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

net/publication/281618670

ROCKFILL DAM SETTLEMENT DATA: PROCESSING AND STATISTICAL

ANALYSIS

Conference Paper · January 2013

CITATIONS READS

2 490

4 authors:

Fjola Gudrun Sigtryggsdottir Jónas Þór Snæbjörnsson

Norwegian University of Science and Technology Reykjavik University
30 PUBLICATIONS   119 CITATIONS    131 PUBLICATIONS   881 CITATIONS   

SEE PROFILE SEE PROFILE

Ragnar Sigbjornsson Lars Grande

Norwegian University of Science and Technology Norwegian University of Science and Technology
135 PUBLICATIONS   2,115 CITATIONS    26 PUBLICATIONS   400 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Structural dynamics of an existing pontoon bridge: monitoring and simulation View project

Embankment dam safety under extreme loading conditions: Stability and breaching View project

All content following this page was uploaded by Fjola Gudrun Sigtryggsdottir on 09 September 2015.

The user has requested enhancement of the downloaded file.

 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

ROCKFILL DAM SETTLEMENT DATA: PROCESSING AND

STATISTICAL ANALYSIS

Sigtryggsdóttir F.G.1, Snæbjörnsson J.Th.2 , Sigbjörnsson R.3, Grande L. 4

1, Reykjavik University, Reykjavik, Iceland, [email protected]
2, Reykjavik University, Reykjavik, Iceland,
3, EERC, University of Iceland, Iceland / Norwegian University of Science and Technology, Norway
4, Norwegian University of Science and Technology (NTNU), Trondheim, Norway

Abstract: The operational safety of large dams is based on both visual inspections and measurements of
important parameters in order to monitor their actual behaviour. Monitoring dam deformations is one of the
main tasks in a dam safety monitoring program. Statistical analysis of the monitoring data provides means of
comparing response from many instruments. Furthermore, prediction models for early warning can be
constructed. Statistical analysis of the recorded data should thus be an integral part of a comprehensive safety
monitoring program. However, the monitoring data is often imperfect and has to be processed to fulfill the
necessary criteria for statistical analysis. The case studied in this paper is the 198 m high Kárahnjúkar
concrete faced rockfill dam (CFRD) in Iceland. The instrumentation of the dam with respect to monitoring
settlements is described as well as the processing of the settlement data. The data presented extends from the
installation of each instrument during the construction of the dam, the first impounding of the reservoir and
through five cycles of operation. The data processing aims at producing time series that can be used for
statistical analysis of the deformation behaviour of the dam and its effect on stresses and strains in the
concrete face slab.
Key words: Rockfill dam, CFRD, settlement, statistical analysis, data processing .

1 Introduction

Monitoring dam deformations such as settlements is one of the main tasks in any dam safety
monitoring program. Statistical analysis of the monitoring data provides means of comparing
response from many instruments. Furthermore, prediction models for early warning can be
developed. The utilization of statistical methods should thus be included in a comprehensive safety
monitoring program. However, often the monitoring data is imperfect and has to be processed to
fulfill necessary criteria required for the statistical analysis. This may especially apply for the
settlement monitoring data which often is collected at irregular time intervals. The processing of
the monitoring data is not only essential for the statistical analysis of the settlement itself but also
for statistical investigation on the effects of the settlement behaviour on other physical factors
monitored within the dam. One consideration is the time interval or step between data points and
the other removal of outliers and smoothing deviations in the original reading on the sensors.
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

The case studied in this paper is the Kárahnjúkar concrete faced rockfill dam (CFRD) in Iceland
(see fig 1). The instrumentation of the dam with respect to monitoring settlements is described as
well as the processing of the settlement data. The monitoring data presented extends from the
installation of each instrument during the construction of the dam, first impounding of the reservoir
and through five cycles of operation. The data processing aims at producing time series that can be
used for statistical analysis of the deformation behaviour of the dam and its effect on stresses and
strains in the concrete face slab.

Hálslón

Reservoir

Kárahnjúkar Dam

Spillway chute

Owner: The National Power Company of Iceland, Landsvirkjun.

