Static Properties and Multiaxial Strength Criterion for Design of Composite Automotive Structures

M. B. Ruggles, G. T. Yak, and R. L. Battiste

Oak Ridge National Laboratory Oak Ridge, Tennessee 3783 1-8051

0 RNL/C P-99 199

f%dF-7fi"s$l3--

ABSTRACT

The Durability of Lightweight Composite Structures Project was established at Oak Ridge National Laboratory (ORNL)by the US. Department of Energy to provide the experimentally-based, durability-driven design guidelines necessary to assure long-term structural integrity of automotive composite components. The initial focus of the ORNL Durability Project was on one representative reference material-an isocyanurate (polyurethane) reinforced with continuous strand, swirl-mat E-glass. The present paper describes tensile, compressive, flexure, and shear testing and results for the reference composite. Behavioral trends and proportional limit are established for both tension and compression. Damage development due to tensile loading, strain rate effects, and effects of temperature are discussed. Furthermore, effects on static properties of various fluids, including water at room and elevated temperatures, salt water, antifieex, windshield washer fluid, used motor oil, battery acid, gasoline, and brake fluid, were investigated. Effects d prior loading were evaluated as well. Finally, the effect 6 multiaxial loading on strength was determined, and the maximum shear strength criterion was identified for design. INTRODUCTION Development of lighter weight, more fbel efficient automobiles represents a technology area where advanced materials, such as polymer matrix composites, can be successfblly applied. However, while significant effort is being devoted to material development and processing, specific design guidance and an understanding of the material performance under actual service conditions is lacking. There is a recognized need for improved structural design methods and criteria that address deformation and failure behavior d composite materials. The Durability of Lightweight Composite Structures Project was established at Oak Ridge National Laboratory (ORNL) by the U.S. Department d Energy to provide the experimentally-based, durability-driven design guidelines [I, 21 necessary to assure long-term structural integrity of automotive composite components. The

project is closely coordinated with the Automotive Composites Consortium (ACC). The ORNL Durability Project initially focused on characterizingand understanding the mechanical behavior of a n SRIM (structural reaction injection molding) isocyanurate (polyurethane) reinforced with continuous strand, swirl-mat Eglass. The isocyanurate resin was DOW MM364, and the reinforcement was Vetrotex Certainteed Unifilo U750. This material was chosen by the ACC and supplied in the form CE 25 x 25 x 0.125 in. plaques. Five layers of mat were used in each plaque, resulting in a fiber content of about 25% by volume (40-50% by weight). The 0" direction is parallel to the roll direction of the original glass mat. The present paper summarizes tensile, compressive, flexural, and shear testing and results. Behavioral trends in tension and compression are discussed and corresponding proportional limits are established. Further, damage development due to tensile loading, strain rate effmts, and effects of temperature are discussed. In addition, effects on static properties of various fluids, such as water at room and elevated temperatures, salt water, antifieex, windshield washer fluid, used motor oil, battery acid, gasoline, and brake fluid, are assessed. Effects of prior loading are evaluated as well. Finally, the effect of multiaxial loading on strength is determined, and the maximum shear strength criterion is identified for design. TENSILE AND COMPRESSIVE ELASTIC CONSTANTS AND STRENGTH PROPERTIES Flat specimens with tabs were used in the tension and compression tests. Tensile and compressive specimens were 1.0 x 0.125 x 8.0 in. and 1.0 x 0.125 x 5.25 in., respectively. Tabs were 1.0 x 0.125 x 2.0 in. for tensile and 1.0 x 0.0625 x 2.25 in. for compressive specimens. In addition to straight-sided tabbed specimens, untabbed dogbone-shaped specimens (0.8 in. wide in the gage section) were used in a small number of tensile tests. A servocontrolled MTS axial-torsion mechanical testing machine together with an MTS digital TestStar Materials Testing Workstation was used for computerized testing and data acquisition. For tensile

Research sponsored by the office of Advanced Automotive Technologies, U.S. Department of Enerw, under contract DE-ACOS960R22464 with Lockheed Martin Energy Research Corporation.

