructs and recreates the over-all building processes to study the disciplinary and col-laboration requirements, derive possible process pat-terns, and create alternative lean construction processes. The technology as well as the generated processes are repeatedly discussed, reflected and evaluated with former project managers and stakeholders. The work so far initiated a new methodology for the ICT sup-ported lean construction which we named Process De-sign Patterns. Process Design Patterns are a formal de-scription of successful solutions to problems. 1 Process Archaeology1.1 Case study and approachWe chose a recently finished four storey residential concrete building for the case study. The aim is to re-construct and recreate the overall building processes with a multi-disciplinary team of two architects, a structural engineer, a construction foreman, three proc-ess managers and a computer scientist. The work is based on 2D architectural drawings and detailed con-struction plans, continuous series of photographs which document the building processes over time, process log books, meeting minutes and work reports. The team creates the necessary 3D-, 4D- and process and organization models with commercial available modeling tools. This approach uses the proposed inte-grated process in Ref. [1] for the efficient use ofReceived:2008-05-30** To whom correspondence should be addressed.E-mail: fritz.haeubi@fhnw.chTsinghua Science and Technology , October 2008, 13(S1): 179-184180 3D/4D models to support the knowledge communica-tion and generation needed during constructability re-views. We used Archicad from Graphisoft and Revit from Autodesk for 3D modeling, Common Point Pro-ject 4D from Common Point Inc and Navisworks from Autodesk for 4D modeling and SimVision from ePM for the process and organization modeling and simula-tion. It is planned to reiterate the 4D- modeling process with recently emerged 4D- software such as the Virtual Construction Software Suite from Vico Software and C3D Framework and Visual Controls from Consoli-dated Contractors Technologies [2]. These applications use advanced modeling technologies and deploy exist-ing data of 3D building models and other sources, such as quantities, cost and performance information to far greater extent. Figure 1 shows a perspector view of the workflow [3].Fig. 1 Workflow to generate a 4D modelThe objectives of our case study are the following: First, determine the necessary level of detail and the granularity of the planning processes. Second, under-stand the inter-disciplinary planning process from the different disciplinary views and objectives. Thirdly, derive and categorize possible Process Design Patterns of recurring tasks. And finally, how to apply lean con-struction principles like continuous flow, tact phases and pull-principle to construction processes and how this can be integrated in a process optimization tool. Building appropriate 3D and 4D POP models serves as first feedback to the constructability of the tasks and the phases. At the same time, benefits and restrictions of the used software applications as well as the neces-sary workflow will be studied. Special attention is paid to the usability, e.g. where does the current deployment of applications differ from “natural” planning proc-esses. The generated (leaner) processes and modelsimulations and predictions are compared with the ac-tual carried out processes and discussed by and with former project managers and stakeholders, who built the project.1.2 Initial findingsx To support design and process decisions, the 3D models need to have a sufficient level of detail to support the granularity of the building processes. As the planning process typically is gaining more precision and granularity through the transforma-tion from rough to detailed design, this require-ment will bring in new challenges to the practice. x Even object oriented 3D CAD applications often lack the necessary process-oriented perspective, i.e. they are designed primarily for producing plans not buildings. Nonetheless, it is already pos-sible to set up walls consisting of different com-ponents like structural wall and outer thermal in-sulation. However, the interoperability with plan-ning software is not ensured because outside of the 3D CAD application such a component is treated as a single geometric object.x The current work-flow of unintegrated 3D/4D modeling software typically needs information ex- and import. We used the IFCs [4] model infor-mation exchange. We observed a multitude of conversion errors reaching from 3D geometry misrepresentation to partial and complete infor-mation loss. The “fine” tuning of the ex- and im-port filters is currently an art and strains the pa-tience of the user. Without extra support from the software providers these try and error attempts of-ten cannot be finished successfully. It was neces-sary to write additional small applications (filters) to correct information exchange errors (see Fig.1). x The handling of different design alternatives in a project with currently available software tools lead to a multitude of unrelated data which will quickly become unmanageable. The strict se-quence of 3D/4D information exchange procedure often distracts user from their intended tasks e.g. the creation of a new design variant might be in-terrupted by an omitted exchange step or an im-port of the wrong model. Linking POP informa-tion in 4D models still includes too much recur-ring manual user interactions. Although theManfred Breit et alġ4D Design and Simulation Technologies and Process Design Patterns to Support (181)greater part of this work can be automated with object and process queries and link propagation the given 3D object hierarchy, missing attribute information or simply inconsistent naming con-ventions require manual user intervention.x Currently available 3D and 4D software do not support the rapid modeling of design alternativesin all three POP dimensions. The modeling proc-ess is time