You are on page 1of 7

Design of Aluminum Bicycle

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


R. Davis
Graduate Student.
Frames
M. L. Hull This paper presents the use of a hybrid stress analysis method for the prediction of
fatigue life of bicycle frames. A finite beam element computer program is used to
Assistant Professor.
calculate frame stresses. In order to correlate the results with actual frame stresses,
Department of Mechanical Engineering, brittle coating experiments are conducted on an aluminum frame. Strain gage
University of California, experiments, guided by the brittle coating results, provide more accurate
Davis, Calif. 95616 measurements of maximum principal stresses. Stress concentration factors are
defined, the fatigue life of an aluminum frame is predicted for a simple loading
condition, and the result is compared with in-service failure data.

Introduction
Recent concerns about the expenditure of energy for human
transportation have accentuated the need for more efficient
passenger vehicles. The result has been an unprecedented
increase in the use of lightweight structural materials in the
manufacture of automobiles in the United States. Another
result has been the increased popularity of bicycles. for
practical transport. ...
With the large interest in bicycle transportation, research
has been conducted at the University of California, Davis in
recent years to develop a lightweight aluminum bicycle.
Aluminum was chosen because preliminary calculations using
a simple finite beam element program showed that a 20
percent reduction of weight could be achieved, while in- Fig. 1 Fatigue lallure In weld alter 1600 km of service
creasing the frame efficiency by about 3 percent above that of
a comparable steel frame [1]. Efficiency indicates the ability
of a frame to absorb as small an amount of energy as possible aluminum truss, and in their specific case devised a prediction
from the total usable effort during pedaling. Hence, a highly method based on simple specimen fatigue data already
efficient frame delivers almost all usable rider effort to the available. White and Patel [6] and Pan and Plummer [7]
drive train with very little energy going into the distortion of used brittle coating and fracture mechanics approaches,
the frame. Other advantages of aluminum include ease of respectively, to determine stress concentration factors for
fabrication by welding and low cost in comparison with new specific joint configurations and loading conditions. Due to
structural materials like carbon or· boron reinforced com- their specific nature, the results are not useful for the bicycle
posites. analysis, but the analysis methods used could be applied to a
Several aluminum frames were built by Mouritsen [2], and bicycle frame.
were put into normal service. Frame structural members were A more general method, advanced finite element analysis,
joined by Tungsten-Inert Gas welding and the frames were has been a major field of interest in tubular frame design. The
ridden in the as-welded condition. One frame failed by evolution of the method can be seen in a sequence of papers
fatigue, however, after about 1,600 km of riding (see Fig. 1). by Greimann, DeHart, Blackstone, Stewart, and Scales [8],
Most steel frames can be ridden for 64,000 km [3]. A more Kuang, Potvin, and Leick [9], and Liaw, Litton, and Reimer
complete analysis of stresses in welded aluminum frames is [10]. The results given, however, are too specific to be ap-
needed in order to develop a frame that will be light, yet able plied to the bicycle problem. The method used is very
to resist premature fatigue failure. powerful and could be applied to the bicycle frame, but the
The majority of previous work on stress analysis of welded limited availability of a large computer system to handle the
tubular joints has been contributed by those interested in enormous computation and storage requirements precludes
design of offshore structures. Bouwkamp [4] discussed some the use of this method.
of the basic ideas for adequate joint design. Sharp and This paper reports a new hybrid stress analysis technique
Nordmark [5] investigated the fatigue strength of a tubular for obtaining accurate measurements of principal stresses in
stress concentration areas. The hybrid technique combines a
simple version of the finite element approach with common
Contributed by the Design Engineering Division for publication in the
JOURNAL OF MECHANICAL DESIGN. Manuscript received at ASME Headquar- experimental stress analysis methods. The new technique is
ters, September 22,1980. used to measure the maximum principal stresses in an

