This page contains a Flash digital edition of a book.

also by 3D-CFD simulations which were then investigated using ModeFrontier to identify better configurations for the transient cooling system. Te coolant circuit and the air path models

represented in the engine model included two clear groups of parameters integral to optimising the design. Five heat transfer coefficients and four heat capacities were identified and modeFRONTIER allowed the engineers to set up an effective optimisation workflow that was capable of automatically interacting with the Kuli engine model and detecting the optimal configuration for the nine parameters. Günther Pessl, head of simulation at BMW,

said: ‘Te easy-to-build integration between the two soſtware enabled faster identification of the best heat transfer coefficients and thermal inertia in the engine analysed.’ Esteco has also been successful in the

optimisation of components for crash simulation in a project with Volvo. Te idea was to optimise material properties of a bumper to reduce failures and develop a better understanding of how the materials responded under varying stresses. Poloni said: ‘Maybe this is something that

relates to high-performance computing, because today one of the most expensive and time consuming simulations is the crash simulation which may take days on several hundred CPUs and we need to run hundreds of these simulations to get to the optimal result. When you are

designing the components of a car, like a bumper in this case, then what you want to achieve is to minimise the mass of the components while fulfilling the mission, which is typically a pre-described load or impact energy when the car hits something. In this situation it is fairly straightforward to define the objective which is the minimum weight but the constraints are playing a major role there.’ Material test data was used to perform

analysis and curve fitting was performed in modeFRONTIER in order to build and validate



models using response surface methods (RSM) of material properties and predict the behaviour of the bumper. Poloni said: ‘Tere is a large spectrum of

numerical methods under the name RSM, in general it means that you have some data and you want to build an artificial model using only the inputs and the outputs generalising this information to create a model of the behaviour.’ Poloni explained that RSM is a very important tool in two circumstances, the first being you

have experimental data but you do not have a model and you want to correlate the inputs and the outputs. Te other situation is if you have a very expensive simulation, like crash simulation for example. You would do some sampling, you would execute a number of pre-described simulations and you build responses for these simulations. You want to predict what happened between the values and get a quick answer.’ In this case, RSM reduced the amount of

simulation that is done as virtual curves can be produced, which are then verified against experimental data. Tis reduces the cost of expensive simulation while still providing a realistic model for crash simulation. Poloni concluded: ‘Response surface methods

is a way of sharing information between different application areas where simulations are particularly expensive. ‘You may build responses for each objective

that you have and then make decisions on the basis of the response surfaces instead of expensive simulation tasks and then go back and verify that your decision is what you would expect, that the actual response is what is predicted from the RSM.’ With the advent of data-centric computing

and the adoption of HPC by smaller companies this type of complex, computationally intensive simulation will become more prevalent as companies innovate in a competitive market.

Conference & Exhibition - April 15th/ 16th 2015 - Derby, UK HPC Processing

Hardware Applications Storage

Reserve your FREE tickets online: @engsimshow

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41