Impact Testing and Simulation of a Crashworthy Composite Fuselage Karen E. Jackson and Edwin L. Fasanella US Army Research Laboratory, Vehicle Technology Directorate
[email protected] [email protected] Hampton, VA
Abstract A composite fuselage concept for light aircraft and rotorcraft has been developed to provide improved crash protection. The fuselage consists of a relatively rigid upper section, or passenger cabin, including a stiff structural floor and a frangible lower section which encloses the crash energy management structure. A 60-in. diameter full-scale fuselage section was manufactured using a composite sandwich construction. Vertical drop tests were conducted at both 0°- and 15°-roll impact attitudes to evaluate the crashworthy features of the fuselage design. The experimental data are correlated with predictions from a finite element model developed using the nonlinear, explicit transient dynamic code, MSC/DYTRAN. Finally, the full-scale impact data are compared with scaled data obtained from similar impact tests of a 1/5-scale model fuselage. This comparison was made to investigate the existence of scaling effects and to demonstrate the application of scale model technology in the development of a composite structural concept. Introduction The present paper describes the development and evaluation of an innovative and costeffective composite fuselage concept for light aircraft and rotorcraft that is designed to meet structural and flight loads requirements and to provide improved crash protection. The two primary design goals for crashworthiness are to limit the impact forces transmitted to the occupants, and to maintain the structural integrity of the fuselage to ensure a minimum safe occupant volume [1, 2]. To meet these objectives, an aircraft or rotorcraft fuselage must be designed for high stiffness and strength to prevent structural collapse during a crash. Yet, the fuselage design must not ________________________ Presented at the American Helicopter Society 56th Annual Forum, Virginia Beach, Virginia, May 2-4, 2000. This paper is a work of the U. S. Government and is therefore in the public domain.
be so stiff that it transmits or amplifies high impact loads to the occupants. Ideally, the design should contain some crushable elements to help limit the loads transmitted to the occupant to survivable or non-injurious levels [3]. The fuselage concept, shown in Figure 1, consists of four different structural regions, each with its own specific design objectives. The upper section of the fuselage cabin is fabricated of a stiff composite sandwich construction and is designed to provide a protective shell enclosing the occupants in the event of a crash. The frangible outer shell is fabricated from a relatively compliant composite material that is wrapped around the upper fuselage section, enclosing the energy absorbing structure beneath the floor, and forming the lower fuselage. The outer shell is designed to provide damage tolerance and aerodynamic shape. Upon impact, the outer shell is intended to deform plastically and to initiate crushing of the subfloor. The energy absorbing subfloor is designed to dissipate kinetic energy through stable crushing. Finally, a key feature of the fuselage concept is the rigid structural floor. The structural floor is designed to react the loads generated by crushing of the subfloor and to provide a stable platform for seat and restraint attachment. The anticipated benefits of the energy absorbing fuselage design over conventional and retrofit fuselage designs include a substantially lower floor-level acceleration pulse, thus permitting lower occupant loads while maintaining cabin integrity during a crash. In addition, the energy absorbing fuselage design provides improved seat and restraint attachment due to the presence of the load-bearing floor. Finally, the crash effectiveness of the fuselage design is much less dependent on the mass of an individual occupant, since the combined masses of all occupants, seats, and the upper portion of the aircraft itself are reacted by the subfloor beams during an impact. These benefits are described in more detail in References 4 and 5.
In 1997, a three-year research program was initiated at NASA Langley Research Center to develop the fuselage concept for potential application to light aircraft and rotorcraft. The fuselage concept was designed and evaluated during the first two years of the research program through fabrication and testing of a 1/5-scale model fuselage [6-10]. During the third year of the research program, a full-scale prototype was fabricated by “scaling up” the geometry and constitutive properties of the 1/5-scale model fuselage. The scaling parameters used in the design and testing of the fuselage concept are shown in Table 1. Comparisons of the 1/5- and full-scale experimental data were made to investigate the existence of scaling effects and to demonstrate the application of scale model technology in the development of a composite structural concept. One of the potential benefits of scale model testing is a significant reduction in the cost of fabrication of the test article. Rigid structural Stiff upper floor fuselage section
Frangible outer shell
Energy absorbing subfloor
Figure 1. Schematic drawing of the fuselage concept. Both structural and impact design requirements were defined for the fuselage concept. The structural design goal for the 1/5-scale model fuselage was to maintain floor rigidity (less than 0.1 inch of floor mid-point deflection) for a 10-psi internal pressure load. This goal was satisfied during the first-year of the research program [6], and the design of the upper section and floor of the 1/5-scale model fuselage was completed. The design goal for crash protection was to limit occupant loads to survivable levels for a 372-in/s vertical impact onto a rigid surface. The 372-in/s vertical velocity requirement is more severe than current regulatory criteria for small aircraft, but it is a realistic, potentially survivable, impact velocity ob-
served in actual crashes and in crash tests conducted at NASA Langley Research Center [1114]. For the 1/5-scale model fuselage, the specific impact requirement was to achieve an average floor-level acceleration of 125-g for the 372in/s vertical impact condition. The average acceleration was chosen for the impact requirement, instead of the peak acceleration, because it provides a physically meaningful measure of the acceleration pulse that is useful in a kinematic evaluation. The impact goal was achieved successfully during the second year of the research program and the design of the 1/5-scale model fuselage was finalized [8-10]. The corresponding impact requirement for the full-scale fuselage section was to achieve an average floor-level acceleration of 25-g for a 372-in/s vertical impact velocity. A secondary requirement was to demonstrate a high level of crash protection for an off-axis (15°-roll) impact at the same velocity. Vertical drop tests of the fullscale fuselage were performed at 0°- and 15°-roll impact attitudes under scaled conditions to verify compliance with the impact design requirements. Also, a finite element model of the full-scale fuselage section was developed using the nonlinear, explicit transient dynamic code, MSC/DYTRAN [15]. Analytical results obtained from the MSC/ DYTRAN simulation were correlated with experimental data obtained from both the 0°- and 15°-roll vertical drop tests. Table 1. Summary of scaling parameters (λ=1/5). Parameter
FullScale
1/5Scale
Scale Factor
Diameter
60 in.
