Recent advancements in injection molding have made it a popu-

lar manufacturing method for cost-effective, high-volume produc-

tion of plastic and particle-reinforced plastic composites. ASTM
Standard D 638 [1], a standard test method for tensile strength of
plastics, has been used widely in the industry to evaluate the tensile
strength of plastic-based materials. In the absence of test methods
for the tensile evaluation of particulate-reinforced plastics, the
ASTM D 638 test method also has been used to evaluate these re-
inforced plastic composites. Figure 1a shows a fractured ASTM D
638 Type IV specimen made of Nylon-11 bonded Nd-Fe-B parti-
cles. This material, which contains 59.7 vol% of brittle Nd-Fe-B
particles, is formulated for permanent magnet applications. The
high percentage of Nd-Fe-B particles is necessary to achieve the
desired magnetic properties for electrical motor and actuator appli-
cations. However, the high percentage of Nd-Fe-B particles also re-
sults in a material with reduced ductility [24].
The tensile evaluation of plastic-bonded Nd-Fe-B magnets has
shown that fracture almost always initiates at a location where the
straight gage section of the specimen ends and the curved transition
region begins. This location is marked as Point B in Fig. 1a. The
failure at Point B, which is outside the uniformly loaded straight
gage section, is due to a stress concentration and can lead to an un-
derestimate of the tensile strength. Such stress concentration is not
a problem for ductile materials because the large plastic deforma-
tion will redistribute the stress. For example, Fig. 1b shows a de-
formed ASTM D 638 Type IV tension specimen made of pure Ny-
lon-11. This specimen continued to deform extensively without
fracturing at the location of initial stress concentration.
As engineered and particle-reinforced plastics become stronger
and in some cases more brittle, the stress concentration in current
tension specimen geometries has become an issue of concern. One
of the objectives in this research is to analyze the stress concentra-
tion factor [5] on specimens specified in ASTM D 638 using the fi-
nite-element method. Five types of specimens, designated as Type
I, II, III, IV, and V, are defined in ASTM D 638. The shapes of Type
III and V specimens are similar, which results in the same stress con-
centration factor at the location where the gage section ends and the
transition region begins. Therefore, only Type I, II, III, and IV spec-
imens were studied. In addition to these four standard tension spec-
imens, the stress concentration factor for other specimen geometries
with single- or double-arc fillets was also analyzed using the finite-
element method to investigate possible ways to reduce the stress
concentration without changing the overall size of specimens.
The photoelastic method has been used in the past to study the
stress concentration factors for shoulder fillets in flat plates [57].
Most of the cases studied have fillet radii much smaller than the
width of the specimen. The shapes of tension specimens investi-
gated in this research have the opposite trend: the fillet radii are
much larger than the width of the specimen in order to reduce the
stress concentration factor. For specimens with very small stress
concentration factors, the finite-element method with a reasonably
fine mesh was applied to analyze the stress distributions and esti-
mate the stress concentration factors. The finite-element method
has been applied previously to the design of flat tension specimens
of advanced composites [8], ceramic matrix composites [9], and
monolithic ceramics for creep testing [10]. Oplinger et al. [11] has
studied the use of streamlined specimens to reduce the tensile and
shear stress concentration in composite specimens.
In this paper, first the finite-element modeling of tension speci-
mens is introduced. The stress concentration factor is defined. The
distribution of von Mises stresses for four types of specimens spec-
ified in ASTM Standard D 638 is presented. The stress concentra-
tion factors of other specimen shapes with single- and double-arc
fillets are analyzed. Finally, a correlation between the magnitude of
the stress concentration factor and the ratio of arc radius and width
of the gage section is established.
Copyright 2003 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. 1
Monika G. Garrell,
1
Albert J. Shih,
2
Edgar Lara-Curzio,
3
and Ronald O. Scattergood
4
Finite-Element Analysis of Stress Concentration in
ASTM D 638 Tension Specimens
ABSTRACT: Experimental results showed that ASTM D 638 Type IV flat tension specimens, made of a Nylon-11 matrix containing a large vol-
ume concentration of Nd-Fe-B particulates, failed at a location where the straight gage section of the specimen ends and the curved transition re-
gion begins. The stress distribution in this specimen was analyzed using the finite-element method, and it was found that there is a stress concen-
tration at this location. The stress distributions in tension specimens with both single- and double-arc transition regions were analyzed and stress
concentration factors were calculated. A linear relationship between the magnitude of the stress concentration factor and the ratio of the width in the
gage section and the arc radius of the transition region is identified. This study shows that it is possible to reduce the magnitude of the stress con-
centration factor for the ASTM D 638 Type IV flat tension specimen by redesigning the specimen geometry without changing its overall size.
