Sei sulla pagina 1di 9

Q. 78 R. ?

C OMMISSION INTERNATIONALE D ES GRANDS B ARRAGES ------VINGTIME CONGRS DES GRANDS BARRAGES Beijing, 2000 ------WINSCAR DAM 24 YEARS OF MANUAL SEEPAGE MEASUREMENT REPLACED BY AUTOMATION (*) Ian CARTER All Reservoirs Panel Engineer, MWH Ltd James Robert CLAYDON All Reservoirs Panel Engineer, Yorkshire Water Services Ltd Andrew Charles ROBERTSHAW Asset Manager - Headworks, Yorkshire Water Services Ltd
GREAT BRITAIN

1. INTRODUCTION Winscar Reservoir is owned by Yorkshire Water Services and is used for public water supply since 1975. It is located in the English Pennines and was the first large dam in Great Britain to have an upstream facing of asphaltic concrete. The re servoir has a surface area of 47 hectares and has a capacity of 8.3 Mm3. It is impounded by a 520 m long rockfill embankment dam that has a maximum height of 53m. There is a grout curtain at the upstream toe with a concrete toe wall through which the reinforced concrete outlet works pass. The membrane consists of two layers of asphaltic concrete 80 mm and 40 mm thick. Further details of the design can be found in Collins & Humphreys (1974). The geology at the dam site comprises a sequence of sandstones and shales of Upper Carboniferous Age, which dip toward the left abutment as shown in Figure 1. A permeable sandstone formation known as the Beacon Hill Flags crops out in the valley floor. Shales underlie most of the remainder of the foundation while another permeable sandstone formation, the Huddersfield White Rock , is present at high level on the left abutment.

(*)

Le Barrage de Winscar 24 ans de mesure manuelle dinfiltration remplac par lautomatisation


1

Q. 78 R. ?

Figure 1 Winscar Dam Geological Section Barrage de Winscar Sction Geologique

Shales Beacon Hill Flags (sandstone) Shales Huddersfield White Rock (sandstone) Grout curtain

Schistes Beacon Hill Flags (grs) Schistes Huddersfield White Rock (grs) cran dinjection

Winscar Dam was provided with many facilities for monitoring performance. Ground water pressures and flows, internal stresses and strains, and surface settlements and displacements were measured during construction and first filling of the reservoir. However the measurement of seepage has remained the principal indicator of satisfactory dam performance. Any seepage through the dam is collected by a series of drains that were dug into the foundation. The drains extend from positions close to the upstream toe wall as far as the downstream toe of the dam, where they are piped into measuring chambers. The original drainage system comprised two branches, as shown on Figure 2, with three measurement stations on the right-hand side and a fourth on the left-hand side. These branches combine before passing through a fifth measuring chamber from which it then discharges to the river downstream. Seepage collected by the underdrainage system arises from one or more of the following sources: reservoir water passing through defects in the asphaltic concrete deck flow past the grout curtain and out through the foundation and relief wells rainwater percolating through the permeable downstream shoulder

Q. 78 R. ?

groundwater from the abutments.

Figure 2 Seepage measurement layout Disposition de measure dinfiltration

Draw off culvert and gallery Line of toe wall and grout curtain Underdrain Measurement chamber Dam centreline Reservoir area

Ouvrage de restitution des debits Axe de mur dorteill et cran dinjection Drainage de fondation Chambre de mesure Axe de barrage Zone de reservoir

2. 2.1
FLOWS DURING FIRST FILLING

SEEPAGE FLOWS

Figure 3 shows a simplified summary of the total flows since the end of construction. Data was gathered daily during the early years of the reservoir but the frequency was reduced in the 1980s to once per week. The current record is discontinuous between 1984 and 1991. The clear relationship between reservoir level and flow is shown in Figure 3. This chart also illustrates the problems experienced during the late 1970s, where a pattern of rising seepage was recorded as the reservoir filled for the first time. The flow increased significantly in early 1977 when the reservoir rose above the level of the Huddersfield White Rock . The flow reached 2,500 m3/day at 333 metres above Ordnance Datum (mOD) and the projected flow would have been about 6,000 m3/day at the Full Supply Level (FSL) of 343.8 mOD. The

