JOURNAL DE PHYSIQUE

Colloque C5, supplCment au n08, Tome 46,aoQt 1985 page C5-455

D E T E R M I N A T I O N OF DYNAMIC FLOW CURVE OF METALS A T A M B I E N T AND ELEVATED

TEMPERATURES BY ROD IMPACT TECHNIQUES*
D.C. Erlich and P. chartagnac**
SRI InternationaZ, 333 Ravenswood Avenue, Menlo Park, California 94025,

U.S.A.
~6sum6 On dicrit une approche combinant rgsultats exp&rimentaux et calculs numgriques pour diterminer la courbe dgformation-contrainte d'un mgtal tempgrature ambiante ou 'a tempgrature GlevGe pour de randes de'formations et des vitesses de dgformation GlevGes (lo4 'a 10 / ) s . Un ichantillon en forme de barre de section circulaire est percut& k haute vitesse et le profil dgforrng rgsultant de l'impact est photographi; avec une cam6ra k haute vitesse. L'essai est ensuite simulg avec un programme bidimensionneldediff6rences finies pour les calculs de propagation d'onde, en variant les donnges pour la courbe diformation-contrainte, jusqut; ce que les profils expirimentaux et calculgs concordent. On prGsente des rgsultats pour l'acier AISI 4340 et pour un alliage de zinc/aluminium.

>

S

Abstract - An experimental/computational procedure is described for determining the compressive flow curve of metals at high strains and strain rates (lo4 to l 5 , , at either ambient or elevated temperatures. A cylindrical o/) specimen rod is impacted at high velocities and the resulting deformation profiles are recorded with a high-speed framing camera. The experiment is then simulated using a two-dimensional finite-difference wave propagation code, varying the input flow curve until the experimental and computational profiles agree. Data is presented for AISI 4340 steel and a zinc/aluminum alloy.

Determination of high strain-rate constitutive behavior of materials is of incteasing interest to researchers in a growing number of fields. However, the availability of experimental techniques that allow such determinations at high strain rates and large plastic strains (20%-150%) has been extremely limited, if not nonexistent. Recently we have developed the rod impact technique (based upon the Taylor test for measurement of dynamic yield strengths /1/ to determine the stress-strain flow curve in compression for a material at ambient or elevated temperature. I

-

HISTORICAL PERSPECTIVE AND TECHNIQUE DEVELOPMENT

Classic Taylor Test In 1947, Taylor and Whiffin /1,2/ accelerated cylindrical specimen rods into a "rigid" plate. The plastic deformation at the impact end shortens the rod, and the fractional change in rod length can, by one-dimensional rigid-plastic analysis, be simply related to the dynamic yield strength. The authors showed this relationship to be independent of both the rod aspect ratio and the impact velocity for a wide variety of materials, including copper, lead, paraffin wax, and various steels. Though appealing in its simplicity, the Taylor test had only a moderate follow-up. Lee and Tupper /3/ in 1953 and Hawkyard et al. /4/ in 1968 attempted to model the *~esearchpartially supported by U S Army Research .. **Current address - ETCA/CEG 465 Gramat, France. Office.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1985556

C5-456

JOURNAL DE PHYSIQUE

Taylor test using various one-dimensional analyses, but were largely unsuccessful. More recently the use of two-dimensional wave propagation codes allowed better understanding of and renewed interest in this technique. In 1972, Wilkins and Guinan / 5 / , using the HEMP code and an elastic-plastic model with work-hardening, were able to correctly simulate the final shapes, as well as the final lengths, of Taylor test specimens of several metallic alloys at ambient temperatures. Their results showed a good correlation between the dynamic yield strength and the fractional change in rod length for a wide range of impact velocities and rod aspect ratios, thus confirming many of the TaylorIWhiffin conclusions. Symmetric Rod Impact Test In the early 1980s, Erlich et al. /6/ implemented two major modifications to the classic Taylor technique. The first was to use ultrahigh speed photography to monitor the deformation history of the specimen rod. This allows intermediate, as well as final deformations to be compared with the computer simulations, thus improving the reliability of the flow curve determination. The second modification was to replace the "rigid" plate with another rod of the same geometry and material as the impacting rod (Fig. l ) a. This arrangement, referred to as the "symmetric rod impact" (SRI) technique, allows the impacting ends of the two specimen rods to deform together symmetrically, thus eliminating boundary condition uncertainties in the analysis that arise from the unknown friction conditions at the rod-plate interface and from the deformation of the "rigid" plate adjacent to that interface. By use of the SRI technique, dynamic flow curves at ambient temperature were obtained for 6061-T6 aluminum / 6 / and 4340 steel /7/.

