See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261119109

Structural Health Monitoring: research and practice

Conference Paper · September 2013

CITATIONS READS

9 17,153

1 author:

Andrea Enrico Del Grosso

University of Genoa
78 PUBLICATIONS 598 CITATIONS

SEE PROFILE

All content following this page was uploaded by Andrea Enrico Del Grosso on 27 March 2014.

The user has requested enhancement of the downloaded file.

Structural Health Monitoring: research and practice

Andrea E. Del Grosso1

1
University of Genoa, Genoa, Italy

ABSTRACT: Structural Health Monitoring is one of the preferred research topics in structural
engineering but practical applications are still behind, at least in the civil sector. The paper is
aimed at reviewing the main research achievements on the subject and to argue about the
reasons because practical applications still encounter difficulties in becoming a standard
practice in civil engineering. Structural health monitoring concepts and current design
approaches are also discussed with consideration of the safety of monitored structures versus
conventional non-monitored ones. Existing standards on structural monitoring and the need for
the development of new standards integrating design, maintenance and management of
constructed facilities are addressed.

1 INTRODUCTION
Observation of structural behavior is a very old discipline that has accompanied theoretical
developments in structural mechanics since its origins (Benvenuto 1991), providing basic
knowledge of physical phenomena and verification of computational procedures. However, in
the last twenty years this discipline has also taken different roles, gradually becoming the basic
tool for facing the so-called time-dependent safety problem ( Mori and Ellingwood 1993) in
civil engineering practice.
The shift from simple experimental observation to Structural Health Monitoring has been driven
by two factors: on the one hand, by the consequences led by degradation of modern construction
materials and functional obsolescence onto infrastructure economics and, on the other hand, by
the availability of cheap, effective and durable innovative instrumentation and
hardware/software tools to accomplish complex data acquisition and signal processing
functions. Structural Health Monitoring (SHM) is indeed just the combination of traditional
experimental/theoretical structural mechanics, electronics, material science, and information and
communications technologies. Applications of this discipline can lead to the definition of
monitored structures, a class of structures the characteristics of which in terms of safety and
reliability indices should be considered differently from traditional structures, where safety
relies on passive resistance only, in order to derive specific integrated design approaches (Del
Grosso 2008 ).
In addition, the integration of monitoring system concepts in structural design is an essential
step in innovative structural engineering, paving the way to the development of smart adaptive
structural systems.
This paper is aimed at reviewing the main research achievements on the SHM subject and to
argue about the reasons because practical applications still encounter difficulties in becoming a
standard practice in civil engineering.

2 MATERIALS DEGRADATION AND OBSOLESCENCE

In developed countries, the greater percentage of infrastructures have been built just after World
War II using steel, reinforced, composite or pre-stressed concrete structural systems. These
techniques still are the most commonly used construction systems worldwide. Materials
degradation and obsolescence are a key issue in infrastructure management not only where
infrastructure stocks are so old (Aktan et al. 2007) but also where, as in recently developed
countries, they represent a problem in perspective. Indeed, the physical and mechanical
properties of these construction materials tend to degrade with time at a relatively significant
speed, thus causing a loss in the economic value of the infrastructure assets. For example,
recent studies have stated that the global economic consequences of corrosion may be evaluated
to reach 3 to 4 GDP points per year (Schmitt et al. 2009).
Considering concrete structures, which are largely the most diffused ones, the most common
and serious in terms of consequences, cause of deterioration in structural members is due to
corrosion of reinforcing steel induced by chloride ion ingress into concrete. Other less common
causes of deterioration in concrete are carbonation induced corrosion, freeze-thaw attack, alkali-
silica reaction, and external and internal chemical attack. Concrete degradation and especially
chloride ion ingress and concrete carbonation is responsible for creating a corrosion potential
for the steel bars, but the actual development of corrosion and the rate of the process are also
dependent on temperature and moisture content in the surrounding concrete (Dangla and Dridi
2009), thus rendering the phenomenon very complex. Besides corrosion, fatigue is also an
important cause of degradation in steel structures subjected to moving loads or vibrations. In
bridges, degradation of joints and supports because of fatigue, corrosion and ageing is also an
important issue influencing management strategies and costs.
Corrosion and material degradation cause a decrease in the resisting section of members and
fasteners which in turn results in a degradation of resistance and stiffness of the whole structural
system. Detection of the presence and progress of the phenomena can be made by direct
monitoring of the electrochemical driving parameters or, indirectly, by analyzing the changes
with time of the structural response (Del Grosso et al. 2008, 2011).
The concept of obsolescence is more related to the evolution of the needs of infrastructure users,
for example (for transportation infrastructures) in terms of commercial speed, traffic volumes,
size and weight of vehicles etc., but obsolescence can also be produced by the unfavorable
levels of maintenance costs induced by degradation. Evaluation of obsolescence results from
complex considerations involving direct, indirect and social costs for decommissioning and
substitution, but the corresponding decision making process is based on parameters that can be
quantitatively estimated from direct and indirect observations.