consuming and intuitive tools which support creative design processes are missing.x The latter two points lead to the requirement of amodel server which can handle different model representations for building elements (e.g. for Virtual Environments) as well as a branching mechanism for design variants and their revisionsover time. The integration of multiple product models has been addressed in Refs. [3] and [5] aswell.2 Process Design Patterns2.1 Definition of process design patternsOne of the goals in the project described above is to improve the design of construction processes by aug-menting measurability and reliability. We approached this by using patterns for process analyses and process design and by applying advanced optimization meth-ods for process optimization.The idea to use patterns as formal descriptions of generic solutions to classes of commonly reoccurring problems was first introduced by the architect Christo-pher Alexander[6] and has been adapted for various other disciplines[7-10]. A pattern records the design de-cisions taken by many builders in many places over many years to resolve a particular problem. Patterns may be collected together into a pattern language that addresses a particular domain[6].While the pattern language idea so far had limited impact in the building industry, it has had a profound influence on software engineering[7-9]. In software en-gineering, a design pattern is a general repeatable solu-tion to a commonly occurring problem in software de-sign. A design pattern is not a finished design that can be transformed directly into code. It is a description or template for how to solve a problem that can be used in many different situations. Object-oriented design pat-terns typically show relationships and interactions between classes or objects, without specifying the final application classes or objects that are involved.In construction and process planning, we use Proc-ess Design Patterns to formally describe actions or events leading the construction or process from one de-fined stage to the next. Preliminary studies, performed in a Swiss construction company in 2006 proved that even complex business processes could be described by a very limited set of process elements[10]. In con-trary to the company’s expectations, most daily opera-tions are performed in more or less equivalent steps, using the same sets of resources.This triggered the idea to use the pattern approach for describing business processes. Typically, such pat-terns are implemented on homogeneous tasks, perfor-med by specific taskforces using a given set of resour-ces, e.g. machines. Thus the scope of change caused by a certain pattern varies with the level of detail in the planning process. The concept of pattern based process design is therefore not limited to a certain stage or scale in the overall planning process.Figure 2 shows a possible generic Process Design Pattern (PDP) applicable to most construction proc-esses. Derived from the single generic top level pattern we recognized seven second level patterns as shown in Fig. 3: ground preparation, forming elements by pour-ing, forming elements by assembling small unified components, assembling site or project specific prefab-ricated elements, assembling adjustable standardized elements or mechanical components, surface treatment and setting into operation.Fig. 2 Generic process design patternThe goal of this step in our project is to create a comprehensive system for the description and defini-tion of construction processes, where the level of detailTsinghua Science and Technology , October 2008, 13(S1): 179-184182Fig. 3 Types of second level Process Design Patterns, examples for the definition of third level patterns and one example of a combined fourth level patternin the description can be consistently varied up- and downwards according the questions related to the ac-tual planning phase.2.2 Lean construction approachLean Management is a set of methods successfully ap-plied in many industries for augmenting process effi-ciency. Since the overall productivity in the construc-tion industry is rather poor [11], lean methods are being introduced under the name of lean construction. One of the key success factors in lean construction is the de-tailed planning and control of onsite activities and lo-gistic processes. The goal is to eliminate re-work, pro-vide continuous work flow and minimize mistakes. 2.3 Synthesizing processes by the use of patterns Patterns define the rules for describing the transition process from one given building state to the next. The overall process evolves in two steps from the process elements defined by the building structure and the ap-plication of patterns.The first step produces a number of distinct process elements or sub processes, generated by applying PDPs on the building structure. The PDPs determine the type and order of actions or tasks to be performed and the Building Information Model (BIM) defines the quantities, shape and restrictions for these actions. Given building structures do not necessarily lead to single process elements. A concrete structure e.g. maybe erected from prefabricated elements or by casting on site. In such cases, the application of PDPs on BIMs creates process elements in two or more variations mu-tually excluding each other. For the purpose of synthe-sizing a construction process, the process elements or sub-processes generated from the BIM are further de-composed into phases, i.e. elementary work units which can be performed in one single step.The overall process setup finally emerges from the integration of the sub processes. Here again, the BIM defines the overall quantities and constraints for the process design and the patterns define the rules for the possible combination of sub processes. Phases or ele-mentary work units described above provide the met-rics for process evaluation. Using very basic process elements for gathering metrics has the advantage that this information may easily