Development of a Lightweight, Crash-Tolerant Fuel Bladder
by Joshua Robbins, Lisa Chiu, John Koval, and Allison Monclova
Crashworthy and ballistic-impact-tolerant fuel systems significantly reduce rotorcraft fatalities due to fires. All Department of Defense (DoD) rotorcraft are now equipped with fuel bladders designed per MIL-DTL-27422 [1]. These requirements include a 65-ft full-scale fuel bladder drop test, as well as a material with puncture, cut, and tear resistance and a self-sealing capability for ballistic projectiles. Crash resistance has typically been achieved via similarity to past designs, or through trial and error, where fuel bladders are progressively tested and reinforced as necessary. This approach, however, adds undue cost, greatly extends the development schedule, and results in designs that are not weight-optimized to meet the requirements.
The purpose of the project discussed herein was to develop a methodology that enabled rapid and cost-efficient definition of lightweight fuel bladders capable of meeting the crashworthiness requirements of MIL-DTL-27422 [1]. Figure 1 illustrates the methodology used, which includes an iterative approach that leverages the application of computer simulations to arrive at a structurally optimized design.
Figure 1. Methodology for Optimization of a Crash-Tolerant Fuel Bladder.
First, structural optimization was completed using both linear and nonlinear finite element (FE) analysis, which generated a thickness distribution. Material testing informed the FE analysis and used lightweight materials developed by the METSS Corporation. The material and structural optimization was repeated until a final design was achieved. Next, the resulting topology optimization was used to define discrete zones for manufacturing. Then, the optimized design was used to build a fuel bladder. Finally, the bladder was subjected to the 65-ft drop test requirements per MIL-DTL-27422 [1], including comparison of the test results to the simulation.
Using this building block approach as opposed to a “trial-and-error” approach to crash resistance lowers risk, while also helping to develop a process that could include additional constraints beyond the single top fitting (e.g., additional side fittings, an increased size, and other geometry features found on bladders in production). Under this effort, two fuel bladders with increasing size and complexity were developed. The first was a Phase 1 cube using the standard dimensions found in MIL-DTL-27422 [1] and a single top fitting that was optimized by Corvid Technologies. The second was a larger bladder, nearly three times the fuel volume of the Phase 1 cube, similar in size to production fuel bladders in currently fielded vertical-lift platforms. The second bladder used a side fitting in addition to an optimized top fitting.
FUEL BLADDER CONSTRUCTION
Typically, fuel bladders follow a similar layup in construction. As illustrated in Figure 2, they include (going from the inside to the outside of the bladder) the inner liner, the nylon fuel barrier coating, the self-sealing rubber, the adhesive, and the reinforcement layers. The specific, lightweight material system used in this effort was developed by METSS. The optimization methodology focused on determining the optimal number of reinforcement plies across the bladder surface.
Figure 2. Typical Fuel Bladder Layup.
The Phase 1 and 2 fuel bladders are illustrated in Figure 3. The Phase 1 bladder was a 93-gal fuel bladder with a single optimized top fitting. The overall size was defined based on MIL-DTL-27422 [1]; however, some simplifications were made for this initial drop test. Namely, the optimized top fitting provided by Corvid Technologies was undersized, and the side fitting specified in Rev. F of MIL-DTL-27422 was not included.
Figure 3. Phase 1 (left) and Phase 2 (right) Fuel Bladders.
The Phase 2 bladder was designed to be more representative of production fuel bladders. The overall size was increased to 272 gal; and the bladder included a larger, optimized top fitting provided by Corvid Technologies, as well as a standard 4-inch-diameter side fitting.
DROP TEST MODELING AND OPTIMIZATION
Drop Test Modeling Approach
To ensure the predictions made by both the modeling and optimization software were indicative of real-world performance, it was necessary to accurately simulate both the material of the fuel bladder as well as the interaction between the water and the structural material. Therefore, the LS-DYNA FE solver was leveraged for its accuracy in dynamic loading environments.
The fluid materials and air domain were explicitly modeled using the Arbitrary Lagrangian Eulerian (ALE) method. The fuel bladder is modeled using a Lagrangian element formulation. The interaction between the fluid domain and the Lagrangian surfaces is controlled by definition of Fluid Structure Interaction (FSI). The definition of the fluid and fuel bladder simulation is shown in Figure 4.