Journal of Mechanical Design OCTOBER 1981, Vol. 1031901


Copyright © 1981 by ASME
= Table 1 Node positions of the bicycle model
O Element Number
Node Coordinate
X Y Z
(cm) (cm) (cm)
1 0 0 3.2
2 0 0 0
3 0 0 -10.2
4 53.7 0 16.4
5 0 0 -6.0
6 53.7 0 -38.8
7 64.4 -2.2 6.7
8 64.4 2.2 6.7
9 53.7 -1.6 -38.8

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


10 53.7 1.6 -38.8
11 58.9 -2.2 -36.0
12 58.9 2.2 -36.0
Fig. 2 Finite beam element representation of the frame 13 91.9 -6.1 -18.0
14 91.9 6.7 -18.0
aluminum bicycle frame and the fatigue life of the frame is
predicted for a simple loading condition.
Table 2 Properties of beam elements'0'
Finite Element Analysis
Element Ix, polar Iy a n d / z , A,
For complex space frames (i.e., frames made up of an # moment of moments of area
assemblage of beams) the simplest analysis method is finite inertia inertia
beam element analysis. Each beam in the frame is a "beam (10~ 7 m 4 ) (10-7m4) (cm 2 )
element," and the elements are connected at "nodes" to make 1,2,3 1.220 0.612 5.70
up the complete structure. Connections between elements may 4,5,6 0.629 0.316 1.89
sustain all six force and moment components. These internal 7,8,11,12,16 0.004 0.020 0.74
loads are assumed to act at the nodes, which are defined by 9,10,13,14,15 0.082 0.004 1.12
17,18 2.835 1.432 7.60
the intersection of center lines of the elements. It must be
noted that this method precludes obtaining a detailed picture For all elements, £•=6.89x1;, MPa
of joint stresses because the nodes are mathematical points y = 0.334
(fl)
within a global coordinate system. SeeFig.2.
An interactive computer program developed at the
University of California, Davis by Sikes and Beadle called direction (nodes 13 and 14) and translation of the head tube in
STAN [11] was used to analyze the bicycle frame. The the x direction (nodes 1 and 3). X, Y, Z directions are shown
program uses the finite beam element approach, and performs in Fig. 2. This unusual choice of coordinates was necessary to
all the necessary matrix manipulation. Coordinates of the allow free rotation at the head tube in the STAN program.
nodes, cross-sectional properties of the beams, the external Translation is not permitted in the Y direction because a
loads, and the boundary conditions must be input by the user. bicycle can be ridden almost upright at all times. Allowing
The idealized model of the 21 in. aluminum bicycle frame that motion in the Y or Z directions would involve dynamic
was analyzed is given in Fig. 2. Inputs necessary for the considerations and wheel and fork stiffnesses. These com-
computer simulation are listed in Tables 1 and 2. The actual plexities were not included in the present analysis. Note that
test frame may be seen in Fig. 5, and tube dimensions are the rear axle is constrained in the X, Y, and Z directions which
given in Table 3. Note that welded joints are ground smooth. is necessary to maintain equilibrium. This is analogous to
The boundary conditions of an operating bicycle frame are defining the rear axle as the origin of a coordinate system
somewhat complex. To determine the sensitivity of stresses to associated with a moving bicycle, with all distortion of the
changes in conditions at the boundaries, the bicycle frame was frame measured relative to that reference frame. Equilibrium
analyzed under three different conditions. Boundary con- is justified because the mass of the aluminum bicycle is very
ditions (A) that most closely approximate actual bicycle frame small (—1.5 kg) and inertial effects can be ignored.
conditions allow rotation at the head tube in the global Z Another set of boundary conditions allowed the rotations
direction (nodes 1 and 3), rotation at the rear axle in the Y aforementioned, but constrained nodes 1 and 3 in the X

