You are on page 1of 12

Fatigue Failures of Light Poles

Geoff Taplin, Principal Engineer, Maunsell Australia Pty Ltd


Glenn Sanders, Assets& Maintenance Manager, CityLink Melbourne Limited
Zoltan Maklary, Manager Infrastructure, CityLink Melbourne Limited
John Connal, Director, Maunsell Australia Pty Ltd

SYNOPSIS
The Bolte Bridge and the Western Link Elevated Road are part of Melbournes
CityLink tollway, and provide vital connections in the transport system west of the
Melbourne CBD.

In September 2003 a light pole on the Western Link Elevated Road failed due to
fracture of two anchor bolts (herein referred to as bolt-failure). The light pole fell
across the northbound carriageway (Figure 1). In June 2004 a light pole near the
crest of the Bolte Bridge failed due to fracture of the mast near the base (herein
referred to as mast-failure). The light pole again fell across the northbound
carriageway (Figure 2). No one was injured in either of these incidents, however
extensive traffic disruption occurred.

CityLink Melbourne Limited (CML) requested that Maunsell Australia Pty Ltd
(Maunsell) investigate the failures, and report on the cause and any recommended
remedial measures. CML committed significant resources to obtaining an
understanding of the underlying cause of these failures so that all reasonable
precautions could be taken to avoid any further failures. The investigation
undertaken by Maunsell on behalf of CML involved original research, leading to
recommendations for remedial work, and recommendations for design procedures to
improve the whole of life performance of light poles. These recommendations are
reported in this paper.

1. DESCRIPTION OF THE LIGHT POLE FAILURES

1.1 Bolt-failure
This light pole, erected in 1999, comprised a vertical tapered steel mast of octagonal
cross section, with outreach mast arms of 4.5 metres and 1.5 metres. The overall
height of mast and mast arm was approximately 17.5 metres. The mast was butted
and fillet welded to a 32 mm thick steel baseplate. The baseplate had four 36 mm
diameter anchor bolts. Figure 1 is a photograph of the fallen light pole, and a detail
at the base of an intact light pole. In September 2003 failure of the light pole
occurred due to fatigue fracture of two of the anchor bolts. Figure 1 shows a
fractured anchor bolt.

As shown in Figure 1, the pole was supported by a double-nut baseplate, which is a


common type of support for a light pole of this height.

Fatigue failure of light poles Taplin Page 1 of 12


Figure 1 Bolt-failure light pole - Light pole base detail - Fracture surface of an anchor bolt

1.2 Mast-failure
This light pole, erected in 1999, comprised a vertical tapered steel mast of circular
cross section with a height of 18.0 metres. The bottom five metres approximately of
the mast was constructed from 4.0 mm thick grade 300 steel. At a height of 15.0
metres the mast had a double outreach mast arm, with lengths of 2.8 metres each
side. The mast was socketed and fillet welded into a 25 mm thick steel baseplate.
The baseplate had four 30 mm diameter anchor bolts. Figure 2 is a photograph of
the fallen light pole and a detail at the base of an intact light pole. This light pole was
also supported on a double-nut baseplate.

In June 2004 the mast fractured at the toe of the circumferential weld where the mast
was attached to the baseplate (Figure 2). Examination of the fracture surface
revealed two distinct fatigue crack zones, each initiated from the weld toe, with the
total crack length extending over approximately 85% of the circumference. Corrosion
on part of the crack surface indicated a relatively long term fatigue failure.

Figure 2 Mast-failure light pole - Light pole base detail Fracture surface of the mast

2. GROUTED VERSUS UN-GROUTED BASEPLATES

There are two common types of base supports for light poles grouted baseplates
that rest directly on concrete or grout, and double-nut baseplates that stand off from

Fatigue failure of light poles Taplin Page 2 of 12


the foundation and bear on levelling nuts. Both the bolt-failure and mast-failure light
poles were installed with double-nut baseplates.

