## Effect of Ballistic Shot Dependency and Weave Matrix on Composite Materials

### by Lt. Jack Morgan, Lt. Alex Ramsperger, and Maj. John Hansen

Modern composite materials increasingly show promise for armor applications throughout the U.S. military, including in combat aviation. Compared to traditional armor, composite armor can provide higher strength per unit mass for meeting weight constraints for force protection and vehicle survivability. The armor’s capability to withstand penetration is perhaps the most significant characteristic for this application, thus underscoring the ongoing need to accurately model the ballistic strength of the advanced materials that composite armor comprises. Modeling and testing standards for hard armor materials used by the Department of Defense (DoD) are contained in MIL-STD-662F, but this standard does not adequately consider the unique properties of modern composite materials.

As most practitioners involved in military armor testing are aware, ballistic standards are commonly centered around the concept known as *V _{50}*, which represents the ballistic velocity at which there is a 50% probability a projectile will completely penetrate a material. MIL-STD-662F uses

*V*and also defines a “fair” impact, when the unyawed projectile of a specified obliquity impacts a distance of at least two projectile diameters from any previous impact or a disturbed area resulting from the previous impact [1]. The work described herein focuses on secondary ballistic impacts, when a projectile impacts in the vicinity of a primary impact point, either within or outside the spacing around the disturbed area generated by the primary, or fair, impact described by MIL-STD-662F.

_{50}Kinsler and Collins [2] investigated the effect of secondary ballistic impacts on the *V _{50} *of Ultra-High Molecular Weight Polyethylene (UHMWPE) and observed instances where the

*V*could increase for secondary shots. However, the study lacked in-depth statistical analysis. Keane [3] methodically investigated this phenomenon to gain more understanding and also observed instances where the

_{50}*V*increased for secondary shots placed relatively close to primary impact points that yielded a partial penetration. Hankins also investigated this topic and found a different result: there was no significant change in

_{50}*V*for all materials tested [4]. Four composite materials in the Hankins study were used: carbon fiber, glass fiber, aramid fiber, and UHMWPE. An additional variable was also added since three different fabric weave matrices were included [4]. Although several materials were compared, the influence of the fabric matrix is unknown and thus warrants further analysis.

_{50}The primary goal of the research described herein is to provide ballistic strength data for composite materials. A more specific goal of this research was to determine the influence of the fabric weave matrix by testing for the ballistic strength of aramid fiber and glass fiber with a polyurethane thermoplastic matrix. The primary ballistic strength estimates were compared to the Hankins results of the same two fiber types, but with a thermoset matrix [4]. Secondary ballistic strength estimates were also made and compared to the primary estimates for each composite material. Knowing the effect of the polyurethane thermoplastic matrix helps to accurately update the MIL-STD-662F for composite materials.

### BALLISTIC IMPACTS

Ballistic impacts are typically defined as an object striking a larger plane or article at high velocity. The object is defined as either a projectile (a device designed for ballistic performance), a penetrator (an object with exclusively terminal ballistic function), or a striker (an object initiating an impact). For this research, the projectiles used for testing were 1/2-inch steel ball bearings, which were considered blunt bodies, having a nose half-angle of greater than 90°. The spherical shape also eliminated any ballistic yaw, the angle the nose has upon impact. Blunt projectiles tend to exhibit plugging, where a section of the article is pushed out, over a cylindrical area. However, due to the large number of substructures, the plies and fibers, in the articles used in this research, some characteristics of sharp projectiles may occur, such as failures about the projectile axis, with delamination spreading about the impact point [5].

When an impact occurs, a hemispherical compression wave is generated in the article. This wave propagates through the article and reflects as a tension wave at the back of the article. The overlap of these two waves creates a high-tension stress, which can result in internal cracks due to tension or ejection of layers of material at the back of the article as spall. This stress could also result in failure in diagonal shear, resulting in a large section being ejected [6]. To fully penetrate an article, the mechanics depend on whether the projectile is blunt or sharp. For blunt impacts, a plug of material forms with the cross section of the projectile. To absorb the impact, the article must absorb the kinetic energy of the projectile. It does this by transforming the kinetic energy into material deformation of the plug, formed at impact. The energy of deformation can be further broken into the energy at the crater that separates the plug from the article and the energy expended by the plug to reach common velocity with the projectile. Through an energy and momentum balance, various parameters can be solved. For sharp projectiles, two damage mechanisms occur, ductile hole enlargement and petaling, where the edges of the article spread around the projectile. As the projectile travels through the hole initiated by the projectile, the article material expands as the cross section of the projectile increases. Furthermore, radial fractures, which travel outward, form petals at the impact point. Analyses of these phenomena are typically done through work analysis and momentum balances [5]. Despite the blunt nature of the projectile used in this research, more petaling than plugging was observed during testing.

