Industrial improvement of CAE through the PIDO - PDF document

Industrial improvement of CAE through th
Industrial improvement of CAE through the PIDO Stefano Odorizzi, University of Padua Abstract The software environments, for the so known PIDO (process integration and multi-objective collaborative design optimisation), highlight technologies for CAE far beyond the exclusively technical and just specific context of application, putting forward methodological innovations of great efficacy in design projects – and associated productive repercussions – not depending on the dimension and the complexity of the industrial environment we are referring to. 1. Numerical methods CAE is the category that includes –or anticipates, in the incessant evolution of the terminology, which characterizes the computational world- software technologies devoted to virtual prototyping/experimentation and process simulation. It refers to the possibility to simulate, through a computational process, a physical phenomenon – more frequently in relation with problems, broadly speaking, associated with design – without needing to fit out the real experiment and/or to build the prototype. The most interesting applications for industry go from mechanics to acoustics, from fluid dynasystems dynamics to impact dynamics, including the simulation of manufacturing processes and of interactions with humans and environment too. The technologies we are talking about, are today globally called “digital prototyping” (DP) also, term which recalls the methodological approach which stands at the implementation basis: numerical methods, as said. These methods firstly appeared in computational programmes, beginning from the 70i

es, on computers devoted to the industry
es, on computers devoted to the industry, and which have progressively demonstrated their practical value with the increase of calculation power of computers. The basic idea is quite elementary and it can be very easily summarized as follows. Those phenomena, which involve a physical continuum, are normally described by laws that find their formulation in differential equations with partial derivatives. Their solution (integration) can be possible (or known) in just a very limited number of cases, and in relation with considerable simplifications, that is, observing very precise boundary conditions. An important mathematic, on the subject, once said: “Differential equations with partial derivatives were invented by God, whereas boundary conditions were thought up by devil.” Numerical methods aren’t influenced by this obstacle, determined by boundary conditions, because – at the cost of controllable approximations – they replace differential equations with an algebraic equations system, which is easily dealable till a huge dimension (number of equations) and solely in dependence on the computational power available. In order to give a transposed, even if a bit rough, interpretation of these methods, we could think about reality as a picture and about the numerical model as a mosaic which tries to recreate it: when the number of 1 tessera grows the mosaic becomes more and more similar to the picture (like on the computer screen, where the images are the result of a pixel matrix). The mosaic, in this case, will represent much more than an image: its colour describes the way in which the physical phenomenon an

alyzed reveals itself in the continuum i
alyzed reveals itself in the continuum included in the image. Obviously, the possible implementations can be much more sophisticated: for instance, to keep the same comparison, in the numerical mosaic it is possible to use tessera in which the colour, instead of being one and homogenous, is changeable following predefined functions. The advantages of the numerical approach are not limited to this: physical laws, referring to discretized continuum, can be reduced to compatibility conditions and balance reactions which can be written in a direct and often elementary way. The material, or materials, behaviour which constitutes the examined continuum and which expresses the quantities relation, there characterized, can take into account, without further complications, unhomogeneity and anisotropy and be indifferently described with arbitral – conventional and non-conventional- models or assumed in relation with reference tests, to be in case standardized . If, to all this, we add the functionalities inserted in commercial calculation programmes, to simplify and fasten the necessary operations to models adjustment – geometric models importation, automatic discretization, graphic description of boundary urbations, materials database, automatic controls on data compatibility, and similar, it is possible to understand how many versatile, powerful and reliable tools, within the reach of designers, are available today. Virtual prototyping technologies permit in detail analysis – and associated designing choices – also in presence of domains and extremely complex phenomena, requiring, on the other side, a use of time, computa

tional, economic and human resources, in
tional, economic and human resources, incomparably reduced than those indispensable to direct physical experimentation. An example is the one presented in picture 1. It refers to a die-casting process simulation and the following fatigue behaviour of a light alloy rim. The forming process is evaluated in the two different phases of the mould filling and following fused casting solidification. The problem is the high complexity, for both the component geometry and, above all, the physics to be treated, which regards both fluid dynamical aspects and metallurgical phenomena. The description is forced to the predictions of imperfection in the casting and of the crystalline structure, which locally develops. In its turn, it recalls mechanic properties distribution, in relation with which the calculation of the life expectancy of the component gains sense. It is a huge step forward in relation with both conventional calculations – that cannot be set aside from the consideration of a behaviour homogeneity, nothing realistic – and direct experimentation, that permits, if practicable, the analysis of an extremely limited number of samples. Nevertheless, if we stop at the solely virtual experimentation, we actually end up using a tool that, how powerful it may be, remains limited to the area of the evaluations and of the possible decisions, taken by the designer or the experimenter- or, if better, by the single affecting the design process and the global evolution of the knowledge, beginning from the society “core knowledge”. The further step can be made using software environments for the PIDO, that is, the “proces

s integration and multi-objective collab
s integration and multi-objective collaborative design optimisation”. 2 2. The PIDO Approach The PIDO approach is the solution pointed out in very recent times, which allows to manage effectively the design process, orienting it to the product-process optimum, in relation with metrics not just technical – but, for instance, of cost, product, marketing – in multidisciplinary contexts, or in case, of parallel engineering. In comparison with other methods- like, for instance the KBE, Knowledge Based Engineering or those specthe company efficiency, like “sigma six”, or other specific for the so called “Evolutionary Economics” – the PIDO uses as “nodes” of the fluxes which express, with a progressive formalization, the logic of processes, calculations and information obtained through numerical models and thanks to virtual experimentation. The PIDO is based, on one hand, on the observation of how the design process evolves, and on the other hand, on the formalization of what can lead to the identification of optimum conditions, which can be contextualized and shared, as best arrangement, in relation with all the different but complementary points of view, in choices regarding industrial product making. As far as the design process, and the way in which it develops, is concerned, the most significant features are: - That of being the main impact on the final product, due to the choices taken in the initial phases of the process, when, on the other side, there cannot be a total clearness on the objectives to be pursued and on the conditions to be satisfied. - That of being diffic

ult, at the beginning of the process, to
ult, at the beginning of the process, to define the measurement criteria for performances to be reached. - The necessity of interacting with the customer and/or with the demand evolution, and/or with the progressive deassociated priorities. - The multidisciplinary of the aspects to be considered under the technical profile, and their dependence on facts and choices of non-technical - Company organization and knowledge transmission. In substance, design is a process, and, as such, it is characterized by subsequent transformations and details additions, which can be separated into phases, receiving the contribution of many professionals and, at the same time, fixing the corresponding decisions. The observation of the way in which the human being works in different design contexts [see, for instance, the study by S. Ahmed, L.T.M. Blessing, K.M. Wallaceon “The relationships between data, information and knowledge based on a preliminary study of engineering designerspresented within the recent ASME meeting on “Design Theory and Methodology”] makes clear that “humans use a basic Generate-Realise-Evaluate process structure which is similar to that which underlies automated search”, even if relevant differences, according to the experience, since "experienced designers operate the realise/evaluate phases in a sophisticated and integrated manner, which shortcuts the need for full realisation and evaluation whenever possible, to home in on any reasons why a design can be eliminated quickly”. The PIDO starts, in this sense, from the idea that the “solution methods can be structured so that the human