Professional Documents
Culture Documents
Ergonomics
Ergonomics
Ergonomics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/terg20
To cite this article: ANDERS KJELLBERG , BENGT-OLOV WIKSTRÖM & ULF DIMBERG (1985) Whole-body vibration: exposure time
and acute effects — experimental assessment of discomfort, Ergonomics, 28:3, 545-554
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
ERGONOMICS, 1985, VOL. 28, No.3, 545-':554
I. Introduction
Most studies of how acute effects of whole-body vibration change with exposure
time have evaluated the effects with measures of discomfort (see reviews by Clarke
(1979) and Kjellberg and Wikstrom (1985». This kind of data has also been quoted as
the main evidence for (von Gierke 1975) or against (Clarke 1979) the time-dependency
proposition of the international standard for whole-body vibration (ISO 1978).
If a comparison is to be made between the ISO standard and empirical data on the
development of discomfort, a measurement technique must be used which gives results
expressed in terms of vibration intensity and not in arbitrary subjective units. For
example, the results should indicate how much a 1 min vibration must be increased to
render a feeling of discomfort similar tothat produced by the same level of vibration
after a 1 hour exposure. This type of result is provided by the matching method, which
requires the subjects to indicate the intensities at which vibrations of varying durations
give rise to the same discomfort as a reference vibration of constant duration. The
matching may be accomplished by an adjustment method in which the subject 'adjusts
one of the vibrations, or by an estimation method in which the experimenter controls
the intensity and the subject has to judge which of two vibrations causes greatest
discomfort.
The matching methods have several other advantages. Subjects are given a natural
task which makes no demands on the ability to handle numbers in contrast to
magnitude estimation. It is also probable that vibration matchings are less influenced
than magnitude estimation by confounding factors such as the monotony of the
situation or strain produced by sitting for a long time in the same position. Finally, it is
much more difficult to remember a vibration setting than a rating (e.g. a number or a
546 A. Kjellberg et at
point on a graphic rating scale). This reduces the risk of a more or less conscious biasing
of results by the subject; an important point since, if discomfort increases over time,
such an increase is probably rather unconspicuous, In such difficult rating tasks there is
a considerable risk of subjects basing their reports on their hypothesis of the time
dependency rather than on the discomfort actually experienced.
However, vibration matching methods have two definite disadvantages. First, they
do not permit subjects to make a series of'matchings at different points of time during a
continuous exposure; the exposure must be interrupted during the matching proce-
dure. Secondly, there is a great risk of the reference vibration inftuencingtheeffect of the
vibration which is to be adjusted or judged. If this happens, the results become
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
uninterpretable. .
The difficulties above are reduced by using cross-modality matching instead of
vibration matching. The subject may be instructed to adjust a sound to the level where
it gives rise to the same discomfort as the vibration at a certain point of time. This
method permits matchings to be made during the vibration exposure and hence saves
time. During one exposure, a series ofmatchings can be made at different points oftime.
There is also far less risk that the vibration exposure will influence the effect of a sound
, rather than that of another vibration. Experiments with sound-vibration matchings
have been performed (Fleming and Griffin 1975, Hempstock and Saunders 1976),
although never with the aim of studying effects of exposure time. If it is shown that the
sound settings increase over time, this would imply that the vibration becomes
increasingly uncomfortable, provided that it can be shown that the sensitivity to sound
is constant during the vibration exposure. The sound settings can be transformed into
vibration levels provided that, for each individual, one knows the increase of vibration
level corresponding to a certain increase of sound level. Transformed in this way the
results may be directly compared-with the time dependency proposed by ISO 2631.
2. Method
2.1. Subjects
Fifteen men -volunteered to participate in the experiment. Their age, height and
weight varied between 23 and 43 years (mean 28'9),167 and 190cm (mean 180),64 and
88 kg (mean 74'1), respectively. None of them suffered from stomach or back trouble
and none had any professional driver experience.
