PhD Student Case Western Reserve University
Modeling Wildfire Dynamics and Resulting Thermal Loads on Water Infrastructure Using Continuous Percolation Theory
Co-Author: Xiong (Bill) Yu (Case Western Reserve University)
Abstract: Wildfire occurrence continues to rise globally due to environmental change, which intensifies hot and dry seasons and leads to more frequent and severe wildfire events. As these hazards escalate, it is increasingly important to understand not only how wildfires spread but also their cascading impacts on critical systems, especially water infrastructure that supports public health, emergency response, and ecosystem resilience. Predictive tools are therefore needed to model wildfire behavior and evaluate how resulting heat loads affect infrastructure performance.
Traditional wildfire models often assume flat terrain and uniform spread, overlooking the nonlinear influences of topography, vegetation heterogeneity, moisture variation, and wind dynamics. These factors strongly govern wildfire direction, spread rate, and intensity, limiting the accuracy of conventional models in complex landscapes.
This research introduces a continuous percolation-based framework that simulates wildfire propagation across heterogeneous terrain. The landscape is represented as a two-dimensional grid of fuel and non-fuel cells, with percolation probabilities dynamically adjusted using elevation, vegetation density, moisture, and wind fields. This allows realistic modeling of slope-driven acceleration, wind-induced anisotropy, preferential spread patterns, and variable fuel structure.
The model computes the evolving fire front, spread rate, and heat generation. These outputs are then coupled with a physics-based surrogate model that represents buried water infrastructure. By linking heat intensity to infrastructure response, the system identifies hazardous thermal thresholds, estimates damage likelihood and spatial distribution, and determines response windows before potential failure.
This integrated approach enables early warning, real-time vulnerability assessment, and long-term resilience planning as wildfire activity intensifies worldwide.
Associate Professor Oregon State University
Dynamic modeling of water systems during wildfires
Co-Authors: Amy Metz and Hannah Farris
Abstract: Recent fires have shown the capacity for urban water distribution systems to depressurize and fail due to high demands from burning homes as well as firefighting efforts. This presentation presents a methodology to simulate water systems during a fire such that these demands can be simulated temporally and regions of vulnerability may be assessed throughout the community. The results can demonstrate when hydrants will no longer support firefighting demands as well as specific regions where mitigation may be most impactful both in the built environment (e.g., homes) as well as in the water distribution system. Often fire modeling does not include the capacity for defensive action, which can greatly under estimate risk as well as potential impact to communities.
Postdoctoral Scholar University of California Berkeley
Simulations of Flame Spread at the Wildland-Urban Interface in a Landscape-Scale Fire Risk Model
Co-Authors: Dwi M J Purnomo, Adam Laird, Maria Theodori, Maryam Zamanialaei, Chris Lautenberger, Arnaud Trouvé, and Michael Gollner
Abstract: This work extends the landscape-scale wildland fire spread simulator ELMFIRE to model conflagrations propagating through wildland–urban interface (WUI) communities. ELMFIRE integrates vegetation, structures and infrastructure, terrain, and weather to simulate fire behavior at 30 m resolution over domains ranging from tens of meters to tens of kilometers. Fire growth is represented using a level-set formulation that captures surface and crown fire dynamics. Recent developments add structure ignition and structure-to-structure fire propagation in the WUI. Structure-to-structure spread is modeled through three primary mechanisms: thermal radiation, direct flame contact, and firebrands. Radiation and flame contact are represented using simplified semi-empirical formulations adapted from the fire safety engineering literature. Firebrand transport is modeled stochastically using a prescribed probability density function for downwind travel distances of flaming/smoldering embers coupled with a simple flight-time model; spot-fire ignition is treated with a semi-empirical criterion based on critical firebrand accumulation. We present simulations of the WUI sub-models in idealized test configurations and landscape-scale simulations of several historical fire events, including comparisons with available post-fire damage data. The results show encouraging agreement in predicted damage patterns and support the use of ELMFIRE for landscape-scale WUI fire risk assessment.
