On The Difference Between Flexural Moduli Obtained by Three-Point and Four-Point Bending Tests

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Polymer Testing 25 (2006) 214–220

www.elsevier.com/locate/polytest

Test Method

On the difference between flexural moduli obtained by three-point


and four-point bending tests
F. Mujika
Department of Mechanical Engineering, Polytechnical University College, University of the Basque Country, Plaza de Europa,
1, 20018 San Sebastián, Spain
Received 12 September 2005; accepted 21 October 2005

Abstract
It has been experimentally seen that flexural moduli obtained by three-point and four-point bending tests are different for the
same specimen. The slope of the load-displacement curve increases as load increases in both tests, showing an apparent stiffening
of the specimen. The present work analyses the effect of the variation of the support span and the load span caused by the variation
of the contact zone between the specimen and support and load rollers. These effects have been analysed by classical beam theory
without taking into account shear effects. Experimental differences greater than 5% for the bending modulus have been obtained
for the same specimen tested in three-point and four-point bending, using two specimens of different carbon/epoxy composite
materials. After corrections based on the analysis developed in this work, the relative differences between three-point and four-
point moduli for the same specimen were under 1% for both specimens.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Three-point bending; Four-point bending; Bending modulus; Unidirectional composite; Carbon/epoxy

1. Introduction configuration due to the bending rotation at supports


as a problem where the superposition principle is not
Bending tests are used for determining mechanical applicable. Theobald et al. [2] carried out an
properties of unidirectional composite materials. Due to experimental analysis in order to analyze the influence
the important influence of shear effects in the displace- of load and geometric configuration in bending tests.
ments, great span-to-depth ratios are used in order to The distance between loading noses or load span in a
eliminate these effects. Three-point and four-point test four-point bending test was varied in order to study the
configurations are used in order to obtain flexural influence of this factor on the flexural strength and
strength and flexural modulus. The rotation of the modulus. Experimental results showed that bending
cross sections in the deformation process leads to the modulus varied in a significant manner when load span
contact zone between specimen and cylindrical supports varied. Theocaris et al. [3] investigated the three-point
changing in a three-point bending test. Furthermore, in a bending test at large deflections including friction
four-point bending test the contact between specimen forces at the supports, axial forces along the beam and
and cylindrical loading noses also changes. the effect of span shortening due to roller supports.
Timoshenko [1] included the effect of the variation Hayat and Suliman [4] performed tensile, three-point
of the support span in a the three-point test bending and four-point bending tests on glass
reinforced phenolic laminates. The specimens used in
E-mail address: impmugaf@sp.ehu.es. four-point bending were wider than the ones in three
0142-9418/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymertesting.2005.10.006
F. Mujika / Polymer Testing 25 (2006) 214–220 215

Nomenclature
a0, b0, L 0 dimensions in the undeformed RA, RC radius at support and at loading noses,
configuration respectively
a, b, L dimensions in the deformed configuration Ef flexure modulus
qA, qB, qC, qD bending angles in the basic system k3p, k4p constants related to error in three-point and
dC, dD displacements in the basic system four-point bending tests, respectively
I moment of inertia with respect to the middle 33p, 34p maximum strains in three-point and four-
plane point bending tests, respectively
P applied load in three-point bending q4p
A ; q4p
C bending angles at supports and at load
QZP/2 applied load at each nose in four-point application points in four-point bending,
bending respectively
w, h width and thickness of the specimen, hA, hC, h1, h2 terms related to support span and load
respectively span reduction in a four-point bending test
q3p bending angle at supports in a three-point x1, x2, x small quantities
bending test d4p
C displacement of the load application points
d3p displacement of the middle point in a three- in a four-point bending test
point bending test r, D geometric factors related to support and
h3p term related to support span reduction in a load application radii
three-point bending test

