Next Generation Fire Model: Auditing a Decade of Development

Dry bay fire has been a major contributor to aircraft ballistic vulnerability dating back to the Vietnam era. While much progress has been made on mitigating fires when they occur, the ability to predict if they occur has remained elusive. This challenge has led aircraft survivability analysts on a quest to develop robust methodologies and predictive models for dry bay fire across the range of operational combat threat engagements. The Next Generation Fire Model (NGFM)—the successor to the legacy Fire Prediction Model (FPM)—is the result of such an endeavor to realize the vision for a credible, physics-based tool that can rapidly predict fire ignitions in support of accurate system-level vulnerability assessments. This article offers a broad look at the last decade of NGFM development, surveying its historical underpinnings, major discoveries and advancements, and continued research challenges for the next decade.

Genesis of NGFM

Entering the 21^st century, the survivability discipline saw an increasing emphasis on credible fire modeling for accurate system-level vulnerability assessments. This emphasis occurred as advancements in high-speed imaging afforded new possibilities in data collection as part of developmental testing (DT) and Live Fire Test and Evaluation (LFT&E). As shown in Table 1, the period from 2010 to 2015 marked a transition from FPM, which had performance limitations with respect to fragment flash and hydrodynamic ram (HRAM) fuel spurt, as well as questions on the overall credibility of the model [1]. During this same period, the KC-46 LFT&E program revealed seemingly anomalous, instantaneous post-impact fuel spurts, which significantly impacted test results and challenged the predominant view of ballistic fire ignition. These instantaneous spurts—or “pre-spurts”—were first documented in 2005 by Disimile [2], but their implication to dry bay fire ignitions and system-level vulnerability assessments was not recognized at the time.

Table 1. Timeline of Significant NGFM Events

Accordingly, a Joint Live Fire (JLF) research project was initiated in 2015 to characterize fuel and water spurts from a generic cubical tank relative to the internal fluid dynamics [3]. This effort, which subsequently became the cornerstone for continued empirical spurt modeling, validated the use of water as a fuel surrogate for interrogating spurt responses without fire obscuration, when the differences in liquid density and sound speed are properly accounted for.

NGFM was formally launched in 2014 when a panel of subject-matter experts across Government, industry, and academia was established to create a strategic plan of action for its development [4]. The panel offered a fresh approach to fire modeling, decomposing the entire ignition chain into four functional areas/phases for dedicated research and development (as shown in Figure 1): (1) penetration, (2) energy deposition, (3) fuel deposition, and (4) ignition. Of these phases, only the penetration phase was assessed to have ample data, knowledge, and robust models (e.g., FATEPEN and ProjPen) to remain essentially status quo. All subsequent phases contained significant gaps in the knowledge base, requiring new research and model development.

Figure 1. Ignition Chain of Events.

In 2017, the Director of Operational Test and Evaluation (DOT&E) prioritized NGFM development, investing in two concurrent Joint Aircraft Survivability Program (JASP) efforts specifically targeting the energy and fuel deposition areas with testing and high-fidelity modeling of fragment flash and HRAM spurt, respectively [5, 6]. Flash testing surveyed duration times for mild steel fragments impacting two-panel arrays of aluminum, while varying projectile size, speed, target thickness, and obliquity. Spurt testing surveyed post-impact delay times from spherical and cubical steel fragments against various sizes of water tanks (no fuel) and across a comparable range of parameters.

Simultaneously, several high-fidelity tools were leveraged to gain a first-principles perspective of the underlying flash and spurt processes. Research by Yang [7, 8] successfully replicated the experimentally observed fluid dynamics of primary spurts using CFD-ACE+ and subsequently developed a novel reduced-order model (ROM) that could accurately produce spurt timing estimates with subsecond runtimes. This research was strictly focused on fluid dynamics and did not incorporate structural dynamics of the impact wall or any fluid-structure interactions (FSI) relevant to pre-spurts. Thus, Lawrence Livermore’s Arbitrary Lagrange Eulerian (ALE3D) hydrocode was leveraged to address this gap, successfully modeling threat penetration and FSI to replicate observed experimental trends in pre-spurt timing with reasonable accuracy [9].

