Motivation. Design of complex systems such as aircraft involves coordination and integration of contributions from multiple work packages (WPs) and suppliers. WPs interact because of interfaces and dependencies in the design, flows of deliverables in the design process, and flows of design change during a program. Such interactions consume management attention and can have time, effort and quality implications.
This paper considers how the effects of interactions in a complex program can be mitigated, for instance by:
1. Adjusting the timing of WPs, phases and suppliers, to minimize the rework caused when preliminary information is released, then updated;
2. Reducing the time taken to review and approve each design change, to avoid continuation of work that will be redone later;
3. Adjusting the assignment of resource across WPs, to avoid bottlenecks caused by unplanned rework and coordination effort.
Such actions seem qualitatively reasonable, but in practice they can sometimes be difficult to evaluate and compare because of the many variables and interactions that determine their impacts. A quantitative approach, grounded in statistics from similar programs, would help to assess improvement scenarios.
Method. A computer model was developed to predict patterns of workload in a complex program and thereby compute the impact of different scenarios on overall time, effort and quality. The model integrates for the first time three issues that have been frequently discussed in the research journals: (1) the initiation, review and approval of design changes; (2) propagation of design changes across WPs; (3) the impact of rework on overall effort and schedule, considering the timing of WPs and the information dependencies between them. It is implemented in an interactive dashboard tool that allows program scenarios to be compared and visualized.
Findings. Due to commercial sensitivity, data and detailed numerical predictions will not be presented. However, the types of information that can be extracted from the model will be indicated. Generic insights gained from the model will also be discussed, including: (1) the flow of change among WPs does not change substantially throughout a program, implying it is largely determined by early architectural decisions; (2) trying to keep a program on track by prioritizing planned work over unplanned rework might result in worse quality, effort and time overall; (3) adjusting the relative timing of work packages may reduce rework in some individual WPs, but the model indicates it might not significantly impact the time, effort and quality of a program overall; (4) the management levers studied can interact with confounding effects, and therefore should be considered in combination.
The authors’ experience suggests the basic structure of the model should apply to any complex development program, thus the insights are expected to be of general interest.
