Utilization of Hydrodynamic Ram Simulator to Determine the Dynamic Strength Thresholds of Structural Joints

By: Brandon Hull


The field of aircraft combat vulnerability is primarily responsible for determining the effects of man-made threat encounters on the operation, performance, and potential loss of an air vehicle. Fuel tank damage is one of the primary contributors to aircraft losses due to the large presented area and inherent volatility. Hydrodynamic ram (HRAM) is one of the leading damage modes related to fuel tanks alongside dry bay fire and ullage deflagration events. During an HRAM event, high-speed projectiles enter the fluid body and generate pressures in excess of 10,000 psi. The interaction of this fluid pressure and the surrounding structure causes high-strain-rate dynamic loading that leads to catastrophic failure, especially at joint locations between skins, spars, and ribs. When these joints are compromised, massive fuel loss through the damage location is likely. Also, if the fuel tank boundary is integral to the structure, primary load paths may be lost, leading to aircraft-level structural failures and the subsequent loss of aircraft. It is therefore imperative to understand the tolerance of fuel tank joints to HRAM loads.

The RamGun at Wright-Patterson Air Force Base (WPAFB) was developed in the mid-2000s to assess this issue. Funded by the Joint Aircraft Survivability Program Office (JASPO), the RamGun was developed to address a need for low-cost evaluation of structural joint designs to HRAM loading [1, 2]. Rather than building and testing an entire fuel tank, coupon-sized joints are manufactured, instrumented, mounted in the test chamber, and exposed to controlled HRAM loads. For this article, all joints were mounted such that the base was normal to the load, simulating a symmetric pull-off event.

Historically, the RamGun has been operated in such a way that the structural joints were driven to complete failure. Single-axis strain gauges were placed on the base and web of the joint. The base strain gauges monitored flexural symmetry, and the web gauges monitored bending and recorded the strain rate and pull-off strain-to-failure. The fluid pressures were also measured at the specimen location using pressure transducers in the test chamber [1]. While this approach permitted the determination of strain-to-failure, it did not enable a determination of load-at-failure. This value is critical for establishing material allowables from which aircraft designs can be influenced.

The approach outlined in this article was modeled after the up-and-down method used to determine the V50 limit of ballistic materials [3]. The goal of such an approach was to establish a pressure range—termed the “zone of mixed results”—in which the composite joint has an equivalent chance of surviving and failing. This range is particularly important for bonded composites since the failure of the adhesive bond is more complex than the failure of bolted joints.


As described previously, the novelty of the information provided in this article is the approach used to assess the tolerance of bonded composite joints to HRAM loads and the corresponding results. The goal of the up-and-down method applied was to establish a zone of mixed results. This type of test procedure is conditional in that the parameters of each test point were influenced by the results of previous testing in the same series. This method is particularly effective when the number of available test assets is limited. Rather than testing conditions that are not particularly informative regarding the joint performance, this adaptive procedure allows for a rapid convergence of test events on the area of interest.

The following is an example test procedure:

  • Test 1: Execute at a pressure for which the coupon is absolutely expected to survive.
  • Test 2: Execute at a pressure for which the coupon is absolutely expected to fail.
  • Test 3: Execute at the midpoint of the pressures used in Test 1 and Test 2.
  • Test 4: If the coupon fails Test 3, execute at the midpoint between Test 3 and Test 1 in hopes of survival. If it survives Test 3, execute at the midpoint between Test 3 and Test 2 in hopes of failure.

This process was repeated for as many specimens as were available for testing. At the end of the test series, roughly half of the specimens should have failed, and roughly half should have survived. Given the continuum nature of failure in bonded composite joints, there should also exist a range of pressures for which some joints failed and some survived. This range is the zone of mixed results. An example visualization of this zone is shown in Figure 1. Identical figures are provided in the results section for the four other configurations tested.

Figure 1. Zone of Mixed Results Example for Test Series.