As a trade-off for entraining structural dynamics, ALE3D results exhibited higher computational losses compared to pure computational fluid dynamics (CFD) tools for this HRAM problem, manifested by suppressed projectile trajectories through the tank over extended simulation times (i.e., 10+ ms). ALE3D was also used to model shock physics in fragment impact testing, but limitations of the code in this capacity forced exploration of an alternative Elastic-Plastic Impact Computation (EPIC) code to more accurately model fragment and target debris states [5].

Due to the significant run times and limited evaluation scope associated with these high-fidelity codes, empirical models were developed as the functional basis of the near-term NGFM in lieu of a physics-based ROM. This engineering-level approach assumes an ignition given a temporal overlap of flash and spurt and is validated against representative dry bay fire ignition tests conducted in 2020 [10]. As such, the approach provides a credible basic capability to predict fires with the predefined goal of 80% accuracy and 80% confidence. Alpha and beta versions of NGFM v1.0 stemmed from the validation effort, with internal reviews conducted between 2021 and 2023. Complicating considerations (e.g., spatial overlap of flash and spurt, latent heat, fragment breakup, chemical interactions, etc.) were deferred to future iterations of the model.

NGFM v1.1 was formally released through the Defense Systems Information Analysis Center (DSIAC) in July 2023 with narrow applicability for steel fragments impacting aluminum targets with only single perforations in the tank wall. A second release (v2.0) in May 2025 expanded the empirical database for this baseline scope and added functionality for armor-piercing incendiary (API) threats.

Key Findings

Quantity and Quality Matters

The foremost realization is that, despite a long history of LFT&E programs, the survivability community is deficient in the quantity and quality of data needed to build robust, empirical ignition models to modern accuracy standards, even with the simplistic NGFM approach adopted thus far. Past LFT&E repositories lack critical information on fragment flash and spurt physics, which is necessary to comprehensively model energy and fuel deposition processes. In addition, amassing suitable NGFM datasets requires not only dedicated test campaigns but also strategic leveraging of LFT&E programs. Considering this fact, the NGFM team recently developed and published a test and instrumentation guide—which accompanied the final report for the NGFM v2.0 release—to facilitate standardization of data collection across the tri-Services [11].

Fuel Pre-Spurt

Accurate prediction of fuel pre-spurt is critical in accurately predicting fragment-initiated fires, due to the timescales relative to flash durations. Fragment impact flashes are brief events compared to the prolonged function durations of API projectiles and often dissipate prior to the arrival of the primary fuel jet generated by the collapse of the internal wake cavity. Formation of pre-spurts—when they occur—is rapid enough to overlap with the flash cloud and cause an ignition. Figure 2 is a still image from high-speed video of a fragment traversing a generic, representative dry bay and impacting an aluminum shot panel attached to a fuel cell, resulting in a pre-spurt event within the visible duration time of the impact flash cloud.

Figure 2. Sample Dry Bay Fire Experiment Showing Pre-Spurt Overlapping With Fragment Flash.

Close inspection of high-speed test videos along with ALE3D modeling have distinguished two subcategories of pre-spurts: (1) shallow jet spurts (SJS), and (2) deep jet spurts (DJS) [9]. These spurts are the axial (shotline-oriented) jets originating by the sealing or pinching of the two ends of the internal cavity (as shown in Figure 3). SJS are those spurts that have been most frequently observed in dry bay testing and are generally the pre-spurts that ignite flash clouds. DJS are less distinguishable in external dry bay views, as their arrival time nearly coincides with the large volume primary spurt upon full cavity collapse.

Figure 3. Origination of Pre-Spurts From Shallow and Deep Cavity Seals.

Timing of SJS, or the post-impact delay to its dry bay arrival, was found by Barlow [11] to be strongly correlated to the tank wall’s natural frequency. This finding was the keystone discovery behind the empirical spurt models currently implemented in NGFM. Further research is needed to refine this correlation to account for the damping effect of fuel backing, as well as true boundary conditions existing in aircraft wing or fuselage dry bays. This finding also has implications for the design and use of replica—or “iron-bird”—test articles used in many LFT&E programs. Typical iron-bird tests use small shot panels (e.g., aluminum or composite) bolted to the spar of an all-steel test article such that the threat impacts representative aircraft materials along the entire shotline. These shot panels must now be sized and installed to match the fundamental frequency of the actual fuel tank wall to accurately evaluate fragment-initiated dry bay fires.

