Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 1 LOW FREQUENCY VIBRATION FROM LIGHT RAIL VEHICLES 1 2 Hugh Saurenman (Corresponding author) 3 ATS Consulting 4 215 N Marengo Avenue, Suite 100 5 Pasadena, CA 91101 6 1-626-710-4400 7 [email protected]8 9 Shankar Rajaram 10 Sound Transit 11 425 S Jackson Street 12 Seattle, WA 98109 13 1-206-903-7336 14 [email protected]15 16 James T. Nelson 17 Wilson, Ihrig & Associates, Inc. 18 6001 Shellmound Street, Suite 400 19 Emeryville, CA 94608 20 [email protected]21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Submitted: August 1, 2016 38 39 Word count: 5241 words text + 9 figures x 250 words (each) = 7491 total 40 41 42
18
Embed
LOW FREQUENCY VIBRATION FROM LIGHT RAIL …docs.trb.org/prp/17-00536.pdf · 1 LOW FREQUENCY VIBRATION FROM LIGHT RAIL VEHICLES ... 15 vibration monitoring system ... 19 The MIA limits
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 1
LOW FREQUENCY VIBRATION FROM LIGHT RAIL VEHICLES 1
Word count: 5241 words text + 9 figures x 250 words (each) = 7491 total 40
41
42
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 2
ABSTRACT 1
The question of how much low-frequency vibration is generated by operations of typical light 2
rail vehicles is an important issue when a proposed light rail system will be in close proximity to 3
vibration sensitive research facilities. In this paper we define low-frequency vibration to be 4
vibration radiated at frequencies below 25 Hz because rail transit vibration often is lower than 5
the vibration from buses and trucks at frequencies below 25 Hz. This paper is largely based on 6
vibration measurements vibration performed in support of the vibration monitoring system that 7
was installed in the University-Link extension of the Seattle Link LRT system. The monitoring 8
system is designed to protect vibration sensitive research facilities that are relatively close to the 9
subway. The testing has shown that there is substantial low-frequency vibration at the vibration 10
monitors from traffic on nearby roadways and that this vibration is often greater than the 11
vibration generated by train operations. In other words, at low frequencies, the vibration from 12
trains passing about 10 ft from the monitors is often less than the vibration created by traffic that 13
is at least 200 ft from the monitors. This is a non-intuitive result, and the primary purpose of this 14
paper is to examine possible reasons for this result and what factors contribute to the low-15
frequency traffic vibration being greater than the low-frequency train vibration. The results of 16
this analysis are applicable to other rail transit projects where low frequency vibration interfering 17
with research equipment is a concern. 18
19
20
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 3
INTRODUCTION 1
This paper is basically a case history of a detailed exploration of low-frequency vibration from 2
train operations in a subway, the University Link (U-Link) extension of the Sound Transit Link 3
light rail system. Researchers at the University of Washington (UW) Seattle campus have been 4
concerned that vibration from the light rail extension through the campus would cause problems 5
with various vibration sensitive research facilities that are distributed throughout the campus. 6
The departments concerned about the low frequency vibration from rail operations affecting their 7
research equipment ranged from the Fisheries Center to the Forest Sciences Laboratory to 8
various physics, engineering, and materials science laboratories. 9
Negotiations between UW and Sound Transit led to several accommodations by Sound Transit to 10
minimize the potential for vibration and electromagnetic fields (EMF) to interfere with the 11
research equipment. This paper focuses on vibration and does not cover the EMF issue. The 12
negotiations led to a Master Implementation Agreement (MIA) signed by the CEO of Sound 13
Transit and the President of the University. This agreement includes specific requirements for a 14
vibration monitoring system located in the subway that provides real-time alerts whenever the 15
vibration levels in nearby buildings may exceed the maximum allowed vibration levels. The 16
MIA provides vibration limits in terms of 1/3 octave band velocity levels over the 2 to 100 Hz 17
range for 24 buildings. 18
The MIA limits are based on measurements of ambient vibration that were performed prior to the 19
