EngOpt 2018 Proceedings of the 6th International Conference on Engineering Optimization

Multi-objective optimization has become an invaluable tool in engineering design. One class of solutions to the multi-objective optimization problem is known as the Pareto frontier. The Pareto frontier is made up of a group of solutions known as Pareto optimal solutions. These solutions are optimal in the sense that any improvement in one design objective must come with the worsening of at least one other. Therefore, the Pareto frontier plays a vital role in engineering design, since it defines the trade-offs between conflicting objectives. Methods exist that can automatically generate a set of Pareto solutions from which the final design can be chosen. For such an approach to be successful, the generated set must truly be representative of the complete design space. This paper offers a new phase in the development of the smart normal constraint bi-directional evolutionary optimization method, which is a recently developed approach that allows the efficient and effective generation of smart Pareto sets to multi-objective topology optimization problems. Currently, only bi-objective topology optimization problems can be solved with this method. Therefore, in this paper the method is generalized to solve topology optimization problems with any number of objectives. This is demonstrated on an example having three objectives.

The use of evolutionary computational techniques has become widespread in many technical disciplines including, but not limited, neural networks and evolutionary algorithms. From these techniques, in the field of global optimization, mainly the evolutionary optimization algorithms are used, especially one of their types – genetic algorithms. From the mathematical point of view, the evolutionary and genetic algorithms are just another representatives of stochastic optimization algorithms. The aim of our research was to describe the basic properties of stochastic algorithms including genetic algorithms, to select suitable candidates from the class of traditional stochastic algorithms and to compare their behaviour with the genetic algorithms. In this paper, we are going to address so-called technical optimization, where we do not know the optimized function directly, but we are able to get the value of an optimized function at any point (for example by measuring a certain quantity). The stochastic optimization algorithms provide the advantage of efficient working even with such functions. An important criterion for optimization is also the ability to parallelize a task. The optimization algorithms can be implemented as a parallel system – we calculate the value of a purpose function at several points at the same time. The paper will also describe the specific described implementation and testing of selected algorithms on analytical functions as well as functions mediated by artificial neural networks, which have been learned on practice data. Furthermore, the algorithm implementation for different environments and their routine user-friendly practical applications are described. The aim of our research was also to select those representatives of traditional stochastic algorithms that would be able to compete with the genetic algorithms by their accuracy or speed, to implement these algorithms and to test them on specific data. Last but not least, the results of testing of each algorithm on the practice data will be presented and, in the final phase, these results will be analysed.

In this research, a combination of global-local optimization with the possibility of multi-objective trade-off solutions is considered for vehicle configuration design. At global-level, many potential configurations are created as initial designs using cellular-division method for design space exploration. At local-level, level-set method optimizes these initial shapes for strict constraint satisfaction. A combination of these two approaches with their individual strengths are synergistically integrated for evolving a configuration from an open-ended design space for better optima. This research demonstrates the framework using several benchmark problems of topology optimization.

Many research efforts have been recently focus to solve large-scale global optimization (LSGO) problems by means of evolutionary algorithms. Cooperative co-evolution has been proposed to solve such problems depending on thousands of variables. This methodology has proved very efficient in solving a wide range of LSGO problems. Nevertheless, it often requires an extremely large number of function evaluations to reach a suitable solution. This is somewhat problematic when the function evaluation is computationally expensive. A globally effective approach to high-fidelity optimization problems based on such expensive analyses lies in the exploitation of surrogate models. They act as cheap-to-evaluate alternatives to the original high-fidelity models reducing the computational cost, while still providing improved designs. This kind of optimization process, referred to as surrogate-assisted optimization, has proved very efficient on small-dimensional problems but suffers from the curse of dimensionality to solve LSGO problems. In this paper, cooperative co-evolution was combined with surrogate-assisted optimization in order to efficiently solve high dimensional, expensive and black-box problems. Experimental results are provided on a wide set of benchmark problems and show promising results for the proposed algorithm.

Bolted connections are important components of steel structures and they must be based on good design, with analytical descriptions of failure mechanisms that are thoroughly researched. A robust and reliable analytical description of a failure mode requires a large base of results, collected experimentally or numerically, usually via FEA. The modelling procedure in FEM simulations can significantly influence result accuracy, and one of the most important parameters is spatial discretization. Mesh size presents a delicate balance between accuracy, efficiency and convergence issues, especially in bolted connection analysis with contact, material and geometric nonlinearity. Due to increased complexity, sensitivity studies are difficult to interpret and time expensive, which is why in this study a mesh size optimization procedure is presented. This optimization is based on a previously verified and validated numerical model parametric study of a single-bolt lap connection. The first step included forming of an optimization model based on an artificial neural network, through which results were extrapolated and compared to experimental. However, estimated solving wall time was significantly longer. In the second step, a nonlinear optimization model was structured to determine a mesh size for which the difference between experimental and optimization results would be equal, whilst having a minimal solving wall time. The optimization model provided insights into mesh size quality to efficiency ratios and rendered that the minimal difference between experimental and numerical results is 1.57%, while having a solving wall time of 1.3 h, instead of the step one optimization modelling estimated 36 h.

The paper is devoted to the problem of trip demand estimation in a road network. Commonly, when solving trip demand estimation problem researchers suppose the presence of so-called a prior origin-destination matrix. Unlike such an approach we assume that the only input data for trip demand estimation problem is traffic load on arcs. Thus, in this paper we intend to avoid using a prior origin-destination matrix for trip demand estimation and show that in such a case one is faced with the problem of combinatorial optimization. Computational complexity of appeared problem is discussed. Heuristic procedure for solving the problem is proposed and it is applied to the test example.

Optimal topologies obtained for structures subjected to deterministic loading can be sensitive to loading variations in terms of both magnitude and direction. Therefore, in this study we consider problem of topology optimization for structures subjected to probabilistic loading. The proposed method applies basic findings from probability theory, which allow to transform the original problem of topology optimization under single probabilistic loading into analogous problem of topology optimization under multiple deterministic loading cases. After recalling the theoretical background of the method,’ its effectiveness is demonstrated on an examples of cantilever structure subjected to horizontally oriented load with randomly varying angle of action.

The data fitting problem of approximating a function in several variables, given by tabulated data, and the corresponding problem for overdetermined (inconsistent) systems of linear algebraic equations are considered. Such problems for measurement of quantities arise, for example, in engineering, physics, etc. A classical approach for solving these two types of problems is the discrete least squares data fitting method. An alternative approach for solving these problems is proposed in this paper: a nonsmooth (nondifferentiable) unconstrained minimization problem is associated with each of these problems, with an objective function, based on discrete absolute deviation norm and/or uniform norm (sup-norm), respectively, that is, the problems under consideration are solved by minimizing the residual using these two norms. Subgradients for the considered problems are calculated, and a subgradient method is applied for solving these problems. Some numerical results, obtained by an appropriate iterative method, are presented, and these results are compared with the results, obtained by the iterative gradient method for the corresponding “differentiable” least squares problems.

Metaheuristic optimization algorithms are characterized by stochastic behavior, making each optimization run unique. In addition, metaheuristic algorithms are usually governed by a number of control parameters that require problem-specific tuning. Many publications on metaheuristic algorithms lack the kind of rigorous statistical convergence assessment that is needed to compensate for the above uncertainties, making it impossible to assess the optimality of the resulting design or the effectiveness of the optimization method. In this contribution, we propose a method to assess the performance (i.e. the ability to find the best known solution and the associated computational cost) of a metaheuristic algorithm that takes into account the influence of its control parameters. First, a large number of simulations (independent optimization runs) are performed, where the values of the control parameters are randomly selected from predefined sets of realistic possibilities. Next, for every value of every control parameter, the corresponding subset of simulations is considered in order to infer the relevant conditional performance statistics. As a example, the approach is used to assess the performance of the genetic algorithm built into matlab for a 25-bar truss test case. It is observed that, for the algorithm and the test case considered, the majority of the control parameters have little influence on algorithm performance.

Recently, the structural optimization has received a strong emphasis that leads in the formulation of the objective function questions regarding the possible combination of various materials. That is, the multi- material optimization in which these materials present different characteristics between them. For example, those referring to the behavior of the material that can be physically linear or non-linear, linear behaviors with different modulus of elasticity, different costs depending on the volume to be used, different behaviors in tension and compression, and so on. The topological structural optimization, particularly, has been receiving efforts in this direction and is extremely adequate to address this type of problem. Another issue in this process is to include the possibility of limiting the number of different materials to be used in the optimized final design. The objective of this paper is to propose a strategy to obtain solutions for structural optimization problems in sizing, shape and topology, where the use of different materials will be incorporated in the formulation of the problem, besides the possibility of the designer choosing the maximum number of these materials. The search algorithm to be used is the Differential Evolution and the control of the maximum number of materials to be used is done through the use of cardinality constraints.

LNG hydraulic turbines used as the replacements of J–T valves play an important role in the LNG industry chain. In this study, an optimization method for hydraulic turbine impeller was established based on experimental design theory, the response surface approximation and genetic algorithm. The external characteristics are obtained by the commercial software ANSYS FLUENT. The efficient and head were chosen as objective functions, and the response surface method (RSM) and NSGA-II genetic algorithm are used to obtain the optimal combination of the parameters of the impeller. The head and hydraulic efficiency of the optimized model rise by 1.5 and 3.2%, respectively.

Search and rescue of people in need after catastrophic events is dangerous and challenging for the rescue personnel. Limited resources and reduced accessibility to danger zones even prevent rescue attempts at all. New technologies to assist and keep safe the search and rescue teams while maintaining maximal success rate are therefore desired.In this work, rescue drones are considered as a swarm of autonomous agents. Each agent follows a certain search protocol and decides from available information which action it will take, for instance delivering first aid kits and utilities or report locations. Agents in the swarm furthermore have a certain ethical protocol that determines to what extent they react to newly discovered victims. The optimal swarm ethics with respect to efficiency is discussed and limits and benefits examined.

Technical components are usually well optimized. However, simply combining these optimized components in a technical system does not necessarily lead to optimal systems. Therefore, focusing on a system perspective reveals new potential for optimization. In this context, we examine thermofluid systems which can be interpreted as fluid systems with superimposed heat transfer. The structure of such systems can be abstracted as a graph – more specifically, a flow network. We translate the underlying optimization problem into a mixed-integer linear program which is designed to obey the physical laws of heat transfer. Typically, fluid systems can be considered as quasi-stationary systems since their dynamic effects are usually negligible. However, for thermofluid systems this assumption does not hold because time-dependency is an issue as storage tanks for heated fluid gain importance. In order to handle the dynamic effects induced by the storage tanks, we further introduce a continuous-time representation based on a global event-based formulation.

Simulation is a helpful tool for analysing and optimising certain problems in various different scenarios. In urban area, traffic simulation could help optimising the placement of the traffic lights or traffic signs. Typically, the simulation based mostly on the linear approximation of the distance and speed. However, drivers’ behaviour could also influence the traffic situation resulting in traffic jam which lead to delay on the arrival time on other drivers. In this work, a simulation framework that focusses on the drivers’ behaviour is presented. Each driver is modelled as an agent which has some driving attitude. The simulation model a daily activity of each agent, where it has a place to live and an office to work. The agents will then do their daily activities which can be an ordinary day or some free day by visiting friends, cinema or shopping center. Based on the daily activity situation the driving style will be influenced from a patient driver which drives carefully to an angered driver that tends to become more careless.

An efficient design method is presented for flow devices with solid structure arrays. It couples macro-scale Navier-Stokes equations with a topology optimization method to optimize the size of individual structures in an array subject to the fluid flow. The presence of structures in the flow is represented by an interfacial force and porosity. The methodology is applied to two optimization problems in flow devices. Intermediate and optimized interfacial force and porosity fields are mapped onto an equivalent physical solid structure distribution. Next, direct numerical simulation (DNS) is performed with the solid structure distribution meshed explicitly in the grid in order to validate the macro-scale modeling approach. A comparison of the cost function values evaluated using both the macro-scale and DNS data shows that macro-scale models can correctly predict trends in cost function value evolution throughout optimization, but there is an offset between the values themselves.

In this work, the synthesis of polyacrylamide based multicomponent hydrogels is modelled using a neuro-evolutive technique which combines Differential Evolution (DE) algorithm and Artificial Neural Networks (ANNs). The neural model is developed in an optimal form applying the DE procedure, meaning the optimization of the weights, biases and architecture (number of hidden layers, neurons in each hidden layer and their activation functions). The data set used in modelling action contains experimental data obtained in our laboratory. Based on the neural model, the process is further optimized with DE algorithm in order to determine the optimum conditions of the synthesis (time, temperature, monomer concentration, initiator, crosslinking agent, inclusion polymer, and type of the polymer added) leading to the maximization of the reaction yield. The optimization of the process, using the stack neural model, provides the working conditions that lead to a reaction yield of 95%.

Multi-objective control with generalized $$\mathcal {H}_2$$ H 2 -norms as criteria is applied to designing optimal isolators for protection of elastic mechanical structures from vibration. Pareto optimal isolators are derived analytically for a single-degree-of-freedom system and synthesized numerically for multi-degree-of-freedom structures by using linear matrix inequalities. Numerical results demonstrate the considerable improvement of protection by means of these optimal isolators.

Offshore wind turbines are subjected to harsh environmental and operational conditions that affect their dynamic properties and cause damage. Visual checks are, for economic reasons, kept as low as possible, thus making the ability to detect damage via transmitted measurements a vital issue.Identifying a structure is considered in essence an inverse problem which can be solved using model-updating techniques, which treat the identification problem as an optimization problem that are well-solved using meta-heuristic optimization schemes.The objective of this study is to investigate the performance of the harmony search algorithm, both basic and modified, in identifying a scaled laboratory model of an offshore wind turbine supporting structure and detect the effects of damage and marine growth. The laboratory model is tested in a wave basin and is subjected to a variety of damage and marine growth levels.