2.2. Apparatus'
An electrohydraulic shaker (Fairey) generatedwhole-body vibration in the vertical
direction (z axis). The shaker was controlled withacceleration signals from an FM tape
recorder (Tandberg 100). The signals were recorded on the floor in the cabin of two
. industrial trucks (see below). The choice of vibration level and the on-off of the shaker
was controlled by the experimenter by means of a personal computer (ABC 80):
Vibrations were of two types. One was recorded in a 12 ton forklift truck during
normal driving outdoors on asphalt. The resonance frequency of this vibration was
3·1 Hz. The other had a resonance frequency of 6·3 Hz and was recorded in a 1·5 ton
forklift truck during normal driving indoors on steel plates. Thus, both vibrations had
random characteristics with a marked resonance frequency (figure 1). Of the recorded
vibration signal a period of 64 s which had a small r.m.s.-level variation was chosen. The
shaker was driven by an endless loop made out of this period.
Whole-body vibration-experimental assessment 547
0.80 A 0.80 /
I
/ \
0.31 / \ 0.31
/\ \
/ V\ '\
0.12
\ 0.12
I r-..
/ 1\ z / \ z
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
1/ r---~ II ~
0.05 0.05
0.4 1.0 2.5 6.3 16 40 100 Hz 0.4 1.0 2.5 6.3 16 40 100 Hz
Figure 1. Third octave spectra of the 3-1 and 6·3 Hz vibrations used in the experiment.
The subject sat on a conventional driving seat with a thin seat cushion but no other
built-in spring and damper system. The seat was mounted on the shaker. A relaxed
sitting position was adopted with a 100° trunk angle, thighs horizontal and knee angle
90-100°. Arms were on the thighs and feet on a footplate. A seat belt was not used.
The subjects matched a sound to the vibration. This was presented by means of a
tape recorder, an amplifier and earphones (Telephonics TDH 39 Pl. The experimenter
determined the sound level by means of a r.m.s.-meter (Briiel and Kjaer 2425). The
sound level was adjusted by the subject with a three-turn linear potentiometer. Sound
was switched on and offby the experimenter. The initial sound level was set high or low
by means of a switch (difference 11 dB).
The sound had a frequency spectrum as shown in figure 2 and could be varied
continuously from 115 dB SPL to less than 15dB SPL. Calibration of the sound
equipment was made with an artificial ear (Briiel and Kjaer 4153). The surrounding
sound level at the subject's ear was 50 dB(A). The infrasound level was 76 dB (linear
2-20 Hz).
dB
100
-- - \
90
80 V
.,.'/
70 / \
I
\
60 I \
20 ·50 125 315 800 2k 5k Hz
Figure 2. Third octave spectrum of the sound used for the cross-modality matching.
548 A. Kjellberg et at
The r.m.s. value of the acceleration signal, frequency weighted with a filter correspond-
ing to the weighting curves in ISO 2631, was 1·4m/s 2 r.m.s. for the 3·I.Hz vibration. and
. ]'6 m/s 2 r.m.s. for the 6·3 Hz vibration. The control condition was primarily included as
a control for a pilot study of physiological responses during the exposure (results not
reported here). A second purpose was to control for possible changes in sound settings
not caused by the vibration exposure.
The order of the three conditions was balanced over the subjects. The day before
both exposure conditions the subjects took part in a sound-vibration scaling session to
establish the individual sound-vibration functions. Thus, each subject participated in
five sessions which took place on separate days.
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
noise emitted from the TV when the transmitter was closed down at the end of the
evening. After a series of training trials six settings of each type were collected.
3. Results
3.1. Sound-vibration calibration
The mean sound settings of the 15 subjects at the different intensities of the 3·1 and
6'3 Hz vibrations are shown in figure 3. For both frequencies sound pressure level
(log Pa) was found to be a linear function of vibration level (log m/s 2 ). The frequency of
the vibration did not affect the regression (the coefficients of regression were I:84 and
1-87 for 3-1 and 6·3 Hz, respectively). Thus, a 1 dB increase of the vibration level
correspondence to an Increase of the sound level by about 1·85 dB. The settings to the
3'} Hz vibration were consistently about 2 dB higher than those to the 6·3 Hz vibration.
