Professional Documents
Culture Documents
ISSN 2190-5452
Volume 4
Number 3
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag Berlin Heidelberg. This e-offprint is
for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.
1 23
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208
DOI 10.1007/s13349-014-0076-5
ORIGINAL PAPER
Received: 18 December 2013 / Revised: 18 March 2014 / Accepted: 21 March 2014 / Published online: 3 April 2014
Springer-Verlag Berlin Heidelberg 2014
Abstract In a bridge design code, the design live loads Keywords Bridge weigh-in-motion Externally
should be based on actual traffic loads. In the past, the restrained deck slab Gross vehicle weight
information about the actual truck loads on highway Steel girder bridge Strain gauge Truck
bridges was obtained from truck surveys, in which the
trucks are stopped for measurement and weighed on ‘static’
weighing scales. During the past few decades, however, the 1 Introduction
information about truck loads is collected either by pave-
ment weigh-in-motion or bridge weigh-in-motion (BWIM) In a bridge design code, the design live loads should be based
systems. This paper provides details of three BWIM on actual truck loads. In the past, information about the
methods, which were calibrated with extensive testing on a actual traffic loads on highway bridges was obtained from
slab-on-girder bridge in Winnipeg, Canada. The first two truck surveys, in which the trucks were stopped for mea-
methods are based on the assumption that truck loads could surement and weighing on ‘static’ weighing scales. During
be represented by an equivalent uniformly distributed load. the past few decades, however, information about truck loads
The first method uses the asymmetry in the shape of the is collected while the trucks move at normal speeds. Some of
bending moment diagram to calculate the gross vehicle the weigh-in-motion (WIM) scales are installed in the
weight (GVW), while the second method compares the pavement. In other WIM systems, a highway bridge is used
girder response at two instrumented transverse sections. for the weigh-in-motion of the trucks; this latter system is
The third method calculates GVW using strain signal area referred to herein as the bridge WIM (BWIM) system, and
as proposed by Ojio and Yamada in 2002. The results show the system installed in the pavement as simply the WIM
that the first method was inconsistent in its accuracy; the system. The WIM systems are significantly more expensive
second method’s accuracy reduced with the length of the and are not more accurate than the BWIM systems.
vehicle. The third method proved consistent and accurate A number of BWIM systems, with varying degrees of
to within 5 % for the test trucks provided that the vehicle accuracy in predicting the axle and gross vehicle weights
speeds could be estimated accurately. Accurate calcula- (GVW) of the trucks, have been introduced. An interna-
tions of the truck speeds were achieved by using the time tional society for weigh-in-motion (ISWIM) is already in
delay in the peak responses of the straps that confine the existence; this society deals with both the WIM and BWIM
externally restrained deck slab of the bridge. systems. This paper introduces three BWIM systems for
determining the gross vehicle weights (GVWs). Two of the
methods were developed by the authors, while the third is
an application of an existing method.
2 Literature review
K. Helmi (&) B. Bakht A. Mufti
ISIS Canada Resource Centre, A250-96 Dafoe Rd, Winnipeg,
MB R3T 2N2, Canada
Moses [1] is the first researcher to introduce a BWIM
e-mail: Karim.helmi@ad.umanitoba.ca system; his method, determines the truck configuration,
123
Author's personal copy
196 J Civil Struct Health Monit (2014) 4:195–208
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 197
[5, 6]. The method depends upon the knowledge of the idealized as a point loads. Positive spikes correspond to the
transverse position of the truck on the culvert and cannot time when each axle load enters and leaves the span, and
measure spacing between axles without the help of tape the negative spikes represent the time when each axle load
switches or other similar devices. It is also important to is over the instrumented section, which was at the mid-span
note that the strain transducers used in the CULWAY for the bridge dealt with by Wall et al. [9]. Unlike the case
system were subject to large drifts, thus requiring frequent of the idealized beam, the spikes in the second derivative of
calibration tests. strains in an actual bridge are not so well defined, because
Znidaric et al. [3] developed a free-of-axle detector (FAD) of which the calculation of speed by the Wall et al. method
WIM method by observing the peak responses of multiple is subject to considerable numerical errors. For their cali-
strain gauges to estimate the number of axles. The speed of the bration tests, they had used a test truck which was installed
truck is estimated by calculating the time delay in the peak with an accurate speed-measuring device. The differences
responses of several sensors at different locations along the between the speeds measured with device and by the sec-
span. The axle spacing is estimated by calculating the time ond derivative method ranged between ?7.8 and -13.5 %.