Considering fail-safe requirements in a topology optimization, where the location of the damage is unknown in advance, leads to a high number of potential damage scenarios to be calculated. Since this is driving the overall calculation time of the optimization, a reduction of the considered damage cases is desirable. In this paper, two strategies to achieve a significant reduction of damage cases are shown: An active-set strategy as an extension to the simplified local damage model first introduced by Jansen et al. as well as a newly developed load-path based algorithm for the placement of damage zones.

Tube-fin heat exchangers (HXs) are widely used in air-conditioning and heat pump applications. The performance of these heat exchangers is strongly influenced by the refrigerant circuitry, i.e. the refrigerant flow path along the different tubes in the HX core. Since for a given number of tubes, the number of possible circuitries is exponentially large, neither the exhaustive search nor traditional optimization algorithms can be used to optimize the circuitry for a given application. Researchers have previously used Genetic Algorithms (GAs) coupled with a learning module or other heuristic algorithm to solve this problem, but there is no guarantee that the resulting circuitry can be manufactured in a cost-effective manner. In this paper, we present an integer permutation-based GA approach for solving the circuitry optimization problem. A finite volume heat exchanger simulation tool is used to simulate the performance of different circuitries generated by the optimizer. The novel genetic operators are designed such that all chromosomes generated by GA can be mapped to a valid circuitry design. As a result, the proposed approach can explore the solution space more efficiently than a conventional GA. The manufacturability aspect is handled using a constraint-dominated sorting technique in the fitness assignment stage. And a hybrid initialization scheme is developed to improve the feasibility of initial individuals. Exhaustive search on small heat exchangers has proved that the proposed Integer Permutation-based GA(IPGA) can find optimal or near-optimal refrigerant circuitry designs using a relatively low population size and iterations. The analyses of several case studies show that the constrained IPGA can generate circuitry designs with capacities superior to those obtained manually and meanwhile guarantee good manufacturability. Overall, a 2.4–14.6% increase in heat exchange capacity is observed by applying the new optimization method to an evaporator from a A-type indoor unit.

Evolutionary optimization approaches such as genetic algorithms are being used increasingly in the optimization of water distribution networks. The number of hydraulic simulations required to find good solutions can be extremely large and time-consuming. It is, therefore, highly desirable to reduce the search or solution space to speed up the optimization process. The new approach herein considers the importance of every path through the network, an intrinsic property of the statistical flow entropy function. Pre-processing, setting or initialization of the reduced solution space is not required a priori. Instead, the reduced solution space is determined adaptively using maximum entropy principles. The methodology comprises two main phases. In the first phase, the entire solution space is explored until a feasible solution is identified. In the second phase, exploitation is effected by means of a reference solution that is updated in every generation. The reduced set of pipe diameter options considered in each generation in the second phase is defined relative to the reference solution. The algorithm was applied to a benchmark network. The solutions obtained were generally less expensive for similar entropy values than solutions from the full solution space. The results revealed that the solution space reduction algorithm limits the search to the areas close to the feasibility boundary. Consistently good results were achieved in terms of the quality of the solutions and computational efficiency.

This paper summarize our work so far on reliability based design optimization (RBDO) by using metamodels and present some new ideas on RBDO using support vector machines. Design optimization of complex models, such as non-linear finite element models, are treated by fitting metamodels to computer experiments. A new approach for radial basis function networks (RBFN) using a priori bias is suggested and compared to established RBFN, Kriging, polynomial chaos expansion, support vector machines (SVM), support vector regression (SVR), and least square SVM and SVR. Different types of computer experiments are also investigated such as e.g. S-optimal design of experiments, Halton- and Hammersley sampling, and different adaptive sampling approaches. For instance, SVM-supported sampling is suggested in order to improve the limit surface by putting extra sampling points at the margin of the SVM. Uncertainties in design variables and parameters are included in the design optimization by FORM- and SORM-based RBDO. By establishing the most probable point (MPP) at the limit surface using a Newton method with an inexact Jacobian, Taylor expansions of the metamodels are done at the MPP using intermediate variables defined by the iso-probabilistic transformation for several density distributions such as lognormal, gamma, Gumbel and Weibull. In such manner, LP- and QP-problems are derived which are solved in sequence until convergence. The implementation of the approaches in an in-house toolbox are very robust and efficient. This is demonstrated by solving several examples for a large number of variables and reliability constraints.

We present a new feasible direction interior point algorithm to compute a numerical solution of a normalized Generalized Nash Equilibrium Problem, (GNEP). The GNEP is an extension of the Nash Equilibrium Problem, (NEP). This one is an equilibrium problem that involves two or more players, each player is associated with a feasible strategy set and a payoff function. We assume that there is not collaboration among players. However, in the GNEP the feasible set depends on the strategies of the others players. The numerical method presented in this paper to compute an equilibrium point of the normalized version of the GNEP is a feasible direction Newton method to solve the necessary conditions characterizing the solution. Given an initial point at the interior of the feasible region, the present algorithm generates a feasible sequence converging to the solution of the normalized GNEP problem.The presented approach was tested on a collection of problems found in the specialized literature. The results suggest that our method is strong and efficient.

Discontinuous Petrov–Galerkin (DPG) methods constitute a modern class of finite element methods, which present several advantages when compared with traditional Bubnov–Galerkin methods, especially when the latter is applied to indefinite or non-symmetric problems. Our objective is to utilize the advantages of DPG methods in the context of topology optimization.The direct application of DPG discretizations to BVPs arising in topology optimization is hindered by the very unusual scaling of the residual, caused by the gigantic jumps in the coefficients of the governing differential equations. In the prototypical case of linearized elasticity with SIMP model the coefficient ratio between the “stiff” and “soft” phases is held at a billion, which is further squared by Petrov–Galerkin methods based on minimizing the squared residual.We introduce a DPG method with appropriately scaled residual norm, which allows us to deal with big contrast ratios in the coefficients. The method is tested on benchmark topology optimization problems.

We propose an algorithm for minimizing a functional under constraints. It uses first order derivatives of both the objective function and the constraints. The step is computed as a sum between a steepest descent step (which minimizes the objective functional) and a correction step related to the Newton method (which aims to solve the equality constraints). The linear combination between these two steps involves coefficients similar to Lagrange multipliers which are computed in a natural way based on the Newton method. The algorithm uses no projection and thus the iterates are not feasible; the constraints are only satisfied in the limit (after convergence). Although the algorithm can be used as a general-purpose optimization tool, it is designed specifically for problems where first order derivatives of both objective and constraint functionals are available but not second order derivatives (as is often the case in structural optimization).

Natural gas and sulfuric acid are two of the most widely consumed products around the world. During the production of sulfuric acid and natural gas dehydration, several gas emissions are released to atmosphere and are negatively affecting the environment. In addition, during the production, the plant units are subjected to the risk of explosion and fire. In order to solve these environmental and safety problems, both processes need to be considered for multi-objective optimization (MOO) with respect to the specified objectives. For natural gas dehydration, the considered bi-objective case is the simultaneous minimization of fire and explosion damage index (FEDI) and photochemical ozone creation potential (POCP). ProMax® 4.0 was used to simulate the process with excel-based NSGA-II algorithm for optimization (EMOO). Many decision variables were considered like flow rates, pressure and temperature of inlet streams. It was found that there exists a potential for improving the process. For the sulfuric acid plant, Aspen Plus was used for simulation with EMOO for minimization of acidification potential (AP) and maximization of the product sales (PRS). Decision variables like columns and reactors operating pressure and raw materials flow rates were considered. Results showed that there is a MOO for this process with respect to the safety and environmental point of view and that process is mainly influenced by the steam flow rate that is used for gas cooling purposes.

The paper presents an HPC implementation of the Multipoint Approximation Method (MAM) applied to problems with uncertainty in design variables as well as in additional environmental variables. The approach relies on approximations built in the combined space of design variables and environmental variables, and subsequent application of a risk measure and optimization with respect to the deterministic design variables, all within the iterative trust-region-based framework of MAM.

In this article, numerical methods for solving engineering problems defined as multicriteria optimization and inverse problem are presented. It deals with the optimization of the design of thermoacoustic engine in the frame of which both types of tasks are solved. In the first step, a populational heuristic is used to find many p-optimal solutions simultaneously, which represents a compromise between usually mutually contradictory goals at work. Based on them, the full Pareto front is approximated. The inverse problem solution reproduces parameters for solutions located on a designated front but those that are not found in multicriteria optimization. In this article, it is proposed to use the RACO heuristics for determining p-optimal solutions and the Bayesian approach as a method for solving ill-conditioned inverse problems. Optimization of the construction of the thermoacoustic engine is aimed at verifying proposed methodology and present the possibility of using both methods in engineering problems. The problem discussed in this article has been formulated and the numerical methods used in the solution have been presented in details.

Facilities installation have been a subject of interest in the production engineering, business management, communities of operational research, transport engineering and logistics companies. The studies of facility location problems deal with questions of minimization of costs in the logistics chain. In many situations, these problems are subject to limitations in the capacity of the facility. Due to these restrictions, logistics demands must be satisfied in order to comply with certain financial interests and meet the service level agreements. In this context, this work aims to evaluate the doubly reactive GRASP procedure integrated with the path-relinking technique, designed to solve the generalized uncapacitated p-medians location problem, using the balanced-sized clusters creation approach, in order to verify the impact of this balance in the context of the total cost of transportation service. The p-median problem refers to locate simultaneously the facilities at different areas, in order to minimize the total transportation distance (and therefore the cost), among each distribution center and the facilities allocated in a certain cluster, for satisfying the demands of its customers. Two reaction parameters are used to control the search for solutions in the GRASP construction phase. The simultaneous use of the two reactive parameters allows the creation of a discipline for the allocation of the customers to the clusters: in general, the nearest customers are allocated first. It was also investigated a relation between the size of the clusters and the median vocation index of each median, which are the centers of distribution of each cluster. For the evaluation of the algorithm and the effectiveness of the balancing of the clusters and the median vocation index, the data concerning the distances among all the provinces of Spain were used.

The conventional systems of reorientation of satellite tracking antenna in portable satellite communications stations are based on serials robots, serial robots have three degrees of freedom and are composed of several kinematic chains joined by rotational articulations. These characteristics can affect the performance of satellite tracking since they limit their movements. For this reason, it is proposed a Stewart platform towards the reorientation system that can accept accelerations and higher speeds during its movement, support large loads that have a higher rigidity and precision to be considered by six actors in parallel and finally, have a better follow-up of your six degrees of freedom. The demand for precision for satellite tracking is high, to determine the accuracy of the Stewart platform, a sensitivity analysis is performed that requires information on the design parameters that must be considered to perform a task with precision. Through the numerical simulations with the Matlab® software, it was found a range for maximum tolerances, varying the geometric parameters and varying the definition of the design applications to satisfy the demand of error tolerance of +/– 0.2° in orientation of portable reorientation systems for tracking satellites.

This paper deals with the monitoring of traffic flow and air pollution on an urban road. Specifically, the location of monitoring stations is studied, looking for points where obtained measures can be representative of the surrounding areas. In order to do it, a 1D mathematical model for obtaining the traffic flow on an urban road network is combined with a 2D model for air pollution. From the numerical estimations of these parameters, the problem of designing the monitoring strategy is formulated as a Mixed Integer Multiobjective Optimization Problem (MIMOP), which is solved by an ad-hoc procedure. Finally, this technique is applied to a simplified but realistic situation in the Guadalajara Metropolitan Area (Mexico).

In this work we deal with the optimization of different issues related to heavy metals phytoremediation techniques, by combining mathematical modelling, optimal control of partial differential equations and numerical optimization. We introduce a 2D mathematical system of nonlinear partial differential equations modelling the concentrations of heavy metals, algae and nutrients in large waterbodies. We formulate an optimal control problem related to the optimization of the phytoremediation process, and propose a full algorithm for computing the numerical solution of the problem. Finally, we present several numerical results for a realistic problem related to: (i) determining the minimal quantity of algae to be used in the heavy metals remediation process, and (ii) locating the optimal place for such algal mass.

This paper focusses on topology optimization for additive manufacturing. Two manufacturing constraints are considered: minimum length scale and maximum overhang angle. The first is needed to ensure that the condition on minimal printable feature sizes is satisfied, while the second eliminates the need for a temporary support structure. Filtering schemes have been proposed in the literature to ensure either a minimum length scale or a maximum overhang angle, but not both simultaneously. In this paper, it is shown that both constraints cannot simultaneously be met by simply applying both filters sequentially, as the effect of the first filter is destroyed by the second. A new, slightly more complex filtering scheme is therefore proposed, which does allow simultaneous control over length scale and overhang angle in a minimum compliance topology optimization problem. The method is successfully applied to a 2D benchmark problem.

The concept “Topology Optimization for Additive Manufacturing” is a recently coined concept that refers to the complete engagement of topology optimization problems and additive manufacturing processes. The idea of coupling both technologies through a specific overhang constraint lies in the idea of a total design freedom, which classic manufacturing processes are unable to reach. If any design can be build, this will enable a continuum “design + manufacturing” process eliminating prost-processing and any interference with the optimized geometry. In the field of structures there are already some ground-breaking approaches, there aren’t however any regarding the optimization of compliant mechanisms. The introduction of the overhang constraint within the topology optimization formulation of compliant mechanisms yields a compromise or inverse relation of functionality and manufacturability. The hinges of the flexible mechanisms are formed by a sharp thinning of the material members and describe a shape that possesses many tangents with different slopes, some of them showing not self supported contours. There is an inverse relation there for functionality and manufacturability. If the hinge is to be corrected so that a direct 3D printing of the mechanisms is possible, the global objective function will be harmed as the optimum-functional shape of the hinge is set aside. This paper introduces the advantages of a flexible overhang constraint for a more accurate topology optimization of 3D printed compliant mechanisms enabling intermediate design for different manufacturability/functionality ratios, and analyses the consequences of fully restricting scaffold structures respect to controlling and reducing them.

