No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Superstructure optimization (MINLP) within ProSimPlus Simulator
Abstract
Although the methodologies of optimization-based process synthesis have evolved significantly during the last thirty years, the ability to robustly and accurately solve industrially-relevant global flowsheet problems remains limited. Engineering expertise and simulation-based sensitivity analysis still remain an essential guide for system alternatives generation and key devices selection. The purpose of this study is to provide a superstructure mixed-integer nonlinear programming (MINLP) optimization within the commercial simulator ProSimPlus. The entire optimization loop is directly managed by the simulator and both continuous variables and discrete integer variables are optimized simultaneously by an external metaheuristic optimizer called MIDACO (Mixed Integer Distributed Ant Colony Optimization).
... It is based on the existence of material nodes, called states, and transformation nodes, known as equipment, which performs a specific task imposed by the optimization algorithm, to connect two different states. In process simulation software, these nodes can be formally represented by classic flow splitters or mixers [29]. For the sake of the reader, these nodes acting as structural switches, will hereafter be mentioned as opening and closing switches, to differentiate them from actual splitting and mixing units of the process. ...
... Solving process engineering superstructures requires to consider continuous and discrete variables using Mixed-Integer Non-Linear Programming (MINLP) algorithms [32]. For this work, MIDACO-Solver is used, as it has proven to be useful for its integration to ProSimPlus software for other applications of superstructure optimization [29,31]. This algorithm is built as a general-purpose solver for mono-and multi-objective optimization problems, acting as a black box capable of handling nonexplicit objective functions, with or without equality or inequality constraints, as well as being able to handle critical properties such as non-linearities or discontinuities. ...
Ammonia has a potential as carbon-free and high-energy density compound for chemical storage of renewable energies and its synthesis from green H2 requires to be as energetically efficient as possible. In this work, a superstructure optimization methodology for process synthesis is proposed and applied to an ammonia production process. The approach covers three different scales: process, equipment, and molecules. Process scale refers to finding the optimal process structure, while the equipment scale is related to the best set of operating conditions, and the molecular scale studies two catalysts (Fe and Ru) with their respective kinetic rates. The optimization intends to minimize the Levelized Cost of Ammonia (LCOA) and maximize the energy efficiency. The best trade-off is found with the Ru catalyst, with an LCOA of 766 €/tNH3 and 57.2 % of energy efficiency. In comparison with reference cases, the LCOA decreases 0.6 % and the energy efficiency increases 1.5%. The main improvement is found in the pressure, 75 bar, reduced in 25 %. The production of 11.6 tNH3/day equals to 2.5 MW of stored power from 3 MW supplied as H2 to the process, with a total energy consumption of 10.67 kW h/kgNH3, including prior H2 and N2 production processes.
... Likewise, the metaheuristic design framework by Geraili et al. (2014) presented a modeling approach for designing energy systems applicated to biorefineries; however, it did not include a strategy for the optimization of multiple simulations in which different configurations of the process flowsheet is possible. Zhao et al. (2018) proposed a superstructure optimization within ProSimPlus simulator using an external metaheuristic optimizer called Mixed Integer Distributed Ant Colony Optimization (MIDACO). ProSimPlus is a process engineering software that performs rigorous mass and energy balance calculations for a wide range of industrial steady-state processes (prosim.net). ...
This paper presents a new optimization approach for the simultaneous structural optimization of process flowsheets with the operating conditions through combining process simulators with metaheuristic techniques. The proposed approach allows optimization of a superstructure in process simulators and reduce the computation time. A superstructure for different configurations for producing solar-grade silicon is considered, which includes three different configurations for solar-grade silicon production (Siemens Process, Intensified FBR Union Carbide Process, and Hybrid Process). The operating conditions with major impact in the performance of each of the proposed configuration were considered as decision variables. The improved multi-objective differential evolution (I-MODE) algorithm was selected as search method from others metaheuristic techniques because its efficiency to solve multi-objective problems in a short central process unit (CPU) time. The optimization algorithm consists in linking the process simulator software Aspen PlusTM with the metaheuristic technique. The results offered attractive options for the considered objective functions in the addressed case study.
