Design and Implementation of a Force

Dynamometer for Friction Stir Welding

J. Y. Sheikh-Ahmad, Dima Ali & Fanyu

Meng

Arabian Journal for Science and

Engineering

ISSN 2193-567X

Arab J Sci Eng

DOI 10.1007/s13369-017-3007-z

1 23
Your article is protected by copyright and
all rights are held exclusively by King Fahd
University of Petroleum & Minerals. This e-
offprint is for personal use only and shall not
be self-archived in electronic repositories. If
you wish to self-archive your article, please
use the accepted manuscript version for
posting on your own website. You may
further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
 Author's personal copy
Arabian Journal for Science and Engineering
https://doi.org/10.1007/s13369-017-3007-z

RESEARCH ARTICLE - MECHANICAL ENGINEERING

Design and Implementation of a Force Dynamometer for Friction Stir

Welding
J. Y. Sheikh-Ahmad1 · Dima Ali1 · Fanyu Meng1

Received: 26 July 2017 / Accepted: 3 December 2017

© King Fahd University of Petroleum & Minerals 2017

Abstract
The locations of the strain nodes on the extended octagonal ring force transducer (EOR) have been examined by many studies
in the literature. However, there is a considerable lack of agreement between the different studies due to different assumptions
and approaches used to define nodal locations. Because an exact solution for the stress and strain distributions in the octagonal
ring does not exist, some researchers used the solution of the classical circular ring as a first approximation. Others used the
finite element method to obtain a solution. But very few studies attempted to investigate the effect of geometry of the ring
itself on the solution. In this work, the stress and strain distributions in an EOR transducers of varying geometries have been
investigated by the finite element method and the locations of strain nodes and those of maximum sensitivity were determined
and compared with the literature. It was found that strain node locations vary with ring geometry, but do not generally coincide
with the classical solution of the circular ring. The strain node locations are also not coinciding with maximum sensitivity
locations for horizontal force measurement. An EOR dynamometer was fabricated based on the results of this study. The
dynamometer was calibrated and tested in friction stir welding. It was shown that the dynamometer response is highly linear
and with very small cross talk.

Keywords Extended octagonal ring · Strain node · Cross talk · Sensitivity · Finite element analysis

1 Introduction on the inner and outer surfaces of the ring at these locations
then the horizontal and vertical forces can be measured inde-
A common low-cost device for measuring forces has been pendently and with high sensitivity [1]. The octagonal ring is
the thin circular ring, also known as the proving ring, which a variation of the proving ring where the outside cylindrical
can be used as a basis for the measurement of two orthog- surface of the ring is replaced by flat faces arranged octag-
onal forces in its plane. An exact solution for the stress and onally around the circumference. The use of an octagonal
strain distributions in the ring can be obtained using elasticity ring gives greater rigidity in the horizontal force direction
theory or strain energy methods. These solutions show that and higher stiffness for both forces. However, this comes
a vertical force applied in the plane of the ring and through at the expense of reducing sensitivity to about 62% of the
its center produces zero bending moment and zero strain at circular ring [1]. The flat outside surfaces make it easier to
the curved surface at an angle of 39.6◦ from the vertical. machine than the circular ring. It is also easier to mount
The application of a horizontal force at the top of the ring strain gages on flat surfaces. The extended octagonal ring
produces zero bending moment and hence zero strain on a transducer (EOR) shown in Fig. 1 gives even more rigidity
diameter parallel to its direction at +/−90◦ from the vertical. and stability to the force measuring rings. The middle part
The angles at which zero stress and strain occur are known as of the EOR is thick to support different loads and can be
nodal angles, φ. This means that if strain gauges are placed designed in various ways for different support and loading
situations. The EOR can generally be made with any width,
and the wider the cross section, the more resistant to extra-
B J. Y. Sheikh-Ahmad
neous forces and moments it will be. The thin ends of the
[email protected]
EOR support strain gages that elongate and contract with
1 Department of Mechanical Engineering, Petroleum Institute, the application of the loads which allow for the measure-
Part of Khalifa University of Science and Technology, ment of strains along their direction. The strain gages are
PO Box 2533, Abu Dhabi, United Arab Emirates