3 MAINTENANCE STRATEGIES AND COST OPTIMIZATION

Due to the large economic effort needed to keep the existing and future infrastructure systems in
efficient and safe conditions, in the recent years several studies and practical applications have
been performed on maintenance strategies and maintenance cost optimization.
The approach that has recently received considerable attention and that is considered the most
attractive for practical applications is based on the use of lifetime functions. A lifetime function
(Figure 1) represents the decay in time of a performance index that may eventually represent the
reliability index or a more complex weighted sum of several indicators.
INDEX

DESIGN VALUE

MAINTENANCE

LIMIT VALUE

DESIGN LIFE EXTENSION TIME

Figure 1. Typical lifetime function and the effect of maintenance.

The use of lifetime functions has been introduced by several Authors; among them it is worth
mentioning the works by Miyamoto et al. (2001) and by Frangopol and Liu (2006) in the
context of lifecycle cost optimization. A more recent review of the approach, performed in the
framework of the European project IRIS (Wenzel et al. 2011) is leading the method to represent
an effective and practical tool for managing constructed facilities.
In synthesis, it is a-priori assumed that the decay of the performance index, originally at the
design value, is such that the limit acceptable value is reached at the end of the design life and
that the lifetime curve is represented by a simple exponential expression. At any time during
the life of the facility, a maintenance intervention should be able to improve the index and, at
the limit, recover the design value of the index itself extending the expected operational life.
Preventive and condition based maintenance can both be considered within the process.
Maintenance can be repeated several times and the operational life can in principle be extended
as long as economically feasible. The above formulation allows to establish a life-cycle cost
optimization process based on heuristics and knowledge-based rules.
All quantities involved in the process are however uncertain in nature; their determination can
be based on statistical knowledge bases and therefore the process can be formulated in
probabilistic terms. It is noted that the whole procedure could be developed in some backward
processing, involving also a re-determination of the safety coefficients to be used at the design
stage.
Assessment of the actual structural conditions allows the a-priori lifetime curve to be
periodically updated with the effect of reducing the uncertainties involved in the process and
transforming the approach in a really effective infrastructure management tool. Structural
Health Monitoring ( Del Grosso and Lanata 2011) can be regarded as a tool for performing this
task (Figure 2).
In current infrastructure management the use of SHM is not however a common practice.
Although in many special cases, like long-span bridges and super-tall buildings, SHM systems
have been efficiently implemented and used for maintenance planning, most of the
infrastructure management applications (e.g.: highway and railway bridges) are still based on
traditional observations (visual inspection and standard NDE). There are many reasons for that.
The following is a tentative list of those reasons.
• Standards and regulations concerning infrastructure safety impose performance of
traditional inspections at fixed time intervals; this obligation cannot be legally
avoided using SHM systems.
• Although a consistent number of damage identification algorithms have been
proposed and validated in the literature, the reliability of the determination of the
structural conditions from the SHM data is still to be widely experienced.
• Although very reliable, durable and stable sensors technologies are nowadays
available on the market, the sensory systems always show some malfunctions; this
needs redundancies at sensor installation and maintenance during operations.
• The operational life of electronics (data loggers, computers, etc.) is shorter than that
of any other system components and much shorter than the operational life of the
structure; this will require frequent substitutions of electronic components.
• Education on SHM systems and global infrastructure monitoring approaches is still
not enough diffused in civil engineering university programs; consequently,
engineers in infrastructure owners organizations are reluctant to rely on SHM.
In synthesis, the economic and technical advantage of using SHM systems in infrastructure
management is still questioned by potential users. Recent discussions held at an academic
workshop (6th IASCM International Workshop on Structural Control and Health Monitoring,
Sydney, 2012) have pointed out such situation and traced research needs for possibly
overcoming the above difficulties in the diffusion of SHM technologies.

