This presentation was made at the NAFEMS Americas Seminar "Model-Based Engineering: What is it & How Will It Impact Engineering Simulation" held on the 1st of October 2019 in Columbus Ohio

Resource Abstract

BACKGROUND: The benefits of Modeling and Simulation (M&S) are well known and documented both within Medtronic and in external literature. However, there is a question concerning the most efficient means to organize the inclusion of M&S into the product development cycle. We tested the hypothesis that a hybrid development process (PDP) involving both Empirical and M&S methods in Centralized or Decentralized organizational structures lead to improvements in throughput and resource utilization.



PROBLEM: Modeling and Simulation, like many engineering and scientific disciplines, requires specific training and skill development. In the Distributed model, all design engineers and scientists are responsible for their own simulations. However, human factors literature supports that for skilled activities, a proficiency curve exists whereby efficiencies improve with repeated practice; and degrade during periods of inactivity. For a Distributed Model, this proficiency curve can be a significant obstacle when M&S activities are only needed occasionally during a project. In the Centralized model, M&S activity occurs within a specific group or department, providing a skilled workforce that remains ‘on top of’ the proficiency curve given their service role in support of multiple projects. A challenge with this organizational method is the speed and efficiency with which learning can be given back to the project team; a challenge that is not experienced by the Distributed model where M&S activity is directly embedded within the team and results are incorporated instantaneously.



METHODS: A discrete event simulation was conducted to evaluate a generic PDP process including concept, development and V&V loops with potential for rework due to design and analysis errors. Organizational models including Distributed, Centralized, and a proposed Mix based on the complexity of the M&S activity were evaluated and compared to a baseline model where M&S was not included. Potential for Design error and Simulation efficiencies are guided by both literature and practical experience. Analyses were conducted stochastically to support statistical evaluation of results. Key comparisons include the number of projects produced via each model in a given period, project time in system, resource utilization and others.



RESULTS & CONCLUSIONS: A discrete event simulation was conducted to compare the organizational modeling efficiencies of multiple strategies for incorporating the benefits of M&S to a project. Results confirmed that the use of M&S is beneficial when compared to the baseline Empirical model. Advantages and disadvantages were identified for Distributed and Centralized organizational models. A proposed Mixed model was identified that balances the key parameters for an overall improved result. Guidance from both literature and practical experience is used to support a finding that a Mixed organizational model outperforms both Distributed and Centralized models and that all uses of M&S outperform the baseline Empirical model.