, Nomenclature

neutral axis of a cross- associated with pedal


X, Y, Z = global coordinates in section crank assembly of
finite element model A = area of cross-section bicycle
x,y,z = local coordinates of an E,ES = modulus of elasticity of (a\s)i = a\ in the sample at the
element in finite element specimen point i
model j<=Poisson's ratio Sf = fatigue strength of
Fx,Fy,F7 = forces in the P=pedal load sample material
corresponding local A,B,C= designations for several Su = ultimate tensile strength
coordinate directions sets of boundary con- of sample material
Ix = polar moment of inertia ditions °a>ffm = alternating and mean
of a cross-section a( = larger principal stress stress, respectively
Mx,My,Mz = moments about the (plane stress) ef = threshold strain of brittle
corresponding local <r2 = smaller principal stress coating
coordinate directions (plane stress) Ks = stress concentration
Iy,I7= moment of inertia about rp ,rf,Zp ,Zf = linear dimensions factor

902/Vol. 103, OCTOBER 1981 Transactions of the ASME


Table 3 Dimensions of 6061-T6 aluminum bicycle frame'"'
F - 'P P
Chain ~ 7~
Tube Length O.D. Wall Thickness
type (cm) (cm) (cm)
r = 17.1 cm Head tube 13.3 3.5 0.64
u = 11.1 cm Top tube 56.2 3.8 0.17
Down tube 62.9 3.8 0.17
z„ = H.O cm Seat tube 55.2 3.8 0.17
V
Bottom bracket 7.1 4.4 0.64
Zj = 4.8 cm Seat stay 51.9 1.6 0.17
Seat stay brace 4.5 1.6 0.17
Chain stay brace 4.5 1.9 0.21
Chain head stay 43.8 1.9 0.21
17 deg angle and seat tube angle.

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


(a)
SeeFig. 5.
Measured to centers of joints defined by intersection of center lines, or
to the free ends of the tubes.
On the Frame

