Design Iteration Control Framework for Offsite Building Projects

Sammanfattning: During the course of the studies this thesis is based upon a design iteration control framework for offsite building projects was conceptualized. The ultimate goal was to develop a robust tool to assist project managers to find the optimal balance between maintaining sufficient project-by-project iteration to solve complex design problems, and avoiding excessive iteration, which unnecessarily complicates design and scheduling. In total three framework components were developed. The thesis describes the problem addressed, presents background information, describes the development of the framework components, and discusses their potential utility in construction contexts. Offsite building has been advocated as an effective means to increase product quality while reducing project duration and cost, provided the design process is efficient. A main challenge in managing the design process is iteration. It must be possible to alter details in order to react to changes in conditions and meet project-specific requirements, but unplanned design iteration should be avoided because it can lead to departures from planned activity sequences, thereby increasing both scheduling and design complexity. In project management literature, two groups of approaches (system dynamic and model-based) for managing design have been established. The first group is used to identify factors that affect design iteration and thus actions that could improve the process, while model-based approaches are used to investigate and predict possible effects of specific design iterations on project outcome. A problem is that current methods do not support attempts to quantify effects of specific improvement actions on project outcome, which could greatly facilitate effective management of resource-constrained projects. Due to the notion that all building projects are unique (which is prevalent in construction management literature) and the complex structure of design processes, it is generally difficult to identify and map iteration phases, cycles or loops in terms of specific actions. However, in offsite building projects it is reasonable to assume that the same design activities have to be carried out in all projects (regardless of the variation in their conditions), albeit to varying extents and, furthermore, that the precedence relations between activities is invariant. Thus, in this thesis (and the underlying studies) offsite building design projects are considered as realizations of a process that is characterized by varying activity extent but invariant precedence relations. The suggested framework is a model-based continuous improvement approach (plan-do-check-act cycle). Essentially it consists of cycles of observing effects of applied actions over the course of several projects, drawing inferences about the effectiveness of the actions from the observations, identifying improvements, applying improved actions and observing their effects. The framework incorporates techniques such as design structure matrix (DSM)-based simulation and Monte-Carlo inverse analysis. It comprises a method to calibrate DSM-based simulation models, a relative measure of design iteration, and a method to identify the most critical process phases (in terms of design iteration). The framework is mainly based on data related to the design process of a two-storey offsite timber frame building, supported by probability density functions for 35 other offsite timber frame multi-storey building projects. The practical applicability of the framework components has been tested in simulation experiments where they were applied to assess design processes related to this type of project, with variations in key conditions. To a lesser degree the framework was also applied to a planning and building approval process. The results of the simulations indicate that the components have high potential practical applicability, provided accurate records of activity execution sequences and corresponding work amounts are available.