About

Theodor Freiheit is a Research Associate Professor in the Department of Mechanical Engineering at the University of Michigan. His research interests encompass manufacturing systems engineering, the design of manufacturing and service systems, reliability and productivity, and the application of Markov chain models including predictive control. He investigates the interaction between manufacturing process parameters and failure modes, as well as product quality and the influence of production system approaches across various industries. His work also involves the application of controls to manufacturing systems, design theory and methodology, and new product development. Freiheit's expertise extends to the design of machine components and systems, with a focus on how designer personality and environment impact design outcomes. He has a particular interest in engineering management, especially in product development, lean product design, and the interaction of business, product, and process within engineering systems. His research supports the development of products for flexible manufacturing systems, contributing to advancements in manufacturing efficiency and product quality.

Research topics

• Computer Science
• Economics
• Engineering
• World Wide Web
• Industrial engineering

Selected publications

• Robot-based Additive Manufacturing of Lego-type Modular Molds for Wind Blades

  2025-04-30

  report
  PublisherDOI

• A discrete state Markov chain model and decomposition solution method for the transient behaviour of buffered long serial transfer lines

  Journal of Manufacturing Systems · 2023-05-25 · 9 citations

  article1st authorCorresponding
  PublisherDOI

• Towards intelligent welding systems from a HCPS perspective: A technology framework and implementation roadmap

  Journal of Manufacturing Systems · 2022-09-29 · 33 citations

  articleSenior authorCorresponding
  PublisherDOI

• Feature-based quality classification for ultrasonic welding of carbon fiber reinforced polymer through Bayesian regularized neural network

  Journal of Manufacturing Systems · 2021-01-01 · 42 citations

  articleSenior authorCorresponding
  PublisherDOI

• Transfer lines with multi-machine stages and non-deterministic processing time: transient performance analysis and control

  Operational Research · 2021-01-11 · 4 citations

  articleSenior author
  PublisherDOI

• Quality detection and classification for ultrasonic welding of carbon fiber composites using time-series data and neural network methods

  Journal of Manufacturing Systems · 2021-10-01 · 22 citations

  articleSenior authorCorresponding
  PublisherDOI

• Robustness Optimization of Product Assembly Architecture for Personalization

  2021-07-16 · 1 citations

  preprint
  PublisherDOI

• Early event detection in a deep-learning driven quality prediction model for ultrasonic welding

  Journal of Manufacturing Systems · 2021-06-23 · 57 citations

  articleSenior authorCorresponding
  PublisherDOI

• Integrating optimal process and supplier selection in personalised product architecture design

  International Journal of Production Research · 2021-03-09 · 18 citations

  articleSenior authorCorresponding

  A key enabler for personalised product design is an open product architecture that allows the integration of personalised modules to create unique products. Decisions regarding product variety, module combinations, and configurations for personalised modules need to be coordinated with the decisions of manufacturing process and supplier selection when developing personalised product architectures. Conventionally, product architecture, processes, and suppliers are independently determined at different product development stages. However, this sequential design process lacks connection between product architecture, process, and supplier, and may lead to suboptimal or even infeasible design solutions with compromised performance. In this study, a concurrent optimisation approach is proposed to integrate manufacturing process and supplier selection into personalised product architecture design. A cost model is developed as a nexus of product architecture, process, and supplier. Then, a mixed-integer optimisation model is established to maximise the potential profit of a product family based on a profit formulation that incorporates customer preference, process resource, supplier, and manufacturing cost. A genetic algorithm is utilised to solve this optimisation problem. The method is demonstrated on the architecture design for a family of personalised bicycles. The result shows that concurrent optimisation can achieve design solutions with higher profitability compared to sequential design strategies.

  PublisherDOI

• Robustness Optimization of Product Assembly Architecture for Personalization

  Volume 2B: Advanced Manufacturing · 2020-11-16

  article

  Abstract Personalization has received extensive attention as a new manufacturing paradigm to address increased market demand for personalized products. An open product architecture that assembling common, customized, and personalized modules is regarded as a key enabler for product personalization, which can deliver one-of-a-kind products for individual customers at near mass production efficiency. Offering the best product architecture should consider the variations in design variables and parameters that influence the performance of a product architecture. This is especially true when designing open architecture for personalized products as many uncertain design quantities need to be considered in early product design stage. A robustness optimization method is proposed to simultaneously optimize product variety, module variant selection, and configuration of personalized module variants for a personalized assembly architecture. First, a profit model is developed to measure the performance of a product architecture, which incorporates individual customer preferences and manufacturing cost. A three-step process is proposed to model heterogeneous customer preferences: conjoint analysis of the preferences of a sample of customers from target market, market segmentation by a multi-variate normal mixture method, and simulation of personal preferences for a broader market by Monte-Carlo simulation. Thus, the simulated individual customer preferences are used to predict the sales and profit of product offerings. Second, the variation of profit associated with a product family architecture due to the uncertainty in customer preference and manufacturing cost estimates is formulated by a sensitivity analysis. A robustness index is defined by combining the objectives of maximizing profit and minimizing its variation. Lastly, a robustness optimization model is established to optimize product architecture by maximizing its robustness index. The proposed method is demonstrated with a personalized bicycle architecture design example.

  PublisherDOI

Frequent coauthors

• Simon S. Park

  10 shared

• Baicun Wang

  First People's Hospital of Yuhang District

  9 shared

• S. Jack Hu

  9 shared

• Changbai Tan

  General Motors (United States)

  5 shared

• Yoram Koren

  University of Michigan–Ann Arbor

  5 shared

• A. Kianimanesh

  University of Calgary

  5 shared

• Deyi Xue

  University of Calgary

  4 shared

• Qinwen Yang

  Hunan University

  4 shared

Education

• PhD, Mechanical Engineering

  University of Michigan