Sensor 1
Data Pre- Damage Identification Process
Processing
Sensor 2

Data Lifetime Function

Fusion Update

Sensor n
Data Pre-
Processing
Sensor m

Figure 2. Lifetime functions update via SHM.

4 MONITORING SYSTEMS
A very large amount of studies and experiences on monitoring systems have been made
available in the recent years but some issues still remain open. A brief summary of research
results and some open questions are presented here.

4.1 Permanent versus periodic monitoring

By permanent monitoring it is intended a monitoring system that is permanently installed and
maintained in operation on the structure, typically from the construction stage. This is the most
complete approach to SHM, allowing to obtain continuous time-series of data comprising
structural response parameters (static and dynamic), environmental parameters, load
characteristics, and other quantities important to the control of materials degradation processes.
The conceptual advantage of permanent monitoring systems is that the time-series of data can
be processed in many different ways, including on-line and multi-stage processing, disclosing
features that may also reveal unexpected structural behaviors. Events like earthquakes, shocks,
storms etc. can be completely described allowing a comprehensive evaluation of the phenomena
and of the corresponding structural response.
This is important not only for assessing the conditions of the single structure under study but
also for characterizing events that have a low probability of occurrence and that are not
consistently modeled in design codes. In addition, data processing can be performed on-line
allowing warnings and alarms to be raised in real-time. Rain-flow counts can be performed on
stress time-histories to provide on-line evaluations of the accumulated damage and of the
residual fatigue life. The disadvantage of permanent monitoring systems is that they are
relatively expensive, they need to be designed very carefully and they produce a very large
amount of data, thus requiring a dedicated organization and complex architectures for data
transmission, management and permanent storage.
Periodic monitoring is performed by temporarily installing an appropriate sensory system on the
structure and gathering data for a short time (from a few hours to a few weeks). Feature
extraction is performed for every measurement campaign and the health conditions of the
structure are determined from the time-histories of the characteristic features of the campaigns.
Periodic monitoring presents several advantages. First of all, periodic monitoring may be
considered a non-destructive evaluation tool more sophisticated than traditional ones but
conceptually consistent with them, and therefore more easy to be understood by infrastructure
owners. Secondly, the cost of acquisition and maintenance of the instrumentation system is
distributed on the number of structures to be monitored. There is no significant difference in the
damage identification algorithms that can be applied but data management is simpler than in the
previous case.
The main disadvantages reside in the fact that the sensor typologies are necessarily limited and
consequently some phenomena cannot be recorded and, of course, accidental events occurring
between subsequent campaigns cannot be recorded as well, although their effects inducing
damages in the structure could be disclosed.
In infrastructure management practice, there is no clear understanding on whether one approach
is superior to the other. It can be noted that, in general, permanent monitoring is to be preferred
for large complex structures, while periodic monitoring is more suitable for SHM applications
on large structure stocks comprising repetitive simple schemes. Table 1summarizes the main
characteristics of the two approaches.
Table 1. Characteristics of permanent versus periodic monitoring

