Search

In this work, an optimisation workflow is presented for uncertainty-based design optimisation using a multi-fidelity Gaussian Process Regression. The motivation of the proposed techniques is to solve a propeller design optimisation problem under uncertainty. This problem involves expensive CFD simulations to evaluate the aerodynamic performance which prohibits the application of standard optimisation techniques and the direct calculation of statistical measures. Therefore, a surrogate-based optimisation workflow is presented. The computational budget limits the number of high-fidelity simulations which makes impossible to accurately approximate the design landscape. This motivates the use of cheap low-fidelity simulations to obtain more information about the unexplored locations of the design landscape. The information stemming from the low- and high-fidelity numerical experiments can be fused together with multi-fidelity Gaussian Process Regression to build an accurate surrogate model despite the low number of high-fidelity simulations. Although Gaussian Process can inherently model uncertain processes, here the design and uncertain parameters are treated separately and only the design space is modelled by a Gaussian Process. The probabilistic space is modelled by a Polynomial Chaos Expansion. The combination of the above techniques allows us to efficiently carry out a design optimisation problem which otherwise would be impractical.

In recent years small medium enterprises (SME) have to manage an increasing number of products, while trying to improve effectiveness of the system to achieve the imposed objectives. The high complexity of the production system and the difficulty to manage the system lead to a low effectiveness regarding the objectives, high inventory and low throughput rate. This paper considers an approach to solve these problems, based on the concept of theory of constraints (TOC). Every organization has at least one constraint. A constraint is any cause that limits the organization’s goals. When focusing on the constraints, TOC gives a substantial improvement without the need for excessive resources. The main goal of this paper is the redesign and control of a production flow that achieves specific performances. This is a typical problem SMEs have. To overcome the constraints, a discrete event simulation (DES) software simulates the production and optimises through optimization software. It includes the analysis, simulation and then subsequently the creation of a new production layout and a potential weekly optimization. For maximizing the desiderated performances a rescheduling of the weekly production period is done using the drum-buffer-rope (DBR) method.

The aim of the present work is to define and to optimize an assembly line that processes parts of medium/large dimension. As the assembly line processes large parts, some operational constraints arise, which for example are assembly to order production, just in time production, etc. The objective functions are minimum Earliness and minimum Tardiness stated to a given level of working stations saturation. To achieve these purposes we use two industrial software: the first is a Discrete Event Simulation (DES) package WITNESS 14 by Lanner Group, the second is a design optimization and process integration package, MODE-FRONTIER by Esteco. The DES software allows us to produce a simulation model for the assembly line. The optimization has carried out in three steps. First step is a multi-objective optimization cycle with genetic algorithm that generates a Pareto Frontier solutions; second step is a methodology that clusters the Pareto Frontier in different groups. A set of simplifications characterises this second step with the aim of reducing both the number of inputs and the objective functions. The third step is a mono-objective optimization cycle that allows us to reach a solution by means of a simplex algorithm. This method offers a way to solve scheduling problem of a complex production plant constrained to respect a management policy.

This paper illustrates a proof of concept of a multi-level optimization, involving parameters related to both geometry and mission control. The goal is indeed to find an optimal boomerang in terms of energy requested to throw the object, optimizing at the same time the shape of the boomerang (high level) and the throw parameters (low level). The optimization problem is accomplished by the modeFRONTIER® platform; the throw parameters optimization takes into account also how the parameters uncertainty affects the boomerang performance (return capability).

The definition of Artificial Intelligence that can be found on the pages of Wikipedia is the intelligence exhibited by machines or software which clearly has a rather broad and vague meaning that in many circumstances has been misunderstood. I would therefore focus more on Artificial Intelligence Tools, i.e. the spectrum of mathematical procedure that can be used to gain, explore and exploit knowledge during a design process. To gain knowledge means to probe design opportunities in a systematic way in order to collect sufficient data to be able to understand and predict product behaviour. To explore knowledge means to be able to drive automatically through the design options using optimization techniques. To exploit knowledge means to be able to take rational decisions about the configuration of a product to be produced. All these actions can be performed by means of software components based on AI-related tools: Neural networks, Evolutionary Computing, Classifier Systems just to name a few. In the development of decision support software for design optimization there is not one technique that would prevail but a blending of tools, including more traditional mathematical algorithms, that contribute to the finding of the best design configuration. In this presentation a selection of industrial application form transportation industry to consumer goods will be used to showcase the use of AItools in daily design activity while possible future needs will be identified by looking at the opportunity offered by collaborative environments.

AlNiCo magnets are permanent magnetic alloys based on the Al-Ni-Co-Fe system. In the present work, Sobol’s algorithm was used to randomly generate alloy compositions so that they are properly distributed in the variable space. The final 80 candidate alloys were selected after examining the phase equilibria and magnetic property predicted by thermodynamic database FactsageTM in the desired temperature of exposure. The magnets were synthesized and tested for desired properties of interest. The various properties were fitted by response surfaces generated by Radial Basis Function module available in commercial optimizer,”modeFRONTIER”. It was followed by optimization for predicting alloy composition for improved properties. This task was simultaneously performed by another commercial optimizer, IOSO. The 5 Pareto optimized candidate alloys were synthesized and tested. One of the optimized candidate alloys dominated the initial 80 candidate alloys in most of the properties. This proves the efficacy of response surface methodology in optimizing the desired properties while minimizing the time and cost in synthesizing the alloys by random experimentation [less]

Obiettivo del presente lavoro è l’analisi del consumo energetico dei forni professionali per la cottura di alimenti, allo scopo di individuare nuove soluzioni per il risparmio energetico e la riduzione dei costi di esercizio. Quindi viene presentato e validato un modello concettuale del processo di cottura basato sull’analisi dei flussi energetici coinvolti durante un generico processo di cottura, considerando il sistema forno + cibo. L’evoluzione energetica del sistema viene descritta suddividendo il processo di cottura in tre fasi ed introducendo il concetto dei livelli energetici di cottura, funzione del tipo di forno e del tipo di ricetta. Tale modello viene poi applicato definendo una strategia di ordinamento ottimale delle cotture per la riduzione dei consumi globali d’esercizio.

In the last decades renewable energy production has been growing exponentially. In this sector, electricity generation through wind turbines has become relevant. The aim of research is to enhance efficiency and energy production. In this article, in fact, the optimization analysis of a 3 MW horizontal axis wind turbine (HAWT) blade is presented. The aerodynamic performances are evaluated through an in house BEM code, developed by Airworks s.r.l., which is coupled with modeFrontier® in order to manage incoming data and post process results obtained in an automatic way. Airfoil type, section length, chord and twist distributions have been taken as input parameters. Power coefficient and annual energy production are the output variables used to judge the aerodynamic performance of the resulting blade. Several runs have been done in order to collect information and find the best design for the blade.

In this work an approach is presented for the analysis of the effects of geomet-ric tolerances in fluid dynamic behaviour of manufactured components. The Tensorial-expanded Chaos Collocation method coupled to Fictitious Domain Method has been used to solve Fluid Dynamic problems with geometric uncertainties. The main advantage of the Tensorial-expanded Chaos Collocation method is its non-intrusive formulation, so existing deterministic solvers can be used. The Least-Squares Spectral Element Method has been employed for the analysis of the deterministic differential problems obtained by Tensorial-expanded Chaos Collocation. This algorithm exploits a Fictitious Domain approach, so it is particularly suitable to solve differential problems defined on stochastic domains. The capabilities of the Tensorial-expanded Chaos Collocation method combined to the Ficti-tious Domain-Least- Squares Spectral Element Method are demonstrated by a numerical experiment.