... The purpose of a recent study is to provide a superstructure MINLP optimization within the commercial simulator ProSimPlus. The entire optimization loop is directly managed by the simulator and both continuous variables and discrete integer variables are optimized simultaneously by an external metaheuristic optimizer called Mixed Integer Distributed Ant Colony Optimization (Zhao et al., 2018) ...
This paper presents the authors’ perspectives on some of the open questions and opportunities in Process Systems Engineering (PSE) focusing on process synthesis. A general overview of process synthesis is given, and the difference between Conceptual Design (CD) and Process Design (PD) is presented using an original ternary diagram. Then, a bibliometric analysis is performed to place major research team activities in the latter. An analysis of ongoing work is conducted and some perspectives are provided based on the analysis. This analysis includes symbolic knowledge representation concepts and inference techniques, i.e. , ontology, that is believed to become useful in the future. Future research challenges that process synthesis will have to face, such as biomass transformation, shale production, response to spaceflight demand, modular plant design, and intermittent production of energy, are also discussed.
The modernization of pharmaceutical manufacturing is driving a shift from traditional batch processing to continuous alternatives. Synthesizing end‐to‐end optimal (E2EO) manufacturing routes is crucial for the pharmaceutical industry, especially when considering multiple operating modes—such as batch, continuous, or hybrid (containing both batch and continuous steps). A major challenge is the ability to compare these manufacturing alternatives across different operating modes, hindering optimal superstructure synthesis. To bridge this gap, this study introduces a hierarchical framework for the synthesis of E2EO manufacturing routes, employing a hybrid rule‐based and optimization‐driven approach. This method optimizes flowsheets modeled using PharmaPy through a simulation‐optimization technique with modest computational requirements. The effectiveness of the proposed framework is demonstrated through a case study on the manufacturing of the cancer therapy drug Lomustine. Two distinct manufacturing scenarios are analyzed to generate E2EO manufacturing campaigns tailored to the specific chemistries and process configurations, considering process efficiency and sustainable manufacturing.
Reverse Osmosis is considered today as the best available technology at industrial scale for desalination as an alternative process to conventional thermal technologies. Nevertheless, there is still opportunity to evaluate and optimize other membrane technologies like membrane distillation. Vacuum membrane distillation has been studied at pilot scale and it is a promising technology to treat seawater in areas with un-expensive heat sources. This study presents a modified method of superstructure optimization to evaluate key performance indicators of reverse osmosis and vacuum membrane distillation. To achieve this objective, Computer aided process simulation is deployed; mathematical model was coded in FORTRAN and added to a superstructure defined in a process simulation software (ProSimPlus). Geometry and material parameters are set according to the commercial module Dow SW30HR-380 data. Finally, superstructure’s operational conditions (pressure and in-process temperature) and specific number of modules in series and parallel are optimized within an ant colony algorithm (MIDACO).
Computer software marketed by companies such as the Heat Transfer Research Institute (HTRI), HTFS, and B-JAC International are used extensively in the thermal design and rating of HEs. A primary objective in HE design is the estimation of the minimum heat transfer area required for a given duty, as it governs the overall cost of the HE. However, because the possible design configurations of heat transfer equipment are numerous, an exhaustive search procedure for the optimal design is computationally intensive. This paper presents a genetic algorithm (GA) framework for solving the combinatorial problem involved in the optimal design of HEs. The problem is posed as a large-scale, combinatorial, discrete optimization problem involving a black-box model. The problem is derived from earlier work on HE design using simulated annealing (SA). SA and GAs are particularly suitable in this black-box model because they lack the crucial gradient information required for other mathematical programming approaches. A methodology based on a command procedure has been modified to run the HTRI design program iteratively coupled to both SA and GAs. In the earlier studies, SA was found to be a robust and computationally efficient technique for the optimal design of HEs subject to infeasibilities and vibration problems. This paper compares the performance of SA and GAs in solving this problem and presents strategies to improve the performance of the optimization framework.