12 in.
λ
Length
60 in.
12 in.
λ
Impact velocity
372 in/s
372 in/s
1
Weight of lead plate Floor loading per fuselage length Avg. subfloor crush stress
1,500 lb.
12 lb.
λ3
25 lb/in. (300 lb/ft)
1 lb/in. (12 lb/ft)
λ2
15 psi
15 psi
1
Pulse duration
38.5 ms
7.7 ms
λ
Floor-level acceleration
25 g
125 g
1/λ
The focus of the present paper is to describe the results from the third year of the research program, including the fabrication and impact testing of the full-scale fuselage section; the correlation of experimental data with analytical predictions from a MSC/DYTRAN crash simulation; and, comparisons of the experimental data obtained from 0°- and 15°-roll vertical drop tests of the 1/5- and full-scale fuselage sections. Full-Scale Fuselage Design and Fabrication Certain geometric and inertial parameters had to be selected before the fuselage concept could be sized and designed. For this study, the fuselage design is based on a full-scale aircraft with a diameter of 60 inches, and a floor load distribution of 25 pounds per linear inch of fuselage length. A schematic drawing of the full-scale fuselage design configuration is shown in Figure 2. The upper section of the fuselage is fabricated using a composite sandwich construction with a 0.00174-lb/in3 closed-cell polyurethane foam core and 0°/90° E-glass/epoxy fabric face sheets. Eglass/epoxy composite material was chosen because of its lower cost and wider use by the light aircraft industry. In addition, a room temperature cure epoxy was selected, thus eliminating the need for a more expensive autoclave cure. The composite sandwich construction in the floor of [ 0°G / 0°E / 45°E ]5
the fuselage consists of an 0.00463-lb/in3 closedcell polyurethane foam core with face sheets consisting of layers of E-glass/epoxy and graphite/epoxy composite fabric. The layers of graphite/epoxy fabric and the higher density foam were used for increased stiffness and improved structural rigidity of the floor. Several energy absorbing subfloor configurations were evaluated for incorporation into the fuselage section. A geometric foam-block design, consisting of five uniformly-spaced, individual blocks of a crushable Rohacell 31-IG closed-cell foam overlaid with E-glass/epoxy face sheets was selected. Each block of foam was 6.5 inches in depth. The cross-sectional geometry of the foam blocks, shown in Figure 2, was chosen to achieve a fairly uniform crushing force. The foam-block subfloor design was selected because it produced an average crushing stress close to the 15-psi design requirement during quasi-static and dynamic compression tests. The required subfloor crushing stress was determined from the parameters of the impact experiment. For a vertical impact of a full-scale fuselage having a floor loading per unit length of 25 lb/in. and a length of 60 inches, a sustained subfloor crushing load of 37,500 lb. would result in a constant 25-g deceleration. This 37,500-lb. load corresponds to a subfloor crushing stress of 15 psi, given an approximate floor area of 2,500 in2. [ 0°E / 45°E / 45°E ]5 [ 0°E ]5
[ 0°G / 0°E ]5
.00174 lb/in3 foam (1.0-in. thick)
[ 45°E ]5 [ 0°E ]5
.00463 lb/in3 foam (2.0-in. thick) .00162 lb/in3 foam (Rohacell 31-IG)
45°E = ±45° E-glass/epoxy fabric (0.004-in. per layer) 0°E = 0°/90° E-glass/epoxy fabric (0.004-in. per layer) 0°G = 0°/90° Graphite/epoxy fabric (0.006-in. per layer)
Figure 2. Schematic drawing of the full-scale fuselage design configuration.
Based on kinematics, a crushing distance of 7.16 inches is needed to dissipate the kinetic energy. Given that the maximum height of the subfloor is 8.52-inches, a crush stroke of approximately 84% is expected. The development and evaluation of the Rohacell foam-block subfloor design, as well as other subfloor configurations, are described in References 7 and 8. A full-scale fuselage section was fabricated according to the design shown in Figure 2. A plywood mandrel, shown in Figure 3, was used to construct the upper section and floor of the fuselage section. The inner face sheets, foam core, and outer face sheets were laid up sequentially by hand, then allowed to cure at room temperature. Photographs of the upper section and floor at various stages of fabrication are shown in Figure 4. The mandrel was built with a slight taper to allow easy removal of the upper section and floor. Next, five Rohacell 31-IG foam blocks were machined to the cross-sectional geometry shown in Figure 2. The inner surfaces of the foam blocks were overlaid with E-glass/epoxy face sheets. Then, the foam blocks were attached to the bottom of the floor using epoxy and allowed to cure at room temperature. Finally, the frangible outer layer of ±45° E-glass/epoxy fabric was wrapped around the foam blocks and attached to the upper section. The upper section and floor weighed 156 lbs. and the fuselage section with subfloor weighed 180 lbs. The completed fuselage was 60-in. in length with a diameter of 60 inches.
