Integrated Design and Multi-Objective Optimization of a Single Stage Heat-Pump Turbocompressor

Small scale turbomachines in domestic heat pumps reach high efficiency and provide oil-free solutions which improve heat-exchanger performance and offer major advantages in the design of advanced thermodynamic cycles. An appropriate turbocompressor for domestic air based heat pumps requires the ability to operate on a wide range of inlet pressure, pressure ratios and mass flows, confronting the designer with the necessity to compromise between range and efficiency. Further the design of small-scale direct driven turbomachines is a complex and interdisciplinary task. Textbook design procedures propose to split such systems into subcomponents and to design and optimize each element individually. This common procedure, however, tends to neglect the interactions between the different components leading to suboptimal solutions. The authors propose an approach based on the integrated philosophy for designing and optimizing gas bearing supported, direct driven turbocompressors for applications with challenging requirements with regards to operation range and efficiency. Using previously validated reduced order models for the different components an integrated model of the compressor is implemented and the optimum system found via multi-objective optimization. It is shown that compared to standard design procedure the integrated approach yields an increase of the seasonal compressor efficiency of more than 12 points. Further a design optimization based sensitivity analysis allows to investigate the influence of design constraints determined prior to optimization such as impeller surface roughness, rotor material and impeller force. A relaxation of these constrains yields additional room for improvement. Reduced impeller force improves efficiency due to a smaller thrust bearing mainly, whereas a lighter rotor material improves rotordynamic performance. A hydraulically smoother impeller surface improves the overall efficiency considerably by reducing aerodynamic losses. A combination of the relaxation of the 3 design constraints yields an additional improvement of 6 points compared to the original optimization process. The integrated design and optimization procedure implemented in the case of a complex design problem thus clearly shows its advantages compared to traditional design methods by allowing a truly exhaustive search for optimum solutions throughout the complete design space. It can be used for both design optimization and for design analysis.

Improved Trust Region Based MPS Method for High-Dimensional Expensive Black-Box Problems

Mode Pursuing Sampling (MPS) was developed as a global optimization algorithm for optimization problems involving expensive black box functions. MPS has been found to be effective and efficient for problems of low dimensionality, i.e., the number of design variables is less than ten. A previous conference publication integrated the concept of trust regions into the MPS framework to create a new algorithm, TRMPS, which dramatically improved performance and efficiency for high dimensional problems. However, although TRMPS performed better than MPS, it was unproven against other established algorithms such as GA. This paper introduces an improved algorithm, TRMPS2, which incorporates guided sampling and low function value criterion to further improve algorithm performance for high dimensional problems. TRMPS2 is benchmarked against MPS and GA using a suite of test problems. The results show that TRMPS2 performs better than MPS and GA on average for high dimensional, expensive, and black box (HEB) problems.

Hierarchical Bayesian Parameter Estimation for Modeling and Analysis of User Affective Influence

Traditional user experience (UX) models are mostly qualitative in terms of its measurement and structure. This paper proposes a quantitative UX model based on cumulative prospect theory. It takes a decision making perspective between two alternative design profiles. However, affective elements are well-known to have influence on human decision making, the prevailing computational models for analyzing and simulating human perception on UX are mainly cognition-based models. In order to incorporate both affective and cognitive factors in the decision making process, we manipulate the parameters involved in the cumulative prospect model to show the affective influence. Specifically, three different affective states are induced to shape the model parameters. A hierarchical Bayesian model with a technique called Markov chain Monte Carlo is used to estimate the parameters. A case study of aircraft cabin interior design is illustrated to show the proposed methodology.

Building of Usage Scenarios Space for Investigating the Fall Situations of the Elderly People

Assessing the number of users defined by a set of specific usage attributes in a given usage contextual situation is not always an obvious task in a market segmentation process. Although new approaches in design and marketing seem to be more sensitive to the adequacy of a design concept with the usage scenarios, these methods do not systematically consider the various usage situations. The present article puts forward a methodology intending to build a usage scenarios space in which the input data is thoroughly collected and validated. This methodology is applied to the complex and multifactorial issue of falls among the elderly in the Metropolitan France. In this paper, numerous medical publications have been made to study influential factors of fall situations. However, even solution providers for fall prevention and teleassistance ignore the real situational coverage of their solutions. As a result, “usage scenarios space” is built using an appropriate segmentation of usage contexts (here, fall situations) and user characteristics. These data are used for a design oracle to predict (simulate) the various and multiple usage scenarios.

Product-Service Integration for the Sustainable Publishing Process

Publishing is the process of developing and producing content for distribution to the public. In the past, the publishing process heavily relies on printing as the method of content production. This causes voracious consumption and waste of natural resources. In today’s sweeping trend of digitization that is featured by the increasing popularity of various smart devices, the publishing process is undergoing a profound transition from the traditional printing-reliant publishing model to the new digital publishing model. Such a transition brings great opportunities for the publishing process to achieve better sustainability by evolving towards a product service system. This paper intends to advance the publishing process from the product service integration perspective. Above all, a general product-service integration framework is developed to describe the interdependent relationships among key stakeholders and elements in the publishing value chain. Furthermore, several specific publishing PSS are discussed. Finally, these publishing PSS are evaluated and compared from the value creation perspective.

