You are on page 1of 5

Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 1009

Leuven, Belgium, 4-6 July 2011


G. De Roeck, G. Degrande, G. Lombaert, G. Müller (eds.)
ISBN 978-90-760-1931-4

Evaluation of Problem Floors Because of Human Induced Vibrations

Thomas M. Murray, PhD1, Brad Davis, PhD2


1
Department of Civil and Enviromental Engineering, Virginia Tech, Blacksburg, VA 24061, USA
2
Department of Civil Engineering, University of Kentucky, Lexington, KY 40506, USA
email : thmurray@vt.edu, bdavis@engr.uky.edu

ABSTRACT: In-situ testing of a lively floor is necessary for the design of an effective remedy. The ideal evaluation is to use
full experimental modal analysis, which requires substantial equipment including at least one floor shaker, to fully define the
natural modes of the floor system; however, it may not be possible, at least for an initial evaluation, because of cost. An
alternate testing procedure is to use a handheld spectrum analyzer with heel-drop and timed walking excitations. This paper
describes the procedure and compares results with experimental modal analysis techniques. The method is further evaluated
with results from a laboratory floor and an occupied building. The method is shown to be economical, convenient and effective.

KEY WORDS: Floor Vibration; Testing; Evaluation; Experimentation; Serviceability.

1 INTRODUCTION In our procedure, the evaluation begins with measurement


Occupant complaints because of lively floor motion are not and recording of ambient floor motion and floor motion
uncommon in North America. In-situ testing of such floors is caused by a member of the measurement team executing a
necessary to accurately determine the level and form of the heel-drop to determine the natural frequency of the problem
vibrations so that an efficient remedy can be determined. The area. A single heel-drop, which is usually applied at mid-bay,
ideal evaluation is to perform full experimental modal provides force excitation between approximately 1 Hz and 20
analysis (EMA) using an electro-magnetic or similar shaker Hz. Acceleration of the floor is recorded and analyzed during
and seismic accelerometers to fully define the natural modes the heel-drop test and is transformed to a spectrum from
of the floor. However, it is the authors’ experience that the which the natural frequencies are inferred. (Because the
cost and time required to mobilize, perform tests, and process applied force is not measured, it is not possible to determine
data from a full EMA testing program is prohibitive in the the frequency response function (FRF) for force and
majority of cases, at least for an initial evaluation. Also, the acceleration applied in the bay.)
disruption caused by shaker based testing may not be An example waveform and corresponding spectrum are
acceptable to the client. The following describes a simplified shown in Figure 1. In this case, the structure has two
testing procedure which has been used extensively by the responsive natural frequencies in the bay being tested: 10.7
authors to successfully evaluate lively floor motion in Hz and 14.3 Hz. The FRF shown in Figure 2 was estimated
buildings. The only instrumentation is a handheld spectrum using EMA techniques and an electrodynamic shaker. It
analyzer and a seismic accelerometer resulting in no shipping indicates natural frequencies at 10.8 Hz and 14.3 Hz which
costs and very rapid mobilization, testing, and data are practically identical to those inferred from Figure 1(b).
processing. The authors’ experience is that heeldrop tests almost
always provide adequate estimates of natural frequencies for
2 TESTING PROCEDURE the purpose of floor vibration evaluation. The major
exceptions are floors that are extremely unresponsive. For
Floor movement may become very large if an occupant walks
those, the resulting vibration will not exceed the ambient
at a subharmonic of the floor natural frequency. (It is the
vibration by an adequate margin, so the spectrum will not
authors’ experience that problem floor occupants report that a
contain peaks that are clear enough to allow natural
certain individual(s) is the cause of the highest level of
frequencies to be identified.
vibrations. The reason is that the individual’s natural gait
Upon identification of natural frequencies, a measurement
matches or nearly matches a subharmonic of a dominant floor
team member walks at a subharmonic of the measured
frequency. To accurately test a floor system for this forcing
dominant frequency (natural frequency that appears to be the
input requires substantial EMA testing equipment. The
most responsive in the bay being tested) while the floor
authors instead use ambient, heel-drop, and walking
motion is recorded. The subharmonic frequency is the
excitations when evaluating a floor to avoid the cost of
dominant frequency divided by an integer so that the resulting
bringing EMA testing equipment to a building. Ambient and
frequency is within normal human walking speeds, 90 to 140
heel-drop excitation measurements are made to determine the
steps per minute. A metronome with the beat set at the sub-
dominant natural frequencies of the floor system problem
harmonic frequency is used by the walker to obtain the
areas. Walking measurements are made to determine
desired step frequency. Timed walking is done along all
expected peak accelerations from normal human walking.
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 1010