start of construction of the U-Link extension. The test results presented in this report were 20
performed as part of the effort to estimate the relationships between vibration at the monitoring 21
system stations in the subway and the train vibration inside the 24 buildings identified in the 22
MIA. The buildings covered by the MIA limits are labeled in red in Figure 1. The four buildings 23
where vibration measurements were performed are the Center for Human Development and 24
Disability (CHDD), the University of Washington Medical Center Cyclotron facility (UWMC), 25
Winkenwerder Forest Sciences Laboratory, and Wilcox Hall that is used by a variety of 26
engineering departments. As seen in Figure 1, the horizontal distances from the subway to these 27
buildings ranged from approximately 600 ft to 1,000 ft. The subway is approximately 100 ft 28
deep. 29
The vibration monitoring system was supplied by International Electronic Machines, Corp. 30
Relevant to this paper are the nine monitoring stations in the subway segment near the UW 31
campus. The monitors are at 150 ft intervals alternating between the northbound and southbound 32
tunnels. Monitors 4-1 through 4-3 are in the bored tunnels, Monitors 5-1 through 5-3 are adjacent 33
to the crossover structure south of the UW Station, and Monitors 5-4 through 5-6 are in the UW 34
Station. Each of the nine monitors are programmed to collect data every time a train is within a 35
set distance of the monitor and to report the maximum 8-second RMS vibration velocity level 36
(Lmax) during the period that the train is within the set distance of the monitor. The data are 37
collected at 1-second intervals and the 8-second RMS levels are calculated from the 1-second 38
data. The 1/3 octave band Lmax spectra of are provided by the monitor software for each train 39
event. 40
The system allows specifying adjustment factors that are used to estimate the building vibration 41
levels from the Lmax levels for each train event. In this report the adjustment factors are referred 42
to as Vibration Adjustment Estimates (VAEs). The monitoring system software adds the 43
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 4
appropriate VAEs at each building to the measured Lmax by each monitor to determine if any of 1
the building vibration levels may exceed an MIA limit. 2
The remainder of this paper outlines the measurement programs, discusses the results of the 3
measurement programs, and presents a discussion of the theoretical considerations relative to 4
when and at what amplitudes vibration in the 2 to 6 Hz range will occur. 5
6
Figure 1. University of Washington Campus Map. 7
Buildings identified by UW as sensitive buildings are labeled in red. 8
Measurements were performed at CHDD, the UW Medical Center Cyclotron, 9
Winkenwerder, and Wilcox. The distance to the subway is shown 10
adjacent to each of the measurement buildings. 11
CHALLENGES TO ESTIMATING VAES 12
The program to develop valid estimates of the VAEs began in July 2015 and needed to be 13
completed in time for the U-Link extension to open in March 2016. Before the system could 14
open the VAE values had to be uploaded to the monitoring system software and tested to confirm 15
that ambient vibration did not generate alarms and that the sum of the measured Lmax at the 16
monitors and the VAEs did not underestimate the train vibration levels at the sensitive buildings. 17
Some of the challenges to developing reasonable estimates for the VAEs were: 18
740 ft
916 ft
840 ft
593 ft
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 5
As seen in Figure 1, the closest sensitive buildings were 600 ft to 1,000 ft from the 1
subway. Ground vibration tends to attenuate relatively rapidly, which meant that 2
relatively low vibration levels at the buildings could be expected. 3
There are a number of structures in the paths between the subway and the buildings. The 4
structures include an underground parking structure, and various buildings, utilities, and 5
roadways. 6
There was considerable interest in the results of the measurements and analysis. The 7
stakeholders included the managers of the vibration sensitive facilities, the UW 8
executives and department heads who wanted to maintain the University’s reputation as a 9
world class research facility that could accommodate modern nanotechnology research, 10
and Sound Transit that wanted to both avoid false alarms and avoid exceeding the MIA 11
