In future, forging companies could design stage sequences significantly more quickly than before. Researchers of the Institut für Integrierte Produktion Hannover (IPH – Institute for Integrated Production Hanover)) are working together with the industry on software that automatically determines the optimal preform.
Evolutionary algorithms thus support the designers in an extremely time-consuming task that, at present, is often carried out according to the principle of trial and error. The transfer project “Preform optimisation” is being funded by the German Research Foundation (DFG), with the results to be tested in practice by Hammerwerk Fridingen GmbH.
Forged components such as connecting rods or crankshafts for engines are manufactured in several stages – from the raw steel component to one or several preforms to the finished part.
At present, it is extremely time-consuming and involves a heavy amount of work to design this so-called stage sequence. Here, the design engineers proceed according to the principle of trial and error:
They produce the preforms in accordance with the knowledge they have gained from experience and then review them by means of FEM simulation or with test forgings. If the forged components are defective – i.e. if, for example, creases form or the form is not correctly filled in – then the engineers must adapt the preforms and test them again.
A process that can last a very long time, especially with complicated components.
In future, software is to find the optimal preform within minutes and thus make the engineers work easier for them. Scientists from the Institut für Integrierte Produktion Hannover (IPH) are working on this together with Hammerwerk Fridingen GmbH. In future, the design engineers are to enter only the desired finished form and the form of the raw component into the software. The stage sequence is then determined automatically.
This is made possible by so-called evolutionary algorithms. They first produce various moulds – more or less at random – and assess them: How heavily does the mould deviate from the raw component? How close is the geometry to the finished forged part? How much force will be required to reshape it and how likely is it that creases will form in the process?
The moulds that perform best are then combined with each other or slightly modified and are assessed again. Thus, better geometries emerge in each generation – and after several stages the software issues one or several optimised preforms.
In the transfer project “Preform optimisation”, the algorithms are now to be readied for use. Hammerwerk Fridingen GmbH, with which the IPH is working closely in the project, will then test the research results in practice: The engineers will design moulding tools for the preforms issued by the software.
They thus forge test components and compare them with components from series production. It will then become evident whether the algorithms are practical – and whether forging companies can save a great deal of time in future in designing stage sequences.