Verktyg för effektivare projektering av betongelement: Gapet mellan program och mötet mellan projektering och programmering
Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
This report mainly covers prefabricated concrete shear walls (precast). In an exploratory study it is investigated how a need to streamline the modeling of wall connections in analysis modelling can be solved with algorithms developed without conventional programming. The work is based on a descriptive feasibility study of the many steps a structural engineer goes through in their analysis process. The procedure covered is based on linear elastic global analysis based on shell theory and the finite element method (FEM). A method that implies an engineering-related idealization and simplification of a complicated real behavior. Reinforced concrete complex behavior is largely derived by how the material cracks under load. The material acts non-linear. Linear elastic analysis is depict controversial in some aspect but generally argued to be project financially feasible. The simplified approach is described in relation to other methods such as the truss-based Strut-and-Tie method and Non-Linear analysis.The work is financed by the author’s employer Sigma Civil AB. Structural departments’ goal is to complete the analysis modelling during tender. The need is derived from the company’s objective. The tendering stage is time limiting, especially for consulting services in the precast industry. Precast structural elements are manufactured off site. Its biggest difference from site casted concrete is the connections weakening effect between elements. The division and connections between elements largely affect the structural behavior. Effects that need to be represented in global analysis. The modelling process of discrete connections in global analysis are however time-consuming. A need to streamline the process exists.A more general purpose is stated regarding the possibilities for structural engineers in consultancy firms to self-develop automated tools. It is noted that similar developments often tend to require programming skills of the engineer, something that isn’t motivated. The CAD- and IT-environment studied in this report allegedly won’t require such skills. In part the aim of the study is to investigate the claim or how structural design and programming would meet.A literature study was conducted covering relevant research by the institution of civil engineering at Lulea Technical University. The study focused primarily on research regarding industrial construction process and development of building systems, launched as platforms. The result indicates that the fully automated platform is viewed as an end goal.In the striking of a note the works scientific basis is described and how it differs from what has already been done. With lessons learned; previous researcher’s results is applied to other areas in the field of structural engineering and another CAD- and IT-environment is used. However, based on a stricter notion, the author considers development of fully automated platforms not suitable for general structural engineering consultancy.While the aim consists of investigating the possibilities for automated tools. The critic of method is based on the fact that structural engineering includes sub-steps that require engineering-related interpretation. The author’s opinion emphases on the boundaries between the structural designer and tool. Tools are developed to automate sub-steps. Conventional software isn’t fully replaced by automated platforms but is rather used alongside smarter tools. Which implies that conventional software is still used. Its communication and the ability to influence this, both outwards and inwards is given. Something that is commonly referred to a as the gap between software.The descriptive feasibility study results in a professionally described checklist-like basis for the analysis process advocated. Important paragraphs from current regulation are cited and interpretations are distinguished. The procedure is divided into steps:Boundary, Idealization, Classification, Structural analysis, Cracking, Second order theory, Design, Local analysis.In addition to the analysis process the work results in algorithms that connect more conventional software to Grasshopper™. Algorithms that form the basis for structural engineers to develop smarter tools. The exchange connections have since been exploratory utilized in the development of smarter tools.What sets Grasshopper™ apart from similar platforms is an extension that allows the use of optimization theory. The optimization theory applied is emphasized as general but slow compared to other theories. Two tools that use the theory have been developed. Optimized Crane Position and Optimized Wall Interface Position. The intention is to streamline the precast structural engineer’s way of work.Optimized Wall Interface Position governs the division of precast concrete shear wall elements. The algorithm’s aim is to minimize the time difference between analysis modelling of a site-casted and precast concrete building. The first step in this process is locating the interface between wall elements.The optimization theory being used by extensions are introduced in a chapter on the CAD- and IT-environment. The software developer’s hope is described as an effort to create a platform for a wider use of evolutionary algorithms for non-programmers. Evolutionary optimization theory and genetic algorithms is depict from a peak-valley problematic. The application is inspired by biological evolution. Genetic algorithms are described as a parable for Genetic programming that’s primarily addressed in context of artificial intelligence. The similarity between them is both their ability to solve problems, which is calculated using a fitness-function. The division of wall elements in the algorithm Optimized Wall Interface Position is based on a fitness-function of least cost for a complete system. Algorithm details are non-disclosure and the report focuses on what drives the cost and optimization of the system. The cost is based on the reinforcement and the work effort with regards to design, production, transport and assembly; and an approximated exterior loading. The cost is however primarily controlled by the number of