A frictionless force transmission between the mechanical components of high production volume systems was accomplished by Rolling Element Bearings, which leads to being an imperative item of the condition-based maintenance (CBM). A prominent fault would cause malfunction, and its substantial growth may lead to catastrophic failure of such machinery ultimately give rise to unscheduled maintenance, and in extreme cases turns into a quite expensive shut down of the same. These decisive situations explore a growing demand for a robust failure diagnosis scheme for bearings. The present work demonstrates a novel approach to develop a dynamic model for vibration response of spherical roller bearings using dimensional analysis with Buckingham's pi theorem (BPT) by considering significant geometric, operating, and thermal parameters of the system. The results obtained have been validated with experiments performed on the developed test rig under diverse operating conditions. Vibration amplitudes considerably enhanced in the presence of rotating speed, applied load, bearing temperature, and volume of defect. Multivariable regression analysis is performed to reveal the effectiveness of the model for precise detection of impending bearing failure. The vibration response obtained by the DA model, experimental runs, and MVRA agreeing with physical perceptive and potentials. Results indicate the simplicity and reliability of the approach. © 2020 Elsevier Ltd