[2], one had been ridden normally to failure. Cracking


occurred at the welded joint connecting the seat tube to the
'f bottom bracket (node 6 in the finite element model, see Fig.
M, = Pz.
1). Note that the weld beads were present on this frame. Using
(A) boundary conditions, and a 445 N pedal load, the
following loads on member 6 at node 6 were obtained:
V-ZP17P
F^ = 332N Mx =0.090 Nm
iV=-77N My =0.768 Nm
F, = - 3 N M = -38.2Nm
Fig. 3 Load inputs to the frame due to pedalling rider. Simulations use
where x, y, and z are local right-handed coordinates, x axial to
a value of P = 445 N.
the member pointing away from node 6, y and z transverse to
the member. Calculations over all members in the frame
direction. These boundary conditions were designated (B). showed that the maximum stress in the entire frame occurs at
Boundary conditions (C) only allowed head tube rotation this location. The finite beam element method predicted a
about the Z axis. The rear axle was completely fixed. point of failure which matches the actual failure, assuming a
The loading condition investigated was an idealization of a simple pedalling load.
rider applying only vertical forces on the pedal with the pedal Analysis of a right pedal load also showed node 6 to be the
crank in the horizontal position. Handlebar forces were not point of maximum stress, but the down tube (member 5) was
considered because those forces would be transmitted directly the most highly stressed. Finite element results indicate that
to the front fork, which was assumed a fixed boundary the bottom bracket is the most failure-prone area for simple
relative to the frame. Also, the rider was assumed to be riding pedal loads.
off the seat. The determination of forces and moments on the In addition to the stresses at the "fictitious" ends of the
frame due to the vertical pedal load is straightforward. The elements (defined by the intersection of tube center lines),
results are shown in Fig. 3. stresses at the actual position of the welded interface between
Data presented by Soden and Adeyefa [12] and Gregor members were computed. Note that such computations do not
[13] concerning pedalling force patterns show that peak account for stress concentration. Table 4 contains a summary
normal pedal force occurs at different angular positions of the of the stresses at the bottom bracket area for the various
pedal crank depending on the rider and the power output boundary conditions and pedal loads at fictitious ends and at
being sustained. Also, the pedal is not always horizontal. welded interface positions.
However, inspection of the data shows that on the average, Circumferential locations of maximum stresses were also
peak normal forces occur near the horizontal crank position, determined. In all calculations, direct shear stress at points of
and pedal tilt is close to zero (10 deg or less) for this con- maximum bending stress were always negligible, so maximum
figuration in all cases studied. Therefore the pedal load stress points always lie on a line perpendicular to the resultant
idealization used here is a good estimation of the actual bending moment vector. Figure 4 shows the points of
maximal pedal load during normal cycling. maximum tensile stress predicted by the STAN simulations.
The bicycle frame was analyzed for both left and right (i.e., Angular displacements (see Section View A-A, Fig. 4) are
reversed) pedal loads. Moments and forces computed ac- listed in Table 5.
cording to Fig. 3 were applied at node 6 of the model in Fig. 2. Inspection of the principal stresses shows that (B) and (C)
Both sets (i.e., left and right) represent configurations that boundary conditions give almost identical results, and the
maximize the loads transmitted to the frame for a given value stresses deviate from those of boundary conditions (A) by up
of P because the moment arm rp of P is maximized. to about 10 percent. The conclusion is that moderate
Note that the loads are assumed to act at the center of the boundary condition changes do not have a large effect on
bottom bracket. Separate analyses were also undertaken with frame stress. An approximate boundary condition set (A) is
the loads being radial forces acting at the positions of the adequate for frame analysis and the problem need not be
pedal crank bearings, but the difference in important element complicated by considering the wheels and forks.
stresses found was negligible (less than 1 percent). Chain force An interesting aspect of the results is the behavior of the
on the rear wheel was not experienced by the frame because chain stay members (elements 9 and 13). When a left pedal
the rear axle was assumed a fixed boundary. A pedal load of load is applied, the left chain stay (element 9) is in tension,
445 N was used in the simulations. The system is linear, while the right (element 13) is in compression. When a right
however, and results may be extrapolated to study any load pedal load is applied, both are in compression. Such behavior
magnitude. is not intuitively expected from the configuration of the pedal
Of several aluminum frames built previously by Mouritsen crank mechanism. The moment Mx generated by the pedal

Journal of Mechanical Design OCTOBER 1981, Vol. 103/903


Table 5 Angular displacements of maximum stress points(a)
Position () for B.C. (A) () for B.C. (B) & (C)
(degrees) (degrees)
1 1.1 2.1
2 3.8 0.5
3 31.0 31.0
4 1.6 0.8
5 1.6 0.7
(a) See Fig. 4

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


LEFT PEDAL LOAD RIGHT PEDAL LOAD
Fig. 4 Locations of maximum principal stresses in the bottom bracket
area ("hot spots")

Table 4 Principal stresses (per 445 N pedal load) at the


bottom bracket area, based on finite beam element method
Boundary Loading Member Node Position 3 al a2
condition type 2 (MPa) (MPa)
type!
5 6 11.8 -0.359
A L 6 6 E 24.8 o
9 9 14.5 -0.114
13 10 0.090 -19.2
5 6 11.5 -0.370
A L 6 6 23.7 o
9 9 10.9 -0.152
13 10 0.110 -15.8
5 6 26.1 -0.003
A R 6 6 E 18.4 -0.260
9 9 0.256 -12.2
13 10 0.250 -12.1
5 6 25.2 -0.004
A R 6 6 17.7 -0.270
9 9 0.343 -9.07
13 10 0.340 -8.87
Fig.5 Bicycle frame test fixture and associated equipment
5 6 12.6 -0.339
B L 6 6 22.6 o
9 9 11.7 -0.141
13 10 0.115 -15.0
5 6 26.3 -0.004
B R 6 6 16.7 -0.288
9 9 0.378 -8.25
13 10 0.381 -7.90
5 6 12.5 -0.340
C L 6 6 22.6 o
9 9 11.9 -0.139
13 10 0.115 -15.0
5 6 26.2 -0.004
C R 6 6 16.6 -0.289
9 9 0.375 -8.30
13 10 0.373 -8.08
1 See text for explanation of boundary condition types.
2L: Leftpedalload, R: Right pedal load.
3E: At fictitious ends of elements, I: At tube-to-tube weld interface. Fig. 6 Bottom bracket area, showing stress pattern developed by
Stresscoat. The dark numbered lines are Isoentatlcs.