The working environment for Helium Circulator in HTR (high temperature gas-cooled reactor) is high temperature about 523 K and high pressure about 7 MPa. So the efficiency of Helium Circulator’s impeller is extremely important. At present, the blade loading parameters and meridional geometry parameters are obtained through the 3D inverse design methods and the widely used Bessel curves. The optimization design strategy was built by combining the 3D inverse design method, CFD analyses, Design of Experiment (DOE), Response Surface Methodology (RSM), Multi-Island Genetic Algorithm (MIGA). The optimization objective is the impeller efficiency at the design point. The input parameters are related to the blade loading, blade lean angle and the meridional channel shape. The results show that the SLOPEH, NCH, Ɵ1, and YH and Ɵ2 are the main influencing factors. The best blade loading distribution is the positive SLOPEH of high value combined with the negative SLOPES. The positive DRVT on the hub and shroud will improve the blade loading at the blade leading edge. The curve with the relative reduced curvature is beneficial to the performance of the impeller. The efficiency of the optimized impeller is 0.77% at Qd and 2.80% at 0.75Qd higher than the original impeller.

Stability is an important issue in topology optimization, since results of the optimization are often framework structures. If some trusses of these structures are subjected to compression, they maybe buckle and the structure fails.A brief review about literature on topology optimization considering linear buckling is given. That includes the material interpolation for linear buckling analysis to avoid spurious modes, mode switching and duplicated eigenvalues.In this contribution a continuously differentiable material interpolation scheme is explained to avoid spurious modes. It is shown how to cope with several modes (mode switching, duplicated eigenvalues) and how to use buckling safety as objective or constraint. The optimized structures are compared with results for compliance design. It is shown, that it is not useful to use buckling and mass as the only structure responses, because substructures subjected to tension can become very thin without a negative influence on the buckling safety. Thus tensile stresses are not limited. Buckling and mass have to be combined with other structure responses, like stress or compliance, to achieve a useful structure. Also the combination with manufacturing constraints for deep drawing is discussed in this contribution. Therefore 3D structures with up to a million design variables are presented.

Uncertainty quantification of identified parameters is an important feature when some quality assessment of the results of model updating procedure is necessary, or when important decisions depend upon these values. In this work, a modification of the conventional sensitivity method is tested along with a Bayesian Monte Carlo framework for identification of system parameters from experimental data, and their probability distributions. First, the updating procedure uses a metaheuristic algorithm (derivative-free) and the Euclidean norm metric. Then, a modification of Markov Chain Monte Carlo method called Transitional MCMC is applied to obtain an approximation of the mean values and probability distributions of the updated parameters based on the scattering of the experimental data. An example is presented with real structure experimental data for updating discrete mass, stiffness and damping parameters, as well as a comparison with previous results yielded by different methods, suggesting equivalent levels of agreement in the updated parameters, but with the advantage of MCMC formulation being practically independent of parameters vectors.

It is common for aircraft design studies to begin with low-fidelity tools and move to higher-fidelity tools at later stages. After early conceptual design stages, designers can take advantage of developments in high-fidelity aerodynamic shape optimization, and more recently, coupled aerostructural optimization to improve their designs. Over the past few years, our research group has developed a framework that allows carrying out high-fidelity aerostructural optimization by coupling a RANS CFD solver to an FEM solver that uses shell elements. In addition, we have recently developed OpenAeroStruct, a light-weight and open-source tool for low-fidelity aerostructural optimization that couples a VLM code to an FEM code that uses spatial beam elements. Due to their low cost, such low-fidelity tools remain useful for design studies. In this paper, we present results from OpenAeroStruct for the optimization of a transport aircraft wing and compare them to results from our group’s high-fidelity framework. Additionally, we describe the simplified wingbox model developed and implemented with OpenAeroStruct for this work.

Hereby are, firstly, presented two global optimized (GO) shapes of two models, Fadet I and Fadet II, which are of minimum drag at cruising Mach numbers 2.2 and, respectively, 3. Her non-classical three-dimensional hyperbolic potential solutions are used as start solutions. The design of GO shape of each FC, lead to an enlarged variational problem with free boundaries, which is solved by using her optimum-optimorum strategy. These two GO models are the source of inspiration of the design of two GO aerospace Catamaran models Cataf I and Cataf II with two twin fuselages, which are almost embedded in the wing thickness are concurrent and are located nearby the leading edges. They meet together in the frontal part of the wing, behind a rescue bar. If a greater and slower GO aerospace model Cataf 1 is considered as a geostationary one and a smaller more rapid GO aerospace model Cataf II as a suborbital model, which up and go on the central upper side of Cataf II, these two new GO vehicles can be useful for a new GO variant of the Saenger project. Further, the non-classical hyperbolic solutions presented here, are used to build reinforced solutions for the Navier-Stokes layer and for the computation of the total drag coefficient of the FCs. A refinement of the design strategy, in form of an iterative optimum-optimorum theory, is also proposed.

Fused deposition modelling (FDM) is an Additive manufacturing (AM) process where the part tensile strength depends on process parameters like layer thickness, part build orientation and infill density to name a few. Layer thickness is an important parameter and in this work, experiments were conducted using specimens build as per ASTM D638 and ASTM D695 specification to find the effect of layer thickness on the tensile strength and compressive strength respectively. A constitutive model in finite element (FE) based on classical laminate theory was then developed using the experimental data that considers the layer effect. An optimization framework was built using this FE model to find optimal layer thickness for each layer without changing the total layer count. The design variables were the layer thickness and the objective function was chosen to minimize the displacement with dimensional constraint for the given loading condition. Sequential quadratic programming was used for search. Example case studies are presented to illustrate the methodology and results.

The aim of this paper is to develop a simplified geometry representation of a generic engineering shape (initially represented by a B-spline surface or B-spline curve) by a simple shape description aiming at application in engineering shape optimization with objective of minimizing the number of shape variables and reducing the shape complexity for potential easier manufacturing.First step in achieving this objective is the analysis of a simple shape which can be described by a B-spline curve and approximated by a single circular arc. The analysis was conducted on an example of shape optimization for Savonius-like vertical axis wind turbine (VAWT). Initially, the shape was described by a B-spline curve and the shape optimization process which includes CFD simulation was conducted to maximize the VAWT power coefficient. The shape parameterization was subsequently changed in order to reduce the parameter set. The performance of the initial design (B-spline) is compared to the shape obtained by the simplified geometry. To test the possibilities of reducing the parameter set in 3D shape we used a point cloud obtained from a 3D scanning of Savonius-like blade. Simple shapes and the B-spline surface were fitted to the same blade and the results were compared.Overall, it was shown that the usage of the simplified shape with reduced parameter set can significantly reduce the number of shape variables for 2D shape without significant loss in the performance of the final optimal result. However, complex 3D shape cannot as easily be described by a simple analytical function.

For structure design mainly parametric optimization is used. But in numerous projects it raises the problems with structure quality evaluation. There are problems with quality criteria formalization. Designer must help in evaluation process with subjective information. To perform it, he must have an overview of structure variants. It is not possibly to use only parameterized structure value function (global quality criterion). To solve this problem, object – relational database structures variation space is used. And the optimization process is moving from starting structure to next “closer” structure in iterative manner. Adaptive multicriterial global optimization algorithm is used. Prototype of such system has been implemented.

Thickness constraint is an important geometrical constraint in topology optimization methods. I present a novel approach of the thickness constraint based on the Fictitious Physical Model (FPM). The FPM is formulated using the similarity of the dispersive coefficient in high order homogenization. The thickness constraint is represented using the solutions of the linear partial deferential equation system. Its design sensitivity is derived using the adjoint variable method. Numerical example is shown to confirm the validity and utility of the proposed method using the level set-based topology optimization method. The main advantage of the proposed method is the allowance of thickness constraint violations during the optimization procedure. Furthermore, the thickness is computed without computing minimum distances from the boundaries of target shape.

The aim of this document is to show a method for the introduction of node to node requirements in the dimensional synthesis of mechanisms when using the minimum distance approach. The presented method allows one to introduce a requirement on the distance of two points belonging to different elements of the linkage, which has multiple practical applications such as the design of grippers. In order to do so, the minimum distance into both nodes is included in the error function for each synthesis point including these kind of requirements. The minimization of the minimum distance function is performed with a sequential quadratic programming algorithm. The required derivatives are obtained analytically to reduce computational cost and improve on convergence. The minimization of the synthesis problem is also solved here using an SQP method, but the error function has been developed taking into account the possibility of applying other methods such as genetic algorithms. The requirement is introduced as an additional term in the minimum distance function, thus allowing one to combine it to other types of requirements, such as node to point or node to line. Although in an initial stage of development, the method shows a good behavior in terms of convergence and computational efficiency. In order to demonstrate this, several examples of both the minimum distance problem resolution and also the synthesis resolution are presented.

Design optimization methods for rotor blades of an axial compressor have been developed by using the computational fluid dynamics (CFD). In order to improve the aerodynamic performances, such as pressure ratio and adiabatic efficiency, three-dimensional Reynolds averaged Navier-Stokes analysis was used for the single stage axial compressor. The optimum design process considering the aerodynamic characteristics consists of designing the shape using the Non-Uniform Rational B-Spline (NURBS) function and performing the optimum design with Gradient-Based Optimization Method (GBOM). For the proceeding of automated optimization, the commercial code ANSYS CFX ver. 16.1 and Design Exploration were applied. Results show that the newly designed model demonstrated better performance than the reference model. In particular, the pressure ratio was found to be higher than that of the reference model.

Aerodynamic shape optimization has established itself as a valuable tool in the engineering design process to achieve highly efficient results. A central aspect for such approaches is the mapping from the design parameters which encode the geometry of the shape to be improved to the quality criteria which describe its performance. The choices to be made in the setup of the optimization process strongly influence this mapping and thus are expected to have a profound influence on the achievable result. In this work we explore the influence of such choices on the effects on the shape optimization of a turbofan rotor blade as it can be realized within an aircraft engine design process. The blade quality is assessed by realistic three dimensional computational fluid dynamics (CFD) simulations. We investigate the outcomes of several optimization runs which differ in various configuration options. We compare the results from the covariance matrix adaptation evolutionary strategy (CMA-ES) with the outcome of a particle swarm optimization (PSO). We also investigate the changes induced by a different initialization of the CMA-ES and by a variation of its population size. A particular focus is put on the variation of the results if we use different number of degrees of freedom for parametrization of the rotor blade geometry. For all such variations, we generally find that the achievable improvement of the blade quality is comparable for most settings and thus rather insensitive to the details of the setup. On the other hand, even supposedly minor changes in the settings, such as using a different random seed for the initialization of the optimizer algorithm, lead to very different shapes. Optimized shapes which show comparable performance usually differ quite strongly in their geometries over the complete blade. Our analyses indicate that the fitness landscape for such a realistic turbofan rotor blade optimization is highly multi-modal with many local optima, where very different shapes show similar performance.

This article addresses the compliance problem along with the piezoelectric actuator design for active vibration control. The structure layout is obtained by solving a compliance minimization problem while the actuators topology is found by the maximization of a controllability index written in terms of the controllability Gramian, which is a measure that describes the ability of the actuators input to move the system state from an initial condition to a desired final state, at rest for instance, in a finite time interval. Also, the polarization direction of each actuator is defined according to the distribution of an additional design variable. Therefore, it is possible to produce both tensile and compressive fields in different points of the structure using the same applied control voltage. In order to achieve this goal, a material interpolation scheme based on the Piezoelectric Material with Penalization and Polarization (PEMAP-P) model is employed and both the optimum structure/actuator layout and polarization profile are obtained simultaneously. The sensitivities with respect to the polarization and design variables are calculated analytically. Numerical examples are presented considering the control of bending vibration modes for a cantilever beam and a simply supported beam in order to show the efficiency of the proposed formulation. The control performance of the designed structures are analyzed by means of a Linear-Quadratic Regulator (LQR) simulation and these results are compared to the ones obtained by a formulation that does not take into account the actuator polarity in the optimization problem, i.e., the polarization profile is stated a priori.

In this research, optimal fin distributions are presented for latent heat storages charged by a constant input power water flow. The limited input power results in non-uniform melting of the Phase Change Material (PCM). Therefore, new designs with non-uniform fin distributions provide the opportunity to outperform the ones with uniform distributions. In this paper, we show that different optimal fin distributions are found depending on the fin width, amount of fins and the input power. The gain in charging performance is discussed by comparing with latent heat storages without heat transfer enhancement.

The multi-criteria and multidisciplinary optimization of the rotor and stator blades of the experimental compressor stage NASA Rotor 37 is carried out. For the optimization the unified parameterized multidisciplinary 3D model was specially made and used. This model includes the air-gas channel of the compressor stage and the finite element model of the rotor blade. This approach allows the consideration of different requirements for the aerodynamic, strength, and mass characteristics within one coupled optimization problem in unified computational space. The goal of this work is the analysis of different blade parameterization schemes and determination of optimum number of variable parameters for compressor stage aerodynamic characteristics improvement with respect to rotor blade static and dynamic strength. As an optimization criterions compressor stage efficiency and the blade mass minimization were used. Aerodynamic limits are: flow rate and pressure ratio values should not exceed base values more than ±0,5%. A static and dynamic strength limits are: maximum stress level should not exceed base level (the original design stress level) and the relative distance between the four natural frequencies and the nearest harmonics should not be less than 20%. As a result of optimization the NASA Rotor 37 version was found, which provide the efficiency increasing by approximately 2% and the blade centrifugal load decreasing by approximately 9%, while all aerodynamic and strength requirements are satisfied. It was also found, that increasing of the blade profile number of variables more than 7 is not rational.