possible walking paths near the area of concern, but is usually of a single degree of freedom system subjected to a harmonic
confined to one bay. forcing function with some modifications for occupant/walker
locations and effective mass. (It is noted that Reference [2]
4
Peak Accel. = 3.95%g
uses very a similar criterion, but rms rather than peak
Meas. Acceleration (%g)

acceleration for evaluation.) Thus, a single peak in a


2 measured record is not directly comparable to the peak
acceleration predicted using the Guide criterion so the
measured waveforms must be converted for realistic
0 comparisons. To do this, the recorded acceleration versus
time waveforms are filtered to a 1-15 Hz bandwidth to
eliminate possible DC offset and high frequency accelerations
-2 which are not felt by humans. Then, a rolling two second rms
acceleration is computed and the center acceleration for each
interval converted to an equivalent sinusoidal peak
-4
0 2 4 6 acceleration (ESPA) by multiplying the rms value by √2. The
Time (sec.) maximum of these values, converted to percent, is then
compared to the Guide criterion, 0.5%g.
(a) Waveform.
3 EVALUATION OF METHOD USING A
0.05 LABORATORY FLOOR
Meas. Peak Acceleration (%g)

X: 10.7 The laboratory floor shown in Figure 3 was used to verify the
0.04 Y: 0.0382
testing procedure. The floor was a single 9.14 m by 9.14 m
X: 14.25
Y: 0.03271
(30 ft by 30 ft) bay constructed and tested at the Virginia Tech
0.03 Thomas M. Murray Structural Engineering Laboratory in
2006. The 222 mm (8-3/4 in.) composite slab consisted of a
0.02 117 mm (4-5/8 in.) steel deck covered with 105 mm (4-1/8
in.) of 24 MPa (nominal 3.5 ksi) normal weight concrete. The
0.01 floor was supported only at the perimeter with W530x66
(W21x44) girders and W360x32.9 (W14x22) beams which
framed into W310x60 (W12x40) steel stub columns.
0
8 10 12 14 16 18 Temporary shoring was provided at the third points of the slab
Frequency (Hz) to limit the dead load deflections during the concrete
placement. The slab supported no nonstructural elements such
(b) Spectrum. as cladding or partitions.
Figure 1. Example heeldrop test result.
0.04 Deck Span
FRF Magnitude (%g/lbf)

X: 14.3
Y: 0.03092
X: 10.8
0.03 Y: 0.02815

W14
W21
0.02

0.01 (a) Photograph.

0
8 10 12 14 16 18
Frequency (Hz)
Figure 2. Example FRF magnitude.
The acceptance criterion commonly used in North America (b) Section
is in the American Institute of Steel Construction/Canadian
Institute of Steel Construction Design Guide 11 Floor Figure 3. Laboratory floor.
Vibrations due to Human Activity, Chapter 4 [1]. The Prior to walking tests, the first natural frequency of the floor
criterion states that the peak acceleration due to walking was found to be 4.98 Hz. Walking tests were then performed
should not exceed 0.5% of gravity (5 milli-g’s) for offices and by measuring the acceleration at the center of the specimen
residences. The Guide provides a method to predict normal due to a single individual walking across the slab as shown in
walking acceleration, which is based on the resonant response Figure 4. Using a metronome, the intended walking speed
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 1011

was 1.66 Hz or 100 bpm (4.98Hz/3 x 60 bpm) which was the differences are not as large because the input energy is spread
third subharmonic. The acceleration record with the over narrow bands instead of being a sinusoid at one spectral
maximum recorded ESPA is shown in Figure 5. The 2 sec. line. However, the differences in acceleration due to small
RMS acceleration is the dashed line and the maximum value step frequency variations are still often sizeable. (Occupied
is 0.753 %g. The ESPA is 1.06 %g, which is reasonable since building floors usually have much higher damping resulting in
EMA determined damping was only 0.5%. a much wider and shorter peak, creating much smaller
acceleration differences due to small step frequency errors.)