To better absorb a ballistic impact, a material must be able to absorb the kinetic energy of the projectile. The greater ductility of the material, the better capability to exchange kinetic energy with energy of deformation. Furthermore, if a material is more ductile, the impact time will lengthen, increasing the impulse applied on the projectile and reducing the momentum. However, should the material be too weak, it will fail under the stress of impact and be unable to make the necessary energy exchange. For this reason, materials that perform well under ballistic impacts must be sufficiently strong to not fail at impact and sufficiently ductile to exchange the kinetic energy of the projectile into deformation energy.

### BALLISTIC STANDARDS TESTING

One of the key measures of performance in ballistic testing is the ballistic limit velocity, *V _{BL}*. This velocity is defined by the previously mentioned MIL-STD-662F, the DoD test standard for ballistic testing. The ballistic limit can be defined as the minimum velocity at which the projectile is expected to consistently penetrate the armor completely, the maximum velocity at which the projectile is expected to fail to penetrate the armor completely, or a velocity within the zone of mixed results (ZMR). The ZMR is the velocity region where two identical shots, under the same circumstances, have two different results. For this research and MIL-STD-662F,

*V*is the desired measure of performance within the ZMR. Other percentages can be designated, with

_{50}*V*designating a velocity with

_{x}*x*% chance of penetrating armor, but

*V*was the focus of this research [1].

_{50}For a ballistic impact, two types of penetrations are possible: a partial penetration (*P _{P}*) and a complete penetration (

*C*) (see Figure 1). For this research, a

_{P}*P*was defined as a penetration where the kinetic energy of the projectile was completely absorbed or reflected by the test article. This means that the residual velocity, the velocity of the projectile behind the test article, was nonexistent. This phenomenon arises if the residual velocity was equal to 0 or the projectile was reflected by the test article. Likewise, a

_{P}*C*was defined as a penetration such that the residual velocity of the projectile behind the test article was greater than 0.

_{P}### 3-POD METHODOLOGY

The 3-Pod methodology [7] was used extensively throughout this research as a process to estimate the *V _{50}*. The Hankins study demonstrated that this method provided the most accurate and efficient results [4]. The methodology consists of three phases to approach sensitivity testing:

- Quickly identify an experimental range
- Optimize parameter estimation
- Choose design levels to be placed near
*x*to obtain information about_{P}*x*_{P.}

For this research, the first phase of the 3-Pod methodology was primarily used. The goal of the first phase is to hone in on the range of the stimulus estimate of the ZMR by using three stages. To initialize the algorithm, initial range of stimulus values must be entered. Wu and Tian [7] recommend for the range to encompass at least six standard deviations from the expected median. Once the program was initiated, the three stages occurred as follows:

- Obtain stimulus bounds (1
*C*and 1_{P}*P*)_{P} - Find overlap for the ZMR
- Improve the ZMR.

Completion of these three stages of the first phase was sufficient to converge on estimated values for *V _{50}*, the average response (

*μ*), and its standard deviation (

*σ*).

### MATERIAL PROPERTIES

The materials tested in this research were 8 hardness satin (8HS) satin weave S-glass fiber and plain weave Kevlar KM2 600 Denier fiber, more commonly known as aramid fiber. The physical material properties of both the glass and aramid fiber materials are outlined in Table 1.

Tensile strength and Young’s modulus are used to describe the stress and strain properties of the material. A higher tensile strength suggests the plate will be stronger, while a higher Young’s modulus suggests the material will be more brittle. It was expected that the aramid fiber plate would have more deformation present than glass fiber due to the lower Young’s modulus. The most optimum material would minimize density and maximize tensile strength while maintaining an appropriate Young’s modulus.

Comparison of the fabric weave matrix properties are significant to analyze as well. Table 2 compares the previous matrix tested by Hankins [4], AF163, to the new UAF-472 polyurethane thermoplastic matrix tested in this research.

The previous AF163 matrix has higher tensile strength than the current UAF-472 matrix, which suggests the latter matrix has more ductile characteristics. Additionally, the Young’s modulus of UAF-472 has 10% of the value of AF163, which suggests the new thermoplastic matrix may exhibit greater deformation upon impact, and possibly larger *V _{50}* values.