vibration limits. Another consideration is that this exercise was basically a dry run for the 12
analysis that will be required for the Northgate Link that is currently under construction. 13
The Northgate Link route will be considerably closer to a number of vibration sensitive 14
laboratories and will include a low-frequency floating slab system to control the vibration 15
levels. 16
In many cases, the train vibration was lower than the ambient inside buildings. The 17
ambient was due to equipment and activities within the buildings, traffic on nearby 18
streets, and various unidentified sources. The low levels of train vibration were due to the 19
distances of the buildings from the subway and the various vibration mitigation measures 20
that were used (e.g. soft fasteners and low impact frogs at the crossover). 21
Limited time to perform measurements and provide results. 22
A key issue is that UW and Sound Transit needed to mutually agree that the approach to and the 23
results were reasonable before the U-Link extension could open, which it did in March 2016. 24
MEASUREMENT PROGRAMS 25
The data available for evaluating ambient vibration at the U-Link subway monitors includes: 26
1. Measurements of train vibration and ambient vibration performed in August 2015. The 27
results of these measurements are documented in Reference 1. 28
2. Certification measurements that were performed by Wilson, Ihrig & Associates (WIA) on 29
August 18 and 19, 2015. These measurements are documented in Ref. 2. 30
3. Continuous monitoring at three VMS monitors from Monday November 23 through 31
Friday November 27, 2015. Thanksgiving Day was Thursday November 26, 2015 and a 32
football game was played at Husky Stadium on Friday November 27, 2015. The data 33
were provided by IEM. 34
4. Measurements performed on February 2016 to further investigate and document the 35
ambient vibration at the monitor sites in the subway and the train vibration levels at the 36
buildings. The results from those measurements are documented in Ref. 4. 37
The locations where train vibration measurements were performed in August 2015 are shown in 38
Figure 2. The measurements were performed on two mornings between midnight and 4 AM. The 39
first morning was focused on locations west of the subway and the second morning focused on 40
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 6
sites northwest of the subway. The measurements were performed with up to 20 seismic 1
accelerometers, four RION DA-20 four channel data recorders, plus a National Instruments 2
LabVIEW recording system with 12-channels. 3
The measurements west of the subway during the first morning of testing consisted of: 4
Four channels between the subway and Montlake Boulevard. 5
Three channels west of Montlake Boulevard. 6
Inside and outside measurements at CHDD and UWMC. 7
Vibration at each of the monitors. 8
The measurements the second morning of testing consisted of: 9
Four channels parallel to the north end of the station and Montlake Boulevard. 10
Two channels in a UW utility tunnel that was approximately halfway between the station 11
and Winkenwerder Hall. 12
Inside and outside measurements at Winkenwerder and Wilcox. 13
14
Figure 2: Surface Measurement Locations for Train and Ambient Testing (August 15
2015) 16
The supplementary measurements performed in February 2016 were just a few weeks prior to the 17
opening in March. Those measurements consisted of two phases: 18
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 7
(1) Measuring ambient vibration at three of the monitors, at the ground surface above the 1
monitors and close to Montlake Boulevard, and on one of the footings of the Montlake 2
Draw Bridge. These measurements were performed during the daytime to obtain 3
vibration from normal traffic. Videotaping of traffic also was performed to allow accurate 4
identification the trucks and buses that caused vibration events. 5
(2) Supplementary train vibration measurements using a test train that was purposely 6
selected because it had one or more wheel flats. 7
The purpose of the February measurements was to identify the source of the ambient vibration at 8
the monitors and to pin down the levels of train vibration in two of the buildings (CHDD and 9
UWMC). 10
A key feature of the test programs was time synching the recording systems. This was achieved 11
by assembling all of the systems in the same location and then triggering record simultaneously 12
