NEXT GENERATION FIRE MODELING
by Jim Tucker
Fires ignited by threats such as warhead fragments and armor-piercing incendiary (API) rounds are a major concern for both fixed- and rotary-wing aircraft. This concern is primarily due to the large presented area of aircraft components containing flammable liquids, such as fuel tanks, fuel lines, hydraulic reservoirs, hydraulic lines, lubricants, and electronics cooling fluids. The Next Generation Fire Model (NGFM) effort was launched in response to a recognized need by the Director, Operational Test and Evaluation (DOT&E) and the aircraft vulnerability community for enhanced fire prediction and modeling capability beyond current capabilities. This need spawned from the increased cost of test and test assets, growing challenges to integrate optimized vulnerability reduction technologies onboard aircraft, and the outgrowth of current fire modeling tools. This last realization is based on the increased knowledge gained in recent years from test programs that have been able to improve data diagnostic information, resulting in a better understanding of the detailed aspects for threat characterization, fluid spray, and, ultimately, ignition. This knowledge has also justified taking a fresh look at fire modeling and the establishment of a path forward.
BACKGROUND AND NEED
There are two main reasons that predicting fire is a top concern to the aircraft vulnerability community: aircrew survivability and economics. The former reason was raised by a review of air combat data in Southeast Asia that showed that fire and explosions contributed to more than 50% aircraft losses. The economic data are more current. As of FY13, the Air Force alone had spent tens of millions of dollars on dry bay fire testing, making fire the largest cost contributor for Live Fire Test and Evaluation (LFT&E) programs. The need for all this testing was driven by a large estimated uncertainty in total platform vulnerable area (Av) driven by probability of kill (PK) due to ballistic-ignited fires. The total uncertainty is a product of the fire uncertainty compounded by the large presented area of components containing flammable liquids in aircraft. Knowledge in understanding all aspects for the physics of fire and the development of models to simulate that understanding are insufficient within the survivability community for supporting advanced aircraft designs and growing requirements.
In the past, the vulnerability community has had two primary methods to evaluate fire. The older method, still in use today, is to use legacy test data if the aircraft conditions align with the test conditions that generated the data. If legacy data are insufficient, then more applicable data must be generated. The cost to run a test program is expensive particularly if production aircraft components are needed and the test requires external airflow to be sufficiently realistic. The second method uses IGNITE and/or the Fire Prediction Model (FPM).
FPM was developed organically over a number of years beginning in the 1990s. The model uses a combination of empirical relationships as well as basic physics, heat transfer, and chemistry to predict the chain of events beginning with penetration, through hydrodynamic ram (HRAM), fuel spray, droplet vaporization, and chemical reaction to predict the probability of ignition. In addition, FPM simulates events beyond ignition, including sustained combustion. IGNITE is a computer library consisting of components (modules) of FPM solely related to ignition. IGNITE was created to be called by higher level vulnerability codes and has been demonstrated with the Computation of Vulnerable Area Tool (COVART). The higher level codes are responsible for computing threat penetration and tank wall damage while IGNITE calculates the remaining elements of the ignition chain and returns a probability of ignition.
Although it has been recognized that existing methods have had their shortcomings, there were no suitable alternatives for modeling ignition. In 2014, DOT&E and the Joint Aircraft Survivability Program Office (JASPO) recognized the increased need for better tools. They also appreciated that, to generate these tools, an increased understanding of fire processes was needed. This appreciation initiated an effort to investigate and execute a plan for the development of an enhanced fire modeling capability—hence, NGFM.
NGFM INITIAL PLANNING
An initial fire model development planning effort was funded by the Joint Aircraft Survivability Program (JASP) with the main focus of forming a tri-Service team to determine community-wide fire modeling requirements and establishing an initial path forward. The project was established as JASP project M-14-11.
