Anita Kovač Kralj

Retrofit of complex and energy intensive processes II: stepwise simultaneous superstructural approach

Simultaneous parameter and structural optimization of an existing complex and energy intensive continuous process have been studied using rigorous models. The method that was recently developed to sequentially optimize retrofits has been extended to a stepwise simultaneous superstructural approach, using available process simulators and optimization software capabilities. An extended procedure has been employed for retrofits using a three-step approach: (i) generation of a process superstructure by pinch analysis; (ii) formulation of a mixed integer nonlinear programming (MINLP) model and its simplification into a relaxed nonlinear programming (NLP) model; (iii) simultaneous optimization, first by a process simulator and than by the NLP algorithm. Pinch analysis offers alternative retrofit designs for postulating the superstructure. An advantage of the relaxed NLP formulation over the MINLP one is that it can be easily generated. Poor local optima due to nonlinear interconnections between alternative solutions have been avoided by applying a direct search complex algorithm using the simulator. The approach has been illustrated by maximizing the annual profit of an existing methanol plant. The optimal solution yields additional savings of 5.17 MUSD per annum (6.75% of the total income) which is 6.5 times the savings obtained by the earlier sequential optimization method.

Simultaneous retrofit of complex and energy intensive processes-III

Abstract A simultaneous structural and parameter optimization model of a total process was used to approach the global optimum solution of the retrofit. Simultaneous nonlinear programming (NLP) optimization can be applied to a complex industrial process giving more economical and profitable operation. The NLP model with variable flow rates of all the streams in different structures has been chosen. The problem has been how to handle the additional effect of nonconvexities, which prevent the problem being converged. The NLP algorithm, which was included as the additional technique using a possibility of minimum variable-error with a lower upper bound in mass and energy balances, was easier to solve. The NLP model of an existing methanol plant has been extended with the process separation section. The additional profit for all the process was 5.262 MUSD/a, while for the process without purification it was 5.173 MUSD/a.

Water condensate collection system by using MINLP model

Abstract This paper presents a retrofitted procedure for wastewater, or more specifically, water condensates separate collection, using an intermediate mathematically-programmed heating design based on a MINLP (mixed-integer nonlinear programming) model. The existing separate condensate collection using an intermediate heating system may no longer be optimal; the basic intention being that minimal changes in the system can efficiently improve the separate collections of low and high-temperature condensates, and the use of available heat. This water condensate collection process was tested on an existing methanol process, by using a MINLP model, which allowed for more effective and additional 5 % water condensate collection system for steam generation.

Study of Silver and Oxide Hybrids of Catalysts Formaldehyde Production by Using NLP Model

Abstract Hybrid catalysts are engineered to contain elements of two or more catalysts. Hybrid catalysts are becoming increasingly important advantages for obtaining novel and desirable catalyst activities and properties. The nature of a hybrid catalyst can be changed with additional characters. The disadvantages can be divided and the advantages multiplied. This study describes a novel technology for the synthesis of formaldehyde production by using hybrid catalysts and the optimization of them. The nature of the hybrid catalyst can be changed by the addition of positive properties. Both, 100 % oxide and 100 % silver formaldehyde production processes can be optimized during hybrid processes. The application of a nonlinear programming, (NLP) mathematical method can be used for optimizing the hybrids of catalysts consisting of percentages of oxide and silver catalysts. The NLP model contains equations for parametric optimization. The optimized silver process modification was based on 80 % silver catalyst material and the addition of 20 % oxide catalyst. A 20 % change of catalyst can enhance the available heat and product production by 3.8 %.

Parameters Influences during Biodiesel Production

Bio-diesel is a clean burning alternative fuel, produced from domestic, renewable resources. Bio-diesel can be blended at any level with petroleum diesel to create a bio-diesel blend. It can be used in compression-ignition (diesel) engines with little or no modification. Bio-diesel is simple to use, biodegradable, non-toxic, and essentially free of sulphur and aromatics. This paper presents the two following identifiable topic areas as key themes: 1. preparation of an aqueous solution of sodium hydroxide – as a catalyst, which can be activated by the most MeO- active groups, and can therefore be converted to methyl esters (biodiesel) from triglyceride. Methoxide (MeO-) was produced from sodium hydroxide (NaOH) and methanol (MeOH) in a batch reactor: NaOH + MeOH = H2O + Na+ + MeO-. During bio-diesel production, methoxide is incorrectly referred to as the product of mixing methanol and sodium hydroxide. An aqueous solution of sodium hydroxide – was prepared as a catalyst, by using different amounts of water at the same temperature. The reaction with lower water took place at the highest and quickest degrees of NaOH conversion and thus more MeO- active groups. The water was effective as an inhibitor.

NLP optimization of gas turbine including experimental catalyst conversion data in methanol plant

This paper presents a study on experimentally measured the degrees of conversion in successive catalyst bed levels which can be included in simultaneous optimization using nonlinear programming (NLP) algorithm. The NLP model is including the equations of structural and parametric optimization of: existing catalyst model, recycled gas stream, reactor, gas turbine, heat exchangers, flash, compressors, splitter and CO2 reuse. The optimization approach is illustrated by a complex process of low-pressure Lurgi methanol production, giving an additional profit of 3,5 MUSD/a.

NLP optimization of a methanol plant by using H2 co-product in fuel cells

Abstract Fuel cells, process heat integration and open gas turbine electricity cogeneration can be optimized simultaneously using nonlinear programming (NLP) algorithm. The NLP model contains equations of structural and parametric optimization. The nonlinear programming model is used to optimize complex and energy intensive continuous processes. The procedure does not guarantee a global cost optimum, but it does lead to good, perhaps near-optimum designs. The optimization approach is illustrated by a complex process of low-pressure Lurgi methanol production, giving an additional profit of 2,6 MUSD/a. The plant, which is producing methanol, has a surplus of hydrogen (H2) flow rate in purge gas. H2 shall be separated from the purge gas by an existing pressure swing adsorption (PSA) column. Pure H2 can be used as fuel in fuel cells.

Hydrogen in the Methanol Production Process.

Hydrogen is a very important industrial gas in chemical processes. It is very volatile; therefore, it can escape from the process units and its mass balance is not always correct. In many industrial processes where hydrogen is reacted, kinetics are often related to hydrogen pressure. The right thermodynamic properties of hydrogen can be found for a process simulation and optimization; they can be estimated by the Grayson-Streed model. In the case studied, a methanol plant with a capacity of 150,000 tons/year, the flow rate of hydrogen (H2) can be optimized using nonlinear programming, but good estimates of thermodynamic properties of hydrogen had to be found. In the case study, the model is in relatively good agreement with experimental measurements in the existing methanol production plant. An additional flow rate of hydrogen can increase the methanol production by 1.2%. Total potential profit increase of the additional methanol production was estimated to be 30 kEUR/a (or 103 euros/year).

Waste heat integration between processes III: Mixed integer nonlinear programming model

Abstract The simulatenous heat and power integration between processes using mixed integer nonlinear programming (MINLP) contains equations of structural and parametric optimization of the nonretrofitted or retrofitted plants. High pressure steam production and generation of electricity using a steam turbine are included in the integrated structure. MINLP can predict: the optimum energy target of the integration between processes, steam and electricity production. The approach has been illustrated by four complex processes (solvent production, formalin production, methanol production and oil refinery) using MINLP. The objective function have maximized the annual profit of heat and power integration to 303,2 kUSD/a.