point bending. Modulus obtained by three-point and In the case of three-point bending, solutions can be
four-point bending tests were different, but the higher obtained by replacing a0Zb0ZL0/2. For four-point
values did not correspond always to the same method. bending, displacements and angles can be obtained
Brancheriau et al. [5] analyzed the influence of the using the principle of superposition. The results needed
shear force, supports and loading head indentation in for the basic system can be obtained, for instance, using
three-point and four-point bending tests on wooden the conjugate beam method without considering shear
samples. The same specimens were tested by both kind effects. The angles and displacements necessary for
of test with maximum loads much lower than those analyzing the cases of three-point and four-point
corresponding to failure. They stated that a three-point bending are:
bending test under-estimates the modulus of elasticity Rotated angles at supports A and B
value in relation to a four-point bending test.
Pa0 b0 ðL0 C b0 Þ
In the present work, the chord slope of the load- qA Z (1)
displacement curve between 0.1 and 0.3% strain points 6L0 Ef I
has been used for modulus calculation. In three-point
Pa0 b0 ðL0 C a0 Þ
bending and four point bending, the displacement used qB Z
has been that corresponding to the loading nose. 6L0 Ef I
Furthermore, quarter point loading has been used in Rotated angles at points C and D
four-point bending. As long as the span-to-depth ratio
Pa0 b0 ðb0 Ka0 Þ
used has been greater than 40, shear effects have not qC Z (2)
been considered. The hypothesis of small displace- 3L0 Ef I
ments has been assumed. Thus, the rotated angles in
bending have been considered small quantities and
second and higher order terms related to these angles
have been neglected.

2. Displacements and angles in the basic system

The system shown in Fig. 1 has been used as the


basic system for displacement and angles calculation. Fig. 1. Basic configuration for angles and displacements calculation.
216 F. Mujika / Polymer Testing 25 (2006) 214–220


Pa0 b20 C 2a0 b0 K3a20 where RA is the support radius. The actual span L
qD Z during the test is
6L0 Ef I
Displacements at points C and D L Z L0 K2h3p (6)

Pa20 b20 The displacement of the middle point in a three-


dC Z (3) point bending test when shear effects are not considered
3L0 Ef I
can be obtained from either dC or dD of Eq. (3) for the

Pa20 b20 C 2a0 b0 Ka20 actual support span, being
dD Z
6L0 Ef I PL3
d3p Z (7)
where Ef is the flexure modulus; I the moment of inertia 4Ef wh3
with respect to the middle plane: IZwh3/12; w the
Replacing Eqs. (6) and (7) and neglecting small
width of the specimen; and h the thickness of the
terms related to bending angles of higher order than 1,
specimen.
results in
 
PL30 h3p
3. Three-point bending d3p Z 1K6 (8)
4Ef wh3 L0
Fig. 2 shows a three-point bending test in the Taking into account that the unique variable in the
undeformed and deformed configurations. A reduction parenthesis of Eq. (8) is the load P, this equation can be
in the support span occurs in the deformed configur- written as
ation, due to the rotation at supports. The rotated angle
at supports in a three-point specimen can be obtained PL30
d3p Z ð1Kk3p PÞ (9)
from either qA or qB in Eq. (1) replacing a0Zb0ZL0/2, 4Ef wh3
obtaining
where
3PL20
q3p Z (4) 9L0 RA
4Ef wh3 k3p Z (10)
2Ef wh3
As long as the rotation angle is small, the span
corresponding to the undeformed configuration is used From Eq. (9), the difference of displacements
in Eq. (4). According to Fig. 2, the error in each support between 2 points is
due to the variation of the contact zone between the DPL30
specimen and the support is Dd3p Z ½1Kk3p ðP1 C P2 Þ (11)
4Ef wh3
h3p Z q3p RA (5)
From Eq. (11) the chord flexure modulus is
m3p L30
Ef Z ½1Kk3p ðP1 C P2 Þ (12)
4wh3
where m3pZDp/Dd3p is the chord slope between the
considered points.
Eq. (12) can be written as

Ef Z E3p ½1Kk3p ðP1 C P2 Þ (13)

where
m3p L30
E3p Z (14)
4wh3
is the modulus usually calculated in three-point flexure.
Considering that it is a small quantity, the maximum
Fig. 2. Undeformed and deformed three-point bending test strain in a three-point bending test is calculated using
configuration. the initial span as
F. Mujika / Polymer Testing 25 (2006) 214–220 217

3PL0 h2 Z hA KhC
33p Z (15)
2Ef wh2
According to Eq. (17) and Fig. 3, supports span
Extracting P from Eq. (15) and replacing in Eq. (13) decreases and load span increases in a four-point
taking into account Eq. (10) results in bending test. The error terms at supports and at load
 