123
 Author's personal copy
Arabian Journal for Science and Engineering

V considerably as the ring was varied from “thin” to “thick”

Mantle member.
H In the present study, we investigate the effect of octago-
Welding direcon nal ring radius and thickness to radius ratio on strain node
locations and sensitivity of EOR transducer. Finite element
analysis was conducted in SolidWorks environment, and the
EOR transducer strain distribution in the sensing part of the octagonal ring
was determined for different combinations of thicknesses
and radii. The numerical analysis was validated by design,
construction and calibration of a 2-component EOR force
Base plate dynamometer. The use of the dynamometer for force mea-
surement in friction stir welding of high-density polyethylene
Fig. 1 Friction stir welding dynamometer was also demonstrated.

usually located at the strain nodes, and connected by isolated 2 Materials and Methods
Wheatstone bridges to allow for independent measurement
of moments and forces in different directions in the plane of 2.1 Design of Force Dynamometer
the ring with the least cross-sensitivity possible.
Unlike the circular ring, there is no exact solution for the The design of the dynamometer for force measurement dur-
stress and strain distributions in the octagonal ring or the ing friction stir welding is shown in Fig. 1. It consists of two
EOR because of the non-uniformity of its thickness. Differ- EOR transducers sandwiched between a base plate, which
ent authors attempted different approaches to determining is fixed to the machine tool table, and the mantle, on which
the locations of strain nodes on an EOR element. Many friction stir welding will take place. The workpieces to be
researchers used as a first approximation the node locations welded, which are in the form of thin plates, are clamped
for the circular ring in designing an EOR [2–4]. Loewen and on top of the mantle using the mounting holes so that the
Cook [5] utilized photoelasticity technique to obtain approxi- weld line joining the butting surfaces is at the center of the
mate solutions for the strain distribution in the octagonal ring. mantle and is aligned along the mantel’s length (see Fig. 8b).
They found that the position of zero strain under the action As the friction stir welding pin is plunged in and traversed
of the horizontal force occurred at an angle of approximately along the weld line, forces V and H are generated in the
50◦ from the vertical, instead of 39.5◦ found in the circular downward and feed directions, respectively, as shown in the
ring. However, it has been widely practiced to place strain figure. Depending on material type and thickness the down
gages at 45◦ for convenience and maximum sensitivity. The force could be several hundred Newtons. The dynamometer
nodes for vertical force measurement remain at 90◦ from the was constructed from AISI 1020 carbon steel. The EOR ele-
vertical [6–9]. Many studies have applied the finite element ments were cut by wire EDM to the desired shape as shown.
method in the design of EOR elements [10–15]. These stud- The surfaces were then sanded and polished for strain gage
ies were concerned with determining the design limits of the mounting. The strain gages are mounted on the inside and
EOR dynamometer and optimizing its dimensions. Despite outside surfaces of the octagonal rings of the transducer in
this extensive and wide research, there remains a great dis- specific locations such that they are sensitive to either the
agreement in the literature as far as the optimum locations vertical force or the horizontal force only. The locations of
for strain gages on an EOR. Chen et al. [11] determined the the strain gages for maximum sensitivity and least cross talk
nodes for measuring the vertical force to be at 100◦ and 105◦ between the two measuring Wheatstone bridges is achieved
from the vertical on the outside and inside surfaces, respec- by strain analysis and identification of the so-called strain
tively, which are different than the +/− 90◦ derived from nodes. In the present work, this is achieved by finite element
the circular ring equations. Girma [16] determined the loca- analysis.
tion of nodes for horizontal force measurement to be at +/−
38◦ instead of 45◦ or 50◦ while Nalavade et al. [12] obtained 2.2 Numerical Models
angles of +/− 34◦ . One reason for these discrepancies is
the effect of the EOR geometry, specifically the thickness to Node identification for the extended octagonal ring (EOR)
radius ratio, on nodal locations. It was shown by Kroencke transducer was conducted by the finite element analysis in
and Hull [10] and O’Dogherty [1] that thin and thick rings SolidWorks Simulation environment. Linear elastic strain
pose different behaviors when deformations and strains are analysis was performed and the locations for zero strain
considered. The nodal points for the vertical force varied for primary and secondary loads were determined. Figure 2

