FlameForge: Combustion of Generalized Wooden Structures

TL;DR

FlameForge tackles the challenge of simulating volumetric wood combustion across generalized structures by unifying air flow, heat transfer, pyrolysis, charring insulation, and combustion within a single simulator. It employs adaptive multiresolution OpenVDB grids and a signed distance field to efficiently represent complex geometries and insulation, while enabling two-way coupling with position-based dynamics. The approach is validated qualitatively on scenes with multiple materials and environments and quantitatively against sub-surface measurements from a real combustion experiment, with an example of coupling to a mechanical simulator. The work demonstrates practical potential for realistic fire-spread modeling in architectural and structural contexts and outlines future directions toward fracture modeling and interactive performance improvements.

Abstract

We propose a unified volumetric combustion simulator that supports general wooden structures capturing the multi-phase combustion of charring materials. Complex geometric structures can conveniently be represented in a voxel grid for the effective evaluation of volumetric effects. In addition, a signed distance field is introduced to efficiently query the surface information required to compute the insulating effect caused by the char layer. Non-charring materials such as acrylic glass or non-combustible materials such as stone can also be modeled in the simulator. Adaptive data structures are utilized to enable memory-efficient computations within our multiresolution approach. The simulator is qualitatively validated by showcasing the numerical simulation of a variety of scenes covering different kinds of structural configurations and materials. Two-way coupling of our combustion simulator and position-based dynamics is demonstrated capturing characteristic mechanical deformations caused by the combustion process. The volumetric combustion process of wooden structures is further quantitatively assessed by comparing our simulated results to sub-surface measurements of a real-world combustion experiment.

Figures (10)

• Figure 1: Overview of the presented FlameForge simulator. The system's state is initialized from the input geometry, its materials and the environmental conditions, and consists of a single vector and seven scalar fields describing air and flames as well as the material. In each update step, several sub-steps compute the next state of the system which is then forwarded to the rendering step to generate the visual output.
• Figure 2: A scene featuring multiple materials simulating fire spread and combustion processes of a burning house which is composed of two types of wood (walls and roof are made up of different wood types), acrylic windows (non-charring), and a stone chimney (non-combustible). The roof catches fire ($t_0$) such as after a lightning strike causing smoke development and fire spread ($t_1$). After some time, the relatively thin wood of the roof has already been burned while the thick walls are barely touched ($t_2$). In contrast, the chimney made out of stone still remains intact at later stages ($t_3$) and even after the fire will be over.
• Figure 3: Illustration of the grid projection in 2D. Left: Fine (orange) to coarse (blue). Black Lines: Material contour. Orange x: Voxels inside the material. Right: Neighborhood look up. Blue dashes: Material occupied voxels.
• Figure 4: Illustration of fire spread and combustion using complex geometry: A traditional, relatively flat-bottomed Chinese boat fully made out of wood has been ignited at the bottom (upper left). The fire then spreads quickly to the roof (upper right) and further progresses (lower left) before the whole boat burns brightly (lower right).
• Figure 5: Simulation of fire spread across multiple objects: A tree is on fire ($t_0$) which further spreads according to the wind direction igniting the roof ($t_1$) of a house (top row) causing severe damage ($t_2$). In contrast, if almost no wind is present, also in the long term ($t_3$), the house remains untouched (bottom row).