Q. 78 R. ?

water level was held down to permit grouting but although the grouting appeared successful, the flows continued to increase. Flows above 3,000 m3/day were recorded during 1979 despite the reservoir level remaining static at 333 mOD.
6000 2 5000 330 350

4000 1 3000

310 2 290

2000 1 1000

270

250

0 Jan-75

Jan-80

Jan-85

Jan-90

Jan-95

230 Jan-00

Figure 3 Total seepage flows since construction D'infiltration d'coulements totale depuis construction

Flow (cubic metres per day) Reservoir level (metres OD)

coulement (mtres cubes par jour) Rservoir de niveau (mtres OD)

Further remedial grouting carried out in 1979 led to a reduction in flow to about 900 m3/day at 332 mOD but the situation remained unsatisfactory. Dye tracer studies were undertaken to investigate leakage flow paths and chemical analyses were carried out on water samples. These latter studies concluded that a defect was probably present in the asphaltic concrete deck. The reservoir was subsequently emptied and a hole was found at about 305 mO D, as described in Routh (1988). The reservoir was refilled after the repair of the membrane. A marked improvement was noted throughout 1981 such that the flow did not exceed 1,500 m3/day at a reservoir level of 338 mOD. However despite this success the situation deteriorated again in subsequent years.

2.2

SEEPAGE FLOWS DURING RECENT OPERATION

The data since 1991 indicates a broadly deteriorating pattern with total seepage flow at Full Supply Level (FSL) rising from about 1,000 m 3/day in 1993 to over 2,000 m3/day two years later. The droughts in 1995/6 and major reservoir drawdown disrupted the pattern, and the leakage was reduced. Remedial works
4

Q. 78 R. ?

on the asphaltic deck were also undertaken during this period and this may have improved the overall watertightness. However since that time the flow has risen again to over 2,000 m3/day with the reservoir at FSL in July 1998. Normalisation of the data to remove the effect of reservoir level (and therefore driving head) indicated a variable response in the flow characteristics with time. The analysis showed that there was an increase of about 75 m3/day per metre rise in reservoir level during the refilling stages in the late 1970s. The response remained relatively stable after completion of the remedial works and 3 ranged between 20 and 40 m /day per metre rise in reservoir level until September 1993. The 1994 refill signalled the start of a period of erratic responses. Since that time the rate has varied between 10 and 70 m3/day per meter rise. The seemingly unpredictable response in total underdrainage flow necessitated detailed examination of the various component parts of flow.

2.3

ANALYSIS OF CURRENT SEEPAGE PATTERNS

The drainage zones within the system are unequal in size and details of the various sub-catchments are outlined in the table below.
Chamber 1 2 3 4 5 Total Note: Asphaltic membrane area m 2 3,000 2,500 3,000 15,000 1,500 25,000 BHF Beacon Hill Flags Foundation surface area, m 2 and geology 5,500 S 5,000 S 5,000 BHF / S 27,000 BHF / S 3,500 S / HWR 46,000 S Shales Catchment area on downstream face m 2 2,500 2,500 2,000 12,000 2,000 21,000

HWR - Huddersfield White Rock

Figure 4 illustrates the typical pattern of behaviour and trends over the three-year period prior to 1998. The records are characterised as follows: Irregular contributions from the two main zones with the relative proportions varying from #4 behind twice that of #3 to almost equivalent. Spikes (i.e. rapid and temporary fluctuations inotherwise normal behaviour) indicating deviations from the overall trend line by as much as 10%. The spikes in the flow record probably result from direct rainfall over the permeable downstream face. The effect varies depending not only on the intensity of precipitation and antecedent rainfall but also on the timing of flow measurement, relative to the rainfall event. The flow into zones #1 and #2 on the right abutment is minimal and accounts for less than 2% of the overall flow. This reflects the relatively impermeability of the shales forming the foundation to this part of the dam. The bulk of all leakage under and through the embankment is concentrated within zones #3 and #4. The relative contribution from each compartment appears to vary with reservoir level. When the reservoir level is high the two zones each
5