123
 Author's personal copy
Arabian Journal for Science and Engineering

horizontal loads applied separately. For each loading condi-

tion, the strains in the model global coordinate system are
determined along the edges 1–8 shown in Fig. 3. Because
strain gages measure strain along their length only, it is nec-
essary to obtain the strain component along this direction.
The strain gages are mounted on the external flat surfaces of
the octagonal ring as well as on the cylindrical inside surfaces
and are aligned with the directions 1–8. To obtain the strain
component in these directions, the following strain transfor-
mation equation was used:

εx + ε y εx − ε y γx y
εy = − cos 2θ − sin 2θ (1)
2 2 2

where ε y  is the strain along the face in local coordinate sys-

tem, εx , ε y and γx y are the normal shear strains in the global
coordinate system, respectively, and θ is the angle between
the x-axis and the normal to the surface on which the strain
gage is mounted, measured counterclockwise.
Fig. 2 Meshed details for the solid models for EOR force transducer

⑤
2.3 Parametric Study

To investigate the effect of EOR geometry on node locations,

⑥
models with different geometries were constructed and ana-
⑧ lyzed in SolidWorks. Table 1 shows the geometric parameters
⑦ of the EOR element. The most important parameters as indi-
cated in previous studies are the ring radius, r, and thickness
to radius ratio, t/r [1,10]. These were varied in three levels
①
Symmetr each according to a 32 factorial design. The combinations of
the factorial design are shown in the first two columns of the
y
② table. However, because the octagonal ring dimensions were
Fixed ④
x
restricted to whole numbers for ease of design and manufac-
z
turing, the actual levels for r/t used in the experiments were
③ slightly different than the ideal factorial design values. These
values are shown in the last column of Table 1. Moreover,
Fig. 3 Boundary conditions and loading details of the transducer
in the current octagonal ring design a thin ring geometry
was maintained in order to minimize strain variation across
shows the meshed solid model for the EOR transducer. The thickness. The width and center-to-center distance of each
element type used was linear tetrahedral solid element. Fine
and coarse mesh sizes were implemented to improve results
Table 1 Design of experiment for numerical analysis of EOR trans-
and resolution in the measuring region of the transducer. In ducer
the ring region, the element size was 1.0 mm, it was 3.0 mm SN r (mm) t/r t (mm) t/r (actual)
elsewhere, and the total number of elements was 165,886.
Figure 3 shows the loading and boundary conditions applied 1 20 0.20 4 0.20
to the EOR element. Because of symmetry of deformation 2 20 0.25 5 0.25
with respect to the midplane parallel to the x−y plane, only 3 20 0.35 7 0.35
half of the width of the EOR element was modeled and sym- 4 25 0.20 5 0.20
metry boundary condition was applied on the midplane of 5 25 0.25 7 0.28
symmetry. The transducer was completely fixed on the bot- 6 25 0.35 9 0.36
tom surface, and the load was applied on the top surface as 7 30 0.20 5 0.17
shown in the figure. This corresponds to the real boundary 8 30 0.25 7 0.23
conditions of the friction stir welding dynamometer as shown 9 30 0.35 11 0.37
in Fig. 1. The analysis was conducted with the vertical and

123
 Author's personal copy
Arabian Journal for Science and Engineering

element were kept the same at 45 and 150 mm, respectively. (a) 4.0E-04
③ ② ①
The width and thickness of the transducer directly influence 2.0E-04
the load capacity. The ring dimensions were selected so that

Strain (μm/μm)
0.0E+00
the load capacity of the dynamometer is in excess of 4 kN. -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70
-2.0E-04