Designer: Palmi Associates as subcontractor to Montgomery
Watson Harza both of USA members of the design team
Kárahnjúkar Engineering Joint Venture (KEJV) headed by
Verkís Consulting Engineers (then VST) of Iceland.
Contractor: Impregilo of Italy Fig. Landsvirkjun

650

600
m a.s.l.

550
Hálslón Reservoir Waterlevel
500
Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time(Date)
Figure 1 Kárahnjúkar dam and variation in reservoir elevation

2 The Kárahnjúkar CFRD

The Kárahnjúkar dam (fig 1.), is a 198 m high and 700 m long CFRD. It is one of three dams
retaining the main storage of the Kárahnjúkar Hydropower Plant (KHP), located in Iceland. The
dam is constructed of palagonite tuff (moberg) rockfill with dam zoning based on State-of-the-Art
practice at the time of design and construction. The main features of the dam are described in [1].
Assessment of the dam’s settlement is presented in [2].

3 Instrumentation monitoring settlement

Instrumentation of the Kárahnjúkar dam with respect to monitoring settlement includes Hydraulic
Settlement (HS) gauges within the dam body and benchmarks at the dam crest and on the
downstream side which are surveyed regularly. Additionally, settlement points on the upstream
side were surveyed during construction and impounding while above water.
The purpose of the settlement instrumentation according to the Designer [3] is to: (1) Monitor
settlement of the embankment fill to compare with design estimates. (2) Allow modulus of the fill
to be calculated to compare with design estimates. (3) Allow revised analysis with the modulus
based on the measured settlement to reevaluate concrete face deflection.
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

The HS gauges are used to monitor the settlement of the dam body. Both instruments from
Geokon (Model 4650 VW Settlement Sensor) and Sisgeo (Model DSM D422F) are used, and in
some locations a second reduntant settlement gage was installed [3,4]. The settlement system is
designed to measure the differential settlement between two points, a reference point and the
settlement point within the dam body where the HS sensor is located. A liquid filled vessel is
located at the reference point and connected to a sensor located at the settlement point by
liquid-filled tubes. The settlement of the sensor within the dam fill is proportional to the liquid
pressure change. [5,6,7] The two sensor models are described in more detail in Section 4.

Figure 2 Hydraulic settlement (HS) gauges and Terminal Structures (TS): Locations.

The HS gauges were installed at three sections through the dam, labelled A, B and C. Their
location is shown on fig. 2. The HS sensors are located at approximate elevations (El.) 545 m a.s.l
and 580 m a.s.l. at sections A and C whereas at section B those still operating are approximately at
El. 510 m a.s.l. and 545 m a.s.l.. The vessel, and thus the reference points for the HS gauges are
provided in so called Terminal Structures (TS) located at the downstream side of the dam (see
fig.2). Depending on location, three to six HS sensor are provided in a line at a level somewhat
lower than the vessel in the TS. This provides a local settlement system measuring the differential
settlement between respective HS sensors and the reference vessel. To obtain the total settlement
with respect to a fixed and stable reference point (benchmark point) the geodetic position of the TS
is required.

4 Instrument readings and settlement calculations

Surveys to determine the geodetic position of the TS were conducted once or twice a week during
construction of the Kárahnjúkar dam and during the first impounding. During the last 30 m of the
first filling (FF), surveys took place with 1-4 days interval. After the first impoundment was
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

complete the surveying intervals were extended to about once every month during the winter and to
a semi-monthly interval during the following summer/fall. After the first year of operation the
interval between surveys to establish the geodetic position of the TS has gradually increased with
these being conducted three times a year (spring, summer and fall) over the last three years.
However, the readings on the HS stations have generally been more frequent and for the past three
years about 5-6 times a year.
Readings on the HS stations are taken manually by a readout box from the respective
manufacturer [3]. A calibration or sensitivity factor, supplied with the sensor, has to be used to
convert the reading to engineering units relating to settlements. The readings on the two models
used differs as one uses a pressure transducer sensor to read the liquid pressure while the other uses
a vibrating wire (VW) pressures sensor. However, the conversion to engineering units and
calculation of the settlements is essentially the same for both models although the presentation of
the results is different as accounted for in the following.