Q. 78 R. ?

contribute about 40% of the total flow. However when the level is drawn down, as it was in 1996, both zones produce less flow but the relative proportion from
2000 2 1600 330 350

1200 1 800
#4 #3

310 2 290

400
#5 #1+2

270

0 Jan-95

Jan-96

Jan-97

250 Jan-98

#3 drops to about 25% while #4 contributes the entire balance. Figure 4


Individual seepage flows since 1991 D'infiltration d'coulements individuelle depuis 1991

Flow (cubic metres per day) Reservoir level (metres OD)

coulement (mtres cubes par jour) Rservoir de niveau (mtres OD)

The contribution from zone #3 could be considered abnormal because of the relative size of the compartment. The effective catchment is much smaller than zone #4 and there is concern that a new concentrated leak may be present in the membrane above zone #3. However it is also possible that there may be some hydraulic interconnection between the two drainage compartments. The response of flow to changes in reservoir level is more pronounced in zone #3 but there is a marked lag in the recovery. This phenomenon might be due to flow through defects in the membrane since groundwater flow through the foundation might be expected to respond without undue delay. A possible explanation of the recent response is that a defect in the membrane beca me enlarged by erosion in the period up to 1993. Stress changes brought about by the 1995/1996 drawdown may have led to deformation of the asphalt resulting in some closing of the defect. However the sustained period of loading under high reservoir levels thereafter may have reversed the healing process in the asphalt.

Q. 78 R. ?

The flow from zone #5 on the left abutment is clearly dependent on reservoir level. It contributes up to 20% of the flow when the reservoir is spilling but reduces to zero below 325 mOD. A similar response is seen with flow into the New Spring area on the left bank downstream of the dam. These flows are clearly linked to groundwater flow, either around or through the grout curtain, via the Huddersfield White Rock. 3. SYSTEM DEFICIENCIES AND IMPROVEMENTS 3.1 PERCEIVED DEFICIENCIES

While the original seepage measurement system has proved robust and invaluable in monitoring the performance of the dam. Advances in data handling and instrumentation mean that its usefulness can be maintained and improved. A 1998 study identified a number of shortcomings in the seepage monitoring system. The relationship between rainfall and seepage flow is poorly understood and weekly rainfall data are inadequate when instantaneous flow measurements are recorded. In addition, man entry into the chambers may be potentially hazardous to personnel because of their confined nature. Another problem with the original system was that of inclusion of erroneous readings within the data. Such readings tend to arise because of systematic errors by different operators in the measurement, recording and transfer of data. Indeed a loss of data between 1984 and 1991 and a change from daily to weekly rainfall measurement has hindered the interpretation of the behavioural patterns. It was considered that a higher level of monitoring would be prudent, in view of the past history of the dam and the recent increase in seepage flow. The understanding of seepage behaviour was relatively poor, as was the ability to predict the timing and amount of flow. It was felt that the situation would be greatly improved by taking measurements more frequently and with greater reliability. The installation of an automated system was therefore proposed as this allowed continuous monitoring. Similarly, the correlation between seepage flows and rainfall would be enhanced if an automated rain gauge could be stationed. Complete rainfall data would allow the run-off response to be identified and permit the true seepage flow to be determined.