3R  applications points are
Ef Z E3p 1K A 33p 1 C 3 3p
2 (16)
h hA Z q4p
A RA (18)
Eq. (16) shows that the error due to span variation
only depends on the strain range adopted for modulus hC Z q4p
C RC
calculation, the radius of the supports and the thickness
of the specimen. where RC is the radius of the load noses. The angle at
supports qA and the angle at load application points qC
are obtained from Eqs. (1) and (2) using the principle of
4. Four-point bending superposition as
There are two sources of error in the case of four- Qa0 b0
q4p
A Z (19)
point bending: the contact between specimen and 2Ef I
loading noses and the contact between the specimen
and supports. Fig. 3 shows a four-point bending test in Qa0 ðb0 Ka0 Þ
the undeformed and deformed configurations. The q4p
C Z
2Ef I
dimensions after deformation and before deformation
are related as where QZP/2 is the load applied at each nose. The
L Z L0 K2hA (17) displacement at load application points is obtained from
Eq. (3) using the principle of superposition as
a Z a0 KhA KhC Z a0 Kh1 Qa2
d4p
C Z ð3b2 C 2abKa2 Þ (20)
h1 Z hA C hC 6LEf I

Replacing a, b and L from Eq. (17) results in


b Z b0 KhA C hC Z b0 Kh2
Qða0 Kh1 Þ2
d4p
C Z ½3ðb0 Kh2 Þ2 C 2ða0 Kh1 Þ
6ðL0 K2hA ÞEf I

!ðb0 Kh2 ÞKða0 Kh1 Þ2  (21)

Neglecting terms of higher order than 1 in the


parentheses of Eq. (21) results in

Q a20 ð1Kx1 Þ  2
d4p
C Z 3b0 C 2a0 b0 Ka20 K2h1
6Ef I L0 ð1Kx2 Þ (22)
!ðb0 Ka0 ÞK2h2 ð3b0 C a0 Þ

where x1Z2(h1/a0) and x2Z2(hA/L0)


Neglecting terms of higher order than 1 in Eq. (22)
and taking into account that when x/0, 1/1KxZ1C
xCx2C. results in

1Kx1
Z 1 C x2 Kx1 Z 1Kx (23)
1Kx2

where xZx1Kx2
Fig. 3. Undeformed and deformed four-point bending test Replacing Eqs. (22) and (23) and neglecting second
configuration. order terms, results in
218 F. Mujika / Polymer Testing 25 (2006) 214–220

Q a20   From Eq. (13) the chord flexure modulus is


d4p
C Z ð1KxÞ 3b20 C 2a0 b0 Ka20
6Ef I L0 m4p L30
Ef Z ½1Kk4p ðP1 C P2 Þ (32)
K2h1 ðb0 Ka0 ÞK2h2 ð3b0 C a0 Þ (24) 8wh3
where m4p Z DP=Dd4p C is the chord slope between the
Eq. (24) is valid for any load span in a four-point
considered points.
bending. In the experimental part of the present work
Eq. (32) can be written as
quarter point loading has been used, being a0ZL0/4 and
b0Z3L0/4. Replacing these values and taking into Ef Z E4p ½1Kk4p ðP1 C P2 Þ (33)
account that QZP/2 results in
   where
PL30 3hA 2hC
d4p
C Z 1K3 C (25) m4p L30
8Ef wh3 L0 L0 E4p Z (34)
8wh3
In order to calculate hA and hC, replacing the is the modulus usually calculated in four-point bending
mentioned values of a0 and b0 in Eq. (19), the angles at when the displacement of the load application points is
points A and C are used.
9PL20 The maximum strain in a four-point bending test
q4p
A Z (26) considering that it is a small quantity is calculated using
16Ef wh3
the initial span as
3PL20 3PL0
q4p
C Z
34p Z (35)
8Ef wh3 4Ef wh2
From Eqs. (18) and (26) the ratio between the terms Extracting P from Eq. (35), and replacing in Eq. (33)
hC and hA is taking into account Eq. (30) results in
 