delay in the peak responses at the same sensor. The method, Since the calculated speed of a vehicle has a direct influ-
however, is only applicable to short span concrete bridges. ence on the accuracy of prediction of the GVW, it is not
A novel BWIM system was introduced in Japan by surprising that the GVWs of the test truck predicted by this
Yamada and Ojio [7], in which the problem of the strain– method are subject to fairly large errors.
time curves with ill-defined peaks was overcome by a The methods mentioned earlier are based on the static
method that the authors call a reaction force method; in this response of the bridge. Several methods were developed
method, the stiffeners in a support cross-frame at one end of using the dynamic response of the bridge, e.g., Leming and
a simply supported girder bridge are instrumented to mea- Stalford [10], Deesomsk and Pinkeaw [11], and Law et al.
sure vertical strains, corresponding to the reactions at the [12]. However these methods are fairly complicated, not
support. The influence lines for support reactions do indeed given to easy implementation. In addition, some of these
have distinct peaks corresponding to each axle load. The methods were developed using simulated data which often
authors calibrated their BWIM system with the help of three involved simplified problems such as a two-axle vehicle.
pre-weighed 3-axle trucks; from the data that they have Rowely et al. [13] implemented a BWIM method based on
provided it was found that their BWIM system could predict the dynamic response at the Vransko Bridge in Slovenia.
axle weights and GVW with accuracies of ±75 and ±15 %, The authors used an algorithm developed by Gonzeales
respectively. It is noted that the accuracy of prediction of et al. [14] based on the moving load identification theory.
this BWIM system might have been poor, because only one The algorithm, however, still needed extra axle detecting
vertical member of the cross-frame was instrumented. equipment to be installed and the finite element modeling
To test the efficiency of the reaction force method of the instrumented bridge.
developed by Yamada and Ojio [7], Bakht et al. [8]
instrumented a support diaphragm of a slab-on-girder
bridge in Manitoba. The instruments consisted of strain 3 Proposed methods
gauges placed on the stiffeners of the diaphragm. The
strain–time plots for the diaphragms had distinct peaks and The purpose of this paper is to provide an accurate free-of-
permitted the calculation of axle loads of moving trucks; axle detector BWIM method. Three methods are investi-
however, the strains were found to be highly sensitive to gated and implemented on a bridge in Winnipeg, MB,
the transverse position of the vehicle on the bridge and, Canada. Two of the methods were developed by the
thus, not very accurate in calculating the axle weights for authors, while the third is an extension of an existing
trucks in random transverse positions. method. The theoretical bases of the three methods are
A BWIM method was recently developed in Connecti- provided in the following.
cut, USA, to predict GVWs, axle weights and axle spacing
of trucks through strains measured near the mid-span of a 3.1 Asymmetry coefficient method
simply supported steel girder bridge with a concrete deck
slab [9]. The unique aspect of this method is that it Similar to the Ontario method [4], this method also uses the
determines the speed of a vehicle from only girder strains, UDL as being representative of all the axle loads of a truck,
i.e., without resorting to ‘intrusive’ devices such as tape which are usually regarded as point loads in theoretical
switches. For calculating vehicle speeds, the authors used calculations. It is assumed that the total load of a truck
the second derivative of recorded strain with respect to GVW is distributed uniformly over a length nL, where L is
time. For an idealized beam, these second derivatives are the span of the bridge and n is a fraction, so that the
pure spikes, each corresponding to an axle load, which is intensity of the UDL is GVW/nL per unit length. The
123
Author's personal copy
198 J Civil Struct Health Monit (2014) 4:195–208
method is based on the assumption that the BWIM system The asymmetry coefficient a is the ratio of the distance
is applied to a right simply supported slab-on-girder bridge of the front end of the UDL inducing Mmax from the left-
having a span length greater than the length of most trucks, hand support and the length of the moment–distance plot,
which is assumed to be a maximum of about 20 m. All so that:
girders are assumed to be parallel to each other.
