Você está na página 1de 2

Controlling the Structural Behaviour of Concrete Structures

Using Model Updating and Low-Cost Monitoring


Tiago Coelho
R&D Grand Holder, MEng1

Joo P. Santos
PHD Student1

Paulo Silveira
Full Research Officer1

The present work addresses the Structural Health Monitoring of large concrete
structures, more specifically of large pre-stressed concrete bridges with high social and
economic importance.
Long concrete structures are prone to damage and malfunctions due to the effects of
concrete rheological effects, namely creep and shrinkage. This fact has motivated
multiple research and practical SHM works to address their study and experimental
validation using embedded strain gages, whether deployed within the concrete medium
or fixed to the steel reinforcement bars. This strategy has been long used and proved
effective for studying and controlling the overall extensions observed throughout
structural systems. However, a proper characterization of these effects requires that
multiple cross-sections be instrumented with a large number of these sensors. This
requirement renders this type of civil structural monitoring systems expensive, not only
due to the large amount of sensors, but also due to the need of deploying acquisition
units with a high number of input and signal conditioning boards.
Strain Gages + Thermometers

Displacement Sensor

100

200

300

400

500

600

Time Index (Days)

100
-50

50

NL2
NL3
NL4

-100

(b)

Displacements (mm)

-50

50

C10m
C2m
C4m

-100

Strain (um/m)

100

(a)

(c)
0

100

200

300

400

500

600

Time Index (Days)

Fig 1 (a) Salgueiro Maia Bridge side view; (b) Average strain measurements; (c) Displacement
measurements.

The present work aims at proposing an alternative and inexpensive solution for
controlling these effects on concrete structures. It relies on few inexpensive
measurements of global structural displacements (in opposition to localized strain
measurement) and in model updating. While strain measurement allows for a direct

Structures Department, LNEC, Lisbon, Portugal

2.5
2.0
1.0

1.5

R=0.9986

0.5

f(Hz) Experimental

3.0

characterization of strain across an entire structural system; the displacements do not,


and are included in an automated model updating to infer the most compatible strain
loadings being measured on a target structure.
The case study addressed herein is the 250m span cable-stayed Salgueiro Maia
Bridge, located in Santarm, Portugal (Fig.1a), which crosses the Tagus river 70km
upstream of Lisbon. The structural system comprises a box-girder deck 2.62m high and
27.7m wide, suspended across a length of 489m by seventy two stay cables sheathed
in steel tubes.
The work described in this paper resorts to strain and temperature measurements
obtained from three cross-sections, one from each pylon and one from the mid central
span (Fig.1a); and to three hydrostatic pressure cells (located in the quarter and mid
central span, Fig.1a) which measure the vertical displacements in comparison to a
reference cell located over one pier. These measurements were conducted during a
period of approximately two years and are shown in Fig.1a,b.

0.5

1.0

1.5

2.0

2.5

3.0

f(Hz) Numerical

Fig 2 Numerical model of the Salgueiro Maia Bridge: correlation between numerical and experimentally
obtained modal frequencies.

The model updating task was conducted using a tri-dimensional numerical model
(Fig.2) of the Salgueiro Maia Bridge which was previously updated using the latest
results from a modal experimental campaign conducted on that structure. Using this
model, combinations of strain loading magnitudes were applied to the deck and pylons
within the range of 0 to 60 um/m, thus leading to a total amount of 216000 different
load cases. The entire analysis was conducted using time-history analyses for
computational simplicity. It was observed that all load cases could be run in only a few
minutes, thus discarding the need for optimization techniques.
From the 216000 analyses, the values of displacements obtained on the quarter and
mid central span of the numerical model were obtained and compared to those
obtained on site (Fig.1c). Several correlation / distance measures were used and the
experimental displacement data was supressed of temperature effects using
regression analysis with displacements as output and the measured temperatures as
input. The load cases which generated displacements sets exhibit larger correlations /
smaller distances to those measured on site were chosen as the ones which more
accurately described the structural evolution generated by rheological effects during
the monitoring period. The comparison of these load cases with the strain
measurements conducted on site (Fig.2c), which were also suppressed of temperature
effects using regression analysis, allowed concluding about the effectiveness of the
proposed strategy.

Structures Department, LNEC, Lisbon, Portugal

Você também pode gostar