3.2

SYSTEM IMPROVEMENTS

The need for improvements to the measurement system was recognised in 1997. The original recommendations were examined in detail and the proposals were refined so that the following benefits might be provided. removal of th e need for man entry into deep, confined spaces continuous automated measurement and recording of data elimination of the need for manual data entry facilitation of data handling and analysis of in a timely manner ability to make direct comparison with rainfall potential to identify trends and dam performance more quickly
7

Q. 78 R. ?

It was not practical to bring about many changes to the underdrainage system as a whole. However it was felt that the construction of four new measuring chambers could improve the interception of flows, particularly in the case of drainage zones #4 and #3, which contribute the bulk of seepage entering the system. The new chambers provide closer resolution of the origin of seepage and also shed light on the apparent hydraulic cross connection between those zones above 295.5 mOD. The new chambers fitted with ultrasonic measuring devices to allow continuous monitoring of the flow over the vee-notches. This data is automatically transferred via telemetry at 15-minute intervals to the water utilities central database. An automated rain gauge measures rainfall at the dam and transfers this data to the same system. The improvements were completed in 1999 and the data collected so far has enhanced understanding of a complex phenomenon. The difficult question of whether or not such flows are acceptable remains to be answered. However the new facilities significantly improve the likelihood that adverse performance will be rapidly identified and that appropriate action will be taken promptly.

4.

SUMMARY

Winscar dam is a 53 metres high rockfill dam with an asphaltic concrete membrane. Seepage through the dam is collected into foundation drains that are piped to measurement chambers at the downstream toe. Seepage flows were recorded manually until 1999, and plotted against reservoir level and rainfall. On first filling the total seepage increased significantly when the water reached permeable sandstone. The reservoir was held down for grouting and then emptied for repairs to a hole in the membrane. By 1998 the seepage was 2,200 m3/day and although this was lower than during the early years, there was nevertheless a rising trend. Loss of data and a change in measuring frequency hindered interpretation of behavioural patterns. Improvements to the seepage measurement system were put in place in 1999 with four new chambers and improvements to four others. The new system incorporates automated measurement of rainfall and flows. Ultrasonic instruments measure head over vee-notches at 15-minute intervals, which are then sent via telemetry to the owners database. The original seepage measurement system has proved to be robust and invaluable in monitoring the performance of the dam. Advances in data handling and instrumentation mean that its usefulness can be maintained and improved.

RSUM

Q. 78 R. ?

Le Barrage de Winscar est un barrage d'enrochement de 53 m de haut avec une membrane de concrete bitumin . L'infiltration travers le barrage est collecte dans les drainages de fondations qui sont conduits aux chambres de mesures situes la bute avale. Les cours d'infiltrations ont t enregistrs manuellement jusqu'en 1999, et tracs contre le niveau du rservoir et la tombe d'eau de pluie. Au premier remplissage, l'infiltration totale augmenta considrablement quand l'eau a atteint le grs permeable. Le rservoir a t maintenu bas pour injection et par la suite vide pour rparation d'un trou dans la membrane. En 1998, l'infiltration tait de 2.200 m 3/jour et bien que ceci tait plus bas q ue les premires annes, il y avait nanmoins une monte de doutes. La perte d'information et un changement dans la frquence des mesures a gen l'interprtation du comportement. Les amliorations dans les mesures des infiltrations ont t mises en place en 1999 avec quatre nouvelles chambres et l'amlioration de quatre autres. Le nouveau systme incorpore la mesure automatique de l'eau de pluie et des coulements. Les instruments ultrasoniques mesurent le surplus au dessus des entailles 15 min d'intervalle, et envoient les details par tlmtrie la base de donnes du propritaire. Le systme original de mesures d'infiltration a prouv tre robuste et de valeur inestimable pour mesurer la performance du barrage. Les avances en traitement de donnes et d'instrumentation permettront que son utilit soit maintenue et amliore. REFERENCES COLLINS, P. C. M. & HUMPHREYS, J. D. (1974) Winscar Dam. Journal IWES, Vol. 28, pp 17-34. ROUTH, C. D. (1988) The investigation, identification and repair of the asphaltic concrete facing of Winscar Dam Trans. 16th Cong. ICOLD, San Francisco, pp 655-677. WILSON, A. C. (1988) & ROBERTSHAW, A. C. Winscar Dam: Investigations and repair to asphaltic concrete membrane Proc. 10 th Conf. British Dam Society, Bangor, pp 292-302.

Potrebbero piacerti anche