on all of the devices using an electronic signal. The recording systems were then run 13
continuously until the end of the measurements, which ranged from 30 minutes to several hours. 14
There was some drift in the time synching between the five systems, but it was on the order of 15
10 msec or less, which was small enough to ignore. 16
Accelerometers were used for the measurements performed in August 2015 and geophones were 17
used for the February 2016 measurements. The accelerometers used had sensitivities of 1 V/g 18
and 10 V/g. In general, the 10 V/g accelerometers were used at the more distant locations. The 19
accelerometers used in the subway were all 1 V/g units. Due to instrumentation noise, the results 20
of the tunnel are not valid at frequencies below 10 to 20 Hz, which is the frequency range 21
focused on in this report. Therefore, the analysis was based on data from the VMS monitors and 22
the surface measurement positions. The geophones used for the February 2016 measurements 23
were similar to those used by IEM at the monitoring stations. 24
The accelerometer and geophone data were processed to obtain 1/3 octave band spectra at 1/2 25
second intervals. The acceleration data were then integrated in the frequency domain to obtain 26
vibration velocity levels. 27
MEASUREMENT RESULTS 28
Figure 3 illustrates typical data from the August 2016 measurements. On all five spectrograms 29
the horizontal axis is time from 1:10:30 to 1:20:30, the vertical axis is frequency from 2 to 40 30
Hz. Color indicates vibration amplitude. The vertical lines show when the test train passed 31
Monitor 5-1. Figure 3(a) shows the vibration at Monitor 5-1. The measurement at Monitor 5-1 32
was performed with an accelerometer that was not capable of accurately characterizing vibration 33
below 10 to 20 Hz and is included to show when the test train passed Monitor 5-1. In general, we 34
used the 40 Hz 1/3 octave band to identify when trains passed the monitors. The odd numbers 35
(T1-T7) are the test train operating in the northbound direction at speeds of 20 to 30 mph. The 36
even numbers (T2-T8) are trains operating in the southbound direction speed of 40 mph. 37
Observations from the results shown in Figure 3 are: 38
Figure 3(b): This position was directly above the tunnel. There are peaks at 40 Hz and 10 Hz for 39
all of the trains and a weaker peak at 20 Hz for trains in the southbound direction. At 40
frequencies below 8 Hz, the vibration appears to be largely uncorrelated to the train 41
operations. 42
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 8
Figure 3(c): This measurement position was on the east sidewalk of Montlake approximately 1
centered on the crossover box. Now the vibration shows substantially less correlation to the 2
train events although there is still an indication that train vibration caused a measurable 3
increase in the vibration levels in the 8 to 40 Hz range, although the variations between the 4
test trains suggests that traffic contributed to the vibration during train events. 5
Figure 3(d): This measurement position was west of Montlake approximately 30 ft from the 6
northbound traffic lane. The 10 volts/g accelerometer was used at this position compared to 7
the 1 volt/g accelerometers used east of Montlake. There is very little change in the vibration 8
levels during trains T1, T4, and T6. T1 was a northbound run at 20 to 30 mph and T4 and T6 9
were southbound runs at 40 mph. The other trains have vibration level increases that are 10
almost simultaneous with the train events, but offset by 10 to 30 seconds. Because these 11
vibration events are not synchronized with the train events and similar vibration signals are 12
not evident during the other train events, it is unlikely that these events are associated with 13
train operations. 14
Figure 3(e): This is the most distant surface accelerometer for this series of vibration tests. There 15
is very little indication of vibration increases during the train events although there are other 16
vibration events that generally are not synchronized with the trains. One interesting 17
observation in Figures 3(d) and 3(e) is about 3 minutes of vibration in the 2 and 3.15 Hz 18
bands starting at 01:13. This vibration fluctuates between about 25 to 40 dB and is not seen 19
for the remainder of the period shown in Figure 3. Similar periods of intermittent low-20
