The embankment dam at Lövön in Sweden was built between 1972 and 1973. The dam is a zoned earth dam, 25m high with a core of moraine. In 1983 a sinkhole occurred on the upstream face followed by a wider sinkhole at almost the same place in 1986. After investigations in 1986 and 1989, it was concluded that internal erosion was taking place in the dam. Grouting was performed in 1992 in the bedrock and

a stabilising downstream berm was constructed in 1994.

Since 1983 several exploratory drillings have been made and weekly water measurements in the standpipes were recorded. In 1993 monthly temperature measurements were taken in order to locate seepage areas and detect seepage changes. In 1998, after examination of all results, the owner, Graningeverkens, decided to repair the dam.

During the summer of 1998 the upper part of the dam was excavated, and several examples of internal erosion, the result

of improper downstream filter and construction error, were discovered.

From the excavation level, a diaphragm wall was constructed in the old core and a sheet pile was driven in the slurry down to bedrock. Additional drilling and grouting was carried out in order to obtain a proper seal between the diaphragm wall/sheet pile and the bedrock. A new core of moraine, with proper downstream filters, replaced the excavated part of the dam.

Three monitoring sections were selected (in sections 0/035, 0/045 and 0/060) to monitor pressures and temperatures in

the foundation, the old core and the downstream part of the new dam. The monitoring system consisted of Geokon vibrating wire piezometers with integral vibrating wire temperature sensors, a data acquisition system, and a PC which was used for data analysis and presentation. Vibrating wire instruments were selected for their small size, ruggedness, proven long term stability and their ability to transmit readings over long cable lengths without signal loss or degradation.

Instrument cables were routed downstream, perpendicular to the dam axis, into the downstream filter, where they were gathered together and run to the data acquisition system. Instrument readings were recorded manually, until the data acquisition system was commissioned in January 1999.

The pressure measurements during the construction period were high, compared with the stationary pressure. The water level in the reservoir started to increase in August 1999 but was always lower than the measured pressure in the core. The high water level, and consequently also water content, resulted in an outflow of water into the downstream filter when the core was drained.

Measurements in section 0/045 are almost the same as in section 0/035, with high levels during the construction period that dissipate to the stationary level. The small variations in the bedrock, old core and downstream filter are also similar.

Only two piezometers were installed in section 0/060, one in the bedrock and one in the core. Their results are generally similar to those described but the pressures are slightly higher. This is expected in the core because the piezometer is installed just outside the sheet pile/diaphragm wall. The head in the bedrock is about 2.5m above the downstream water level, ie about 1m higher than in the other sections. The small variations which are seen to occur here follow the variations of the downstream water level. The sudden increase at the time of filling the reservoir is remarkable and is only observed in this section. The upstream water level has little influence on the pressures measured here which indicates that the bedrock is more permeable than in the other sections;

this is only to be expected as section 0/060

is located just outside the theoretical grouted area.

Temperatures measured by the piezometers exhibited stable values with daily variations of less than 0.1°C. A general decreasing temperature trend in the core was observed for the first months, until summer 1999, when the temperature started to increase. The linear decreasing temperature in the beginning was most likely a consequence of heat built-in during the summer, during a slow adjustment to the mean annual air temperature.

Temperature measurements in the bedrock indicate a significantly larger temperature variation in section 0/060 than in other sections. This indicates seepage somewhere in the bedrock. The seepage flow can be estimated from these data (amplitude and lagtime) and are found to be about 0.25mm/sec using the methods presented by Johansson (1997). Due to the heat conduction a thermal impact will occur outside the seepage zone. Seepage might be larger if the measuring point is within the seepage flow.

It can therefore be seen that the combined information from both temperature and pressure measurements provides useful information from which seepage flows can be estimated. Small seepage flow changes can also be detected before they affect the dam, thus improving overall safety.