4.1 Liquid pressure measured using vibrating wire (VW) pressure sensors

The Geokon settlement system consists of a VW sensor (model 4650), located in the dam fill,
which is hydraulically connected via two tubes to a reservoir, i.e. a liquid filled reference vessel
located in the TS [3]. The VW sensor senses the liquid pressure in the tube and thus provides a
measure of the hydraulic head of liquid indicating the elevation difference between the sensor and
the vessel. The two liquid filled tubes allow flushing to remove any air bubbles that might form.
Furthermore, a third tube, a vented cable, runs from the sensor to the vessel so that barometric
pressure fluctuations do not affect the readings (fig. 3).

Figure 3 Typical installation of vibrating wire settlement system (Geokon) (fig from [5])

The change in the vessel (reservoir) liquid level is measured visually using a scale, whereas the
sensors are read using a Geokon read out model GK-403 [3], which displays directly the period
(digit) of the vibrating wire which can be converted to micro strains using the formula below:

E=Eo – (R1-Ro) G – ERES (1)

Here Eo is the sensor elevation at installation, ERES is the change of the fluid level in the reservoir,
Ro is the initial sensor reading, R 1 is the subsequent sensor reading, G is the calibration factor
supplied with the sensor. [5]

4.2 Liquid pressure measured with pressure tranducer

The Sisgeo settlement system consists of a series of level sensors (model D422F) within the dam
body, which are hydraulically connected to a liquid filled reference vessel (model D422S080)
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

located in the TS [3,4]. The liquid pressure at each sensor location is read by a pressure transducer
that is referred to the liquid level in the reference vessel which is read by a load cell. The liquid
pressure provides measure of the elevation difference between the sensors and the vessel.
The instruments in the Kárahnjúkar dam are read manually using a Sisgeo readout model
C6004VW, displaying 4 to 20 mA reading [3]. Readings are taken both on the HS station within
the dam body and at the reference vessel. The differential settlement of the sensor is calculated
from the following formula:
D = (L-Lo)/S (2)

Where D is settlement in mm; L is current reading in mA, Lo is the initial reading in mA and S is
the sensitivity factor from the compliance certificate, mA/mm
To obtain the differential settlement of each HS station within the dam relative to the reference
vessel the following expression is used:

H = (D-Do)dam – (D-Do)Ref (3)

Where D is the current settlement measure, Do is the initial or “zero measure”, (D-Do)dam is the
difference between D and Do for the HS station within the dam., (D-Do)Ref is the difference between
D and Do or the reference vessel inside the TS.[6]

4.3 Temperature corrections

The manufacturers of both instruments describe in their manual how the measurement can be
corrected for temperature variations, however, they also state that these are insignificant, especially
for buried sensors, and can be neglected in most cases. For the instruments in the Kárahnjúkar dam
temperature variations are modest since the sensors are buried and the reference vessel is enclosed
in the TS. Thus no temperature correction has been done for the monitoring data presented here.

4.4 Absolute settlement

As mentioned before, the terminal structures (TS) where the reference vessels for the differential
settlement sensors are located, are settling with the dam. Geodetic surveying of a reference point on
the TS is therefore required to monitor the continuing movements of the TS, which needs to be
accounted for to obtain the absolute settlement of each HS station.
When evaluating the absolute settlement from the raw instrument readings using the formulas
above, it is important to take note of all changes that might have occurred during the construction
of the dam. This could include cutting of cables and tubes which would result in new setting of
initial value (Ro or Lo) and potential relocations or late installation of reference vessels. It is
essential that all such changes are carefully registered by the person reading the instruments.
In summary to obtain the absolute settlement at each HS station the following measurements are
combined: (1) Readings on the sensor within the dam body; (2) Readings of the reference vessel
liquid level; (3) Geodetic survey on the Terminal structures