For this configuration, exactly half of the coupons survived the HRAM event, and the other half failed. However, as shown, the highest pressure for which a coupon survived was 810 psi, and the lowest pressure for which a coupon failed was 716 psi. The region between these two values is the zone of mixed results.

The terms survival and failure were qualitative assessments of the joint status following testing. After each event was completed, the coupon was inspected for visible signs of damage. In all test cases categorized as a failure, this damage was a separation at the adhesive interface between the web and the base of the joint coupon. If there was no visible sign of damage, the coupon was categorized as a survival.

There was an additional hypothesis that joints that survived but fell within the boundary of the zone of mixed results would have degraded residual strength properties. To investigate this hypothesis, all coupons that survived the RamGun were preserved and tested in a quasi-static test fixture. The test methodology applied to these coupons was the same as that applied to two pristine coupons of the same configuration. This approach allowed for a comparison of failure load between pristine coupons and potentially compromised (i.e., RamGun-tested) joints of the same configuration.

If the pull-off load at failure for the RamGun-tested coupons was significantly less than the pristine coupons, it could be deduced that the internal microstructure of the composite or the adhesive bond was compromised in the RamGun testing despite the absence of a catastrophic failure. This hypothesis is further examined in this article.


Test Articles

To fully test out the methodology, five different article configurations were manufactured. They comprised three main constituents: a base, a web, and a pi preform. The base was constructed from unidirectional carbon fiber tape, the web was constructed from two-dimensional woven carbon fiber fabric, and the pi preform consisted of three-dimensional woven carbon fiber. All of the joint constituents were pre-impregnated with an epoxy resin system.

Figure 2. Pi Joint Schematic of Constituents.

A film adhesive was used to bind these three constituents together into a single joint coupon. The adhesive was formed into a pocket for the web and the pi preform clevis and laid flat to adhere the pi preform to the base. These components were all cured according to the process instructions inherent to the epoxy resin and adhesives. A schematic of the pi joints is shown in Figure 2.

The dimensional properties of these constituents were varied to generate five unique configurations. For each configuration, eight coupons were constructed for RamGun testing and two coupons were constructed for quasi-static testing. Twelve additional Configuration 1 coupons were tested in the RamGun as well, and results from those tests are included.


As previously introduced, the HRAM simulator (RamGun) is located at WPAFB’s Aerospace Vehicle Survivability Facility (AVSF) and operated by the Air Force’s 704th TG/ OL-AC. The RamGun was developed and has undergone improvements for the past decade, largely funded by JASPO. The apparatus was developed to impart HRAM loading on structural joints and evaluate their high-strain-rate failure in a more cost-effective manner than assembly-level testing. Since 2003, it has been used to test numerous joint designs, ranging from welded metal to bonded composites.

The RamGun is capable of generating fluid pressure waves similar to those expected during HRAM events. These pressure waves travel down a water-filled test chamber and are imparted onto the test coupon mounted at the end of the chamber. The test chamber of this apparatus is shown in Figure 3.

Figure 3. RamGun Test Chamber.

During operation, pressurized air is pumped into the gas chamber. This is the controlled variable during testing. When this gas pressure is released, it forces a Delrin puck down the gas gun barrel at velocities that range from 100 to 1,400 ft/s. At the end of the barrel is a steel plate, onto which the puck impacts. This causes fluid pressure waves to be generated in the water reservoir and travel toward the coupon location, as shown in the simulation in Figure 4.

Figure 4. Pressure Wave Model for the RamGun Chamber [1, 2].

The coupon is placed at the end of the test chamber. The base of the coupon is not restrained and rests against a foil membrane, which retains the fluid in the chamber. The web of the coupon is held in place by a web holder and affixed using torqued bolts. As the fluid pressure imparts itself on the base of the coupon, the web is restrained and the base is forced through the foil membrane, leading to a separation of the web and base in a failure event. If the joint survives, it is retained by the web holder.