The test articles used in this research came from the same manufacturer and manufacturing batch. Each had a thickness of 0.25 inches and had a UAF-472 polyurethane thermoplastic matrix. The Hankins test articles were constructed in the same manner but with an AF163 matrix [4]. The material properties of these plates were consistent with the material properties from Hankins (as shown in Table 3). This shows that the only difference between the two groups of each fiber material was the matrix material.

### RANGE TESTING

To organize the test point matrix, all secondary shots must have a qualitative definition for their placement relative to a primary impact location and extent of damage. These definitions are designed to align with the test standards set by Hankins [4] to ensure a fair and repeatable test while isolating as many design variables as possible. This organization distinguishes each shot as close, medium, and far impacts, whose definitions are expanded upon in this section.

As the armor in this investigation was a plied composite, the material was expected to delaminate under ballistic impacts. In contrast, homogeneous materials are assumed to show no delamination. Under MIL-STD-662F, the standard describes that a shot-to- shot distance of two projectile diameters from any previous damage is sufficient to qualify as a far spacing and should perform as a clean shot. The expected material performance under each shot type was developed based on the results from Hankins [4].

Close impacts were shots impacting two projectile diameters from the initial shots. This ignored the damage specifications of MIL-STD-662F but follows the two-projectile-diameter specification. These were taken such that the secondary shots impacted within the zone of delamination of the initial shot. The zone of delamination from the secondary shot was expected to overlap the impact point of the initial shot.

Medium impacts were shots impacting at a minimum distance following the standards set down in MIL-STD-662F. For these shots, delamination regions were expected to overlap between initial and secondary shots, but again, this is not a requirement. The ballistic limit for secondary shots was expected to be statistically similar to primary shots in this region.

Far impacts were shots impacting such that the delamination regions of the initial and secondary shots did not overlap. These shots were treated as fair shots, acting as independent events. The distance required is material dependent as the delamination changes with material. The ballistic limit was expected to remain statistically the same in this region. These shots were used to maximize the utility of each test specimen.

There were 25 plates of each composite material available for testing. The 3-Pod methodology was used to find all *V _{50}* estimates, which typically took up to 12 shots to accurately determine this value. If the

*V*was found in less than 12 shots within Phase I of 3-Pod, the remaining shots were used in Phase III to gain a more accurate value. The location of the clean shots were placed in each quadrant of the test plate having a distance of 3 inches from the bottom and side. The velocities of the shots were recorded, and the shots were labelled with either a

_{50}*C*or

_{P}*P*. Clean shots were taken first until the plate had one shot in each quadrant.

_{P}Next, up to four secondary close and medium shots were taken adjacent to those initial clean shots. A single plate had a maximum of eight total shots. The secondary shots were organized into four different types for *V _{50}* estimation: Close

*C*, Close

_{P}*P*, Medium

_{P}*C*, and Medium

_{P}*P*. These types describe the result of the initial shot and the shot-to-shot spacing of the secondary shot.

_{P}The primary data relevant to this research were the velocity before impact, shot-to-shot distance, and delamination post-impact. The residual velocity, the velocity behind the test article, was also recorded for the *C*’s but was not relevant to data analysis. This section details the data collection methods. The methods and equipment for determining the relevant velocity measurements were the same as those used by Hankins. A thorough explanation of these methods and equipment can be found in his thesis [4].

The extent of delamination was determined using an audiovisual tap test on articles for which *V _{50} *was calculated.

Delamination was measured on a damage diameter basis. For visual inspection, calipers were used to measure the delamination on the front and rear of each plate, centered on the impact location. Measurements were taken both after initial shots and secondary shots. The audio tests for material delamination consisted of a manual tap test. To perform the test, the material was supported by an 80/20 aluminum extrusion. A box-end wrench was used to tap the article outside the possible extent of delamination. The article was tapped progressively closer to the impact location until a discernible difference was noted and the extent of the delamination was marked. This process was repeated until the entire delamination region was characterized. The process was then repeated on the rear of the test article. The delamination that was detected using audio was greater than the delamination present during visual inspection. Given the inability to visually inspect internal delamination, the auditory delamination was the one recorded and marked. The *V _{50 }*estimates for each test series were then compared by conducting a hypothesis test, assuming normality, using the null hypothesis that the ballistic limits were equal.

