A FRAMEWORK BASED ON MULTI-HAZARD FRAGILITY MODELS FOR RISK ASSESSMENT OF BURIED PIPELINES SUBJECTED TO EARTHQUAKES AND EXPANSIVE SOIL CONDITIONS

Abstract

This dissertation develops and demonstrates a two-phase, multi-hazard framework to assess the risk and resilience of buried water pipeline networks exposed to the compound effects of earthquakes and expansive soils. Motivated by the underestimation of pipeline damage when hazards are modeled in isolation, this work integrates engineering mechanics with spatial analytics to produce decision-making insights for utilities. Phase 1 develops finite-element (FE)–based fragility surfaces that jointly quantify damage as a function of seismic shaking and moisture-induced soil swelling; Phase 2 operationalizes these fragilities within a GIS platform to generate city-scale risk maps, compute system-level resilience metrics, and test mitigation strategies. The methodology begins with inventory compilation (pipe diameter, material, burial depth; soil plasticity and active-zone depth) from public datasets and local records. Seismic demands are represented by a target response spectrum consistent with regional hazard and implemented through spectrum-matched synthetic accelerograms. Swelling demands are characterized by using soil swelling index and active-zone depth to estimate total heave. A 3-D FE model (soil as Mohr–Coulomb continuum, pipe as shell with frictional soil-pipe contact) imposes swelling as an upward displacement boundary and then applies the earthquake time history; the engineering demand parameter is the peak axial tensile strain, which governs fracture in buried pipes. Incremental dynamic analysis across a grid of peak ground acceleration and swelling levels yields response datasets that are translated, via statistical generalization, into lognormal, two-variable fragility surfaces for discrete damage states (minor and major). Phase 2 ingests these fragilities into a GIS workflow that overlays pipeline attributes with spatial fields of seismic and swelling intensity to assign segment-level damage probabilities, constructs network-wide risk maps, and estimates a Serviceability Index (SI) after an event. Break-rate models are tied to segment length distributions to approximate system degradation without full hydraulic simulation, enabling rapid scenario analysis.The framework is applied first to Lawton, Oklahoma, a city near the Meers Fault with extensive swelling soils and a typical active-zone depth of ~2 m. Sixteen fragility scenarios spanning pipe materials (cast iron, PVC), diameters (6, 8, 10, and 12 in.), and soils (CL, CH) reveal strong multi-hazard interaction: swelling alone can drive major damage in highly expansive zones, while moderate swelling combined with moderate shaking produces a nonlinear escalation in failure probability. Cast-iron (CI) pipes exhibit brittle behavior and high fragility even at modest demands; PVC’s ductility accommodates larger strains and shifts probability mass from major to minor damage over a broad range of hazard intensities. GIS risk maps for Lawton show that segments under CH soils dominate expected major damage, and a parametric study of soil remediation (reducing effective moisture-change depth from 2.0 m toward 1.0 m in 10 cm steps) indicates substantial SI gains once remediation exceeds ~60 cm; at 100 cm, swelling-induced deformation is effectively removed and remaining risk is seismic. To test transferability, Lawton-based fragilities are applied to Norman, Oklahoma, similar in swelling soils but more heterogeneous (CI, ductile iron, PVC; clays and sands). With conservative assumptions where needed (treating DI as CI; sandy soils as CL for swelling), three scenarios are compared: actual material mix, all-PVC, and all-CI. Results confirm that accurate inventory data is pivotal: assuming all CI materially overestimates risk and mitigation needs, while assuming all PVC can be unconservative. Nonetheless, the overall resilience patterns and the benefits of remediation remain consistent, supporting the cautious reuse of regional fragilities across cities that share active-zone depth and swelling characteristics. Collectively, the dissertation contributes: (i) a new class of FE-based multi-hazard fragility models that capture seismic–swelling interactions; (ii) a scalable GIS-based model that converts engineering fragility into actionable, city-scale risk and resilience intelligence; (iii) empirical evidence from two Oklahoma cities that PVC materially improves resilience relative to CI, and that remediation depths >60 cm meaningfully increase SI; and (iv) a demonstration of model transferability under justified assumptions, enabling efficient expansion of resilience planning beyond a single municipality. Limitations include material and soil generalizations (e.g., DI treated as CI, sands treated as CL), reliance on representative FE geometries rather than exhaustive field-specific models, and dependence on public data resolution; validation against break records is a priority for future work. Extensions are outlined to broaden material coverage (e.g., DI, HDPE), incorporate time-dependent deterioration and ground-failure hazards (liquefaction, landslides, fault rupture), integrate economic and equity metrics, and apply the framework to other buried lifelines. By unifying mechanics-based fragility with geospatial decision tools, the research enables risk-informed replacement, targeted soil remediation, and material transition strategies that enhance the resilience of water distribution systems under compound natural hazards.