4.5 Example of instrument readings

Fig 4 provides examples of instrument readings from both Sisgeo and Geokon sensor monitoring
settlement at location HS-A-6. The Sisgeo instruments are denoted HSA6s at this location and
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

those from Geokon HSA6g. The readings on the Geokon reference vessel are stable compared to
readings on the Sisgeo reference vessel, which the authors believe can be explained by the effect of
the vent tube provided as a part of the Geokon system to remove barometric pressure effects. The
steps referred to in legends on the figures are discussed in Section 6.
Instrument HSA6g Geokon Instrument HSA6g

Reference Vessel Readings (m)

9100 0.08
Step 1
Sensor Readings (digit)

Step 2 and 3
9000

8900 0.075

8800
Step 1
Step 2
8700 0.07
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12 Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date) Time (Date)

Instrument HSA6s Sisgeo Instrument HSA6s

8.5 13

Reference Vessel Readings (mA)

Step 1
Sensor Readings (mA)

Step 2
12.5

12

Step 1
Step 2 and 3
7.5 11.5
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12 Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date) Time (Date)

Figure 4 Instrument readings for location HS-A-6.

5 Criteria for statistical analysis

Statistical analysis of time series requires: (1) All the time series must have the same number of
data points, n; (2) The time series must all start at the same point in time and end at the same point
in time; (3) The time interval between data point i and data point i+1, where i=1,2,…n, must be the
same for all the time series. However, the time interval does not necessarily have to be the same
between all data points. If these requirements are not fulfilled the time series needs to be rearranged
or processed to meet these criteria. When all the relevant time series have been processed to fulfil
these conditions they can be analysed statistically. For example the correlation between individual
time series can be calculated, a principal component analysis can be conducted and a multivariate
regression performed provided that appropriate predictor functions have been defined. Also,
standardization (normalization) of the data becomes straight forward
Time series with uneven time step can be analysed and compared as outlined above, however
regenerating the time series with a fixed time step, provides further possibilities for data analysis.
The settlements of the Kárahnjúkar dam are measured manually and with varying time intervals (1
to 114 days), while some other physical factors are read automatically with a fixed interval in time.
Instruments to monitor strain in the concrete face slab of the Kárahnjúkar dam are for instance
collected automatically every day. The settlement of the dam induces stresses and strains in the
concrete face slab. To investigate this interaction between the concrete face slab and the rockfill a
multivariate analysis could be applied which preferably requires a settlement time series with the
same time step as the time series for the strain.
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

6 Processing of monitoring data for statistical analysis. Fixed time-step.

At the beginning of this study of the Kárahnjúkar dam settlement it was anticipated to use the
absolute settlement time series obtained directly from an automated data processing system
(ADPS). However after these had been reviewed and used in preliminary analysis it was clear that
there were unexplained discrepancy in the data compared to time series obtained during the
impounding phase prior to the establishment of the ADPS (termed original on figures below). Thus
the raw data was requested from the Owner which readily submitted the data. It has proven vital for
the statistical analysis performed hitherto to be able to scrutinize and process the raw readings and
geodetic surveys rather than to have only the absolute values calculated by the ADPS to work with.
The processing of the settlement data is outlined below defining 4 steps in the processing of the
data to obtain time series with fixed time step (1 day), smoothing of deviations and same starting
and end points in time. Steps 1 to 3 are divided in part a) and b). In part b) of these Steps the
settlement is calculated so that these can be compared at all Steps. The data processing was
conducted using the software Matlab.
1
TS A1 EoC
Section B
TS A2 Section C
Section A
0.8 TS B2
TS B3
TS C1
Settlement (m)

0.6 TS C2

First
0.4 IMP

0.2
End of first impounding

0
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date)

Figure 5 Terminal structures survey data (dots) and interpolation (line) between survey data
smoothing irregularities in the survey (applied from about EoC End of Construction). (IMP: impounding)
Instrument HSA5g Instrument HSA6g
0.66 1.12

1.08
Settlement (m)

Settlement (m)

0.62

1.04

0.58
1
Original TS Original TS
Step 1 Step 1
0.54 0.96
Oct07 Oct08 Oct09 Oct10 Oct11 Oct12 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date) Time (Date)

Figure 6 HS time series obtained from both original TS and corrected TS data (Step 1).