DISCLAIMER
any of their employees, make any warranty, express or implied, or assumesany legal liability or respom-bilityfor the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not neeessanly constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
States Government. Neither the United States Government nor any agency thereof, nor

This report was prepared as an account of work sponsored by an agency of the United

DISCLAIMER
Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.

I
1 I
~

tests, specimens were mounted in Instron mechanical wedge grips and strain measurement was accomplished with an MTS f extensometer of 1.0 in. gage length. In the case c compression tests, specimens were mounted in a Wyoming IITRI fixture, and strain gages were used for strain measurement. The average in-air room temperature values d stifiess (modulus of elasticity), E, ultimate tensile strength, UTS, and failure strain, E f , are based on 277 tensile tests, 185 in the 0" direction, and 92 tests in the 90" direction. The average values are given in Table 1. The average value cf Poisson's ratio, v, for loading in the 0" direction is 0.31. Results of the tensile tests were further assessed to establish an average proportional limit for the material, defined as the point (oPL, in stress-strain space where the tensile stress-strain EPL) curve departs from linearity.

The following simple straight-line approximation may be used to represent reductions in both stifiess and strength with changing ambient temperature, T(OF):

(dE,Aa,,)

x 100%= 11.35 - 0.19 T.

The average in-air room temperature values d compressive stifhess, Ec, ultimate compressive strength (UCS), and compressive failure strain, Erc, are given in Table 3. These values are based on twelve compressive tests, seven in the 0" direction, and five in the 90" direction.

Table 1. Room temperature tensile elastic constants and strength properties 0" direction 90" direction Values E (Msi) 1.37 1.68 UTS ( h i ) 21.3 28.5 Ef (%) 2.12 2.17 OPL &si) 5.08 6.19 EPL (%) 0.38 0.37

Table 3. Room temperature compressive elastic constants and strength properties Values 0" direction 90" direction Ec P s i ) 1.37 1.67 UCS @si) -24.1 -3 1.5 EK (%) -2.41 -2.60

The UTS was found to increase with increasing stiffhess. The relation between UTS and stifiess can be reasonably well approximated with a linear law:

It can be seen that the stress-strain behavior is symmetric; that is, compressive and tensile properties are very similar, with compressive failure strain being slightly higher than the tensile value. The tensile and compressive sets of data can be approximated with straight lines of the same slope; that is, linear laws with the same coefficients can be used to describe the relation between strength and sti&ess for both tension and compression. FLEXURAL ELASTIC CONSTANTS AND STRENGTH Five 0.125-in.-wide by 1.0-in.-high beams, two in the 0" and three in the 90" direction, were tested under threepoint loading to determine the in-plane modulus of rupture (MOR), equal to the elastically calculated maximum bending stress at failure. The average MOR values were 30.6 ksi and 33.5 ksi for the 0" and 90" directions, respectively. The MOR value for the 0"-direction specimens was 22% higher, and the MOR value for the 90" direction was 5.4% higher, than the tensile strength for the corresponding orientation in the same plaque. Six 1.O-in.-wide by 0.125-h-high beam specimens were tested to determine the average out-of-plane flexural properties given in Table 4.

oms A666 + 0.02E. =

The proportional limit stress is 24% and 22% of the UTS f a the 0" and 90" directions, respectively. The proportional limit strain constitutes 18% and 17% of the failure strain for the 0" and 90" directions, respectively. Tensile properties as a function of temperature were established based on tensile tests conducted at -40, 70, 135, 190, and 250F. Average values of stiffhess and UTS obtained at elevated temperatures given in Table 2 are noticeably lower than those obtained at .and below room temperature. Likewise, tensile strength and stifiess appear to increase with decreasing temperature, resulting in "stronger" material at -40OF.