Plywood 3/32" Plywood cover
Experimental Program Test Set-up Photographs of the fuselage section prior to the 0°- and 15°-roll vertical drop tests are shown in Figure 5. A 1,574-lb. lead plate was mounted to the floor to represent the inertia of seats, occupants and other masses. In the original plan for the full-scale test, a 1,500-lb. lead plate was supposed to be used to obtain a floor loading of 25 lb/in. However, a heavier lead plate was needed to achieve the correct mass scaling with the 1/5scale model test. An explanation for this weight difference is given in the Scaling Effects section of the present paper.
6' 72-in.
Plywood ribs
Cover
3/4" Plywood Ribs
Figure 4. Photographs showing the fuselage at different stages of fabrication.
5' 60-in.
Figure 3. Schematic drawing of the mandrel used to fabricate the full-scale fuselage section.
Four lifting brackets were attached to the upper section of the fuselage, two on either side, as shown in Figure 5. For the 0°-roll impact, four cables of equal length were attached to the lifting brackets and were secured to an A-frame support on the drop tower. For the off-axis impact condition, the cable lengths were adjusted to achieve the desired 15°-roll attitude. For the 0°-roll impact test, the fuselage section was instrumented with
front and rear accelerometers, which were secured to the lead plate along its centerline to record the vertical acceleration response. For the 15°-roll impact, the accelerometers were placed at the midpoint of the lead plate, one on the right side and one on the left side of the centerline. Data were collected at 10,000 samples per second using a PC-based digital data acquisition system. For both impact conditions, the fuselage section was raised to a height of 180 inches and dropped onto a rigid surface to achieve a 372-in/s vertical velocity at impact.
based on the time delay, the velocity at impact, and the distance between the two accelerometers. Also, it should be noted that the response recorded by the front accelerometer is clipped at 70-g. Consequently, only the rear acceleration data is analyzed for this test. An average acceleration of 29.9-g was calculated by determining the area under the acceleration response curve and dividing by the pulse duration of 36.9 ms. This value of average acceleration is 19.6% higher than the 25-g impact design requirement. Several factors may have contributed to this discrepancy including some deviations from the test condition specified by the design requirement. These factors will be discussed in more detail in the Scaling Effects section of the present paper. The acceleration response was integrated to obtain the velocity time history shown in Figure 6(b). This plot indicates that a rebound occurred due to the release of stored energy within the fuselage section. The fuselage section rebounded to a height of approximately 5 inches following the initial impact event. The fuselage section came to a rest following the secondary impact, which occurred at about 390 ms. The velocity response was integrated to obtain the displacement time history, shown in Figure 6(c). A maximum displacement of 7.9 inches occurs at approximately 31 ms. This value was not verified by motion picture analysis and may include some elastic deformation of the floor in addition to actual displacement due to crushing of the subfloor. The maximum distance available for crushing is 8.52 inches, which is the height of the subfloor at the centerline. An estimated crush stroke of 92.7% is obtained by dividing the maximum crush distance by the subfloor height.
Figure 5. Photographs of the fuselage prior to the 0°-roll (top) and 15°-roll (bottom) impact tests. 0°-Roll Vertical Drop Test Results The acceleration responses obtained from the front and rear accelerometers that were mounted to the lead plate on the floor of the fullscale fuselage are shown in Figure 6(a). A time difference of approximately 1.5-ms is observed between the two acceleration curves, indicating that the fuselage section did not experience a flat impact. A pitch angle of about 1° is estimated
A sequence of photographs taken from the video footage is shown in Figure 7 illustrating the fuselage deformation during the impact event. Some ovalizing of the upper section of the fuselage occurred during the impact. The amount of deformation observed may be attributed to the fact that an open fuselage section was tested. It is expected that this deformation would be significantly reduced for an actual aircraft having a closed fuselage. Minor damage to the fuselage was observed in the transition region between the upper fuselage section and floor on the right side of the fuselage. The damage consisted of a disbond between the inner face sheet and foam core. Another sequence of photographs is shown in Figure 8 highlighting the crushing and
Acceleration, g 100
Rear accel. Front accel.
80 60 40 20
failure response of the subfloor. As shown in the middle photograph of Figure 8, the E-glass/epoxy face sheets are delaminated from the foam blocks as crushing begins. The foam blocks are then compressed and failed. Following the test, the crushed subfloor was removed and replaced with a new subfloor of the same design prior to the 15°roll vertical drop test; however, the minor damage to the section was not repaired.
0 -20 -40 0
10
20
30
40
50
Time, ms (a) Acceleration versus time response. Velocity, in/s 100
0
-100
-200
-300
-400
0
10
20
30
40
50
Time, ms (b) Velocity versus time response.
Displacement, in. 2
0
-2
-4
Figure 7. Sequence of photographs depicting fuselage deformation during impact.
-6
-8
0
10
20
30
40
50
Time, ms (c) Displacement versus time response.
Figure 6. Experimental data from 0°-roll drop test.
