Seon B. Kim

Integrated design and control under uncertainty: Embedded control optimization for plantwide processes

Abstract High performance processes should operate close to design boundaries and specification limits, while still guaranteeing robust performance without design constraint violations. Since design chemical process is operating close to tighter boundaries safely; much attention has been devoted to integrating design and control, in which the design decisions, dynamics, and control performance are considered simultaneously in some optimal fashion. However, rigorous methods for solving design and control simultaneously lead to challenging mathematical formulations which easily become computationally intractable. In an earlier paper of our group, a new mathematical methodology to reduce the combinatorial complexity of integrating design and control was introduced ( Malcolm et al., 2007 ). We showed that substantial problem size reduction can be achieved by embedding control for specific process designs. In this paper, we extend the embedded control methodologies to plantwide flowsheet. The case study for the reactor-column flowsheet will demonstrate the current capabilities of the methodology for integrating design and control under uncertainty.

Integration of Design and Control for a large scale flowsheet

Abstract In order to design a process that operates close to tighter boundaries safely, much attention has been devoted to the integration of design and control, in which the design decisions, dynamics, and controlled performance are considered simultaneously in optimal fashion. However, rigorous methods for solving design and control simultaneously meet challenging mathematical formulations which become computationally intractable. In an earlier paper of our group, a new mathematical formulation to reduce the combinatorial complexity of integrating design and control was introduced. We showed that substantial reduction in the problem size can be achieved by embedded control for specific process designs. In this paper, we extend the embedded control to the plantwide process. This case will demonstrate the current capabilities of the methodology with integrated design and control under uncertainty.

Pinch Point Calculations and Its Implications on Robust Distillation Design

Abstract Rising energy costs and growing environmental awareness motivate a critical revision of the design of distillation units. Systematic design techniques, such as the rectification body, column profile map, and temperature collocation methods, require exact knowledge of all pinch points in a particular system, because these stationary points delineate the possible composition trajectories realizable in separation columns. This paper demonstrates novel methods for rigorously determining all pinch points for the constant relative volatility, ideal and non-ideal systems. Constant relative volatility and ideal solution systems are transformed into one-dimensional polynomial and nonlinear functions, regardless of the number of the components. A deflation method is proposed to locate all zeros in ideal and non-ideal zeotropic problems. For more challenging non-ideal problems, a novel hybrid sequential niche algorithm is used to solve hard azeotropic problems successfully. Finally, the design implications of these pinch point locations are investigated to show how new separation configurations can be devised. Methodically the paper points out the use of rigorous pinch point computations in conjunction with continuous composition profiles for robust distillation design.

Optimization of Complex Column Networks with Hybrid Genetic Algorithm

Abstract Complex column networks including Petlyuk, Kaibel, divided wall columns, side-stripper and rectifiers have better energy efficiency than simple separation networks. We present automatic and systematic computer-aided design and synthesis of complex column networks with heat-integrated configurations to separate multicomponent mixtures. Automatic synthesis of complex column is a challenging problem because the rigorous mass, equilibrium, summation and heat (MESH) equations lead to non-smooth fragmented search spaces. Therefore global optimization methods involving both structural and parametric degrees of freedom can currently not handle this problem and find realizable optimal energy efficiency in practice. To tackle the network synthesis problem, we proposed the following design and optimization approaches. First we apply an inverse design methods based on temperature collocation which substantially reduces the problem dimension. This method identifies the design feasibility keeping final product purities. Second, we developed hybrid genetic algorithm, in which stochastic GA elements can be combined with a local deterministic search. At first stage, GA explores the design space until individual network solutions begin to coalesce and confines the region with niche technique. The gradient-based local optimizer refines a solution to local optimality. This hybrid scheme enables to find out all global and local solutions and thus improves computation speed and accuracy of the solutions. Our novel methods allow the global search for feasible, effective and optimal energy efficient designs of the complex distillation column network in a fully automatic fashion. We used the rigorous commercial flowsheet simulator AspenPlus to validate the result and both results are very close each other. The automatic and rigorous flowsheet synthesis is apt to systematically address industrial-size process design problems such as the synthesis of energy-efficient separation networks, layout of biorefineries with novel feedstocks or sustainable process for reduction of greenhouse gases emissions.

Robust Thermodynamically-guided Algorithms for Synthesis of Energy Efficient Separation Networks

Abstract Distillation has a particularly high potential for energy savings, as it accounts for around 40–70% of capital and operating costs in petrochemical and commodity industries. However, numerous existing techniques that have been proposed to facilitate the design often lack the robustness and reliability needed to rigorously solve the problem, largely due to simplifying assumptions. In this contribution, we present computational methods which exploit a novel thermodynamically motivated problem transformation, entitled temperature collocation, rather than using classical tray-by-tray models which fail in pinched regions. This novel methodology applies to ideal, non-ideal and azeotropic mixtures, and is independent of the number of components. Due to the generalised nature of the method, we have also addressed the synthesis of complex column configurations as well as heat integration of distillation trains. Furthermore, the robust synthesis algorithm is presented, a method that automatically synthesizes a distillation network for given product purity requirements. Entire separation flowsheets are generated with rigorous thermodynamic models without the need to introduce limiting simplifications as is the case with existing shortcut techniques. The computational approach guarantees to identify realizable columns with a finite number of trays and operating conditions, which may easily be validated with industrial flowsheet simulator software packages like Aspen Plus.

Synthesis of Energy Efficient Complex Separation Networks

Abstract Separation processes account for about fifty percent of capital and operating costs including the highest energy demand in the chemical industry. The rise in energy consumption, the high cost of the fuels, and combined with the environmental impact increases the demand for energy saving separation processes such as optimal complex column networks which are estimated to achieve energy savings of up to 70%. This paper introduces a novel algorithm to create optimal complex column arrangements which encode the cost and states of global solutions with minimum user input. In the complex column networks, several combinations and internal connections increase the difficulty to optimize these processes. The proposed configuration algorithm generates all possible network configurations expressed as a continuous sequence of column section profiles and discriminates suboptimal solutions taking the column network structure as input. Simultaneously, a robust feasibility test is applied to the generated systematic network based in thermodynamic transformations called temperature collocation where the operating conditions, structure, and length of the separation network for realizing the desired product purities are not predefined. This robust hybrid algorithm combines the advantages of deterministic and stochastic search techniques. This computational approach guarantees rigorous column profiles validated with industrially accepted simulation software such as ASPEN. The capabilities will be illustrated using multicomponent realistic case studies especially quaternary systems.