In October 2014, the Institute for Defense Analyses (IDA), an early proponent of improved fire modeling tools, hosted an initial working-level meeting with representatives from across the Department of Defense (DoD), as well as industry, academia, and the national laboratories. The panel of subject-matter experts (SMEs) were convened to establish, and then evaluate, the functional areas of the fire chain. The group’s collective experience included penetration, fire testing, fire protection, vulnerability analysis, hydrodynamics, aerodynamics, and combustion. The panel included participants from the U.S. Army Research Laboratory’s Survivability/ Lethality Analysis Directorate (ARL/ SLAD), the Air Force Life Cycle Management Center’s Combat Effectiveness and Vulnerability Analysis Branch (AFLCMC/EZJA), the Naval Air Warfare Center Weapons Division at China Lake (NAWCWD-CL), the Johns Hopkins University Applied Physics Laboratory (JHU APL), the Lawrence Livermore National Laboratory (LLNL), the Air Force Institute of Technology (AFIT), and the SURVICE Engineering Company. The team was led out of the Air Force Materiel Command’s (AFMC’s) 96th Test Group (96 TG/OL-AC), which is now the 704 TG/OL-AC, with oversight from DOT&E.
From the beginning, the ultimate goal of NGFM was to provide the analysis and test community with a model that is:
- Fast-Running: The model must support higher level vulnerability analysis codes, which need to run as many as tens of thousands of scenarios for a single threat at a single velocity. Therefore, NGFM must be capable of running in the submillisecond timeframe (i.e., faster than real time).
- Credible and Validated: A key reason for the lack of confidence in current tools is the lack of validation. To avoid this problem, NGFM must be validated at the most basic level as part of the development process.
- Modular: Modular development has many benefits, including supporting validation, allowing for parallel development efforts, and aiding in incremental development, where improved modules can easily replace less effective versions.
IGNITION: PRIORITY 1 IN THE FIRE KILL CHAIN
When most engineers and analysts outside of the vulnerability analysis community think of fire modeling, they think of a tool such as the National Institute of Standards and Technology’s (NIST’s) Fire Dynamics Simulator (FDS) or the Sandia National Laboratories’ codes ARIA and FUEGO, which can be coupled to look at complex fires involving composite materials. However, these codes concentrate only on the sustained combustion phases of fire. The user must already have knowledge of the initial events, the ignition phase. Given the complexity of threat-initiated ignition and the fact that there are currently well-supported codes that focus on combustion, it was determined early on in the NGFM plan development initiative that understanding and modeling the ignition phase of the fire kill chain would be a priority.
The amount of information available to help support solving the unique threat-initiated ignition problem is minimal so the planning team began definition of the requirements by establishing definitions of the discrete ignition events referred to as functional areas. These categories for API rounds are shown in Figure 1. (Similar definitions were generated for fragment penetrators.) The categories were further broken down by elements that could encompass a process or that could be broken down further into subelements, such as a property or state at the end of a process. Examples include residual mass, residual velocity, function type, fuel droplet size, etc. Based on the community’s current state of knowledge, lists of functional areas were developed for both API rounds and warhead fragments. High-explosive incendiary (HEI) rounds were not evaluated.
For one reason, API rounds and fragments have several common elements along their respective ignition chain of events, while HEI rounds are so dissimilar that little leveraging of API and fragment data can be done. Adding HEI rounds would greatly expand the scope, while the interest in the aircraft community is primarily concerned with API rounds and fragments.
IGNITION FUNCTIONAL AREA BASE KNOWLEDGE AND PRIORITY
After the functional areas were established, the team worked to develop evaluation criteria for how each element affected each of the four functional areas (i.e., how important is it), as well as to evaluate the state-of-the-art in terms of the community’s understanding and ability to model each of the elements.
The elements varied slightly depending on whether the threat was an API or fragment, and therefore the two chains were evaluated separately. SMEs filled out worksheets, recording their judgments using a numerical scale, which was color-coded using shades of red and green with yellow, indicating a rating midway along the scale. The darker shade of red in the Importance evaluation indicated the more critical the element was to a given functional area, while green meant no effect. In addition to providing ratings, SMEs provided justification, commentary on the current state-of-the-art, and recommendations for future work to improve the state-of-the-art.
