BearingSolver: Physics-Based Bearing Analysis

Before programming, I identified the physical phenomena and outputs that BearingSolver had to model. I then conducted a structured literature review using SKF technical resources, British Gear Association training, the Harris bearing-analysis references, ISO 281:2007, and relevant scientific publications.

For each candidate model, I documented its required inputs, calculated outputs, assumptions, applicability limits, and available validation references. This established a traceable calculation chain covering bearing geometry, operating clearance, Hertzian contact behavior, load distribution, deformation, stiffness, life, lubrication indicators, and frictional losses. The resulting 65-page technical note became the implementation specification for the computational models.

I converted the selected physical models into modular Scilab functions, progressing from bearing geometry and operating clearance to Hertzian contact, radial and axial response, combined loading, life, and frictional losses. Each function used defined engineering inputs and returned traceable outputs that could be reused by the interface and tested independently during validation.

Radial equilibrium was solved iteratively by adjusting bearing displacement until the forces generated by the loaded rolling elements balanced the applied load. Axial response used the Jones stiffness formulation and Scilab's nonlinear solver to determine contact angle and deformation. Combined loading required a coupled three-equation equilibrium system for axial, radial, and angular behavior.

I reformulated the coordinate system to make the internal kinematics easier to interpret and used prescribed misalignment as a practical engineering input. When the advanced ISO fatigue-limit formulation did not converge reliably, I retained the documented simplified method instead of delivering an unstable solver. The excluded capability and its consequences remained visible in the model documentation and interface warnings.

I transformed the computational models into a Scilab GUI organized around the engineer’s complete analysis sequence: define the bearing geometry and materials, create multiple operating cases, execute the calculations, inspect the results, and explore the detailed contact behavior.

Every field, control, and result panel was programmed and connected to the underlying calculation functions. The interface presents load distribution, deformation, stiffness, contact pressure, and validity information through structured tables and plots. Contextual help, input checks, and explicit warnings—such as identifying truncated contact surfaces—help users distinguish a numerical result from one that is valid for engineering use.

I validated BearingSolver through a progressive chain of independent evidence. I first benchmarked the implemented calculations against SKF SimPro, then cross-checked their physical behavior and reference cases against NASA technical publications, Harris bearing-analysis examples, and the ISO 281 bearing-life standard. The final stage compared the model predictions with the available evidence from the company’s bearing test-bench process.

The test-bench results were not used to calibrate or correct the computational models. Instead, I recorded every validation test, its reference conditions, its results, and the observed differences in a 200-page validation note covering the complete campaign. This body of evidence allowed me to quantify the precision of each model and make its expected accuracy traceable to the engineering team. A separate user guide documented the complete application workflow.