0
Research Papers

Predicting Change Propagation in Complex Design Workflows

[+] Author and Article Information
David C. Wynn

Department of Engineering,
University of Cambridge,
Trumpington Street,
Cambridge, Cambridgeshire CB2 1PZ, UK
e-mail: wynn@cantab.net

Nicholas H. M. Caldwell

Suffolk Business School,
University Campus Suffolk
Waterfront Building,
Neptune Quay, Ipswich IP4 1QJ, UK
e-mail: N.Caldwell@ucs.ac.uk

P. John Clarkson

Department of Engineering,
University of Cambridge,
Trumpington Street,
Cambridge, Cambridgeshire CB2 1PZ, UK
e-mail: pjc10@eng.cam.ac.uk

This elimination is the underlying reason for distinguishing planned from unplanned dependencies in the workflow notation. The effect is that a task is forced to wait if any currently pending change may propagate to it indirectly, but not if the propagation would only occur following an exception—for instance, if a gateway reveals the need for iteration. Distinguishing between planned and unplanned dependencies thereby resolves a potentially cyclic task network into an acyclic one, ensuring that tasks are executed in a sequence that guarantees no rework unless information flow risks are manifested. Thus, the planned/unplanned distinction fulfills a similar role to process architecture, i.e., sequence of tasks in a DSM prepared for simulation, in the model reported by Browning and Eppinger [20]. However, the approach in this article provides a more natural way to specify sequencing constraints on a graphical workflow comprising concurrent streams and explicit flow logic, such as that shown in Fig. 4.

In our case studies, the task duration could only be estimated based on experts' opinions. Because of this, the information that could be used to select and parameterize a probability distribution was limited. The normal distribution was believed to be a reasonable choice by the industry sponsors, because it is well understood by most engineers and allows expression of variability alongside the expected value of task duration. Different distributions can be used in the model by altering Eq. (4). Some experimentation with the triangular and uniform distributions indicated that the output of the model is not greatly influenced by the precise shape of the input distributions, as long as the center and spread are similar across the compared distributions.

1Corresponding author.

Contributed by the Design Theory and Methodology Committee of ASME for publication in the JOURNAL OF MECHANICAL DESIGN. Manuscript received September 9, 2013; final manuscript received March 29, 2014; published online June 2, 2014. Assoc. Editor: Irem Y. Tumer.

J. Mech. Des 136(8), 081009 (Jun 02, 2014) (13 pages) Paper No: MD-13-1402; doi: 10.1115/1.4027495 History: Received September 09, 2013; Revised March 29, 2014

A simulation model to help manage change propagation through design workflows is introduced. The model predicts resource requirements and schedule risk of a change process, accounting for concurrency, multiple sources of change, and iterations during redesign. Visualizations provide insight to answer common management questions. The approach is illustrated using an aircraft design workflow from Airbus and a more complex turbine disk design workflow from Rolls–Royce.

FIGURES IN THIS ARTICLE
<>
Copyright © 2014 by ASME
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Fig. 1

Example workflow network. Tasks are depicted as rectangles, deliverables as cylinders, and gateways as circles. Black arrows depict planned dependencies and red dotted arrows (dark shading if viewed in grayscale) depict unplanned dependencies. Black lines with circular endpoints depict resource constraints. Red deliverables (dark shading if viewed in grayscale) indicate selected points of change injection into the workflow. Depth of shading on a task indicates the mean effort incurred in that task for this change case, calculated by the algorithm as explained in Sec. 6.2. GTA = global thermal aircraft model. ENTM = equipment thermal model. PTS = purchase technical specifications. OEM A/C = original equipment manufacturer of aircraft.

Grahic Jump Location
Fig. 2

Properties dialog for task 14 in Fig. 1, showing duration and output sensitivity options that are generated automatically according to the task's context in the workflow

Grahic Jump Location
Fig. 3

Examples illustrating how K determines the effect of the sensitivity level on θ, and thus on the duration of a task as specified in Eq. (4), for selected values of K and the extent of change Δ

Grahic Jump Location
Fig. 4

Workflow diagram for the turbine disk design process, illustrating the outline structure and complexity level of a real-world CPiW model. Swimlanes indicate assignment of the process across teams who must frequently transfer information. Green deliverables (light shading if viewed in grayscale) represent information received from other parts of the design process, i.e., points of potential change initiation for this workflow. Blue deliverables (dark shading if viewed in grayscale) represent information delivered to other parts of the design process, i.e., points of potential change out-propagation. These details may be viewed by magnifying the electronic version. All labels are obscured due to commercial sensitivity.

Grahic Jump Location
Fig. 5

Task DSM of the turbine disk design process, sequenced to ensure that unplanned dependencies, which may cause iteration if they occur, appear as marks above the diagonal. These dependencies are shown as triangles. Detail is obscured due to commercial sensitivity.

Grahic Jump Location
Fig. 6

Probability distribution of the change process duration, for the workflow and initiation points shown in Fig. 1. Numerical values have been modified due to commercial sensitivity.

Grahic Jump Location
Fig. 7

Probabilistic Gantt chart of the change implementation process of Fig. 1, divided according to resource. Green bars (light shading if viewed in grayscale) show when a task is executed for the first time. Red bars (dark shading if viewed in grayscale) show rework. Numerical values have been modified due to commercial sensitivity.

Grahic Jump Location
Fig. 8

Expected effort required from each team over time to implement a selected change in the turbine disk design process. Numerical values have been modified due to commercial sensitivity.

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In