International Journal of Scientific Research and Engineering Development

Continuously welded rail (CWR) has displaced the fishplated joint as the standard permanent way construction throughout the world because the mechanical discontinuity at a bolted joint is the origin of track deterioration, rolling noise, vehicle dynamic loading, and rail-end fatigue. This study establishes a data-driven diagnostic framework to evaluate the sufficiency of current scalar process acceptance parameters against a documented single-case joint failure that passed traditional non-destructive inspections but suffered a flat brittle fracture during press straightening. By constructing a systematic process signature database of 79 weld cycles across R260 and R350HT rail grades on two nominally identical Schlatter installations, this work evaluates thirteen scalar descriptors— including newly derived velocity, energy, and volumetric index features. Statistical evaluations via Welch’s unequal-variance t-test reveal that despite identical processing standards, the two machines exhibit significant divergent baselines (p < 0.01) across 11 key operational parameters, proving that raw numerical acceptance thresholds are non-transferable. Most critically, an out-of-sample sufficiency evaluation demonstrates that all thirteen summary scalar parameters of the failed joint resided safely within their corresponding ±3σ control limits, meaning traditional scalar-value statistical process control is fundamentally blind to intra-cycle variations. Continuous trace analysis using a Schlatter WeldAnalyzer confirms the fracture originated from a two-second platen velocity discontinuity during final flashing, driving flashing arc extinction and a localized 8% secondary current spike that left hidden unexpelled oxides at the bond line. To address this blind spot, a three-layer transferable solution framework is proposed: upstream calibration gating rules, intra-cycle trace monitoring rules operating on temporal velocity/current envelopes, and machinestratified scalar control charts.

Shobhit Sharad Gottlieb, "In-Process Quality Prediction in Rail Flash Butt Welding: Testing the Sufficiency of Scalar Process Signatures" International Journal of Scientific Research and Engineering Development, V9(4): Page(1466-1477) July-August 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.