### GLASS FIBER RESULTS AND DISCUSSION

The clean series for glass fiber consisted of seven test series (as shown in Figure 2), which took an average of 12.6 shots. For the clean test sample series, 88 shots were used. An additional 10 shots were taken for a total of 98 shots, with 50 *C *’s and 48 *P’*s. Shots were placed 4.384 inches apart such that the zones of delamination did not overlap. This met the definition of far shots, and each was a fair impact. The results of each clean test series can be seen in Table 4, and the characterization of delamination zones for *C*’s and *P _{P}*’s for clean shots can be seen in Table 5.

To compare the results of the various glass fiber test series, hypothesis testing was also performed. Each test series was assumed to be normal, and a one-tailed hypothesis test was performed with the null hypothesis being that the ballistic limits were equal and the alternative hypothesis being that the ballistic limits were not equal. All hypothesis tests were conducted at a 95% confidence interval with *ρ* = 0.05 (1 minus the confidence interval). If the cumulative probability (*α*) of any one value compared to the normal distribution was less than *ρ*, the null hypothesis was rejected. The *α* value was calculated using the *z* score for the value being compared to the normal distribution. If the results of the hypothesis test concluded that the null hypothesis failed to be rejected, it was assumed that the results could be combined with a new average and standard deviation.

When comparing the clean series among each other, the first clean series had a maximum *α* value of 0.023. This result was enough to reject the null hypothesis and conclude that the other ballistic limits were different from the first series. The second clean series had an *α* value greater than 0.05 for all series except for clean series one. This result was enough to fail to reject the null hypothesis for with all series except the first series. The third series followed the same trend. The fourth series had an *α* value less than 0.05 when compared to the first, third, and fifth series. The fifth series had an *α* value less than 0.05 when compared to the first, fourth, sixth, and seventh series. The sixth series had an *α* value less than 0.05 when compared to all series except for the fourth and seventh series. The seventh series had an *α* value greater than 0.05 whencompared to every series. Based upon these results, the first series was unique enough to remain alone. The second, third, and fifth series were averaged together as “Clean Average 1,” and the fourth, sixth, and seventh series were averaged together as “Clean Average 2,” as shown in Table 6.

When compared to the Hankins data (using the same hypothesis test process as described previously), the first clean test series and both clean averages were statistically distinct. For the first test series, there was a 69.87 fps increase, which is a 10.71% increase. For the first clean average, there was a 46.37 fps increase, which is a 7.11% increase. For the second clean average, there was a 57.49 fps increase, which is an 8.81% increase [4]. This comparison highlights the effect of the matrix material on the primary *V _{50} *ballistic limit.

The team then completed two test series for each type of secondary shot. The results of each test series are shown in Table 7. The series with close relative spacing were aimed at 0.5 inches from the relevant primary shot. The exception being if that distance would place the shot outside the zone of delamination. This resulted in the Close *C _{P}* shots having an average of 0.49 inches separation from the primary shot and the Close

*P*shots having an average of 0.43 inches separation from the primary shot. The series with medium relative spacing were aimed 1 inch from the edge of the zone of delamination from the relevant primary shot, with an exception being if that distance would place the shot closer to the zone of delamination from a different primary shot. This resulted in the Medium

_{P}*C*shots having an average of 1.24 inches separation from the primary shot and the Medium

_{P}*P*shots having an average of 1.41 inches from the primary shot. While this did not meet the standard for separation from MIL-STD-662F, further separation increased the likelihood of a secondary shot interacting with multiple primary shots. This separation led to secondary shots impacting outside the zone of delamination while still having an overlap of damage.

_{P}Just as was done with all clean series, each pair of secondary series were compared using a hypothesis test to determine if each pair could be averaged together or if they should be considered separately when comparing with other groups. The results of this comparison showed that each pair of the Close *C _{P}*, Close

*P*, and Medium

_{P}*C*series could be combined but that the Medium

_{P}*P*series should be considered separately. This result is shown in Table 8.

_{P }These average secondary series were compared to the ballistic limits of the three clean series using the same hypothesis test process described previously. An example of this comparison is shown in Figure 3, where the Close *C _{P} *average is similar enough to Clean Series 1 and Clean Average 2 that it fails to reject the null hypothesis. However, the Close

*C*average rejects the null hypothesis when compared to Clean Average 1, showing a statistically significant increase in secondary

_{P}*V*for that one case. The summarized results of all comparisons between secondary series and clean series are provided in Table 9. As shown in the table, none of the secondary series showed a statistically significant difference compared to all of the clean series.