load and the angular positions of the members combine to


give this somewhat counter-intuitive result. mathematically straightforward. Durelli, Phillips, and Tsao
[14] offer an extensive treatment of the theory behind brittle
Experimental Stress Analysis coating analysis. More advanced methods of stress calculation
are outlined that are based on the presence of a significant
To both assess the accuracy with which the finite beam minor principal stress. For the bicycle frame, however, a2 is
element method predicted the maximum principal stress expected to be very small compared to a" because most stress
points and examine the effects of stress concentration, the is either axial or bending. Therefore, a simple yet approximate
stress state in the bottom bracket area of the aluminum test formula may be used to calculate the stress at a crack in the
frame was determined experimentally. Brittle coating coating.
(Stresscoat) experiments were conducted first. The advantages
(I)
of brittle coating over other methods such as photoelastic
models is that a picture of the stress field is obtained with a is the principal stress in the specimen perpendicular to the
al s
minimum of special equipment and analysis of results is crack at the cracking load, E s is the modulus of elasticity of

904/VoI.103, OCTOBER 1981 Transactions of the ASME


Table 6 Comparison of principal stresses in three stress stress points predicted by the finite element program for the
concentration areas (o) left pedal load. Accordingly, the finite beam element method
Source appears adequate for the prediction of trouble spots in a space
( f f ls)l (C|s)2 ( C T ls3>
(MPa) (MPa) (MPa) frame. The actual values of stress from the (C) boundary
conditions do not agree with the strain gage measured values
Finite Element 22.6 12.5 11.9 because stress concentration at the joints is not accounted for
Analysis
Stresscoat 51.2 48.9 31.6 by the simple finite beam element theory.
Test
Strain Gage 39.7 38.1 30.4 Fatigue Analysis
Measurements Accurately estimating the fatigue life under the simple left
<a)
SeeFig.4 and right pedal loading would entail mounting strain gages in
all areas of maximum stress concentration. Left and right

