The recent advances in centrifugal pump design do not only require a better suction performance but also there have been attempts to reduce design time at a lower cost. The traditional trial-and-error optimization design method, however, depends on the designer's experience, which requires longer cycles. This is because the computational process of calculating the net positive suction head required (NPSHr) involves several calculation steps and this consumes a lot of computational time. An investigation was therefore carried out to test a novel NPSHr prediction method in a double-suction centrifugal pump using unsteady numerical simulations. In the new approach, a new boundary pair was introduced and an algorithm was used to estimate a good value for a static pressure value that correlates to a 3% drop in pump head to determine the critical cavitation point. Experiments were conducted to validate the hydraulic performance and the cavitation model. The NPSHr and the characteristic “sudden” head-drop were very well predicted by the novel approach in only three simulation steps. The internal flow analysis showed that for 0.6 Q d, the flow around the volute tongue was uneven at NPSH = 10.06 m, inducing flow separation and recirculation at the tongue region. Attached cavities were also observed around the suction ring in the spiral suction domain. The pressure fluctuations were analyzed also and the dominant frequency at the pump outlet and tongue region was the blade passing frequency. Consequently, the novel approach proved very robust and efficient in NPSHr prediction and would be a good alternative to shorten simulation time during cavitation optimization design process in centrifugal pumps.