15°-Roll Vertical Drop Test Results Plots of the acceleration time histories obtained from the 15°-roll vertical drop test are shown in Figure 9. The experimental responses were obtained from the accelerometers located on the right and left sides of the lead plate. Due to the rolled impact attitude, the measured acceleration responses represent the component of the
acceleration that is normal to the floor, which is rotated 15° from the vertical direction. Another component parallel to the floor is also present, but was not measured in the experiment. The acceleration response measured by the right accelerometer, which is closer to the point of impact, exhibits a higher magnitude and shorter pulse duration than the acceleration response measured by the left accelerometer for a 15°-roll impact attitude. The response measured by the right accelerometer has an average acceleration of 28.6-g for a pulse duration of 40 ms. The response measured by the left accelerometer has an average acceleration of 19.4-g and a pulse duration of 56 ms. Post-test photographs of the fuselage section are shown in Figure 10. During the test, no observable rebound of the fuselage occurred. However, the fuselage rotated upon impact, due to the initial 15°-roll attitude at impact. Damage to the fuselage occurred in the transition region between the curved upper section and the floor on both sides; however, the damage on the right side was much more severe. Upon impact, a large disbond between the inner face sheets and the foam core developed and grew upward toward the lifting brackets on the right side of the fuselage. The lifting brackets appeared to arrest the growth of the disbond. Also, some large pieces of the foam core in the transition region on the right side were fragmented. Similar damage, though to a much lesser extent, was observed on the left side of the fuselage. It is important to note, however, that even for this severe impact condition, the fuselage maintained cabin integrity to provide a livable volume space for occupants.
Acceleration, g 100
Left accelerometer
80 60 40 20 0 -20
0
20
40
60
80
Time, ms Acceleration, g 100
Right accelerometer 80 60 40 20 0 -20
0
20
40
60
80
Time, ms Figure 9. Left and right acceleration responses from the 15°-roll impact test. MSC/DYTRAN Crash Simulation and Experimental Correlation
Figure 8. Sequence of photographs highlighting the subfloor crushing process.
Model Development A detailed three-dimensional finite element model of the full-scale fuselage was developed using MSC/DYTRAN [15]. MSC/DYTRAN is a commercially available, nonlinear explicit transient dynamic finite element code, marketed by MSC Software Company. The complete undeformed model, shown in Figure 11(a), consists of 14,992 nodes, 18,240 elements, and 60 concentrated masses. The inner and outer face sheets of the upper section and floor are modeled using CQUAD4 shell elements, and the foam core in the upper section, floor, and subfloor is represented by CHEXA solid elements. The material proper-
ties of the 0°/90° and ±45° E-glass/epoxy fabric material were determined from coupon tests and are modeled using a linear elastic-plastic material model with strain hardening. The 0.00174- and 0.00463-lb/in3 polyurethane foam cores in the upper section and floor are modeled as DMATEL linear elastic solid materials. The foam material properties were obtained from crushing tests of individual blocks of foam, without face sheets. The more complicated multi-layered face sheets in the floor are modeled as laminated composite materials using the PCOMP feature in MSC/ DYTRAN. The specific material properties used in the model are provided in Table 2.
trated masses, each weighing 26.23 lb., are distributed in a centralized rectangular region on the floor to represent the inertial properties of the lead plate. The entire model weighed 1,754.3 lb., which is close to the actual 1,769 lb. weight of the fuselage section. A master-surface to slave-node contact is defined between the subfloor and the impact surface. The impact surface is modeled as a 12-in.thick plate with all of the edge nodes fixed. For the 0°-roll impact simulation, an initial vertical velocity of 372 in/s is assigned to all elements in the model except for those elements forming the impact surface. To represent the orientation of the fuselage section at contact, the impact surface was pitched by 1° (front end down) while the orientation of the structural model was unchanged.
(a) Full model.
Figure 10. Photographs of the fuselage section after the 15°-roll drop test. The Rohacell foam blocks, which are located in the subfloor region of the MSC/DYTRAN model, are shown in Figure 11(b). The five 6.5in.-deep Rohacell 31-IG foam blocks are represented using DYMAT24 solid elements with properties of a linear elastic, plastic material with strain hardening and an ultimate failure strain of 80%. The 0°/90° E-glass/epoxy face sheets on the foam blocks in the subfloor are represented as DMATEP shell elements with linear elastic material properties up to a yield stress of 12,000 psi with strain hardening to ultimate failure. Sixty concen-
(b) Foam blocks located in the subfloor. Figure 11. Undeformed model of the full-scale fuselage section.
Table 2. Material property data used in the MSC/DYTRAN model of the fuselage section. Formulation
ρ (lb-s2/in 4)
E (psi)
ν
Aluminum
DYMAT24
2.65e-4
10.e6
.33
55,000
±45° E-glass
DMATEP
1.73e-4
1.5e6
.49
9,000
117,650
0°/90° E-glass
DMATEP
1.73e-4
2.75e6
.11
12,000
117,650
Foam 3 lb/ft3
DMATEL
4.5e-6
1,300
650
Foam 8 lb/ft3
DMATEL
1.2e-5
8,000
3,200
Graphite/epoxy
DMATEP
1.45e-4
9.1e6
.06
Rohacell foam
DYMAT24
4.2e-6
2,000
0.3
90.
54.
