(635c) Data-Driven Feasibility Analysis for Modular Design Under Demand Variability

For years, the design of chemical process facilities has followed a traditional cost reduction paradigm relying on economy of scales [1].The so called 2/3 power law implicates that as chemical plants grow bigger (scale up), the cost is being reduced following a 2/3 power low. In simple terms, large plants exhibit better efficiencies and lower cost due to more efficient process integration. However, recent studies on modular and distributed manufacturing have introduced a new angle to economies of scale [2]. Small scale, modular and continuous plants could shorten planning and construction times, while distributed facilities could compensate increases in investment costs with decreases in operational (distribution) costs.

Modular design involves the use of small standardized modules of fixed size in a production process. Multiple identical devices may be assembled to achieve a certain throughput. The key-question in modular design then becomes to define a process based on a limited number of different modules [3]. Modular and distributed processes may not only contribute to decreases in distribution costs, but also provide additional flexibility to production processes when compared to large-scale plants. Flexibility is a fundamental concept of design under uncertainty. During conceptual design, there are often process parameters that are not well known, such as kinetic rate constants, demand or product and feedstock prices. Ensuring flexible designs allows one to systematically hedge against exceptional realizations of process parameters [2].

In this work, a framework for modular design under demand variability is presented. It is assumed that different module options for different equipment are available, that the equipment are arranged in a sequential process, and that the sequence of equipment is known and fixed. Then, given a certain demand space, the goal is to determine the optimal selection of module options that minimize investment costs while ensuring that the entire demand space can be covered. The framework consists of two basic steps: first, the feasible region of different module options will be defined using a data-driven approach. Then, the simultaneous design and flexibility evaluation problem is formulated as a multi-objective optimization problem and solved to optimality.

The proposed approach will be illustrated using the process synthesis optimization and flexibility evaluation of an air separation unit (ASU). The process consists on separating air into oxygen, nitrogen and argon through liquefication followed by low-temperature distillation. This process may be modularized into the four basic operations of heat exchange, expansion, distillation, and compression [4]. Then, given different module options for each equipment, varying on size and process specification (e.g., number of columns, pressure, etc.), the goal is to define a set of options that can achieve product specifications at a minimum investment cost, while ensuring that the operation remains flexible and the entire demand space can be covered.

1. Kim, Y.-h., et al., Modular chemical process intensification: a review. 2017. 8: p. 359-380.
2. Chen, Q. and I.E. Grossmann, Recent Developments and Challenges in Optimization-Based Process Synthesis. 2017. 8(1): p. 249-283.
3. Seifert, T., et al., Small scale, modular and continuous: A new approach in plant design. Chemical Engineering and Processing: Process Intensification, 2012. 52: p. 140-150.
4. Sirdeshpande, A.R., et al., Process synthesis optimization and flexibility evaluation of air separation cycles. 2005. 51(4): p. 1190-1200.