BEFORE IMPACT
Identical Specimen Rods
(a) Symmetric Rod l mpact Test

AFTER IMPACT

Asymmetric l mDact Test

Specimen Rod

FIGURE 1 SCHEMATIC OF TWO ROO IMPACT CONFIGURATIONS Asymmetric Rod Impact Test For Elevated Temperatures In the early 1980s, Gust I81 used a reverse ballistics variation of the classic Taylor test to measure fractional changes in the length of various metallic rods at initial temperatures up to about 100O0C. In this variation, the "rigid" plate is launched into a stationary specimen rod preheated to the desired temperature. Concurrently, researchers at SRI International, stymied in their attempts to use the symmetric rod impact technique to determine dynamic flow curves at elevated temperatures (because of the infeasibility of heating the moving specimen rod), were investigating several alternatives. One, impacting a stationary heated rod with an identical but unheated rod, proved unsatisfactory because the gross difference in deformability of the two rods created large uncertainties in the boundary conditions at the impact interface.

A second alternative was to combine the reverse tiallistics impact of a heated rod with high-speed photographic measurements of the resulting deformation, to form the "asymmetric rod impact" or ASRI technique (Fig. l ) b. However, the boundary condition uncertainties inherent in the ASRI technique needed to be examined. Computer simulations of the proposed ASRI test showed that the presence or absence of friction had a profound effect on the rod deformation, whereas the use of ideal rigid or realistic material properties for the impactor plate had a noticeable but minor effect

.

Identical 4340 specimen rods at room temperature were therefore impacted in two experiments: one an SRI test and the other an ASRI test at approximately half the impact velocity (so as to attain similar stresses). Computer simulations of both tests yielded identical flow curves to well within the experimental uncertainties, provided that realistic material properties were assigned to the impactor plate and a frictionless rod-plate interface condition was used. Thus it was concluded that frictional effects were insufficient to noticeably influence the rod deformation profiles (future tests are planned to validate this conclusion for a wider range of specimen materials), and thus the ASRI technique could be used with confidence to obtain dynamic flow curves. I1

- EXPERIMENTAL TECHNIQUE

SRI Test at Ambient Temperature A diagram of the experimental arrangement for the SRI test at ambient temperature is shown in Fig. 2 The specimen rods are identical right circular cylinders, 44.4 mm . long by 9 5 mm in diameter (these dimensions are arbitrary), whose ends are machined . flat and parallel to within about 0.01 mm. The impacting rod is mounted on the front end of a projectile, which is accelerated by expanding helium in a 6.35-cmdiameter gas gun. The stationary rod is held in place by six ceramic fingers attached to a target mounting fixture, which in turn is affixed to the muzzle of the gun. Alignment of the two rods is critical to ensure that the impacting ends are parallel and coaxial. Misalignment by as little as 0 1 mm can have a noticeable . effect on the deformation profiles.
Hole for Glass

5 cm

SIDE VIEW

END VIEW

FIGURE 2 SYMMETRIC ROD IMPACTTESTSAT AMBIENT TEMPERATURE The specimen rods are backlit by a variable-duration fast rise- and fall-time highintensity xenon flash tube triggered just before impact. The silhouettes of the deforming rods are recorded by a high-speed framing camera at framing rates between one-half and one million frames per second. Selected frames from an SRI test of 4340 steel (Rc31) are shown in Fig. 3 .

