You are on page 1of 10

THE REWORK OF THE JOINT MONITORING INDEX BLOCKS

RESPONSE OF THE DAM (IMCRB)


Luana E. Felippi*, Sheila R. Oro1 and Cleonis V. Figueira2
*
Student of Specialization in Numerical Methods in Engineering
Rua Brasília, 80, 85670-000 Salto do Lontra, Brasil.
e-mail: luana_felippi@hotmail.com

Keywords: Concrete Dam; Monitoring of Structures; Factor Analyses.

Abstract. The Joint Monitoring Index Blocks Response of the Dam (IMCRB) was created by
analysing the series of sensors for monitoring the displacements of structures and foundations,
from two buttresses blocks in Excerpt D located in Itaipu’s dam, considering the action of the
ambient temperature and the water level of the reservoir. This index contributes to the
diagnosis of changes in the standard behaviour of the dam structure, and after its formulation
was validated in this same section. In the following study, the IMCRB is reformulated and
validated for the two blocks of Excerpts F, the manual measurement data of the following tools:
pendulum, ancillary bases, shaft extensometer and Piezometers acquired between January
1990 until November of 2017. The IMCRB reworking involved the using of statistical methods:
factor analyses, analyses of series, confidence intervals, with a big participation in the
modeling and the remaining set for index validation. The reworked Joint Monitoring Index
Blocks Responses (IMCRB) of dam can help in the overall performance analyses of selected
blocks concrete structures, preview its behavior and help in the anomalies data identification,
felicitating in the decision making.

1 INTRODUCTION
Any structure constructed transversely to a river or talweg with the function of raising water
levels and / or raising a reservoir of water accumulation for the regulation of river basins or of
another receiving fluid is called dam (MSIB, 2002, p.15). This type of construction forms an
artificial obstacle with the ability to reuse water or other liquid, rejects and debris, with the
purpose of construction or control that can vary from small sand dams, frequently used in
farms, to enormous concrete structures, generally used for water supply, electric energy, flood
control, irrigation, among other purposes (SETTI et al, 2000).
Dams can be classified according to their purpose, their design and the materials used in
their construction. In general, those that are constructed of wood, masonry of stone or gabion,
and conventional ones of concrete type - being of the gravity type, relieved gravity, buttress or
arc -, of earth-type, rock-type or tailings, and mixed concrete and landfills (ORO, 2016).
The structural safety of dams requires special attention. The term security refers to the ability
of dams to meet the behavioral demands necessary to prevent incidents and accidents in
structural, economic, environmental and social aspects (MSB, 2002, p.17). The constant
assessment of their structural health can assist in the determination of anomalies quickly, and
consequently make decisions about it. This activity is the responsibility of the engineering area

1
Academic Department of Physics, Statistic and Mathematics. Universidade Tecnológica Federal do
Paraná – Francisco Beltrão. E-mail: sheilaro@utfpr.edu.br
2
Academic Department of Physics, Statistic and Mathematics. Universidade Tecnológica Federal do
Paraná – Pato Branco. E-mail: cleonis@utfpr.edu.br
1
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

and should be performed with the aid of monitoring instrumentation and visual inspections
(ZUCULIN, 2012, p.9).
The main instruments used for monitoring of the structures of a concrete dam are: selector
box, direct pendulum, inverted pendulum, gasket meter, basemeter, electrical joinmeter,
reinforcement strainmeter, concrete stressmeter, concrete strainmeter and resistance
thermometer. As foundation instrumentation, multiple shaft extensometers, standpipe
piezometers, IPT settlement gauges, total pressure cells, triorthogonal meters and water level
gauges are used.
The general aim of this work is the usage of multivariate analysis techniques to manage
concession and collaboration for decision making about the structural stability of dams. For
that purpose, instrumentation sensors that will improve the relativity of two variables have been
identified, control charts for the instrumentation of monitoring structures and foundation of
concrete dams were elaborated and the overall behavior of concrete dam structures were
estimated through IMCRB (joint monitoring index of responses of the blocks).
Monitoring of dams involve a huge quantity of information and data, generally collected
starting from the time of its construction, being their interpretation empirically done. With that
said, the subject addressed is an alternative to collaborate on these structures monitoring
proccess and help on the decision making process about their safety.