AASHTO (2001), NCHRP (2002), and Cook (2000) make the following points about
grouted baseplates:
Compressive load from the baseplate should be supported directly by the
anchor bolt levelling nuts neglecting bearing of the baseplate on grout.
Experience has indicated that anchor bolts may experience corrosion if
cracking occurs in the grout, or if adequate drainage is not provided.
Under dynamic loading, pretension against the grout may be quickly lost due
to wear of the grout from movement, so no benefit can be accrued for fatigue.
Grout, when specified under baseplates, should be nonshrink and should not
contain any chlorides or other harmful additives that could cause corrosion of
the anchor bolts.

3. FATIGUE LOADING DUE TO WIND

Failure of both the bolt-failure and mast-failure light poles was due to metal fatigue.
Possible causes of the fatigue loading considered were oscillation of the light poles
caused by deflection of the bridge structure, oscillation of the light poles caused by
truck-induced gusts, and oscillation of the light poles due to wind loading.

Fatigue failure of light poles caused by deflection of the bridge structure is known to
have occurred in Japan, however it is understood that the bridge was a relatively
flexible steel structure. The Bolte Bridge in particular is a very rigid structure, and the
energy imparted to the light poles as a result of bridge deflection is extremely low.
Oscillation of overhead cantilever mast arms supporting traffic signals, signs and
variable message signs due to the air movement caused by passing traffic is known
to have caused fatigue failures of mast arms in the USA. However no failures of light
poles have been attributed to this cause, presumably because of the lower wind area
and the greater distance from the traffic.

Oscillation of light poles due to wind loading is known to have caused fatigue failures,
and was considered the most likely cause of the failures reported here. The natural
gusting of the wind can cause vibration of the light pole in the same direction as the
wind is blowing (along wind response) or, due to the phenomenon of vortex
shedding, the wind can cause vibration of the light pole in a direction transverse to
the direction that the wind is blowing (cross wind response).

3.1 Wind Conditions


Gust wind speed data at Melbourne airport for the period January 2002 to September
2004 showed that the light pole failures in September 2003 (Figure 3) and June 2004
did not occur under exceptional wind conditions. This is consistent with the
observation that the failures were due to fatigue, and not a result of overload.

Fatigue failure of light poles Taplin Page 3 of 12


30

25

20
wind speed m/s

15

10

0
1-Sep-03 6-Sep-03 11-Sep-03 16-Sep-03 21-Sep-03 26-Sep-03 1-Oct-03
date and time

Figure 3 Three second gust wind speeds September 2003 (dashed line is failure date)

3.2 Quantifying the wind loading


Assessing the loads which will cause the stress fluctuations leading to fatigue failure
is very difficult. Wind loading is a random process, so it is not possible to determine
the magnitude and number of cycles of stress that a light pole will be subjected to.
We must rely on averaged wind properties to estimate this. Converting the wind
speed to the force on a light pole is also imprecise. It depends upon the local wind
climate at the site, and it also depends upon the response of the light pole and how it
interacts with the gusting wind. The behaviour of the light pole is affected by its
resonant behaviour, so that when the frequency of loading from the wind is near to
the natural frequency of vibration of the light pole, dynamic amplification of the stress
in the light pole will occur. Finally, even if the complete loading history of the light
pole is known (by measurement for example) there is still great uncertainty in the
prediction of the time to fatigue failure. Fatigue prediction models are not precise.

The propensity for the light poles to undergo vortex shedding was assessed against
the provisions of AASHTO (2001). The wind speed at which vortex shedding would
occur at the first natural frequency is 0.9 m/s for the bolt-failure light pole and 0.8 m/s
for the mast-failure light pole. Even at the second mode of vibration the wind speeds
are still low at 2.4 m/s and 3.1 m/s respectively. Therefore it was concluded that
vortex shedding would generate very low stress ranges, and certainly less than the
stress ranges due to natural wind gusts. This left natural wind gusts as the likely
source of the fatigue loading.

4. STRUCTURAL ANALYSIS

AS5100 (Standards Australia 2004) requires that light poles be designed to prevent
excessive vibration leading to fatigue. AS/NZS4676 (Standard Australia 2000)
requires that oscillations due to dynamic wind which may induce fatigue should be

Fatigue failure of light poles Taplin Page 4 of 12


investigated and, if significant, due account should be taken of them in design.
Neither Australian standard provides advice as to how fatigue life can be assessed.

