ThC3

2:00pm-2:15pm

Interferometric inclinometer for structural monitoring

Daniele Inaudi, Branko Glisic

SMARTEC SA
Via a1 Molino 6,CH-6916 GRANCIA, Switzerland
Tel: +41 91 993 09 24, Fax: +41 91 993 09 40
email: ,Web: www.smartec.ch

ABSTRACT

The introduction of long-gauge fiber optic sensors has opened new opportunities for the monitoring of large structures.
Using these sensors it is possible to observe the local elongation of a structure and calculate its global deformations.
The use of intemal sensors detecting integrated strain is however unable to detect the rigid-body motion of a structure,
relative to an external reference system. This would require the installation of long deformation sensors connecting the
structure with extemal reference points. Another possibility is to use gravity as a reference to measure vertical rotations
of the structure. A gravity-referenced rotation sensor is conventionally called an inclinometer and measures the angle
variations between a nominally vertical reference line in the structure and the direction of gravity. It is possible to use
uni-axial inclinometers or bi-axial ones, measuring rotations in the two perpendicular vertical pianes. Today’s
inclinometers are based on a variety of techniques and can achieve sensitivities in the micro-radian range. When using
fiber optic deformations sensors in a given structure, it would however be interesting to use the same technology to
obtain inclination reading [I]. This prompted us to develop a fiber optic-based inclinometer compatible with the SOFO
system [2].
This contribution describes the development of this novel sensor and the first measurements obtained with it.

1. MEASUREMENT PRINCIPLE
The proposed design is based on the use of a pendulum as schematically represented in Figure 1. The hinge of the
pendulum is connected to structure to be monitor, while its end is free to swing according to the structure’s rotations.
Two mirrors are installed at the free end of the pendulum. Two optical fibers with a collimating optical system are
facing the mirrors and are connected to the structure. These two fibers are connected to a coupler and constitute
therefore the two arms of a Michelson interferometer. When the structure rotates, the pendulum will remain vertical,
while the two fibers will move left or right while conserving their relative distance unchanged. This push-pull
configuration therefore allows the transformation of a rotation in the vertical plane into a correlated change in the path
unbalance between the arms of the Michelson interferometer. This path unbalance is than measured with a standard
SOFO reading unit.

To reading unit

I
-
Collimator *
Figure 1, Principle of the SOFO inclinometer

The SOFO reading unit is based on low-coherence interferometry in a pair of Michelson interferometers PI.The
broadband light coming from the SLED source @1330nm in the SOFO reading unit is split by the coupler contained in
the sensor. The two paths go to the two fiber ends where they are collimated to the mirrors installed on the pendulum
and reflected by the mirrors. The light is than recoupled in the fibers, recombined by the coupler and sent back to the
reading unit. The reading unit uses a second and adjustable Michelson interferometer to match the path difference
IEEE
0-7803-7289-1/02/$17.0OO2002 39 1
introduced by the sensor’s interferometer with an equal and precisely known path unbalance. Only in this matching
condition, interference fringes will be detected on the photodiode. This measurement techniques has proven especially
adapted for long-term monitoring, allowing absolute measurements with a resolution in the micrometer range and
excellent stability over long periods of time. One measurement requires less than ten seconds.

2. SENSOR DESIGN
The basic design presents a number of difficulties that have to be addressed in the realization phase. The hinge is a
critical point, since it has to guarantee a complete freedom of the pendulum movements and should not be affected by
friction effects. This is usually done by using blade hinges that have the drawback of leaving the pendulum unconnected
form the casing structure and therefore vulnerable to shocks (requiring careful handling and transportation). The blade
design also minimizes the damping of the pendulum oscillation making measurements impossible in the presence of
vibrations. Another challenge is constituted by the path of the two optical fibers that are mounted opposed to the
pendulum. This d e s the optical system sensitive to temperature changes that could affect the two fibers differently
and complicates the casing design. Another challenge is constitute by the rotation of the mirrors that affects the back-
coupling efficiency of the light into the fibers. Finally, a relatively long pendulum arm is necessary to achieve good
rotation sensitivity.
All these challenges have been addressed and solved using the monolithic aluminum structure presented inFigure 2.

Weight

Mirror

Neck-down

Collimator

Coupler

Fixation Holes

Reading Unit
Figure 2, Monolithic pendulum structure

The whole sensor is obtained by machining a single aluminum block, increasing its solidity and minimizing all
temperature influences. In this design, the pendulum is replaced by a double arm that forms a parallelogram with the
weight. This eliminates the mirror rotations and allows all optical elements to be placed on the inside of the moving
elements. The hinge is replaced by four “neck-downs” or thin laminas that introduce a well-controlled mobility in the
system. Finally the system is mounted upsidedown, allowing an amplification of the movements of the mirrors. To
further damp the oscillations, magnets are used and produce Foucault currents that allow quick absorption of the
oscillation energy.
It is interesting to notice that that design is completely symmetric and that the same martial is used in the whole
construction. This makes the system particularly insensitive to temperature variations that will affect both arms of the
interferometer equally, thus balancing themselves out at the central position (for more detail see the section on
temperature sensitivity). The other effect of temperature is linked to the temperature dependence of the young’s
modulus of the aluminum laminas. This remaining second-order effect can be neglected for practical applications.
The monolithic structures doubles as casing for the inclinometer, allowing a very compact sensor. The full dimensions
are of 110 mm x 130 mm x 21.5 mm for a total weight of 600 g. The inclinometer is shown inFigure 3.