A new and universal penalty method is introduced in this contribution. It is especially intended to be applied in stochastic
metaheuristics like genetic algorithms, particle swarm optimization or ant colony optimization. The novelty of this method
is, that it is an advanced approach that only requires one parameter to be tuned. Moreover this parameter, named oracle, is easy and intuitive to handle. A pseudo-code implementation of the method is presented together with numerical results
on a set of 60 constrained benchmark problems from the open literature. The results are compared with those obtained by common
penalty methods, revealing the strength of the proposed approach. Further results on three real-world applications are briefly
discussed and fortify the practical usefulness and capability of the method.
KeywordsConstrained optimization-Global optimization-Penalty function-Stochastic metaheuristic-Ant colony optimization-MIDACO-Solver-Mixed integer nonlinear programming (MINLP)
This article first reviews recent developments in process synthesis and discusses some of the major challenges in the theory and practice in this area. Next, the article reviews key concepts in optimization-based conceptual design, namely superstructure representations, multilevel models, optimization methods, and modeling environments. A brief review of the synthesis of major subsystems and flowsheets is presented. Finally, the article closes with a critical assessment and future research challenges for the process synthesis area. Expected final online publication date for the Annual Review of Chemical and Biomolecular Engineering Volume 8 is June 7, 2017. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
Downstream processing of biofuels and bio-based chemicals represents a challenging problem for process synthesis and optimization, due to the intrinsic nonideal thermodynamics of the liquid mixtures derived from the (bio)chemical conversion of biomass. In this work, we propose a new interface between the process simulator PRO/II (SimSci, Schneider-Electric) and the optimization environment of GAMS for the structural and parameter optimization of this type of flowsheets with rigorous and detailed models. The optimization problem is formulated within the Generalized Disjunctive Programming (GDP) framework and the solution of the reformulated MINLP problem is approached with a decomposition strategy based on the Outer-Approximation algorithm, where NLP subproblems are solved with the derivative free optimizer belonging to the BzzMath library, and MILP master problems are solved with CPLEX/GAMS. Several validation examples are proposed spanning from the economic optimization of two different distillation columns, the dewatering task of diluted bio-mixtures, up to the distillation sequencing with simultaneous mixed-integer design of each distillation column for a quaternary mixture in the presence of azeotropes.
The potential performance of carbon dioxide as working fluid is recognized to be similar to that of steam, which justifies thorough thermodynamic analysis of possible cycles.
The substantially better results achievable with CO² with respect to other gases are due to the real gas behaviour in the vicinity of the Andrews curve. Simple cycles benefit from the reduced compression work, but their efficiency is compromised by significant losses caused by irreversible heat transfer. Their economy, however, is appreciably better than that of perfect gas cycles. More complex cycle arrangements, six of which are proposed and analyzed in detail, reduce heat transfer losses while maintaining the advantage of low compression work and raise cycle efficiency to values attained only by the best steam practice.
Some of the cycles presented were conceived to give a good efficiency at moderate pressure which is of particular value in direct-cycle nuclear applications.
The favourable influence on heat transfer coefficients of the combined variation with pressure of mechanical, thermal and transport properties, due to real gas effects, is illustrated. Technical aspects as turbo-machines dimensions and heat transfer surfaces needed for regeneration are also considered.
Cooling water requirements are found to be not much more stringent than in steam stations.
Copyright © 1969 by ASME Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material Stature
Use interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternal
This paper introduces a simultaneous process optimization and heat integration approach, which can be used directly with the rigorous models in process simulators. In this approach, the overall process is optimized utilizing external derivative-free optimizers, which interact directly with the process simulation. The heat integration subproblem is formulated as an LP model and solved simultaneously during optimization of the flowsheet to update the minimum utility and heat exchanger area targets. A piecewise linear approximation for the composite curve is applied to obtain more accurate heat integration results. This paper describes the application of this simultaneous approach for three cases: a recycle process, a separation process and a power plant with carbon capture. Case study results indicate that this simultaneous approach is relatively easy to implement and achieves higher profit and lower operating cost and, in the case of the power plant example, higher net efficiency than the sequential approach.
