BFB 0030473

Dam Safety: Improving Management
Eliane Alves Portela and Jo/to Bento t

Laborat6rio Nacional de EngenhariaCivil (LNEC), Dams Department
Av. do Brasil 101, 1799 Lisboa CODEX, Portugal
e-mail: e_ljane@lnec.pt
t Instituto SuperiorT6cnico(tST), Civil Engineering Department
Av. RoviscoPais, 1096 LisboaCODEX, Portugal
e-mail: o~~ 2 0 vil2.ist,utl.pA
Abstract. Dam engineering emerged as a structured branch of engineering

in the earlier parts of this century. Because many large dams are now over
50 years old, special attention has been devoted to the issue of their
efficient management. The act of managing large dams is very complex
and requires the effort of multi-disciplinary teams involving highly
specialised knowledge. In the case of dams, such knowledge is
accumulated through a process of learning and understanding how each
structure responds to the current and exceptional actions. Material
properties may change during the course of their life, which means the
structural responses to certain actions may change with time. Through the
surveillance activity the changes in structural behaviour can be monitored
and any anomalous situation should be identified by a thoroughly analysis.
As old structures tend to deteriorate and request special attention from
managers and as more structures are built every day, managers are faced
with large amounts of information which need to be evaluated in order to
highlight any anomalous situation which may endanger the structure
serviceability or its safety, and trigger possible remedial actions. The
present work has been dedicated to investigate how knowledge based
system technology could help managers in the everyday duty of the dam
safety control activity. The system is meant to support dam managers in
the decision-making process but not to act as a replacement of sound
human and engineering judgements.
1. Introduction
Safety and serviceability requirements of existing dams must endorse all relevant
aspects of its history, namely: design, construction and operation. Such task is very
comprehensive, demanding and often very puzzling. To ensure that a dam remains in
good health, surveillance must be continuos and performed at a professional level. To
assure timely perceptive analysis, those who process surveillance data must be
selective. They must be able to sort out what may be important and study it quickly.
Otherwise, there may be a tendency to bog down in the voluminous detail that can be
generated by a comprehensive system of observation. After initial basic reviews,
those chunks of data that may indicate questionable trends should be examined in
greater depth [ 1].
AI concepts and techniques, namely knowledge based systems, are tools with a
recognisable potential to assist the engineering activities related to monitoring,
interpretation and diagnosis.
456
A few interesting associations between expert system technology and dam safety
control activities have been reported with success in the literature [2, 3].
The work reported in the present paper concerns the development of one such
knowledge based system aimed at identifying abnormal structural dam behaviour
through the analysis of data originated both at the automatic monitoring system and as
a result of regular visual inspections. The system is intended to encapsulate different
types of information and knowledge about dam safety. It relies mainly in the results of
both quantitative and qualitative monitoring data, design parameters, specific tests,
operation history, site characterisation, expert knowledge on dam safety and on the
historical behaviour of the specific structure being monitored.
2. Safety Assessment
The lifecycle of monitoring data starts with its acquisition on site, either by
manual or automatic acquisition means. The original raw data suffers a first level
check against basic parameters, such as the instrumentation range, and the sensor
readings are translated to engineering quantities. On a subsequent second level such
"raw" engineering quantities are validated against pre-defined mathematical models.
Once the data is considered "reliable", the actual safety assessment may take place.
Traditionally, most of the safety assessment tasks have been carded out by very
experienced engineers, who have accumulated a great amount of experience about
dam engineering and most specifically about each particular dam being assessed.
One of the main goals of the present research is to elicit engineering reasoning
processes out of such a few dam experts, so that a model of that expert reasoning may
be developed. By doing so, one expects to support less experienced staff and relieve
experienced engineers from the everyday duty of routine safety assessment, allowing
the expert more time for more complex problems.
Basically, three different levels of knowledge are identified: i) general
background; ii) specific knowledge and iii) expert knowledge. The first type of
knowledge corresponds to general scientific knowledge, such as structural mechanics,
hydraulics and geotechnical engineering. Such general background knowledge has a
very broad base of application, even though it is applied here within the relatively
specialised context of dam engineering. The second form a separate chunk since each
particular dam is a unique system and must be treated as such; its particularities must
be identified to build the specific knowledge base on the dam. Finally, expert
knowledge forms a fundamental part of the system, combining general background
knowledge and specific knowledge about the dam. It enables reasoning processes
leading to an overall evaluation of the whole system.
It is therefore obvious that the efficiency of the system depends deeply on the
experience and adherence of the experts involved.
3. System Development
The system under development is being applied to a specific case of a double
curvature arch dam in Portugal during its operation. It attempts to monitor a number
of specific data values and spot abnormal ones or trends suggesting remedy measures
to solve or mitigate foreseen problems. Monitored quantities are water level,
457
temperatures, displacements, strains, stresses, joint movements, seepage and uplift.