For many applications numerical simulations are available in varying degrees of fidelity and computational expense. On the one hand, in an accurate and time-consuming high-fidelity (HiFi) version, that is used as a reference for design and optimization, and on the other hand, in a low-fidelity (LoFi) version that is faster but less accurate. In computational fluid dynamics (CFD) of turbine airfoils an accurate 3D solver accounts for the HiFi model and a faster 2D solver for the LoFi model. Assuming the LoFi model captures the fundamental physics reasonably well, many inexpensive LoFi computations may be coupled with a few expensive HiFi computations to enhance the accuracy of a surrogate model based solely on the HiFi data. In this way multi-fidelity surrogate models can be used to speed up the optimization.Conventional multi-fidelity surrogate models approximate the HiFi objective function by aligning the low-fidelity predictions to the high-fidelity results. This is commonly achieved by a kriging based interpolation of the error between both fidelity models. The design space variables are the independent variables of this interpolation.In this paper a different approach is presented. Instead of using the design space variables directly, LoFi coordinates in a low-dimensional subspace are used for the interpolation of the HiFi objective values. The low-dimensional subspace representation is obtained by Proper Orthogonal Decomposition (POD) of the LoFi computational domain to identify the most important modes of variation.Based on these surrogates multi-fidelity optimizations of a gas turbine second stage vane are carried out utilizing a kriging based Expected Improvement strategy. Especially the POD based method shows a fast convergence and outperforms a single-fidelity optimization.

As one of the main renewable energy sources, wind energy has gained an important role in the generation of sustainable energy. For this reason, the aim to achieve a high degree of utilization as well as the aim to enhance the durability of wind turbines became vital research topics. Therefore, the ability to identify structural damage and consequently prevent component failure is a significant tool of interest in relation to flexible service intervals and condition-based maintenance of wind turbines. As rotor blades are related to about twenty percent of the overall costs of a wind turbine, monitoring their condition is of high interest for the reduction of operation and maintenance costs. Thus, the focus of this contribution is on the detection, localization and quantification of structural damage of wind turbine blades.There has been a development of many different non-destructive damage localization techniques over the past decades, whereby vibration-based damage localization techniques have successfully been used to monitor wind turbine blades. Vibration-based methods assume that damage-induced variations in the structural properties, namely mass, stiffness and damping, cause detectable changes in the structural behavior. In the presented work, a reference model of a parameterized offshore rotor blade is created. In order to predict damages in this reference model, the stiffness of certain cross sections is reduced. The simulation of the corresponding structural behavior creates a data set, representing the measured response of a damaged state. To detect, locate and quantify these changes, the structural properties of the reference model are adapted to the ‘measured’ response by comparing modal parameters. Then, cross sections with varied properties indicate the area where damage has occurred. To analyze the considered model updating procedure, different sets of sensor positions and different numbers of design variables are compared utilizing various optimization algorithms.

In the field of two-dimensional image and video processing, convolutional neural networks have been successfully applied to generate novel images by composing content and style of two different sources, a process called artistic or neural style transfer. However a usage of these methods for three-dimensional objects is not straightforward due to the unstructured mesh representations of typical shape data. Hence efficient geometry representations are required to use neural network based style transfer concepts for three-dimensional shapes and to enable the fast creation of style options for instance in a product ideation process. In this paper an overview of current state-of-the-art shape representations is presented with respect to their applicability of neural style transfer on three-dimensional shape data. Combinations of three-dimensional geometric representations with deep neural network architectures are evaluated towards their capability to store and reproduce content and style information based on previously proposed reconstruction tests.

A procedure to optimize the distribution of infill material in the core of a three-dimensional object manufactured through additive manufacturing is presented. The goal is to maximize stiffness and it requires a prior subdivision of the infill region into non-overlapping subdomains, separated by thin walls. Each subdomain shares the same lattice cell size and topology while the spatial distribution of lattice material within each subdomain is determined through two optimization problems. The first problem makes use of effective properties of the cell to estimate the optimum relative density. The second problem optimizes the area of the lattice bars individually. The methodology is illustrated through the use of a simple example.

The grillage optimization problem can be originally posed as a pair of infinite dimensional variational problems: equilibrium and kinematic, where a function of bending moment and out-of-plane displacement is sought respectively. In the current paper those problems are analytically reduced to a pair of discrete forms being counterparts of the two forms that are well-established in optimization of trusses. In the process the two new mutually dual norms defined on a plane emerge.

In the simulation of crashworthiness structures different kinds of nonlinearities like large deformations, contact and material nonlinearities appear. In order to optimize the topology of crashworthiness profile structures, the Graph and Heuristic Based Topology Optimization (GHT) was introduced. This method uses heuristics that are derived from expert knowledge to change the topology through adding or removing walls from the profile.The original approach of the GHT is limited to extrusion profile structures, as the graph syntax, which describes the profile cross-section and the heuristics, were developed for the extrusion manufacturing process. Due to the demand for using different manufacturing processes, the approach is extended by sheet metal compound structures. In the extension, which is described in this contribution, the structure is split into multiple less complex parts, which can be separately manufactured and later bonded to form the complete structure.In general all kind of joining processes can be used to connect the parts. In this first extension, the approach is limited to adhesive bonding. Depending on the joining process, it is necessary to create flange connections to increase the joining surface area and thus the strength of the bonding.Because the joining technique has a huge influence on the mechanical behavior, the flanges and the bonding have to be included in the analysis model of the optimization. Therefore, new modeling schemes are used to define the number and positions of the flanges.In this paper the necessary adoptions to optimize bonded profiles are described. In addition, optimizations of an aluminum profile that is impacted in a drop tower are carried out to illustrate the procedure. The optimization results for the manufacturing processes extrusion and the sheet metal compound are compared.

Nowadays, the environmental consequences of global warming are of major concern. Worldwide, efforts are being made to reduce the CO2 (carbon dioxide) emissions in order to mitigate the greenhouse effect. Construction sustainability is gaining increasing relevance in the last years to cope with the large environmental impact of construction industry. Considering this the structural design should aim to obtain economical, structurally efficient and “environmentally-friendly” solutions. In this work a numerical model for the sustainable optimum design of reinforced concrete (RC) frames was developed. The structural analysis includes all the actions and relevant effects, namely, dead and live loads, the time-dependent effects and the geometrical nonlinearities. The analytical discrete direct method is used for sensitivity analysis. The sustainable design of RC frames is formulated as a multi-objective optimization problem with objectives of minimum construction cost, minimum CO2 emissions, minimum deflections and stresses and a Pareto solution is sought. The minimax solution is found by the minimization of a convex scalar function obtained through an entropy-based approach. The displacements and stresses design goals are established according to the Eurocode 2 recommendations for the design of framed structures. The design variables considered are the beams and columns cross-sectional dimensions and the steel reinforcement area. The features and applicability of the developed numerical model are demonstrated by a numerical example concerning the optimization of a real sized RC frame.

The Limited Resistance Rigid Perfectly Plastic Hinge (LRPH) are special steel connections mainly usable to join beam elements of plane or spatial steel frames. The fundamental characteristics of these devices are the mutual independence of their own resistance and stiffness features as well as the respect of assigned constraints related to the elastic and limit behaviour of the joined elements. Within the frame structural scheme, the device plays the role of a rigid perfectly plastic hinge, constituted by a suitably sized sandwich section. The efficient use of the LRPH in the relevant frame depends on the appropriate design of the device geometry. In the present paper, a new approach devoted to the optimal flexural design of the LRPH is presented, according with the imposed mechanical constraints as well as with further suggested technological ones. The optimization procedure is based on a genetic algorithm approach and different applications are reported confirming the good applicability of the computational method as well as the reliability of the relevant device.

The stability analysis and optimization of elastic web travelling between two rollers with a constant velocity are presented. The mathematical model for a layered travelling web (continuous isotropic composite plate) is developed restricting the consideration to one open draw. The layered plate with various mechanical properties of layers is considered and analytical expressions for the effective characteristics are derived. As a result the composed structure can be considered as an isotropic homogeneous plate and the obtained formulas for computation of critical velocity can be applied. Then the isoperimetric optimization problem is formulated and studied. The total mass of the layered plate is considered as an isoperimetric condition. The critical divergence velocity is taken as an optimized quality criterion. To this end consisted in maximization of the web stability and for maximization of the divergence velocity with respect to material distribution, the evolutionary optimization method (genetic algorithm) is applied. The number of materials is supposed to be given. Applying the genetic algorithm these materials are distributed on the plate thickness (provide the optimal plate consisted of some layers of different thickness) and the critical velocity is maximized under the constraint on the total mass of the structure. Numerical results are presented for different sets of problem parameters.

As recent power modules with utmost performance are required in small and light-weight structures, the design and the manufacturing of heat sinks that offer maximal cooling performance have become a challenge due to continuously evolving miniaturization and smaller cooling areas. In line of the above, and for practical realizations, heat sinks having the ability to operate in limited volume, as well as having simple geometry are highly desirable due to existing limitations in cooling area and machining processes. We propose a methodology to design heat sinks that aim at minimizing temperature and pressure loss based on fin considerations, as well as design of experiments and response surfaces of objective functions about design variables. Numerical simulations and sensitivity analyses based on Finite Element Method have shown the improved cooling performance in terms of thermal resistance when compared to the conventional heat sinks with straight fins. The unique point of our proposed approach lies not only in the simplicity of the geometry of the heat sink, but also in the higher cooling performance, which is key to realize enhanced cooling mechanisms by existing metal-based 3D printers. We believe our approach offers the building blocks to enable the design and realization of heat sinks with manufacturable geometry and utmost cooling performance. As such, our proposed approach has the potential to capitalize on the benefits of metal-based 3D printing technologies, enabling the possibility to realize challenging geometries.

The identification or optimization of structures in macro scale is widely used nowadays. The goal of the paper is to apply identification techniques to obtain information about material distribution on the micro level. The presented methods open new possibilities. The structures build with the use of materials with optimal microstructure can obtain the best performance. The parameters of microstructure can be identified taking into account loads of the macro structure. The identification of microstructure parameters is not easy currently, but in future, in applications where performance of the structure is very important, the presented approach may be used with success. A bio-inspired method based on the artificial immune system (AIS) is used to solve the identification problem. Immune computing provides a great probability of finding the accurate solution. It is developed on the basis of a mechanism discovered in biological immune systems.

Various forms of geometric imperfections of compressed rods, which are subjected to loss of stability, have been considered in the normative literature. Similar investigations are also intensively carried out for the shells with imperfections, taking into account sub- and post-critical behavior of the systems, as well as an inelastic deformation of materials. However, there are no recommendations regarding the choice of the worst forms of such imperfections. The existing intuitive hypothesis regarding the choice of a shape with which the rods lose their stability can only be considered as a prerequisite for a proof or disproof of this hypothesis. Besides, references to extensive engineering experience are not sufficient; furthermore, already known various counterexamples contradict this hypothesis. In this paper, we formulate a general optimization mathematical model for selecting worst geometric imperfections for rods and shells as a problem of optimum control. Analytical solution of this model has been obtained for subcritical behavior of elastic rods with different boundary conditions. It is shown that the optimal solutions for worst geometric imperfections do not belong to smooth functions, but to the discontinuous piecewise smooth functions. Differences between behavior of simply supported rods with the obtained anti-optimal (worst) geometric outlines and that of those with the currently used imperfect functions reach 19.2%. The presented formulation makes it also possible to evaluate and obtain more precise recommendations for the shells with imperfections.

The aim of this paper is to provide a reliable simplified numerical model for analysis of dynamic behaviour of railway tracks. This is motivated by the demands of railway companies that prefer simplified models to complex numerical models, thanks to their numerous advantages. As an adequate simplified model, the discrete support model (DSM) is selected. The range of its applicability and general formulas for identifying properties of its constituents are proposed and validated. To calibrate the DSM, a 3D detailed finite element (FE) model is created and validated by comparison with published experimental measurements. The objective function in optimization runs is formed by the L2-norm of the difference between the vertical displacement of the rail obtained by the DSM and 3D FE model. Based on conclusions from optimization runs, formulas for identifying the properties of the DSM are proposed and validated. These formulas cover a large range of typical properties of the railway tracks, which in turn identifies the range of applicability of the DSM.

Earthquakes are sources of seismic loading on structures with probabilistic nature of their records. As a result, no single earthquake is reliable for decision making in consequent structural design. A common solution to reduce such probabilistic effects is spectral matching of earthquake records with a target design spectrum within a prescribed period interval. Selection of the corresponding scale factors in such a process is formulated here as an optimization problem while the objective is to achieve the best compatibility of the mean spectrum with the target. A new algorithm is proposed to solve this problem. It utilizes switching between movement of a candidate design vector and its bound-based opposite toward the current best solution. The algorithm is designed for simplicity and efficiency for such a practical engineering task. Numerical tests exhibit considerable reduction of spectral matching error with respect to common practice in application of uniform scale factors.

In this work, a novel topology optimization problem formulation is proposed.The case of the concurrent topological optimization of two different components sharing a portion of the design space is considered. The design problem represents the relevant design situation in which more than one component has to be fitted in an enclosed space and each component has its own load carrying function.The proposed algorithm assigns the shared space to one or the other body depending on the relative sensitivity of each element to the total compliance of the system. After each element has been assigned to one of the two design domains, the connectedness of the two domains is enforced. The volume fraction is enforced at system level, i.e. the volume fractions of the two domains can be different, but the total volume fraction complies with the set value. In this way, the available mass can be allocated in the most convenient way among the two bodies.Some examples are presented to show the performance of the proposed algorithm. In one example, two structures for which the optimal solutions are known from the literature have been considered. The two design domains are overlapped as to allow the two optimal solutions to be found. The two optimal solutions are obtained by the concurrent topological optimization algorithm. Moreover, by imposing some reasonable, but not optimal, divisions of the design domain, structures with higher compliances have been obtained.In the last part of the paper, the proposed algorithm is used to optimize a tool support arm made by two components and modeled by a non linear finite element model.