-4.0E-04
3 Results and Discussion
-6.0E-04 εy’-v
ey'-v
3.1 Numerical Results εy’-h
ey'-h
-8.0E-04
Angle (Deg)
Simulation of the deflection of EOR under vertical and
horizontal loads, applied independently, was conducted in (b) 8.0E-04
④ εy’-v
ey'-v
SolidWorks Simulation environment. The applied load for 6.0E-04 εy’-h
ey'-h
each simulation was 2 kN. Strains εx , ε y and γx y were deter-
4.0E-04

Strain (μm/μm)
mined along the specified edges in the global coordinated
system and then were transformed to the local coordinate 2.0E-04

system using Eq. (1). Figure 4 shows the variation of the 0.0E+00
transformed strain ε y  along the different outside and inside -2.0E-04
-90 -70 -50 -30 -10 10 30 50 70 90

surfaces of the EOR-2 (r = 20 mm, t = 5 mm). These

-4.0E-04
surfaces correspond to the numbers 1–8 in Fig. 3. Similar
distributions were obtained for the other EOR geometries. -6.0E-04

Strains resulting from application of the vertical and hori- Angle (Deg)
zontal forces are labeled ε y  -v and ε y  -h, respectively. Strain
(c) 8.0E-04
nodes are determined from this and other similar figures as ⑤ ⑥ ⑦

those locations where the strain component crosses zero. The 6.0E-04 εy’-v
ey'-v
εy’-h
angle φ in this figure is measured from the horizontal in a ey'-h
Strain (μm/μm)

4.0E-04
counterclockwise direction. Figure 4a shows that no strain
2.0E-04
nodes exist on surface 1, quiet contrary to the common prac-
tice reviewed in the previous literature of using angle 45◦ or 0.0E+00
50◦ from the vertical for strain gage placement on this face. 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

Face 2 has a strain node at 7.2◦ where ε y  -h = 0 and ε y  -v is -2.0E-04

near its maximum and positive. This location will have high -4.0E-04
sensitivity to the vertical force but will not sense the hori- Angle (Deg)
zontal force. Similarly, face 3 has a node at −34.5◦ where
ε y  -v = 0 and ε y  -h is negative, but not close to its maximum. (d) 6.0E-04
⑧
This location is not sensitive at all to the vertical force but 4.0E-04

has moderate sensitivity to the horizontal force. Examining

Strain (µm/µm)

2.0E-04
the remaining plots in Fig. 4b–d allows the determination of
0.0E+00
all strain nodes for vertical and horizontal sensitivity. These 90 110 130 150 170 190 210 230 250 270
-2.0E-04
nodes are tabulated in Tables 2 and 3, respectively. The tables
also list the sensitivity at each nodal location. It is noted that -4.0E-04
the EOR generally has high sensitivity, and that the sensitiv- -6.0E-04 εy’-v
ey'-v
ity for the horizontal force is almost twice as much as that εy’-h
ey'-h
-8.0E-04
for the vertical force. In comparison, the sensitivities of an Angle (Deg)
EOR of similar r/t ratio were reported by Godwin et al. [17]
as 1.72E−7 and 1.91E−7 N−1 for the vertical and horizon- Fig. 4 Strain distribution on the surfaces of the octagonal rings on both
tal forces, respectively. These values are about 27% different sides of the transducer. Angle φ is measured counterclockwise from the
horizontal. a Strains on surfaces 1, 2 and 3. b Strains on surface 4. c
from the numerical values obtained in the present work. Strains on surfaces 5, 6 and 7. d Strains on surface 8
Examining Fig. 4 also shows that the nodal locations for
V and H are not the same as the locations for maximum
sensitivity (i.e., maximum strain). The nodal locations for from the conventional −40◦ by 3◦ –5◦ . Tables 4 and 5 list
the vertical force are shifted from the horizontal by 6◦ –7◦ , the locations for maximum sensitivity for the different faces
while the nodal locations for the horizontal force are shifted of the EOR and the corresponding sensitivities. The tables