0.8
0°/90° E-glass w/failure
DMATEP
1.73e-4
2.75e6
.11
12,000
117,650
.001
Material
A similar crash simulation was performed to predict the acceleration response of the fullscale fuselage during the 15°-roll drop test using MSC/DYTRAN. The undeformed model, shown in Figure 12, is the same model that was used to perform the 0°-roll impact simulation. However, some modifications were made to account for the 15°-roll impact attitude. In the experiment, the fuselage section was rolled by 15° and dropped vertically onto a flat impact surface. However, for the analysis, it was more expedient to rotate the impact surface by 15°, than to rotate the fuselage model. As a result of using this approach, it was necessary to change the initial condition from a pure vertical velocity of 372 in/s to a velocity vector with a horizontal component of 96.3 in/s and a vertical component of 359.3 in/s. Simulation Results for the 0°-Roll Vertical Drop Test The MSC/DYTRAN-predicted acceleration response is correlated with the rear accelerometer data from the vertical drop test of the fullscale fuselage section in Figure 13(a). In general, the simulation predicts the overall shape and magnitude of the experimental curve well. However, the analysis under predicted the magnitude of the peak acceleration that occurred at about 28 ms in the experimental response. The average acceleration of the predicted response is 36.8-g, which is 23% higher than the experimental value of 29.9-g. The pulse duration of the predicted response is 35 ms, which is close to the experimental pulse duration of 36.9 ms.
G (psi)
σy (psi)
Eh (psi)
ε psf (in/in)
Figure 12. Front view of the MSC/DYTRAN model used to simulate the 15°-roll impact test. The predicted and experimental velocity time histories are plotted in Figure 13(b). The plot indicates that the simulation is dissipating or removing energy at a faster rate than the experiment, causing the model to stop sooner than the actual fuselage. For example, at any given time, the velocity of the experimental response is lower (more negative) than that of the predicted response. The predicted and experimental displacement responses are shown in Figure 13(c). The maximum displacement predicted by the MSC/DYTRAN simulation is 6.5 inches, compared to a maximum displacement of 7.9 inches observed in the experiment.
Acceleration, g Experiment MSC/DYTRAN
100 80 60 40 20 0 -20 -40 0
10
20
30
40
50
Time, ms (a) Acceleration response. Velocity, in/s 200
Experiment MSC/DYTRAN
100 0 -100 -200 -300 -400
0
10
20
30
40
50
Time, ms
(b) Velocity response. Displacement, in. 2
Experiment MSC/DYTRAN
0
-2
-4
-6
-8 0
10
20
30
40
50
Time, ms (c) Displacement response.
Figure 13. Analytical/experimental correlation.
Simulation Results for the 15°-Roll Vertical Drop Test The predicted acceleration responses are correlated with the experimental data obtained from the right and left accelerometers during the 15°-roll vertical drop test of the full-scale fuselage section in Figure 14. The MSC/DYTRAN predictions were obtained from nodes located on the floor at the approximate locations of the two accelerometers. For the right accelerometer location, the MSC/DYTRAN simulation predicted a large spike in the acceleration response with a peak acceleration of 135-g occurring at 16 ms, as shown in Figure 14(a). The spike in the predicted acceleration response has a much higher magnitude and longer duration than the experimental response. Following the initial spike, the MSC/ DYTRAN response closely matches the overall magnitude and shape of the experimental curve. The average acceleration of the MSC/DYTRAN response is 37.2-g with a pulse duration of 42 ms. The predicted average acceleration is 30% higher than the experimental value of 28.6-g; however, the pulse duration of both responses is approximately 40 ms. For the left accelerometer location, the predicted acceleration response, shown in Figure 14(b), closely matches the general shape and magnitude of the experimental curve. The average acceleration of the MSC/DYTRANpredicted response is 19.5-g for a pulse duration of 52.8 ms. The correlation with the experimental data is excellent, given that the average acceleration of the experimental curve is 19.4-g for a pulse duration is 56 ms. Modeling Inaccuracies The discrepancies between the predicted and experimental acceleration responses for the 0°-roll drop test may be attributed to inaccurate modeling of the subfloor crushing process and other deficiencies in the model. The material properties of the Rohacell foam used in the model were obtained from compression tests on individual cubic blocks of foam. Consequently, the material properties represent the compressive response and failure of the material only. During the initial impact, the Rohacell foam blocks were subjected to a more complex loading scenario, including bending and shear. It is expected that the Rohacell foam would fail at much lower loads when subjected to shear and flexure. However, material property data were not available for the Rohacell foam for these loading conditions. Another deficiency in the model is that it did not allow disbonding of the face sheets from the Rohacell foam in the subfloor. Because of these factors, the subfloor in the model had a relatively stiffer
response than that of the actual subfloor causing the model to dissipate more energy initially than the test, as indicated in Figure 13(a). In addition, the material model for the Rohacell foam in the MSC/DYTRAN simulation did not represent the stiffening effect that occurs for large compressive strains, as the foam becomes compacted. Based on compression test results, the foam begins to compact at 70-80% stroke. Some portions of the subfloor were compacted since the full-scale fuselage section exhibited a crush stroke of greater than 90%. Once the foam becomes compacted, the loads reacted by the floor increase dramatically. Since the simulation did not account for the effects of foam compaction, the actual subfloor had a relatively stiffer response than the model near the end of the pulse.
One explanation for the over prediction of the magnitude of the right side acceleration response during the 15°-roll drop test, shown in Figure 14(a), is that the pre-existing damage was not incorporated into the MSC/DYTRAN model. Also, the model did not allow for a disbond between the inner face sheets and the foam core and no failure properties were assigned to the 0.00174- and 0.00463-lb/in 3 foam in the upper section and floor, respectively. Thus, the model exhibited a relatively stiffer response with a higher magnitude peak acceleration than did the experiment. However, since relatively minor damage occurred on the left side of the fuselage during the 15°-roll drop test, the deficiencies in the model for predicting disbonding and foam failures were less important and a high level of correlation was obtained.