The present study applies the method of combination of elementary mechanisms to the evaluation of collapse conditions of planar frames. The collapse load is evaluated seeking the absolute lowest value among all the mechanisms that can be obtained combining the elementary ones. The optimization procedure is developed through three different bio-inspired optimization algorithms. In particular, genetic, immune and ant colony algorithms are considered. Original codes developed in the agent-based programming language NetLogo allow building into a virtual metrical space and visualizing in the user interface every single mechanism and the correspondent collapse load. The elementary mechanisms are then combined and the minimum collapse load, together with the corresponding collapse mechanism, is obtained. Several applications have been performed with reference to frames of different size subjected to a seismic load scenario consisting of horizontal forces with increasing magnitude acting on each floor, and permanent vertical loads applied to the beams. The collapse loads and related mechanisms, obtained by means of the proposed optimization procedures, have been compared to the correspondent ones provided by nonlinear push over analysis, showing a very good correspondence.

Computer simulations are widely used in science and engineering to evaluate models, as a partial substitute to real-world experiments. Especially, it is required to complex technological process modelling. However, in this case, the application of computer simulation technologies often leads to high computational cost. The solution of this problem is using high-performance computing equipment and effective parallel algorithms. In this paper, the problem of computer modeling of water purification processes by electromagnetic method with the help of modern high-performance computing systems with modern architecture is considered. The challenge is discussed for three-dimensional calculation of the flow of an aqueous medium in a cleaning tank of real geometry. As a mathematical model of the motion of the aquatic environment, we chose the Navier-Stokes equations under conditions of fluid incompressibility and isothermal purification process. For this model, we proposed an original numerical algorithm using prismatic grids. This algorithm was parallelized and implemented as a code for supercomputers with central and vector processors. Within the testing of the developed software, the convergence of the numerical algorithm and optimization of the parallelization scheme were carried out. To increase the efficiency of parallelization, dynamic load balancing of the calculators was implemented. The analysis of the obtained results confirmed the advantages of the proposed approach to solving the problems of modeling the processes of water purification by an electromagnetic method.

Sparse models are preferred in machine learning problems because of their computational interpretability and it is seen in many applications such as in Google Page Rank, classification and regression problems, in the method of Principal Component Analysis (PCA) that finds the most important features and further applications in graphical models. In this study, we derive optimality conditions for the quadratic problem which has cardinality constraint imposing sparse solution. Our Quadratic model is a special application of ensemble pruning model in ensemble learning algorithms. Here, we refer to our previous study on this application in ensemble selection for clustering problems. The quadratic model proposed in this study optimizes trade-off between accuracy and diversity of discriminant functions (classifiers) simultaneously so that the best candidates of ensemble are selected for prediction step. The selection of the best classifiers in the ensemble is crucial for the overall performance of ensemble learning algorithms since redundant/outlier solutions in the ensemble library will decrease the overall prediction accuracy. In order to eliminate such candidates, both accuracy and diversity are taken into account when selecting the best subset of the ensemble. The cardinality constraint is further relaxed by considering various approximations such as $$l_1$$ l 1 -norm regularization and student t-log likelihood approximations. Under these considerations and approximations, we build optimality criteria for our quadratic optimization problem with a cardinality constraint.

Regarding shape optimization of structural components, topology optimization has become one of the most popular methods to achieve significant reductions in mass and volume, while maintaining stiffness. The basic topology optimization algorithm considered in this paper and a heuristic updating scheme are described in Bendsøe [1].Topology optimization is an iterative process, which requires the Finite Element Method (FEM) to obtain the objective function and constraints. In case of structural optimization, the objective function is the compliance, also known as strain energy, and has to be minimized. A better performance of FEM can be achieved if all the elements have a uniform shape. This is usually impossible for any design space with the standard boundary-conforming FEM. In this work, we propose the use of the cartesian grid Finite Element Method (cgFEM) [2] instead of the standard FEM. The main feature of this method is the use of Cartesian FE grids independent of the geometry that allow for the use of an efficient hierarchical data structure that reduces the amount of calculations, this resulting in a high performance FE analysis.In addition, we propose the implementation of h-adaptative mesh refinement [3] to the main loop of optimization. To improve the accuracy in the definition of the boundary.In this work the SIMP method (Solid Isotropic Material with Penalization) [4] is used to avoid the ill-posedness of the problem [5]. The iterative process begins with a uniform distribution of densities. In each step, a gradient-based algorithm has been used to obtain the new material layout. Before fully reaching the convergence, the elements with intermediate density values are refined. Then the process continues until full convergence or until another mesh refinement is needed.As a result of adding h-adaptative refinement to the optimization algorithm, we are able to reproduce an optimized geometry similar to the one obtained with a coarse mesh, but with greater definition of its boundary.

Graph-based design languages aim at the holistic digital design and representation of complex industrial products. These languages are based on the structure of natural languages, in which the vocabulary and the rules define a language grammar. The translation of the abstract graph-based design language into multiple specific domain-dependent engineering models occurs in a framework of a design compiler. This design compiler compiles the design rules automatically and allows the easy reuse of design and production knowledge. The product development engineers are thus relieved of most of their routine work by generative means: all the domain-specific models such as finite element models (FEM) and geometry models (CAD) are generated automatically based on the translation of the design language into a central data model. The central data model is the core result of the compilation of a design language and holds all the information necessary to generate all domain-specific models, which are specified in the design language. This paper describes the use of this kind of design language for the embodiment and dimensioning of a frame for a Formula Student racing car. This implementation extends the optimization design language, which has already been presented in part at the WCSMO12 [1].

This paper is concerned with the analysis of framed structures with inextensible and rigid members, i.e., members without axial and bending strains. Rigid and inextensible members may be useful in educational software because they capture the essence of the structural behavior with a reduced the number of variables. In addition, they allow a comparison with results obtained by hand calculation using classical structural analysis methods. There are three main approaches to constraint handling: transformation, penalty function and Lagrange multiplier methods. The transformation methods do not allow the determination of axial forces in inextensible member or bending moments in rigid members. On the other hand, the penalty function method allows the computation of internal forces in constrained members and its implementation is simple. However, as the penalty factor increases, the stiffness matrix becomes increasingly ill-conditioned, which may lead to large solution errors. This paper presents a methodology for considering structural member deformation constraints using Lagrange multipliers. It consists of adding strain constraints into the total potential energy minimization, leading to a quadratic programming problem. In addition, this approach is very suitable for computational implementation because it does not affect the generic characteristic of a matrix structural analysis. The solution gives rise to one Lagrange multiplier per constraint, which is essential for computing member internal forces. However, there are situations in which inextensible and rigid member constrains may be redundant, which prevents the determination of dependent Lagrange multipliers. Although not implemented, a special treatment is indicated for the solution of this problem.

Stability analyses of slopes have been a challenge for engineers, requiring development of complex numerical models to assess the risk levels and potential hazards. The numerical models involve combination of analysis methods and integrated optimization approaches, which generally induce intense engineering calculations with upscale time complexity. To obtain good and quick solutions, a robust optimization algorithm is necessary, leading to an efficient and reliable stability analysis framework. Within this context, various optimization techniques involving deterministic and metaheuristic approaches were proposed in the past decades. The proposed methods often suffer from convergence issues have time deficiencies, which highlights a necessity of development of an effective optimization algorithm. In this study, a modified version of Differential Evolution (DE) algorithm named Elitist Differential Evolution (EDE) is proposed to solve slope stability analysis problems. To develop a complete analysis framework, EDE is integrated with a non-circular failure surface generation method and limit equilibrium based stability analysis techniques. Its performance is compared with other optimization algorithms such as conventional DE, Particle Swarm Optimization and Grey Wolf Optimizer using benchmark problems reported in the literature. The experiments demonstrate that EDE greatly improves the results of other alternatives, validating the applicability of the algorithm to slope stability analysis. Furthermore, statistical performance of EDE has become prominent in the experiments, which further emphasizes its robustness.

Authors present a topology optimization procedure for steady-state forced vibration problems where a weighted sum between active input power and static compliance is used to obtain anti-resonances of displacement at load points in vibrating structures, at frequencies close to those of interest. The reactive input power, converted to a relation between kinetic energy and potential energy, helps to improve the procedure. Several examples are presented to illustrate the potential of the proposed method.

We develop a topology optimization method including high-cycle fatigue as a constraint. The fatigue model is based on a continuous-time approach, which uses the concept of a moving endurance surface as a function of the stress history and back stress evolution. The development of damage only occurs when the stress state lies outside the endurance surface. Furthermore, an aggregation function, which approximates the maximum fatigue damage, is implemented. As the optimization workflow is sensitivity-based, the fatigue sensitivities are determined using an adjoint sensitivity analysis. The capabilities of the presented approach are tested on numerical models where the problem is to maximize the stiffness subject to high-cycle fatigue constraints.

Reliability-based methods have been established to take into account, in a rigorous manner, the uncertainties involved in analysis of engineering systems. The failure probability and reliability index are used to quantify risks and therefore evaluate the consequences of failure. First/second-order reliability method (FORM/SORM) is considered to be one of the most reliable computational methods to deal with reliability in engineering systems. Basically, the idea is to overcome the computational difficulties in determination of the reliability index and approximating the constraints. In this contribution, a new methodology to deal with uncertainties in engineering systems is proposed. This approach, called Second-Order Inverse Reliability Analysis (SOIRA), consists in the use of first and second order derivatives to find the solution associated with the highest probability value (inverse reliability analysis). In order to evaluate the proposed methodology, three reliability approaches (FORM, SORM and IRA - Inverse Reliability Analysis) are applied in two test cases: (i) W16X31 steel beam problem and (ii) beam problem. The obtained results demonstrated that the proposed strategy represents an interesting alternative to reliability design of engineering systems.

In this paper the Topological Derivative for crash loaded structures is derived with the adjoint sensitivity analysis. The main idea of the adjoint sensitivity analysis is to circumvent the direct calculation of the sensitivity of the displacement field. Instead, the adjoint equilibrium equation has to be solved. In this approach, material derivation and partial integration in the time domain are applied to the Topological Derivative. This ensures, that the inertial effects are kept, as they are important for a reliable crash simulation. The result is a backward integration scheme for the adjoint state.With implicit time integration, a numerical scheme to solve the primal and the adjoint problem is demonstrated. The specific adjoint equation as well as the Topological Derivative for a displacement functional and the internal energy are presented.

The currently available computational power and improvements of high-fidelity numerical simulations have lead to an increased use of computational fluid dynamics (CFD) in the analysis of turbomachinery flows, particularly in design environments. The optimization cases often contain up to thousands of design variables and gradient-based (GB) optimization algorithms are typically selected due to their efficiency. The adjoint method is key to efficiently compute the derivatives required by the GB algorithms, with a computational cost nearly independent of the number of design variables. In this paper we present the details of the development of an adjoint multirow interface based on the mixing-plane treatment to extend an already existing adjoint solver using the ADjoint approach. The mixing-plane treatment allows the steady simulation of multiple rows, taking their interaction between one another into account and thus providing more realistic results. A stator/rotor turbine stage of a commercial jet engine is analyzed and some representative sensitivity results are presented and discussed.

We propose in this paper to deal with squeal noise reduction of brake systems through their shape optimization during the design step. We first expose the FEM model used to generate the stability diagram representing the squeal noise behavior of a given brake system shape. We then propose an objective function able to be included in a minimization problem and based on the stability diagram. We use then a parallel code to browse the objective function response surface through a Latin Hypercube Sampling design of experiment. A Self Organizing Map is then generated to expose the sensibility of our objective function to seven shape parameters of the FEM brake system. We present and analysis the SOM results for further optimization steps.

Multidisciplinary design and optimization is a promising methodology for the efficient design of complex systems, in particular when it combines coupled analyses with gradient-based optimization techniques. In this case, it requires the derivatives evaluation of the functions of interest with respect to the design variables, which is the most demanding computational task in the process, so the goal of this work is to develop an efficient optimization framework to solve aerodynamic design problems using exact gradient information. To this end, the aerodynamic model based on the panel method is reformulated into five smaller modules, in which the respective sensitivity analysis blocks are constructed using exact gradient estimation methods: automatic differentiation, symbolic differentiation and the adjoint method. After the aerodynamic and corresponding sensitivity analysis tools are verified numerically, aerodynamic optimization problems are solved using the new tool with remarkable success since, when compared to the finite-differences method, the optimization time can be reduced by 90%.

Equivalent electric circuits are widely used in electrochemical impedance spectroscopy (EIS) data modeling. EIS modeling involves the identification of an electrical circuit physically equivalent to the system under analysis. This equivalence is based on the assumption that each phenomenon of the electrode interface and the electrolyte is represented by electrical components such as resistors, capacitors and inductors. This analogy allows impedance data to be used in simulations and predictions related to corrosion and electrochemistry. However, when no prior knowledge of the inner workings of the process under analysis is available, the identification of the circuit model is not a trivial task. The main objective of this work is to improve both the equivalent circuit topology identification and the parameter estimation by using a different approach than the usual Genetic Programming. In order to accomplish this goal, a methodology was developed to unify the application of Cartesian Genetic Programming to tackle system topology identification and Differential Evolution for optimization of the circuit parameters. The performance and effectiveness of this methodology were tested by performing the circuit identification on four different known systems, using numerically simulated impedance data. Results showed that the applied methodology was able to identify with satisfactory precision both the circuits and the values of the components. Results also indicated the necessity of using a stochastic method in the optimization process, since more than one electric circuit can fit the same dataset. The use of evolutionary adaptive metaheuristics such as the Cartesian Genetic Programming allows not only the estimation of the model parameters, but also the identification of its optimal topology. However, due to the possibility of multiple solutions, its application must be done with caution. Whenever possible, restrictions on the search space should be added, bearing in mind the correspondence of the model to the studied physical phenomena.