KEYWORDS: stress concentration factor, finite-element analysis, tensile strength
Journal of Testing and Evaluation, Vol. 31, No. 1
Paper ID JTE11402_311
Available online at: www.astm.org
Manuscript received 5/2/2002; accepted for publication 11/13/2002; pub-
lished 11/18/2002.
1
Department of Mechanical and Aerospace Engineering, North Carolina
State University, Raleigh, NC 27695.
2
Department of Mechanical Engineering, University of Michigan, Ann Ar-
bor, MI 48109.
3
High Temperature Materials Laboratory, Oak Ridge National Laboratory,
Oak Ridge, TN 37831.
4
Department of Materials Science and Engineering, North Carolina State
University, Raleigh, NC 27695.
Finite-Element Modeling
The two-dimensional plane stress six-node and twelve degree-
of-freedom triangular linear elastic element was selected to model
the flat-plate tension specimens. The commercially available AN-
SYS finite-element analysis software and its mesh generator were
used. Figure 2 shows the finite-element meshes used to model the
four types of ASTM D 638 tension specimens.
Due to the symmetry of the specimen geometry, only one quar-
ter of the specimen is modeled. Point O is the center of the speci-
men. Lines OA and OE are the symmetry planes of the specimen.
Point O is fixed in both the X- and Y-directions. Nodes on Line OA
have zero displacement in the X-direction and are free to move in
the Y-direction. Nodes on Line OE have zero displacement in the
Y-direction and are free to move in the X-direction. Nodes on Line
ED are prescribed a uniform 0.1-mm displacement in the X-direc-
tion and zero displacement in the Y-direction. This is intended to
simulate the holding of both ends in the wide section of the tension
specimen using perfectly rigid grips. The extension of the speci-
men under this condition is 0.2 mm. In actual tensile tests, the spec-
imen is clamped and pulled using a set of mechanical or hydraulic
grips. The actual loading condition near the gripped area is there-
fore the combination of the normal compressive stress and in-plane
tensile and shear stresses.
Dimensions of finite-element meshes and the number of ele-
ments and nodes are summarized in the table in Fig. 2 for each type
of ASTM D 638 specimen. OA is one half of the width of the gage
2 JOURNAL OF TESTING AND EVALUATION
FIG. 1Pictures of ASTM D 638-98 Type IV tension specimens: (a) Ny-
lon-11 bonded Nd-Fe-B (59.7 vol%) material tested with at 100C, and (b)
pure Nylon-11 tested at 100C.
FIG. 2Finite-element mesh for four types of specimen in ASTM D 638-98.
FIG. 3Geometric model of the single and double radius transitional
area.
section, DE is one half of the width of the wide section, OE is one
half of the length of the specimen, and AB is one half of the length
of the gage section. Between Points B and C, the specimen changes
from the narrow to the wide section. As shown in Figs. 2a, 2b, and
2c, Type I, II, and III specimens consist of a circular arc of radius
r
1
between Points B and C. These are known as single-arc speci-
mens. As shown in Fig. 2d, two radii, r
1
and r
2
, are used to blend
two arcs in the transitional region between Points B and C for the
Type IV specimen. This is designated as the double-arc specimen.
The geometrical relationships of the single- and double-arc
shape between Points B and C are shown in Fig. 3; w
1
and w
2
are
widths of the narrow and wide section, respectively. As shown in
Fig. 3a, the single-radius specimen has a continuous slope at Point
B and a discontinuous slope at Point C. The double-arc specimen,
as shown in Fig. 3b, has a continuous slope at all three transitional
points: B, F, and C. Arcs BF and FC have the same angle, indicated
by . The center of Arc BF, denoted as C
1
in Fig. 3a, has the same
X-coordinate as Point B. Similarly, the center of Arc CF, denoted
as C
2
, has the same X-coordinate as Point C. These geometrical
constraints together ensure a continuous slope at Points B, F, and C
for the double-arc specimen.
A hydrostatic stress-dependent yield criterion is usually used to
analyze stress concentrations in plastics [12,13]. The yield criterion
for magnetic particle reinforced plastics has not yet been studied.