Fragment Flash

Prior testing has already demonstrated that the fragment flash duration on the second panel of a two-panel plate array (where the second panel is representative of the fuel tank wall) is significantly different than the flash duration on the impact of the first panel (representing the aircraft outer skin). Recent fragment impact testing has also revealed a significant difference in flash durations between air- and liquid-backed second panels. Initial comparisons indicate these flashes last two to three times longer, on average, in a liquid-backed vs. an air-backed configuration. This revelation could have implications for how baseline panel flash testing is conducted for future LFT&E programs, recognizing the limited set of existing data warrants additional examination for how extensive the effect is [11].

Test Instrumentation

One final highlight from the development of NGFM v2.0 is the potential utility of high-speed infrared (IR) imaging to broaden flash datasets beyond the visible spectrum and fold in thermal factors to the NGFM methodology. IR cameras were successfully demonstrated to record the latent heat of fragment impacts where the visible flash cloud had dissipated (as shown in Figure 4). In a small number of cases, ignitions have been observed to occur shortly after the visible dissipation of flash events, indicating a need for heat mapping as one future refinement for NGFM. Standardizing IR data collection, however, is limited by the speed of current IR cameras. Maximum speeds are currently an order of magnitude lower than high-speed monochrome cameras, thus presenting a significant synchronization burden, as well as data accuracy issues, until technology improves.

Figure 4. IR vs. High-Speed Imaging of a Fragment Penetration Through Two Aluminum Panels (Right-to-Left Shotline).

Continuing Challenges

Amassing accurate data will continue to be a necessity for strengthening the integrity and scope of empirical models, in addition to gaining further insights and validation points for physics-based models. The primary challenge here will be dealing with the inherent noise and nuisance variables of dry bay fire testing. Both flash and spurt phenomena are quite variable (flash more so than spurt), which demands significant sample sizes to establish meaningful datasets. Furthermore, fragment shatter, multiple impact holes, and tank wall ruptures are just a few of the additional factors that complicate empirical model development. Regions of the experimental design space dominated by these effects were not prioritized in early versions of NGFM, but they will have to be adequately addressed for NGFM to be relevant across the necessary range of operational conditions. Current deterministic models will also need to adopt probabilistic approaches until a breakthrough in physics-based research occurs.

Developing a fast-running, physics-based model is an overarching challenge, specifically for SJS, in fragment impact scenarios. Yang’s ROM for primary spurt is the archetype for translating a complex process into a simplified fast-running tool that would otherwise be restricted to computationally intensive simulations [8]. An analogous ROM for SJS will be a substantial breakthrough in realizing the vision for NGFM. Current high-fidelity simulations of the HRAM fuel deposition process face challenges of accurately and efficiently modeling FSI.

Figure 5 shows example ALE3D simulations that proved useful in SJS trend analyses but require continued research for resolving the domain with refined accuracy. Fluid dynamic and structural dynamic aspects of the problem can be independently modeled with dedicated finite volume and finite element architectures with superior accuracy, but a fully validated, integrated FSI simulation has yet to be achieved. Once established, these simulations will be essential in overcoming experimental limitations for elucidating the underlying physics of SJS.

Figure 5. Projectile Penetration and Cavity Formation in ALE3D Simulations [9].

One final looming challenge for NGFM, at least from an empirical modeling standpoint, will be incorporating additional pertinent factors, such as spatial overlap of flash and spurt, latent heat of nonvisible flash clouds, and chemical radicals influencing fire ignitions. While assuming ignitions on the basis of temporal overlap of flash and spurt provides an 80% solution, incorporating additional complexities will be necessary to capture the last 20% of the solution.

Conclusion

The combat aircraft survivability community’s +40-year investment in fire assessment is not only a statement of technological progress but also a testament to the complexity and difficulty of the field of research. A new foundation was successfully established over the last decade for the next generation of fire modeling, anchored by new insights into energy and fuel deposition processes as well as revelations of their complexity. A baseline empirical model is the latest product of the foundational plan of action, which, although limited in scope, has addressed known deficiencies of legacy methodologies with sufficient accuracy. Research challenges persist for future releases as the NGFM team presses forward over the next decade of development with the vision of a credible, physics-based tool underpinning higher fidelity in system vulnerability assessments.

