|SFPE Guidelines for Substantiating a Fire Model for a Given Application|
Issue 54: SFPE Guidelines for Substantiating a Fire Model for a Given Application
By Craig E. Hofmeister, P.E., FSFPE & Stephen M. Hill, P.E. Rolf Jensen and Associates, Inc.
The use of computer fire models in fire hazard and fire protection analyses has gained an increasing level of acceptance in recent years, with predictive capabilities spanning a wide variety of applications. Additionally, individual fire models have been continually developed and refined to provide more sophisticated tools with impressive visual graphics and output. In broad terms, fire modeling and fire simulation can include basic simple algebraic correlations, lumped-parameter models (zone models) and computational fluid dynamics models (field models), which are used to predict, or replicate, various fire phenomena within an established set of boundary conditions.
In order to be an effective tool for a fire related analysis, the model user and involved stakeholders must have confidence in the model results. There has been significant work in recent years to verify and validate individual fire models; however, there has been limited guidance for both the model user and the reviewer/authority having jurisdiction/consumer to assess whether the selected model was appropriate for the particular application. Therefore, the Society of Fire Protection Engineers (SFPE) developed the Guidelines for Substantiating a Fire Model for a Given Application1 (Engineering Guide), which was published in early 2011.
In the past year, the Engineering Guide and the process outlined therein has gained some acceptance in performance-based design applications. The Engineering Guide defines a process by which all stakeholders can establish a basic level of confidence in the modeling approach and results.
The Engineering Guide establishes a methodology with specific steps to review the suitability of a fire model for a specific application including:
The methodology is summarized in the following figure taken from the Engineering Guide.1 The figure outlines the process in flow chart form.
Define the Problem of Interest
The initial step in reviewing a fire modeling application is to fully understand the scope of the problem, the factors that may influence the calculation or model, and if considered necessary, review and/or search the literature for previous work in the subject area. The basic problem definition is important to establish the fundamental requirements of the application regardless of the user's familiarity with the analysis.
As an example, a "typical" atrium smoke control application can include a variety of parameters and physics such as smoke movement, smoke temperature, smoke concentration, smoke layer location and depth, compartment pressure, compartment ventilation, visibility, and flame height. The analysis may also include an evaluation of heat release rate, heat flux, ceiling jet temperature and velocity and sprinkler/detector response as part of an initial design fire study. Several other parameters need to be considered based upon the available information, including the geometry of the space or domain, the timeline of the analysis, time-based events, impact of materials, initial conditions and boundary conditions.
The final component in a comprehensive evaluation of the problem of interest is to clearly define the objectives and output of the fire model analysis.
Select a Candidate Model
The baseline intent for substantiating a fire model for an application is to check whether a model with the required capabilities (governing equations and assumptions) and level of accuracy are appropriate for the problem of interest. Many models and model types are available, often with overlapping capabilities. In order to adequately review a candidate model, the user must establish the available model input data and the desired outputs.
The general options for choosing a fire model include algebraic equations/correlations, zone models (lumped parameter), or field models (computational fluid dynamics). Model selection often can be a balance between obtaining the necessary level of output resolution and the level of uncertainty associated with available input and the model's use. A more complicated solution is not always a better solution unless the problem definition has confirmed that greater detail is needed.
Verification and Validation
Regardless of the candidate model selected by the user, the Engineering Guide emphasizes the importance for the user to consider and assess the predictive capability of the model to be sure it is appropriate for its intended use.
The process of verification is intended to check whether the mathematics of the model are correct and that the physics will be correctly described by appropriate equations. Because a model is often comprised of equations that are packaged into a program developed by others, the practicality of the user's role in verification is generally limited. It is incumbent on the user during the verification process to have a thorough understanding of the underlying assumptions and limitations of the calculations performed by the model.
In the event that existing V&V studies are not available for the application of interest, it may be necessary to either review additional experimental data or commission specific fire experiments in order to collect measurements against which model predictions may be compared. The blind application of a model must be avoided at the cost of reporting erroneous results. Ultimately, the results obtained from the most basic of calculations to more complex fire models are only as good as their accuracy of the physics that are being described.
Additional levels of uncertainty are often integrated into a model by the user in the form of assumptions, input parameters, and the spatial definition of the modeled space. User effects often result in additional error outside the predictive capability of the calculations. As such, a modeler should have a good understanding of the sensitivity of user input parameters and the impact they may have on the model output.
The user effects can include chosen parameters, including the spatial domain, design fire and location, suppression effects, or output parameters. The impact of user effects often increases with the complexity of the chosen model.
While the accuracy of the model calculations has a direct impact on the validity of the model (V&V), the detail and accuracy of the model inputs and assumptions can have just as large an impact on the model results.
During the initial stages of any fire modeling
application, it is considered good practice to document the steps
outlined in the Engineering Guide. Such documentation provides one with a
template that can be used as thorough justification for qualifying a
model as appropriate for the desired applications. Depending on the
project and the requirements of the client, stakeholder or AHJ, it may
be necessary to provide varying levels of detail or types of
The guide can be purchased here.
Craig Hofmeister & Stephen Hill are with Rolf Jensen and Associates, Inc.
For questions concerning delivery of this e-Newsletter, please contact our Customer Service Department at (216) 931-9934 or magazine.sfpe.org.