Acceleration, g 150
Scaling Effects Experiment MSC/DYTRAN
100
50
0
-50
0
10
20
30
40
50
60
70
80
Time, ms
(a) Right accelerometer. Acceleration, g 60
Experiment MSC/DYTRAN
40
20
0
-20
-40
0
10
20
30
40
50
60
70
80
Time, ms
(b) Left accelerometer. Figure 14. Predicted and experimental acceleration responses for the 15°-roll impact test.
Comparison of Acceleration Responses As mentioned in the introduction of the present paper, the fuselage concept was originally designed and validated as a 1/5-scale model. The full-scale fuselage is a geometric and constitutive replica of the 1/5-scale model design that was completed at the end of the second year of the research program. Thus, by scaling the experimental data obtained from the vertical impact tests of the 1/5-scale model fuselage and comparing with the full-scale data, it is possible to investigate potential scaling effects and to determine the usefulness of scale model technology in the development of structural concepts. The acceleration response obtained from the 0°-roll vertical drop test of the full-scale fuselage section is plotted with the “scaled up” acceleration response from the vertical drop test of the 1/5-scale model fuselage section in Figure 15. The scaled response, shown in Figure 15, was obtained by multiplying the experimental acceleration data for the 1/5-scale model by the scale factor, λ, and by multiplying the corresponding time by the scale factor, 1/λ, where λ equals 1/5. In general, the scaled acceleration response has the same magnitude and overall shape as that of the full-scale data for the first 22 ms of the pulse. However, the scaled response does not exhibit the large peak acceleration that occurs at 28-ms in the full-scale response. The average acceleration of the 1/5-scale model response is 25.4-g, which is 15% lower than the 29.9-g average acceleration of the full-scale data. However, the pulse durations of both responses are nearly identical.
Using the same approach, the acceleration responses obtained from the 15°-roll vertical drop tests of the 1/5- and full-scale fuselage sections are shown in Figure 16 for both the left and right accelerometers. In general, the 1/5-scale model test data predicts the magnitude and shape of the full-scale responses well. The right accelerometer over ranged during the 1/5-scale model drop test. However, the data is included to show the level of agreement prior to and following that portion of the response that is missing. Comparisons of the average accelerations and pulse durations for the 1/5- and full-scale data are shown in Table 3. The minor discrepancies between acceleration responses of the 1/5- and full-scale fuselage sections may be attributed to some scaling anomalies resulting from the testing program, as explained in the Scaling Anomalies subsection of the present paper.
Acceleration, g
inches. The maximum crush obtained from the full-scale test data was 7.9 inches. The estimated value of maximum crush obtained from the 1/5scale model data is only 3.5% lower than the actual full-scale value. Thus, no scaling effects in the subfloor crushing behavior were evident.
Acceleration, g 100
Full-scale data 1/5-scale data
80 60 40 20 0
Full-scale data 1/5-scale data
100
-20 0
80
10
20
30
40
50
60
70
80
Time, ms 60
(a) Left accelerometer.
40
Acceleration, g 100
20 0
80
-20
60
-40
0
10
20
30
40
50
Time, ms Figure 15. Acceleration responses from the 0°-roll drop tests of the 1/5- and full-scale fuselages. Comparison of Damage Levels and Subfloor Crushing Response Comparisons of the damage levels sustained by the 1/5- and full-scale fuselage sections are also indicators of the presence of scaling effects. For the 0°-roll vertical drop test, the upper section and floor of the 1/5-scale model fuselage showed no damage by visual inspection. The fullscale fuselage exhibited only minor damage in the transition region between the curved upper section and the floor. During the 1/5-scale model drop test, a maximum crush distance of 1.525inches was determined by double integration of the acceleration response. This value, multiplied by the scale factor 1/λ, or 5, gives an estimated full-scale maximum crush distance of 7.625
Full-scale data 1/5-scale data
40 20 0 -20
0
10
20
30
40
50
60
70
80
Time, ms (b) Right accelerometer. Figure 16. Acceleration responses from the 15°roll drop tests of the 1/5- and full-scale fuselages. A more dramatic difference in the level of damage sustained by the 1/5- and full-scale fuselage sections was observed for the 15°-roll impact condition. The 1/5-scale model fuselage exhibited minor damage in the form of a disbond in the transition region on the right side of the fuselage.
Table 3. A comparison of data obtained for the 1/5- and full-scale fuselage vertical drop tests. Average acceleration, g
Pulse duration, ms
Test 1/5-Scale
Full-scale
1/5-Scale
Full-scale
0°-Roll
25.4
29.9
38.5
36.9
15°-Roll (left accel.)
18.6
19.4
44
56
15°-Roll (right accel.)