0.5

FRF Magnitude (%g/lbf)


X: 4.975
0.4 Y: 0.3638

0.3
X: 5.025
0.2 Y: 0.1743

0.1

0
4 4.5 5 5.5 6
Figure 4. Walking tests. Frequency (Hz)
Figure 6. Laboratory floor FRF magnitude.
Measured Acceleration (%g)

1.5 Peak Accel. = 1.33%g


The ESPA from each measurement was plotted versus the
1 step frequency to investigate the effect of step frequency on
the response. First, an indirect step frequency measurement
0.5 method was developed. Figure 7 shows a portion of the
0 measured acceleration spectrum (acceleration plotted vs.
frequency) in the vicinity of the step frequency and double the
-0.5 step frequency for one of the walking tests. In this case,
characteristic peaks can be seen at 1.69 Hz and 3.38 Hz which
-1
are approximately the first and second harmonic of the
-1.5 Max. 2 sec. RMS Accel. = 0.753%g intended step frequency, 1.66 Hz. The third harmonic in this
case is then estimated to be 4.98 Hz and it is this harmonic of
0 2 4 6 8 10
Time (sec.) walking that is nearest to the specimen’s natural frequency
and causes resonance.
Figure 5. Laboratory floor walking acceleration trace.
Even when a metronome is used, most walkers attempting 0.02
Measured Acceleration (%g)

to match a set beat can be expected to err, perhaps 1-2 bpm,


X: 3.375
especially when only taking a few steps as is the case when Y: 0.01841

walking across the 9.14 m (30 ft) laboratory floor. Note that 0.015
usually the test walking speed attempts to cause resonance by
matching the third or fourth harmonic to the dominant X: 1.688

frequency. A 1 bpm error in step frequency translates into a 3 0.01 Y: 0.009069

bpm (0.05 Hz) and 4 bpm (0.067 Hz) error at the third and
fourth harmonics, respectively. Further, because a very
0.005
lightly damped system such as the laboratory floor or an
unoccupied building has a very narrow and tall FRF peak,
walking speed error can result in less than the theoretical peak 0
acceleration. For example, using a shaker and force plate, 1 2 3 4
measured accelerance (ratio of acceleration output to force Frequency (Hz)
input, both at mid-bay) FRF is shown in Figure 6. From this Figure 7. Step frequency identification.
plot, it is obvious that very small differences in sinusoidal
force input frequency result in very large differences in Figures 8 and 9 show the ESPA for all parallel and
acceleration response. For example, at 4.975 Hz and 5.025 perpendicular to the deck span walking tests, respectively.
Hz (a mere 3 bpm frequency difference), the accelerance The maximum acceleration occurred at 1.69 Hz, with the third
values are 0.364 %g/lbf and 0.174 %g/lbf, respectively. harmonic nearly matching the measured natural frequency of
Walking acceleration differences due to small step frequency 4.98 Hz. The absolute maximum accelerations were 1.06 %g
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 1012

and 0.836 %g for walking parallel and perpendicular to the Figure 11 shows the heeldrop test waveform and spectrum
slab span, respectively. which indicates that the dominant natural frequency is 5.85
The maximum ESPA, 1.06 %g, exceeds 0.5 %g which is Hz and another natural frequency exists at 7.95 Hz, both of
the recommended peak acceleration limit for residence and which can be excited by one of the first four harmonics of the
office occupancies in Reference [1]. If the slab was used in an walking force.
actual structure, the damping would likely be 2.5% to 5% of
critical damping as compared to 0.5% which was determined
during EMA tests. Therefore, this slab would surely be 6.5m 6.5m 6.5m
adequate in an occupied building.
W610x174 W610x174 W610x174