Permanent Monitoring Periodic Monitoring

Sensor types Extended Restricted

Data management Complex Simple

Accidental events Recorded Not recorded

Damage identification On-line Off-line

Warnings & Alarms Real-time Deferred

Fatigue life evaluation Direct Indirect

Installation costs High Low

Operational costs High Low

4.2 Diagnostic and Prognostic Algorithms

The development of damage identification or diagnostic algorithms is a very common topic in
SHM research. For damage identification it is intended a procedure able to analyze the
monitoring data and determine occurrence, location and intensity of damage. Hundreds of
journal and conference papers have proposed a large variety of such procedures. Their
effectiveness is usually proven by analyzing computer simulated data, benchmark studies and
small scale laboratory experiments. Relatively few papers are reporting about damage
identification on real structures subjected to artificially induced damages, normally using
measurements of dynamic response before and after a known damage level has been induced in
the structure. In the Author’s knowledge, there is no case reported in the literature where
algorithms of this type have revealed insurgence of damage in real structures but cases are
reported where behavioral anomalies with respect to predictions given by design models have
been detected. In the Author’s opinion, the development of diagnostic algorithms has reached a
substantial maturity and the preparation of a comprehensive review paper will be very fruitful
for disseminating them to potential practical users and identifying the needs for future research.
All algorithms need a period of observation in which the structural health conditions can be
considered unchanged (reference period). The effectiveness of a diagnostic algorithm can be
measured in terms of: a) length of the reference period, b) minimum detectable damage for
given signal to noise ratios, c) time of observation after damage needed for detection, d)
capability of locating damage, e) capability of determining the intensity of damage, f) capability
of identifying multiple damages occurring at different locations, and g) reliability. This latter
aspect has been recently investigated (Del Grosso and Lanata 2012) but further research is still
needed. A synthetic categorization of the algorithms can be found in (Del Grosso 2012).
The computational complexity of the different algorithms is also very different and the
influence of environmental conditions encountered in real cases is largely influencing their
effectiveness. In practical applications, SHM operators privilege the use of the most simple of
them, consisting in frequency analysis, various types of correlation and simple predictive
models, leaving the more complex to successive stages of processing. It is noted that simple
algorithms can be easily implemented in smart sensing systems to provide quick on-line
detection of anomalies.
As concerning prognostic algorithms, i.e. algorithms able to estimate the remaining life of the
structure, they can be grouped in two classes. A first class makes use of finite element
structural models that include material degradation models. In these models the static or
dynamic parameters are optimized in order to reflect the real structural response and the
evolution of the structural conditions. The other class comprises heuristic models. A simplified
and very practical approach is to use the updated lifetime functions to predict the expected life.
This approach avoids the computational complexity of the first class of methods but provides
very useful information to support engineering decisions.

5 GUIDELINES AND STANDARDS

A limited number of guidelines and standards has been released to date. The only official
international standards are the ISO 14963:2003 – Mechanical Vibration and shock – Guidelines
for dynamic test and investigations on bridges and viaducts and the ISO 18649:2004 –
Mechanical vibrations – Evaluation of measurement results from dynamic tests and
investigations on bridges. These standards refer to the use of dynamic measurements to perform
periodic SHM functions on bridges. Other guidelines, more widely addressing the issue of
SHM and the design of monitoring systems have been published by research organizations like
ISIS Canada (ISIS Manual n. 2 – Guidelines for structural health monitoring) or have been
produced in the framework of international research projects like the European SAMCO and
IRIS. In the IRIS framework, a proposal for standards covering the use of lifetime functions has
been elaborated by CEN WG 63.
An interesting standard has been recently issued in Russia (GOST P 53778 2010 Building and
Structures – Technical inspections and monitoring regulations). This standard is mandatory in
the Russian Federation and broadly addresses structural and geotechnical inspection and
monitoring during service life.
Rules for inspection and management of various types of infrastructures have been issued by
several agencies in the world, but they do not expressly address issues related to structural
health monitoring as described in this context.
It is however recognized that the lack of international standards and regulations on buildings
and structures considering the use of SHM represents an obstacle to the diffusion of the
applications. The need for working on this subject is therefore pointed out.
A particular aspect that still need to be investigated from the theoretical standpoint in view of
impacting on design standards is related to the reliability of monitored structures versus non-
monitored ones. In conventional structural design codes according to the European limit state
format or the American LRFD, characteristic values of loads and resistance of materials are
deduced from standard probability distributions and, in addition, safety verifications are
performed by applying appropriate safety factors to characteristic values, to reflect the
uncertainties involved in the process.
A question now arises regarding the appropriateness of those safety factors when uncertainties
are reduced by the presence of a permanent monitoring system on the structure providing
information on the structural conditions and allowing interventions to be made for keeping the
 probability of failure below the acceptable limits. To date there is no study, in the Author’s
knowledge, addressing this question in a systemic way. It is envisaged that the backward use of
the lifecycle functions could provide a useful approach.