The maximum von Mises stress was used to calculate the stress con-
centration factor and to identify the location that possibly initializes
the fracture. In addition to the von Mises stress, the maximum prin-
cipal stress, which is more suitable for the analysis of brittle materi-
als, was also applied to calculate the stress concentration factor.
The finite-element analysis provided results of stress components
and the von Mises and principal stress for each node. The nodal
stress was used to calculate the stress concentration factor, K
f
.
K
f

no
m
m
a
i
x
nal
(1)
where
max
is either the maximum nodal von Mises stress or max-
imum first principal stress, and
nominal
is either the average von
Mises stress or the average first principal stress on nodes in the
gage section. Finite-element analysis results show that the variation
of von Mises stress among nodes in the gage section is very small,
less than 1.5%. An elastic modulus of 130 MPa and Poissons ratio
of 0.3 were used in the finite-element analysis.
GARRELL ET AL. ON ASTM 638 TEST SPECIMENS 3
FIG. 4Distribution of von Mises stress in ASTM D 638 tensile test specimens. (Value by the node represents the ratio of nodal von Mises stress rela-
tive to the maximum von Mises stress at node MX in Pa.)
Finite-Element Analysis of Four Types of Tension Specimens
Figure 4 shows the distribution of von Mises stress for ASTM D
638 Type I, II, III, and IV standard tension specimens. For each
specimen, an overview and another enlarged view of the von Mises
stress distribution is presented. The enlarged view is centered about
Point B to identify the increase in von Mises stress. In Fig. 4, MX
and MN mark the nodes with maximum and minimum von Mises
stress, respectively. As shown in the overview of the specimen,
Node MN is close to Point C. In the enlarged view, the location of
Node MX relative to Node B can be seen. Node MX is next to Node
B, as in Type I and II, or a node away from Node B, as in Type III
and IV. It is noted that the mesh shown in the enlarged view repre-
sents the exaggeratedly deformed finite-element mesh from the
ANSYS output. Node B, originally located at the intersection of a
straight line and an arc, can be seen on a slightly curved region af-
ter deformation.
The ratio of nodal von Mises stress relative to the maximum von
Mises stress at Node MX is also presented in the enlarged view on
nodes around Points B and MX in Fig. 4. For Type I, II, and III
specimens, the overview of von Mises stress distribution does not
show the stress concentration. Upon closely examining the distri-
bution of von Mises stress around Point MX in the enlarged view,
all four types of ASTM D 638 tension specimens demonstrate dif-
ferent levels of stress concentration at Node MX.
The von Mises stress at Node MX (
max
) and at nodes in the gage
section (
nominal
) were extracted from the finite-element output and
substituted into Eq 1 to calculate the stress concentration factor, K
f
.
Results of K
f
are listed in Table 1. Type IV and II specimens have the
highest and lowest K
f
, 1.112 and 1.014, respectively. It was found that
K
f
has an almost linear relationship versus the parameter w
1
/r
1
, as
shown in Fig. 5, for the four types of ASTM D 638 specimen.
4 JOURNAL OF TESTING AND EVALUATION
TABLE 1The stress concentration for different single- and double-
radius tension specimen geometry.
Designation r
1
(mm) r
2
(mm) w
1
/r
1
K
f
ASTM Type I 76.0 . . . 0.171 1.046
D 638 Type II 76.0 . . . 0.079 1.014
Type III 76.0 . . . 0.250 1.068
Type IV 14.0 25.0 0.429 1.112
Single S-I 6.5 . . . 0.923 1.245
radius S-II 9.0 . . . 0.667 1.174
tension S-III 13.0 . . . 0.462 1.119
specimen S-IV 19.0 . . . 0.316 1.083
S-V 28.0 . . . 0.214 1.056
S-VI 39.0 . . . 0.154 1.040
S-VII 52.0 . . . 0.115 1.030
Double D-I 9.0 30.0 0.667 1.181
radius D-II* 14.0 25.0 0.429 1.112
tension D-III 17.0 22.0 0.353 1.091
specimen D-IV 19.5 19.5 0.308 1.081
D-V 22.0 17.0 0.273 1.072
D-VI 25.0 14.0 0.240 1.061
D-VII 30.0 9.0 0.200 1.051
* The same geometry as Type IV.
FIG. 5The stress concentration factor versus ratio of gage section width and radius for five types of ASTM D 638 specimens and the single- and dou-
ble-radius specimens.
