Data-driven models are essential tools for the development of surrogate models that can be used for the design, operation, and optimization of industrial processes. One approach of developing surrogate models is through the use of input−output data obtained from a process simulator. To enhance the model robustness, proper sampling techniques are required to cover the entire domain of the process variables uniformly. In the present work, Monte Carlo with pseudo-random samples as well as Latin hypercube samples and quasi-Monte Carlo samples with Hammersley Sequence Sampling (HSS) are generated. The sampled data obtained from the process simulator are fitted to neural networks for generating a surrogate model. An illustrative case study is solved to predict the gas stabilization unit performance. From the developed surrogate models to predict process data, it can be concluded that of the different sampling methods, Latin hypercube sampling and HSS have better performance than the pseudo-r... [more]

This work focuses on the determination of the design and operation parameters of a thermal system depending on the optimization objective set. Its main objective and contribution concern the proposal of a generalized methodological structure involving multiobjective optimization techniques aimed at providing a solution to a practical problem, such as the design and dimensioning of a solar thermal system. The analysis is based on system operation data provided by a dynamic simulation model, leading to the development of multiple surrogate models of the thermal system. The thermal system surrogate models correlate the desired optimization objectives with thermal system design and operation parameters while additional surrogate models of the Pareto frontiers are generated. The implementation of the methodology is demonstrated through the optimal design and operation parameter dimensioning of a solar-assisted geothermal heat pump that provides domestic hot water loads of an office building... [more]

As forced-air cooling for heat sinks is widely used in the cooling design of electrical and electronic equipment, their thermal performance is of critical importance for maintaining excellent cooling capacity while reducing the size and weight of the heat sink and the equipment as a whole. This paper presents a method based on the combination of computational fluid dynamics (CFD) simulation and surrogate models to optimize heat sinks for high-end energy storage converters. The design takes the thermal resistance and mass of the heat sink as the optimization goals and looks for the best design for the fin height, thickness and spacing, as well as the base thickness. The analytical and numerical results show that the thermal resistance and mass of the heat sink are reduced by the proposed algorithms, as are the temperatures of the heating elements. Test results verify the effectiveness of the optimization method combining CFD simulation with surrogate models.

As an important combustion aid for aerospace vehicles, subcooled liquid oxygen of high density can be used to increase loading capacity of a spacecraft. Providing a large amount of cryogenic propellant in a short time with a strict energy consumption limitation is a challenge in the design of the fuel filling system. The authors proposed a vacuumed subcooling system combined with an ejector and liquid ring pump to vacuum a liquid oxygen tank and obtain subcooled liquid oxygen. After the liquid oxygen tank is vacuumed to an intermediate pressure by the ejector, it is further vacuumed to 10 kPa using the liquid ring pump. The infinitesimal method was used to simulate the thermodynamic processes involved. Taking the ejector working fluid mass flow rate, jet pressure, intermediate pressure, initial tank liquid level, and liquid ring pump speed as optimizing variables, optimization was conducted to determine the optimal vacuuming time, remaining liquid level in the tank, pumping speed diffe... [more]

In order to enhance the hydraulic performance of the volute pump, the Kriging model and genetic algorithm (GA) were used to optimize the 3D diffuser of the volute pump, and the hydraulic performance of the optimized model was compared and analyzed with the original model. The volute pump diffuser model was parameterized by BladeGen software. A total of 14 parameters such as the distance between the leading and trailing edges and the central axis, and the inlet and outlet vane angle were selected as design variables, and the efficiency under the design condition was taken as the optimization objective. A total of 70 sets of sample data were randomly selected in the design space to train and test the Kriging model. The optimal solution was obtained by GA. The shape and inner flow of the optimized diffuser were compared with those of the original diffuser. The research results showed that the Kriging model can effectively establish the high-precision mathematical function between the desi... [more]

It is paramount to optimize the performance of a chemical process in order to maximize its yield and productivity and to minimize the production cost and the environmental impact. The various objectives in optimization are often in conflict, and one must determine the best compromise solution usually using a representative model of the process. However, solving first-principle models can be a computationally intensive problem, thus making model-based multi-objective optimization (MOO) a time-consuming task. In this work, a methodology to perform the multi-objective optimization for a two-reactor system for the production of acrylic acid, using artificial neural networks (ANNs) as meta-models, is proposed in an effort to reduce the computational time required to circumscribe the Pareto domain. The performance of the meta-model confirmed good agreement between the experimental data and the model-predicted values of the existent relationships between the eight decision variables and the n... [more]

This study proposed a kind of optimization design for a reversible axial-flow pump based on an ordinary one-way pump. Three-dimensional (3D) Reynolds-averaged Navier−Stokes (RANS) equations was used to predict the pump performance, and the optimized design was validated by an external characteristic test. Six main geometry parameters of an impeller and diffuser based on an orthogonal experiment were set as design variables. The efficiency and head under forward and reverse design conditions were set as the optimization objective. Based on 120 groups of sample designs obtained from Latin hypercube sampling (LHS), a two-layer artificial neural network (ANN) was used to build a non-linear function with high accuracy between the design variables and optimization objective. The optimized design was obtained from 300 groups of Pareto-optimal solutions using the non-dominated based genetic algorithm (NSGA) for multiobjective optimization. After optimization, there was a slight decrease in the... [more]

Organic Rankine cycle (ORC) power systems have recently emerged as promising solutions for waste heat recovery in low- and medium-size power plants. Their performance and economic feasibility strongly depend on the expander. The design process and efficiency estimation are particularly challenging due to the peculiar physical properties of the working fluid and the gas-dynamic phenomena occurring in the machine. Unlike steam Rankine and Brayton engines, organic Rankine cycle expanders combine small enthalpy drops with large expansion ratios. These features yield turbine designs with few highly-loaded stages in supersonic flow regimes. Part A of this two-part paper has presented the implementation and validation of the simulation tool TURAX, which provides the optimal preliminary design of single-stage axial-flow turbines. The authors have also presented a sensitivity analysis on the decision variables affecting the turbine design. Part B of this two-part paper presents the first applic... [more]

Artificial neural networks (ANNs) are used in various applications for data-driven black-box modeling and subsequent optimization. Herein, we present an efficient method for deterministic global optimization of ANN embedded optimization problems. The proposed method is based on relaxations of algorithms using McCormick relaxations in a reduced-space [\textit{SIOPT}, 20 (2009), pp. 573-601] including the convex and concave envelopes of the nonlinear activation function of ANNs. The optimization problem is solved using our in-house global deterministic solver MAiNGO. The performance of the proposed method is shown in four optimization examples: an illustrative function, a fermentation process, a compressor plant and a chemical process optimization. The results show that computational solution time is favorable compared to the global general-purpose optimization solver BARON.