123
 Author's personal copy
Arabian Journal for Science and Engineering

Table 2 Nodal locations for

Node φ (◦ ) Strain ε y -v Sensitivity (με/N) EOR surface
vertical force measurement for
EOR-2 1 7.2 2.086E−04 1.04E−07 2 outside
2 6.6 − 2.903E−04 − 1.45E−07 4 inside
3 172.8 2.085E−04 1.04E−07 6 outside
4 173.4 − 2.904E−04 − 1.45E−07 8 inside

Table 3 Nodal locations for

Node φ (◦ ) Strain ε y -h Sensitivity (με/N) EOR surface
horizontal force measurement
for EOR-2 5 − 34.5 − 3.598E−04 − 1.80E−07 3 outside
5 74.4 − 3.869E−04 − 1.93E−07 4 inside
6 − 36.9 4.945E−04 2.47E−07 4 inside
7 216.9 − 4.946E−04 − 2.47E−07 8 inside
8 214.5 3.599E−04 1.80E−07 7 outside
8 105.6 3.869E−04 1.93E−07 8 inside

Table 4 Maximum sensitivity

φ (◦ ) Strain ε y -v Strain εy  -h Sensitivity (με/N) EOR surface % Increase
for measuring vertical force
37.7 1.90E−04 2.259E−04 9.49E−08 1 outer
4.5 2.11E−04 − 2.971E−05 1.06E−07 2 outer 1.2
5 − 2.92E−04 1.833E−05 − 1.46E−07 4 inner 0.6

Table 5 Maximum sensitivity

φ (◦ ) Strain ε y -h Strain ε y -v Sensitivity (με/N) EOR surface % Increase
for measuring horizontal force
50.6 3.14E−04 1.308E−04 1.57E−07 1 outer
− 50.1 − 5.70E−04 − 1.639E−04 − 2.85E−07 3 outer 58.3
− 52.1 6.63E−04 1.643E−04 3.31E−07 4 inner 34.0
50.1 − 4.01E−04 − 2.006E−04 − 2.00E−07 4 inner 3.6

also list values of the secondary strain component at the same Figure 6 shows the numbering system of the different nodes
location (i.e., for the vertical strain as the primary, the hor- and the direction of measurement of the location angle φ.
izontal strain is the secondary). It is noted that the nodal It can be seen that nodal locations are greatly affected by
locations for the vertical force are shifted slightly from the the radius and thickness of the octagonal ring. Furthermore,
locations of maximum strain (about 1◦ ) and thus the sensi- nodal locations appear to vary linearly with ring thickness and
tivity at the nodal location is only slightly different from that radius, with possible interaction between the two for nodal
the maximum strain location. However, for the horizontal locations 5, 5 and 6 (for horizontal force measurement). To
force, the locations of maximum strain on faces 3 and 4 are determine the effect of geometry on nodal locations, non-
shifted considerably from the nodal locations. The locations linear regression of the results was conducted. The nodal
for maximum strain are at +/ − 50◦ from the horizontal (i.e., locations were fit to a partial second-order polynomial of the
+/ − 40◦ from the vertical, which is closer to the theoretical form:
angle of +/ − 39.6◦ reported for the circular ring). The dif-
ference in sensitivity between nodal locations and maximum ∅ = a0 + a1r + a2 t + a12 r t +  (2)
strain locations in this case varied from 34 to 58%. Therefore,
mounting strain gages at maximum strain locations for hor- where a0 , a1 , a2 , a12 are empirical constants and  is error.
izontal force measurement results in considerable increases The goodness of fit of the regression model is determined
in sensitivity. But this comes at the expense of greater cross by the regression coefficient Radj 2 . Tables 6, 7, 8 and 9
talk as it can be seen from the values of the counter strain show the model coefficients for the different nodes and the
component. This cross talk might reach as high as 50%. corresponding regression coefficients. Only statistically sig-
Figure 5 shows the effect of the EOR geometry on node nificant regression coefficients (with P < α = 0.05) were
locations for the different combinations of EOR transducers. shown in these tables. It can be seen that a partial second-

123
 Author's personal copy
Arabian Journal for Science and Engineering

(a) 10.0 (c) -30.0

20
r (mm)
Node 1 Location (Deg)

