Designing Systems Visualisations for Decision Support

Agzam Idrissov

Research output: Book/ReportPh.D. thesis

126 Downloads (Pure)

Abstract

Visual representations are integral to designing. The ever increasing digitalisation of design processes has led to the wide use of systems visualisations. Systems visualisations are abstract, computer-supported, interactive and data-rich visual representations that display product data across multiple design domains, such as requirements, product architecture or verification tests. They combine visual principles, such as collapsible tree diagrams or parallel coordinates plots and visual techniques, such as filtering or highlighting to facilitate decision making for designers by conveniently representing large amounts of heterogeneous and dynamic product data.
There is little research in engineering design on how to build systems visualisations. Such lack of knowledge about how to build systems visualisations can lead to ineffective systems visualisations. This prevents designers from making sense of often complex product data and affects the quality of subsequent design decisions.
With the overall aim to inform the building of visualisations to be used in designing, in this Ph.D. thesis, I formulate and address three research questions. The first research question (RQ1) asks: “What are the challenges encountered by designers when using systems visualisations in practice?”. To tackle RQ1, through semi-structured interviews and document analysis in three large manufacturing companies, the Visual Challenges paper (A) shows that a variety of systems visualisations, such as hierarchical linked lists, adjacency matrices, or collapsible tree diagrams are used throughout various design activities, including product configuration and requirement verification. The paper illustrates that the use of such systems visualisations presents multiple challenges to designers, such as the absence of interactive visual techniques, inefficient visualisation layout, or inadequacy with respect to the data represented. The findings suggest that a more nuanced understanding of visual principles and techniques is required to produce systems visualisations in accordance with designers’ needs.
The second research question (RQ2) inquiries: “Which aspects have to be considered when building systems visualisations?”. In response to RQ2, this thesis argues that these three aspects are system features of represented product data, visualisation tasks performed by designers when using systems visualisations, and design activities in which these visualisations are used. While understanding design activities helps to elicit knowledge and information needs of designers, considering system features and visualisation tasks help to select appropriate visual principles and techniques when building systems visualisations.
Each of these three aspects are inspected via four research papers. First, the LinkedTree Visualisation paper (B) takes an exploratory approach, and after conducting a literature review, it proposes a novel systems visualisation, a linked tree, to support traceability between multiple design domains, such as stakeholders, requirements, product architecture, and verification tests. Taking a vacuum cleaning robot as an example, the paper illustrates how hierarchical data about design domains and relationships between respective elements can be represented using systems visualisations. Then, the Validating System Features paper (C) explores system features, such as hierarchies, relationships, patterns, and processes, to understand the nature of the product data represented. The results suggest that systems visualisations are recognised as representing the aforementioned system features, even by study participants without prior familiarity with the visualisations used. Further, the Mapping System Features paper (D) continues this line of thinking by mapping system features to different types of design activities. By reviewing the Design and Information Visualisation literature, the paper connects system features and visual principles using Gestalt Principles. Further, the paper identifies design activities that are supported by different systems visualisations and are built on various visual principles, such as network diagrams or adjacency matrices. Such a mapping allows the selection of suitable visual principles based on the system features of the product data to be represented. The Visualisation Tasks paper (E) uses a typology of visualisation tasks from the field of Information Visualisation to describe visualisation tasks performed by designers when using systems visualisations. Using semi-structured interviews of six product development students and using document analysis, the visual representations used during the design of an autonomous robot were analysed. Then, using the typology of visualisation tasks, a system visualisation was built to support the designers. The study showed the utility of analysing visualisation tasks to inform the selection and adjustment of visual principles and techniques within systems visualisations.
The third research question (RQ3) asks: “How to inform the building of systems visualisations?”. To address this question, the Conceptual Framework paper (F) brings together the three aspects identified in RQ2, namely, system features, visualisation tasks, and design activities, to propose a conceptual framework to inform the building of systems visualisations. By conducting semi-structured interviews with practitioners and document analysis in a manufacturing company, the currently used visual representation was modified into a systems visualisation using the proposed conceptual framework. The study illustrated the application of the proposed conceptual framework to inform the building of systems visualisations.
The core contributions of the thesis have been to bring insights from the field of Information Visualisation into engineering design research, to understand challenges encountered by designers in practice when using systems visualisations, and to propose a conceptual framework that describes the interconnections between system features, visualisation tasks, and design activities, ultimately to suggest a systematic approach to building systems visualisations. The conceptual framework has also been pilot applied outside engineering design.
Original languageEnglish
PublisherTechnical University of Denmark
Number of pages222
Publication statusPublished - 2021

Fingerprint

Dive into the research topics of 'Designing Systems Visualisations for Decision Support'. Together they form a unique fingerprint.

Cite this