2 MATERIALS AND METHODS


This work is characterized as a descriptive research. The descriptive research, according to
Gil (2008), "have as main objective the description of the characteristics of a certain population
or phenomenon or, therefore, the establishment of relations between variables".
The data were collected together with the CEASB (Center for Advanced Studies in Dam
Safety), responsible for the safety of the Itaipu Dam, which made the measurements of the
instrumentation installed in blocks F19 and F20, of Excerpt F of the Itaipu Dam available.
Following, information on the Itaipu Dam and the methods used for data analysis are shown.

2.1 Itaipu dam


The Itaipu Hydroelectric Power Plant is a bi-national venture between Brazil and Paraguay,
located on the Paraná River, near the city of Foz do Iguaçu. The Itaipu dam is 7,919 meters
long and has a maximum height of 196 meters. It is built with a mixed structure, consisting of
concrete, earth and rockfill, with the main purpose of generating clean and renewable energy.
(ITAIPU BINACIONAL, 2018).
Throughout its extension, whether in structures or foundations, instruments are used to
monitor the safety of the Itaipu Dam. There are approximately 2,400 instruments installed, of
which 1358 are in concrete, 881 in foundations and 161 in geodesy. However, only 270 have
automated readings, the others only have manual readings (ITAIPU, 2018). The CEASB -
Center for Advanced Studies in Dam Safety - is the department responsible for the structural
monitoring of the dam and the development of strategic solutions in this area.

2
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

Figure 1: Itaipu Dam

ROOT: ITAIPU BINACIONAL, 2018

The Itaipu dam (Figure 1) is a concrete, gravity relief type of dam that extends for 600
meters through the Paraná river bed, where some blocks reach 196 m of height, having received
a detailed instrumentation for the observation of the repression (SILVEIRA, 2003, page 163).
The left link dam is also concrete, buttress type. The other types will not be approached on this
work due to its approach.

2.2 Statistical Methods


For this work, we used time series models, multivariate method of factor analysis and
control charts.
According to Neto (2017), time series are a set of observations generated sequentially in
time and that present a serial dependency between them. The factorial analysis refers to the
identification of implicit sources of variation common to two or more variables, called factors
(LATTIN, CARROLL, GREEN, 2011). According to Neto and Chaves (2017), control charts
are used to detect particular factors of variation outside control limits and also to non-random
systematic behavior.
The research carried out consists of two phases, the first one related to the reformulation of
the Joint Monitoring Index of Dam Blocks Response (IMCRB) for key blocks F19 and F20 of
Excerpt F of the Itaipu dam, and the second with the validation of it.
The CEASB made available a database composed of the values of the manual measurements
of displacements, rotations and percolations measured by the instruments considered in this
work, namely: direct pendulums, inverted pendulums, multiple rod extensometers, piezometers
and extensor bases, totaling 36 sensors analyzed , which are installed in the key blocks F19 and
F20 of Excerpt F. The data used refer to the manual measurements performed from January
1990 to December 2017.
As most of the instruments had several measurements during the month, the monthly
average of the measurements of each instrument was initially calculated, forming a matrix of
size 336x36, in which each column represented an instrument and its measurements over the
specified period.
Some gaps in the data matrix were observed, which were filled by forecast values
determined through time series. Data from this matrix was standardized and separated into two
subgroups, the first with data between the period January 1990 and December 2015, referred

3
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

to as the "first group", and used for the formulation of the IMCRB, and the rest of the data
between the period of January 2016 and December 2017, denominated as "second group", used
for validation.
The factorial analysis was performed with the first data set, through R Software, in order to
rank the instruments according to their importance. The factors determined by the analysis
generated scores, which were used as input data for the IMCRB, which is described as follows:
∑𝑚
𝑘=1 𝜆𝑘 ƒ𝑖𝑘
IMCRBi = ; i = 1, 2, 3, ..., n
𝑝

𝜆𝑘 = k-nth value of the correlation matrix;


𝑓𝑖𝑘 = i-nth score do k-nth extracted factor;
p = quantity of utilized sensors (number of variables);
n = number of readings for each sensor.