PhD Student UC Berkeley
Coupling Cell2Fire with Downscaled High Resolution Wind Data
Co-Author: Marta Gonzalez (UC Berkeley)
Abstract: We couple Cell2Fire (state-of-the-art cellular automata fire spread simulator) with WindNinja. Specifically, we generate spatially varying wind fields, downscaled to 30-m resolution by leveraging high resolution weather model analysis data. To evaluate, we simulate the 2025 Palisades Fire and find that our coupled model can accurately capture the early stages of the fire's propagation. In addition, we use the fire spread simulation to model fire potential polygons and networks, which can be used to aid decision-support during wildfire management and operations.
Associate Professor University at Buffalo
Influence of Built Environment Characteristics on Fire Spread Dynamics and Community-Scale Damage
Co-Author: Nisha Saharan (University at Buffalo)
Abstract: Extreme wildfires increasingly impact wildland-urban interface (WUI) communities, underscoring the need to understand how built environment characteristics influence fire spread and losses. This study examines the role of building-level and community-scale attributes on fire spread dynamics and subsequent damage using physics-informed simulations. Results indicate that clusters of ignition-resistant buildings can alter both the speed and spatial pattern of fire progression within communities. Ignition-resistant structures are classified based on key construction attributes, distinguishing vulnerabilities to radiant heat exposure (affecting siding and windows) and fire spotting mechanisms (affecting roofing, vents, decks, fencing, etc.). The simulation framework and building classifications are evaluated using two recent wildfire events, the 2021 Marshall Fire and the 2023 Lahaina Fire, where clusters of ignition-resistant buildings were observed. In both cases, these clusters affected local fire dynamics, and structures within the clusters predominantly survived the fire. The approach is then extended through scenario-based analyses for the area impacted by the 2025 Eaton Fire in Los Angeles County, which experienced widespread urban destruction. Hypothetical configurations of ignition-resistant structures near the wildland boundary are examined to assess their influence on potential damage at the community scale. The findings highlight the importance of integrating building design, spatial configuration, and fire spread simulations to inform strategies that enhance wildfire resilience under extreme conditions.
PhD Student Utah State University
A multi-hazard computational approach for damage assessment of steel buildings under fire-following-earthquake scenarios
Co-Authors: Negar Elhami-Khorasani and Mohsen Zaker Esteghamati
Abstract: Fire-following-earthquake (FFE) is a cascading hazard where post-earthquake fires exploit compromised fire protection systems, amplifying damage and extending downtime. The 2024 Noto Peninsula earthquake (M7.5) triggered 17 fires, three of which spread significantly and burned 49,000 m2 in Wajima City. Despite these consequences, existing probabilistic FFE frameworks have critical limitations, such as simplifying damage to passive fire protection (e.g., spray-applied fire-resistive material (SFRM) delamination) and active fire suppression systems. This study develops a probabilistic approach for multi-hazard damage assessment of steel buildings under FFE scenarios through state-dependent vulnerability functions that connect structural and non-structural damage under seismic loading to subsequent fire spread and intensity. The approach integrates seismic damage assessment with fire scenarios conditioned on post-earthquake building states using Pelicun and OZone software. This approach explicitly models fire protection system degradation, including SFRM delamination, compartmentation breach, and sprinkler system impairment, as functions of peak floor acceleration and inter-story drift ratio. Component-level damage modeling is expanded to incorporate a broader range of non-structural components beyond partitions, glazing, and doors. The methodology is demonstrated on a database of steel moment-resisting frames (1-19 stories) subjected to ground motion records and subsequent parametric fire scenarios. Results quantify how earthquake-induced degradation of fire protection systems amplifies structural and non-structural fire damage, enabling systematic assessment of cascading FFE risk. The implementation will be released via DesignSafe-CI to enable broader adoption.