Application of the singular value and pivoted QR decompositions to reduce experimental efforts in compressor characterization

Compressor characterization, either by running experiments in a turbocharger test rig or by detailed CFD modelling, can be expensive and time-consuming. In this work, a novel method is proposed which can be used to build a complete compressor map from a reduced number of measured operating points combined with a previously collected database. The methodology is based on the application of the Singular Value Decomposition (SVD) method to acquire the orthonormal bases of a matrix which contains the information of previous compressor observations. These bases are used along with pivoted QR decomposition to obtain the minimum number of measurement points which are required to implement this technique as well as its optimal placement within the map. The reconstruction of two different compressor maps was made to validate the method. The results show a substantially better trade-off between number of testing points and accuracy compared to standard equidistributed sampling.