Consider a simply supported beam with span L, with an x2 x1 n þ ð1 nÞa
a¼ ¼ ð6Þ
instrumented section at a distance aL from the left-hand x3 x1 ð 1 þ nÞ
support, where a is a fraction of the span length (Fig. 2). As The bridge response collected for BWIM systems is
shown in Fig. 2, the influence line for bending moments at usually in terms of strains plotted over time, which can be
the instrumented section is triangular, and the area under converted into the distance scale only if the speed of the
the influence line is a(1 - a)L2/2. From the records of the vehicle is known. In the absence of the knowledge of the
instrumented section, the objective is to determine n, from vehicle speed, the asymmetry coefficient a can be calcu-
which GVW can be determined as explained later. lated from the time records as follows:
Assuming that the UDL travels from left to right, the t2 t1
complete record for a given truck has the length (1 ? n)L, a¼ ð7Þ
t3 t1
as it covers the leading edge of the UDL coming to the left-
hand support and leaving the beam when the trailing edge where t1 is the time when the leading edge of the load just
of the load just leaves the right-hand support. comes on the beam, t2 is the time at which Mmax occurs,
As shown in Fig. 3, the maximum moment at the and t3 is the time when the trailing edge of the load just
instrumented section, Mmax, occurs when the leading and leaves the bridge.
trailing edges of the UDL have the same height ordinate on Real data contains noise and oscillations caused by
the influence line, so that a fraction a of the UDL is to the vibrations and it is difficult to establish the start and end
left of the instrumented section and a fraction (1 - a) to time of a truck event accurately. Instead, the start and end
the right. Mmax for the UDL is given by the following time are determined by a threshold value that is a fraction
equation: of the maximum moment.
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 199
123
Author's personal copy
200 J Civil Struct Health Monit (2014) 4:195–208
the time interval between the two peak responses is aLðL aLÞ L aLðL aLÞ
denoted as t. Since (x2 - x1) = Vt, Eq. (16) can be written A ¼ ðP1 þ P2 þ P3 Þ ¼ GVW
L 2 2
as: ð19Þ
Vt ¼ ð1 nÞrL ð17Þ
where GVW is the sum of all three loads.
Denoting t/T as s and dividing Eq. (12) by Eq. (17), the The total area A under a response diagram can be easily
following equation is obtained: obtained by numerical integration. The advantage of using
rs numerical integration is that A becomes relatively insen-
n¼ ð18Þ sitive to the smoothing process. By using Eq. (19), GVW is
rþs
given by the following equation:
After determining n from Eq. (18), GVW can be found
2Amoment
using Eq. (11). GVW ¼ ð20Þ
aLðL aLÞ
3.3 Beta method The relation between the moment and strain below the
proportional limit is linear, so that:
This method was proposed by Ojio and Yamada [15], the basis
M
of which could be explained with the help of Fig. 6 which e¼ y ¼ CM ð21Þ
IE
shows bending moments at a reference point in a simply
supported beam due to a set of three moving point loads. If the strain area is used instead of the moment area, Eq.
It is well known that the influence line for each of the (20) could be rewritten as:
point loads has a triangular shape with maximum moment 2Astrain
being PaL(L - aL)/L, where P is the magnitude of the GVW ¼ C ð22Þ
aLðL aLÞ
load, aL is the distance of the reference section from the
left support, and L is the span of the beam. Although the For practical application, the strain readings are recor-
moments due to P2 and P3 are offset from the left support ded in the time scale. To convert strain into the distance
by x1 and (x1 ? x2), respectively, the length of the bending scale, the strain is multiplied by velocity, v, so that Eq. (22)
moment diagram remains L for each load. It can be shown could be rewritten as:
that the total area, A, of the bending moment diagram under 2Astrain
GVW ¼ Cv ¼ bA ð23Þ
the three moving loads is given by: aLðL aLÞ
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 201
Although referred to as a ‘constant’, b is really a vari- resistance strain gauges on the underside of the top flanges,
able that depends upon: (a) the speed of the vehicle, (b) the as well as strain gauges on the external steel straps,
direction of travel, (c) the location of the instrumented restraining the deck slab. Strain gauges at the girders are
transverse section, and (d) sometimes on the transverse identified by the letter ‘‘G’’, while strain gauges installed at
position of the vehicle. the straps are identified by the letter ‘‘S’’.
The steel straps of the externally restrained deck slab of
the bridge were instrumented at Sections AA and CC.
