You are here: Applications > Projects > Code Assembling
MIPTO (Management of an Integrated Plateform for auTomatic
Optimization): Shape optimization of aeronautical combustion
chambers
Strict regulations on pollutant emissions and the need of reducing the times between design and marketing push aeronautical engine manufacturers to reconsider the concepts of the next generation of combustion chambers as well as their design methodologies. In the design cycle of aeronautical combustion chambers, engineers have used reactive and turbulent simulation codes based on the RANS approach for some years in the recent past. Their use has allowed reducing development times and costs mostly by decreasing the number of experimental tests. The way to integrate these tools in an efficient design framework is still an open issue.
The aim of this work done during the PhD of Florent Duchaine (http://www.cerfacs.fr/~duchaine) is to provide a multiobjective optimization based methodology to develop a fully automated tool using numerical simulation codes for the evaluation of different design alternatives. The studies deal with the automation of the simulation processes and focus particularly on the automatic mesh generation aspects. In order to reduce the overall response time due to the use of optimization techniques with expensive simulation codes, a strategy based on metamodelling is proposed. The resulting tool is developed with the PALM coupler granting performance and flexibility to the application. After some validations and evaluations on case studies, a full size application on an industrial combustor has proved that the tool is able to identify the most promising design solutions.
SNECMA-PALM. Automatic thermal loop based on N3S-Natur / ASTRE /
Abaqus for the prediction of combustion chambers wall
temperature.
Prediction of walls temperature is a problem of primary importance for the estimation of the life cycle of aeronautical combustion chambers. For this reason, SNECMA uses numerical simulations to compute thermal equilibrium that involve convective, radiative and conductive fluxes. The 3D solvers generally used for these computations are N3S-Natur for the turbulent reactive two phase flow predictions, ASTRE for the radiative fluxes and Abaqus for thermal conductivity in solid structures. The convergence of these multi-physics studies implies an iterative process to reach a stationary state. The classical way to carry out such computations is to sequentially run the different codes and manually interfacing them by exchanging data files after having performed the needed format, numbering and unit conversions. Beyond the tiresome aspect of the task, such a procedure is time consuming in term of human and computer power. Moreover, such a manual process can lead to an important uncertainty due to human manipulations of computational data.
In this context, the use of the PALM coupler aims at integrating the 3D solvers and the different interfaces between the codes in an automated application. In addition to the obvious overall time reduction of the resulting process, such an application can help the designers to easily discern the important physical contributions and to optimize the tool: it will be relatively easy to implement different computing strategies, to tune the most significant parameters and to determine the convergence conditions.