The design trade-offs between thermodynamics and economics of energy conversion systems can be more effective by combining a superstructure and mixed-integer non-linear programming (MINLP) techniques. The front of decision space showing the optimal sets of economic behavior and system efficiency with different corresponding optimal system structures and process variables can provide additional and useful information on cost-effective design of thermal systems. In this paper, this idea was successfully applied to supercritical coal-fired power plants to investigate the economically-optimal designs at each efficiency level. The superstructure involving up to ten feedwater preheaters, up to two reheatings and a secondary turbine with steam extractions (ET) was built. An improved differential evolution algorithm was used to simultaneously solve the parametric and structural optimization problem. The differences among the fronts of various types of plants, the front changes with plant efficiency and the effects of introducing an ET were discussed in detail. For a single reheating unit, a decrease of 2% in cost of electricity can be achieved. The optimal pressure ratios of reheatings are 0.15-0.25 (for single reheating), 0.2-0.3 and 0.15-0.3 (for double reheatings).
The developments obtained in recent years in the field of mathematical programming considerably reduced the computational time and resources needed to solve large and complex Mixed Integer Non Linear Programming (MINLP) problems. Nevertheless, the application of these methods in industrial practice is still limited by the complexity associated with the mathematical formulation of some problems. In particular, the tasks of design space definition and representation as superstructure, as well as the data collection, validation and handling may become too complex and cumbersome to execute, especially when large problems are considered. In an earlier work, we proposed a computer-aided framework for synthesis and design of process networks. In this contribution, we expand the framework by including methods and tools developed to structure, automate and simplify the mathematical formulation of the design problem. Furthermore, the models employed for the representation of the process alternatives included in the superstructure are refined, through the inclusion of the energy balance. Finally, the features of the framework are highlighted through the solution of two case studies focusing on food processing and biofuels.
A framework is presented for the automated superstructure generation and optimization of distributed energy supply systems (DESS). Based on a basic problem description (specifying load cases, available technologies, and topographic constraints), the presented framework first employs the P-graph approach for the generation of an initial superstructure containing exactly one unit of each feasible technology. DESS, however, require to account for multiple redundant conversion units, and thus a successive approach is employed to automatically expand the initial P-graph superstructure. In addition, topographic constraints are incorporated. The expanded superstructure is automatically converted into a mathematical model using a generic component-based modeling approach. Here, a robust MILP formulation is used to rigorously optimize the structure, sizing and operation of DESS. The employed MILP formulation accounts for time-varying load profiles, continuous equipment sizing, and part-load dependent operating efficiencies. In the present implementation, a GAMS model is generated that can be readily optimized. The methodology is applied to a real-world case study. It is shown that the framework conveniently and efficiently enables automated grassroots and retrofit synthesis of DESS identifying unexpected and complex designs.
This work addresses the optimization of ammonia–water absorption cycles for cooling and refrigeration applications with economic and environmental concerns. Our approach combines the capabilities of process simulation, multi-objective optimization (MOO), cost analysis and life cycle assessment (LCA). The optimization task is posed in mathematical terms as a multi-objective mixed-integer nonlinear program (moMINLP) that seeks to minimize the total annualized cost and environmental impact of the cycle. This moMINLP is solved by an outer-approximation strategy that iterates between primal nonlinear programming (NLP) subproblems with fixed binaries and a tailored mixed-integer linear programming (MILP) model. The capabilities of our approach are illustrated through its application to an ammonia–water absorption cycle used in cooling and refrigeration applications.
The economic optimization of a distillation column involves the selection of the number of trays and the feed- and side-streams locations, as well as the operating conditions that minimize the total investment and operation cost. In this paper, we present a superstructure-based optimization algorithm that combines the capabilities of commercial process simulatorstaking advantage of the tailored algorithms designed for distillation and property estimation implemented in these simulatorsand generalized disjunctive programming (GDP). The proposed algorithm iterates between two types of subproblems: A nonlinear programming (NLP) subproblem, in which the trays are divided into existing and nonexisting (nonexisting trays behave like simple bypasses without mass or heat exchange through the use of Murphree efficiencies), and a specially tailored master mixed integer linear programming (MILP) problem. The NLP subproblems are solved by integrating the process simulator with an external NLP solver. Several examples are presented which show promising results.