Figure 9 shows the locations of the electrical resistance
4 Calibration tests strain gauges on the straps and girders at Section AA.
The instrumented sections were calibrated for BWIM
To apply the proposed BWIM system, a bridge in Manitoba under two trucks, one of which had a single steering axle
was instrumented and tested for calibration using two and two 2-axle groups; this truck, designated as truck A, is
vehicles of known configurations. The following sections shown in Fig. 10.
describe the bridge and the tests. The other truck, designated as truck B, had a single
steering axle, followed by two 2-axle and 3-axle groups.
4.1 Details of bridge and calibration trucks Details of truck B are shown in Fig. 11a and b. Figures 10
and 11 also show GVW and Bm values for the two test
The bridge used in this study was the South Perimeter trucks. As noted in previous sections, Bm is the length of the
Bridge in Winnipeg, Canada, having seven spans, of which equivalent uniformly distributed load that induces nearly
four are continuous and the remaining three simply sup- the same maximum moments and shears in a beam as the set
ported. Figure 7, which identifies the span numbers, shows of idealized point loads. In the context of the asymmetry
that the simply supported span 2 was instrumented at two coefficient and the two-station methods, Bm = nL.
transverse sections, DD and CC, in addition to the trans- The calibration tests were conducted under two sets of
verse section AA over the third pier from the western end. loads. In the first set, truck A was made to travel at a
The bridge has both west-bound and east-bound traffic. crawling speed in 14 different transverse positions, which
Sections CC and DD are located at a distance of 18.8 and are identified in Fig. 12 as crawling speed load cases
5.6 m from the west support of span 2, respectively. (CSLC). The purpose of this series of tests was to deter-
As shown in Fig. 8, the bridge comprises an externally mine the value of the constant C that relates the sum of
restrained concrete deck slab designed according to the theoretical peak girder moments M to the corresponding
Canadian Highway Bridge Design Code (2006) [16] on sum of observed girder strains, e, according to Eq. (21).
eight steel girders. At Section DD, which is closer to the For the second series of tests, both trucks A and B were
left-hand support of the bridge, electrical resistance strain made to travel at different speeds in the middle of the
gauges measuring longitudinal strains were installed on the normal and passing lanes. The purpose of the second series
upper side of the bottom flanges of all girders (Fig. 8b). At of tests was to determine the accuracy of the various
Section CC (Fig. 8a), the girders also included electrical BWIM methods described earlier. The load case numbers,
123
Author's personal copy
202 J Civil Struct Health Monit (2014) 4:195–208
speeds of the trucks and other information are provided in The percentage difference from the average in the cal-
Table 1. culated values of C is plotted in Fig. 13 against the trans-
verse positions of the center of truck A. It can be seen in
4.2 Calculation of bridge constant C Fig. 13 that the magnitude of C is little affected by the
transverse position of the vehicle. The average value of
By using the observed data, smoothed using a low pass C for Section CC was found to be 11.307 9 1012N.mm for
filter, under crawling speed load cases 1 through 14 the east-bound direction and 12.077 9 1012N.mm for the
(Fig. 12) and Eq. (21), values of constant C were calculated west-bound direction, with a maximum variation of 1.5 %,
for both Sections CC and DD for truck A traveling in each and the average value of C for Section DD was found to be
of east-bound and west-bound directions. 9.836 9 1012N.mm for the east-bound direction and
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 203
(a)
(b)
Fig. 12 Transverse positions of truck A, moving at crawling speed for calibration tests
123
Author's personal copy
204 J Civil Struct Health Monit (2014) 4:195–208
Fig. 14 Strains in straps due to trucks A and B during high-speed load case 2
strains calculated from strain rosettes, or even one inclined 5.1 The asymmetry method
strain gauge in the web could produce the same desired
effect of distinct sharp peaks that appears with the passing Table 3 presents the summary of the percentage of
of individual axle or axle groups, and the same procedure deviation from the scale weighed GVW for trucks A and
could be used. The speeds were also calculated from the B, using the asymmetry method while taking the start and
two-station method using Eq. (12). end times as 2 % of the maximum sum of strains for all
Table 2 compares the velocity calculated by the strap girders.
and two-station methods with those reported by the truck The results in Table 3 show that that the asymmetry
driver. It is noted that in the absence of an accurate mea- method produces inconsistent results and in many cases the
suring device, the reported speeds can only be used as deviation from the measured GVW is quite large. In the
guides. The table shows that the variation in the velocities case of test 1 for truck B at Section CC, the value of a was
calculated by the strap method for trucks A and B is calculated as 0.59, which produced a value of n larger than
smaller than that of the two-station method. 1, and therefore a negative value for the GVW which is
meaningless.