4.1 Fatigue assessment based upon rigid baseplate models


Rigid baseplate models use the assumption that the stress in the anchor bolts and in
the mast can be directly calculated from the overturning moment at the base of the
light pole.

AASHTO (2001) specifies an equivalent wind pressure to be used to calculate the


fatigue stress range from natural wind gusts. For the bolt-failure light poles this
equated to a wind speed of 26 m/s, and for the mast-failure light pole 29 m/s. The
stresses induced by this wind pressure must be below the constant amplitude fatigue
threshold (CAFT), ie low enough that the detail in question can sustain infinite cycles
of loading. The results of the analysis are presented in Table 1. According to
AASHTO (2001) the bolt-failure light poles have adequate fatigue capacity, but the
mast-failure light poles do not.

In his textbook Holmes (2003) presents a method specifically developed for the
fatigue assessment of light poles. This method predicts a fatigue life for fluctuating
stress due to natural wind gusts. The results of this method (assuming a rigid
baseplate), and using the fatigue life curves from AS4100 (Standard Australia 1998),
are shown in Table 1. According to the Holmes method, the bolt-failure light poles
have a marginally acceptable fatigue life, but the mast-failure light poles do not.
Table 1 Fatigue assessment assuming rigid baseplate

According to AASHTO (2001): calculated CAFT


bolt-failure light poles - stress in anchor bolt 35 MPa 48 MPa
mast-failure light poles - stress in mast at base 50 MPa 18 MPa

According to Holmes method: predicted fatigue life


lower limit upper limit
bolt-failure light poles - anchor bolts 35 years 90 years
mast-failure light poles - mast at base 4 years 10 years

4.2 Advanced structural analysis


The above analysis was based upon the simplifying assumption that the baseplate is
rigid. This section describes the further analysis undertaken by Maunsell in order to
more accurately model the behaviour at the base of the mast. The analyses
described in this section were undertaken using non-linear finite element analysis,
using the software Strand 7. The non-linearity in the analysis was included to
account for grout providing support when loaded in compression, but not when
loaded in tension.

Figure 4 shows isometric and side views of the bolt-failure light pole bending under
the AASHTO (2001) fatigue design wind speed. The different colours represent
different steel plate bending elements. The displacements are magnified 300 times.
The bending of the baseplate can be clearly seen.
Fatigue failure of light poles Taplin Page 5 of 12
mast

baseplate

bracket

Figure 4 Light pole bending under serviceability wind loading (300X)

The baseplate bends upward on the windward side of the light pole, and downward
on the leeward side. This bending causes the anchor bolts on the windward side to
bend outwards, and causes the anchor bolts on the leeward side to bend inwards.
Note that this bending in the anchor bolts is at right angles to the wind direction
(transverse bending). Transverse bending of the baseplate causes stresses in the
bolts which are far greater than the stresses calculated in the bolts if the baseplate is
assumed to be rigid. Bending of the baseplate is the action which leads to bolt
fatigue failure

Although not so easily seen in the Figures, the bending of the baseplate also causes
high local stresses in the mast near the base, because the wall of the mast bends
locally as the baseplate bends. As is the case for the bolts, the bending of the
baseplate causes stresses in the mast which are far greater than the stresses
calculated in the mast if the baseplate is assumed to be rigid. Bending of the
baseplate is the action which leads to mast fatigue failure.

The finite element analysis predicts much higher stresses, and therefore much worse
fatigue performance, than the simplified analysis assuming a rigid baseplate that was
presented above (Table 2). The stresses in the anchor bolts are particularly high,
and can be considered to be an overestimate of the actual stresses. This is because
the analysis model assumes that there is no relative rotation between the anchor bolt
and the baseplate. Even with fully tightened bolts there will be some relative
movement.
Table 2 Fatigue assessment - based upon finite element analysis

According to AASHTO (2001): calculated CAFT


bolt-failure light poles - stress in anchor bolt 120 MPa 48 MPa
mast-failure light poles - stress in mast at base 150 MPa 18 MPa