Using the two factors as inputs for the index, a series with 312 values was determined, with
the same quantity of measurements of each instrument up to december of 2015. Starting with
the series, limits of 95% and 99% confidence were stipulated. The values of IMCRB are within
these limits.
With the help of R Software, the model that describes the values of first group series was
stipulated, made of the manual measurements from January 1990 to December 2015. The
estimated values of the model, forecasts and confidence limits were pltted on a control letter
and analyzed if within the limits.
The method was validated calculating a new IMCRB series with the values obtained from
two scores of the factor data from the second group, from between January 2015 and the
December 2017.

3 RESULTS AND DISCUSSION


The analyzed data is related to 36 sensors called as variables X1, ..., X36. These include:
direct pendulums, inverted pendulums, extensometer meters, piezometers and bases of
measure. As descriptive measurements: mean and standard deviation (SD) of each sensor are
presented on Table 1. It can be observed that the data is quite heterogeneous, because of the
differences of greatnesses and units of measurement. Some sensors will exhibit high variability,
such as the X10 sensor, being the main cause of this characteristic the seasonality of the present
series (winter-summer cycles).

Table 1: Descriptive Measures


X1 X2 X3 X4 X5 X6 X7 X8 X9 X10
Mean 1,94 2,02 2,75 0,86 4,69 -0,72 7,03 -1,14 9,45 -0,64
DP 0,47 0,37 0,92 0,31 1,42 0,44 1,73 0,45 1,86 0,45
X11 X12 X13 X14 X15 X16 X17 X18 X19 X20
Mean 14,39 -1,91 3,53 1,79 5,06 5,71 -11,29 -9,07 -7,00 -8,20
DP 2,17 0,28 0,34 0,23 0,24 0,22 0,30 0,29 0,26 0,16
X21 X22 X23 X24 X25 X26 X27 X28 X29 X30
Mean -6,91 -9,26 -7,08 1,31 -0,45 1,27 -2,09 0,52 -0,27 148,77
DP 0,16 0,15 0,13 0,05 0,02 0,04 0,09 0,32 0,02 7,53
X31 X32 X33 X34 X35 X36
Mean 89,74 139,22 97,44 35,67 18,98 19,28
DP 6,75 20,46 13,39 2,22 0,12 0,12

SOURCE: OWN AUTHORSHIP

4
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

After obtaining the matrix of the monthly averages and performing the standardization of
the data, the first set of data was submitted to factorial analysis with the aid of R Software. For
this, covariance and correlation matrices were estimated, and the sphericity test of Bartlett to
test the hypothesis that the correlation matrix is equal to the identity matrix. The test showed
that the value of the T (Tcalc) statistic was greater than the critical value with 95% confidence
level (T_95):
Tcalc = 26085,2 e T_95 = 689,5012

Thus, the hypothesis that the correlation matrix is equal to the identity matrix is rejected,
and the response variables of the sensors are correlated.
To verify if the factorial analysis of the sample data was adequate, the KMO index was
calculated, which returned as a result the value:
KMO = 0,9296022

Showing that it is appropriate to perform the factorial analysis, in such a way that its value
exceeds the established minimum of 0.5 for the adequacy of the analysis.
The factorial analysis was performed with 6 factors, a quantity indicated by the Kaiser
criterion, and corresponded to 90.11% of the total data explanation. The Kaiser criterion took
into account the eigenvalues determined from the correlation matrix, which were greater than
one.
The commonalities of the sensors of each instrument (Table 2) were considered high
because they were equal to or greater than 0.55, which means that all variables (sensors) are
well represented by the factorial model. The graphs of these factors are shown in Figure 2 from
left to right.

Table 2: Communalities
X1 X2 X3 X4 X5 X6 X7 X8 X9 X10
0.83 0.86 0.94 0.84 0.96 0.91 0.98 0.93 0.96 0.89
X11 X12 X13 X14 X15 X16 X17 X18 X19 X20
0.98 0.72 0.79 0.68 0.97 0.81 0.99 0.97 0.98 0.99
X21 X22 X23 X24 X25 X26 X27 X28 X29 X30
0.99 0.98 0.97 0.84 0.55 0.76 0.93 0.81 0.55 0.90
X31 X32 X33 X34 X35 X36
0.93 0.78 0.83 0.99 0.81 0.93
SOURCE: OWN AUTHORSHIP