JOURNAL DE PHYSIQUE

After the deformation is complete (from 40 to 100 ps after impact, depending on the material), the specimen rods fly into a recovery pipe filled with rags and other energy-absorbing materials that minimize additional deformation. The pipe is narrow enough to prevent the projectile from entering and reimpacting the specimen. The recovered rods are then sectioned along the axis and examined metallographically to ascertain the extent of internal damage. The impact velocity must be low enough to suppress the formation of tensile voids (which may occur at early times by the focusing of the radial release waves on the rod axis) or shear bands (which may occur at later times as a result of large plastic deformation near the impact ~1thou~h.a small amount of incipient damage can be tolerated, any significant end). amount of damage will affect the shape of the deforming rod profiles or cause rod failure. ASRI Test at Elevated Temperature The ASRI technique for specimens at elevated temperatures differs only slightly from the SRI technique. The impactor is a 5-cm-wide, 1-cm-thick disk of ultrahigh strength maraging steel backed by a thicker aluminum projectile head. The stationary specimen rod needs to be aligned parallel to the direction of impact, but the co-axiality condition critical to the SRI test is not necessary here, since the rod can impact anywhere near the middle of the disk. The specimen is heated before impact by three infrared line heaters arranged around the rod (at a distance of 15 cm) so that radiation from the linear filaments is focused by elliptical reflectors onto the rod. The temperature of the specimen rod is monitored by a Chromel-Alumel thermocouple attached to the nonimpacted end. By use of a variable 280-V, 100-A power.supply for the heaters, a temperature of 1000°C can be attained in about 150 s. Thermal uniformity of the rod is quite good because of the relatively short thermal equilibrium times (a couple of seconds) for metallic rods of this diameter. Specimen rods from ASRI tests are usually recovered with negligible additional damage and with the impact interface very nearly perpendicular to the rod axis. Such is not the case with specimens recovered from SRI tests, where the two rods decelerating together can re-impact, causing further deformations, and where slight misalignments in co-axiality can skew the impact interface. 111

- ANALYTICAL TECHNIQUES

Measurement of Rod Deformation Profiles The first step in the analysis is to digitize and plot the rod profiles for comparison with computer simulations at various times during the deformation. For SRI tests, the profiles are obtained exclusively from the framing camera records. For ASRI tests, however, in addition to the photographic records, we can measure, with a far greater accuracy, the recovered specimens and obtain the final deformation profile. For the framing camera records, errors may be caused by photographic aberrations and nonlinearities, target misalignments (SRI tests only), and parallax errors, but mostly by the fuzziness of the image. If we assume that the error band has a constant width, M y , independent of axial position (Fig. 4, we can estimate the ) magnitude of the uncertainty in radial expansion AylAR, both for the photographic and the recovered rod measurements. The results, shown in the table in Fig. 4, show that the error in radial expansion, for a particular axial position, can vary from 2% for the sharpest photographs of large deformations (100% AR/R ), to 25% for the poorest quality photographs of smaller deformations (20% A R / ~ ). For the same two deformations, measurements of the recovered rod profiles woufd yield radial expansion uncertainties of 0.3% and 1.5%, respectively.

(b) +1 IU

(f) +15IU

I

5

a Q

n

0.5 0

-

\--7
Contour
1. O

: 't : li na il
0

-

I
4.0

6
(dl +7 pa
(h) +30@
FIGURE 3 SILHOUETTES OF 4340 STEEL (Rc31) RODS DURING SYMMETRIC l MPACT AT 456 m/s
Times shown are approximate

(c) +4 m

2. O 3.0 AXIAL DISTANCE (cm)

FIGURE 5 DEFORMATION CONTOURS 2 5 ~ AFTER 4340 s STEEL SYMMETRIC ROD IMPACTS AT 457 m/s

t m s from ie

impact.

AY/AR for

AY
(mm)

AY Ro

AR AR - =20% -= Ro 0

AR 50% -R0

- 100%

Framing Camera Silhouettes Recovered Rod Poie rfls
FIGURE 4

0.1 - 0.25

2-5%

10-25%

4-1 0%

2-5%

0.015

0.3%

1.5%

0.6%

0.3%

UNCERTAINTIES IN OBTAINING DIGITIZED ROD PROFILES

Computer Simulations of Rod Impact Tests The dynamic flow curve of a material is determined from a rod impact test by computationally simulating the experiment and varying the input flow curve until the computed profiles agree with those determined experimentally at various times during the deformation history. The computer code we have used for these simulations is the recently developed C-HEMP /9/, a two-dimensional finite-difference code for treating stress wave propagation, in either planar or axisymmetric flow, caused by impacts or explosive detonations. The specimen rods (and the impactor, for ASRI test simulations) are divided into a series of rectangular zones, or computational cells, each of which, in the axisymmetric geometry appropriate to the rod impact test, represents an annulus of revolution about the rod axis. The corners, or nodes, of the cells are given an appropriate initial velocity. Then the subsequent node velocities and the resulting deformations are determined for successive small time increments by solving the Lagrangian equations of motion for a continuous medium.

