Use of Comparative Vacuum Monitoring Sensors For Automated, Wireless Health Monitoring of Bridges and Infrastructure

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

This paper describes objective technical results and analysis.

Any subjective views or opinions that might be expressed


in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

Use of Comparative Vacuum Monitoring Sensors for Automated,


SAND2018-5926C
Wireless Health Monitoring of Bridges and Infrastructure

Dennis Roach
Sandia National Laboratories, New Mexico, United States ofAmerica

ABSTRACT: Economic barriers to the replacement of bridges and other civil structures have created an aging
infrastructure and placed greater demands on the deployment ofeffective and rapid health monitoring methods.
To gain access for inspections, structure and sealant must be removed, disassembly processes must be com-
pleted and personnel must be transported to remote locations. Reliable Structural Health Monitoring(SHM)
systems can automatically process data, assess structural condition, and signal the need for specific maintenance
actions. They can reduce the costs associated with the increasing maintenance and surveillance needs of aging
structures. The use of in-situ sensors, coupled with remote interrogation, can be employed to overcome a myr-
iad of inspection impediments stemming from accessibility limitations, complex geometries, the location of
hidden damage, and the isolated location ofthe structure. Furthermore, prevention of unexpected flaw growth
and structural failure could be improved ifon-board SHM systems were used to regularly, or even continuously,
assess structural integrity. A research program was completed to develop and validate Comparative Vacuum
Monitoring(CVM)sensors for crack detection. Sandia National Labs,in conjunction with private industry and
the U.S. Department ofTransportation, completed a series ofCVM validation and certification programs aimed
at establishing the overall viability of these sensors for monitoring bridge structures. Factors that affect SHM
sensitivity include flaw size, shape, orientation and location relative to the sensors, along with operational en-
vironments. Statistical methods using one-sided tolerance intervals were employed to derive Probability of
Flaw Detection(POD)levels for typical application scenarios. Complimentary, multi-year field tests were also
conducted to study the deployment and long-term operation ofCVM sensors on aircraft and bridges. This paper
presents the quantitative crack detection capabilities of the CVM sensor, its performance in actual operating
environments, and the prospects for structural health monitoring applications on a wide array of civil structures.

1 INTRODUCTION mandated, nondestructive inspections. These pro-


cesses are not only time consuming but they provide
Multi-site fatigue damage and hidden cracks in hard- the opportunity to induce damage to the structure.
to-reach locations are among the major flaws encoun- The use of in-situ sensors for monitoring the condi-
tered in today's extensive array of aging structures tion of structures and mechanisms, coupled with re-
and mechanical assemblies. The costs associated mote interrogation, can be employed to overcome in-
with the increasing maintenance and surveillance spection difficulties including the remote location and
needs of aging structures are rising. The application expansive distribution of critical structures such as
of Structural Health Monitoring (SHM) systems us- railway bridges and rail cars. Prevention of unex-
ing distributed sensor networks can reduce these costs pected flaw growth and structural failure could be im-
by facilitating rapid and global assessments of struc- proved if on-board health monitoring systems are
tural integrity. These systems also allow for condi- used to more regularly assess structural integrity [2,
tion-based maintenance practices to be substituted for 3]. The ease of monitoring an entire network of dis-
the current time- or cycle-based maintenance ap- tributed sensors means that structural health assess-
proaches thus optimizing maintenance labor [1]. ments can occur more often, allowing operators to be
Other advantages of on-board, distributed sensor sys- even more vigilant with respect to flaw onset.
tems are that they can eliminate costly, and poten-
tially damaging, disassembly, improve sensitivity by
1.1 Comparative Vacuum Monitoring
producing optimum placement of sensors with mini-
mized human factors concerns in deployment and de- Comparative Vacuum Monitoring (CVM) is a
crease maintenance costs by eliminating more time- simple pneumatic sensor technology developed to de-
consuming manual inspections. Current structural tect the onset of cracks. CVM sensors are perma-
maintenance operations require personnel entry into nently installed to monitor critical regions of a struc-
Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned
normally-inaccessible
subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s NationalThe
or hazardous areas to perform ture. CVM
Nuclear sensor
Security is based
Administration under on theDE-NA0003525.
contract principle that
a steady state vacuum, maintained within a small vol- external impact, off-design conditions or malevolent
ume, is sensitive to any leakage [4]. A crack in the attacks.
material beneath the sensor will allow leakage result-
ing in detection via a rise in the monitored pressure. 700