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


pedal loads would then be applied and the maximum mean
the specimen material, and eds is the tensile strain in the and alternating principal stress would be measured. Inasmuch
calibration bar necessary to crack the coating. Brittle coating as the principal stress under a left pedal load was measured in
always cracks perpendicular to the tensile principal stress. the testing reported herein, these data are only partially
Errors in using equation (1) will be very small as long as a2 is available.
negligible (see [14], Section 15.7). Despite the lack of complete data, the principal stresses
One drawback involved in brittle coating is its sensitivity to under a right pedal load can be approximated by defining
various parameters. For example, the threshold strain e^ may stress concentration factors K,
be affected by temperature, coating thickness, time between (ji^gage
application of coating and testing, speed of load application, K,= (2)
(a,,) STAN
humidity, and other factors. Therefore, even under the most
controlled conditions, accuracy of principal strain for left pedal loads. Because stress concentration is primarily
calculations is questionable. geometry dependent, the stress concentration factor for a
Nevertheless, because of the ability of brittle coating to particular region is not expected to vary under right and left
show overall stress trends by drawing isoentatics, the bottom pedal loads. Accordingly, principal stresses from a right pedal
bracket area was coated and studied. Loads were applied by load may be obtained by computing the principal stresses in
hanging weights from a mock pedal crank as shown in Fig. 5. "hot spot" areas from STAN and then multiplying these by
Relaxation of load was accomplished with a hydraulic floor the appropriate value of Ks.
jack. Left pedal loads only were applied. The clamping of the Comparing computed stress values at the joint interfaces
frame most closely matches the (C) boundary conditions. A for boundary conditions (C) (see Table 4) with the ex-
regular steering head bearing set was used with a special shaft perimental values, stress concentration factors at the "hot
to allow rotation at the head tube. spots" are
The results of one of several brittle coating tests are shown (AT,), =1.76
in Fig. 6. The dark numbered lines are isoentatics. The lowest (tf,) 2 =3.04
numbered isoentatics indicate the earliest cracking zones. (AT, ) 3 =2.56.
Three separate zones of initial cracking are present, and these
zones correspond closely to the maximum stress points Recall that boundary condition changes did not have a large
computed from the finite beam element analysis (Fig. 4). effect on the stress state; stress concentration factors would
Also, the concentration of isoentatics within the fillet areas of not be expected to change appreciably. Using boundary
the joints is high. This indicates the intense stress con- condition (A) results, which most closely approximate an
centration in these areas. actual bicycle, the corrected predicted stresses for a left pedal
load become
Calculation of the actual principal stresses at the initial
cracking zones is somewhat uncertain. Principal stresses on (ff„),=41.8MPa
the first isoentatic line may be easily calculated by a simple ((7^)2 = 35.0 MPa
modification of equation (1), noting that magnitude of stress (a, J 3 =28.1 MPa
at a point increases linearly with the load. The stresses inside
the first isoentatic can only be approximated, however. A Because an actual bicycle frame failed at the seat tube
simple interpolation technique can be used to approximate the connection to the bottom bracket, the fatigue life of the frame
load at which the first crack appears. Errors up to 10 percent at that point ("hot spot" 1) is of interest. Using the ap-
are possible (assuming no error in ef) when 90 N increments of propriate value of Ks to correct the STAN computed stresses
load are utilized. For a 445 N pedal load, a sample of the for the right pedal load, the stress at point 1 ranges between
stresses calculated in this manner at the "hot spots" 1, 2, and the values
3 is given in Table 6. These values vary greatly from the finite
element results mainly due to the stress concentration of the (°\S)L: 83.6 MPa
joints. (O]S)R: -45.2 MPa
Due to problems with temperature control, thickness of the during the pedalling cycle for pedal loads of 890 N. This
coating, and coating creep, stress values from three different loading magnitude is realistic according to Soden and
Stresscoat tests varied up to about 6.9 MPa. Therefore Adeyefa [12] for riding on a 1 to 10 incline. The fact that
Stresscoat magnitudes of stress can be accepted with limited these stresses represent the maxima and minima of the
confidence. To obtain higher accuracy in principal stress principal stress was shown by simulating several loading
magnitudes, small (1.6 mm) strain gages oriented to measure conditions intermediate between horizontal left and
principal stresses were mounted in the maximum stress areas horizontal right pedal positions. Values of o^ are less than 1.5
delineated by the Stresscoat tests. Results for the 445 N pedal percent of au and can be neglected. The mean and alternating
load are given in Table 6. The strain gage results must be components of the principal stress are
considered the most accurate values. Note that only the (ffi5)3
value agrees well with the Stresscoat determinations. , = 18.8 MPa
The Stresscoat test results correlate well with the maximum Oh = 64.4 MPa.