Tension specimens with low w
1
/r
1
ratio usually have large arc
radius and are longer than specimens with high w
1
/r
1
. For example,
as shown in Fig. 2 and Table 1, Type II and IV specimens have the
same width in the gage section and different radius r
1
. The Type II
specimen has a larger arc radius, which makes the specimen longer
and has lower w
1
/r
1
and K
f
. Longer and bigger tension specimens
are unattractive because they require more material and a bigger die
for injection molding and cost more to produce.
Instead of using the von Mises stress, the maximum principal
stress and nominal principal stress were extracted from the finite-
element output and substituted into Eq 1 to calculate K
f
. For all four
types of specimens, the difference in K
f
based on von Mises and
maximum principal stresses is very small, less than 0.001.
Finite-Element Analysis of Single- and Double-Arc
Tension Specimens
In this section, the investigation of stress distribution in alterna-
tive tension specimen geometries that maintain the same overall
size as the Type IV specimen is discussed. The objective is to iden-
tify the geometric parameters that would reduce the magnitude of
K
f
. The width (w
1
) and length of the gage section are 6 and 13 mm,
respectively. The width of the wide section, w
2
, is 19 mm.
Two groups of sample geometry were studied. One group con-
sists of single-arc specimens, similar to those in Type I, II, and III.
Another group has the double-arc geometry, similar to the Type IV
specimen.
Single-Arc Specimens
As shown in Table 1, seven single radius curves are used to
blend between Points B and C. The minimum radius of Arc BC is
6.5 mm ( (w
2
w
1
)/2). The specimens with radii of 6.5, 9, 13,
19, 28, 39, or 52 mm at Arc BC were designated as S-I to S-VII, re-
spectively. The shape and finite-element mesh of three selected
specimens, S-I, S-IV, and S-VII, are shown in Fig. 6. For speci-
mens with arc radius equal to or less than 39 mm, the overall length
of the original Type IV specimen can be maintained. The K
f
of the
39-mm radius single-arc specimen (S-VI), as shown in Table 1, is
1.040. This is significantly lower than K
f
of 1.112 of the Type IV
specimen.
Plots of K
f
versus w
1
/r
1
for these seven specimens are shown in
Fig. 5. The larger radius of arc BC lowers both w
1
/r
1
and K
f
. Fig-
ure 5 also shows that the relationship between K
f
and w
1
/r
1
is al-
most linear and matches the trend of Type I to IV specimens.
Double-Arc Specimens
Seven double-arc specimens, denoted as D-I to D-VII, were stud-
ied as well. As shown in Fig. 3, the values of r
1
and r
2
in the double-
arc geometry can be reversed and still maintain the continuity of
both position and slope at Points B, F, and C. This special geomet-
rical characteristic of the double-arc shape was used to define con-
jugate pairs of specimens. As shown in Table 1, D-I (r
1
9 mm, r
2
30 mm) and D-VII (r
1
30 mm, r
2
9 mm) form a conjugate
pair of double-arc specimens. D-II (r
1
14 mm, r
2
25 mm) and
D-VI (r
1
25 mm, r
2
14 mm) as well as D-III (r
1
17 mm, r
2
22 mm) and D-V (r
1
22 mm, r
2
17 mm) are another two sets
of conjugate double-arc specimens. It is noted that the D-II speci-
men has the same geometry as the ASTM D 638 Type IV specimen.
The intent is to investigate the change in K

of a conjugate double-
arc specimen. The D-III specimen has the same r
1
and r
2
, both equal
to 19.5 mm. The shape and finite-element mesh of three selected
specimens, D-I, D-IV, and D-VII, are shown in Fig. 6.
Finite-element analysis results of K
f
versus w
1
/r
1
for seven dou-
ble-arc specimens are listed in Table 1 and plotted in Fig. 5. It is
important to note that the value of r
2
does not affect K
f
. By just
switching the two arcs of the original Type IV specimen, the K
f
was
reduced from 1.112 to 1.061. By further increasing r
1
to 30 mm
without changing the overall size of the Type IV specimen, K
f
was
reduced to 1.051.
Modeling of the Stress Concentration Factor
As shown in Fig. 5, a straight line can be used to determine K
f
for various w
1
/r
1
. Linear regression analysis was performed on all
17 data points, including the four types of specimen specified by
ASTM D 638 and the seven single-arc and six double-arc speci-
mens. The following model is obtained:
K
f
0.268 (w
1
/r
1
) 0.998 (2)
Equation 2 can be used to predict the K
f
based on w
1
/r
1
and may
be useful for new tension specimen design.