Figure 1 shows top-view and side-view schematics of 0


.1 500
Crack Det71(vacuum unachievable)

the self-adhesive, elastomeric sensors with fine chan- Ia°


•300
nels etched on the adhesive face along with a sensor No Crack(vacuum achieved)
loo
being tested in a lap joint panel. When the sensors
o s lo 15 20 25 30 35 40 a 50 55 60 65 70 75 SO a a
are adhered to the structure under test, the fine chan- Time(s)

nels and the structure itselfform a manifold of galler-


ies alternately at low vacuum and atmospheric pres- Figure 2. Crack detection monitoring with CVM system and
sure. Vacuum monitoring is applied to small galleries pressure response used to indicate the presence of a crack.
that are placed adjacent to the set of galleries main-
tained at atmospheric pressure. If a flaw is not pre- In the matter of bridge refurbishment alone, the Na-
sent, the low vacuum remains stable at the base value. tional Bridge Inventory Database (Fed. Highway Ad-
If a flaw develops, air will flow from the atmospheric min. 2003)indicates that 30% ofthe 600,000 bridges
galleries through the flaw to the vacuum galleries. in the United States are "structurally deficient." In
When a crack develops, it forms a leakage path be- addition, a majority ofthe rail bridges in U.S. are op-
tween the atmospheric and vacuum galleries, produc- erating beyond their initial design life. A bridge that
ing a measurable change in the vacuum level. This is "structurally deficienr is still strong enough and
change is detected by the CVM monitoring system stable enough for use; however, closer scrutiny ofthe
shown in Figure 2. It is important to note that the bridge is required to ensure its continued, safe opera-
sensor detects surface breaking cracks once they in- tion. In 2006, the American Society of Civil Engi-
teract with the vacuum galleries. Since the sensor neers(ASCE)issued a report on the status ofthe U.S.
physics is based on pressure measurements, there is infrastructure. It assessed everything from roads to
no electrical excitation involved. These sensors can hazardous waste systems and gave the country's in-
be attached to a structure in areas where crack growth frastructure an overall grade of"D". Steel superstruc-
is known to occur. On a pre-established engineering ture bridges built during the interstate construction
interval, a reading will be taken from an easily acces- boom of the 1950s and 1960s are reaching or surpas-
sible point on the structure. sing their initial design lifetime. Depending on their
level of maintenance, some bridges are showing visi-
CV M Sensor
ble signs of deterioration.
Minirn ize distanee
horn rivet head to
produce smallest
On September 30,2006, part ofan overpass collapsed
crack detection
in Montreal. On August 1, 2007, an Interstate 35
Fatig ue Creeks
bridge crossing the Mississippi River in Minneapolis
failed. The collapse of the Interstate 35 bridge
Sensor Pad
CVM Sensor Adjacent to
Crack Initiation Site
prompted many questions regarding the health of
similar structures around the world and their associ-
ated maintenance programs. Figure 3 shows three
bridge failures — in Minneapolis, Montreal and Con-
necticut — and one bridge in Delaware with a large
Figure 1. Schematics depicting operation of CVM sensor and fatigue crack that was discovered and repaired prior
polymer sensor mounted on the outer surface of a riveted lap
joint. to any catastrophic failure.