elements produces by the element division. How cost factors can be adjusted for each project but examples are given in the reports appendices. In the appendices derivations of mean values are presented based on governmental statistical record and public annual reports.The development intention is to streamline the design process in a regulatory manner. Experiential reasons have led general development of tools that can be applied to all types of walls. An underlying idea driving the development is to incorporate more features. Future features will be added and result in a more complete automated tool. The tool is described as inefficiently developed for its current singular purpose.Continuous evaluation with performance-based analyzes governs the development of Optimized Wall Interface Position. A mean value is determined from five time-lapse trials where the tool solves two different scenarios. The scenarios describe trivial cases for an experienced precast structural engineer. The algorithm does not differentiate from what a designer considers to be difficult, it attacks all problems similar. The two cases being tested contains of two short wall lines with different wall openings and two long straight wall lines requiring exactly 20 wall elements. A precast engineer would solve both cases monotonous without further reflection, limited only by the six mouse clicks each division requires. Three versions of Optimized Wall Interface Position have been analyzed. At first being unjustified; to be modified. More information was incorporated outside the optimization. The solution space was reduced in ways that makes the tool less general but still including appropriate wall divisions. Changes resulting in a feasible tool. In a third version, the optimization found an appropriate wall division in a couple of seconds for all trials. However, to succeed with streamlining the tool and obtain a more automatic work process; conventional programming was applied in sub-tasks. A combined work process between the structural engineer and the programmer is proposed. The structural engineer lays the foundation, the programmer acting support and becomes more involved when tools are automated and implemented.The discussion begins with a wider view on society and the benefits of efficiency and automation. Industrialization of the service sector, fewer can now do more and jobless growth occurs. A recent study is related: more than 53 % of today’s jobs will be automated and gone in 20 years. How the structure of society handles the transition and education in adulthood is discussed. Parallels and risks from other industries are introduced. An example of high-frequency trading with financial instruments is given. A risky business that uses algorithms based on the same evolutionary optimization theory. Intelligence Augmented is covered in relation to Artificial Intelligence. Intelligence Augmented is about strengthening the human operator instead of replacing it. The focus then lands on the boundary between man and machine. When algorithms is used in symbiosis with the operator smart tools are described instead of automated processes. If the operator, the structural engineer, is introduced in the development the transition can also be made. The discussion continues, more specified to the reports result. It is proclaimed that smarter tools that streamline design can be developed by structural engineers with Grasshopper™. The appropriation is built on the principle of information exchange in the gap between more conventional software. Suitable boundaries between tool and structural engineer are important to establish. Especially for calculations with simplified methods that require much interpretation. Grasshopper™ can be used by structural engineers to assess, add or manipulate parametric information between software. By bridging the gap between these, great potential is given for the development of smarter tools.The discussion turns to continued research where it is mentioned that more features can be incorporated in the algorithm for Optimized Wall Interface Position. The step isn’t far from allowing the algorithm to model wall connections. It is also possible for the tool to interpret the analysis result. With such a development approximated external loading wouldn’t need to be considered, actual loading could. If so, the algorithm gets more general and the possibilities for a more holistic automated tool appears. Finally, Grasshopper™ is described as a visual node-based algorithm editor with the potential to change the structural engineer’s way of work. The intuitive visual work methodology in Grasshopper™ is argued being greatly beneficiary to the user’s learning curve. The threshold knowledge that usually is required in programming before results could be presented is drastically reduced. The complete work bears witness to this.Structural engineers can use Grasshopper™ in a lot of ways. Between global analysis, local analysis and design. To exchange data between popular, in some ways similar, tools like Excel™ or MathCAD™. Information could be prepared and collated to the structural report, for documentation and quality control.At last automation is achieved with optimization theory.
Place, publisher, year, edition, pages
2015. , 75 p.
Teknik, Förtillverkade element, Prefab, Grasshopper, parametrisk design, plattform, industriellt byggande, Betong, linjär analys, global analys, bärverksanalys, algoritmer, finita element metoden, fackverksmetoden, icke-linjär analys, FEM, FEA, väggskarv, väggkoppling, indelning av element, väggelement, stabilisering, shear wall, SS-EN 1992-1-1, konstruktion, smarta verktyg, artificiell intelligens, augmenterad intelligens, system, projektering, beräkning, skalteori, sandwichmodell, plan spänning, plant spänningstillstånd, topologioptimering, evolutionär optimeringsteori, genetiska algoritmer, Galapagos, Tekla, Revit, Strusoft FEM-Design, Model Code, Eurokod, EKS, automatisering, Strut-and-Tie, Betongprefab
IdentifiersURN: urn:nbn:se:ltu:diva-46208Local ID: 3da37018-f1cc-442e-8316-0e5780c33ea6OAI: oai:DiVA.org:ltu-46208DiVA: diva2:1019521
Subject / course
Student thesis, at least 30 credits
Architectural Engineering, master's level
Validerat; 20150524 (global_studentproject_submitter)2016-10-042016-10-04Bibliographically approved