frequency vibration occurred during most of the measurements. We have not identified the 21
source of this vibration. 22
The key conclusion drawn from this close inspection of the vibration data at the ground surface 23
is that there is no indication that the train operations caused any vibration at frequencies of 8 Hz 24
and lower that was greater than the normal, middle of the night, background vibration. 25
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 9
1
Figure 3. Example Spectrograms of Surface Train Vibration 2 The vertical lines indicate when the test train passed Monitor 5-1. Odd numbers are northbound 3 runs at 20 to 30 mph. Even numbers are northbound runs at a speed of 40 mph. 4
(a)
(b)
(c)
(d)
(e)
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 10
Figure 4 shows the vibration from the two-car test train measured at Monitor 4-2. Figure 4a shows the vibration 1 from all the test runs that did not have simultaneous vibration from traffic (blue lines), the average train vibration in 2 red, and the L1, L5, and L10 exceedance levels during the train testing. The periods of the train operations were 3 removed before determining the exceedance levels. Train testing was performed between 1 AM and 4 AM when 4 traffic on Montlake was at a minimum. Figure 4a shows that even in the early morning hours when traffic was at a 5
minimum, train vibration was often lower than the ambient vibration at frequencies below about 16 Hz. Figure 4b 6 compares the average level of train vibration at Monitor 4-2 with the measured levels of truck and bus vibration 7 measured at Monitor 4-2 as the vehicles passed over the Montlake Bridge. From Figure 4b, it is evident that 8 vibration from trucks and buses passing over the Montlake Bridge usually exceed the train vibration at Monitor 4-2 9 over the 2 to 25 Hz 1/3 octave bands. 10
Because the vibration from traffic on Montlake Boulevard is lower than the vibration from vehicles passing over the 11 Montlake Bridge, the ambient vibration at the other monitors is lower than at Monitor 4-2. As a result, train 12 vibration in the 12 Hz and higher 1/3 octave bands could be determined without subtracting out the effects of 13 ambient vibration. 14
Figure 4. SB train vibration at Monitor 4-2 vs. ambient vibration 15
COMPARISON OF BUS AND LRV VIBRATION 16
In past projects when potential for interference with vibration sensitive equipment has been a 17
concern, the vibration from existing automobiles, buses, and trucks often has been overlooked. 18
Figure 5 shows a comparison of the force density levels derived from vibration testing that was 19
performed for the Central Corridor LRT project in Minneapolis (Ref. 3). The measurements were 20
performed in a corridor where both light rail trains and buses operated in parallel. Figure 5 shows 21
that bus vibration can be expected to be equivalent to and frequently greater than LRT vibration 22
at lower frequencies. The testing in Minneapolis shows that the crossover is in the 20 to 30 Hz 23
range. 24
Of course there are a number of factors that will affect the relationship between vibration 25
generated by buses and LRVs. These include speed, roadway condition, roadway construction, 26
track construction, track condition, wheel condition, and localized geology. 27
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 11
1
Figure 5. Comparison of Bus and LRV Force Density Curves, Minneapolis 2
3
THEORETICAL CONSIDERATIONS, MOVING LOAD IN TUNNEL 4
This section presents some theoretical considerations relative to the vibration in the tunnel and 5
when the quasi-static deflection of the tunnel by the train would cause measurable low-frequency 6
tunnel motion at the U-Link monitors. 7
The ground vibration displacement response to a sinusoidal load is essentially uniform or slightly 8
increasing with frequency for an infinite half-space. Thus, the velocity response of the ground 9
increases with increasing frequency at the rate of at least 6 dB per octave. Ignoring the effects of 10
soil layering, the displacement response of the ground is roughly inversely proportional to the 11
shear stiffness. For sources below grade, the dominant wave types are shear waves, compression 12
waves, and non-radiating near-field responses. Rayleigh surface waves form at a horizontal 13
distance of roughly five times the tunnel depth. Because the U-Link tunnel is approximately 14
100 ft deep, Rayleigh surface waves are not expected to be significant for the vibration in the 15