hC 2 9DRA  4p 4p
Z r (27) Ef Z E4p 1K 31 C 32 (36)
hA 3 4h
where rZRC/RA is the ratio between the loading noses Eq. (36) shows that the error due to the load span and
and the support radius. support span variations depends on the strain range
From Eqs. (18), (25) and (27) the displacement at adopted for modulus calculation, on the radius of the
load application points is supports and the loading noses and on the thickness of
  the specimen.
PL30 hA When the same strain range is used in four-point and
d4p
C Z 1K3 D (28)
8Ef wh3 L0 three-point bending tests, the ratio between error terms
in Eqs. (16) and (36) is:
where DZ(9C4r)/3
Taking into account that the unique variable in the 9DRA =4h 3
Z D (37)
second term of the parenthesis in Eq. (28) is P, this 3RA =h 4
equation can be written as
The ratio in Eq. (37) only depends on the ratio
PL30 between loading noses and supports radii. When the
d4p
C Z ð1Kk4p PÞ (29) load and supports have the same radius, i.e. rZ1 the
8Ef wh3
ratio in Eq. (37) is 3.25. Even in the limit case that r/0
where the ratio in Eq. (37) is 2.25. Therefore, the error in
27L0 DRA modulus calculation in four-point bending is much
k4p Z (30) greater than the corresponding error in three-point
16Ef wh3
bending. It is worth underlining that this fact is
According to Eq. (29) the increment of displacement independent of material properties, support span and
between two points is specimen dimensions; it only depends on the parameter
D, related to the radii of loading noses and supports.
DPL30 The error corresponding to third point loading in
Dd4p
C Z ½1Kk4p ðP1 C P2 Þ (31)
8Ef wh3 four-point bending can be determined in a similar
F. Mujika / Polymer Testing 25 (2006) 214–220 219

manner than the followed in this section replacing the (26) and (35) as
values a0ZL0/3 and b0Z2L0/3 in Eqs. (19) and (24). 3L0 4p
q4p
A Z 3 (40)
4h
5. Experimental procedure Having fixed maximum strain at 0.3% in both cases,
the angle for the thinnest specimen in Eq. (39) is 0.08
5.1. Materials, tests and apparatus rad and in Eq. (40) is 0.12 rad. Thus, the hypothesis of
small displacements is satisfied in three-point bending
AS4/8552 and IM7/8552 carbon/epoxy uni- and in four-point bending for both specimens.
directional composite materials from Hexcel Compo-
sites have been used in order to check the analytic 5.3. Strains, loads and displacements
predictions. One specimen of each material with fibres
oriented longitudinally was tested in three-point and When two strain points are used for the chord
four-point bending in a universal testing machine, modulus calculation, loads and displacements corre-
INSTRON 4206, at a displacement rate of 1 mm/min. sponding to both points are necessary. As strain gauges
The data acquisition system was Testworks 4 from are not usually used in bending tests, the value of strain
MTS. Five tests were performed for each specimen. can be obtained approximately from the displacement
The radius of supports and loading noses were: according to Eqs. (9) and (15) in three-point bending
Supports RA Z 2 mm and Eqs. (29) and (35) in four-point bending, without
considering span variations. The relation between strain
Loading noses RC Z 5 mm and displacement for both cases is