6 CONCLUSIONS AND DIRECTIONS OF FUTURE RESEARCH

The paper has summarized the main research and applications achievements in SHM
technologies. Several open problems still remain unsolved and may be the subject of future
research. Apart from standardization needs, as already mentioned in the previous paragraph, the
first and probably most important issue is related to the safety coefficients that should be
adopted in the design of monitored structures. This relates to refurbishment design of existing
structures as well as to the design of new structures, whereby the presence of the monitoring
system can redefine the probabilistic modeling of design uncertainties. A second issue is related
to the use of updated behavioral models (FE models) that constantly reflect the state and the
evolution of structural conditions. Reliable techniques to construct and use these models,
sometimes referred to as “numerical twins” of the real structure, still have to be developed and
experienced.

REFERENCES
Aktan, AE, Ellingwood, BR and Kehoe, B. 2007. Performance-based engineering of constructed systems.
Journal of Structural Engineering 133 (3): 311-323.
Benvenuto, E. 1991. An Introduction to the History of Structural Mechanics. Springer Verlag, Berlin
Dangla, P and Dridi, W. 2009. Rebar Corrosion in Carbonated Concrete Exposed to Variable Humidity
Conditions. Interpretation of the Tuutti’s Curve. Corrosion Science, 51:1747-1756.
Del Grosso, A. 2008. On the Reliability of Smart Monitored Structures, 14th World Conference on
Earthquake Engineering, Beijing, China.
Del Grosso, A E. 2012. On the static monitoring of bridges and bridge-like structures. Bridge
maintenance, Safety, Management, Resilience and Sustainability, F. Biondini & D. M. Frangopol eds.,
CRC Press/Balkema, Leiden. 362-367.
Del Grosso, A, Lanata, F, Pardi, L, Mercalli, A. 2008. Health Monitoring for Corrosion Detection in
Reinforced Concrete Bridges, 4th Int. Conf. on Bridge Maintenance, Safety and Management, Koh
And Frangopol eds., Taylor and Francis.
Del Grosso, A, Inaudi, D, Lanata, F, Posenato, D. 2011. SHM of Ageing Reinforced Concrete Structures,
First Middle East Conference on Smart Monitoring, Assessment and Rehabilitation of Civil Structures
SMAR 2011, Dubai, UAE, paper n. 265.
Del Grosso, AE and Lanata, F. 2011, Uncertainties in Damage Identification and Lifetime Functions of
Ageing Concrete Structures. Structural Health Monitoring 2011, F-K Chang Ed., DESTech
Publication, Lancaster. 1871-1878.
Del Grosso, AE and Lanata, F. 2012. Reliability estimate of damage identification algorithms. Reliability
Engineering and Risk Management (3), Y.G. Zhao, J. Li, Z.H. Lu, T. Saito eds., Central South
University Press, Shanghai. 350-355.
Frangopol, DM and Liu, M. 2006. Life-cycle cost and performance of civil structures. McGraw-Hill 2006
Yearbook of Science and Technology, McGraw-Hill, New York.
Miyamoto, A, Kawamura, K, and Nakamura, H. 2001. Development of a Bridge Management System for
Existing Bridges. Advances in Engineering Software, 32:821-833.
Mori, Y and Ellingwood, BR. 1993. Reliability-based Service-Life Assessment of Aging Concrete
Structures. Journal of Structural Engineering 119 (5): 1600-1621.
Schmitt, G, Schütze, M, Hays, GF, Burns, W, Han, E-H, Pourbaix, A, and Jacobson, G. 2009. Global
Needs for Knowledge Dissemination, Research, and Development in Materials Deterioration and
Corrosion Control, White Paper, The World Corrosion Organization.
Wenzel, H, Veit-Egerer, R and Widmann, M. 2011. Risk Based Civil SHM and Life Cycle Management.
Structural Health Monitoring 2011, F-K Chang Ed., DESTech Publications, Lancaster. 717-724.

View publication stats