6.1 Step 1: Processing the geodetic survey data on the TS

Results from a geodetic survey on the TS are presented in fig 5. As mentioned before the readings
on the HS stations are more frequent than the geodetic surveys on the TS. The original processing
of data from the HS settlement gauges by the ADPS (termed Original TS on fig. 6) used the last
surveyed geodetic TS position as reference value for all HS readings taken between TS surveying
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

intervals. This results in loss of details in the settlement behaviour. Interpolation between actual
surveying data points on the TS structures is required to obtain values applicable for calculating the
settlement time series for each HS station. The dots on fig 5 are the surveying results, from which
the improved data set is constructed, represented by continuous lines on the figure. The
reconstructed data set has been generated by reducing the influence of outliers, linear regression
between actual data points and smoothing of deviations. The smoothing operation is primarily used
for the data obtained after the end of construction (EoC). The smoothing procedure is only accepted
after a cautious comparison with the original data to validate the adjusted curve. Part a) of Step 1
(Step 1-a) is the reconstruction of these time series of the TS settlement while part b) of Step 1
(Step 1-b) is the utilization of this in calculating the absolute settlement of the HS stations. Fig. 6
shows the settlement time series at Geokon stations HSA5g and HSA6g using both the original and
the adjusted/corrected TS data obtained after Step 1 (both part a) and b)). Clearly, both correction
and interpolation of the TS values is2 required to observe the details in the settlement behaviour.

2
Step 1 a and b FF
EoC HSB EL. 510
Settlement 1.5
HSB EL. 545
calculated using the HSB 11
HSA EL. 545
corrected TS 1.5 HSB 17g
HSA EL. 580
curves. HSC EL. 545
1
Settlement (m)

HSC EL. 580

Time interval varies
from about 1 day to 1
114 days
0.5
After Step 2 a and b
with linear 0.5
interpolation the
0
plotted curves look Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
exactly the same.
0
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date)
2

2
Step 3 a and b FF
Linear interpolation EoC HSB EL. 510
1.5
HSB EL. 545
between instrument HSB 11
HSA EL. 545
readings to obtain 1.5 HSB 17g
HSA EL. 580
data point for each HSC EL. 545
1
Settlement (m)

day and smoothing HSC EL. 580

of the sensor
1
readings (not
reservoir readings) 0.5
approximately from
the EoC (varies). 0.5

0
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12

0
Oct05 Oct06 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date)

Figure 7 Settlement at HS stations (EoC: End of Construction; FF: First filling)

Steps in the data processing.

6.2 Step 2: Time series with fixed time step

All the HS settlement time series are shown on fig. 7 demonstrating to some extent the effects of
Step 1 to 3 in the data processing. On fig. 8 the same steps for stations at location HS-A-5 and
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

HS-A-6 (HSA6s and HSA6g) are shown in more detail since the first filling (FF). Step 2 is
discussed in this subsection.
The absolute settlement is calculated from firstly the TS settlement and secondly the instrument
readings, R and L, for the Geokon and Sisgeo sensors respectively. With the time series for the TS
settlement curves processed as described above for Step 1-a, time series with fixed time step
already exists for the TS. The time step was here chosen to be one day.
Step 2, part a) (Step 2-a) is the removal of outliers in the readings if required followed by
interpolation between: Firstly, data points of the sensor readings (R or L) to obtain time series with
a fixed time step of one day. Secondly, data points of the readings on the reference vessel inside the
TS. Step 2, part b) (Step 2-b) is the calculation of the absolute settlement using the data from Step
1-a and Step 2-a.
Provided linear interpolation functions are used in Step 2-a, the time series obtained after Step
2-b should contain the same settlement data points obtained after Step 1-b. The difference is in the
additional data points generated at fixed time steps (one day or as specified) between these original
measuring points.
Instrument HSA6g Instrument HSA6s
1.12 1.14

1.08 1.1
Settlement (m)

Settlement (m)

1.04 1.06

1 1.02
Step 1 Step 1
Step 2 and 3 Step 2 and 3
0.96 0.98
Oct07 Oct08 Oct09 Oct10 Oct11 Oct12 Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date) Time (Date)

Instrument HSA5g
0.66

Legend
HSA5g Geokon sensor at location HS-A-5
Settlement (m)

0.62
HSA6g Geokon sensor at location HS-A-6
HSA6 Sisgeo sensor at location HS-A-5
0.58

Step 1
Step 2 and 3
0.54
Oct07 Oct08 Oct09 Oct10 Oct11 Oct12
Time (Date)

Figure 8 Settlement at locations HS-A-5 and HS-A-6 since the end of the first filling.