Journal of Mechanical Design OCTOBER 1981, Vol. 1 0 3 / 9 0 5


Modified Goodman Diagram
6061 Aluminum Frame, 4043 Weld Filler

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


Mean Stress, MPa
Fig. 7 Fatigue diagram for aluminum frame joint. Modified Goodman 1.51 104 10 s lo« 10J 10» i09
Line corresponds to safety factor of 1.
Number of Cycles to Failure

Fig. 8 Rotating beam fatigue curves for 6061-T6 aluminum, transverse


butt welds, 4043 filler wire

To determine the fatigue life of the joint, the values above


are plotted on a fatigue diagram. In Fig. 7, two Modified
Goodman lines corresponding to a safety factor of 1 have loading conditions may be used to obtain a better estimate of
been drawn. One corresponds to 6061-T6 Aluminum with fatigue life for a given frame configuration. Dynamic analysis
4043 filler in the as-welded condition (ultimate tensile strength of the frame-rider system may be used to determine maximum
Su = 213.7 MPa [15]) and the other is for the T6 re-heat- forces imparted to the frame. Frame stresses could then be
treated case (Su = 303.4 MPa [16]). The resulting alternating computed as discussed.
fatigue strength values are 70.7 MPa and 68.7 MPa for the as- 2. Fatigue Data. During the literature search phase of
welded and heat treated cases, respectively. the project, a large deficiency was found in available data on
Fatigue life data for welded aluminum is somewhat sparse, aluminum fatigue behavior. Welded aluminum data were
and re-heat treated weld fatigue data are virtually nonexistent. especially difficult to locate, and in several cases different
Because the tensile properties of T6 heat treated welds almost quantitative data sources were contradictory. Because
match those of solid 6061-T6 alloy, a fatigue life curve for aluminum is becoming much more popular as a structural
pure 6061-T6 is expected to closely approximate the actual material, a clear, concise, and complete source of fatigue data
heat treated weld behavior. Fatigue data for the as-welded is greatly needed. Without such data, fatigue performance of
case were found [15] and are plotted in Fig. 8 along with the aluminum structures can be predicted with limited accuracy.
fatigue life line for the 6061-T6 alloy [17].
Applying the alternating fatigue strengths found previously
to the respective plots in Fig. 8 shows that life for the joint of References
the test frame is greater than 108 cycles in both cases of weld
treatment. Because 108 cycles is roughly equivalent to 3 x 105 1 Hughes, K., and Hull, M.L., "Efficiency of Aluminum Bicycle
Frames," Dept. of Mech. Engr. Report, University of California, Davis, Mar.
km of riding for a typical 10-speed bicycle, the frame certainly 1977.
seems adequately designed for the simple pedal load analyzed. 2 Mouritsen, J., and Hull, M.L., "Construction of an Aluminum Bicycle
Therefore, it is possible that the in-service frame failure was Frame," Dept. of Mechanical Engineering Report, University of California,
solely due to the stress concentrations caused by the presence Davis, Apr. 1977.
of weld beads (the test frame studied herein had smooth 3 Burke, L.P., "The Development of a High Performance Bicycle Frame
Using a New Chrome. Molybdenum Seamless Tubing," presented at ASM
joints), and grinding of the welds may provide a simple Conference in Materials Innovations in Sporting Goods, Cleveland, Oct. 1976.
solution to the fatigue problem. It is also possible that a more 4 Bouwkamp, J.G., "Concept of Tubular Joint Design," Journal of the
severe loading condition was experienced by the frame which Structural Division, ASCE, Vol. 90, No. ST 2, Apr. 1964, pp. 77-101.
failed than was analyzed. 5 Sharp, M.L., and Nordmark, G.E., "Fatigue Strength of Welded
Tubular Aluminum Truss," Journal of Structural Division, ASCE, Vol. 103,
No. ST 8, Aug. 1977, pp. 1619.
Concluding Remarks 6 White, A., and Patel, N.M., "Brittle Lacquer Studies and Recom-
mendations for Strengthening Rectangular Tubular Joints," Fifth Offshore
1. Frame Loading. In the actual process of riding a Technology Conference, Proceedings Preprints, No. OTC 1822,1973.
bicycle, the frame experiences many different types of 7 Pan, R.B., and Plummer, F.B., "A Fracture Mechanics Approach to
loading. Seat loads, handlebar loads, and loads due to road Non-overlapping Tubular K-Joint Fatigue Life Prediction," Eighth Offshore
Technology Conference, Proceedings Preprints, No. OTC 2645,1976.
irregularities may all affect the frame stresses. Even normal 8 Greimann, L.F., DeHart, R.C., Blackstone, W.R., Stewart, B., and
pedalling involves more complex loading besides a simple Scales, R.E., "Finite Element Analysis of Complex Joints," Fifth Offshore
vertical load, especially when toeclips are used to pull on the Technology Conference, Proceedings Preprints, No. OTC 1823,1973.
pedals, as well as push. Therefore, a frame actually sees a 9 Kuang, J.G., Potvin, A.B., and Leick, R.D., "Stress Concentration in
Tubular Joints," Seventh Offshore Technology Conference, Proceedings
spectrum of loads applied at various points. More exact Preprints, No. TC 2205,1975.
fatigue analysis may involve the study of such a loading 10 Liaw, C.Y., Litton, R.W., and Reimer, R.B., "Improved Finite
spectrum and the cumulative fatigue damage that results. A Elements for Analysis of Welded Tubular Joints," Eighth Offshore
probabilistic approach may prove useful for this problem. In Technology Conference, Proceedings Preprints, No. OTC 2642,1976.
order to employ such a method, a large data base of rider 11 Sikes, D.F. and Beadle, C.W., STAN: A Three-Dimensional Finite
Beam Element Code for Interactive Computer-Aided Design, Dept. of Mech.
behavior, typical road surfaces, and other factors affecting Engr. Report, University of California, Davis, Oct. 1978.
riding loads would be needed. 12 Soden, P.D., and Adeyefa, B.A., "Forces Applied to a Bicycle During
In light of the requirements of the probabilistic approach, Normal Cycling," Journal of Biomechanics, Vol. 12,1979, pp. 527-541.
13 Gregor, R.J., "A Biomedical Analysis of Lower Limb Action During
the analysis method described herein, applied to other loading Cycling at Four Different Loads," Ph.D. thesis in Physical Education, Penn-
conditions, may provide enough information to design a sylvania State University, Aug. 1976.
frame in a much shorter time. Superposition of several 14 Durelli, A.J., Phillips, E.A., and Tsao, C.H., Introduction to the