The primary data sets collected from a RamGun test are the fluid pressure (measured through a combination of five pressure gauges) and strain response of the specimen mounted at the end of the fluid chamber. For the purposes of this test series, seven strain gauges were mounted to each specimen. The visual indication of failure and information from these instruments was used to draw conclusions on the performance of the joints and their resistance to HRAM loads.

Quasi-Static Fixture

The purpose of quasi-static testing was to identify coupons whose residual strength was degraded after exposure to RamGun testing. This testing was conducted using a four-point, roller-restraint setup with a constant applied strain of 0.04 in/min until joint failure. Two pristine specimens of each joint configuration were tested, and the load measured by the fixture’s load cell was used to establish a baseline performance. The same method was also applied to each test coupon that did not fail during RamGun testing. The test apparatus is shown in Figure 5.

Figure 5. Quasi-Static Test Fixture.


During RamGun and quasi-static testing, all joint coupons failed at the adhesive layer between the bottom of the pi preform and the base. There were no undesired failures in the web or at the clamp location. For some failed coupons, the entire pi appeared to disbond uniformly and immediately, while others appeared to peel and delaminate a few layers of the base during failure. Although this difference is interesting, there was no discernible effect on the pressure at which the joints failed.

Examples of a survived and failed specimen are presented in Figure 6, followed by a description of the RamGun test results and the quasi-static test results.

Figure 6. Survived and Failed Test Coupons After RamGun Testing.

RamGun Results

As stated previously, the purpose of the methodology employed during this experiment was to test each configuration at discrete pressure values such that a zone of mixed results for the pressure load at failure could be established. The results are shown in Figures 7–11 for the tested configurations. A brief description of each test series is provided after each figure.

The top of each figure reports the tested pressures that resulted in a failure, and the bottom reports the pressures that survived the RamGun. Each discrete test event is plotted on the corresponding axis. The green box represents the pressure range for which there were no failures, and the red box represents the range for which there were no survivals. The zone of mixed results is the overlap of these two regions.

Configuration 1 had 20 test points. Of all points tested, 13 coupons failed and 7 survived. The lowest pressure at the coupon location that caused a failure was 256 psi. The highest pressure for which a coupon survived was 484 psi. Several coupons were tested at extremely high pressures to explore the influence of strain-rate effects on absolute strain at failure. However, information related to this portion of the test is not included in this article.

Configuration 2 had eight test points. Of all points tested, four coupons failed and four survived. Given the construction of this configuration, it was expected to outperform Configuration 1, and the results matched the expectations. The lowest pressure at the coupon location that caused a failure was 716 psi. The highest pressure for which a coupon survived was 810 psi. The average pressure in the zone of mixed results was 763 psi, which is more than double the Configuration 1 performance for pressure load at failure.

Configuration 3 had eight test points. Of all points tested, two coupons failed and six survived. Although the up-and-down method was applied to these coupons, the variability of the RamGun caused more survivals than desired. The puck velocity correlates well with the air chamber pressure; however, there is variance in the puck impact angle and fluid pressure wave attenuation through the water column that affect the pressure at the coupon location. Configuration 3 testing was the most affected test series by this variance. The zone of mixed results for this configuration had the smallest range of all configuration tested, from 692 to 707 psi.

Configuration 4 had eight test points. Of all points tested, three coupons failed and five survived. This was the only configuration tested for which the zone of mixed results did not emerge. The gap between the highest survival pressure and the lowest failure pressure was 80 psi. If more test coupons were available, it is probable that a zone of mixed results would emerge by testing around this region. Configuration 4 survived the highest pressures of the five configurations tested, with an average of the survival and failure gap pressures of 1,252 psi.

Configuration 5 had eight test points, an even number of which passed and failed during RamGun testing. The average pressure in the zone of mixed results was 781 psi. This result was similar to the performance of Configurations 2 and 3.