Figure 4. Baseline Fuel Bladder Simulation Definition.
Quarter symmetry was used to define the geometry representative of a basic Phase 1 test cube with a single top fitting. The cube material properties were based on a traditional three-ply bladder wall construction. The pressure environment and resulting structural response due to the hydrodynamic ram are a function of the FSI predicted by the analysis. The baseline structural response for the 65-ft drop test is shown in Figure 5.
Figure 5. Baseline Fuel Bladder Simulation Results.
Optimization Methodology and Results
Optimization of the fuel bladder wall was accomplished by using an iterative process that leveraged both nonlinear, transient explicit LS-DYNA analysis as well as a topology optimization using an implicit and linear analysis in Altair OptiStruct. The topology optimization approach allows for the thickness of each element to be optimized separately.
An overview of the optimization process is shown in Figure 6. LS-DYNA was first used to obtain the displacement fields as well as the expected stresses and strains at numerous time steps that would be used as optimization points. Then the displacement fields were used to calculate the equivalent static loads (ESL). The ESL are defined as the static loads that generate the same displacement field as the nonlinear response of interest. Finally, using the ESL, linear static optimization was completed in OptiStruct to determine the thickness distribution for the fuel bladder. This approach uses a combination of both principal stress and principal strain constraints. The LS-DYNA bladder definition is then updated, and the process is repeated as necessary to converge on an optimal solution. The initial displacement fields used resulted from a simulated drop test using a three-ply bladder comprising the METSS material with self-sealing layers.
Figure 6. Optimization Method.
The results of the optimization effort are shown in Figure 7, with the blue regions representing the minimum thickness allowed and the red representing the maximum thickness allowed. Iteration 1 (Figure 7a) used a compliance minimization objective that attempts to find the stiffest bladder while using a specified volume fraction. Iteration 2 (Figure 7b) used an objective to minimize the mass, which resulted in additional weight efficiency. The results of Iteration 3 (shown in Figure 7c) show a similar trend, indicating that the analysis is converging on a solution. Through a final LS-DYNA run of this last thickness map, it was confirmed that no stress or strain constraints were violated. The results of the topology optimization were then used to define discrete zones for the manufacturing process (Figure 7d). The bladders were manufactured using traditional hand layup techniques with plies cut from a roll. The discrete zones were defined with guidance from METSS and Floats and Fuel Cells (FFC), which were the manufacturing partners.
After the bladder was discretized into manufacturable zones, the next step was to investigate the effect that the ply orientation had on the overall bladder performance and determine what the optimal orientation was for this configuration. The optimization routine for LS-DYNA (LS-Opt) was used as it was able to analyze the nonlinear simulation and optimize for it without any simplifications. The discretized bladder (Figure 7d) was analyzed to determine the optimal ply orientation across the layers. The analysis varied the orientation of the structural plies within the defined limits and explored all possible combinations.
Figure 7. Optimization Results.
A design of experiments approach was used to gauge the accuracy of the optimization. This leveraged a second-order polynomial metamodel and evaluated all possible combinations of layer orientations (given the relatively low number of combinations). The feasibility of a given orientation was also determined by setting the peak stress and strains as constraints (as was done during the topology optimization) and also optimizing for the lowest overall stress since that would yield the optimal orientation.
Figure 8 shows the comparison between the value of the peak stress that the metamodel predicted and the actual value computed during the simulation. This comparison provides an evaluation of the accuracy of the metamodel. As can be seen in Figure 8, all of the combinations analyzed were feasible. The ply combinations that resulted in the lowest peak stresses were ultimately selected for fabrication.
Figure 8. LS-Opt Metamodel Accuracy Plot.
TEST DEMONSTRATION AND CORRELATION
Fabrication
The build plan detailing the number of reinforcement plies for the Phase 1 fuel bladder is shown in Figure 9. The top, bottom, and sidewalls include a single layer of reinforcement across all surfaces where ply overlaps are required to join discrete plies. Subsequent layers then include additional structural plies. The second layer down (shown in green) is the corner reinforcement. The third layer down (shown in light gray) is referred to as the “belly band,” covering the entire circumference of the fuel bladder.