N/A
28.6
28.5
40
Also, some fragmentation and crushing of the foam occurred near the disbond. In general, the damage to the 1/5-scale model fuselage was localized to the relatively small transition region on the right side of the fuselage. The damage to the full-scale fuselage was more severe with a large disbond that developed in the right side transition region and grew upward toward the lifting bracket. Some large pieces of the foam core were fragmented and broke loose from the section during the test. In addition, similar damage was observed near the left mounting bracket. Due to the extent of the damage, the upper section of the full-scale fuselage showed significantly more deformation during the impact than did the 1/5-scale model fuselage. Scaling Anomalies An average acceleration of 29.9-g was determined from the acceleration time history obtained from the 0°-roll vertical drop test of the fullscale fuselage section. This value is 19.6% higher than the 25-g design goal. The large difference is significant for the full-scale fuselage section, especially given that the average acceleration obtained from the 1/5-scale model fuselage impact test was only 1.6% higher than the design goal [8-10]. Some scaling anomalies existed that may account for the differences observed in the experimental results obtained from the 1/5- and full-scale fuselage drop tests, including: Mass scaling. The 1/5-scale model fuselage section, without lead weight on the floor, weighed 2.23 lbs. This value multiplied by 1/λ 3, or 125, gives an expected weight of the full-scale fuse-
lage section of 278.75 lbs. However, the actual weight of the full-scale fuselage section, without the lead plate, was 180 lbs. Given that the number of plies for each laminate in the full-scale fuselage was scaled exactly, this difference is attributed to excessive epoxy in the 1/5-scale model fuselage. The mass discrepancy was accounted for by increasing the amount of lead weight placed on the floor of the full-scale fuselage section from 1,500 lb. to 1,574 lb. such that the combined weight of the upper section, floor, and lead plate was exactly 125 times that of the similar combined weight of the 1/5-scale model fuselage section. This approach was taken assuming that the combined weight of the upper section, floor, and lead plate was the important scaling parameter and not the individual weights of each component. However, the weight of the lead plate is roughly ten times the weight of the upper section and floor. As a result, the lead plate provides about 90% of the inertia during the drop test. While the increased weight was necessary to achieve the correct mass scaling, it resulted in a more severe impact condition than that specified by the impact design goal, by producing a floor loading greater than 25-lb/in. Consequently, the full-scale fuselage section experienced a relatively more severe impact than that of the 1/5-scale model fuselage section. Geometric scaling. Based on slight differences in geometry between the 1/5- and full-scale fuselage sections, the actual scale factor was determined to be 1/5.03, not 1/5. However, the scale factor of 1/5 was used for the results in the present paper.
Energy scaling. Typically in a drop test, the total potential energy is equal to the drop height plus the crush stroke distance multiplied by the weight (mass times gravity). The potential energy is converted to kinetic energy minus the total work, including dissipative work performed in crushing of the subfloor and elastic work performed in deforming (ovalizing) the fuselage. According to the scaling law, the scale factor for potential energy is 1/λ 3. Thus, the ratio of total potential energy for the 1/5- and full-scale fuselage drop tests should equal 125. For the full-scale fuselage, the total potential energy is: PE f = Mg(h + δf )
(1)
where PE f is the total potential energy for the fullscale fuselage, M is the total mass of the full-scale fuselage, g is gravity, h is the drop height, and δf is the maximum crush distance for the full-scale fuselage. The total potential energy for the 1/5scale model drop test is given by a similar equation: PE m = mg(h + δm)
(2)
where PE m is the total potential energy for the 1/5scale model fuselage drop test, m is the total mass of the 1/5-scale model fuselage, g is gravity, h is the drop height, and δm is the maximum crush distance for the 1/5-scale model fuselage. The drop height for both the 1/5- and full-scale fuselage sections was 180 inches, as measured from the bottom of the fuselage to the impact surface. The weight of the 1/5-scale model fuselage (mg) is 1/125 of the weight (Mg) of the full-scale fuselage. Also, the crush stroke distances for the 1/5- and full-scale fuselage drop tests are 1.525 and 7.9 inches, respectively. Consequently, the scale factor for potential energy for these experiments is: PE f = 125(180-in. + 7.9-in.) PE m (180-in. + 1.525-in.)
model fuselage test. The relative increase in potential energy translates into increased kinetic energy and total work performed by the full-scale fuselage section. Pre-existing Damage State. One explanation for the difference in the amount of damage sustained by the 1/5- and full-scale fuselage sections for the 15°-roll impact condition is the presence of the existing disbond in the full-scale fuselage from the 0°-roll drop test. The disbond occurred in the right side transition region, which was a critical location for the 15°-roll drop test. As mentioned previously, this damage was not repaired prior to the 15°-roll impact. Likely, the presence of the existing disbond contributed to the severity and extent of the damage by weakening the transition region, and serving as the initiation site for damage growth. Mounting technique. To perform the vertical drop tests of the 1/5-scale model fuselage, a lifting bracket and four tabs were attached to the fuselage section using epoxy. The lifting bracket was located on top of the fuselage at its midpoint. The tabs were located at the top and bottom of the fuselage, two on either end. Guide wires were passed through holes in the tabs to maintain the fuselage orientation during the drop test [8-10]. For the full-scale drop tests, four mounting brackets were bolted to the upper section of the fuselage, requiring four holes to be drilled in the upper section for each bracket. For the 0°-roll drop test, the presence of the holes had no influence on the full-scale test results and no damage originated at the holes due to stress concentration. However, for the 15°-roll drop test, damage was observed at both the right- and left-side brackets. Thus, the mounting technique influenced the integrity of the composite sandwich structure in the upper section of the full-scale fuselage. Concluding Remarks