1.2 Tested Bay


Peak Acceleration (%g)

0.8

0.6

0.4

0.2 Walking Paths

0 W610x92
1.5 1.6 1.7 1.8 1.9 2 W610x82 W610x92
Step Frequency (Hz)
Figure 8. Laboratory floor ESPA vs. step frequency (all tests,
walking parallel)
Figure 10. UK Pharmacy partial framing plan

1.2 1
Peak Acceleration (%g)

Meas. Acceleration (%g)

1
0.5
0.8

0.6 0
0.4

0.2 -0.5

0 Peak Accel. = 0.912%g


1.5 1.6 1.7 1.8 1.9 2 -1
Step Frequency (Hz) 0 2 4 6
Time (sec.)
Figure 9. Laboratory floor ESPA vs. step frequency (all tests,
walking perpendicular) (a) Waveform
Meas. Peak Acceleration (%g)

4 CASE STUDY
A large bay in the Second Floor of the University of Kentucky 0.03
College of Pharmacy building was also tested and those
results are used to further illustrate the testing and evaluation X: 5.85
Y: 0.01912
procedures. 0.02
The tested bay supports tables, very light book shelves,
computer desks, a small medication storage area, and no
partitions. The floor was constructed using a conventional 0.01 X: 7.95
Y: 0.007192

composite slab with 190 mm (7-1/2 in.) total depth and 76


mm (3 in.) deck. As shown in Figure 10, the slab is supported
by 20 m (66 ft) long W920x201 (W36x135) beams at 3.25 m 0
(10 ft 8 in.) spacing and 6.5 m (21 ft, 4 in.) long W610 (W24) 2 4 6 8 10
girders. Walking paths traversed the bay parallel to and Frequency (Hz)
perpendicular to the beams as shown in Figure 10. (b) Spectrum
Figure 11. Heeldrop test results for UK Pharmacy
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 1013

Figure 12 shows the waveform and spectrum that were REFERENCES


recorded during the walking test with maximum response. [1] T.M. Murray, D.E. Allen, and E. Unger, Steel Design Guide Series 11,
The waveform in Figure 12(a) clearly shows a resonant build- Floor Vibrations due to Human Activity,, American Institute of Steel
up due to the third harmonic of the walking force. Note the Construction, Chicago, IL, USA 1997.
repeating pattern of an increase in the vibration peak [2] P.J. Devine, A.L. Smith, and S.J. Hicks, Design of Floors for Vibration:
a New Approach, SCI P354, The Steel Construction Institute, Silwood
amplitude followed by two lower peaks and then another
Park, Ascot, Berkshire, UK, 2007.
increase. The spectrum shown in Figure 12(b) also indicates
that the majority of the response is due to the third harmonic
excitation of the 5.85 Hz dominant natural frequency.
Interestingly, the fourth harmonic of the walking force almost
matches the 7.95 Hz natural frequency, so it contributes
significantly to the response as well. The maximum 2 second
RMS acceleration, 0.160 %g is converted to 0.226 %g ESPA
which does not exceed the recommended limit of 0.5 %g for
offices and residences. Therefore, no vibration complaints
should be expected due to walking in this area and none have
been reported.

0.4 Peak Accel. = 0.376%g


Meas. Acceleration (%g)

0.2

-0.2

-0.4 Max. 2 sec. RMS Accel. = 0.16%g


0 2 4 6 8 10
Time (sec.)
(a) Waveform
Meas. Peak Acceleration (%g)

X: 5.8
Y: 0.03065

0.03

0.02 X: 7.725
Y: 0.0161

0.01
X: 1.95
Y: 0.002823

0
2 4 6 8 10
Frequency (Hz)
(b) Spectrum
Figure 12. Walking test results for UK Pharmacy

5 CONCLUSION
It is shown that a simplified experimental program using only
a handheld spectrum analyzer and human induced floor
excitations can be used to evaluate floors for possible
annoying vibrations. The proposed procedure is economical,
fast, and causes only minor interruptions in an occupied
building.

You might also like