Upon conclusion of the RamGun test series, the up-and-down method was shown to be successful for identifying failure pressure regimes using discrete test events. Configuration 1 clearly withstood the lowest pressures, and Configuration 4 withstood the highest pressures. Configurations 2, 3, and 5 all performed similarly and roughly averaged the performance of Configurations 1 and 4.

Figure 7. Configuration 1 RamGun Results.

Figure 8. Configuration 2 RamGun Results.

Figure 9. Configuration 3 RamGun Results.

Figure 10. Configuration 4 RamGun Results.

Figure 11. Configuration 5 RamGun Results.

Quasi-Static Results

After the RamGun testing was completed, quasi-static testing was conducted on two pristine coupons of each configuration and all RamGun-tested coupons that survived. The first item of note pertains to the quasi-static pristine coupon pull-off load compared to the RamGun pressure loads at failure. Table 1 shows the results.

The general trend of configuration performance discovered in the RamGun testing was similar to that exhibited during quasi-static testing. In general, Configuration 1 failed at the lowest loads and Configuration 4 failed at the highest loads. However, during quasi-static testing, the Configuration 2, 3, and 5 performances were substantially different from one another, whereas their average failure pressure during RamGun testing were all within 100 psi. This result indicates that the dynamic nature of RamGun testing and the corresponding high strain rates affect the performance of the joint, and those effects are construction-specific.

At the outset of this test, it was hypothesized that a relationship may have been made between quasi-static performance and dynamic performance. After the completion of both test series, no numerical correlation between the two values could be made given the available test data. However, given the apparent and consistent nature of the general trends, qualitative determinations of joint design characteristics that result in the worst and best performance can be made. This information is pivotal during the iterative design process to achieve a survivable fuel tank configuration.

The second purpose of the quasi-static testing was to identify coupons that did not completely fail but had residual strength losses following RamGun testing. All 17 coupons that survived RamGun testing were subjected to the same quasi-static test as the pristine coupons. A load rate of 0.04 in/min was applied until separation of the web and base occurred. Of all coupons tested, there were three that failed at a significantly lower load than the pristine coupons: two were Configuration 3, and one was Configuration 5. Figure 12 displays the pull-off load for all quasi-static tested coupons.

Figure 12. Pull-Off Load for All Quasi-Static Tests.

The three coupons that suffered residual strength losses are highlighted in the figure. All three of these coupons fell within the zone of mixed results for their respective configurations. The Configuration 3 coupons were tested at 707 and 698 psi. As shown in the results, both of these values are within the extremely tight zone of mixed results for this configuration (692–707 psi). Their failure loads were 73% and 44% less than the pristine sample, respectively. The third joint that displayed compromised residual strength was a Configuration 5 coupon that was exposed to 804 psi and failed at a load 23% less than the pristine sample. This coupon is also in the zone of mixed results for Configuration 5 (695–867 psi).

This result indicates that the microstructure of these coupons was compromised during RamGun testing even though complete failure did not occur. This indication is understandable since these coupons all fell within the zone of mixed results, meaning that other coupons of the same configuration completely failed at or below these test pressures. However, since there were many coupons that also fell within the zone of mixed results but did not exhibit degraded performance, additional investigation would be necessary to understand the cause of the strength loss. It is suspected that a loss of bond integrity initiated but did not fully propagate for those coupons. Techniques such as nondestructive inspection (NDI) could be used in future testing to check for the presence of anomalies in the bond line.