According to Holmes method: predicted fatigue life


lower limit upper limit
bolt-failure light poles - anchor bolts 0.12 years 0.31 years
mast-failure light poles - mast at base 0.06 years 0.15 years

Fatigue failure of light poles Taplin Page 6 of 12


Using finite element modelling, the variations in stress in the bolts and the mast due
to changes in the connection details were investigated. Increasing the baseplate
thickness from 32 to 50 mm reduced the calculated bolt stress by 40% and the
calculated mast stress by 30%. These results demonstrate that the major contributor
to the stress in the bolts and the mast is the bending of the baseplate. The most
direct way to reduce the stresses and increase the fatigue life is to increase the
stiffness of the baseplate. Highways Agency (1999) contains guidance on the design
of column baseplates. Following the procedures in that document would have
resulted in baseplate thicknesses that avoided the high stresses in the anchor bolts
and mast that occurred in the failed poles.

Non-linear finite element modelling was used to assess the effect of grouting under
the baseplate, which is effective only in compression (Table 3). On the side of the
baseplate that is lifting up, the stress in the mast near the base is virtually unchanged
by grouting (because grout is considered ineffective when the baseplate lifts off).
However the stress range that the mast goes through is approximately halved. This
leads to an improved fatigue performance. On the side that is lifting up the grout
provides some benefit to the anchor bolts because it stiffens them and reduces the
bending. This effect is not able to be modelled realistically, and therefore the benefit
cannot be quantified.
Table 3 Fatigue assessment according to AASHTO (2001) non-linear fe analysis with grout

According to AASHTO (2001): calculated CAFT


mast-failure light poles - stress in mast at base 40 MPa 18 MPa

According to Holmes method: predicted fatigue life


lower limit upper limit
mast-failure light poles - mast at base 2 years 5 years

Grouting may therefore increase the fatigue life of the mast at the base by a factor of
approximately 30 times in the case of the mast-failure light poles. The increase in
the fatigue life of the bolts cannot be estimated, but it would likely be of a similar
order of magnitude.

5. FIELD MEASUREMENTS

To determine the effectiveness of modifications intended to provide a satisfactory


fatigue life for the light poles, field measurements were made in three adjacent light
poles on the Bolte Bridge (mast-failure light poles). The three poles comprised:
an original light pole (ungrouted 25 mm thick baseplate)
a grouted baseplate light pole (grouted 25 mm thick baseplate)
a grouted-and-thickened baseplate light pole (grouted 40 mm thick baseplate)

The field measurements were obtained by VicRoads (acting on behalf of CML).


During field measurements, strains in the vertical direction were measured on the
mast immediately above the baseplate weld, and on the mast 500mm above the
baseplate. The field data related only to mast failure - bolt strains were not
Fatigue failure of light poles Taplin Page 7 of 12
measured. Field date were recorded in time frames of varying length. A typical plot
of measured strain versus time, for a 30 minute time frame, is presented in Figure 5.
In the data analysis, each time frame was analysed separately for maximum stress
and fatigue damage.

150
1-PNL

100

50
microstrain

-50

-100

-150
13:19:12 13:26:24 13:33:36 13:40:48 13:48:00 13:55:12 14:02:24
time

Figure 5 Strain versus time on the original light pole. 30 minute time frame

5.1 Data Analysis


The data for each time frame was analysed to determine the strain ranges by the
method of Rainflow counting (Downing 1982). A histogram of strain ranges was
produced for each measurement position for each time period. Figure 6 is a typical
histogram.

Rainflow Analysis 030405-13_27pm to 13_57pm 1PNL

2500

2000
Frequency

1500

1000

500

0
5

3
.7

.1

8.

1.

5.

8.

2.

5.

8.

2.
21

65

10

15

19

23

28

32

36

41

Strain Ranges (microstrain)

Figure 6 Histogram of strain ranges on the original light pole.