be gathered from the analysis of existing processes or from industry bench-marks.The final design of the resulting overall process is done by advanced optimization techniques [12,13] in three steps. First the necessary performance indicators for the process optimization are defined in the behavior section of the underlying POP-Model.Then the process elements gathered from the build-ing structure are brought into a primary sequence, based on geometric constraints from the BIM. This preliminary process candidate does not have to be cor-rect in the terms of feasibility.In a final step, optimized processes candidates areManfred Breit et al ġ4D Design and Simulation Technologies and Process Design Patterns to Support (183)generated by using genetic algorithms [12,13]. This final step takes process constraints like resource usage or process behaviors into consideration. This simulation technique allows producing and evaluating many dif-ferent instances of the target process by variation and recombination of sub processes.Figure 4 shows the basic proposed algorithm used forsynthesizing construction processes from given build-ing structures by the use of Process Design Patterns.2.4 Exemplary step from building components tobuilding process models To decide whether a PDP can be assigned to building components depends on the properties of these compo-nents. Such properties can be the type of the building component, its material, its location in the structure, its structural properties, etc.The PDP provides the criteria which define whether it can be assigned to a specific building component. Based on these properties, the PDP assignment algo-rithm builds a Property Breakdown Structure (PBS) of the components. This PBS specifies those buildingcomponents to which the PDP can be assigned.Fig. 4 From an architectural 3D model to a lean construction processesIt is possible to define several PDPs which can be used to do the same sub-process. An example might be in place assembling and installation versus pre-fabrication installation. Consequently, one needs to de-cide which PDP should be used in order to perform the sub-process.Similarly, if a subgroup of geometric components assigned to a specific Process Design Pattern need a special treatment, a more specialized Process Design Pattern overwrites a more general PDP (see Fig. 5).Fig. 5 Specialized Process Design Patterns can overwrite more general Process Design PatternsAssigning Process Design Patterns to each buildingcomponent separately enables a very detailed processplanning. At the same time, it is also possible to plan aproject on a higher level. This can be done by composing building components which have the same PDP assigned into groups. A possibility of such a grouping could be a group of building components which are of the same type and are potentiallyTsinghua Science and Technology, October 2008, 13(S1): 179-184 184performable in one process step.3 ConclusionsThe aim of the presented research is to determine how pattern based process design may successfully support lean construction methods. Therefore, we initiated a “Process Archeology” in order to reveal the require-ments for tools that can support lean construction. First findings show that existing tools provide only limited support. To overcome these shortcomings, we started to develop new methodologies and technologies. Through the introduction of Process Design Patterns, we propose a methodology to synthesize optimized construction processes from architectural models. By providing process templates that integrate problem so-lution and expert knowledge, Process Design Patterns may have the potential to generate construction proc-esses meeting defined optimization criteria based on relevant performance indicators. We are convinced that the proposed approach can be deployed widely in the AEC industry and that the findings of the case study have a more general character: e.g. industrial facility projects often have reocurring processes with pattern characteristics. Actually, we successfully applied the Process Design Patterns to the construction of new prototypes of waiting room facilities for the Swiss Railway Corporation.References[1]Hartmann T, Fischer M. Supporting the constructabilityreview with 3D/4D models. Building Research & Informa-tion, 2007, 35(1): 70-80.[2]Jreije R. The promise of visual controls: Introducing C3D.In: CCC Bulletin 4th Quarter. 2007.[3]Kiviniemi A, Fischer M, Bazjanac V. Integration of multi-ple product models: IFC servers as a potential solution. In:Conference on Information Technology in Construction.Dresden, 2005.[4]Haymaker J, Suter B, Fischer M, et al. The perspective ap-proach: Enabling engineers to construct and integrate views and generate an evolving project model. WorkingPaper Nr 081, Center for Integrated Facility Engineering,Stanford University, 2004.[5]IAI: International Alliance of Interoperability. http://www./, 2008.[6]Alexander C, Al E T. A Pattern Language. England: Ox-ford University Press, 1977.[7]Gamma E, Helm R, Johnson R, et al. Design Patterns. Bos-ton, USA: Addison-Wesley, 1995.[8]Coplien Jo, Schmidt Dc. Pattern Language of Program De-sign. Boston, USA: Addison-Wesley,1995.[9]Gabriel R. Patterns of Software. England: Oxford Univer-sity Press, 1996.[10]Häubi F, Righeti M. Analyse der Geschäftsprozesse einermittelständischen Schweizerrischen Bauunternehmung.Institute for 4D Technologies, unpublished, 2005.[11]Teichholz P. Labor Productivity Declines in the Construc-tion Industry: Causes and Remedies, AECbytes Viewpoint#4, April 14. /viewpoint/2004/is- sue_4.html, 2008.[12]Märki F, Vogel M, Breit M, et al. Interactive toolbox for4D modeling. In: Proceedings of the 11th International Conference on Computing and Decision Making in Civiland Building Engineering (ICCCBEXI). Montréal, Canada,2006.[13]Märki F, Fischer M, Kunz J, et al. Decision making forschedule optimization. Department of Civil Engineering, Stanford University. /online.publi- cations/TR169.pdf, 2007.。