906/Vol. 103, OCTOBER 1981 Transactions of the AS ME


Theoretical and Experimental Analysis of Stress and Strain, McGraw-Hill, 16 Welding Kaiser Aluminum, First Edition, Kaiser Aluminum and
1958, Chapter 15. Chemical Sales, Inc., 1969.
15 Person, N.L., and Wolfer, G.C., "Fatigue Properties of Butt Weld- 17 Craighead, CM., Eastwood, L.W., and Lorig, C.H., "An Appraisal of
ments in 6061-T6 Plate," Interim Project Report #MS DR 59065, Kaiser the Usefulness of Aluminum Alloys for Supersonic Aircraft and Guided Missile
Aluminum & Chemical Co., July 24, 1959. Construction," Rand Report #R-104, The Rand Corp., Aug. 8,1948, p. 57.

Downloaded from https://asmedigitalcollection.asme.org/mechanicaldesign/article-pdf/103/4/901/5592886/901_1.pdf by University Of Wisconsin Madison user on 04 March 2020


Change of Address Form for Journal of Mechanical Design

Present Address—Affix Label or Copy Information from Label

If you are planning


To Move, Please
Notify The
ASME-Order Dep't
345 East 47th St.
N.Y..N.Y. 10017

Don't Wait! I^rint New Address Below


Don't Miss An Issue!
Allow Ample Time to Name
Effect Change.
Atten finn
Addn ;ss
City State or Country ZiD

Journal of Mechanical Design OCTOBER 1981, Vol. 103/907

You might also like