This paper deals with the development of a mixed integer nonlinear programming (MINLP) synthesizer for sequential modular simulators. Firstly, a variant of the outer approximation/equality relaxation/augmented penalty (OA/ER/AP) algorithm for MINLP problems is presented that makes use of Benders cuts in previous or subsequent iterations. An automatic process synthesis environment is then described for the public version of the ASPEN simulator using this algorithm, with the decomposition strategy by Kocis and Grossmann. The application of this new capability is demonstrated with several examples including the structural optimization of the hydrodealkylation of toluene process.
The problem of integrated process and control system design is discussed in this paper. We formulate it as a mixed integer nonlinear programming problem subject to differential-algebraic constraints. This class of problems is frequently non-convex and, therefore, local optimization techniques usually fail to locate the global solution. Here we propose a global optimization algorithm, based on an extension of the Ant Colony Optimization metaheuristic, in order to solve this challenging class of problems in an efficient and robust way. The ideas of the methodology are explained and, on the basis of different full-plant case studies, the performance of the approach is evaluated. The first set of benchmark problems deal with the integrated design and control of two different wastewater treatment plants, consisting on both NLP and MINLP formulations. The last case study is the well-known Tennessee Eastman Process. Numerical experiments with our new method indicate that we can achieve an im-proved performance in all cases. Additionally, our method outperforms several other recent competitive solvers for the challenging case studies considered.
Evolutionary algorithms, derived by observing the process of biological evolution in nature, have proven to be a powerful and robust optimizing technique in many cases. This paper describes the use of evolutionary algorithms for simultaneous structural and parameter optimization in process synthesis in a modular program environment. The commercial simulator ASPEN PLUS™ was integrated for the determination of the target function value. The ModelManager™ was utilized for a comfortable graphical problem definition. The simulations and the cost calculations exploit the complete process modeling accuracy without the necessity of simplifications due to restrictions imposed by the optimization method. The user can be sure that the best solution does not get lost by the application of crude models. In this paper the optimization method will be demonstrated by the synthesis of separation sequences and the overall process synthesis with limited degrees of freedom.
Chemical process synthesis methods and tools developed over the last several decades have reached a level of maturity that have provided advantage to practitioners in an environment of increased costs and shrinking margins. Future growth within the chemical process industries is likely to involve even keener competition with greater impact from factors such as raw material and energy availability, climate change mitigation, sustainability, and inherent security. The future will probably see an expanded role for the systematic generation process synthesis paradigm, including an increased interdependency with process and catalytic chemistry on one hand and operability and control expertise on the other. Advances from artificial intelligence may inspire new process synthesis paradigms incorporating more effective representations of the underlying physical sciences and engineering art, new social concerns, new design strategies, and new computerized implementations. The future may also see a collaboration of the systematic generation and superstructure optimization process synthesis paradigms in which systematic generation is used to create the superstructure for simultaneous discrete and continuous variable optimization. As the resulting process designs will certainly be evaluated from additional points of view including social considerations, superstructure optimization will need to produce families of good designs for multi-criteria Pareto optimization. There are many challenges, but continued progress will be made and these challenges will be met.
An optimization framework is proposed in this work for the synthesis and design of complex distillation sequences, based on a modified genetic algorithm (GA) coupled with a sequential process simulator. The use of a simulator facilitates the formulation of rigorous models for different process alternatives, while the genetic algorithm allows the solutions of the complex non-convex mathematical problem, involving discrete and continuous decisions. To reduce the computational requirements of the optimization procedure, several strategies are proposed, including a novel stopping criterion, which provides an efficient way to end the calculations when the optimal solution has been found. The implementation of these strategies resulted in reductions up to 60% in CPU time for the synthesis of complex distillation systems, succeeding in problems where deterministic mathematical algorithms had failed.
This paper proposes a novel extractive dividing wall distillation column, which has been designed using a constrained stochastic multiobjective optimization technique. The approach is based on the use of genetic algorithms to determine the design that minimizes energy consumption and total annualized cost. Several case studies are used to show the feasibility of performing extractive separations in dividing wall distillation columns. The simulation results show the effect of the main variables on the complex extractive distillation process.
The supercritical thermodynamic power cycle Douglas Paper No
- E Feher
E. Feher, 1967, The supercritical thermodynamic power cycle Douglas Paper No.4348, in:
Proceedings of the Intersociety Energy Conversion Engineering Conference, 13-17