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 205
Truck A % Deviation CC 18.89 78.46 71.55 119.61 -5.57 -9.65 -7.64 -15.78
% Deviation DD -9.84 -0.61 -3.90 -11.24 13.28 13.50 12.83 25.50
% Deviation Av. 4.52 38.93 33.83 54.19 3.86 1.93 2.59 4.86
Truck B % Deviation CC N/A -11.08 28.01 74.33 9.42 -32.87 -31.18 -30.95
% Deviation DD -6.21 -17.24 -0.30 24.61 24.31 34.74 24.83 29.44
% Deviation Av. N/A -14.16 13.86 49.47 16.86 0.93 -3.17 -0.75
Table 4 GVW results using the Travel direction East bound West bound
two-station method
Test no. Test 1 Test 2 Test 3 Test 4 Test Test 6 Test 7 Test 8
5
The values of percentage deviations for Sections CC and 5.3 The beta method
DD were calculated using n, based on the values of T obtained
from the data at Sections CC and DD, respectively. Figures 15 and 16 present the calibration of beta for the
The percentage deviation in calculating the GVW for east-bound and west-bound directions, respectively. It
truck A from the measured value, using the averages from could be shown that the speeds from straps give smaller
Sections CC and DD, was a maximum of about 4.5 %. scatter for beta than the speeds calculated from the two-
Deviations calculated using data from Sections CC and DD station method, thus providing more accurate calibration
were below 5 % in all but three cases, with once case for the parameter beta. This could be demonstrated in
barely exceeding 5 % and the other two with a maximum Table 5, which presents the percentage variation of GVW
deviation of 8.6 %. from the measured value, calculated using the beta
On the other hand for truck B, the average deviation parameters obtained from Figs. 15 and 16.
ranged from 9.8 to 21.5 %, with individual deviations The results in Table 5 show that the percentage devia-
consistently above or slightly below 10 %. tion from the measured value in calculating the GVW for
123
Author's personal copy
206 J Civil Struct Health Monit (2014) 4:195–208
truck A ranged from 0.23 to 8.9 % when using speeds measured values ranged from 0.6 to 4.4 % when using the
calculated with the two-station method, while the per- speeds calculated from the straps.
centage deviation from the measured values ranged from
0.36 to 4.4 % when using the speeds calculated from the
straps. 6 GVW estimate for permit trucks
The results also show that for truck B the percentage
deviation from the measured value ranged from 0.19 to A structural health monitoring (SHM) system has been installed
18.5 % when using the speeds calculated using the two- on the South Perimeter Bridge and the data are continuously
station method, while the percentage deviation from the collected, to be used for identifying information about truck
123
Author's personal copy
J Civil Struct Health Monit (2014) 4:195–208 207
Truck A Speed (two station) -6.87 4.65 0.23 -1.16 2.70 -7.71 -8.91 -8.89
Speed (straps) -0.97 0.36 0.58 -1.71 -1.27 -3.55 -4.44 -1.92
Truck B Speed (two station) 8.29 -15.03 0.12 5.70 1.19 -13.00 -18.26 -18.45
Speed (straps) 1.24 0.72 1.12 -0.60 -2.61 -4.38 -3.98 -1.89
Table 6 674 kN permit truck results summary • The asymmetry method gives large errors and is not
Calculation GVW GVW Average % reliable.
method calculated calculated GVW variation • The two-station method provides good prediction for
from from (kN) of average the GVW when the length of the truck is less than half
Section Section from
CC (kN) DD (kN) measured
the length of the bridge. Further investigation is needed
to estimate accurate limitations of the method.
Two-station 384 425 405 -40.0 • The beta method provides predictions with errors\5 %
method of the GVW, provided that accurate estimates of
Beta 382 382 382 -43.4 vehicle speeds could be obtained.
method(v from
2-station • Responses of the straps of the externally confined deck
method) slab provide reliable estimates of the truck speeds.