Whenever one of these quantities is detected to fall out of the expected range, the
system should trigger an inference process leading to both a diagnosis and proposal
for action [4].
Following the data collection, the monitoring data must be immediately processed
and analysed with the results being checked in order to ascertain whether they
conform to the pre-established behaviour models; this comparison may give rise to a
reformulation of the models and a reassessment of the safety conditions of the works.
To explain an abnormal value found in the monitoring data, a dam engineering
expert searches for a cause of the abnormality using his own expertise and general
knowledge of the structure. Trying to mimic such reasoning processes, the system
uses a set of complex and very large causal networks to identify possible scenarios
associated to the detected anomalies. The causal networks and the relevant associated
scenarios have been produced as a result of a thorough knowledge acquisition
process.
Three groups of scenarios were defined:
• General scenarios - related mainly with the unusual or exceptional loads and
deficiencies of design and construction;
• Foundation scenarios - related mainly to the dam foundation, such as weakness of
the foundation under permanent or repeated loading, deterioration of grout
curtains, erosion and solution, sealing of fractures, clogging of drainage system,
plastic displacement and sliding of foundation, settlement and swelling of
foundation material, and
• Dam body scenarios - related mainly to problems on the dam body, such as
contraction, creep, swelling of concrete, loss of structural continuity, loss of
structural strength under permanent or repeated actions and degradation due to
chemical reactions of materials with the environment.
For each of the above scenarios a set of symptoms is established to confirm the
relation between the abnormal measurement and the identified scenario, which will
lead to a diagnosis. The set of conditions established for each scenario is based on the
available cause-effect relationship, which may also affect the properties of the
materials; and may be reflected in the monitored quantities in a specific way.
For each of the conditions it is defined a general descriptor, referring to its state:
an example being a slow decrease or a rapid increase. The description by qualifiers
is, by definition, subjective and a formal means of propagating subjective information
is required so that consistent results can be obtained. Nevertheless, it is not an easy
task to find "indicators" univocally connected with the safety state of the structure,
and consequently one has to rely on rather indirect and tortuous procedures of
"inference", on the basis of easily readable "indicators".
Each scenario is also linked to more specific symptoms: geological and
geotechnical conditions, operation schemes, visual inspection data, water quality data,
concrete properties, design and construction data, etc.
The correct identification of a scenario depends on the description of the
appropriate symptoms. Part of the ability of the system to focus the search for
possible scenarios on the ones that might afflict the structure, results from an
automated identification of which symptoms are of primary importance and which are
of secondary importance for the diagnosis of abnormal behaviours.
458
Primary symptoms are sure signs that a particular scenario has indeed taken or not
taken place and, therefore, present strong evidence for the diagnosis of a particular
type of scenario. Secondary symptoms provide some information about possible
scenarios and represent weaker evidence for such particular scenario. Secondary
symptoms become relevant when a fine-tuning of the diagnosis is required or if
insufficient information is initially available to make a diagnosis.
Once a scenario is identified by the system through association of symptoms, the
system compares its findings with a knowledge source containing a series of problems
with possible recommendations. If a match occurs, the system displays the result.
Recommendations may consist of remedial measures, methods leading to further
investigation to confirm a diagnosis, a preventive measure, or even a mere (justified)
statement that no action is required.
4. Conclusions
The only way to prevent deterioration of structm'es and its natural consequences -
danger of loss of serviceability and safety- is by enforcing a proper, thorough and
timely assessment of the structure.
A knowledge-based system which identifies abnormal dam behaviours through the
analysis of data coming from a monitoring system and visual inspection was
presented. The system is intended to help managers overcoming the difficulties
related with this activity and to support the full understanding of a dam's behaviour,
while promptly identifying, at early stages, a path to a possible incident. Indeed, an
immense set of domain knowledge needs to be invoked, including that of previous
cases, in order to spot a problem, diagnosing it, and enabling the proposal of
therapeutic action. No such task could be easily and effectively performed without the
aid of A1.
Acknowledgements
The present work has been funded by the PRAXIS XXI Program, through
research grants no. BD/278/94 and 3/3.1/CEG/2547/95.
References
1. Jansen, g. B. (1983) Dams and Public Safety. A Water Resources Technical
Publication. U.S. Department of the Interior, Bureau of Reclamation, 332 p..
2. Salvaneschi, P.; Cadei, M.; Lazzari, M. (1997) "A Causal Modelling Framework
for the Simulation and Explanation of the Behaviour of Structures" in Artificial
Intelligence in Engineering, 11,205-216, Elsevier.
3. Zaozhan, S.; Zhongru, Wu (1994) "Cause Analysis in Dam Safety Assessment",
in Dam Engineering, V-I, 31-41.
4. Portela, E.A.; Bento, J.; Ramos, J.M.; Silva, H.S. (1998) "The Safety Control of
Dams: Improving Management Through Expert System Technology",
Proceedings of the Fourth World Congress on Computational Mechanics, Buenos
Aires, 29/June-2/July/1998 (to appear).

BFB 0030473

Uploaded by

Copyright:

Available Formats

BFB 0030473

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BFB 0030473

Uploaded by

Copyright:

Available Formats

Dam Safety: Improving Management

Eliane Alves Portela and Jo/to Bento t

Abstract. Dam engineering emerged as a structured branch of engineering

temperatures, displacements, strains, stresses, joint movements, seepage and uplift.

You might also like