This roadmap outlines research plans for advancing the capabilities of the fire model CFAST. Continuing support for CFAST is via internal funding at the National Institute of Standards and Technology (NIST). In addition, support is provided by other agencies of the U.S. Federal Government, most notably the Nuclear Regulatory Commission (NRC) Office of Research and the U.S. Department of Energy.
How We Got Here
The CFAST model has its roots in at least two earlier zone fire models. The "Consolidated Compartment Fire Model," developed by Cooper and Forney and originally intended to be a collection of reference algorithms for fire modeling, forms structure of CFAST and provides some of the most basic algorithms in the model, including convective and radiative heat transfer and natural vent flows. The most direct predecessor to CFAST, the "Fire and Smoke Transport" or FAST model developed by Jones defined the necessary inputs to and outputs from the model, and other algorithms that make up the model, most importantly fire chemistry and plume flow. Originally released in June 1990, the model has undergone at least five major revisions and numerous minor revisions over the last two decades to become one of the most consistently and widely used fire models worldwide.
With its long history, CFAST is a relatively stable and robust model. Still, with continuing research in fire and advancing computer capabilities, its role has become both a tool for direct analysis and a screening tool through numerous CFAST simulations to identify specific scenarios that require a more detailed analysis with more complex CFD models such as FDS. With this in mind, several areas of research need to be addressed in order to improve CFAST. First, the underlying algorithms need to be improved to ensure consistent calculations for hundreds of variants of a chosen scenario. Second, efforts in improving the validation of the model are necessary to better understand and quantify the accuracy of the predictions of the model. Third, the user interface needs to evolve along with the fire model to ease the use of the model for use as a screening tool and as a standalone analysis tool. Finally, visualization of the output of the model needs to improve to allow quick screening analyses of multiple CFAST simulations.
Improvements in the Fire Model
In FY 2014, we streamlined the underlying source code for CFAST and updated it to be consistent with the Fortran 2003 syntax. In addition, the combustion chemistry was totally rewritten with an eye towards more accurate calculations and simpler user inputs. This has provided a solid basis for further improvements to the model. Experience over the years has shown the model to be applicable to a broad range of scenarios, but there are a small number that cause the model to fail with often only small changes in the input required for successful calculation. For use as a screening tool, the model needs to be able to perform dozens or even hundreds of variations on a single scenario so that even a small fraction of failures becomes important.
Areas for improvement include:
- Investigate improvements in the model numerics including
- replacing the DAE solver with newer algorithms ,
- changing the equation set to allow control volumes to have zero volume and mass ,
- modification or removal of reduced Jacobian algorithms (which were added to speed up calculations when computer resources were orders of magnitude slower than those available today) to simplify the underlying algorithms, and
- possibly adding additional equations (that are now done as side calculations) to the solution set to ensure more robust calculations.
- replacing the DAE solver with newer algorithms ,
- Improve physical algorithms including
- improve the heat and mass transfer between layers in a compartment which have been shown to be important particularly in post-flashover fires ,
- include multi-layer conduction for surfaces and targets (still in the code, but unused due to limitations in the user interface), and
- add more comprehensive calculation diagnostics to facilitate and speed analysis of problem cases.
- update the HVAC algorithm to eliminate outdated code and provide a better dynamic solution rather than the current quasi-steady-state algorithm,
Improvements in Model V&V
With support from the U.S. Nuclear Regulatory Commission , we have considerably more validation data for CFAST than was available in previous versions, with an automated process that allows us to check all the V&V comparisons with each new build of the model. At least three areas of research are appropriate going forward:
- Add additional data sets to the validation suite to extend the range of existing validations and to extend the validation to quantities that are not currently included. For the former, we have continued to work with NRC and coordinate the effort with FDS validation efforts to be as efficient as possible.
- Study additional techniques for comparing model predictions with experimental data. At present, comparisons are based on peak values. More information comparing entire curves of data and curve shape would provide a more complete picture of model accuracy.
- Extend CFAST V&V by extending the suite of verification tests for the model to verify correct operation of individual and overall calculations in model.
Improvements in the User Interface
Some part of the popularity of CFAST is a result of a focus on its user interface as a typical Windows program. Typically through user feedback, this part of the software is updated when appropriate to ease the use of the model, maintain compatibility with evolving inputs for the CFAST model, and to improve analysis and visualization of the model outputs. Again, recognizing the changing purpose of the model, several areas for additional development are necessary:
- Add additional visualization capability for CFAST data with Smokeview. This includes adding visualization for all vents types, smoke and heat alarms, and adding graphical display of simulation results. Some of this comes from ongoing collaborations with researchers worldwide.
- Include ability in the user input to create, run, and analyze multiple variations on a user-selected base scenario to facilitate the use of CFAST as a screening tool.
- Combine the fire model simulations with CFAST with those from a simple egress model. Again in keeping with the focus of providing a more simple calculation tool for use as a screening tool for more complex calculations, the recent NIST Egress Estimator software (based on calculations from the SFPE handbook)  could be included as part of the CFAST GUI to form CFAST+Egress for simple estimates of stairwell and elevator egress from buildings. By combining the two models, the impact of fire conditions on egress could be included as well.
User support is an ongoing effort for any software product, including CFAST. CFAST is developed and maintained using fairly common software development tools. The source code and manuals are stored in an external repository and can be accessed by all members of the team. User support is provided via a Discussion Forum hosted by GoogleGroups and an Issue Tracker hosted by GitHub. In short, a typical CFAST user merely goes to a website and enters a question or bug report, and we all can answer the question or track the bug until it is fixed. This has dramatically improved our ability to help users, as before it was simply one or two people answering emails.
Over the years, CFAST has benefited from collaborations with numerous individuals and organizations. We encourage collaboration, even though direct funding from NIST and other sources is rarely available. Over the past decades, perhaps the most important lesson we've learned from our development is that proposed additions need to made with an understanding of the relative level of detail in the rest of the model. Those interested in working with CFAST should contact us to discuss their interests to ensure the direction of the effort matches current efforts from NIST and others and to ensure that any proposed effort is in keeping with the general complexity / simplicity of other parts of the model.
 see http://www.netlib.org/ode/
 Forney, G. P. and Moss, W. F. Analyzing and Exploiting Numerical Characteristics of Zone Fire Models. Natl. Inst. Stand. Technol., NISTIR 4763, 1992.
 Zhang, X. and Hadhisophocleous, G., "An improved two-layer zone-model applicable to both pre- and post-flashover fires." Fire Safety Journal, 53, pp. 63-71, 2012.
 Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications, Volume 5: Consolidated Fire and Smoke Transport Model (CFAST), U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research (RES), Rockville, MD, and Electric Power Research Institute (EPRI), Palo Alto, CA, NUREG-1824 and EPRI 1011999, 2007.
 Reneke, P.A., Tofilo, P., Peacock, R.D., and Hoskins, B.L., Simple Estimates of Combined Stairwell / Elevator Egress in Buildings, Natl. Inst. Stand. Technol., Technical Note 1722, 2012.