tunnel due to surface traffic along Montlake Blvd. Most of the vibration energy is carried in 16
shear. Thus, the ground vibration propagates through increasing depths of the ground with 17
decreasing frequency. The shear stiffness of the ground tends to increase with increasing depth 18
due to confining stress, further decreasing the response with decreasing frequency. 19
At the UW campus, the geologic materials consist of over-consolidated sands and gravels, 20
glacial tills, and silts that were deposited by glaciers. These overlay sediments at depth that also 21
have high shear stiffness, though slightly less than the overlying glacial deposits, as indicated by 22
shear wave velocity data for the Sound Transit alignment on the UW campus illustrated in Figure 23
6. The shear stiffness increases with depth down to about 50 feet below grade, below which the 24
shear wave velocity lessens slightly. At some depth, an increase of shear stiffness with increasing 25
depth is expected due to increasing confining stress. 26
The shear stiffness of the ground is proportional to the product of soil density and the square of 27
the shear wave velocity. As shown in Figure 6, the shear wave velocity on the UW campus is 28
about 1,500 feet per second, roughly twice the shear wave velocity of alluvial soils where over-29
consolidation has not occurred. Thus, shear stiffness is roughly four times that of typical alluvial 30
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 12
soils, which means that the response of the ground at low frequencies is perhaps 12 dB lower 1
than the response of alluvial materials in locations such as the Central Valley of California 2
assuming that all other factors are unchanged. The increased density of over-consolidated soils 3
also increases the shear stiffness and reduces the response. 4
The compression wave velocities shown in Figure 6 are generally much higher than the shear 5
wave velocities, reaching 7,500 feet per second, which is about 60% greater than the speed of 6
sound in water (4,700 feet per second). The high compression wave velocity relative to shear 7
wave velocity is typical of saturated soils. The soil is virtually incompressible. As a result, the 8
amplitude of the compression wave is very low compared to the shear wave amplitude. 9
These over-consolidated materials exhibit low material loss factor, with the result that dissipation 10
of propagating wave energy on the UW campus is low compared with alluvial or sandy materials 11
not subject to over-consolidation. This is one of the reasons that the vibration velocity at 12
frequencies above 30 Hz is more apparent than at frequencies below 20 Hz at the UW campus. 13
Observed vibration energy dissipation at low frequencies was negligible due to masking by 14
vibration from traffic on Montlake Boulevard and other sources within several hundred feet of 15
the tunnel. 16
The response at the ground surface is dominated by the non-propagating near-field at low 17
frequencies, where the source-receiver distance is roughly less than a wavelength. At 10 Hz, the 18
shear wavelength is about 150 feet, so that vibration at receivers at the ground surface above the 19
tunnel or at the tunnel bench adjacent to the track are significantly affected by the near-field. The 20
near-field is inversely proportional to the square or cube of the source receiver distance, and thus 21
attenuates rapidly with distance, and is non-propagating. The static deflection field of the moving 22
loads of the vehicle also produces a very low frequency non-radiating ground motion at very low 23
frequencies (see below). 24
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 13
1
Figure 6. Seismic velocities at various borings on the UW campus 2
Figure 7 illustrate the line source responses (LSR) measured at the tangent track south of the UW 3
cross-over box. A shaker was used at frequencies below 20 Hz and impact forces above 10 Hz. 4
The shaker data were taken at 1/3 octave band center frequencies, while the impact data involve 5
a continuous spectrum over the entire frequency range. Impact test results below 12.5 Hz are not 6
shown, due to poor coherence in the impact data. These data indicate that the LSR decreases 7
rapidly with decreasing frequency below 50 Hz, for the reasons cited above. At frequencies 8
below 8 Hz, the response is much less uniform, possibly due to near field effects, but also due to 9