In this paper, we investigate the ability of genetic representation methods to describe two-dimensional outline shapes, in order to use them in a generative design system. A specific area of mechanical design focuses on planar mechanisms. These are assembled of mechanical components, e.g. multiple levers, which transmit forces and torques over their contour. The shape of the contour influences the performance of the overall system. The genetic representations are based on floating-point chromosomes, where each value maps to a specific parameter of a resulting shape. In order to evaluate the performance of each representation method, a set of target shapes was defined. These consist of simple symmetric and asymmetric shapes with edges and curves, and also of more complex mechanical lever shapes, extracted from an automotive device. An evolutionary algorithm with crossover and mutation operators is used to search for the best approximation of these target shapes. The fitness function is based on two penalty values: first, calculated by comparing the area of a candidate solution with the area of a target shape; and second, based on the intersection area between a candidate solution and a target shape compared to the entire area of the target. Experiments were undertaken to investigate the capabilities of the representations in terms of search space coverage; compatibility with evolutionary operators; and the ability to produce shapes with mechanical characteristics. The results show the benefits and drawbacks of using each of selected methods of representation, and their suitability of reassembling different outline shapes.

Forest fires cause many changes in environment and in climate, becoming a huge concern related with environment, as your prevention and control. The fire risk calculation supports the planning of activities to prevent forest fire, as it determines the probability of fire occurrence in certain place. This article has the aim of mapping fire risk areas of Belo Horizonte, one of the most populous cities from Brazil, located in the Minas Gerais State, in the Southeast Region of Brazil. The proposed modeling is to create an artificial neural network with supervised training. A neural network to do the prediction of most propitious fire areas is expected, where it can be introduced the input variables at any period that desire to be determined. This estimate will provide the outline of priority areas for prevention activities and allocation of brigade teams, seeking to minimize possible damages caused by fires.

Topology optimization is one of the most intensively developed and frequently implemented in practice engineering design tools. The idea is to find within a defined design domain the distribution of the material that is optimal according to the assumed criteria. The typical solutions regard structures made of one material, but allowing for implementation of multi-material structures may open new possibilities for improving existing solutions. The conventional approach is a redistribution of material or materials within a whole design domain. This concept is extended in the present paper by introducing the idea of ‘subdomain oriented multi-material topology optimization’. The design domain is divided into regions for which different types of material are defined and through the optimization procedure the multi-material structure is created. The aim of the present research is therefore to find optimal topologies, under restriction that redistribution of material can be performed only within subdomains selected for employed materials. What is important, in terms of practical applications, it is possible to impose constraints on volume fraction of each defined material. Obtained results of preliminary numerical studies show, that this approach produces different results as compared with classical single-material problems. In addition, included self-weight loading makes considered design problems more practical and realistic. As the optimization tool efficient and versatile heuristic method based on Cellular Automata (CA) concept is utilized. The main advantage of the CA algorithm is that, it is an easy to implement, fast convergent technique and usually requires less iterations, as compared to other approaches, to achieve the optimal solution.

This paper presents a preliminary metaheuristic approach for underwater swarm robotic parameter configuration applied to optimal plume detection under time-variant scenarios. In this work the plume scanning has followed a collaborative approach that has been modelled following a real-based scenario obtained within the European project SWARMs (“Smart and Networking Underwater Robots in Cooperation Meshes”) [1]. The proposed optimization algorithm is designed aiming at minimising the overall time of the mission while assuring an optimal plume detection. Preliminary results show that this proposed approach can assist the operator when designing the mission and configuring the optimal swarm robotic parameters.

Predicting bankruptcy is one of the most challenging subjects and research topics in economic and financial areas especially in these last decades. Making a financial early warning system to evaluate firm’s failure risk depending on their financial behavior can be a crucial key indicator for making decision. One of the most popular and performer tools to predict financial distress is Artificial Neural Network (ANN). In this paper, a financial warning system is proposed based on a hybrid ANN model to predict bankruptcy and risk scoring, this hybrid model considers the firms’ behavior for three years to predict risk failure. Taking into consideration in one hand that ANN is a powerful tool to approximate nonlinear function if it is designed with appropriate parameters, and in the second hand, the problem of local minima, we propose a topology design algorithm based on an improved Particle swarm optimization and simulated annealing to define an optimized ANN architecture. Taking in consideration feature selection, a sensitivity analysis in made to catch the relevance of the discriminant variables used in the proposed financial warning system. A comparative performance study is reported. The results showed that the proposed model represents a valid alternative to give an early risk failure warning.

This paper discusses the influence of different Radial Basis Function (RBF) in metamodel construction to be applied to 3D aerodynamic wing optimization problems using the Constrained Optimization with Response Surface (CORS) methodology in conjunction with a stochastic Controlled Random Search Algorithm (CRSA). CORS methodology is based on the iterative construction and optimization of response surfaces with a robust search pattern application. In the CORS methodology the response surface may be generate by at least three types of methods: (i) Classical (polynomials and parametric surfaces), (ii) Statistical (K-Nearest, Kriging and Gaussian Processes) and (iii) Advanced (RBF and Neural Network). The response surfaces used in this paper are constructed using six different types of RBF: Gaussian, Hardy’s Multiquadric and Inverse Multiquadric, which depend on a shape parameter, and linear, cubic and thin plate spline, which do not depend on a shape parameter. The RBF’s are directly applied on the response surface construction inside the CORS structure. Thus the choice of a RBF and of a shape parameter (being the case) may influence significantly the methodology efficiency. The CORS methodology is here applied for accelerating the optimization process of wing aerodynamic designs with a solver based on a first order 3D panel method and a 2D boundary layer model. Since the main objective of this paper is of prospective nature, the choice of a relatively low-fidelity flow computation solver is justified. One considers problems of minimizing the aerodynamic coefficient relation (CD/CL) and the inverse of lift coefficient (1/CL). Comparative influence of the RBF choice on the acceleration induced by CORS methodology is investigated taking into account the number of expensive objective function evaluations necessary to find the minimum value in each problem.

The simulation of the pantograph-catenary dynamic interaction has become an essential tool for the design of the overhead contact line. With the help of an efficient simulation strategy, the geometry of the catenary can be optimised in terms of the current collection quality. This work is a first attempt to obtain robust optimised catenaries in which the uncertainty caused by the installation errors is taken into account in the simulations. The optimisation problem is solved by means of a Bayesian Optimisation algorithm and the stochastic objective function is evaluated via Monte Carlo simulations. The results show, on the one hand, the good performance of the Bayesian Optimisation technique when compared with a Genetic Algorithm, and on the other hand, the coincidence between the deterministic and the robust optimal catenaries.

This paper presents an application of multiobjective optimization for the damage quantification in beams. The simulation of damage relies on the finite element analysis of Euler-Bernoulli beams and is carried out by considering a reduction in the Young’s modulus of specific elements. The damage is quantified by minimizing two objective functions. These two functions are based on the difference in the Frobenius norm of matrices containing the modal displacements and the modal rotations of a beam in the undamaged and damaged states. The solution of the optimization problem thus defined is solved by a direct multisearch algorithm, which is an extension of the direct search algorithm to multiobjective optimization. This algorithm does not need any derivatives information about the objective functions. The validity, robustness and efficiency of the present application is tested for different boundary conditions of the damaged beam and high levels of noise in the simulated measured data.

An application of the fast multipole boundary element method (FMBEM) and an artificial immune system (AIS) to the optimization of porous structure effective elastic properties is presented. The FMBEM allows one to model complex geometries with much lower number of degrees of freedom in comparison to the finite element method, that is usually applied in computational homogenization. Representative volume elements (RVEs) are modelled, with displacement boundary conditions corresponding to a given strain state in the macro scale. Effective elastic constants of the material are calculated by using the averaged strains and stresses. Design variables considered in the optimization problem describe the geometry. The minimized objective function involves a metric that allows one to calculate the distance between two elasticity tensors: a current solution and a reference tensor that defines the desired properties. A benchmark problem of porous structure with maximized effective bulk modulus is solved.

Composite materials have gained prominence as an intensively used material in the aerospace and mechanical industry due to their characteristics of stiffness and low weight. The possibility of designing composite specimens with continuous orientation of the fibers at the ply level, following smooth contours, makes this material even more attractive, as it assess, in a more rational way, the whole reserve of fiber stiffness in the directions of main loadings. This work presents a methodology for the optimization of composite materials by the definition of a continuous fiber orientation. Parameterized curves are used to define the continuous orientation of the fibers and the control points are assumed as design parameters during optimization. A Multiobjective Quantum Particle Swarm Optimization (MO-QPSO) algorithm is used as an optimizer due to desirable characteristics of good convergence and lower likelihood of being stuck in local minima. Two examples are presented; both have as one of the objective functions the Practicality Index (a value that defines the ease of execution of the continuous orientations of the fibers). The first one is a dynamic analysis where the previous objective is confronted with the maximization of the first natural frequency. The next example is a trade-off with the Tsai-Wu failure criteria and the Practicality Index. The results are compared with literature solutions (which uses an improved NSGA-II algorithm). In the end, the orientation of the fibers found in the composite material was very similar to those reported in the literature, confirming the validity of the proposed methodology.

This work aims at investigating the applicability of the level-set based thickness optimization method, earlier proposed by the authors, to a realistic structure. The design has to have sufficient stiffness and strength while the structural mass is minimized. The concerned composite structure is subjected to multiple load cases. The proposed method guarantees the fulfillment of the design guidelines, namely symmetry, covering ply, disorientation, percentage rule, balance, and contiguity of the layup. The stiffeners divide a composite structure into several smaller panels. The manufacturability of a resulting design is guaranteed as plies are continuous among adjacent panels (the design is blended). The proposed method is successfully applied to the mass minimization problem of the stiffened top and bottom skin of a wing torsion box. The structure, subject to two load cases, is optimized where local buckling and allowable strain are the constraints of the problem.

This work proposes a methodology to design composite materials, considering two distinct materials phases and one void phase simultaneously, in order to minimize the thermal expansion coefficients. The design of composite material is treated as a topology optimization problem with a multi-material and multi-scale approach. The Bi-directional Evolutionary Structural Optimization method (BESO) is used to solve the optimization problem and the homogenization method is applied to obtain the equivalent properties for the designed material. In order to show the suitability of the implemented methodology, it is presented one example for the minimization of the homogenized thermal expansion coefficients considering a two-dimensional state of stress. A setting using two material phases, and void was performed resulting in an orthotropic material with thermal expansion less than 10% of the case composing the domain with any of the material phases used.

In a process of designing any ultra-efficient mechanical object, one of the most important issues is to minimize an overall resistance forces in each of its components, that reduce the efficiency and performance. However, the process is complex and time-consuming, as each element should be treated separately and a whole object must be also considered as a unity. With a usage of different optimization tools, a designer is able to obtain the best possible solution with given constraints and conditions. In this paper, an optimization approach for an ultra-efficient electric vehicle is presented. The analyzed object is a supporting system for an electric vehicle powered by a hydrogen fuel cell designed by Smart Power Team from the Silesian University of Technology, which is dedicated to participating in Shell Eco-marathon competition, in a UrbanConcept class. In a given project, the optimization process is aimed at reducing total mass with ensuring sufficient stiffness. In case of the supporting system based on shell elements, three main areas of optimization can be distinguished: body’s shape, other subassemblies’ placements and their connections to the system and an inner and outer structure of shell elements. As the supporting system is not a part of a drivetrain, its impact on the vehicle’s total efficiency is not direct and should be analyzed in two ways: aerodynamic properties of the outer shape and mechanical properties of the supporting system that is mass and stiffness. In the paper, the optimization process, as well as obtained results, are presented.

Continuous development of technologies, state of the environment and market demands emerging from them require creating energy efficient vehicles for the daily commute purposes. The goal of this paper was to present optimization process of the load bearing structure designed as spatial frame, which was a part of a hydrogen fuel cell powered vehicle, that can be described as a small urban car. Due to its purpose, it was not travelling with high velocities and did not had to cope with major road irregularities. The structure had to obey rules and regulations of Shell Eco-Marathon competition, in which it was going to take part in under the custody of Smart Power Team from the Silesian University of Technology in Gliwice. Impact on efficiency, from the structure point of view, came from aerodynamics, weight and stiffness, which may influence different subassemblies efficiency. The analysed vehicle already possessed shape of the fuselage, which was previously optimized to improve aerodynamics, therefore its shape was not taken into consideration. Goal of the structure was to obtain low weight, therefore materials used in the load bearing system were mainly composite materials like carbon fibre tubes and sandwich structures, but also 3D printed elements. The secondary objective of the conducted work was to develop a structure, which could be manufactured and assembled by students. To obtain this goal, without sacrificing the safety of the vehicle and its driver, a special type of joints between carbon fibre tubes were applied in this design.

An innovative research front for composites is their production by additive manufacturing (AM), also referred to as 3D printing. AM, which was previously mainly used for prototyping, is now evolving towards functional components. Part of the motivation in the search for AM developments to fibre composites is to explore the inherent flexibility of the related processes, for example the possibility of laying curved fibre-paths within a component. This work aims to present a variable stiffness curve-based design parametrization to the optimisation of a 3D printed aeronautical lug, made of thermoplastic polymer reinforced by continuous fibres. Results on structural modelling for stiffness and strength show that these quantities behave in different manners for the assumed parametrization. An optimisation problem is proposed, and it is expected that optimised designs for stresses may further enhance the applicability of 3D printed composites in load bearing situations.