About the Authors

Dr. Adam Goss is a U.S. Air Force engineer with nearly 20 years of Federal service. He currently heads the 704th Test Group’s Aerospace Vehicle Survivability Facility at Wright-Patterson AFB, OH, where he is responsible for executing Title 10 LFT&E, as well as research, development, test, and evaluation initiatives to enhance aircraft vulnerability assessments and reduction technologies. Dr. Goss is also the Chair of the Survivability Technical Committee of the American Institute of Aeronautics and Astronautics. He holds B.S. and M.S. degrees in aerospace engineering from The Pennsylvania State University and a Ph.D. in aerospace from The Ohio State University.

Mr. Timothy Staley is the Technical Expert in the Vulnerability Analysis Branch of the U.S. Air Force’s Life Cycle Management Center at Wright-Patterson AFB, OH. He has more than 23 years of experience in aircraft LFT&E, vulnerability reduction, and modeling and simulation (M&S), supporting numerous Air Force and Joint acquisition programs, including the F-35, KC-46, F-22, and F-15EX. He has also developed methods for integrating test and analysis data into M&S tools and provides technical direction on the development of the COVART ballistic vulnerability tool. Mr. Staley holds B.S. and M.S. degrees in mechanical engineering from the University of Kentucky.

References

1. Bestard, J. “Survivability Assessments—The Fire Prediction Model (FPM).” Aircraft Survivability journal, spring 2011.
2. Disimile, P. J., J. M. Davis, J. M. Pyles, and J. R. Tucker. “Hydrodynamic Ram Spurt Discovery Phase.” Joint Aircraft Survivability Program Technical Report M-05-02-001, AFRL-WS-WP-TR-2006-9001, October 2005.
3. Wacker, S. R., E. T. Brickson, and J. A. Sawdy. “Joint Live Fire (JLF) Aircraft Systems Detailed Test Report for Ballistically Induced HRAM Spurt Characterization.” Joint Live Fire Technical Report JLF-TR-14-05, August 2016.
4. Wacker, S. R., J. R. Tucker, and R. Anderson. “Next Generation Fire Model Plan of Action.” Joint Aircraft Survivability Program Technical Report M-14-11-01, January 2018.
5. Joint Aircraft Survivability Program. “Fragment Flash Characterization.” Technical Report for V-17-02, September 2020.
6. Barlow, B., and C. Leakeas. “Hydrodynamic Ram Induced Spurt Engineering Model Development.” Joint Aircraft Survivability Program Technical Report V-17-01-001, January 2022.
7. Yang, H. Q., P. J. Disimile, and G. J. Czarnecki. “A Multiphase and Multiphysics CFD Technique for Fuel Spurt Prediction with Cavitation and Fluid-Structure Interaction.” American Institute of Aeronautics and Astronautics Paper 2015-3419, June 2015.
8. Yang, H. Q., S. Y. Yang, and P.J. Disimile. “A Validation Study of Hydrodynamics Ram and Fuel Spurt Using CFD Tool.” American Institute of Aeronautics and Astronautics Paper 202-0356, January 2020.
9. Goss, A. E., J. P. Bons, and G. C. Burton. “An Arbitrary Lagrange-Eulerian Investigation of HRAM Shallow Jet Pre-Spurt Formation and Time Sensitivities to Impact Plate Dynamics. International Journal of Impact Engineering, vol. 167, 2022.
10. Tucker, J., and R. Stewart. “JASP Project V-20-01: Next Generation Fire Model (NGFM) Validation Project Final Report.” Joint Aircraft Survivability Program Technical Report JASPO-V-20-01-001, May 2023.
11. Stewart, R., J. Tucker, and C. Leakeas. “JASP Project V-23-02 and V-23-05: Next Generation Fire Model (NGFM) 2.0 Development Consolidated Report.” Joint Aircraft Survivability Program Technical Report JASPO-V-23-02-002, April 2025.