Electronic Thesis/Dissertation
Generative Independence: A Programmatic Analysis Framework for Complex Systems Engineering Open Access
In complex systems developments, engineers often need programmatic analysis models to help manage and make design and integration decisions. Programmatic analysis models assess cost, schedule and risk for a development program. Where there are significant choices available to an analyst about what models to select and how to assess them, the systems engineering literature provides mostly ad hoc guidance about how to achieve sufficiently diverse and independent approaches to simulate a complex system and provide critical guidance to managers.This dissertation research develops theory and heuristic guidance to help engineers decide how to choose and use programmatic analysis models. To do so, I compare two functionally similar programmatic analysis models that model the same phenomena but do so using different approaches. The key research question was to understand how differences – or independence – among programmatic analysis models shape model analysis results, both among individual models and model ensembles. Based on a literature review and application to a case study of programmatic models, I identified ten key dimensions of independence that can lead model assessments to produce important results. This is the basis for the Programmatic Analysis Framework of Generative Independence, which provides engineers a way to examine what types of model attributes differences are likely to generate different model results. Several points of guidance follow: 1) I show how the selected programmatic models (Cash Flow (CF) and Joint cost and schedule Confidence Level (JCL)) have complementary strengths and blindspots, introduced by how a model’s attributes are comprised along the dimensions of independence. 2) By varying the CF and JCL models, I identified that the causal logic and idealization approach is the key determinant of the utility of a programmatic model in replanning around key programmatic challenges. 3) Changes to one dimension of a model affect other dimensions, such as with the strong connection between model structure and the manageability of the model. 4) Engineers can consider how to change their models during the analysis process, potentially choosing to switch to alternate models or to use new models in parallel. Collectively, I extrapolate to discuss the main value of using independent approaches to increase search capability and increase likelihood of accurate claims.
- License
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
Relationships
Items
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
Pirtle_gwu_0075A_14957.pdf | 2019-12-17 | Open Access |
|