Validation and uncertainty estimation of carbon fiber epoxy composite model

Sarah Scott, Andrew Kurzawski, Victor Brunini, John Hewson, Juan Hidalgo-Medina, Rory Hadden, Stephen Welch

Research output: Contribution to conferencePaperpeer-review


Carbon fiber epoxy composites are increasingly used in systems requiring a material that is both strong and light weight, as in airplanes, cars, and pressure vessels. In fire environments, carbon fiber epoxy composites are a fuel source subject to oxidation. This study addresses modeling the thermal response of a carbon fiber composite material through heating and pyrolysis. Using TGA data, a decomposition mechanism is proposed to describe the decomposition and smoldering. This is then combined with a finite element conduction-radiation model with a porous media model for gas advection. Mass loss results are compared to cone calorimeter experiments where the composite was exposed to heat fluxes of 30 kW/m2 and 80 kW/m2. Two backing materials are compared, aluminum (a heat sink) and ceramic (an insulator). Two thicknesses of the sample are examined, 29mm and 4.5mm. A material sensitivity study is conducted to understand uncertainty associated with input parameters, and how these sensitivities change with the heat flux, backing material, and thickness. It is shown that uncertainty increases with heat flux and is higher for the aluminum backing material. Contact resistance plays a large role in the uncertainty for the 4.5mm aluminum backed samples.
Original languageEnglish
Number of pages12
Publication statusPublished - 2019
Event11th U. S. National Combustion Meeting: Organized by the Western States Section of the Combustion Institute - Pasadena, Pasadena, United States
Duration: 24 Mar 201927 Mar 2019


Conference11th U. S. National Combustion Meeting
Country/TerritoryUnited States
Internet address


  • fire
  • carbon fibre epoxy composite
  • smoldering
  • pyrolysis


Dive into the research topics of 'Validation and uncertainty estimation of carbon fiber epoxy composite model'. Together they form a unique fingerprint.

Cite this