Due to the interdependent nature of the chain (e.g., penetration elements can affect three of the four functional areas), as well as its sequential nature (i.e., down-the-chain elements do not really affect up-the-chain elements), it became challenging to evaluate the importance of some of the elements. An element such as the Ballistic Limit (V50) initially seemed to be the most important element for understanding ignition.
Table 1 shows the final Importance ratings for API rounds while Table 2 shows the Importance ratings for fragments. The Knowledge ratings were also captured but are not reproduced in these tables. They followed the same rating logic as the Importance ratings. In general, the more ratings that are redder (or higher numerical value) across both categories indicate the more motivation there is to study that element, as it would be both important and we lack the knowledge/ capability. In theory, improving our understanding of any one of these highest ranked elements has the potential to lead to dramatic improvements in prediction capability. Granted, improving understanding of a particular element may show the overall ignition chain is less sensitive to that element.
However, reducing the number of elements that are critical is almost as important since resources may be better applied studying other elements.
At the end of the evaluations and discussions, the rankings resolved into to three broad priorities for further work:
- Any Threat Penetration elements that affect either Energy Deposition or Fuel Deposition–HRAM elements.
- Any Energy Deposition elements that affect the Temporal or Spatial Overlap comparisons between the flash/function cloud and the liquid spray.
- Fuel Deposition–HRAM elements that affect Fuel Deposition–Spurt elements, including Spray Geometry, Spurt Ejection Time, and Droplet Distribution.
Three methods for advancing the community’s understanding were explored: additional research, testing, and advanced modeling. Regardless of the method, solid documentation would be required to continually build our understanding.
Additional research was recommended in elements that had no counterpart outside of the vulnerability community, or if there was a counterpart but the applicability to the ignition chain problem was unknown.
Exploratory testing could also be considered a form of research while more focused testing could serve model development or validation. Regardless, the key was the testing had to either be dedicated or have a primary objective of supporting the model. The number of tests and the detailed data required would always be at odds to piggyback on an LFT&E event.
Modeling using advanced codes would be too computationally intensive to solve the entire ignition chain problem on their own. However, the use of validated advanced modeling (e.g., hydrocodes, three-dimensional [3-D] multi-phase computational fluid dynamics [CFD] codes, etc.) could complement testing by allowing detailed study of phenomena of certain elements or processes. Advanced models could also be used to develop and “validate” faster running algorithms that could be used in NGFM.
There were also discussions about the foundation for each of the modules, empirical vs. first-principles/physics-based. In theory, empirical models can be developed more quickly, but there needs to be a wide breadth as well as depth of testing to develop enough data to cover the spectrum of scenarios and ensure test results have sufficient repeatability to make the model credible. A physics-based approach would demonstrate how well we understand a process but could be challenging for processes that are unstudied outside of our community. If feasible, a series of physics-based modules is the desired outcome. But it is recognized that for simulating some processes, an empirical approach may be the only solution, at least in the near term. Fortunately, with the modular approach, NGFM is never locked into a particular approach permanently. If a better, physics-based module can be developed, then it can replace an empirically based one.
F17 NGFM-RELATED PROGRAMS
The first two modules scheduled for development are fragment flash duration for aluminum striker plates and HRAM fuel spurt timing/spray characterization. The two coordinated three-year efforts are funded by JASP and closely monitored by IDA and DOT&E.