5
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

Figure 2: Factor graphs

ROOT: SOURCE AUTHORSHIP

Each factor represents the behavior of a set of sensors, which had the highest correlation
with it.
The main characteristics perceived in each factor are:
 The first factor presented a decreasing tendency in the average, but with a small
variation, around 1 unit. This factor is considered to be of greater importance in relation
to the response of the dam movement, since it represents 49.3% of the total;
 The second factor indicates that the data is distributed approximately around the same
average, but in this case the variation presented is larger, around 2 units. This factor
corresponds to 14.2% of the dam movement response;
 The third factor presents a change, with an increase in the average, between 1992 and
1998, after which it keeps it constant until approximately 2008, showing a slight
increase as of this year. This factor corresponds to 10.5% of the dam movement
response;
 The fourth factor shows a decay in the average until approximately 1999, after which
it presents a more stable behavior. This factor corresponds to 7.3% of the dam
movement response;
 The fifth factor initially shows an increasing trend in the average until about the year
of 1999, and soon after that this growth converges to decay until approximately the year
of 2011, after which the average becomes more stable. This factor corresponds to 5.2%
of the dam movement response;

6
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

 The sixth and last factor presents a behavior different from the others, representing a
greater irregularity in the average, with some high and low peaks. This factor
corresponds to 3.6% of the dam movement response.
From the factorial loads, and with the aid of the R Software , scores were generated for each
factor, which were entered as input data in the IMCRBi, together with the eigenvalues
determined in the correlation matrix. Thus, the IMCRB series containing 312 values was
generated, whose graph showed a trend (Figure 3a), that the series was non-stationary. To make
it stationary, the series was differentiated again and its graph was constructed (Figure 3b).

Figure 3a: IMCRB chart Figure 3b: Differentiated IMCRB chart

SOURCE: OWN AUTHORSHIP SOURCE: OWN AUTHORSHIP

The IMCRB series showed seasonal behavior, which is explained by the behavior of the
dam structure during the year. This is represented by the high peaks, noted in the winter months,
and lows, noted in the summer months.
In addition, the data presented a tendency of decay of the average during the whole period
observed, but from approximately the year of 2003, this decay becomes less accentuated. The
justification for this fact may represent that in the course of time, the responses of the dam
structure were differentiated, presenting a change in the average, possibly caused by the
accommodation of the structure and by the inexistence of specific factors.
From the data of the different IMCRB and IMCRB, control and alert limits were determined,
so that the data had 95% and 99% confidence.
Through the Statgraphics Software, the ARIMA (2,1,2) model was stipulated for the
IMCRB series and the ARIMA (2,0,2) model for the differentiated IMCRB series. Then, 24
future values of each of the series were predicted and it was identified that both the model data
and the stipulated forecasts were within the confidence limits. It was soon expected that the
actual data from the second group series would also be within these limits.
For the validation of the method, the scores for the data factors of the second group, which
correspond to the measurements from January 2015 to December 2017, were initially stipulated
and used as input data for the IMCRBi, generating a series of updated data. The same was done
with the differentiated scores of this series.
With the series, two control charts were generated, containing the data from the first group
and the confidence limits. In addition, the data of the second group, which are highlighted in
Figure 4, were plotted on the same control charts.

7
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

Figura 4: Control charts for IMCRB and Differentiated IMCRB

SOURCE: OWN AUTHORSHIP

It is observed that the data of the second group, used for the validation of the method, was
adjusted almost entirely within the control limits, and in their totality within the alert limits, for
the values of the IMCRB. This indicates that the process is under control.
The IMCRB chart shows a tendency, which is possibly related to the strong influence that
Factor 1 exerts on the index, representing 49.3% of the total. However, from the middle of the
timeline, this trend tends to become smoother, presenting greater stability around a mean. This
shows that measured displacements become increasingly predictable and balanced.
Both high and low extremities are shown on the IMCRB chart, occurring with some
frequency. The upper extremities refer to the measurements that occurred in the winter months,
and the lows referring to the measurements that occurred in the summer months. In this way,
it is perceived that the temperature variation interferes with the movement of the dam.
The series for the distinctive IMCRB was almost complete within confidence limits.
Atypical points are observed on isolated dates, which did not interfere with the overall behavior
of the structure, and according to Oro (2016) may be the result of extreme environmental
conditions, faulty or maintenance instrument, measurement errors, or instrument calibration .
The model described in this paper represents satisfactorily the responses of the F19 and F20
block movement of the Itaipu dam Excavation F, and contributes to the interpretation of the
dataset referring to the measurements of the installed instrumentation.