From the histogram, maximum stress ranges were obtained for each measurement
position, for each time period. Also from the histogram the fatigue damage at each
position was calculated, for each time frame. The fatigue damage was calculated
using the Palmgren-Miner rule for cumulative damage (Fisher 1997), assuming that
the mast to baseplate weld detail was Fatigue Category 40 in accordance with
AS4100 (Standards Australia 1998). Knowing the cumulative damage within the time

Fatigue failure of light poles Taplin Page 8 of 12


frame, and the length of the time frame, it was possible to calculate the theoretical
fatigue life. This calculation assumed that the wind conditions that occurred during
the time frame remain unchanged for the life of the structure. The prevailing wind
conditions during some of the field work were certainly more severe than average, so
this assumption is conservative, leading to an underestimate of the fatigue life. As
well as performing the cumulative damage and fatigue life calculation for each time
frame, a separate calculation was undertaken over the total periods of measurement,
which were 9.22AM to 3.30PM on 3 April 2005, and 5.51AM to 9.03AM on 24 May
2005.

5.2 Wind Speed and Direction


Figure 7 plots the mean wind speed (30 minute average) over the period of field
measurements at three weather stations surrounding the site, as well as the
maximum stress range for each time frame. The stress in the light pole is well
correlated with the wind speed. This confirms that wind loading is the cause of the
stress and fatigue cracking in the light poles.

20 100

18 90

16 80

14 70 maximum stress range (MPa)


wind speed (m/s)

12 60

10 50

8 40

6 30
Fawkner Beacon
4 Laverton 20
Melbourne Airport
2 10
max stress range
0 0
8:24 AM 9:36 AM 10:48 AM 12:00 PM 1:12 PM 2:24 PM 3:36 PM 4:48 PM
time

Figure 7 Mean wind speed during field measurement on 3 April 2005

5.3 Grouted-and-thickened baseplate versus original


Figure 8 plots the maximum stress ranges at the lower measurement positions, for
each time frame. The grouted-and-thickened baseplate light pole (dashed lines) has
lower stress ranges than the original pole (solid lines) at nearly every time frame.

On 3 April 2005 the maximum stress range measured in the original pole was 87
MPa, and the maximum stress range measured in the adjacent grouted baseplate
pole was 65 MPa. On 24 May 2005 the maximum stress range measured in the
original pole was 58 MPa, and the maximum stress range measured in the adjacent
grouted-and-thickened baseplate pole was 38 MPa. The constant amplitude fatigue
limit was 29 MPa for the weld detail used, therefore, based upon the maximum
recorded stress ranges, the light poles would not have an infinite fatigue life.

Fatigue failure of light poles Taplin Page 9 of 12


70
originalLNE grout&bpLNE originalLNW grout&bpLNW
originalLSE grout&bpLSE originalLSW grout&bpLSW
60
maximum stress range (MPa)

50

40

30

20

10
5:31:12 6:00:00 6:28:48 6:57:36 7:26:24 7:55:12 8:24:00 8:52:48 9:21:36
AM AM AM AM AM AM AM AM AM
time

Figure 8 Maximum stress range at each time frame lower measurement positions grouted-
and-thickened baseplate versus original

5.4 Fatigue Life Results

Figure 9 plots the estimated fatigue life based upon each time frame of measured
stress. The fatigue life is extremely sensitive to wind speed, decreasing rapidly for
the time frames corresponding to higher wind speeds. Estimated fatigue lives in
excess of 50 years are not plotted. The fatigue life is calculated separately for each
time frame, and assumes that the prevailing wind conditions will be maintained for
the life of the structure. This is conservative when measurements are taken in high
wind conditions, as here. The usefulness of the plot is in comparing the original light
pole (solid lines) to the grouted-and-thickened baseplate light pole (dashed lines).
The modified light pole has significantly greater fatigue life.