Beta method 723 723 723 7.2
(v from straps)
Acknowledgments The authors wish to acknowledge the contri-
bution of the late Dr. Leslie Jaeger in formulating the two-station
weights. To test the BWIM methods against trucks under actual method.
traffic conditions, Manitoba Infrastructures and Transportation
(MIT) was asked for data about the permit trucks; the required
data included truck weights, speeds and the time of crossing the References
bridge. Two heavy trucks were identified with precise weight
1. Moses F (1979) Weigh-in-motion system using instrumented
measurement provided for only one truck, which had a steering
bridges. ASCE Trans Eng J 106:233–249
axle followed by two 2-axle, one 3-axle and one 2-axle groups 2. Quilligan M, Karoumi R, O’Brien EJ (2002) Development and
and a single axle. The total GVW was 674 kN and the speed testing of a 2-dimensional multi-vehicle Bridge-WIM algorithm.
provided by the driver was 40 km/h. Using the provided 3rd int.conf. on weigh-in-motion (ICWIM3)
3. Znidaric A, Kalin J, Lavric I (2002) Bridge weigh-in-motion
information, the strain data for the time period were extracted
measurements on short slab bridges without axle detectors. 3rd
and analyzed using the two-station and beta methods. The speed int.conf. on weigh-in-motion (ICWIM3)
calculated using the two-station method was 30 km/h, and the 4. Harman DJ (1984) A study of commercial vehicles and their
speed calculated using the strap data was 48.6 km/h. Table 6 multiple presence. OJTCRP Project No. 23141, U. of W. Ontario,
Faculty of Eng. Science, London, ON, Canada
provides a summary of the analysis results.
5. Scott, G. (1987) Weighing-in-motion technology—status CULWAY
The permit truck results show the same pattern as truck in Australia. Australian Research Board Seminar on Road Traffic
B. The beta method using speeds from straps provided data collection using weighing-in-motion, Melbourne, Australia
relatively accurate results, while the two-station method 6. Heywood RJ (1991) Bridge applications of WIM data. Proc, of
the AUSTROADS Bridges Conf., In: ed. Heywood RJ, (ed.),
and the beta method calculated using speeds from the two-
Austroads, Haymarket, Australia, p 139–150
station method provided large errors. It should be noted 7. Yamada K, Ojio T (2003) Bridge weigh-in-motion system using
that the errors were significantly larger than in the case of reaction force method. Proc. of the Int. Workshop on Structural
truck B. Although axle spacing was not provided consid- Health Monitoring of Bridges/Colloquium on Bridge Vibra-
tion’03. Japan Society of Civil Engineers, pp 269–276
ering the configuration, it could be concluded that the truck
8. Bakht B, Mufti AA, Tadros G, Eden R, Mourant G (2006)
was very long and that the accuracy of the two-station Weighing-in-motion of truck axle weights by Japanese reaction
method was reduced with the increase in the truck length. force method. 3rd Int. Conf. on Bridge Maintenance, Safety and
Management, Porto, Portugal, July 16–19, pp 981–982
7 Conclusions 9. Wall CJ, Christenson RE, McDonnell A, Jamalipour A (2009) A
non-intrusive bridge weigh-in-motion system for a single span
steel girder bridge using only strain measurements. Report no.
Based on the analysis of the test data and the results of CT-2251-03-09-5. Connecticut Dep. of Transportation
calculating the GVW using the three proposed methods, the 10. Leming Sarah K, Stalford Harold L (2003) Bridge weigh-in-
following conclusions could be made: motion system development using superposition of dynamic
123
Author's personal copy
208 J Civil Struct Health Monit (2014) 4:195–208
truck/static bridge interaction. American control conf., proc. of 14. González A, Rowley C, OBrien EJ (2008) A general solution to
the 2003IEEE. p 815 the identification of moving vehicle forces on a bridge. Int J
11. Deesomsuk T, Pinkaew T (2010) Evaluation of effectiveness of Numer Methods Eng 75(3):335–354
vehicle weight estimations using bridge weigh-in-motion. IES J 15. Ojio T, Yamada K (2002) Bridge weigh-in-motion systems using
Part A Civil Struct Eng 3(2):96–110 stringers of plate girder bridges. 3rd int.conf. on weigh-in-motion
12. Law S, Bu J, Zhu X, Chan S (2004) Vehicle axle loads identi- (ICWIM3)
fication using finite element method. Eng Struct 26(8):1143–1153 16. Canadian Standards Association (2006)CAN/CSA-S6-06, Cana-
13. Rowley CW, OBrien EJ, Gonzalez A, Žnidarič A (2009) dian highway bridge design code
Experimental testing of a moving force identification bridge
weigh-in-motion algorithm. Exp Mech 49(5):743–746
123