Coping with Complexity

: Approaches, Tools and Frameworks for Systems Engineering

Abstract

How can system engineers deal with the increasing complexity of the systems that they strive to engineer? In this thesis we explore approaches, tools and frameworks for managing the challenges presented by system complexity.

First, the application of network science methods to the analysis of two real‐world enterprise architectures is explored to assess the relative importance of different architecture components and vulnerability to vertex removal and failure cascades. Several significant challenges are identified and are summarised as five guiding principles for utilising network science concepts for architecture evaluation.

Given these challenges, a systems engineer may ask: How useful a lens is a complexity perspective when evaluating candidate systems? The likely merits, and challenges, in evaluating system complexity are analysed using an industrial decision-support tool as a case study, supported by 16 semi-structure qualitative interviews with users. This analysis is bolstered by a survey of 55 systems engineers which reveals competing views on the relative and absolute importance of terms relating to system complexity.

To capitalise on these findings, a complexity evaluation framework is derived; a structured approach to complexity
evaluation which emphasises; collaboration, shared understanding of contextually relevant complexity factors, iterative complexity (re-)evaluation, and progressive refinement of tools and processes through linking these to project outcomes in the form of a wider organisational learning cycle. This framework is mobilised as a Complexity Register, a novel decision-support tool. Preliminary validation of the tool suggests that it delivers an ability to structure collaborative discussions of the potential impact, and mitigations, of various sources of system complexity.

The thesis concludes with a discussion of the challenges that remain for organisations in assuring that the framework is being implemented effectively, and, relatedly, in collecting sufficient empirical evidence to conduct cost-benefit analyses of the impact of system complexity evaluation.

Date of Award	25 Jan 2022
Original language	English
Awarding Institution	University of Bristol
Supervisor	Seth Bullock (Supervisor)