50

45

40

35 originalLNE
fatigue life (years)

grout&bpLNE
30
originalLNW
25
grout&bpLNW
20 originalLSE
15 grout&bpLSE
originalLSW
10
grout&bpLSW
5

0
5:31:12 6:00:00 6:28:48 6:57:36 7:26:24 7:55:12 8:24:00 8:52:48 9:21:36
AM AM AM AM AM AM AM AM AM
time

Figure 9 Fatigue life per time frame - grouted-and-thickened baseplate vs original

Fatigue failure of light poles Taplin Page 10 of 12


In addition to calculating the fatigue life for each time frame, the fatigue life was
separately calculated using the measured strains for the entire measurement period.
The results for each of the lower measurement positions are given in Table 4. These
results show a clear benefit from grouting, and a greater benefit from grouting and
providing a thicker baseplate.
Table 4 Estimated fatigue life (rounded to whole years)

Data - 3 April 2005 9:22AM to 3:30PM Data - 24 May 2005 5:51AM to 9:03AM
position original grouted grouted position original grouted-and- g-and-t
on pole baseplate on pole thickened
original baseplate original
N 2 years 13 years 6.5 NE 3 years 26 years 8.7
W 7 years 28 years 4 NW 2 years 40 years 20
S 3 years 3 years 1 SE 1 years 16 years 16
SW 2 years 15 years 7.5

6. REMEDIAL MEASURES

As a result of the findings from the finite element analysis and field measurements,
CML replaced all light poles similar to the mast-failure pole. The replacement poles
had increased mast wall thickness in the lower section (6mm vs 4mm), thicker
baseplates (40mm vs 25mm) and the baseplates were grouted. The baseplates of
all light poles similar to the bolt-failure pole were grouted. A program of regular
inspection and ultrasonic testing has been implemented.

7. CONCLUSIONS

1. The light poles collapsed as a result of fatigue failure caused by stress


fluctuations due to natural wind gusts.
2. Finite element analysis and field measurements confirmed that stress fluctuations
in the anchor bolts and the mast near the base of the light poles were excessively
high, and would lead to fatigue failure.
3. The high stress fluctuations appeared to be primarily caused by inadequate
thickness of the baseplates. Design methods which assume rigid baseplates will
give gross underestimates of the stresses and hence of the fatigue lives of the
bolts and the masts.
4. New poles should be designed with thicker baseplates, and should have a fatigue
assessment which allows for the baseplate bending. Highways Agency (1999)
contains useful guidance on the thickness of baseplates.
5. Although there is a strong body of expert opinion that grout should not be relied
upon as part of the structural system because of a loss of capacity over time,
grouting does lead to reductions in the magnitude of the stress cycles. If grouting
is used, the installation quality must be high, and both short and long term
monitoring of the grout condition should be undertaken.

Fatigue failure of light poles Taplin Page 11 of 12


6. Because of the uncertainty in the predictions of fatigue life, light poles which may
fall across major roads should be monitored for the development of fatigue
cracks.
7. Further failures of similarly designed and constructed light poles in exposed
locations around Australia should be expected, unless remedial measures are
undertaken.

8. REFERENCES AND BIBLIOGRAPHY

1. AASHTO (2001) Standard Specifications for Structural Supports for Highway


Signs, Luminaires and Traffic Signals AASHTO, Washington DC USA
2. Cook R. A. et al (2000) Use of grout pads for sign and lighting structures Part 1 -
Structural evaluation. Prepared for Florida Department of Transportation and
Federal Department of Transportations USA
3. Downing S. D. Socie D. F. (1982) Simple Rainflow Counting Algorithms.
International Journal of Fatigue, Jan. 1982 pp 31-40.
4. Fisher J. W. et al (1997) A Fatigue Primer for Structural Engineers ATLSS Report
97-11 Lehigh University, Pennsylvania USA
5. Highways Agency (1999) Design Manual for Roads and Bridges Volume 2
Section 2 Part 1 BD26/99 Design of lighting columns, Highways Agency UK
6. Holmes, J.D. (2003) Wind Loading of Structures Spon, London
7. NCHRP (2002) Fatigue-resistant design of cantilevered signal, sign and light
supports National Cooperative Highway Research Program Report 469,
Washington DC USA
8. Standards Australia (1998) AS4100 Steel Structures, Sydney Australia
9. Standards Australia (2000) AS/NZS 4676 Structural design requirements for utility
service poles, Sydney Australia
10. Standards Australia (2004) AS5100 Bridge Design, Sydney Australia

Fatigue failure of light poles Taplin Page 12 of 12

You might also like