4 CONCLUSION
This paper presented the reformulation of the Joint Monitoring Index of the Dam Blocks
Response (IMCRB) for the data of blocks F19 and F20 of Excerpt F of the Itaipu Dam, and the
due interpretation of its results.
The method used allowed to identify that the instruments involved in the research, being:
direct and inverted pendulums, extensometers, extensometers, and piezometers, are appropriate
to make predictions about the dam responses. In addition, it can be seen that the instruments
correlated with factor 1 are more important in relation to the dam responses.
Through the IMCRB it was observed that over time the data regarding the instrumentation
measurements of these blocks tend to show stability around a mean, showing that the measured
displacements become more and more predictable. In addition, the IMCRB data show
seasonality, with high and low peaks, representing the temperature effects of winter and
summer seasons.
Data from the series of the second set of data, for the period from January 2015 to December
2017, were distributed within the confidence limits stipulated at the beginning of the process,
indicating that the measurement values are predictable and that the process is stable.

8
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

The IMCRB proved to be an adequate alternative for the monitoring of the instrumentation
of concrete dams, being a method that returns reliable and fast results, facilitating the agility in
decision making in case of anomalies.

ACKNOWLEDMENT

I thank God first for the gift of life and for my family for their support and encouragement.
To the PTI - Ceasb for the opportunity, grant of the research grant, and also for the provision
of the necessary data for its accomplishment.
To my counselor, Professor Sheila Regina Oro, for her doctrine, readiness for attendance,
patience, encouragement, and for believing in my potential. The realization of this work would
not be possible without its referrals.
To the Federal Technological University of Paraná - UTFPR for the availability of space for
the guidelines.
I am also grateful to the Federal University of Paraná - UFPR, and to the teachers who teach
in it, through the knowledge acquired during classes in Numerical Methods specialization,
which helped to carry out the research.

REFERENCES

Agência Nacional das Águas (Brasil). Guide And Forms For Dam Safety Inspections / Agência
Nacional Das Águas. -- Brasília: ANA, 2016.

CEASB. Centro de Estudos Avançados em Segurança de Barragens. Disponível em: <


https://www.pti.org.br/pt-br/ceasb>. Acesso em: 02 jan. 2018.

GIL, A. C. How to design research projects / Antônio Carlos Gil. - 4.ed.- São Paulo: Atlas,
2002.

ITAIPU BINACIONAL. Itaipu Binacional: The Largest Generator of Clean and Renewable
Energy of the Planet. Disponível Em: < https://www.itaipu.gov.br/>. Acesso Em: 04 jan. 2018.

LATTIN, J.; CARROLL, J. D.; GREEN, P. E. Multivariate Data Analysis. São Paulo: Cengage
Learning, 2011.

MSIB. Manual of Safety and Inspection of Dams. Brasília: Ministério da Integração Nacional,
2002. 148p.

NETO, A. C. Time Series Analysis: Class notes. Foz do Iguaçu. 2017.

NETO, A. C.; CHAVES, I. A. Statistical Techniques Applied to Quality Engineering: Class


Notes. Foz do Iguaçu. 2017.

9
Luana E. Felippi, Sheila R. Oro, and Cleonis V. Figueira.

ORO, S. R. Structural Behavior Monitoring Index of the Concrete Blocks of Dams - A Multi-
Purpose Approach / Sheila Regina Oro. – Curitiba, 2016. 139 f.

SETTI, A. A.; LIMA, J. E. F. W.; CHAVES, A. G. M.; PEREIRA, I. C. Introduction to Water


Resources Management. 2. Ed. Brasília: ANEEL/ANA, 2000. 207 p.

SILVEIRA, J. F. A. Instrumentation and behavior of foundations of concrete dams / João


Francisco Alves Silveira. – São Paulo: Oficina de Textos, 2003.

ZUCULIN, S. Dams Safety Course: Module II: Inspection and Auscultation of Dams: Unit 2:
Instrumentation of Dams. 2012. 71 p.

10

You might also like