A standard Mie-Gruneisen formulation relates the pressure to the specific volume and
internal energy, and a rate-independent elastic-plastic model with work hardening is

JOURNAL DE PHYSIQUE

used to describe the plastic deformations in each cell. Bulk and shear moduli are obtained from Hugoniot data, if available, but computations have shown that the deformation profiles are sensitive to variations in the flow parameters only, and not the elastic moduli. The flow curve is inputted in terms of a flow stress (Y), equal to the yield strength in a uniaxial stress test, as a function of the equivalent plastic strain, which is defined as the square root of two-thirds the sum of the squares of differences of the three principal strains. A quasi-static compressive or tensile flow curve is often used as the first trial flow curve (since there are few dynamic curves to be found in the literature), and the flow curve is then modified in subsequent computations until the computed deformation profiles agree with the experimental results to within the experimental error. This process may take from two to more than twenty computational iterations. Considerations of Strain Rate and Temperature Computer simulations have shown that the equivalent plastic strains (as defined in the previous section) and strain rates vary as a function of time and of axial and radial position within the specimen rod. For positions near the impact interface, the average strain rate along the rod axis is two to three times higher than that near the edge. For positions along the rod axis, the average strain rate near the impact interface is about four times higher than that one rod radius distant. It is thus important to note that the experimentally measured radial rod expansion at a particular axial position is not the result of a region of material undergoing deformation at a specific strain rate, but rather the result of various regions of material deforming over a moderate range of strain rates. Therefore, the rod impact technique should not be viewed as a method for accurately determining the strain rate sensitivity of the flow curve within the range of strain rate observed in the test (most of the deformation occurs between lo4 and 5 x 104/s), but rather as a means of determining the average flow curve over that range. Furthermore, if a material's flow stress exhibits a very large strain rate sensitivity within the observed strain rate range, then no matter what flow curve we would try in our rate-independent simulation, we could not match the experimental profiles throughout the deformation history, and in fact, we might not be able to match the profiles well at any time much after impact. So conversely, if we are able, by using a particular flow curve, to match the experimental profiles well throughout the deformation history, then theathe material exhibits no significant strain rate sensitivity within the range noted above. The computer simulations have also shown that large temperature increases caused by plastic work are created in the highly deformed regions of the specimen rods. This is as expected for any high-strain-rate process, where the deformation is too rapid to allow temperature equilibration throughout the specimen. For the 4340 steel SRI test discussed previously, temperatures in the region near the impact interface increased by as much as 430°C during the deformation process. The rod impact test is thus an adiabatic, rather than an isothermal process, and there is no way to separate the temperature effect on the flow curve (thermal softening) from the plastic strain effect (work hardening). This makes it difficult to compare the flow curve obtained from rod impact tests with those obtained at a specific temperature from other tests. We do not consider this a serious limitation, however, because the flow curve we obtain is appropriate for a material undergoing high strain and strain rate deformations, which are always accompanied by a large temperature increase. IV

- ROD

IMPACT TEST RESULTS

Results are presented for some SRI and ASRI tests at ambient and elevated

temperatures. Space limitations prevent comparisons with other data or detailed discussion of the metallurgical significance of these results. SRI Tests on 4340 Steel at Ambient Temperatures Ambient temperature SRI tests at identical impact velocity (457 m/s) were made using 4340 steel specimens of three initial hardnesses: Rb94, Rc31, and Rc39. The latetime profiles (approximately 28 ys after impact) for all three of these tests are compared in Fig. 5. Although the three profiles do not precisely coincide, they all lie within the band of experimental uncertainty, M y . Computer simulations were performed, first using the Rb94 and Rc39 static flow curves obtained in the literature /10,11/, and then varying the input flow curves until the computed rod profiles matched the experimental data to within +9y. Fig. 6 shows the various flow curves used in these calculations and compares their resultant late-time profiles with the experimental rod profiles (an average from the three tests). The static curves provide a very poor match to the data. Elasticperfectly plastic curves at flow stresses (Y) of 13 and 14 kbar (Case 1 and 2, respectively) are significantly better. A flow curve that work hardens from Y = 11 . kbar at zero strain to Y = 13 kbar at a 0 2 strain and is perfectly plastic at further strains (Case 3) provides the desired match. To further substantiate this result, we compared the Case 3 simulated profiles at several intermediate times with the corresponding experimental profiles. Results for the Rc31 case (Fig. 7 ) , indicate good agreement throughout the deformation history.

t \=+a;

I

I

I +

........

