| 1. |
- Vidner, Olle, 1990-
(författare)
-
On Multi-Disciplinary Optimization in Engineer-to-Order Product Configuration
- 2023
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Customized products are becoming increasingly common, and increasingly important for maintaining a competitive advantage in certain industries. Being able to quickly and accurately respond to unique customer requirements can provide a competitive edge or even be the only path to survival. In practice, configurators are commonly used to manage the customization process, gathering the customer’s requirements and suggesting feasible solutions to the customer’s problem.Fostering and maintaining a viable product customization offering is not easy. A particularly challenging category of products is one where an extensive engineering effort might be needed to even produce a reliable estimate of the product’s price. These products are usually referred to as engineer-to-order (ETO) products.Prior work has pointed out the potential of using optimization as part of configuration solutions for ETO products, but the literature is limited in its extent and does not clearly prescribe how to structure and approach such solutions.This thesis outlines a conceptual and technical architecture for implementing optimization-based configuration solutions. Reusable primitives for supporting the routines involved in this architecture are provided. These findings are verified through application and evaluation within two industrial case studies, also yielding important industrial needs to cover in the future research and development of the proposed framework. By examining three additional case studies, common issues in the development and deployment of design automation (DA) systems are identified.Successful implementation of the proposed framework for optimization-based configurators can lead to two main benefits. First, engineering configurator prototypes can be developed rapidly, to test the viability of configurator projects – a category of projects prone to expensive failures. Second, optimization-based configurators can be used to support rapid design space exploration in early product development stages, leading to enhanced product knowledge in a critical phase, and in turn, increased product value.
|
|
| 2. |
- Eek, Magnus, et al.
(författare)
-
A Framework for Early and Approximate Uncertainty Quantification of Large System Simulation Models
- 2015
-
Ingår i: Proceedings of the 56th Conference on Simulation and Modelling (SIMS 56), October, 7-9, 2015, Linköping University, Sweden. - Linköping : Linköping University Electronic Press. - 9789176859001 ; , s. 91-104
-
Konferensbidrag (refereegranskat)abstract
- Uncertainty Quantification (UQ) is vital to ensure credibility in simulation results and to justify model-based design decisions – especially in early development phases when system level measurement data for traditional model validation purposes are scarce. Central UQ challenges in industrial applications are computational cost and availability of information and resources for uncertainty characterization. In an attempt to meet these challenges, this paper proposes a framework for early and approximate UQ intended for large simulation models of dynamical systems. A Modelica simulation model of an aircraft environmental control system including a liquid cooling circuit is used to evaluate the industrial applicability of the proposed framework.
|
|
| 3. |
- Gustafsson, Erik, 1993-, et al.
(författare)
-
Comparison of Design Automation and Machine Learning algorithms for creation of easily modifiable splines
- 2020
-
Ingår i: Proceedings of NordDesign 2020, Lyngby, Denmark, 12th - 14th August 2020. - : The Design Society. - 9781912254088
-
Konferensbidrag (refereegranskat)abstract
- In order to enable easy modification of results from a design optimization process in a CAD tool, a flexible representation of the geometry is needed. This is not always trivial however, since many file formats are not importable as modifiable geometry into the CAD tool, and if they are, they might not represent the geometry in a way that enables easy modification. To mitigate this problem a design automation (DA) and a machine learning (ML) approach are developed and compared using a test case from an optimization process used to optimize hose routing in tight spaces. In the test case used, the geometry from the optimization process consists of center curves represented as a large number of points. To enable easy modification a more flexible representation is needed such as a spline with a few well-placed control points. Both the DA and ML approach can approximate center curves from the optimization process as splines containing a varying number of control points but do show different properties. The DA approach is considerably slower than the ML but adds a lot of flexibility regarding accuracy and the number of control points used.
|
|
| 4. |
- Gustafsson, Erik, 1993-
(författare)
-
Exploring Data-Driven Methods to Enhance Usability of Design Optimization
- 2022
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Developing high-performing products at a low cost while keeping development time down is increasingly important in today’s competitive market. The current state presents a need for efficient product development processes. One of the challenges is knowledge often being limited in early stages where the cost of making changes is still relatively low. As the process progresses more knowledge is gained to better support decisions; however the cost of making changes increases, limiting the design freedom. To increase knowledge while retaining design freedom, several computer-based tools are available to both generate and evaluate designs in order to make iterations faster and more accurate.Design Optimization (DO) can be utilized to explore the design space and find optimal designs. A Computer-Aided Design (CAD) model is often required as input to analysis tools evaluating the designs. By utilizing Design Automation (DA) several tasks involved in creation and modification of CAD models can be automated. For this reason, DA is sometimes considered an enabler for DO although its use is far wider, covering several aspects of the design process mainly focusing on automating repetitive and routine tasks.Machine Learning and other data-driven methods are becoming increasingly viable in the context of DO and DA. This thesis explores the use of data-driven methods to enhance the usability of DO in different ways such as a faster process, new use-cases, or a more integrated and automated process. Literature in the area is reviewed, identifying applications, trends and challenges. Furthermore, two support tools are developed, incorporating data-driven methods tied to an industrial case. The applications focus on parameterizing geometry and predicting design performance respectively. Potential benefits, limitations, and challenges are discussed based on the literature review and insights from the two support tools. The focus of the thesis is mainly on how data-driven methods can facilitate automation and integration in the design process, specifically for complex products requiring significant engineering efforts.