Large variations were found between the individual sound-vibration functions.
The coefficients of regression varied between 1 and 3'1 for the 3·1 Hz vibration (11 ofthe
coefficients lay between 1·4 and 2'1) and between 1·1 and 2·8 for the 6·3 Hz vibration
(with nine of the coefficients lying between 1·4 and 2'1). In the transformation of each
subject's sound settings to vibration levels these individual coefficients of regression
were used.
Sound
dB r-------------------,
85
dB
Vibration'
Figure 3. Mean sound pressure levels (± S.D.) at which the sound gave rise to the same degree
of discomfort as the four intensities of the two vibrations (0, 3·1 HZ; 0,6'3 Hz).
550 A. Kjellberg et al.
separate analyses. One was based on nine points of measurement (1, 2,3,4,6,8, 10, 12,
and 14) and the other on all points starting from point four. Both analyses yielded the
same results. Log m/s 2 increased linearly as 'a function of log time in both cases
(F(I, 14)= 17·2 and 18-4, respectively, p<O'OOI) and this linear change did not differ
between the two frequencies, Furthermore, there was no tendency for the rate of
increase to change during the exposure period, i.e. the quadratic trend component was
not significant in any case. .
The individual results are given in figure 6. It is evident from this figure that the
individual results did not in all cases correspond to the average trend. In both
conditions the settings of 12 subjects showed an increasing trend of the level with
exposure time and three showed a decreasing trend. For some subjects, however, the
trend was rather irregular. The figure also shows that the two frequency conditions
tend to give similar results within individuals; the correlation between the individual
regression coefficients in the two conditions being 0·64 (p<O·OI).
m/s' ,....---------------------,
rms
1.4
+
.25 4 16 64 min
2~H
1.4
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
.25 4 16 64 min
Figure 5. The development of discomfort during the exposure to the 6·3 Hz vibration.
Mean sound settings (transformed into vibration levels) as a function of exposure time
(± S.E. of the mean).
~ ~ ~
.~
~ ,.
~.-- ~ -'v--f'/I.- ~
~
.
~
~ / ~
->
. M u
3.1 Hz
-----.J'.- ~
~ .~ --vJ\--
A C •
~ ~ ~
~
F
-----------
G J
~
~ ~ ~ ~
. 6.3 Hz
~
Figure 6. Individual sound settings of subjects A-O (transformed into vibration levels) at
0·25--64 min exposure time.
552 A. Kjellberg et at
.from the exposure session of the settings of the TV and vacuum-cleaner noise were,
however, on the average about 1dB higher in sound pressure level after the 64 min
vibration exposure than before. For about half of the subjects the trend was in the
opposite direction, and therefore the differences between the two settings before and
after the exposure were not significant. The four correlations between the settings
before and after varied between 0·83 and 0·87. The control settings of sound thus imply
that there was no consistent change in sensitivity to sound in the group during the
64 min experiments.
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
4. Discussion
In summary; the experiment showed that the sound was adjusted to increasingly
higher levels as a function 'of exposure time. The control condition and the control
settings before and after exposure imply that this increase was unlikely to be a
consequence of a lowered sensitivity to noise. Thus, discomfort appears to grow as a
function of exposure time at least up to 1 hour's exposure. The alternative interpret-
ation of the results would be that the sensitivity to noise changed in some other way
during vibration exposure than in the control condition and that this change
disappeared immediately after the end of exposure (and thus could not be detected by
the control settings). This interpretation appears rather implausible. Furthermore, the
experiment showed that the change of discomfort expressed as log m/s 2 may be a linear
function of log exposure time.