(3)
PE f = 125 (187.9) PE m 181.525
(4)
PE f = 125(1.035) PE m
(5)
The result obtained in Eqn. 5 indicates that the potential energy change experienced by the fullscale fuselage section is actually 3.5% higher than the expected “scaled” value, which is 125 times the potential energy change of the 1/5-scale
A composite fuselage concept for light aircraft and rotorcraft was fabricated and tested to demonstrate improved crash protection. The fuselage concept consists of a relatively rigid upper section, or passenger cabin, including a stiff structural floor and a frangible lower section which encloses the crash energy management structure. A 60-in. diameter full-scale fuselage section was fabricated using composite sandwich construction and drop tested from a height of 180-in., for an initial impact velocity of 372 in/s. An average floor-level acceleration of 29.9-g was obtained
from an accelerometer placed on a lead plate that was mounted to the floor of the fuselage. Only minor damage was observed in the transition region between the upper section and floor of the fuselage. The maximum displacement of the energy absorbing foam-block subfloor was 7.9 inches, providing a crush stroke of greater than 90%. A second drop test was performed at 372 in/s vertical velocity with a 15°-roll attitude, to determine the crashworthy performance of the fuselage under off-axis impact conditions. The average acceleration ranged from 28.6-g on the right side to 19.4-g on the left side of the floor of the fuselage. Crash simulations were performed using the nonlinear explicit transient dynamic code, MSC/DYTRAN, to predict both the 0°- and 15°-roll vertical drop tests. The MSC/DYTRAN model contained 14,992 nodes; 18,240 elements; and 60 concentrated masses to represent the inertia of the lead plate on the floor of the fuselage section. In general, the MSC/DYTRAN simulation predicted the overall magnitude, shape, and duration of the experimental acceleration pulses for both impact conditions well. Finally, the acceleration response of the full-scale fuselage section was compared with “scaled up” experimental data obtained from drop tests of a 1/5-scale model fuselage. The 1/5-scale model fuselage was a geometric and constitutive replica of the full-scale prototype, and the impact experiments were conducted under scaled conditions. In general, the scaled acceleration response had the same overall shape, magnitude, and pulse duration as that of the full-scale acceleration response. The average acceleration of the 1/5-scale model response was 25.4-g, which is 15% lower than the 29.9-g average acceleration of the full-scale data. For the 15°roll impact condition, the average acceleration for the 1/5-scale model fuselage was 18.6-g, which is only 4% lower than the similar full-scale response. Some scaling anomalies existed that may account for the small differences in the experimental data and damage levels exhibited by the 1/5- and fullscale fuselage sections. References 1. Desjardins, S., et al., “Aircraft Crash Survival Design Guide,” USAAVSCOM TR 89-D-22A through E, 5 volumes, December 1989. 2. Eiband, A. M., “Human Tolerance to Rapidly Applied Accelerations: A Summary of the Literature,” NASA TM 5-19-59E, Washington, DC, June 1959.
3. Cronkhite, et al., “Investigation of the CrashImpact Characteristics of Advanced Airframe Structures,” USARTL-TR-79-11, Sept. 1979. 4. Jackson, K. E., “A Comparative Analysis of Three Composite Fuselage Concepts for Crash Performance,” Proceedings of the 52nd AHS Forum and Technology Display, Washington DC, June 4-6, 1996. 5. Jackson, K. E., “Analytical Crash Simulation of Three Composite Fuselage Concepts and Experimental Correlation,” Journal of the American Helicopter Society, Vol. 42, No. 2, April 1997, pp. 116-125. 6. Jackson, K. E., and Fasanella, E. L., “Innovative Composite Fuselage Design for Improved Crashworthiness,” 54th American Helicopter Society Forum and Technology Display, Washington DC, May 20-22, 1998. 7. Fasanella, E. L., and Jackson, K. E., “Analytical and Experimental Evaluation of Composite Energy Absorbing Subfloor Concepts,” Proceedings of the AHS National Technical Specialists’ Meeting on Rotorcraft Crashworthiness, September 14-17, 1998, Phoenix, AZ. 8. Jackson, K. E., and Fasanella, E. L., “Crashworthy Evaluation of a 1/5 Scale Model Composite Fuselage Concept,” Proceedings of the 55th American Helicopter Society (AHS) Forum, Montreal, Canada, May 25-27, 1999. 9. Jackson, K. E., and Fasanella, E. L., “Crash Simulation of a 1/5-Scale Model Composite Fuselage Concept,” 1999 MSC Aerospace Conference Proceedings, Vol. 1, Long Beach, CA, June 7-10, 1999. 10. Jackson, K. E., and Fasanella, E. L., “Crashworthy Evaluation of a 1/5 Scale Model Composite Fuselage Concept,” NASA/TM-1999-209132, ARL-MR-441, April 1999. 11. Jones, L. E., and Carden, H. D., “Overview of Structural Behavior and Occupant Responses from a Crash Test of a Composite Airplane,” SAE Technical Paper 951168, May 3-5, 1995. 12. Thomson, R. G., Carden, H. D., and Hayduk, R. J., “Survey of NASA Research on Crash Dynamics,” NASA Technical Paper 2298, April 1984.
13. Williams, M. S., and Fasanella, E. L., “Crash Tests of Four Identical Low-Wing Twin-Engine Airplanes with Truss-Reinforced Fuselage Structure,” NASA Technical Paper 2070, 1982. 14. Alfaro-Bou, E., Williams, M. S., and Fasanella, E. L., “Determination of Crash Test Pulses and Their Application to Aircraft Seat Analysis,” SAE Technical Paper 810611, April 1981. 15. MSC/DYTRAN User’s Manual Version 4.0, The MacNeal-Schwendler Corporation, Los Angeles, California, November 1997.