the response of the ground, which may involve nodes in the response function related to layering, 10
or tunnel/soil dynamics. Background vibration may have also affected the data at some 11
geophones. 12
13
0 1000 2000 3000 4000 5000 6000 7000 8000
0
50
100
150
200
250
Seismic Velocity Summary
NB-123-P
NB-123-S
NB253-P
NB253-S
NB-255-P
NB-255-S
NB-259-P
NB-259-S
NB-354-P
NB-354-S
NB-356-P
NB-356-S
VELOCITY - FT/S
DE
PT
H -
FT
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 14
1
Figure 7. Comparison of LSTM’s Measured at Southbound Tangent Track South of UW 2
Station 3
Vibration forces produced by transit vehicles are roughly proportional to the acceleration of the 4
mass of the vehicle truck at frequencies between the primary resonance frequency of about 8 to 5
10 Hz and secondary suspension resonance frequency of around 2 Hz. Above the primary 6
resonance frequency, the un-sprung mass of the wheel set is expected to dominate the load up to 7
the resonance frequency of the resilient wheels, axle and direct fixation of the rail, which is about 8
40 to 60 Hz depending on rail fastener stiffness. The track forces due to a sinusoidal roughness 9
between the wheel and rail will thus decrease at a rate of 12 dB per octave with decreasing 10
frequency at frequencies between the primary resonance frequency and the wheelset/track 11
resonance frequency. 12
A peak occurs at the primary resonance frequency, which is controlled by the truck mass and the 13
parallel combination of the primary and secondary suspension stiffnesses. The frequency appears 14
to be about 8 to 10 Hz for the Kinkysharyo vehicle. Below this frequency, the track loading 15
again decreases with decreasing frequency at 12 dB per octave decrease in frequency until the 16
secondary suspension resonance frequency is encountered at about 2 Hz. These characteristics 17
are illustrated in Figure 8. In this figure of the theoretical invert force estimate for a one micro-in 18
amplitude sinusoidal roughness, the secondary, primary, and wheelset/track resonances are at 2, 19
9, and 36 Hz, respectively. 20
The rail roughness tends to decrease with increasing frequency at a rate of 3 to 6 dB or more per 21
doubling of frequency. This is partially compensated by the increasing bandwidth of one-third 22
octave band filters. The bandwidth is approximately 23% of the nominal center frequency, so 23
that the response of the third-octave filter increases by 3 dB per doubling of frequency. 24
-10
0
10
20
30
4 8 16 31.5 63 125 250
LSR
(D
B)
FREQUENCY (HZ)UWT LSR (TRAP RULE)
SHAKER & IMPACTS
SHAKER LSR (SB 1199+50)
SHAKER LSR (SB 1199+66)
IMPACT LSR (SB 1199+50)
IMPACT LSR (SB 1199+66)
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 15
In view of the above, one may expect rates of attenuation of the vibration velocity at the ground 1
surface to be of the order of 18 dB per halving of frequency, notwithstanding peaks in the 2
spectrum at the wheelset/track resonance, primary resonance, and secondary resonance 3
frequencies. 4
The roughness of asphalt or concrete road surfaces is generally much greater than the roughness 5
of high quality steel rails used on transit systems. The static load of the LRV at AW4 loading is 6
about 160,000 lbs, which may be compared with the static load of a heavily loaded tractor-trailer 7
of around 80,000 lbs. If the LRV were operated on rails with roughness equal to that of asphalt 8
or concrete pavement surfaces, the low frequency vibration produced by the LRV would likely 9
be substantially greater than that produced by trucks and buses. The smoothness of the wheels 10
and rails relative to that of the road surface is a major factor in the difference between vibration 11
produced by road vehicles and transit vehicles. 12
Tunnel motion due to the moving static loads of the vehicle trucks may be very apparent within 13
the tunnel in which the train travels, with spectral peaks at frequencies below 4 Hz. The 14
estimated 1/3 octave velocity levels for a 2-car train with three trucks per vehicle traveling at 40 15
mph are shown in Figure 9 along with the estimated train vibration levels observed at monitor 4-16
2. The peak at 1.6 Hz is due to the static loads of the individual trucks passing by. The estimated 17
levels of train vibration at monitor 4-2 that are included in Figure 9 are in the same general range 18
as the predictions based on static loading analysis over the 2 to 5 Hz range. At higher 19