392
 3. TESTING
The following tests were carried out in order to certify the sensor performances:

Sensitivity, range and reproducibility

The prototype was mounted on a rotation table allowing a controlled vertical rotation. Since the readings of the rotation
table would not have been sufficiently precise, a reference electronic inclinometer (Wyler Zerotmnic) was mounted on
the same support and used as reference.
The results are shown in Figure 4, Response of the SOFO Inclinometer.
95 >

E
E
'-
m
g IS..

3
:
I 8..
E
-2
c 75.

B
3

0 02 04 06 06 1 12 14 16 18 2

Rotation r)
Figure 4, Response of the SOFO Inclinometer

The results show a very linear response with a resolution of 20 p a d and a precision of 44 p a d over a measurement
range of 2", corresponding to a maximal error of 0.042% full scale over the full 5" measurement range. This compares
with the performances of the best electronic inclinometers. The response was of 1.12 mm/O. The tests showed no sign of
hysteresis and the reproducibility was within the precision figure.
By rotating the inclinometer so that the weight and pendulum are facing downwards, it is possible to increase the range
of measurement to 17" at the cost of reducing the resolution to 70 p a d .

Temperature sensitivity
There are two main possible sources of temperature sensitivity: thermal expansion of the flexure structures and
temperature dependence of the elastic modulus of the flexure structure. Since the structure is symmetric and the sensors
works in pus-pull, the first effect only appears when the inclinometer is not at the central position, otherwise the thermal
expansion in one arm is balanced by the one in the other arm. This effect is maximal when the inclination is maximal, in
either direction. The absolute value of the error is of the order of 700 p a d for a temperature variation of 100°C, a
negligible value that could be easily corrected if even a crude temperature value is available.
The effect of a temperature-dependant elastic module of aluminum is of changing the recall forces that define the
position of the inversed pendulum. No reliable data on that effect was found in the literature, but we suspect that the
393
effect is also small. An experimental investigation of this hypothesis is ongoing, but made difficult by the lack of a
reliable, accurate and temperature -insensitive reference system.

Stability
Testing the stability of the system proved difficult because of the high sensitivity of the system. Figure 5 shows
measurements performed during 34 hours with the sensors installed rigidly on an industrial building during the
weekend. A trend in the measurements shows a small rotation of the building during this period, but the 24-hour
periodicity indicates that the system did not suffer form a drift at least during this short period of time. Additional long-
term tests will be required to validate the long-term stability of the system. However the used aluminum type is not
expected to show any creep that would induce a drift in the measurements. The SOFO reading unit has already been
shown to be free form drift even after years of operation.
- _.
*
15 382 . -
Y I m Y - m..
15 380‘
a*
-
.
)

1 5 3 7 8 . ~

E
I .).MI I I1 0

1 15374.
- 0 0-

E I ni-aai. in.4 i1e.1~4

- 15372‘
I

9 - ~ o a e n Y

2 15370.

15368-

4. CONCLUSIONS
The presented sensor extends the use of the SOFO system, which is mainly applied to the measurement of elongations
in structures. Using a pendulum parallelogram machined from a single aluminum block and measuring its deformation
with a push-pull interferometric configuration, it is possible to obtain an inclinometer compatible with the SOFO
reading unit. The proposed inclinometer has a resolution of 20 p a d and a precision of 0.0042% full scale. The total
measurement range i of 5’ and the temperature cross-sensitivity is expected to be very low. These values compare
favorably to the high-end of electronic inclinometers and are well suited for the monitoring of civil structures such as
bridges, towers and anchored walls. The small size of the sensors and the hermetic design make them suitable for open-
air use, while the low temperature sensitivity makes them ideal for measuring temperature-induced rotations.

ACKNOWLEDGEMENTS

The Authors are indebted to Axel Bertholds (Sensoptic SA, Losone Switzerland) and Stefan0 Bottinelli Mecartex SA,
Losone) for designing monolithic structure and to Marco Cerulli for carrying out part of the measurements. We would
also like to acknowledge the support of IMM (Grancia, Switzerland) for the use of the climate chamber.

REFERENCES

[I] P. Ferdinand, S. Rougeault, “Optical Fiber Bragg Grating Inclinometry of SMART Civil Engineering and Public
Works”, OFS2000, Venice, SPIE Vol. 4185, pp.13-16
[2] D. Inaudi “Field testing and application of fiber optic displacement sensors in civil structures”, 12th International
conference on OFS ‘97- Optical Fiber Sensors, Williamsburg, invited paper, OSA‘(1997), Vol 16, p 596-599.
[3] D. Inaudi, A. Elamari, L. Pflug, N. Gisin, J. Breguet, S. Vurpillot, “Lowcoherence deformation sensors for the
moqitoring of civilengineering structures“ Sensor and Actuators A‘ (1994), Vol. 44, p 125-130.

394