This approach aims for decreasing costs in lightweight structures using FRP composite materials based on a hybrid construction where expensive and high-stiffness materials performs together with inexpensive and low-stiffness material. The proposed optimal design of hybrid composite stiffened structures addresses sizing, topology and sustainable material selection in a multi-objective optimization framework. Minimum weight (cost) and minimum strain energy (stiffness) associated with sustainable factors are the objectives of the proposed structural robust design approach. The model performs the trade-off between the performance targets against sustainability, depending on given stress, displacement and buckling constraints imposed on composite structures. The design variables are ply angles and ply thicknesses of shell laminates, the cross section dimensions of stiffeners and the variables related to material selections’ and structural distribution. A Multi-objective Memetic Algorithm (MOMA) searching Pareto-optimal front is proposed. MOMA applies multiple learning procedures exploring the synergy of different cultural transmission rules. These rules are associated with some kind of problem knowledge and learning classified as Lamarckian or Baldwinian. The memetic learning procedures aim to improve the exploitation and exploration capacities of MOMA as shown by the numerical simulations.

An automated prepreg fabric draping system is being developed which consists of an array of actuated grippers. It has the ability to pick up a fabric ply and place it onto a double-curved mold surface. A previous research effort based on a nonlinear Finite Element model showed that the movements of the grippers should be chosen carefully to avoid misplacement and induce of wrinkles in the draped configuration. Thus, the present study seeks to develop a computationally efficient model of the mechanical behavior of a fabric based on 2D catenaries which can be used for optimization of the gripper trajectories. The model includes bending stiffness, large deflections, large ply shear and a simple contact formulation. The model is found to be quick to evaluate and gives very reasonable predictions of the displacement field.

The knee prosthesis development has tried many approaches including passive, dynamically damped and powered designs. Nevertheless, the passive approach still has some very important aspects to contribute to this complex design process and must be restudied before applying any powered approach. The synergistic combination of this two perspectives is necessary to conceive the knee prosthesis implementation as a biomechatronic design and to take advantage of the best from each approach.The purpose of this study is the dimensional synthesis of a polycentric six-bar mechanism to be implemented as a knee prosthesis. For this reason, a constrained numerical optimization is proposed where the main objective is the minimization of the error when following the desired trajectory, previously obtained as a result of the Mexican anthropometry studies. Additionally, this study adds, as a requirement, the avoidance of the knee hyperextension with the mechanism itself instead of using mechanical stops added into other stages of the knee design. This work handles this new requirement by applying a set of new constraints in the optimization problem to impose the desired behavior as well as forcing the synthesized mechanism to be a Non-Grashof mechanism. The optimization problem is solved by using an evolutionary technique, because of its simplicity in the implementation and the good results reported in literature when solving real-world engineering problems. The Differential Evolution algorithm is applied in combination with some simple but efficient rules to handle the constraints.By following this route the obtained mechanism follows the desired trajectory, avoids the hyperextension and its configuration is more compact due to the folding characteristics of a Non-Grashof configuration. Finally, the obtained mechanism simulation is presented and it is modeled using CAD tools.

The multiscale modelling of materials plays important role in engineering design and biomechanical analysis of tissues. The multiscale methods allow to take into account influence of microstructure of the macro model behavior. The paper deals with inverse problems solution in multiscale modelling. The bioinspired algorithm are used for optimization and the Finite Element Method used direct problems solving.

Rolling noise emitted by railway wheels is a problem that affects human health and limits the expansion of the railway network. This problem is caused by the wheel-rail contact, and it is predominant over the rest of noise sources from the vehicle/track system for the usual speed conditions in urban areas. The minimization of rolling noise through changes on the wheel shape by means of the finite element method is discussed in this work, which focuses on potential shape modifications in existing wheels in the form of an optimal wheel web perforation distribution. Such a modification is a cost-effective solution that can be performed in a relatively short term in already manufactured and operating railway wheels. To this end, two objective functions with different computational costs are studied and analysed with several configurations of a genetic algorithm-based optimizer. Both approaches focus on minimizing rolling noise. Approach 1 is based on the minimization of the area below the sound power vs. frequency curve of the wheel, and thus requires solving the system dynamics. On the other hand, Approach 2 is based on the maximization of the natural frequencies of the wheel in order to shift its resonances out of the excitation range, and therefore it only requires a modal analysis. The acoustic radiation analysis is performed through the computation of the normal surface velocities, using a time-domain approach and including a contact filter applied in the track roughness, considered as excitation. Moreover, the structural requirements for fatigue strength in wheels proposed by the optimizer are ensured according to actual standards. Results using Approach 1 reflect that an optimized distribution of perforations on the web of a railway wheel, can reduce significantly the sound power level in the entire studied frequency domain (0–5 kHz). This is related to the high sensitivity of the acoustic radiation response with the perforation pattern. Such a phenomenon appears to have a higher impact on noise minimization than that associated with the reduction of the radiating surface due to perforations. The high reduction of the radiated sound power is primarily due to the fact that certain wheel vibration modes with high acoustic contribution are shifted out of the excitation range corresponding to the contact force, this effect being observed in the best solution of Approach 1. Less significant sound power reduction is obtained with Approach 2, although its associated computational cost is considerably lower.

Regarding acoustic attenuation, reactive silencers show a good behaviour at low to mid frequencies, whereas dissipative silencers, i.e., those containing absorbent material, are used in exhaust system applications due to their good performance in wide frequency bands, essentially at high frequencies. The combination of both types of configurations results in the hybrid silencer, which shows considerable attenuation in almost all the frequency range of interest. As an initial approach, plane wave models with low computation cost can be employed for acoustic attenuation prediction mainly at low frequencies. However, for more general problems with complex geometries and heterogeneous absorbent material it is necessary the use of multidimensional numerical analysis methods, such as the Finite Element Method (FEM). In the present work, the axisymmetric FEM model using a pressure formulation [1] is used in order to predict sound attenuation in dissipative and hybrid silencers in terms of transmission loss (TL). Next, a Topology Optimization (TO) gradient-based algorithm is implemented in order to get the maximum attenuation in a certain target frequency range, assigning a different filling density to each element of the central chamber, while keeping constant the total mass of absorbent material. The adjoint method is used during the TO process in order to speed up the computation of the objective function sensitivities with respect to the design variables (bulk density of each element or group of elements). This TO method is combined with the geometry optimization of the dissipative and reactive chambers. Thus, the implemented algorithm is capable of obtaining not only the optimal absorbent material layout but also the optimal geometry of both chambers (radius and length) which maximizes the silencer transmission loss. In order to combine both techniques, the Method of the Moving Asymptotes (MMA) described in reference [2] is implemented. The FEM solver algorithm is validated with experimental results available in the bibliography. Finally, it is demonstrated that the optimization process results in substantial improvement of the silencer sound attenuation at the target frequency range.

In this paper, we address the problem of workload distribution in Integrated Modular Avionics (IMA) systems. IMA system is a real-time computer system consisting of computing modules connected by a communications network. Each module has a multicore central processor unit (CPU). Workload for an IMA system is a set of periodic computational tasks grouped into partitions. Tasks from different partitions communicate by message passing. During workload distribution, each partition is assigned to a single CPU core of some module. Message passing between partitions assigned to cores of different modules creates network load. To guarantee system scalability, the network load must be minimized, while keeping all core loads within specified limits. We represent the workload distribution problem as a modified multiple knapsack problem and propose a branch and bound algorithm for solving this problem, along with a specific scheme of upper estimate calculation and an optimization which improves the algorithm performance. Results of experimental evaluation of the algorithm are also presented. The proposed algorithm is implemented in a tool system accepted for operation by one of the leading Russian aircraft design companies.

Object of research is the adaptive gear variator. In present the concept “variator” is used only for frictional mechanisms. In the article it is proved, that the gear mechanism with constant engagement is capable to execute the variator functions. The adaptive toothed variator represents the planetary gear mechanism with constant engagement of cogwheels and with the variable transfer ratio. The variator works without a control system. Basic difference of a gear variator is the mechanism has two degrees of freedom and only one input. Definability of motion provides the planetary four-bar closed contour creating additional power constraint with circulation of energy. The variator is created on the basis of a discovery of professor Ivanov K.S. «Effect of force adaptation in mechanics». The essence of the discovery is a planetary kinematic chain with two degrees of freedom in which cogwheels form the mobile closed contour, adapts for variable loading by an independent regulation of motion speed of cogwheels in a contour under the influence of loading. The basic theoretical regularities of creation of a gear variator, the description of a development type of a variator and the test-bed, technique of conducting of tests and results of tests of the adaptive gear variator which are confirming effect of force adaptation are presented.

This paper presents a novel algorithm for power distribution network reconfiguration (DNR) and optimal allocation of solar powered distributed generation (DG) sources taking as objective function the reduction of power losses in the system. The proposed algorithm allows for simultaneous solving of the DNR and optimal DG allocation tasks. The adopted approach is a customized evolutionary algorithm which utilizes recombination operators for the preservation of the radial structure of the network, integer-based operators to determine the bus for DG placement, and floating-point operators for handling the power output of the DG sources. The algorithm is tested on a 69-bus network. A comparative analysis reveals an outperformance of state of the art methods presented in the literature. As a measure for result quality, a statistical analysis from multiple algorithm runs is reported and analyzed.

Common Structural Rules for Bulk Carriers and Oil Tankers (CSR) issued by IACS provides requirements for loads, hull girder strength, hull local scantling, prescriptive buckling, etc. Literatures show that there are lots of studies on mid-section structure optimization of Oil tanker. However the design constraints in these studies are usually simple instead of considering a wide variety of requirements in CSR, thus the optimization results are not very practical for engineering applications. Since the requirements in CSR are complicated and updated continuously, software by classification societies is designed and used to assess the prescriptive requirements of the CSR. In this paper, MARS2000 by BV is integrated and combined with methods of design of experiment (DOE) and approximate model, to carry out a mid-section structure optimization of Oil tanker. The above strategy and procedure provide an available and effective method for the mid-section structure optimization of Oil tanker based on CSR prescriptive analysis.

The paper industry is the industry with the third highest energy consumption in the European Union. Using recycled paper instead of fresh fibers for papermaking is less energy consuming and saves resources. However, adhesive contaminants in recycled paper are particularly problematic since they reduce the quality of the resulting paper-product. To remove as many contaminants and at the same time obtain as many valuable fibres as possible, fine screening systems, consisting of multiple interconnected pressure screens, are used. Choosing the best configuration is a non-trivial task: The screens can be interconnected in several ways, and suitable screen designs as well as operational parameters have to be selected. Additionally, one has to face conflicting objectives. In this paper, we present an approach for the multi-criteria optimization of pressure screen systems based on Mixed-Integer Nonlinear Programming. We specifically focus on a clear representation of the trade-off between different objectives.

The improved design of recirculated aquaculture systems (RAS) is needed facing the demand for increased fish production as well as increased concern of fish wellbeing. Here, we make a step towards using of computational fluid dynamics (CFD) for the optimization of fish tanks. The proposed CFD based methodology allows the modeler (the designer) to manipulate both the tank geometry and operating conditions, in order to solve a multicriteria optimization problem. The individual objective functions are quantifying multiple criteria, dealing mainly with the rearing conditions for fish (a determined average velocity and low velocity variance), fast biosolids removal and the cost of energy and place. As an example of our methodology, we present a study involving the CFD analysis of four different RAS: (i) the circular, (ii) the octagonal, (iii) the square, and (iv) the rectangular multivortex RAS, either with or without additional plate baffles between two water inlets. Based on the two-dimensional description of the flow field within each RAS configuration, the set of efficient solutions for RAS is inferred.

To increase pressure to supply all floors of high buildings with water, booster stations, normally consisting of several parallel pumps in the basement, are used. In this work, we demonstrate the potential of a decentralized pump topology regarding energy savings in water supply systems of skyscrapers. We present an approach, based on Mixed-Integer Nonlinear Programming, that allows to choose an optimal network topology and optimal pumps from a predefined construction kit comprising different pump types. Using domain-specific scaling laws and Latin Hypercube Sampling, we generate different input sets of pump types and compare their impact on the efficiency and cost of the total system design. As a realistic application example, we consider a hotel building with 325 rooms, 12 floors and up to four pressure zones.

The increase in quality standards in the automotive industry requires specifications to be propagated across the supply chain, a challenge exacerbated in domains where the quality is subjective. In the daily operations of ThyssenKrupp Presta AG, requirements imposed on the vibroacoustic quality of steering gear need to be passed down to their subcomponents. We quantify the influence of ball nut assemblies on the steering gear by finding optimal encodings of their respective vibroacoustic signals, iteratively maximizing the correlation of their order spectra under orthogonality constraints. We compare the performance of linear and non-linear variants of this approach known as Canonical Correlation Analysis and establish the superiority of the neural network based variant in terms of attainable correlation. The practical relevance of our findings is guaranteed, since the visualization of the weights enables the identification of core influence areas.

The automotive future demands light and affordable designs and components. To develop lighter parts in a decreased time structural optimization is used in many R&D departments.To ensure an ideal design of cast parts in the topology optimization a casting simulation is integrated in the optimization procedure. With this additional simulation the castability can be optimized in parallel to the mechanical properties and the weight. The result is a manufacturable and light structure, which can be transferred into a real part easily. To describe the geometry a regular voxel mesh in combination with a binary design variable per element was used in the past.In this paper an approach is shown to use the available routines of the topology optimization to optimize the shape of the part on a small scale, too, and provide a smooth optimization result. This is done by using a design variable per element, which can be continuous between zero and one in the surface layer. In this way all implemented functions and manufacturing restrictions can be used for both topology and shape optimization. To avoid numerical difficulties the FEM simulations are done with a converted tetrahedron mesh. This increases the quality of the simulation by guaranteeing a smooth surface and avoiding sharp edges. The CFD based casting process simulation is still done with the very fast, structured hexahedron elements. The results of the simulations are mapped on the optimization model to change the part design again. Beside the increased degrees of freedom for doing shape optimization and the higher simulation quality with adapted meshes this optimization procedure allows easier the smoothing of the surface.

