A reproducible SSI-contact benchmark and amplification index for seismic pounding of adjacent buildings

Seismic pounding between adjacent buildings remains difficult to assess because the separation demand depends not only on peak displacement, but also on soil-structure interaction, phasing and contact-law assumptions. This study develops a reproducible benchmark for two adjacent reduced-order first-mode building models, meaning that each building is represented by the dominant lateral modal mass, stiffness and damping needed for screening-level comparison. Twenty-one PEER/NGA-West2 record sequence numbers, with two unscaled horizontal components per record, are analyzed separately and conservatively aggregated at record level. Five cases are compared: fixed-base linear contact; flexible-base soil-structure interaction without force transfer; flexible-base Kelvin-Voigt contact; flexible-base Hunt-Crossley contact; and parametric soil-structure-interaction/contact models. For the baseline 6 cm gap and medium foundation stiffness, the Hunt-Crossley soil-structure-interaction model increased mean peak impact force by 170.8% relative to the fixed-base baseline and increased mean impact count by 42.5%. It also increased mean peak force by 124.6% relative to the Kelvin-Voigt case while avoiding tensile rebound artifacts. The proposed amplification indices classify amplified, reduced, neutral and soil-structure-interaction-created impacts, including undefined ratios when the fixed-base denominator is zero. The results support separation screening and identify when fixed-base pounding models can be non-conservative.