From August 2004 to February 2010, Mr Lemouedda carried out research into the numerical investigation and optimization of gas-liquid heat-exchangers. In his research, the term “optimization” is to be understood to mean “the targeted, systematic search for the best possible solution”.

The dissertation was developed in close cooperation between the Faculty of Process Engineering and the Chair of Fluid Mechanics at the University of Erlangen-Nuremberg Dres. (Prof. Breuer, Prof. Delgado, Prof. Durst). This cooperation – and thus Mr Lemouedda’s research – was funded by the Bavarian State Ministry of Science, Research and the Arts as part of the first two programmes of the Competence Network for Scientific High-Performance Computing (KONWIHR I and II). The complete financing was secured through the STAEDTLER Foundation, the BayME programme Technologiebrücke [technology bridge], and funds from Nuremberg Tech.

Mr Lemouedda limited himself to dealing with gas-liquid heat exchangers, as these are very often used in the relevant industry. The method he developed can be applied to heat exchangers using other fluid combinations without any basic difficulties.

The optimization of heat exchangers is particularly difficult because a large number of objective functions (criteria) and parameters have to be taken into account. This is what is known as multi-criteria optimization, where the individual criteria can contradict each other. Examples include the combinations heat transfer coefficient and pressure loss, energy input and gain, or production and operating costs. The parameters are often discretely defined variables such as different, standardized pipe dimensions, different materials, different geometries of heat transfer-enhancing elements, and so on. Continuously defined parameters are rather rare in this application. Conventional optimization methods using Lagrange multipliers are therefore not suitable.

Methods that artificially generate a single-criterion optimization by means of a weighted linear combination of criteria are also unsuitable, because a certain choice of weighting factors can lead to sensible results for one problem definition, but to unreasonable results for quite similar problems. Mr Lemouedda recognized that the combination of Pareto optimization with genetic algorithms is ideally suited to the optimization of heat exchangers. He essentially restricted his criteria to the energy expenditure related to the construction volume and the transferred heat, which is also related to the construction volume. His method can be applied to different or additional criteria without any fundamental difficulties.

The optimization criteria can be determined in various ways in the optimization method selected by Mr Lemouedda. For example, experimental determination is just as possible as the use of analytical relationships, if available, for example from the VDI heat atlas. Mr Lemouedda used computational fluid dynamics (CFD) to calculate the objective functions mentioned. The genetic algorithms he used proved to be particularly helpful in this respect, because with the large parameter spaces available they can drastically reduce the number of criteria calculations to be performed via CFD.

Besides the scientifically based derivation of the described method and the exemplary investigation of different heat transfer-enhancing elements, the aim of the thesis was to develop a process that automatically combines geometry generation via parameterized CAD, CFD, Pareto optimization, and genetic algorithms.