2 APPLICATIONS FOR CRACK DETECTION 3 CVM PERFORMANCE ASSESSMENT USING


USING CVM SENSORS ONE-SIDED TOLERANCE INTERVALS

Recent events have demonstrated the need to address The Federal Aviation Administration's Airworthi-
critical infrastructure surety needs [5]. The applica- ness Assurance Center at Sandia Labs,in conjunction
tions for CVM sensors can include such diverse struc- with industry and airline partners, completed valida-
tures as: buildings, bridges, trains and subway vehi- tion testing on the CVM system in an effort to adopt
cles, mining structures, railroad cars, trucks and other Comparative Vacuum Monitoring as a standard NDI
heavy machinery, pressure vessels, oil recovery practice [5-6]. Fatigue tests were completed on air-
equipment, pipelines, steel transmission towers, craft components to grow cracks in representative
ships, tanks and a wide array of military structures. structure while the vacuum pressures within the vari-
Damage can arise from service loads as well as from ous sensor galleries were simultaneously recorded. A
fatigue crack was propagated until it engaged one of
the vacuum galleries such that crack detection was chosen as 90%), and 2)the degree of confidence de-
achieved and the sensor indicated the presence of a sired (normally chosen as 95%). The data analyzed
crack by its inability to maintain a vacuum. Probabil- here consisted of fatigue cracks that were propagated
ity of flaw detection assessments were coupled with in various metal specimens with the direction of
on-aircraft flight tests to study the performance, de- growth aligned with the CVM mounted sensors. The
ployment, and long-term operation of CVM sensors data captured is that of the flaw length at the time for
on aircraft. The result was a series of flaw detection which the CVM provided sustainable detection. With
curves that can be used to propose CVM sensors for these assumptions, there exists a distribution on the
aircraft crack detection. One set of test specimens flaw lengths at which detection is first made. In this
were wing box fittings from the Boeing 737 which context, the probability of detection for a given flaw
was the chosen CVM application from Delta Air length is just the proportion of the flaws that have a
Line's fleet. Figure 4 shows the details of one test detectable length less than that given length. That is,
series addressing a wing box fitting application along the reliability analysis becomes one of characterizing
with installation of CVM sensors for the flight test the distribution of flaw lengths and the cumulative
program. distribution function is analogous to a Probability of
Detection(POD)curve. Assuming that the distribu-
tion offlaws is such that the logarithm ofthe lengths
has a Gaussian distribution, it is possible to calculate
a one-sided tolerance bound for various percentile
flaw sizes. To do this, it is necessary to find factors
Kn,7,a to determine the probability y such that at least
a proportion (1-a)ofthe distribution will be less than
X— Kn,y,,, where X and S are estimators of the mean
-1r-tt and the standard deviation computed from a random
ill .11 sample of size n. The data captured is the crack
length at CVM detection. From the reliability analy-
sis, a cumulative distribution function is produced to
Figure 3. Applications for in-situ crack detection using CVM provide the maximum likelihood estimation (POD).
sensors. This stems from the one-sided tolerance bound for the
flaw of interest using the equation:
T POD(90,95)= X (K n,y,a)(S) (1)
where T = tolerance interval for crack length corre-
sponding to 90% POD with a 95% confidence; X =
mean of detection lengths; K = probability factor (—
sample size and confidence level desired); S = stand-
ard deviation of detection lengths; n = sample size; 1-
a = detection level; and y = confidence level.
Because of physical, time or cost constraints, it is
often impractical to inspect an entire population. Due
to the limited number of data points, the reliability
calculations induce a penalty by increasing the mag-
nitude of the K (probability) factor. As the number
ofdata points increases,the K value will decrease and
the POD numbers could also decrease. The formula
in equation (1) is set-up to produce the upper bound
