(106g) Ensuring Process Safety of Heat Exchanger Networks Under Uncertainty

Heat exchanger network synthesis (HENS) offers a cost effective means of achieving a plant’s energy demand [1,2]. Often overlooked, however, are the process safety metrics, including the potential for overpressure. One possible consequence is a tube rupture, an overpressure event whereby the low pressure shell side is pressurized by the high pressure tube side [3,4,5]. If the pressures are not considered, the chances of selecting unsafe exchanger configurations that are susceptible to tube rupture increase. To overcome this, prior work has incorporated a Safety Rating (SR) for individual heat exchangers [6,7]. This metric captures the severity of a tube rupture for a single exchanger and can serve as a benchmark when comparing the safety of different exchanger configurations. The specific challenges of safety-ensured HENS, however, include the complexity in incorporating complicated SR calculations, ensuring safety of connected exchangers in a network setting, and the difficulty in imposing the safety constraints under uncertainty. The deterministic SR is obtained via dynamic simulation that requires specifications and highly nonlinear thermodynamic property predictions for liquid-liquid, vapor-liquid, and flashing liquid-liquid systems. We tackle this challenge by developing a data-driven piecewise linear (PWL) underestimation of SR, which is then incorporated in the HENS formulation that retains the MINLP structure of the synthesis problem. Originally proposed by Rebennack and Kallrath [8], the PWL formulation ensures an efficient and tight but, at the same time, a conservative underestimation (vs. approximation) of SR. In the context of process safety, this is important because of the severity of erroneous approximation. The simplicity of PWL allows a minimum SR constraint to be applied to all exchangers to obtain a final HEN configuration that meets a plant’s level of risk.

We further extend the HEN synthesis to account for the uncertainty of SR predictions. This uncertainty is assumed to be in the form of pressure variations across a HEN’s streams. These uncertainties, which are common during normal operation, can lead to lower SR values. If left unaccounted for, the SR minimum threshold originally proposed may not perform as expected in a plant setting and instead give a false sense of security. As the SR is itself a function of pressure, we use a robust counterpart of the safety constraint to tackle the worst-case scenario and ensure that the designed HEN can tolerate a certain level of variance that may be present in a plant. One benefit of this approach is that plants could use historical operating data to assume a level of variance and subsequently determine the most cost effective HEN configuration. We will demonstrate the overall safety-ensured HEN synthesis framework using several case studies.

References:

[1] Yee, T. F., & Grossmann, I. E. (1990). Simultaneous optimization models for heat integration—II. Heat exchanger network synthesis. Computers & Chemical Engineering, 14(10), 1165-1184.

[2] Ponce-Ortega, J. M., Jiménez-Gutiérrez, A., & Grossmann, I. E. (2008). Optimal synthesis of heat exchanger networks involving isothermal process streams. Computers & Chemical Engineering, 32(8), 1918-1942.

[3] Grim, L. (2017). CSB Investigation: Williams Geismar Olefins Plant Reboiler Rupture and Fire. In ASSE Professional Development Conference and Exposition. American Society of Safety Engineers.

[4] API Standard 521. (2014). Pressure‐Relieving and Depressuring Systems.

[5] Hellemans, M. (2009). The safety relief valve handbook: design and use of process safety valves to ASME and International codes and standards. Elsevier.

[6] Harhara, A., & Hasan, M. F. (2020). Dynamic modeling of heat exchanger tube rupture. BMC Chemical Engineering, 2(1), 1-20.

[7] Harhara, A., & Hasan, M. F. (2019). Incorporating Process Safety into Heat Exchanger Network Synthesis and Operation. In Computer Aided Chemical Engineering (Vol. 47, pp. 221-226). Elsevier.

[8] Rebennack, S., & Kallrath, J. (2015). Continuous piecewise linear delta-approximations for univariate functions: computing minimal breakpoint systems. Journal of Optimization Theory and Applications, 167(2), 617-643.