No abstract

No abstract

This paper considers the effect of an upstream compressor stage on a compressor inter-spool duct. The duct geometry must be fixed early in the engine design process, well before the design of the upstream stages. It is therefore important that the designer has a good physical insight into how engine representative inlet conditions affect the limits of the duct design space. An experimental and computational investigation of two strutted inter-spool S-ducts was undertaken. Both were tested with and without an upstream stage present. The first duct is of a conventional axisymmetric design with a radius change to length ratio ΔR/L = 0.50. This duct is characteristic of the most extreme ducts considered in modern engine design. The second duct is of a non-axisymmetric design and is 20% shorter, ΔR/L = 0.625. This is well beyond the design limit of axi-symmetric strutted ducts. The paper shows that the presence of the upstream stage increases the duct loss by 54%. The rise in loss occurs on the hub wall and is the result of the incoming stator wakes pooling onto the hub wall, forming a row of contra-rotating streamwise vortex pairs adjacent to the hub wall. These vortices pump boundary layer fluid into the free stream, thus raising the mixing loss. In the non-axisymmetric duct an extra mechanism was observed. The streamwise vortex pairs act to ‘re-energise’ the boundary layer. This reduces strut secondary losses caused by the endwall contouring. The net result is that on the non-axisymmetric duct the presence of an upstream stage only increases the duct loss by 28%. Comparing the two ducts, it is shown that with engine representative inlet conditions, the conventional symmetric duct and 20% shorter non-axisymmetric duct have identical performance. This shows that low loss ducts can be designed which are significantly more extreme than current design limits.

SUMMARYDespite its robustness, the design and optimization of aerodynamic shapes using genetic algorithms su ers from high computing cost requirements, due to excessive calls to Computational Fluid Dynamics tools for the evaluation of candidate solutions. To alleviate this problem, either the use of distributed genetic algorithms or the implementation of surrogate evaluation models have separately been proposed in the past. A distributed genetic algorithm relies on the handling of population subsets that evolve in a semi-isolated manner by regularly exchanging their best individuals. It is known that distributed schemes generally outperform single-population ones. On the other hand, the implementation of less costly surrogate evaluation tools, such as the autocatalytic radial basis function networks developed by the authors for the purpose of getting rid of most of the 'useless' exact evaluations, reduces considerably the computational cost. The aim of the present paper is to employ a surrogate evaluation model in the context of a distributed genetic algorithm and to demonstrate that the combination of both results in maximum economy in CPU cost. In addition, whenever a multiprocessor system is available, the gain is much more pronounced, since the new optimization method maximizes parallel e ciency. The proposed method is used to solve inverse design and optimization problems in aeronautics and turbomachinery.

SUMMARYIn aerodynamic shape optimization, the availability of multiple evaluation models of different precision and hence computational cost can be efficiently exploited in a hierarchical evolutionary algorithm. Thus, in this work the demes of a distributed evolutionary algorithm are ordered in levels, with each level employing a different flow analysis method, giving rise to a hierarchical distributed scheme. The arduous task of exploring the design space is undertaken by demes consisting the lower hierarchy level, which use a lowcost flow analysis tool, namely a viscous-inviscid flow interaction method. Promising solutions are directed towards the higher level, where these are further evolved based on a high precision/cost evaluation tool, viz. a Navier-Stokes equations solver. The final, optimal solution is obtained from the highest hierarchy level. At each level, metamodels, trained on-line on the outcome of evaluations with the level's analysis tool, are used. The role of metamodels is to allow a parsimonious use of computational resources by filtering the poorly performing individuals in each deme. The entire algorithm has been implemented so as to take advantage of a parallel computing system. The efficiency and effectiveness of the proposed hierarchical distributed evolutionary algorithm have been assessed in the design of a transonic isolated airfoil and a compressor cascade. Remarkable superiority over the conventional evolutionary algorithms has been monitored.