Table 2. Tensile properties as a function of temperature E Change in UTS (hi) Change in Temp. (OF) (Msi) E (%) UTS (YO) 250 0.94 -36.1 19.4 -25.5 190 1.00 -34.3 19.2 -28.9 135 1.14 -24.9 21.1 -22.2 70 1.48 0 26.0 0 -40 1.83 23.6 34.8 33.7

Table 4. Room temperature out-of-plane flexural properties Values 0" direction 90" direction EF(Msi) 1.60 1.79 MOR @si) 40.5 54.0 E@(%) 2.98 3.07

In an effort to account for wide-beam effects, the material was assumed to be isotropic and the equation

E = [(l- #)/]a,

(3)

was used to calculate the stifkess values. In addition, 0.375-in.-wide by 0.125-in.-high beams were tested in threepoint bending, yielding average MOR values of 45.14 ksi and 50,ll ksi for the 0" and 90' directions, respectively. It is seen that the 1.0-in.-wide and the 0.375-in.-wide beams produce comparable average MOR values. It is recognized that the out-of-plane MOR values are more than a factor of two greater than the UTS. That is, the load-canying capacity is more than twice that which would be indicated by limiting the elastically calculated maximum stress to the UTS. It is thus appropriate to provide a higher allowable for primary out-ofplane bending than for primary membrane loading in the design guide [l]. A factor of 1.5 is suggested. It must be noted that this observation only applies to out-of-plane bending. In the case of in-plane bending, the MOR values are much closer to the UTS values.
IN-PLANE SHEAR MODULUS AND STRENGTH The in-air room temperature values of the elastic shear modulus, ultimate shear strength, and failure shear strain were determined using Iosipescu specimens [3-51 oriented in the 0" and 90" directions. As expected, the data from the two directions were in good agreement. The average in-plane shear modulus, Gl2, shear strength, $12, and shear strain at failure, f12, were G12 = 0.65 Msi ,#U = 16.0 ksi, and .uf12 = 3.2%. It should be noted that the average tensile moduli rf the plaque that the shear specimens were cut f?om were higher than the averages for all plaques for both 0" and 90" directions. A corresponding adjustment in the measured in-plane shear modulus provides a value for G 2 of 0.59 Msi. A 1 corresponding adjustment in the shear strength based on the tensile strength data gives a value of 14 h i . The shear modulus of an isotropic material is related to the modulus of elasticity and the Poisson's ratio by the following equation:
(4)

25000

5
v)

22500-

20000

STRAIN RATE (Us)

Figure 1. Effect of strain rate on ultimate tensile strength.

EFFECTS OF PRIOR LOADING Damage development in tensile loading, as manifested by changes in stiaess with increasing stress and strain, was explored in tensile tests with intermittent stifkess checks. During this rest a specimen was loaded to 20% UTS and then unloaded to zero stress, at which point a stifkess check was performed. This sequence, consisting of loading to a specific loadstress level, unloading to zero stress, and a stiffhess check, was repeated for the load levels of 40, 60, and 80% UTS. Stiffhess check, loading to 80% UTS, and unloading to zero stress were followed by loading to failure. Percent change in stiffhess vs maximum prior tensile load is shown in Fig. 2.

The calculated modulus of elasticity based on the average G12 of 0.59 Msi and a Poisson's ratio of 0.31 is 1.55 Msi. This is consistent with the measured values of 1.37 Msi in the 0" direction and 1.68 Msi in the 90" direction.

EFFECT O F STRAIN RATE

The effect of strain rate on tensile behavior was assessed through constant strain rate tests performed at strain rates of IO", lo4, 1WZ,and 10 s". While no significant efi'ixt of rate on stiffuess and failure strain was observed, the UTS appeared to increase considerably with increasing strain rate as seen in Fig. 1. The e f c of rate on compressive behavior w s fet a investigated in tests conducted at the following constant load rates (with equivalent strain rates given in parentheses): 7.5 Ib/min (10" s-I), 2330 Ib/min (3 x IO4 s-'), and 13,980 Ib/min (2 x l o 3s-'). Observations were similar to those made in tensile tests.
0

2 5

SO

75

100

% UTS

Figure 2. Loss of stiffness due to prior tensile loadings to indicated levels.

As seen in Fig. 2 no significant changes in stifikess occur ' below 29.6% UTS. Above this threshold, stifiess decreases with increasing prior load, and this decrease may be described with the following linear law:
(T AE x 100%= 5.5 - 0.186-1TTC x 100%.