_{50}### ARAMID FIBER RESULTS AND DISCUSSION

The clean series for aramid fiber consisted of 8 test series, each using 12 shots with each shot placed 4.384 inches apart to ensure it met the far shot definition for fair impact. From the clean series, 32 shots were *C _{P}*’s, and 68 were

*P*’s. This large imbalance of

_{P}*C*’s and

*P*’s added difficulty in performing the secondary shot series. Table 10 outlines the eight clean series

_{P}*V*estimates, standard deviations, and number of shots used to complete the series.

_{50}Primary impact damage in aramid fiber was characterized by relatively large delamination regions (averaging greater than six projectile diameters across). Despite the large delamination regions, the clean series did not have any delamination overlap, adhering to the far shot definition and fair impact. Figure 4 shows an example of the front and rear delamination extent. Notice the medium secondary shots were placed slightly outside the delamination zone while close shots were placed within the zone. Furthermore, Table 11 outlines the average delamination extent of damage on both the front and rear side of the test plate.

One-tailed hypothesis testing was performed for aramid fiber results as well. This analysis of the clean *V _{50} *series justified three average groups. As shown in Table 12, the first, fifth, and sixth series were averaged together as “Clean Average 1”; the second, third, and fourth series were averaged together as “Clean Average 2”; and series seven and eight were averaged together as “Clean Average 3.”

These clean *V _{50} *average estimates were used to determine the impact of the thermoplastic matrix (UAF-472) by comparing data with Hankins, which used a thermoset matrix (AF163) [4]. For aramid fiber, there was an increase in the

*V*value across all shot categories. Because the thermoplastic matrix was the only variable changed between the current research and that of Hankins, it can be concluded that the thermoplastic matrix increases ballistic strength for aramid fiber. On average, there was a 226.36 fps (28.26%) increase in clean shots. Table 13 outlines the

_{50}*V*value increases from the Hankins data [4].

_{50}The team then completed two test series for each type of secondary shot. The results of each test series are shown in Table 14. Due to the low number of *C _{P}*’s from the clean series, the secondary series shot counts were reduced for

*C*test points. These secondary series were halted as soon as the 3-pod algorithm reached Phase I completion (rather than extending the series to the full 12 shots, like the others). The series with close relative spacing were aimed at 0.5 inches from the relevant primary shot. The Close

_{P}*C*shots had an average shot-to-shot distance of 0.61 inches, and the Close

_{P}*C*shots had an average shot-to-shot distance of 0.57 inches. Each shot for the medium series was placed outside the delamination region, positioned approximately 0.5 inches from the edge of the previous delamination region. If the delamination region was too large, the secondary shot was placed closer to avoid interference from other delamination regions. The Medium

_{P}*C*had an average shot-to- shot distance of 1.49 inches, and the Medium

_{P}*P*shot had an average shot-to-shot distance of 1.53 inches.

_{P}Just as was done with all clean series, each pair of secondary series were compared to determine if each pair could be averaged together or if they should be considered separately. The results of this comparison (given in Table 15) showed that each pair of the Close *C _{P} *and Medium

*P*series could be combined but that the Close

_{P}*P*and Medium

_{P}*C*series should be considered separately.

_{P}These average secondary series were compared to the ballistic limits of the three clean series to identify statistically significant differences. An example of this comparison is shown in Figure 5, where the Close *C _{P} *demonstrated a significant increase for all clean averages. The Close

*C*

_{P}*V*value increased 113.66 fps (11.11%), 142.5 fps (14.33%), and 72.12 fps (6.77%) when compared with Clean Average 1, 2, and 3, respectively. This increase can be seen by the far-right Close

_{50}*C*average normal distribution shown in the figure. The summarized results of all comparisons between secondary series and clean series are provided in Table 16. The secondary shots for aramid fiber showed a significant increase in

_{P}*V*when compared to the clean shots. All

_{50}*α*values of secondary shots compared with clean shot averages were functionally 0. This means that every secondary shot series was significantly different from the clean series shots.

### CONCLUSIONS

In summary, the research discussed herein conducted ballistic strength testing on the composite materials 8HS satin weave S-glass fiber and plain weave aramid (Kevlar KM2 600 Denier) fiber. The test material matrix was UAF-472 thermoplastic fabric weave matrix for both fiber materials. The goal was to observe the effect of the change in matrix on the *V _{50} *ballistic limit velocity of the composite panels. The effect is determined by (1) comparing the

*V*of primary shots to those of previous research, and (2) identifying if any difference was present between the

_{50}*V*of primary and secondary shots on the same material. Understanding such characteristics of composite armor enables better designs and improved aircraft survivability.