Experimental Static (RB94) Case 1 (13 kb) Case 2 (14 kb) Case 3 (11-13 kb)

1

I i i l Contour nta AXIAL DISTANCE (crn)

IL

5
4

0

0.2 0.4 0.6 08 . 1.0 EQUIVALENT PLASTIC STRAIN

FIGURE 6 COMPARISON OF EXPERIMENTAL PROFILE 2 8 ~ AFTER SYMMETRIC IMPACT s OF 4340 STEEL RODS WITH COMPUTED PROFILES, USING DIFFERENT FLOW CURVES

FIGURE 7 CALCULATED (SOLIDLINE)AND EXPERIMENTAL (DASHED LINE)PROFILES AT INTERMEDIATETIMES AFTER IMPACT ROD AT 457 mls OF 4340STEEL (Rc31)

C5-462

JOURNAL DE PHYSIQUE

Considering the large variation in the static flow data as a function of initial hardness, the insensitivity of the dynamic flow curve was at first surprising. However, hardness tests on the recovered rods indicated that the material from the R 94 and R 31 specimens hardened to nearly the Rc39 level at rather low strains (gelow 20%$. So, for most of the deformation history, the three materials behaved similarly. ASRI Tests on Zinc/Alumimum Alloy Six ASRI tests were performed using Zn-22%A1 alloy at two grain sizes: fine-grained (FG) and coarse-grained (CG), with mean grain diameters of 0 6 and 3.3 ym, . respectively. Tests on both FG and CG specimens were conducted at ambient temperature, 150°C, and 250°C. Flow curves that provided an accurate match between the final computational profiles and the recovered specimen rod profiles are shown in Fig. 8. Comparisons of computational profiles and framing camera profiles at earlier times in the deformation histories (as, for example, in Fig. 5) showed agreement within the larger experimental uncertainty. The dynamic flow curves seem to indicate that, for the CG specimens, work hardening dominates over thermal softening, whereas for the FG specimens, the reverse is true.

1 .o
-

. :
--A

-

-----C G

I

I

I

I

I

I

I

I

I / J ] I 211-22 Al Alloy

,

-

6 U1

0.5 .

\- Photographs from

-Measured Profile

. .

\

20% .

FG

F = Fine Grained G C = Coarse Grained G

---C-HEMP 20°c-. Simulated Profile

C G

-- 20°c -150°c

Q
0

O 05 .
250% . .

L ~- A ~ b
15Ooc. .

-

I--------

,

- --->--------. --_ -._ '-_G -,F
' .

_

0 P 0

2
I

250"~.

05 .
T

C G

-0

I

l

Denotes Maximum Strain Level Reached by C-HEMP Simulation I I I I I I I I I l

AXIAL DISTANCE (cm)

AXIAL DISTANCE (crn)

,

(a) 25 ps After Impact

(b) 60 fis After Impact

0

0.4 0 8 1.2 1 6 2 0 2.4 2.8 . . .
EQUIVALENT PLASTIC STRAIN
FIGURE 9 COMPARISON OF CALCULATED AND EXPERIMENTAL (FROM PHOTOGRAPHS) FG Zn-22AI ALLOY ROD PROFILES AT TWO TIMES DURING ASRl TESTS

FIGURE 8 FLOW CURVES OBTAINED FROM ASRl TESTS OF Zn-22AI ALLOY

REFERENCES Taylor, G. I., Proc. Roy. Soc. A 194 (1948) 289-299. Whiffin, A C , Proc. Roy. Soc. . . (1948) 200-232. Lee, E H. and Tupper, S J., J. Appl. Mech.Z(I954) 63-70. . . Hawkyard, J. B., Eaton, D. and Johnson, W., Intl. J. Mech. Sci. E(1968) 929948. Wilkins, M L., and Guinan, M. W., J. Appl. Phys.a(l973) 1200-1206. . Frlich, D. C , Shockey, D. A., and Seaman, L., AIP Conf. Proc. 78, 2nd Topical . Conf. on Shock Waves in Condensed Matter (1981) 402-406. Erlich, D C , and Shockey, D. A, Proc. APS Topical Conf., Shock Waves in . . . Condensed Matter (1583) 129-132. Gust, W H., J. Appl. Phys. 53(5) (1982) 3566-3575. . Seaman, T., , Cooper, T. and Erlich, D C., "User's Manual for C-HEMP, a Two. Dimensional Wave Propagation Code," SRI International Final Report for Ballistic Research Laboratory, Aberdeen Proving Ground, M) (1984). Metals Handbook, Amer. Soc. for Metals, 9th Edition (1978). Idarson,F. R. and Nunes, J., ASM Trans. 5 (1960) 663-682. 3 -

AE -

Sign up to vote on this title
UsefulNot useful