Figure 9. Phase 1 Fuel Bladder Build Plan.
The finished Phase 1 fuel bladder, shown in Figure 10, included an 8-inch x 12-inch structurally optimized fitting provided by Corvid Technologies. The lightweight construction means that the empty bladder doesn’t hold a perfect shape, as is visible by the slight irregularity in Figure 10. The final, as-built Phase 1 cube was 36% lighter than the conventional baseline. The weight for the conventional baseline was based on a bladder material weight of 1.353 lb/ft2 [2].
Figure 10. Phase 1 As-Built Fuel Bladder.
Drop Test Setup
The drop test was performed at the FFC facility in Memphis, TN, and was done in accordance with MIL-DTL-27422 [1]. The drop test tower and crash impact support fixture are shown in Figure 11.
Data acquisition during the test included high-speed video, Digital Image Correlation (DIC), as well as two additional views of video (front and back) using GoPro cameras. The DIC was used to measure the structural response of the fuel bladder walls, including surface displacement and strain. The test setup is shown in Figure 12. The DIC used two high-speed video cameras at 15,000 frames/second. Tracking of the surface response was enabled by a high-contrast pattern of dots painted on the bladder walls. The pattern was applied using a white basecoat and irregular dots approximately 0.50 inch in diameter applied with a permanent marker. In addition, a plastic sheet was put on the ground to prevent dust from obscuring the cameras. This practice is not recommended for future testing, however, as the sheet shredded during the test and nearly obscured the cameras.
Figure 11. FFC Drop Test Tower (top) and Crash Impact Support Fixture (bottom).
Figure 12. Phase 1 Fuel Bladder Test Setup With DIC.
Test Results and Correlation
The fuel bladder was dropped and allowed to fall freely, guided by the wires on either side of the crash impact fixture. The Phase 1 fuel bladder, which was dropped from a height of 65 ft in accordance with MIL-DTL-27422 [1], remained intact, with no evidence of leakage post-test. Pre- and post-test photographs of the fuel bladder are shown in Figure 13.
Figure 13. Phase 1 Fuel Bladder Pre-Test (left) and Post-Test (right).
Post-test evaluation included visual inspection as well as review of the high-speed video and post-processing of the DIC data. The DIC data was used to compare to the pre-test predictions generated by the LS-DYNA simulation. Figure 14 shows a comparison of the measured displacement compared to the simulation results at the time when the maximum strain occurs. Figure 15 shows a comparison of the peak strains for the measurement locations, with the time histories compared in Figure 16. The simulation compares extremely well with the measured time history, with the simulation underpredicting the measured peak strains by 0 to 13.5%. For the point at the lower corner location, the majority of the DIC data were lost during the drop test, thus precluding comparison to the simulation. In addition, the plastic sheet that was on the ground to prevent dust shredded on impact and blew upward, as can be seen in the DIC still image in Figure 14. The plastic sheet partially obscured the view, but the peak strains at the majority of the locations were still captured. Other interruptions in the DIC data occurred throughout, as can be seen in the time history in Figure 16. These interruptions are due to variables such as the plastic sheet, the folding of the bladder, and changes in lighting.
Figure 14. Displacement Comparison (in Inches) Between Simulation and Test.
Figure 15. Peak Strain Comparison Between Simulation and Test.Figure 16. Surface Strain vs. Time Plot Comparison Between Test and Simulation.
Figure 16. Surface Strain vs. Time Plot Comparison Between Test and Simulation.
Phase 2 Demonstration
Per MIL-DTL-27422 [1], the Phase 2 crash impact test is required for a full-size production bladder. Therefore, the second part of this effort was to repeat the use of the optimization methodology on a more complex, production-representative bladder design. As mentioned, the Phase 2 bladder had a volume of 272 gal (nearly three times that of the Phase 1 bladder), as well as the addition of a side fitting. The optimization methodology was used and resulted in an overall weight savings of 27%. Figure 17 shows the optimized Phase 2 bladder with the number of reinforcement plies specified (including the resulting overlaps from manufacturing).