Note, that Eq. ( 5 ) predicts 13.1% stifhess loss at the end cf life. Loading to two-thirds of the UTS, which is the specified design allowable, will result in a 6.9% stiffness degradation. Effects of prior cyclic loading on tensile properties were investigated. A series of specimens were subjected to tension-tension fatigue cycles to various fhctions of the predicted cyclic life. The specimens were then monotonically tested to failure. The residual tensile strength was found to decrease linearly with the fatigue usage factor, n/Nf,where n, is the number of cycles the specimen has been subjected to and, Nf, is the allowable number of cycles. The predicted end-ofWe stiffness reduction due to prior fatigue loadings was 16.5%, which was somewhat higher than the 13.1% end-of-life stiffness loss predicted for static tensile loadings. Effects of prior creep on tensile strength and Stiffness were explored. Specimens were subjected to a sustained load for 20, 41, and 61% of rupture life, and then monotonically loaded to failure. Prior creep.had no effect on subsequent tensile strength and stiffness. EFFECTS OF MOISTURE ON STRENGTH AND STIFFNESS The study of environmental effects on strength and stifhess demonstrated that exposure time and weight gain provide measures of degradation for a given set of conditions. Moisture effects on tensile strength and stiffness are shown in Fig. 3. Conditions shown include (1) exposure in distilled water, (2) exposure in distilled water with superimposed loads of 25 and 50% UTS [ h m 61, and (3) exposure in 180F demineralized water. Exposure in 180F demineralized water was introduced for the purpose of accelerating the sorption process. Exposure under load is expected to serve a similar purpose. Observations of weight gain with exposure time c o n f m these expectations. For a given exposure time, weight gain due to moisture absorption is significantly increased by superimposed load as well as by elevated temperature. Fig. 3 shows that strength and M e s s decrease with increasing exposure time. It should be noted, that because the reductions were referenced to average values fkom several plaques, the solid curve fits in Fig. 3 do not pass through zero, as they should Therefore, they were adjusted (see dashed lines) to represent zero reduction in strength and/or stifhess at zero exposure time. Furthermore, correlations representing strength and stifhess reduction due to room-temperature exposure without load, can be combined into a single equation:

3 a

20

y = 3.479LOG(x)+ 3.44%

- 100 1000

10

10000

:
y = 3.405LOG(x) + 0.145

EXPOSURE TIME (h)

-20

10

100

1000

10000

EXPOSURETIME (h)
EXPOSURE IN DISTILLEDWATER EXPOSURE IN 180F DEMINERALIZEDWATER

9
0

EXPOSURE IN DISTJUED WAER UNDER LOAD = 50 %UTS (ORNL)

EXPOSURE IN DISTILLEDWATER UNDER LOAD = 25% UTS. HENSHAWel al EXPOSUREIN DISTILLEDWATER UNDER LOAD = 5~2%UTS. HENSHAW el al

Figure 3. Effect of moisture exposure time on tensile strength and stiffness.

FACTORS FOR OTHER FLUIDS

In addition to distilled and demineralized water, the following seven automotive fluid environments were investigated in this study: saltwater, antifreeze, windshield washer fluid, used motor oil, battery acid, gasoline, and brake fluid. Details of the fluid composition as well as sources cf fluidsare given in Ref. [3]. The average changes in strength and stifkess due to exposure for 1080 h and 7540 h are presented in Fig. 4. Strength results in Fig. 4 demonstrate that battery acid has the largest degrading effect followed by windshield washer fluid and distilled water, and then saltwater. Motor oil and brake fluid appear to increase the strength. Stifhess results in Fig. 4 indicate that battery acid and windshield washer fluid have the largest degrading effect (about a 20% reduction) for 7540-h exposure. The other changes in m e s s are all less than 10%. As in the case of moisture, the longer exposure times cause greater degradation.

AR x 100% = 3 4 log,, t E +3.44 . .4

where AR is reduction i strength and/or stifhess and t is n E exposure time in hours (tE > 0.1 h). Note, that Eq. (6) yields a 17% reduction in strength and/or stiffness after one year c f moisture exposure. Observations regarding moisture effects on tensile strength and m e s s can be extended to compressive properties. Exposure to moisture resulted in degradation cf compressive strength and stiffness. Presoaking in 180F demineralized water has a significantly greater effect on both tensile and compressive properties than presoak in roomtemperature distilled water.

