(183h) Mathematical Programming Formulations for Design and Manufacturing of Process Families: Applications to Green Energy Systems

Significant, timely reduction in carbon emissions requires rapid deployment of a large number of decentralized process systems, including, for example, carbon capture and clean energy production facilities. Considering carbon capture, the range of required installations can be viewed as a family of similar processes that must, collectively, meet a range of design characteristics, including feed properties, capture rates, and other performance criteria. Traditional process design approaches are not appropriate for this kind of broad, rapid deployment of many similar processes. Classical techniques focus on optimizing individual designs for each installation, failing to exploit opportunities for shared components, decreased engineering effort, and reduced manufacturing costs across these installations. While modularization aims to reduce capital costs and decrease deployment timelines[1], capacity requirements are achieved by “numbering up”, which impacts economies of scale and can lead to increased overall costs and decreased flexibility to meet rapid deployment goals.

In contrast, in this presentation, we focus on the simultaneous, integrated design of process families for low-cost manufacturing and rapid deployment of green energy systems. Similar to product family design approaches, we consider the trade-off between cost-effectiveness, similarities among processes, and satisfaction of requirements and demands of the process[2]. We present multiple mathematical programming formulations for the simultaneous optimization of process families. A mixedinteger linear programming formulation (MILP) utilizes a discretized set of operating conditions and potential sizing options[3]. The MILP formulation is scalable to large problems because of favorable theoretical properties that allow us to significantly reduce the number of binary variables in the formulation. However, the time-consuming bottleneck of this approach is computation of the input data
obtained by sampling across the potential design space. In contrast, we also propose a mixed-integer nonlinear approach that directly works with the process model to determine optimal sizes across operating conditions while avoiding the need to precompute large numbers of simulations. We provide computational comparisons between these formulations, along with a discussion of the theoretical properties of these formulations. We also present case studies that demonstrate the value of this approach in addressing the challenges of rapid, broad deployment of related process systems.

Acknowledgements and Disclaimers: Sandia National Laboratories is a multimission laboratory managed
and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary
of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security
Administration under contract DE-NA0003525. This work was partially funded by the Institute for the
Design of Advanced Energy Systems (IDAES) with funding from the Office of Fossil Energy, Cross-Cutting
Research, U.S. Department of Energy.

References
[1] Baldea, M., Edgar, T. F., Stanley, B. L. & Kiss, A. A. (2017), ‘Modular manufacturing processes: Status,
challenges, and opportunities’, AIChE journal 63(10), 4262–4272.
[2] Simpson, Timothy & Siddique, Zahed & Jiao, Roger. (2006). Product Platform and Product Family
Design: Methods and Applications. 10.1007/978-1-4614-7937-6.
[3] C. Zhang, C. Jacobson, Q. Zhang, L.T. Biegler, J.C. Eslick, M. A. Zamarripa, D. Miller, G. Stinchfield, J. D.
Siirola , C. D. Laird, “Optimization-based Design of Product Families with Common Components”, to
appear in proceedings of PSE 2021+