Figure 17. Phase 2 Fuel Bladder Design.
As mentioned, in accordance with MIL-DTL-27422 [1], this bladder was dropped from 65 ft. The bladder survived with no leaks post-test. While this test didn’t use DIC for the surface strains (due to budget limitations), the high-speed video was post-processed to track the displacement of the sidewall. Table 1 shows the results of the measured vs. predicted displacement, where the magnitude agrees within 15%. Figure 18 shows a qualitative comparison between the simulation and test at the approximate point of maximum downward displacement (just before the bladder begins to rebound). Application of the optimization and analysis approach further demonstrates the methodology beyond the standard cube and provides confidence in its use for future bladder optimization efforts.
Table 1. Phase 2 Displacement Comparison
Figure 18. Phase 2 Simulation vs. Test Comparison.
CONCLUSIONS
Through execution of this effort, a structural optimization methodology was developed to efficiently manage the hydrodynamic ram in the 65-ft drop impact event through the bladder wall construction. The methodology was developed and demonstrated for both a Phase 1 and Phase 2 fuel bladder, which included increasing size and complexity. The designs included incorporation of the Corvid-optimized top fitting and use of the METSS lightweight material system. In addition, FFC provided its services for practical expertise in fabrication and test.
The optimization methodology developed under this effort enables more reliable, rapid development of weight-efficient designs. It successfully demonstrated the ability to predict hydrodynamic ram events during a fuel bladder drop test, and subsequently used those results to perform structural optimization of the bladder wall construction. This methodology resulted in a weight savings of up to 36% when compared to a baseline conventional bladder design and eliminated the need for a trial-and-error approach, thereby reducing cost and weight without compromising safety.
ABOUT THE AUTHORS
Mr. Joshua Robbins is a structural engineer with The Boeing Company. He has 3 years of experience in the aerospace industry, with specific experience in explicit nonlinear finite element analysis to enhance crash resilience and ballistic tolerance of composite and metallic structures. He also is experienced in production support and post-production damage tolerance analysis. Mr. Robbins has a bachelor’s and master’s degree in civil engineering from Arizona State University.
Ms. Lisa Chiu has worked as a structural analyst for The Boeing Company for 15 years, focusing on analysis, simulation, and testing of composite structures. She currently serves as the principal investigator for several research and development programs focusing on survivability, durability, and damage tolerance. She is also a Boeing-designated expert in ballistic structural analysis and ballistic test execution. Ms. Chiu holds a bachelor’s and master’s degree in mechanical engineering from the University of Massachusetts Lowell.
Mr. John Koval is a material, process, and physics engineer with The Boeing Company. He has 15 years of experience supporting various defense and space programs, including the F-22, F/A-18, B-1B, and International Space Station. He leads new material development and in-fleet support for many material systems, as well as materials standardization and qualification processes for new development programs. Mr. Koval holds a bachelor’s degree in materials science and engineering from the University of Illinois and a master’s degree in mechanical engineering from Washington University.
Ms. Allison Monclova is a vulnerability test engineer with the Naval Air Warfare Center Weapons Division – China Lake. She has 4 years of experience, with a focus on aircraft survivability, vulnerability reduction technology, live fire testing and evaluation, and project management for both Department of Defense (DoD) and non-DoD customers. She has a bachelor’s degree in mechanical engineering from the New Mexico Institute of Mining and Technology.
References
- “Detail Specification for the Tank, Fuel, Crash-Resistant, Ballistic-Tolerant, Aircraft.” MIL-DTL-27422, Rev F, 2014.
- Heater, K., J. Macarus, R. Watts, and B. Pilati. “Lightweight, High Performance Aircraft Fuel Bladders.” Aircraft Survivability, Summer 2011.
Acknowledgments
The authors would like to thank project team members Mark Wilenski (Boeing Research & Technology), Vladimir Balabanov (Boeing Commercial Airplanes), Jon Macarus and Ken Heater (METSS Corporation), Fred Tavoleti and Alex Hernandez (FFC), and Aaron Ward (Corvid Technologies). Appreciation is also expressed to JASPO for providing funding and leadership.