GARRELL ET AL. ON ASTM 638 TEST SPECIMENS 5
FIG. 6Finite-element meshes for selected single- and double-arc specimens (3 mm half-width in the gage section for all specimens).
The linear trend indicates the possibility that an analytical solu-
tion for the stress distribution in this geometrical configuration may
exist. This requires further investigation.
Concluding Remarks
The stress distribution of four types of tension specimens rec-
ommended in ASTM D 638 was investigated through the use of the
finite-element technique. It was found that the location of maxi-
mum von Mises stress for Type IV geometry coincided with the lo-
cation of experimentally observed failures in brittle specimens
made of Nylon-11 bonded Nd-Fe-B particles. This study identified
a solution to reduce the magnitude of stress concentration by in-
creasing the radius of the arc in the transitional area. A linear cor-
relation of stress concentration factor K
f
versus w
1
/r
1
was identified
and a linear model was established to predict the K
f
based on w
1
/r
1
.
Results obtained in this research can help establish standards and
aid in the design of specimen geometry for tension testing of brit-
tle particle-reinforced composites.
Acknowledgments
A portion of this research was sponsored by the User program of
the High Temperature Material Laboratory of Oak Ridge National
Laboratory and by National Science Foundation Grant 9983582
(Dr. K. P. Rajurkar, Program Director).
References
[1] ASTM D 638-99: Standard Test Method for Tensile Strength
of Plastics, Annual Book of ASTM Standards, ASTM Interna-
tional, West Conshohocken, PA, 1999.
[2] Garrell, M. G., Ma, B. M., Lara-Curzio, E., Shih, A. J., and
Scattergood, R. O., Mechanical Properties of Nylon Bonded
Nd-Fe-B Permanent Magnets, Journal of Magnetism and
Magnetic Materials, accepted for publication.
[3] Garrell, M. G., Ma, B. M., Lara-Curzio, E., Shih, A. J., and
Scattergood, R. O., Mechanical Properties of Polypheny-
lene-Sulfide (PPS) Bonded Nd-Fe-B Permanent Magnets,
Materials Science and Engineering: A, submitted.
[4] Garrell, M. G., Mechanical Properties of Injection Molded
Permanent Magnets, M.S. thesis, North Carolina State Uni-
versity, Raleigh, NC, April 2002.
[5] Pilkey, W. D., Petersons Stress Concentration Factors, 2nd
ed., John Wiley & Sons, New York, 1997.
[6] Wilson, L. H. and White, D. J. Stress-Concentration Factors
for Shoulder Fillets and Grooves in Plates, Journal of Strain
Analysis, Vol. 8, 1973, pp. 4351.
[7] Frocht, M. M. Factor of Stress Concentration Photoelasti-
cally Determined, Transactions of ASME, 1935, Vol. 57, pp.
A-67A-68.
[8] Worthem, D. W., Flat Tension Specimen Design for Ad-
vanced Composites, NASA Technical Report CR-185261,
NASA Lewis Research Center, Cleveland, OH, 1990.
[9] Shi, J., Brady, L., and Ireson, A., Design of Room Temper-
ature Tension Specimen Made of Ceramic Matrix Compos-
ites, Key Engineering Materials, Vol. 164165, Trans Tech
Publications, 1999, pp. 257260.
[10] Wang, Z, Chiang, C. K., and Chuang, T.-J., Optimum De-
sign of a Ceramic Tensile Creep Specimen Using a Finite
Element Method, Journal of Research of the National In-
stitute of Standards and Technology, Vol. 102, 1997, pp.
1528.
[11] Oplinger, D. W., Parker, B. S., Gandhi, K. R., Lamothe, R.,
and Foley, G., On the Streamline Specimen for Tension
Testing of Composite Materials, Recent Advances in Com-
posites in the United States and Japan, ASTM STP 864, J. R.
Vinson and M. Taya, Eds., ASTM International, West Con-
shohocken, PA, 1985, pp. 532555.
[12] McCrum, N. G., Buckley, C. P., and Bucknall, C. B., Princi-
ples of Polymer Engineering, 2nd ed., Oxford University
Press, Oxford, 1997.
[13] Courtney, T. H., Mechanical Behavior of Materials, 2nd ed.,
McGraw-Hill, New York, 2000.
6 JOURNAL OF TESTING AND EVALUATION