Node 5 Location (Deg)

8.0 -32.0 25
30
6.0 -34.0

4.0 -36.0
r (mm)
20
2.0 -38.0
25
30
0.0 -40.0
2 3 4 5 6 7 8 9 10 11 12 2 3 4 5 6 7 8 9 10 11 12

t (mm) t (mm)
10.0 80.0
(b) (d) 20
r (mm) 25

Node 5' Location (Deg)

Node 2 Location (Deg)

8.0 78.0
30
6.0 76.0

4.0 r (mm) 74.0

20
2.0 72.0
25
30
0.0 70.0
2 3 4 5 6 7 8 9 10 11 12 2 3 4 5 6 7 8 9 10 11 12

t (mm) t (mm)

(e) -30.0
20
Node 6 Location (Deg)

-32.0 r (mm) 25

30
-34.0

-36.0

-38.0

-40.0
2 3 4 5 6 7 8 9 10 11 12

t (mm)

Fig. 5 Variation of strain node locations with radius and thickness of octagonal ring. a Node 1, b node 2, c node 5, d node 5 , e node 6

order polynomial gives an excellent fit to the parametric 3.2 Experimental Validation
analysis results with regression coefficients ranging from
0.9653 to 0.999. Node 1 and node 2 locations are influenced The complete dynamometer assembly as shown in Fig. 1 was
by both radius and thickness of the octagonal ring, with the fabricated from AISI 1020 steel which has a yield strength
effect of radius being the greatest as indicated by the t test val- of 350 MPa and an ultimate tensile strength of 420 MPa.
ues. Node 5 and node 6 locations are mostly affected by the The two EOR elements were cut to the required geometry
thickness of the ring and the interaction term between radius and dimensions using wire EDM. The inside circle diameter
and thickness. These models allow the design of different of the EOR is 40 mm and the ring thickness is 5 mm. The
EOR transducers with minimum cross talk as demonstrated center-to-center distance of the two inside circles is 150 mm
in the next section. and the EOR width is 45 mm. The inside and outside sur-
faces of the EOR were prepared for strain gage mounting

123
 Author's personal copy
Arabian Journal for Science and Engineering

Fig. 6 Strain gage locations for V

EOR-2. Gages 1, 2, 3 and 4 are
for vertical force measurement.
5 t
Gages 5, 6, 7 and 8 are for
H
horizontal force measurement
8’ 5’
r
3 4 2 1

6
7
8 5

Table 6 Regression results for node 1 (a) 2.0E-05

0.0E+00
Coefficient Std. error t test P
-2.0E-05 εh = 1.7196E-10x - 1.6536E-07
a0 12.6907 0.1446 87.7506 <0.0001 R² = 3.3488E-01
-4.0E-05
a1 − 0.2076 0.0055 − 37.5946 <0.0001
-6.0E-05
a2 − 0.4138 0.0236 − 17.5048 <0.0001
Strain

-8.0E-05
a12 0.0075 0.0009 8.6196 0.0003
-1.0E-04
R 2 adj 0.9996 Standard error of estimate 0.0198
-1.2E-04
-1.4E-04
Table 7 Regression results for node 2 εv = -8.4338E-08x - 2.7823E-06
-1.6E-04
Coefficient Std. error t test P R² = 9.9568E-01
-1.8E-04
0 500 1000 1500 2000 2500
a0 12.5217 0.1075 116.4369 <0.0001
Force V (N)
a1 − 0.1882 0.0041 − 45.8377 <0.0001
a2 − 0.5368 0.0176 − 30.5420 <0.0001
(b) 2.0E-05
a12 0.0050 0.0006 7.8210 0.0005
R 2 adj 0.9999 Standard error of estimate 0.0147 0.0E+00

-2.0E-05 ε v = 2.0556E-09x - 5.5733E-07

Table 8 Regression results for node 5 R² = 4.7785E-01
-4.0E-05
Coefficient Std. error t test P
Strain