6.3 Step 3: Smoothing of deviations

After Step 2-a smoothing of deviations in the instrument readings can be considered for the sensor
and also for the reference vessel, this is referred to as Step 3-a in the data processing and Step 3-b
is the calculation of the settlement time series using the data from Step 1-a and the interpolated
time series from Step 2-a smoothed in Step-3-a.

6.4 Step 4: Extracting time series with the same start and end point in time

Step 4 involves selecting a suitable starting point in time for either all the settlement time series or
some group of time series that one indends to analyse statistically. Generally, the most suitable
 Hydropower 2013--CHINCOLD 2013 Annual Meeting
The 3rd International Symposium on Rockfill Dams

starting point would be close to the end of construction when the installation of all instruments,
sensors and reference stations, at their final location is completed and their operation tested.

7 Processing of monitoring data for statistical analysis. Varying time-step.

Statistical analysis can be performed on time series with uneven time steps provided the criteria set
forth in section 5 are fulfilled. If it is preferred to use the time series obtained in Step 1-b described
above the following should be kept in mind: (1) Manual readings on the HS stations might
generally be conducted the same day, however there may be missing data points that either need to
be interpolated from existing ones or all other measurements that day deleted from the time series.
(2) Select a suitable starting point in time for all the series. (In most cases this would be close to the
EoC). (3) Select one sound time series to compare to the others to confirm that the varying time
intervals between data points are the same for all the time series.

8 Summary and Conclusions

The original settlement curves presented (from the ADPS) are sufficient for the simplest form of
safety monitoring of the settlement, which was the purpose originally set for the design of the
instrumentation (see section 3). However, further processing of the monitoring data is essential for
both statistical analysis of the settlement itself and to investigate the effect of the settlement
behaviour on other physical factors monitored within the dam, such as strains and stresses in the
concrete face slab of a CFRD. For the data processing it is important to use the raw readings and
survey data in order to calculate the settlement at each measuring point within the dam.
Furthermore, the settlement time series must fulfill the criteria set for statistical analysis. The steps
taken to improve the processing of Kárahnjúkar dam settlement data have been outlined in this
paper and it is demonstrated how settlement time series with both fixed and varying time steps can
be constructed to fulfill criteria for statistical analysis. Examples of statistical analysis for the
Kárahnjúkar dam are given in [2].

Acknowledgements

This work is partly funded by Landsvirkjun's Energy Research Fund and Viðlagatrygging Íslands
The first author thanks Landsvirkjun for providing and allowing the usage of the monitoring data.

References

[1] Johannesson P, Perez H and Stefánsson B, Updated behavior of the Kárahnjúkar CFRD in Iceland,
23rd ICOLD Congress, Brasilia, Brazil June 2009
[2] Sigtryggsdóttir FG, Grande L, Sigbjörnsson R, Evaluation of the Kárahnjúkar dam settlement
behaviour using monitoring data, Proceedings of ICOLD 2013 International Symposium, Seattle 2013.
[3] Perez, H, Kárahnjúkar CFRD. Design Memorandum C12-Update, KEJV/MWH, Report LV-2009-107.
[4] Impregilo, KHP Contract KAR-11 Instrumentation Report, Dec 2005/Jan-Feb 2006; 2006
[5] Geokon, Instruction Manual-Model 4650 VW Settlement Sensor, 2013
[6] Sisgeo, Hydraulic settlement gauges. Profiler. Instruction Manual. (09/04, Rev.2, D422F,…)
[7] Sisgeo, DSM Settlement monitoring system; Data sheet, 2006

View publication stats