_{50}The glass fiber satin weave panels with the thermoplastic UAF-472 matrix showed a statistically significant (7−10%) increase in primary *V _{50} *ballistic limit velocity over the same fiber and weave with the thermoset AF163 matrix tested by Hankins [4]. However, statistical analysis of secondary

*V*estimates showed no clearly significant difference for the glass fiber panels. Any difference observed was within the variation of primary

_{50}*V*estimates.

_{50}The aramid fiber plain weave panels with the thermoplastic UAF-472 matrix showed a statistically significant (24−32%) increase in primary *V _{50} *ballistic limit velocity over the same fiber and weave with the thermoset AF163 matrix tested by Hankins [4]. Additionally, statistical analysis of secondary

*V*estimates showed a significant increase for the glass fiber panels for all combinations of close or medium shot-to-shot spacing from either partial or complete primary shots. No such difference of secondary

_{50}*V*was observed for the same aramid fiber plain weave panels with AF163 matrix tested by Hankins [4].

_{50}### ABOUT THE AUTHORS

Lt. Jack Morgan is currently a student in undergraduate pilot training at Vance AFB, OK. He holds a bachelor’s and a master’s degree in aeronautical engineering from the U.S. Air Force Academy and AFIT, respectively.

Lt. Alex Ramsperger is currently a student in undergraduate pilot training at Vance AFB, OK. His previous research includes a sixth-generation fighter concept design and a maneuverable hypersonic reentry vehicle design. He holds a bachelor’s and a master’s degree in aeronautical engineering from the U.S. Air Force Academy and AFIT, respectively.

Maj. John Hansen is an assistant professor at AFIT and specializes in aircraft combat survivability research. His research and development experience has ranged from the Air Force Research Laboratory to the Japanese Technical Research & Development Institute to flight testing the AC-130 and CV-22 in the Air Force Test Center. He holds a bachelor’s degree, two master’s degrees, and a doctorate in engineering from Brigham Young University, AFIT, the Air Force Test Pilot School, and the University of Michigan, respectively.

**References**

**[1]** U.S. Department of Defense. “*V**50* Ballistic Test for Armor.” Department of Defense Test Method Standard MIL-STD-662F, 18 December 1997.

**[2] ** Kinsler, R., and J. Collins. “Comparison of *V**50* Shot Placement on Final.” *Proceedings* *Personal Armour Systems Symposium *(PASS), https://doi.org/10.13140/2.1.4269.8567, 2012.

**[3] ** Keane, M. P. “Ballistic Limit Shot Dependency Testing in Composite Materials.” Master’s thesis, Air Force Institute of Technology, Wright-Patterson AFB, OH, 2019.

**[4] ** Hankins, C. C. “The Effect of Shot Dependency on Composite Materials Subject to Ballistic Testing.” Master’s thesis, Air Force Institute of Technology, Wright-Patterson AFB, OH, 2021.

**[5]** Backman, E., and W. Goldsmith. “The Mechanics of Penetration of Projectiles Into Targets.” *International Journal of Engineering* *Science*, vol. 16, issue 1, https://doi.org/10.1016/0020-7225(78)90002-2, 1978.

**[6]** Ball, R. E. *The Fundamentals of Aircraft Combat* *Survivability Analysis and Design*. Second edition, American Institute of Aeronautics and Astronautics,

**[7]** Wu, F. J., and Y. Tian. “Three-Phase Optimal Design of Sensitivity Experiments.” *Journal of* *Statistical** Planning and Inference*, vol. 149, https://doi.org/10.1016/j.jspi.2013.10.007, 2014.

**[8]** 3M Company. “AF 163-2 Data Page.” St. Paul, MN, 2009.

**[9]** Adhesive Films Inc. “Product Data UAF-472.” Pine Brook, NJ, 2022.

**[10] ** Martinez-Rubi, Y., B. Ashrafi, M. B. Jakubinek, S. Zou, K. Laqua, M. Barnes, and B. Simard. “Fabrication of High Content Carbon Nanotube-Polyurethane Sheets With Tailorable Properties.” *ACS** Applied Materials & Interfaces*, vol. 9, no. 36, https://doi.org/10.1021/acsami.7b09208, 2017.

**Acknowledgments**

The authors would like to thank the Joint Aircraft Survivability Program Office for its support in funding this investigation, TenCate Advanced Armors for supplying the test samples, and all of those who helped complete this research.

**Note:**

**Note:**

*A version of this article was presented at the American Institute of Aeronautics and Astronautics (AIAA) SciTech Forum in January 2023.*