-6.0E-05
a0 − 31.6831 0.1947 − 162.7082 <0.0001 -8.0E-05
a2 − 1.0460 0.0746 − 14.0299 <0.0001
-1.0E-04
a12 0.0253 0.0022 11.5779 <0.0001
-1.2E-04
R 2 adj 0.9653 Standard error of estimate 0.1524
εh = -6.7619E-08x - 6.8696E-07
-1.4E-04
R² = 9.9483E-01
Table 9 Regression results for node 6 -1.6E-04
0 500 1000 1500 2000 2500
Coefficient Std. error t test P Force H (N)
a0 − 32.2397 0.1694 − 190.3177 <0.0001 Fig. 7 Calibration curves for EOR-2, a vertical force, b horizontal force
a2 − 1.5979 0.0649 − 24.6369 <0.0001
a12 0.0337 0.0019 17.7177 <0.0001
by sanding and polishing. One EOR member was wired for
R 2 adj 0.9912 Standard error of estimate 0.1325
measuring the vertical force only using Wheatstone bridge 1,
2, 3, 4 as shown in Fig. 6. Gages 1 and 3 are subjected to ten-

123
 Author's personal copy
Arabian Journal for Science and Engineering

(a)
(b)

FSW tool

EOR HDPE plates

Dynamometer Mantle

(c)
Welding
Plunge V

Dwell

End

Fig. 8 Application of EOR dynamometer in measuring friction stir welding forces (a, b) and measured down and feed forces (c)

sile strains while gages 2 and 4 are subjected to compressive KFH-6-350-C1 from OMEGA, with grid length of 6 mm,
strains. The other EOR member was wired to measure the nominal resistance of 350 ohms and a gage factor is 2.05.
horizontal force using Wheatstone bridge 5, 6, 7, 8. Gages 6 Calibration of the dynamometer was performed by loading it
and 8 are subjected to tensile strains while gages 5 and 7 are in the vertical and horizontal directions, independently using
subjected to compressive strains. It is noted based on nodal the z-axis of a milling machine and a Kistler Type 9257B
analysis that alternative nodes exist for 5 and 8 on the inside dynamometer for reference. For vertical force calibration, the
surface of the ring. These are labeled 5 and 8 . This finding Kistler dynamometer was placed on top of the EOR assem-
was also reported by Chen et al. [11]. However, in the present bly and the vertical load was applied by moving the milling
study, we decided not to use these nodes because of the dif- machine z-axis downward against the Kistler dynamometer.
ficulty of mounting strain gages on the inside surface of the For horizontal force calibration, the Kistler dynamometer
ring and to avoid congestion. The strain gages used were type was mounted vertically against the edge of the mantle and a