The individual results show that the trend over time varied widely both within and
between subjects. This probably reflects the difficulty of the task rather than genuine
differences.The scaling session showed that subjects were obviously able to make fairly
reliable matchings in this cross-modality matching task, where the relatively short
intervals between vibration stimuli made the task one of relative judgements. During
the exposure session, however, the changes were so much less obvious and the time
intervals so much longer that such relative judgements were not possible. Instead the
subjects had to make an absolute judgement, namely, to find the point of subjective
equality between the vibration and the sound at a certain moment. This judgement,
being of a more arbitrary character, is obviously more difficult to make, and could be
expected to result in greater measurement variability.
The functions obtained between vibration and sound with respect to their
subjective effects imply that a 1dB increase of a vibration corresponds to slightly less
than a 2 dB increase of the sound. This fits reasonably well with results presented by
Flemming and Griffin (1975) and with the theoretically expected function (Stevens
1975). Hempstock and Saunders (1976) let their subjects adjust either the sound or the
vibration. They obtained widely different functions in the two conditions, neither of
them corresponding very well to the present results.
The observed time dependency deviates from that proposed by ISO 2631 in two
respects. First, the standard suggests that the effect is almost constant during the first
10 min of exposure, whereafter it starts to increase. The results give not support for such
a notion. Secondly, the rate of increase given by the standard is considerably higher (a
regression coefficient of 0,5) than that observed in the experiment (a regression
coefficient of 0-05). If the rate of increase really were as high as proposed by ISO 2631,
then many of the earlier studies should have demonstrated a change over time, in spite
of the fact that' the methods used were probably less sensitive than that used in the
present study.
Whole-body vibration-experimental assessment 553
However, the regression coefficients also depart from those obtained in an earlier
study from this laboratory where very short exposure times were investigated with
vibration matching technique (Kjellberg and Wikstrom i 984).The change over a 2 min
exposure to a 6·3 Hz sinusoidal vibration was there found to follow a linear function
between log acceleration level and log exposure time with a regression between 0·1 and
. 0'17. It is probable that this constitutes a better estimate of the regression over time
since the short exposure period permitted the use of a method which is considerably
easier for the subject; the matching was done with an identical vibration instead of a
sound.
As has been pointed out by Poulton (1979)and Stevens (1975), the contraction bias
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
(also called the regression effect) is strengthened when the reference and matching
stimuli are made more qualitatively different. This contraction effect is a general
problem in connection with subjective measurements and denotes the fact that subjects
tend to make underestimations at high levels and overestimations at low levels. It
should be noted that a contraction bias almost certainly occurred during both the
exposure session and the scaling session, thus affecting the regression coefficients in
both sessions. However, as was pointed out above, the task during the exposure session
was more difficult, which makes it probable that the contraction bias in this case was
stronger (see, for example, Stevens 1975, p. 273). The contraction bias therefore could
have lead to an underestimation of the growth rate of discomfort over time.
The important result of this experiment is thus -that the increase in subjective
discomfort probably is linear up to at least 1hour. This means that the rate of increase
observed during short exposures (Griffin and Whitham 1980, Kjellberg and Wikstrom
1985,Miwa 1968)could at least be extrapolated to the first hour of exposure. However,
the results from these studies differ considerably regarding the regression coefficients,
and further studies are therefore needed.
Die Methode des Kreuz-Modalitats Vergleichswurde eingesetzt, urn die Entwicklung von
Diskomfort wiihrend einer einstiindigen Ganzkorperschwingungs.~xposition zu untersuchen.
Die Aufgabe der Person bestand darin, ein Breitbandgerausch auf das Niveau einzustellen, bei
dem es den gleichen Grad an Diskomfort bewirkt wie eine Schwingung. Regellose vertikale
. Schwingungen mit einer Resonanz von 3,1 oder 6,3 Hz wurden verwendet, Die. Gerausch-
einstellungen wurden in Schwingungsniveaus iibertragen durch Bestimmung von Gerausch-
Schwingungs-Funktionen, die fur jede Person berechnet waren. Die Gerauschniveau-
Einstellungen stiegen als eine Funktion der Expositionszeit an und Kontrollmessungen zeigten,
daB das nicht die Auswirkung einer geringeren Empfindlichkeit gegeniiber Gerausch sein
554 Whole-body vibration-experimental assessment
konnte. Der Anstieg, ausgedriickt in log Beschleunigung, war eine lineare Funktion der log
Expositionszeit. Daraus wird geschlossen, daB Ergebnisse von Untersuchungen mit kiirzerer
Expositionszeit auf Expositionsperioden von mindestens einer Stunde extrapoliert werden
miiBten. Obwohl die benutzte Methode wahrscheinlich eher zu einer Unterschatzung des
Anstiegs des Diskomfort iiber der Zeit fuhrte, liiBt das Experiment darauf schlieBen, daB der
Anstieg in dem ISO-Standard 2631 uberschatzt ist.