frequencies, the estimated values from the measurements diverge sharply from the theoretical 20
value. This is the frequency range that the vibration shifts from non-radiating vibration related to 21
the moving static load to radiating vibration that is caused by the interaction of the rolling wheels 22
with the rail roughness. The ground motion due to moving static loads attenuate rapidly with 23
distance and are unobservable beyond roughly 50 ft. 24
Discrepancies between the measured and predicted vibration due to a moving static load may be 25
related to the distances between the trucks, here assumed to be 30 feet, and the actual distances 26
and various other factors. The secondary and primary suspension resonances should not be 27
factors for moving static loads. One may also expect the vibration response of an actual tunnel 28
wall to differ from the response of a semi-infinite half space with a vertical load as modeled here. 29
The moving loads are assumed to be at 100 feet below the surface. The receiver is assumed to be 30
10 feet from the track center and at mid tunnel height, 10 feet above the load application point. 31
The load of each truck is assumed to be 50,000 lbs. The soil parameters are assumed to be 32
similar to those found at the UW station. The calculations are based on an exact theory (R. D. 33
Mindlin, Physics, v7. 1936, pg. 195) for a uniform half-space. 34
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 16
1
Figure 8 Transmitted Force for Relative Roughness of 10−6 in. 2
3
4
1.00E-03
1.00E-02
1.00E-01
1.00E+00
1.00E+01
1 10 100
TRA
NSF
ER F
OR
CE
-LB
/MIC
RO
-IN
HCDF-MAGNITUDE
HCDF-MAGNITUDE
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 17
1
Figure 9. Estimated Vibration Velocity Level at Tunnel Wall for 2-car train traveling at 40 2
mph in the same tunnel 3
4
CONCLUSIONS 5
1. There is not much low frequency train vibration evident in the U-Link measurements. The 6
results show that ambient vibration from traffic and other sources at the monitor locations on 7
the tunnel safety walk often exceeded the low frequency vibration generated by test trains. 8
This non-intuitive result indicates that the Sound Transit LRVs do not generate much low-9
Saurenman, H.J., Nelson, J.T., Rajaram, S. Page 18
frequency vibration and trucks and buses tend to generate higher levels of low-frequency 1
vibration. Similar results have been observed at other transit systems. 2
2. The limited low-frequency vibration energy observed at the subway monitoring stations in 3
and south of the UW Station are consistent with other measurements of train vibration 4
measured in locations with overly consolidated soils. 5
3. Based on the soil properties in UW area, this result is consistent with the models that would 6
be used to predict the vibration properties. 7
4. Although this report focuses on when low-frequency vibration is not observed, it is important 8
to recognize that there are situations when low-frequency energy will be generated by LRV 9
operations. This report does not include specific examples of this because the focus is on why 10
low-frequency vibration from LRV operations has not been observed at the ground surface at 11
the north end of the U-Link extension. 12
5. For the U-Link extension in the University of Washington area, at frequencies below 10 to 13
20 Hz, we expect the ground vibration from buses and trucks to exceed the vibration 14
generated by LRVs. The observation includes the following qualifiers: (1) LRVs with good 15
condition wheels, (2) similar distances between the measurement positions and the vehicles, 16
and (3) the buses/trucks and LRVs operate at similar speeds. 17
6. The trigger thresholds for the U-Link vibration monitoring system should be set so that they 18
not are triggered by the vibration caused by the moving static loads of the LRVs, as these are 19
non-radiating vibration sources, nor by background vibration from external sources. 20
REFERENCES 21
1. ATS Consulting, “Final Report: Derivation of Transfer Functions for IEM Vibration 22
Monitoring System,” December 18, 2015. 23
2. Wilson, Ihrig & Associates, “Vibration Threshold Compliance Testing at the University 24
of Washington,” September 18, 2015 25
3. ATS Consulting, Central Corridor Light Rail Transit Project Final Environmental Impact 26
Statement, Appendix J4, “Vibration Measurements and Predictions for Central Corridor 27