The fragment flash project (Fragment Flash Characterization, V-17-02) is being led by AFLCMC/EZJA, with involvement from the Naval Surface Warfare Center Dahlgren Division (NSWCDD), 704 TG/OL-AC, ARL/SLAD, and industry. The purpose of the project is to improve the ability to predict front face and back face flash durations from single and two-panel aluminum target arrays. Fragment flash characterization is critical to understanding if the fuel spurt and threat flash coincide in time and space. Given the shorter duration of fragment flashes and lower temperatures vs. API function durations, the fragment ignition chain is highly sensitive to this particular element. The approach is to use a combination of shock physics M&S combined with an extensive test series. The end product will be a library/module for predicting flash durations. Ideally, it will be physics-based, but the test data can be used to develop an empirical model.
The 704 TG/OL-AC will lead the HRAM project (HRAM Spurt Model Development and Validation, V-17-01), with involvement from AFLCMC/EZJA, ARL/SLAD, LLNL, and industry. The objective of this effort is to be able to accurately predict fuel spurt timing for flash/function cloud overlap calculations. As with V-17-02, the approach to the problem will be two-pronged: development of test data for validating an advanced model as well as statistical data analysis to develop an alternate empirical approach. Both this effort and V-17-02 will begin with detailed modeling to ensure the proper data are collected in each test series. Similar to V-17-02, the preference will be to develop a physics-based module using the advanced model to develop a set of faster-running reduced complexity algorithms; but at a minimum, an empirically based module will be developed.
Another NGFM effort taking place under the HRAM project is to develop the first version of the NGFM Software Development Plan (SDP). The SDP is a living document that will define the model framework and flesh out all the modules, including the data going into and out of each. As new modules are developed or modified, the SDP will be updated. Where it is known, the accuracy requirements for each module will be defined, and an evaluation of the state-of-the-art will be documented. Although not traditionally the role of an SDP, future versions of this SDP will also be used to identify the path forward in terms of which modules are in most need for updating.
As part of the NGFM planning efforts, an overall model structure was envisioned. Figure 2 shows the planned flow of information through NGFM, including the use of the two existing penetration codes: Projectile Penetration Model (ProjPen) for API rounds and Fast Air Target Encounter Penetration (FATEPEN) for warhead fragments. Their outputs along with the shotline boundary conditions and other properties would all enter the model. Simulations of in-tank HRAM events would predict the timing and nature of fuel spurts that would interact with the thermal energy (Energy Deposition) generated by either incendiary function or fragment flash. The outcome of this prediction in turn would lead to a check to determine if the spray and function/flash cloud simultaneously overlap. If not, then no ignition can occur. If they do coincide, then additional calculations are made to
check into calculations to determine if the duration is long enough to vaporize and ignite the fuel spray. Then the model will check if the heat generated by the ignition is greater than the heat lost to the bay or vaporizing additional droplets. If yes, then ignition is assumed to occur. In general, the model’s structure is not dissimilar from IGNITE, which is not surprising since the order of ignition chain events dictates which pieces of information are available as reactions unfold.
Because the first draft of the SDP is being developed in parallel with the flash and HRAM projects, the requirements for those two modules will be defined first to ensure their plans are well-synced with the overall NGFM effort.
The vulnerability community has come a long way in improving our understanding of the different elements of the ignition chain. We are now heading toward the ability to credibly predict these different processes and eventually reliably predict ignition across all ballistic scenarios of interest to the fixed- and rotary-wing communities. The first steps may appear to be limited in scope, but as identified by the original team, they will fill a big gap in our knowledge/capability in addressing fuel spray and threat flash overlap. From there, we can move onto composite materials and other scenarios. Then we can start drilling down into other elements to which the ignition chain is also sensitive. As modules are developed, NGFM will improve at each iteration, giving the community the capability it needs.
ABOUT THE AUTHOR
Jim Tucker currently serves as a senior engineer for the SURVICE Engineering Company. He has more than 21 years of specialized experience in aviation-related fire research and modeling, as well as the development, testing, and modeling of fire protection tools and methodologies. He is a current member of the NGFM Integrated Product Team. Mr. Tucker holds a B.S. in mechanical engineering as well as an M.S. in fire protection engineering from the Worcester Polytechnic Institute.