In industry, the use of large centrifugal compressors is widespread inmany vital applications such as petrochemical, natural gas distribution,and fertilizers. The increasing rate for using suc_h machines implies toconsider them as the backbone of the production process. In theseprocesses, profitability can rest directly upon the efficiency and operabilityof the plant’s compressor. The continuous monitoring and control of thesemachines are important factors influencing the efficiency and reliability of them. In the present work, the performance of a centrifugal compressor isanalyzed in order to identify the operation instability during surge androtating stall. A mathematical model is proposed in order to predict surgeas a transient phenomenon. This model requires the identification of thecompressor performance during steady state and reverse flow. So, anothermathematical model is also proposed to predict the performance duringsteady state operation and reversed flow. The proposed model is a modifiedversion of the model presented by Galvas [NASA TN D-7487, 1973]. Themodified model has a higher coefficient of correlation than Galvas’ model.Moreover, this model predicts the performance of multi-stage centrifugalcompressors, multi-compressors arranged in series and the reverse flow.This model was highly recommended to be proposed because themeasurements of the reverse flow performance is very difficult specially inindustrial compressors. Consequently, the integration between theproposed models results in an essential tool for studying the dynamic andsteady state performance of industrial multi-stage centrifugalcompressors.In order to validate the proposed models, the results are comparedof the cross sectional area of wall piers along the height and resting on rigid foundations.A test program involving sixteen stepped coupledshear wall models js formulated to examine the effect of (i) irregularity, as denoted by the ratio of the cross sectional area of wall piers in the upper segment to that in the lower segment, and (ii) coupling, as denoted by the rigidity of the connecting beams. The significant response parameters of interest are (i) shear forces in connecting beams, ii) distribution of normal and shear <stress components along the height particularly at base and level of variation, and (iii) lateral deflection. The experimental setup is described and the photoelastic raw test data are obtained. The photoelastic braw test data include complete patterns of isochromatic and isoclinic fringes. The shear difference method is applied to the photoe las tic raw test data to obta in the.