i:;:i:tlllHliJO) 1 ~/lllIliR~O)l'~~O)~fttO: -:n, -r~« ¢ t: Ib ~0::7 0 A ' 'f::1" I) 7" -{ '"7 ';I 'f /';:fit. ~
fIl~' -rlilf;l:{rfi'? te; *~;gO)W~'W:, a;:'rff~W!-a0)V«lv2:'~JlIJtJ>(, t: <? "t1'~~ I:: Iii] J; to: t.t. ¢~t;:
~§ii"t ¢ t O)1.'~ s, ~iR7JrtiJO)'7 /':1"AlillJ~ fIl ~), ;I;l;liJ1!311UHl: 3.1 Hz tJ> 6.3HzO) I::'"t:> <? tJ'1.'~
s, {fO)lN:)i::Ij:~*~;gt;:':n, -r)i::lt.> t.:lf-lillJml~ 10: J: I) liJllJ V«/ldO:~~ L..t.: .:
lfO) v«lvlj:llilll~/lllO)ml~ I:: L- -rmtJDL-, :J /' )- 0 -lvilll)i::J: I). ~ i1..1j:W!lfto:j;j"t -!>~gril!:rO)
Downloaded by [Universitaets und Landesbibliothek] at 13:26 21 November 2013
References
CLARKE, M. 1.,1979, A study of the available evidence on duration effects on comfort and task
proficiency under vibration. Journal of Sound and Vibration, 65, 107-123.
FLEMMING, D. B., and GRIFFIN, M. 1., 1975, A study of the subjective equivalence of noise and
whole-body vibration. Journal of Sound and Vibration, 42, 453-461.
GRIFFIN, M. J., and WHITHAM, E. M., 1980, Discomfort produced by impulsive whole-body
vibration. Journal of the Acoustical Society of America, 68, 1277-1284.
HEMPSroCK, T. I., and SAUNDERS, D. 1., 1976, Cross-modality determination of the subjective
growth function for whole-body vertical, sinusoidal, vibration. Journal of Sound and
. libration, 46, 279-284.
ISO, 1978, Guide for the evaluation of human exposure to whole-body vibration. Standard 2631,
International Organization for Standardization.
KJELLBERG, A., and WICKSTROM, B.-O., 1984, Subjective effects of whole-body vibration with
short duration. Journal of Sound and Vibration (in the press).
KJELLBERG, A., and WIKSTROM, B.-O., 1985, Whole-body vibration: exposure time and acute
effects-a review. Ergonomics, 28, 535-544.
MIWA, T., 1968, Evaluation methods for vibration effect. Part 7: The vibration greatness of the
pulses. Industrial Health, 6, 143-164. .
POULTON, E. C., 1979, Models for biases in judging sensory magnitude. Psychological Bulletin, 86,
777-803. '.
STEVENS, S. S., 1975, Psychophysics. Introduction to Its Perceptual Neural and Social Prospects
(New York: WILEY). .
VON GIERKE, H. E., 1975, The ISO standard: 'Guide for the evaluation of human exposure to
whole-body vibration'. Paper presented to '1975 Ride Quality Symposium', NASA TM X-
3295, DOT-TSC-75-40.
. WINER, B. J., 1970, Statistical Principles in Experimental Design (New York: MCGRAW-HILL):