|
|
| 5. |
- Hallberg, Peter, 1974-
(författare)
-
On Knowledge Creation and Learning at the Intersection of Product Development and Engineering Education
- 2021
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Today’s demands on higher engineering education given the rapid transformation of society are, to say the least, multifaceted. Rapidly increased complexity of technology as well as adaptation to sustainability requirements are causing major transformations and mergers of whole domains of technology that strongly impact current and future engineering workforces within these domains, in particular, the need for new competencies. To adapt to industry demands for engineering competence that fits new constellations of technology domains, providers of advanced engineering training – i.e., engineering faculties – need to inventory their toolbox for ways to support knowledge creation processes. Product development theory is a central part of many types of academic engineering programs. However, as the product development process itself is a process of knowledge creation, it also has strong relations to theories of learning. This thesis explores the idea that some of the tools of engineering that are also taught at engineering faculties and therefore are familiar to their members can beneficially be applied to the development and management of engineering curricula. This thesis explores the domains of product development theory, engineering education and learning analytics in search of overlapping approaches to knowledge creation. The outcome of this search, which are also the result of this thesis, is a set of proposed tools, measures, and approaches for the development, management, content, and arrangement of engineering curriculum. The main contributions focus on the use of physical artifacts and their contribution to engineering educational frameworks, such as the Conceiving – Designing – Implementing – Operating (CDIO) initiative. For this purpose, the thesis picks up on a previously developed concept of low-cost demonstrators for the establishment of a formalized learning and enabling platform that promotes implementation and execution of the CDIO framework. Furthermore, by adopting a similar approach to product development and learning theory, additional tools are identified and developed for curriculum adaptation, enhancement, and management. In particular, by examining the curriculum from previously unexplored perspectives followed by establishment of novel measurements, this thesis demonstrates how curriculum developers and program managers can increase their insights into the outcomes of their decisions.
|
|
| 6. |
- Persson, Johan, et al.
(författare)
-
A Framework for Multidisciplinary Optimization ofa Balancing Mechanism for an Industrial Robot
- 2015
-
Ingår i: Journal of Robotics. - : Hindawi Publishing Corporation. - 1687-9600 .- 1687-9619. ; , s. 1-8
-
Tidskriftsartikel (övrigt vetenskapligt/konstnärligt)abstract
- The paper presents a framework that can be used to design and optimize a balancing mechanism for an industrial robot. The framework has the capability to optimize three different concepts - a mechanical, a pneumatic and a hydro-pneumatic. Several disciplines are included in the framework, such as dynamic and static analyses of the robot performance. Optimization is performed for each concept and the obtained optimal designs are all better then the reference design. This means that the framework can be used both as a tool to optimize the balancing mechanism and also to support concept selection.
|
|
| 7. |
- Persson, Johan, et al.
(författare)
-
A Modified Complex Algorithm Applied to Robust Design Optimization
- 2011
-
Ingår i: 13th AIAA Non-Deterministic Approaches Conference. - Reston, Virigina : American Institute of Aeronautics and Astronautics. - 9781600869518 ; , s. 2011-2095
-
Konferensbidrag (refereegranskat)abstract
- Today there is a desire to perform optimizations in order to receive optimal system properties. However, for computationally expensive simulation models, an optimization maybe too tedious to be motivated. This paper proposes a modification of the Complexoptimization algorithm to enable the creation and usage of local meta-models during theoptimization. Its performance is demonstrated for a few analytical problems and a reliabilitybased design optimization is conducted for an aircraft example.
|
|
| 8. |
- Persson, Johan, et al.
(författare)
-
Comparison of Different Uses of Metamodels for Robust Design Optimization
- 2013
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- This paper compares different approaches for using kriging metamodels for robust design optimization, with the aim of improving the knowledge of the performance of the approaches. A popular approach is to first fit a metamodel to the original model and then perform the robust design optimization on the metamodel. However, it is also possible to create metamodels during the optimization. Additionally, the metamodel need not necessarily reanimate the original model; it may also model the mean value, variance or the actual objective function. The comparisons are made with two analytical functions and a dynamic simulation model of an aircraft system as an engineering application. In the comparisons, it is seen that creating a global metamodel before the optimization begins slightly outperforms the other approaches that involve metamodels.
|
|
| 9. |
- Persson, Johan, et al.
(författare)
-
Comparison of Sampling Methods for a Dynamic Pressure Regulator
- 2011
-
Ingår i: 49th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition. - Reston, Virigina : AIAA American Institute of Aeronautics and Astronautics. - 9781600869501
-
Konferensbidrag (refereegranskat)abstract
- Concepts for complex products are often developed using computer models, introducinguncertainties both in design and model accuracy. There exist several methods forapproximating these uncertainties and this paper presents and compares some of them. Thefocus is on sampling based methods including or excluding response surfaces, and they arecompared by accuracy and computation time, using a Monte Carlo sampling as reference.The application is a simplified system model of a dynamic pressure regulator that controlsthe air supply in the environmental control system of an aircraft.
|
|
| 10. |
- Persson, Johan, et al.
(författare)
-
Comparisons of Different Methods for Robust Optimization in Engineering Design
-
Annan publikation (övrigt vetenskapligt/konstnärligt)abstract
- This paper compares the performance of five methods for robust design optimization of computationally demanding models including one novel method. The comparison is made using several mathematical functions and two engineering problems. The performance metrics are the mean value and standard deviation of the optimum as well as an index that weights together the required number of simulations of the original model and the chance of finding the optimum. The result of the comparison shows that sequential robust optimization is the most effective method.
|
|