This article has presented a method for establishing the dynamic failure limit of bonded composite joints exposed to HRAM loading. This method was inspired by the up-and-down procedure used to determine the V50 ballistic limit of armor [4]. Using the RamGun, various joint configurations were exposed to HRAM pressure loads, and their survival or failure was recorded. The goal of the test was to establish failure regions of pressures, or zones of mixed results, in which various joint configurations could potentially fail or survive. This information could be used to influence aircraft analysis and design to achieve a more survivable product with respect to HRAM event tolerance. After the completion of RamGun testing and assessment of the data, it was apparent that the proposed up-and-down method was successful in identifying failure regimes for all configurations tested. Of the five configurations, four had clearly identified zones of mixed results. For the fifth configuration (Configuration 4), the gap between the highest pressure for survival and lowest pressure for failure is only 80 psi. If more coupons were available for test, it is likely that a zone of mixed results for this configuration would have emerged as well.

There was a secondary hypothesis that, for joints that survived but fell within the boundary of the zone of mixed results, the residual strength properties would be compromised. This hypothesis was examined through the use of quasi-static testing. Pristine coupons of each configuration were pull-off-tested with quasi-static loading (0.04 in/min) to establish a baseline performance for failure load. All joints that survived RamGun testing were subjected to the same procedure, and their failure loads were determined.

Of the 18 coupons that survived RamGun testing and existed in the zone of mixed results, three coupons showed significantly degraded performance compared to the pristine coupons during quasi-static test. Two of the joints were Configuration 3 coupons. Their failure loads were 73% and 44% less than the pristine sample, respectively. The third joint failed at a load 23% less than the pristine sample. All of these coupons were RamGun-tested in the zone of mixed results for their respective configurations. It is likely that the internal microstructure of the composite or the adhesive bond for these three coupons was compromised in RamGun testing, although a catastrophic failure did not occur.

Qualitative conclusions regarding joint configuration performance to HRAM loading can be made from the test results. Configuration 1 is the most susceptible to failure to HRAM, and Configuration 4 retains its integrity to the highest loads. This information matches the quasi-static test results and can be used to influence designs and achieve a more survivable fuel tank configuration. Unfortunately, the test did not allow for the development of a numerical relationship between quasi-static pull-off performance and RamGun pressure loads at failure. This result is likely due to the complexities of high-strain-rate effects during dynamic events.


The RamGun is managed and operated by Mr. Jason Sawdy, 704th TG/OL-AC, at WPAFB, OH. The guidance of Mr. Sawdy and his team during test planning and execution was pivotal to the test’s success. In addition, Mr. Ron Hinrichsen of Skyward Ltd. offered extensive assistance related to HRAM modeling.

Finally, significant input related to the approach was provided by Mr. Tim Staley, AFLCMC/EZJA, and incorporated into the final product described herein.


Mr. Brandon Hull has worked as an aircraft vulnerability engineer at Northrop Grumman Corporation’s Aerospace Systems since 2016, performing vulnerability analyses, supporting live fire test planning and execution, analyzing test results, providing design guidance for vulnerability reduction, and evaluating vulnerability reduction features and technologies for implementation. Prior to that, he served as a structural and ballistic survivability engineer at Boeing Phantom Works and completed his master’s thesis at NASA Langley Research Center. Mr. Hull holds bachelor’s and master’s degrees in aerospace engineering from Virginia Polytechnic Institute and State University.


[1] Czarnecki, G., M. Maxson, J. Sawdy, M. Miller, and R. Hinrichsen. “Evaluation of Skin-Spar Joint Resistance to Hydrodynamic Ram.” JASPO-V-4-04-001, March 2006.

[2] Hinrichsen, R., S. Stratton, A. Moussa, and G. Zhang. “Hydrodynamic Ram Simulator.” JASPO-V-07-06-001, AAC-TR-08-17, September 2008.

[3] MIL-STD-662F. “V50 Ballistic Test for Armor.” 18 December 1997.

[4] Ball, R. E. The Fundamentals of Aircraft Combat Survivability Analysis and Design. 2nd edition, American Institute of Aeronautics and Astronautics, Reston, VA, 2003.


This article was presented at the American Institute of Aeronautics and Astronautics (AIAA) SciTech Forum in January 2019. Reprinted with permission. Copyright 2019 by AIAA, Inc. All rights reserved.