Quality Variability and Compositional Discrepancies in Long-Term Supply of Liquefied Stabilized Soil for Pile Removal Backfilling

• Field UCS variability (CV 0.33-0.59) exceeded plant controls (0.14-0.25) • XRF revealed 6.6% calcium in source soil vs 1.0% expected from contamination • 40% of samples failed specifications, bleeding rates most critical parameter • XRF composition estimation validated for rapid field quality assessment • Grade-based quality framework proposed with enhanced traceability protocols This study investigates the quality characteristics and compositional variability of commercially supplied liquefied stabilized soil (LSS) used for existing pile removal backfilling in the Tokyo metropolitan area. Quality inspections were conducted over approximately one year at three construction sites receiving materials from four permanent production plants. Results revealed significant discrepancies between construction site measurements and plant quality control data, with field-measured unconfined compressive strength (UCS) exhibiting coefficients of variation (CVs) ranging from 0.33 to 0.59 compared to plant CVs of 0.14 to 0.25. Out of a total of 124 LSS batches evaluated for UCS, density, and bleeding performance at the construction sites, 49 batches (39.5%) did not satisfy at least one of the Tokyo Metropolitan Government specification limits, with the most frequent cause of failure being bleeding rates exceeding 1%. Unlike previous studies that focused on short-term or single-plant evaluations, this study provides the first long-term, multi-plant dataset that directly compares plant-reported quality with field performance for pile removal backfill applications. The study also integrates portable X-ray fluorescence (XRF)-based composition estimation with laboratory validation to quantify the impact of cement-contaminated recycled source soils on UCS variability. Based on these findings, the study proposes a graded quality control framework tailored to pile foundation backfill, linking material quality classes to structural performance requirements and traceability protocols. XRF analysis revealed substantial deviations between design mix proportions and delivered compositions, primarily attributable to unexpectedly high calcium content (6.6% vs. assumed 1.0%) in construction-generated source soils. Laboratory validation confirmed that elevated calcium contents associated with recycled soils containing cement residues affect the accuracy of XRF-based composition estimation and suggest a potential contribution to compromised strength predictability. Based on these findings, a grade-based quality control framework is proposed, recommending Grade A (semi-structural) specifications with enhanced traceability and standardized testing protocols for pile foundation backfill applications.