Figure 4. CVM testing on wing box fitting with close-up show- for the tolerance interval which represents the actual
ing fatigue crack crossing into CVM sensor and installation of
CVM sensors on Delta Air Lines aircraft for flight tests.
POD value. With the same parameters described
above, the maximum likelihood estimate describing
SHM reliability calculations will depend greatly on the optimal performance on the Probability of Detec-
the complexity of the structure and geometry of the tion for the OSTI approach can be calculated as:
flaw profile. Since it is based on a sample of the en- POD = EXP
(-Ori(x)-X)2
)
(2)
tire population (n data points), the confidence is less xs-v 2n 2.52
than 100%. Thus, the One-Sided Tolerance Interval
(OSTI)is greatly affected by two proportions: 1) the As an example, the data acquired from CVM fatigue
percent coverage which is the percent of the popula- tests on 2.54"(0.1") thick 2024-T3 aluminum struc-
tion that falls within the specified range (normally ture were used to calculate the 90% POD level for
CVM crack detection. This POD curve, representing
the 95% confidence level, is plotted in Figure 5. The 4 MULTI-CVM SWITCH-BASED SYSTEM FOR
maximum likelihood estimated POD function, repre- REMOTE BRIDGE MONITORING
senting the optimum performance for CVM crack de-
tection, was calculated from equation(2)and is plot- A real-time monitoring system was developed for re-
ted alongside the 95% confidence bound. The overall motely interrogating a distributed array of CVM sen-
POD value (95% confidence level) for CVM crack sors on a transportation bridge structure. It uses a se-
detection in 2.54 mm thick aluminum skin was deter- ries of pressure switches that can continuously
mined to be 0.58 mm (0.023"). In this particular in- monitor structures remotely via a wireless transmit-
stance, it was desired to achieve crack detection be- ting device. Sensors were placed in known fatigue
fore the crack reached 0.1" in length so this goal was critical locations on the bridge structure. When a
achieved. In over 200 fatigue tests conducted using crack breaches a sensor,the pressure switch is opened
CVM sensors there were no false calls produced by and,in turn, triggers a message that is sent to a central
the sensors. maintenance center. Up to 50 switches can be pow-
1 00
ered by one vacuum pump. The CVM monitoring
system, shown in Figure 6, was mounted at a central-
0 90 -
POD Maximum Likelihood Estimate ized point on the structure ofinterest. Sensors can be
0 80 -
made in almost any shape and out ofa material to suit
0 70 -
POD Uncertainty — 95% Confidence Bound the required environment. Multiple sensors can be
0 60 -
arranged to monitor the growth of a crack. It may be
.5
2 0.50 -
O
-1: . ;
-f-.4
;
that there is a known crack and a sensor placed ahead
o. 0.40 - of the crack will be triggered if the crack grows. Of-
0.30 - •• • ••••L ten there are known critical locations at joints or
0.20 -
4 ; welds that require monitoring. The CVM monitoring
0.10 - — POD system can continuously update web sites or send au-
—95% Coma
0.00 tomated text messages or e-mails so that operators
0 0.02 0.04 0.08 0.08 01 can quickly and remotely ascertain the condition of a
flaw length(inch) structure and determine if maintenance action is re-
quired. Solar cells were used to recharge the on-
Figure 5. Probability ofcrack detection curves showing detecta-
board batteries so that the system can provide contin-
ble flaw lengths for CVM sensors.
uous, real-time monitoring for 3-4 years without
The results sited above are valuable for thin-walled maintenance. Multiple years of field operation has
structures such as those used in aircraft, automotive, revealed excellent bridge health monitoring capabili-
and some pipeline construction. However,many civil ties of such an SHM system.
structures use thick steel members. Earlier studies re-
vealed that the thickness of the plate can affect CVM Vacuum Communication device
performance so additional performance tests studied
CVM crack detection in thick-walled structures. Air-
craft use thinner materials and have crack detection
requirements of 1.27 mm to 2.54 mm in length. Civil
structures contain thicker materials, have higher
safety factors and can tolerate longer cracks such that
