A non-smooth variational framework for optimization of chemical process systems
Keywords:
Non-smooth optimization, variational inequality, clarke subdifferential, generalized gradient, chemical process optimization, projected subgradient method, nonsmooth stationarity, process systems engineering.Abstract
Chemical process optimization frequently requires design and operational choices guided by mathematical models that are continuous and everywhere non-differentiable. The non-smoothness may come from either absolute-value penalty terms, changes in the operation regime, limits due to saturation, maxima and minima functions, phase-selection rules, or the piecewise representation of the process. Classic gradient-based optimization methods break down in the case where the functions that define process performance and constraints contain kinks or regime-dependent derivatives. The present paper develops a non-smooth variational model of chemical process optimization employing generalized sub-differentials and variational inequalities. The decision-making problem of the process is formulated in a compact convex feasible set, while the objective function describing process performance is taken to be Lipschitz continuous with nonsmooth penalty terms. The first-order necessary conditions of optimality are derived using Clarke's generalized derivative and the optimization problem is formulated as a variational inequality involving the generalized derivative and the normal cone of the feasible region. Under continuity assumptions, local Lipschitz continuity, convexity of the feasible region and pseudomonotonicity of the associated generalized mapping, the existence of optimality results is shown. Also, the computational complexity is evaluated with regards to the number of process variables, active constraints and nonsmooth evaluations. Experimental investigations are performed on a typical problem of chemical process optimization, where different states of operation are examined with respect to reaction temperature, residence time, feed rate and intensity of separation. The framework presented achieves values of 0.184, 0.219, and 0.287 in the testing cases while satisfying the restrictions on yield for the product. Comparative analysis with various approaches, including smooth projected gradient, standard penalty for optimization, and derivative-free search, shows the effectiveness of the combined application of generalized first-order information. The results provide mathematic basis for the framework of optimizing the systems in case of multi-mode processes, which cannot be reasonably described by globally-smooth functions.
Downloads
Downloads
Published
Issue
Section
License
You are free to:
- Share — copy and redistribute the material in any medium or format for any purpose, even commercially.
- Adapt — remix, transform, and build upon the material for any purpose, even commercially.
- The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms:
- Attribution — You must give appropriate credit , provide a link to the license, and indicate if changes were made . You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
- No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.
Notices:
You do not have to comply with the license for elements of the material in the public domain or where your use is permitted by an applicable exception or limitation .
No warranties are given. The license may not give you all of the permissions necessary for your intended use. For example, other rights such as publicity, privacy, or moral rights may limit how you use the material.