1080 h
~

-0

Table 5. Effects of prior exposure in automotive fluids on shear properties Property Distilled Battery 180F Water Acid Demin.
7
9 2 12

4 a

r 0
8

-5

12

-14

-8

-16

-3 0 8

Water

-13 -35 -27

-10

-1-

Visual examination of the Iosipescu specimens indicated that the battery acid had attacked the matrix, Ieaving fibers exposed on the surfhce. However, the soak in battery acid caused an 8% increase in the shear strain at failure. The soak in 180F demineralized water was the most damaging; causing a 35% decrease in the shear strength. Note that the corresponding reduction in tensiIe and compressive strength w s approximately 50%. a

MULTIAXIAL STRENGTH CRITERION

FLUID
10

g a
8

-10

-20

-30

I
FLUID

STFENGTH
STIFFNESS

Figure 4. Effect of prior exposure to various automotive fluids on tensile strength and stiffness.

Observations made for tensile properties can be extended to compression. Compression tests on specimens presoaked in battery acid for 1080 h produced a 17% reduction in stiffness and a 14% reduction in strength. These values are somewhat higher than the corresponding ones in tension. Iosipescu specimenswere soaked in distilled water for 2200 h, battery acid for 2200 h, or 180F demineralized water for 1080 h. Percent changes in shear modulus, shear strength, and shear strain at failure for the different environments are given in Table 5 .

Because the designer is usually only provided with uniaxiaI strength values, while the stresses in components are often multiaxial, it is necessary to have a multiaxial strength criterion. Because the material in this study is generaIly in a plate or shell form that is loaded so that the stresses are in the plane of the plate, only a biaxial strength criterion is required. Many merent strength criteria have been proposed. Some of them account for different strengths in tension and compression and for merent strengths in different directions. However, such criteria become quite complex for the designer to use and require extensive materials testing to determine all the required constants. To keep the criterion simple and easy for the designer to use, the strength is assumed to be the same in tension and compression and in all directions in the plane of the plaque. Thus, the only strength value used is the tensile strength in the 0 direction. ' Five candidate strength criteria are shown in Fig. 5. Note that all five criteria predict the same uniaxial strengths. They differ most under equal biaxial tension or compression and when the maximum principal stress is tension and the minimum principal stress is equal in absolute value but opposite in sign. This is the condition of pure shear. According to classical elastic plate theory, when a concentric ring-load is applied to a simply-supported circular plate, the stress state at the bottom surface is equal biaxial tension everywhere inside the ring load. Thgrefore, this test w s chosen for evaluating the failure criterion in the first a quadrant. Tests were run on 3.7-in.-diam. circular disks supported on a plate with a 3.5-in.-diam. hole and loaded with a plunger that made contact at a radius of 0.75 in.. Because the deflection at failure was more than twice the disk thickness, it was necessary to perform a large-deflection elastic finiteelement analysis to determine the stresses at failure. The disks failed where the plunger contracted the disk. That is the location of maximum stress according to the analysis. The analysis showed that the radial stress was somewhat higher than the tangential stress at the location where the disks failed.

Maximum shear stress criterion

strain criterion

I *90"
Figure 6 Available data indicate maximum shear . criterion applies.

Figure 5. Candidate failure criteria.

It was necessary to account for the observed difference in tensile strength and out-of-plane flexure strength (MOR) became the disks were basically subjected to out-of-plane bending. The corrected maximum stresses in the disks at failure were 22.5 ksi in the radial direction and 17.0 hi in the tangential direction. The average maximum stress in the disk is compared to the failure criteria in Fig. 6 The average failure point was . plotted at two locations in stress space that are symmetrical about a 4 5 O line. The average maximum stress in the disk agrees well with the maximum shear and maximum stress theories of failure. The maximum work theory agrees almost as well, but is unconservative. The modified maximum work theory is extremely conservative, and the maximum strain theory is extremely unconservative. "herefore, either the maximum stress or maximum shear theory would be a reasonable failure criterion in the first quadrant, where they are identical. Note, however, that those two criteria are very different in the second and fourth quadrants. The principal stresses in the case of pure shear are tension in one direction and equal compression perpendicular to the tensile stress. Therefore, shear strength provides the data needed for evaluating the criteria in the second and four& quadrants. The average data point fiom the shear test is plotted in both the second and fourth quadrants in Fig. 6. The average 90" direction tensile strength and compressive strengths are also plotted. Clearly, the maximum stress theory is unconservative in the second and fourth quadrant. Only the maximum shear and modified work criteria are both conservative relative to all the average data points. The maximum shear criterion is more conservative than the modified work criterion in the second and fourth quadrant. However, the modified work criterion is extremely conservative in evaluating the first quadrant. Unfortunately, data in the third quadrant are not available for the M r u e criteria. The maximum shear criterion is recommended because it is widely used by designers and is very simple.