The length of both specimens was 150 mm, the 6dh


3Z (41)
support span was 100 mm and in the case of four-point L20
bending quarter loading was used. The dimensions of
where 3 is the maximum strain in three-point and four-
the specimens were:
point bending (33p or 34p) and d is the displacement of
AS4=8552 w Z 16:6 mm h Z 1:81 mm the loading nose in both configurations (d3p or d4p
C ). In
spite of the displacements measured by the machine
IM7=8552 w Z 15:2 mm h Z 2:37 mm having errors related to the driving system and possible
indentation of the specimen, considering that these
5.2. Small displacements condition errors do not vary in a significant manner between the
two points adopted, they are partially eliminated when
ISO 14125:1998 [6] specifies that the ratio of the difference between displacements is used for chord
maximum displacement to span must be less than slope calculation. Nevertheless, Eqs. (16) and (36) are
10% for considering small displacements. From Eqs. not recommended for modulus correction, as long as
(4) and (7) without considering the change in span, the error in strain caused by the mentioned errors of
results in displacement data are additive. Eqs. (13) and (33) are
preferred for modulus correction as long as they are
d3p q3p
Z ! 0:1 (38) based on load data, in spite of an iterative procedure
L0 3 being necessary due to Ef not being known, as
Therefore, the maximum bending angle must be less explained below.
than 0.3 rad (178). This angle limitation can be
considered also for four-point bending. In the case of 6. Results and discussion
three-point bending, the relation between the angle at
supports and the maximum deformation can be Replacing the initial modulus E3p and E4p of Eqs.
obtained from Eqs. (4) and (15) as (14) and (34) obtained from tests in Eqs. (10) and (30)
as flexural modulus Ef, the constants k3p and k4p,
L0 3p
q3p Z 3 (39) respectively, were obtained. Replacing these constants
2h in eqs. (13) and (33), the corrected modulus
In the case of four-point bending, the relation corresponding to three-point and four-point were
between the angle at supports and the maximum obtained, respectively. The process was repeated until
deformation can be obtained from the first of Eqs. constant values of flexural modulus were obtained.
220 F. Mujika / Polymer Testing 25 (2006) 214–220

Table 1 bending. Having fixed the strain range for modulus


Uncorrected and corrected flexure modulus obtained by three-point determination, error terms do not depend on material
and four-point bending tests
properties or load span used. They only depend on the
Experimental Corrected thickness of the specimen and on the radius of supports
E3p (MPa) E4p (MPa) Ef3p (MPa) Ef4p (MPa) and loading noses.
AS4/8552 131,077 137,802 129,428 129,551
Differences in experimental modulus calculated
(G128) (G211) without correction of greater than 5% for the same
(E4pKE3p)/ 5.1% 0.1% specimen tested in three-point and four-point bending
E3p!100 decrease to below 1% after corrections considered in
IM7/8552 150,813 158,866 149,116 150,355 the present work.
(G227) (G151)
(E4pKE3p)/ 5.3% 0.8% If modulus calculation is carried out without
E3p!100 corrections, both test methods, but particularly four-
point bending, overestimate the flexure modulus.
Table 1 shows the initial modulus E3p and E4p, and
the constant values Ef3p and Ef4p after the iterative Acknowledgements
process applied to the mean values. For both specimens
the constant value was obtained after two iterations in The author wish to thank the Ministry of Education
three-point bending and after four iterations in four- and Science of Spain for its financial support on the
point bending. After the first iteration step the values research project DPI 2004-02642, ‘ Flexure and
obtained were very close to the end constant value for Interlaminar Fracture Behaviour of Carbon/Epoxy
both test configurations. Multidirectional Laminates’.
According to Table 1, the initial difference between
modulus obtained by three-point and four-point References
bending is above 5% in both cases and is reduced
[1] S. Thimoshenko, Strength of Materials, Part 2 (Spanish edition),
below 1% after the correction process. The final
Espasa-Calpe, Madrid, 1978.
difference is greater for the thicker specimen (0.8%), [2] D. Theobald, J. McClurg, J.G. Vaughan. Comparison of Three-
probably due to the greater influence of shear effects in Point and Four Point Flexural Bending Tests, International
the displacements. Composites Expo 1997, Washington, 1997, pp. 1–9.
[3] P.S. Theocaris, S.A. Paipetis, S. Paolinelis, Three-point bending
7. Conclusions at large deflections, Journal of Testing and Evaluation 5 (6)
(1977) 427–436.
[4] M.A. Hayat, S.M.A. Suliman, Mechanical and structural proper-
Due to the bending rotation at supports, support span ties of glass reinforced phenolic laminates, Polymer Testing 17
decreases in both, three-point and four-point bending (1998) 79–97.
tests. Furthermore, due to the rotation at loading noses [5] L. Brancheriau, H. Bailleres, D. Guitard, Comparison between
in a four-point bending test the load span increases in modulus of elasticity values calculated using 3 and 4 point
bending tests on wooden samples, Wood Science and Technology
such a test. 36 (2002) 367–383.
The effect of the mentioned span variations is much [6] ISO 14125:1998, Fibre-reinforced Plastic Composites, Determi-
greater in four-point bending than in three-point nation of Flexural Properties, 1998.

You might also like