Batch annealing process is highly automated and commonly used heat treatment process which is eligible for industrial applications. In this process, steel coils are vertically stacked one on top of each other and heated up to 720 °C between 13–25 h. Temperature distribution and annealing time of each coil depends on weight, thickness, outer diameter, height, vertical position, etc. Having an optimal combination of coil batches increases the productivity of the process and to create the optimal batches, annealing times of each combination should be predicted. For this purpose, a validated heat transfer model is used to simulate past 2 years’ processes. More than 29,000 data are obtained which covers different coil parameters and annealing programs. A neural network model is constructed to predict the annealing time. Error rates of annealing time prediction model are between 30 and 50 min which is quite successful. After the validation of the annealing time prediction model, an Integer Programming formulation is developed for the optimal batch creation problem. An alternative approach is proposed for selecting the best batch combinations in a dynamic production environment. Increase in production efficiency is demonstrated in the computational study.

The article presents a method for improving the characteristics of fuel pump of rocket engine by the joint usage of mathematical optimization software IOSO, meshing complex NUMECA and CFD program ANSYS CFX. The optimization software was used for automatic change of the geometry of low-pressure impeller, transition duct and high-pressure impeller to find the optimal design. The original variant of the remaining parts of the pump was kept. For this reason, only geometrical parameters of the blades were varied without changing the contours of the pump meridional flow part. The investigated pump consists of five parts: inlet duct, low-pressure screw centrifugal stage, transition duct, high-pressure screw centrifugal stage and volute outlet duct. The pump main parameters with water as the working fluid (based on experiment data) were the following: high-pressure stage rotor speed was 13300 rpm; low-pressure rotor speed was 3617 rpm by gearbox; inlet total pressure was 0.4 MPa; outlet mass flow was 132.6 kg/s at the nominal mode.Comparison of calculated characteristics of the basic pump with the experimental data was performed before the optimization.The analysis of characteristics for the obtained optimized pump geometry was carried out. It was found that pump with optimized geometry has greater efficiency in comparison with the original pump variant. The obtained reserve can be used to boost the rocket engine, and/or to reduce the loading of the main turbine, which operates in aggressive oxidizing environment.

Reversible cold milling is a widely used industrial process for metal forming. It consists of backward and forward motion of the sheet metals between the rolling mills. As the process is highly automated, optimization of the process parameters is essential for cost minimization. However, it is a complex optimization problem because maximization of the rolling speed increases the motor mill currents and rolling force, and physical constraints of the line may not allow to achieve that. In this paper regression and artificial neural network models are developed to predict process parameters. Genetic algorithm with the fitness function based on regression model is used to optimize total time of process. Performance of both prediction method and genetic algorithm are validated with past industrial data and field experiments. The results show that the developed method is an effective tool for the process optimization of reversible cold rolling process.

Since 1913, the Harris’ model is adopted within intermittent production systems to size the batches to produce and purchase. For each product, the model sets the so-called Economic Order Quantity (EOQ) as the quantity optimally trading-off the cost of orders and the average stock cost. Traditionally, the EOQ from the Harris’ model is a milestone for make-to-stock (MTS) production systems. In addition, existing extensions of the base model are in the direction of including multiple actors of the supply chain, i.e. joint economic lot size, and tailored product management policies, i.e. consignment stock. A basic hypothesis behind the lot size models is that the production line productivity is higher than the average market demand so that a dynamic equilibrium becomes feasible. Nevertheless, in the case of permanent or temporary high product request, the productivity of a single production line can be insufficient. This case makes of interest the adoption of multi-line production systems. Such systems are made of parallel production lines able to produce the same product at the same final qualitative standards so that the output is a unique batch of identical products.This paper investigates MTS multi-line systems presenting two formulations of the EOQ model for the case of identical lines (1) and the case of lines with different productivity and setup cost (2). Finally, an application of the model is done with data taken from a leading company operating in the beverage packaging sector.

The article presents one optimization method for improving of the working process of an axial compressor of gas turbine engine. Developed method allows to perform search for the best geometry of compressor blades automatically by using optimization software IOSO and CFD software NUMECA Fine/Turbo. Optimization was performed by changing the form of the middle line in the three sections of each blade and shifts of three sections of the guide vanes in the circumferential and axial directions. The calculation of the compressor parameters was performed for work and stall point of its performance map on each optimization step. Study was carried out for seven-stage high-pressure compressor and three-stage low-pressure compressors. As a result of optimization, improvement of efficiency was achieved for all investigated compressors.

Smart and distributed energy micro-production is the new pattern for the electric energy supply, joining high service level and sustainability issues. Within such a context, the renewables, i.e. solar photovoltaic (PV), micro-wind, etc., play an increasing role as part of the source mix because of their capillary presence and the decrease of the required initial technology investments. On the contrary, the renewable intermittence is the key weakness to overcome to make a turning point to their final spread. To this purpose, hybrid energy systems join the plus of having renewable modules to the plus of having backup traditional units activated in the case of lack of energy.This study presents and applies to an Italian rural context a linear programming model to best design and manage a local off-grid renewable smart energy system. The power system may include PV and micro-wind technologies together with a battery bank and diesel generator as the backup system. Starting from the expected average load profile, the environmental conditions and the technical features of the energy modules, the model selects the most suitable energy sources, optimizes the power rates of each unit and manages the energy flows within the system. The final goal to achieve is to minimize the levelized cost of the produced electricity (LCOE) making such a system competitive respect to fully fossil fuel based energy systems.The aforementioned case study exemplifies the model application focusing on a remote scientific center requiring electric energy for its daily research activities. The area where the center is located is badly connected to the national grid and, actually, a fossil fuel generator is used, only, to provide electricity. An as-is vs. to-be differential analysis assesses the effect of introducing a dedicated renewable smart energy system finding its economic feasibility over a 15 year lifetime. Evidences show the convenience of exploiting the solar source, while little convenience is for micro-wind installation because of low available wind power and the increasing system complexity. Globally, the LCOE is close to 0.14 €/kWh making competitive the hybrid energy solution, close to the evident environmental benefit.

Supply chain is a group of distributed entities interacting with each other and aiming to improve their business strategies. However, due to globalization, supply chain’s companies grow more and more connected as they face challenges related to complexity. Consequently, in order to keep on improving, companies must cooperate and act as a centralized system; this cooperation is mostly accomplished through information sharing. Indeed, information sharing is pivotal in achieving coordination and visibility along the supply chain. Furthermore, effective planning activities in a supply chain depends highly on the sales forecasts, which is a very complex task due to the uncertain character of the costumer’s demand and various other internal and external parameters.This paper tackles the issue of information sharing by studying its effect on demand’s prediction in the context of a multi-echelon supply chain. It presents an agent-based approach to model coordination by sharing demand forecasts in a supply chain. Multi-agent system is used to model cooperation among multi-echelon supply chain’s members. Moreover, neural network forecasting ability is applied to train agents in order to predict the costumer’s demand of a certain product by demonstrating the importance of information sharing. The proposed agent-based system models sharing demand forecasts as a coordination tool in a multi-echelon supply chain with a concrete numerical experimentation using real supermarket data. Besides, sales forecasting is realized based on different architectures of neural networks including recurrent and feed forward structures.

Trajectory planning is considered a fundamental concern in robotics. In this paper, we discuss the use of optimization techniques to obtain optimum trajectories of industrial robots. We use the flexibility of optimization techniques to address different formulations and solve them using the Feasible Direction Interior Point Algorithm (FDIPA). This method essentially solves two linear systems in each iteration to compute a descent and feasible direction of the problem, then performs a line search procedure that assures global convergence and feasibility of all iterates. Initially, it will be presented the physical description of the tasks to be executed by the serial robotic manipulator. At first, we discuss point-to-point collision-free paths, that is, given an initial pose of the robot and a final target point, find an optimum trajectory that minimizes time, total displacement, energy or other performance index while avoiding collision with an obstacle. Then we discuss the path-following cases, where, given the desired trajectory of the end-effector, optimum joint trajectories are calculated, such that velocity or acceleration peaks are minimum. Further, both cases are formulated as optimization problems, which we also deal with joint mechanical limits (maximum displacements, velocities and accelerations) as constraints. Finally, we use a 4 degrees-of-freedom (DOF) planar manipulator to present numerical examples. Our results prove the effectiveness of the proposed approach and ensure robustness and applicability of the optimization method in the context of robotics.

Integrated in the LEEUAV project, the objective of this work was to optimize the propeller-driven propulsion system previously implemented. A propeller was parametrized in terms of planform and airfoil shape and the software QPROP used to evaluate the performance of in terms of thrust, power and thrust coefficient and propeller efficiency. Experimental tests were conducted for three different propellers to study study the performance sensitivity to propeller diameter and pitch, electric motors, and also to validate the numerical model. Following those tests, a multi-objective shape optimization using MATLAB $$^{\textregistered }$$ ® , for cruise and climb conditions, was performed. At the end of this optimization, a system motor $$+$$ + propeller with an higher efficiency was obtained.

Today’s industrial optimisation process for aircraft structures does not consider flight control systems. In this paper a method to couple flight control systems with aerostructural design in the scope of optimisation is presented. Further, an overview on necessary methodical enhancements coming with controller integration is given. The integration is demonstrated on a flexible aircraft model. A generic pitch controller is coupled with the aeroelastic solver. It is demonstrated how a simple disturbance of a trimmed state can be controlled and how critical time steps can be detected. Wing skin composite layers are sized with the selected loads. Structural responses from the controlled system are compared to those of the uncontrolled system for both the baseline and the optimised model.

The Modified Cuckoo Search was used to customise the design of a pre-mixer for use in a pressure let-down station heater. This is a true industrial test application for this novel optimisation algorithm. A commercial-off-the-shelf (COTS) mixer was used as a bench mark to assess the impact of Modified Cuckoo Search. In addition to customisation of the pre-mixer’s design using Modified Cuckoo Search, the commercial Ansys Design Exploration v16.0 software was applied to the problem. This package utilises the Non-Dominated Sorted Genetic Algorithm-II (NSGA-II). The results of our study showed that by using these optimisation algorithms at least a 30% improvement in the performance as compared to the COTS design could be achieved. Further to this, it was found that the free and open source Modified Cuckoo Search performed comparatively to the commercial package Ansys. This will be significant for small to medium sized enterprises who are often limited by budget constraints to using off the shelf devices.

Thermal cracking is one of the most energy-consuming process in the chemical industry and its optimization has become a real challenge for the research community. In this context, this paper proposes two metaheuristic approaches based on the Genetic Algorithm (GA) and the Harmony Search (HS) algorithms for minimizing the sum of the Energy Consumption and the Water Use in the overall thermal cracking process. Simulation results show that HS achieves best average minimum and mean values than its counterpart GA.

The efficient and flexible design of renewable power plants is key to increase competitiveness of clean technologies and accomplish climate targets. However, renewable power plants that deliver energy only when the renewable source is available produce large fluctuations and increase cost of integration in the wider electricity system. Optimised design of renewable power plants with energy storage increases reliability and decreases integration cost of sustainable technologies. Here we show a two-stage multi-objective optimisation framework to optimise the design and the operation of power plants that combine two or more generation technologies and energy storage, with the aim of producing firm or dispatchable electricity. With the optimisation framework it is possible to handle different technologies to design a sustainable, cost competitive, flexible and dispatchable power plant. Besides, the post-optimisation analysis handles other key performance indicators and provides detailed information that improves the decision making.

Optimal feed temperature was determined for a non-isothermal fixed-bed reactor performing hydrogen peroxide decomposition by immobilized Terminox Ultra catalase. This feed temperature was obtained by maximizing the average substrate conversion under constant feed flow rate and temperature constraints. In calculations, convection–diffusion–reaction immobilized enzyme fixed-bed reactor described by a set of partial differential equations accounting for parallel and thermal deactivation was taken into account and based on kinetic, hydrodynamic and mass transfer parameters previously obtained in the process of decomposition. The simulation showed the optimal feed temperature to be strongly dependent on hydrogen peroxide concentration, feed flow rate and diffusional resistances expressed by biocatalyst effectiveness factor. It has been shown that the more significant diffusional resistance and the higher hydrogen peroxide conversions are, the higher the optimal feed temperature is.

In a wide number of engineering applications, the interpolation of a lookup table (LUT) can substitute expensive calculations. The generation of a LUT consists of pre-calculating a set of quantities from a collection of points that covers the study domain where the interpolation is possible. Nevertheless, the selection of points where the exact calculation is performed is of utmost importance for the LUT size and accuracy. Thus, the goal of this paper is to provide a dedicated optimization tool for the generation of accurate and efficient LUT. Here, the domain of the LUT is structured by the so-called layers, in which, the thicknesses of each layer define the distance between pre-calculated points. The optimization problem consists of maximizing the layer thicknesses, that is, minimizing the LUT size, such that the interpolation errors within the layer domain are kept under specified tolerances. Thus, a sequential design approach is applied to design each layer of the LUT until the layers cover the study domain. To achieve reliable LUT generations, a new optimization algorithm has been implemented to reach optimal layers with minimum iterations. The strategy proposed here is applied to generate a LUT to substitute a selected analytical function. The optimization procedures demonstrate not only the performance of the optimization algorithm, but also its convenience in the generation of multidimensional LUT.

Many infrastructures were built in 1950’s and 60’s intensively in Japan. They got older in half a century, and now it is an urgent task to inspect, monitor and maintain such aging infrastructures. The Ministry of Land, Infrastructure, Transport and Tourism (MLIT) renewed the guideline of tunnel inspection and monitoring in 2014. In this research, we proposed optimization models to level the yearly budget for tunnel inspections and monitoring activities under the MLIT’s guideline. It is very important to resolve the bias of the cost of inspection and monitoring from year to year because Japanese local governments adopt the single-year budget system and would like to set aside the constant budget for inspection and monitoring each year. We conducted some numerical experiments with the practical data of tunnels and confirmed the effectiveness of our proposed models.