SUMMARY

In-air room temperature tensile elastic constants and strength properties were established for both 0" and 90" directions. Proportional limit was established and found to constitute 22 to 24% of the UTS. Ultimate tensile strength was found to increase with increasing stifhess, and a linear law correlating UTS and stifihess w s proposed. a Tensile properties as functions of temperature were established for temperatures between -40 and 250F. An equation for predicting changes in elastic properties with ambient temperature was proposed. Compressive properties and proportional limit were established and found to be very similar to those in tension. The in-plane MOR was no more than 22% higher than the tensiIe strength, but the out-of-plane MOR was more than twice the tensile strength. The room temperature in-plane shear 'modulus and strength were determined. No significanteffect of loading rate on stifhess and failure strain was observed. However, UTS increased considerably with increasing strain rate. This observation was extended to compression. E f c s of prior static, cyclic and creep loadings were fet investigated. Static loadings above a 29.6% UTS threshold reduced the subsequent sti&ess (up to 13.1% at 100% UTS). Residual tensile strength decreased linearly with fatigue cycling (up to 22% at n/Nr = 0.8). With a design factor CE 20 on cycles to failure, the maximum reduction is negligible (about 1.5%). A factor of 20 on cycles to failure will limit the stifness loss during fatigue cycling to 10% or less.

Prior creep strains (up to 61% of the creep ductility) have no effect on subsequent tensile strength and stifkess. E f c s of moisture exposure on tensile. and compressive fet ee strength and stiflkess w r explored. Exposure time and weight gain were shown to provide measures CE degradation for a given set of conditions. A single equation w s proposed to represent strength and m e s s a reductions due to moisture exposure. E f c s of seven automotive fluid environments on tensile fet strength and stiffness were investigated. Battery acid had the largest degrading effect followed by windshield washer fluid, distilled water, and fmally saltwater. Shear and bending tests on circular disks provided data for evaluating the applicability of various multiaxial strength criteria. The maximum shear stress criterion using the zero direction tensile strength as the base value was recommended for design. REFERENCES

1 J. M. Conun, et. al., Durability Based Design Criteria . for an Automotive Structural Composite: Part I . Design Rules, OWL-6930, Lockheed Martin Energy Research Corporation, Oak Ridge National Laboratory, February 1998.
2. J. M. Conun, et. al., Durability-Based Design Criteria for an Automotive Structural Composite, Proceedings of the 13 Annual ESD Advanced Composites Technology Conference, September 28-29,1998.

3. ASTM Standard D 5379/D 5379M-93, Standard Test Methodfor Shear Properties of Composite Materials L y the V-Notched Beam Method, American Society ftr Testing and Materials.
4.

5 . Adams, D.

F., and Walrath, D. E., Further Development of the Iosipescu Test Method, Experimental Mechanics, Vol. 27,No. 2 June 1987, pp. , 1 13-11 . 9

Walrath, D. E., and Adams, D. F., The Iosipescu Shear Test as Applied to Composite Materials, Experimental Mechanics, Vol. 23,No. 1 March 1983,pp. 105-110. ,

6. Henshaw, J. M., Meyer, L. J., Houston, D. Q., and Hagerman, E. M., Stressed Environmental Degradation of Automotive Composite Materials, Advanced Composites Conference Proceedings, ESD - The Engineering Society, April 1997,pp. 367-380.
7 . J. M. Corum, et.al, Durability of Lightweight Composite

Structuresfor Automotive Applications: Progress Report for Period Ending September 30, 1995, ORNLITM13176, Lockheed Martin Energy Research COT., Oak Ridge National Laboratory, March 1996,