123
 Author's personal copy
Arabian Journal for Science and Engineering

horizontal force was applied against it using the feed axis. In References
this loading mode, it was ensured that the horizontal force is
in the plane of the mantle and that no bending moment was 1. O’Dogherty, M.J.: The design of octagonal ring dynamometers. J.
Agric. Eng. Res. 63, 9–18 (1996)
generated. Figure 7 shows the calibration curves for the ver-
2. Kheiralla, A.F.; Yahya, A.; Zohadie, M.; Ishak, W.: Design and
tical and horizontal forces. It is shown that the dynamometer development of a three-point auto hitch dynamometer for an agri-
performance is linear with maximum cross talk of 0.1% and cultural tractor. AJSTD 20(3&4), 271–288 (2003)
2.6% for the vertical and horizontal forces, respectively, at 3. Afzalinia, S.; Roberge, M.: An extended octagonal ring transducer
for the compression chamber of a large square baler. J. Agric. Sci.
full scale. Figure 8 shows the application of the dynamometer
Technol. 11, 377–389 (2009)
in actual friction stir welding of high-density polyethylene. 4. Abbaspour-Gilandeh, Y.; Khanramaki, M.: Design, construction
The measured down and feed forces shown in Fig. 8c exhibit and calibration of triaxial dynamometer for measuring forces and
a dynamically stable and stiff measuring platform that is able moments applied on tillage implements in field conditions. J.
Metrol. Soc. India 28(2), 119–127 (2013)
to capture the different phases of the welding process. The
5. Loewen, E.G.; Cook, N.H.: Metal cutting measurements and their
vertical force during the welding process started at 2 kN and interpretation. Proc. Sot. Exp. Stress Anal. 13(3), 57 (1956)
decreased with welding time due to material softening. The 6. Korkut, I.: Adynamometer design and its construction for milling
feed force was about half of the vertical force. operation. Mater. Des. 24, 631–637 (2003)
7. Yaldiz, S.; Ünsacar, F.; Sağlam, H.; Isik, H.: Design, development
and testing of a four component milling dynamometer for the mea-
surement of cutting force and torque. Mech. Syst. Signal Process.
4 Conclusions 21, 1499–1511 (2007)
8. Karabay, S.: Design criteria for electro-mechanical transducers and
arrangement for measurement of strains due to metal cutting forces
The strain node locations for an extended EOR transducer acting on dynamometers. Mater. Des. 28, 496–506 (2007)
were determined by the finite element method, and the 9. Dandage, R.V.; Bhatwadekar, S.G.; Bhagwat, M.M.: Design, devel-
effect of the octagonal ring geometry on node locations was opment and testing of a four component milling tool dynamometer.
investigated. A two-force dynamometer based on the EOR Int. J. Appl. Eng. Technol. 2(1), 45–50 (2012)
10. Kroencke, M.; Hull, M.L.: A Method for designing multiload com-
transducer was constructed and tested in friction stir welding. ponent dynamometers incorporating octagonal strain rings. Exp.
The findings of this study can be summarized as follows: Mech. 29(2), 195–204 (1989)
11. Chen, Y.; McLaughlin, N.B.; Tessier, S.: Double extended octag-
1. Strain node locations for the EOR transducer vary greatly onal ring (DEOR) drawbar dynamometer. Soil Tillage Res. 93,
462–471 (2007)
with ring radius and thickness. This variation is closely 12. Nalavade, P.; Salokhe, V.; Niyamapa, T.: Strain node identifica-
captured by a partial second-order polynomial relating tion by FEM and measurement of disc forces by double extended
node location to radius and thickness. octagonal ring transducer. J. Food Agric. Environ. 8(2), 730–735
2. The node locations for vertical force measurement are (2010)
13. Shishvan, S.H.; Akdeniz, R.C.; Gilandeh, Y.A.; Akkus, G.: The
shifted counterclockwise from the horizontal position by evaluation and comparison of stress node on extended octagonal
5◦ –8◦ . They move closer to the horizontal position as the ring transducer by strain energy theory and finite element method.
ring thickness and radius are increased. J. Agric. Mach. Sci. 7(3), 299–303 (2011)
3. The node locations for measuring the horizontal force are 14. Zhao, Y.; Zhao, Y.; Liang, S.; Zhou, G.: A high performance sensor
for triaxial cutting force measurement in turning. Sensors 15, 7969–
shifted counterclockwise for the known − 39.6◦ position 7984 (2015)
of the circular ring and they move closer to this position 15. Uddin, M.S.; Songyi, D.: On the design and analysis of an
as the ring thickness and radius are increased. No strain octagonal-ellipse ring based cutting force measuring transducer.
node existed at + 39.6◦ . Measurement 90, 168–177 (2016)
16. Girma, G.: Multicomponent dynamometer for measurement of
4. The locations for maximum sensitivity do not coincide forces on plough bodies. J. Agric. Eng. Res. 42, 85–96 (1989)
with the node locations for no-cross talk. For the hor- 17. Godwin, R.J.; Reynolds, A.J.; O’Dogherty, M.J.; Al-Ghazal, A.A.:
izontal force measurement, the locations for maximum A triaxial dynamometer for force and moment measurements on
sensitivity coincide with the theoretical node location for tillage implements. J. Agric. Eng. Res. 55, 189–205 (1993)
the circular ring.
5. The force dynamometer performance in actual welding
shows dynamic stability and sensitivity to the process
changes.

123