Development of a Common Platform for Testing Metamodel Based Design Optimization Methods

Metamodel based design optimization (MBDO) algorithms have attracted considerable interests in recent years due to their special capability in dealing with complex optimization problems with computationally expensive objective and constraint functions and local optima. Conventional unimodal-based optimization algorithms and stochastic global optimization algorithms either miss the global optimum frequently or require unacceptable computation time. In this work, a generic testbed/platform for evaluating various MBDO algorithms has been introduced. The purpose of the platform is to facilitate quantitative comparison of different MBDO algorithms using standard test problems, test procedures, and test outputs, as well as to improve the efficiency of new algorithm testing and improvement. The platform consists of a comprehensive test function database that contains about 100 benchmark functions and engineering problems. The testbed accepts any optimization algorithm to be tested, and only requires minor modifications to meet the test-bed requirements. The testbed is useful in comparing the performance of competing algorithms through execution of same problems. It allows researchers and practitioners to test and choose the most suitable optimization tool for their specific needs. It also helps to increase confidence and reliability of the newly developed MBDO tools. Many new MBDO algorithms, including Mode Pursuing Sampling (MPS), Pareto Set Pursuing (PSP), and Space Exploration and Unimodal Region Elimination (SEUMRE), were tested in this work to demonstrate its functionality and benefits.

A Concept Selection Framework for Early Sorting of Reconfigurable System Designs

A seven-step framework for sorting proposed concepts of system changes / reconfigurations is presented that seeks to characterize the overall ramifications on system architecture. This framework is intended for use immediately following a concept generation phase. The framework uses three simple questions: “What level of the system design does this concept apply to?” “What levels of the system design does the concept impact?” and “What is the severity of this impact?” A flowchart leads the designer through these questions and assigns each concept a classification from one to five based on the answers. Class one concepts have little to no impact on the rest of the system architecture. They can be included with little fear of massive change propagation and system redesign. Class five concepts carry large changes to system architecture and therefore should be included only if they can be shown to be highly beneficial, or if there remains enough design freedom such that the cost of changing the system architecture is minimal. Meanwhile, class five concepts are likely to have much higher potential to create revolutionary design. A case study is used to demonstrate the application of the sorting framework in the context of a Mars rover mission. Several example concepts are provided to illustrate key insights from the case study. Convergence of the framework is explored by comparing the authors’ results to a second test done by a new design team.

Advanced Robust Optimization Approach for Design Optimization With Interval Uncertainty Using Sequential Quadratic Programming

Uncertainty is inevitable in real world. It has to be taken into consideration, especially in engineering optimization; otherwise the obtained optimal solution may become infeasible. Robust optimization (RO) approaches have been proposed to deal with this issue. Most existing RO algorithms use double-looped structures in which a large amount of computational efforts have been spent in the inner loop optimization to determine the robustness of candidate solutions. In this paper, an advanced approach is presented where no optimization run is required to be performed for robustness evaluations in the inner loop. Instead, a concept of Utopian point is proposed and the corresponding maximum variable/parameter variation will be obtained by just solving a set of linear equations. The obtained robust optimal solution from the new approach may be conservative, but the deviation from the true robust optimal solution is very small given the significant improvement in the computational efficiency. Six numerical and engineering examples are tested to show the applicability and efficiency of the proposed approach, whose solutions and computational time are compared with those from a similar but double-looped approach, SQP-RO, proposed previously.

A Maximum Confidence Enhancement Based Sequential Sampling Scheme for Simulation-Based Design

This paper presents a maximum confidence enhancement based sequential sampling approach for simulation-based design under uncertainty. In the proposed approach, the ordinary Kriging method is adopted to construct surrogate models for all constraints and thus Monte Carlo simulation (MCS) is able to be used to estimate reliability and its sensitivity with respect to design variables. A cumulative confidence level is defined to quantify the accuracy of reliability estimation using MCS based on the Kriging models. To improve the efficiency of proposed approach, a maximum confidence enhancement based sequential sampling scheme is developed to update the Kriging models based on the maximum improvement of the defined cumulative confidence level, in which a sample that produces the largest improvement of the cumulative confidence level is selected to update the surrogate models. Moreover, a new design sensitivity estimation approach based upon constructed Kriging models is developed to estimate the reliability sensitivity information with respect to design variables without incurring any extra function evaluations. This enables to compute smooth sensitivity values and thus greatly enhances the efficiency and robustness of the design optimization process. Two case studies are used to demonstrate the proposed methodology.

Using a Goal-Switching Selection Operator in Multi-Objective Genetic Algorithm Optimization Problems

This paper demonstrates how solution quality for multiobjective optimization problems can be improved by altering the selection phase of a multiobjective genetic algorithm. Rather than the traditional roulette selection used in algorithms like NSGA-II, this paper adds a goal switching technique to the selection operator. Goal switching in this context represents the rotation of the selection operator among a problem’s various objective functions to increase search diversity. This rotation can be specified over a set period of generations, evaluations, CPU time, or other factors defined by the designer. This technique is tested using a set period of generations before switching occurs, with only one objective considered at a time. Two test cases are explored, the first as identified in the Congress on Evolutionary Computation (CEC) 2009 special session and the second a case study concerning the market-driven design of a MP3 player product line. These problems were chosen because the first test case’s Pareto frontier is continuous and concave while being relatively easy to find. The second Pareto frontier is more difficult to obtain and the problem’s design space is significantly more complex. Selection operators of roulette and roulette with goal switching were tested with 3 to 7 design variables for the CEC 09 problem, and 81 design variables for the MP3 player problem. Results show that goal switching improves the number of Pareto frontier points found and can also lead to improvements in hypervolume and/or mean time to convergence.