their crack detection requirements are in the range of
12.7 mm to 25.4 mm in length. Additional tests stud-
ied CVM sensors monitoring cracks in 9.5mm thick
steel. For the loaded structure, CVM crack detection
occurred when the fatigue cracks ranged from 1.02
mm to 1.78 mm in length. For the unloaded condi- Figure 6. Real-time, remote monitoring system for a network of
tion, CVM crack detection occurred when the fatigue CVM sensors.
cracks ranged from 1.52 mm to 9.65 mm in length.
However, regardless of whether the sensor monitor-
ing is completed during a loaded or unloaded condi- 5 CONCLUSIONS
tion, the results indicate that CVM sensors could reli-
ably detect fatigue cracks well before they reach 12.7 The effect of structural aging and the dangerous com-
mm (0.5") in length. bination of fatigue and corrosion has produced a
greater emphasis on the application of sophisticated
health monitoring systems. Corrective repairs initi-
ated by early detection of structural damage are more
cost effective since they reduce the need for subse- situ crack detection. Proceedings of Air Transport Assoc.
quent major repairs and may avert a structural failure. Nondestructive Testing Forum.
Through the use ofin-situ CVM sensors, it is possible U.S. White House document(2003). Nationalstrategyforphys-
ical protection ofcritical infrastructures and key assets.
to quickly, routinely, and remotely monitor the integ- Wheatley, G., J. Kollgaard, J. Register, & M. Zaidi (2003).
rity of a structure in service and detect incipient dam- Comparative vacuum monitoring as an alternate means of
age before catastrophic failures occur. These sensors compliance. Proceedings of FAA/NASA/DOD Aging Air-
can be attached to a structure in areas where crack craft Conference.
growth is known to occur. On a pre-established mon- Roach, D., T. Rice, S. Neidigk, & D. Piotrowski (2015). Estab-
itoring interval, a reading can be taken remotely or lishing the reliability ofSHM systems through the extrapo-
from an easily accessible point on the structure. Each lation of NDI probability of detection principles," Proceed-
time a reading is taken, the system performs a self- ings of International Workshop on Structural Health
Monitoring.
test. This inherent fail-safe property ensures the sen-
sor is attached to the structure and working properly
prior to any data acquisition. In several structural cat-
egories studied, the CVM sensors provided crack de-
tection well before the crack propagated to the critical
length determined by damage tolerance analyses. In
addition, there were no false calls experienced in over
200 fatigue crack detection tests. The sensitivity, re-
liability, and cost effectiveness of the CVM sensor
system was demonstrated in both laboratory and field
test environments.
Global SHM, achieved through the use of sensor
networks, can be used to assess overall performance
(or deviations from optimum performance) of large
structures such as bridges, pipelines, transport vehi-
cles, and buildings. The ease of monitoring an entire
network of distributed sensors means that structural
health assessments can occur more often, allowing
operators to be even more vigilant with respect to
damage onset.

6 ACKNOWLEDGEMENTS

Sandia National Laboratories is a multi-mission la-


boratory managed and operated by National Tech-
nology and Engineering Solutions of Sandia, LLC,
a wholly owned subsidiary of Honeywell Interna-
tional, Inc., for the U.S. Department of Energy's
National Nuclear Security Administration under
contract DE-NA0003525.

7 REFERENCES

Roach, D.(2006). Health monitoring of aircraft structures using


distributed sensor systems. Proceedings of
DoD/NASA/FAA Aging Aircraft Conference.
Beral, B.& H. Speckmann (2003). Structural health monitoring
for aircraft structures: A challenge for system developers
and aircraft manufacturers. Proceedings of International
Workshop on Structural Health Monitoring.
Roach, D.(2004). Use of distributed sensor systems to monitor
structural integrity in real-time," Quality, Reliability, and
Maintenance in Engineering, Oxford,UK:Professional En-
gineering Publishing Ltd.
Roach,D,J. Kollgaard,& S. Emery(2006). Application and cer-
tification ofcomparative vacuum monitoring sensors for in-

You might also like