THE 15MW Guajoyo hydro power plant in the north east corner of El Salvador was built in the early 1960s and is one of the oldest hydro plants in the country. In the 1990s it became obvious that the plant, which has operated continuously since it was commissioned, could no longer maintain safe and reliable operation.

Rehabilitation work was undertaken in 2000 and included sealing of the seeping dam block joints, consolidation and contact grouting of the highly fissured concrete lining of the tunnel, the replacement of the turbine runner blades and the replacement of the stator winding to increase the unit output and efficiency.

Plant history

Guajoyo’s reservoir is formed by an old valley whose exit was blocked by volcanic flows and deposits in prehistoric times, creating the natural lake Lago Guïja. The lake water has cut a small discharge canal through the volcanic deposits which were blocking the lake exit.

Across this canal a dam was constructed to increase the active reservoir volume, and create adequate inflow conditions to the power tunnel. The dam is built on highly fissured and fractured volcanic deposits; the underground seepage across the dam is still substantial and can approach one-tenth of the rated turbine discharge when the reservoir is at maximum operating level. Despite various grouting attempts along the dam axis in the past, the seepage rate could not be substantially reduced. However, the dam stability is not affected.

The fairly large reservoir can be labelled yearly pondage, and is actually used as such. During the wet season (May to December) the plant is maintaining the maximum lake level and frequently operates in run-of-river mode. In the dry season (January to April) the plant operates predominantly as a peaking plant and gradually draws down the lake to the minimum operating level.

The discharge facilities at the dam structure comprise one bottom outlet, one gated spillway, one side channel spillway in the extension line of the main dam, and a fuse plug in an auxiliary dam some distance away from the main dam. Originally, all these structures together were capable of discharging some 2000m3/sec. During the planning and construction period, the spillway discharge areas were entirely uninhabited. However, in the last forty years, settlers moved into this area and a national highway was constructed inside the reservoir area, thus cutting off the fuse plug. Neither the function of the side channel spillway nor the fuse plug was ever tested. Of the original 2000m3/sec spill capacity now only some 700m3/sec can be considered safe spill capacity, far below the required discharge for a maximum probable flood event. In fact, the tropical storm ‘Mitch’ in November 1998 was a strong reminder to reconsider the spill facilities of the plant.

The relation between the power conduit length and the turbine head of the Guajoyo plant is about eight. Normally, when the relation power conduit/turbine head exceeds the value of four, a surge chamber is required to maintain waterhammer and speedrise of the generating unit after full load rejection within economic and technical limits. However, the topographic conditions of the Guajoyo plant excluded the construction of a surge chamber. For that reason, the water velocity in the tunnel had to be drastically reduced to about 1.4m/sec at rated conditions – with a corresponding increase of the tunnel diameter – to keep waterhammer and speedrise of the generating unit under control.

The original turbine installed at the project had a rated head of 42m and was designed for an unusual head variation from 27 to 54m. In the original planning stage it was thought that another plant would be erected downstream of Guajoyo to raise the tailwater. However, the construction of another plant was never realised, and from its initial operation the actual operating head of Guajoyo is in the range of 42 to 52m.

Rehabilitation works

The rehabilitation works were contracted in three lots: civil work (Salvadoran contractor), electromechanical work (Italian contractor) and supervisory and monitoring equipment (Spanish contractor). The civil works at the waterways were scheduled to take place entirely in the safe dry period January to April. Since the power plant approach canal is actually the dewatering outlet of Lago Guïja, the dewatering of the approach canal also blocks the natural outlet of the lake. This meant that the canal could only be dewatered in the dry period when the lake level is not expected to rise.

In order for work to begin, coffer dams were constructed across the approach canal and the tailrace and the waterways were drained. The essential rehabilitation works are listed below:

Concrete lined tunnel

The first visual inspection of the tunnel works showed very long continuous cracks in the unreinforced concrete lining. No further detailed investigations were made apart from studying the original geological drawings and grouting records. Based on the assessment of the first inspection, contact grouting only at a regular pattern was specified. During grouting it was discovered that at certain tunnel sections very large takes had to be injected. Consequently, at those sections core drillings were performed. The cores revealed highly fractured and disintegrated rock behind the lining. For that reason, consolidation grouting in those sections was ordered, almost doubling the cost for the tunnel lining rehabilitation. These measures were successful as after filling and emptying the tunnel, no sign of seepage could be observed.

Spillway stoplogs

One spillway vane is 4m wide and 7m high. The maximum thickness of the watersheet over the sill is about 4m. The vane is to be closed by four roller stoplogs each 4m wide and 1.10m high. The stoplogs were supposed to be set by a grappling beam and crane against the natural flow of water across the spillway. When the watersheet approached 4m, the grappling beam and the engaged stoplog could not be lowered any more – the set stopped halfway down in the flowing water.

Despite considerations about uplift, downpull, impulse and deflection forces, friction and streamlining, the actual water flow defied all those considerations. In the end only an increase of the grappling beam deadweight by about three times solved the problem, and with practice and experience the stoplogs could be set to the lowest position against the full flow of water.

Inundation of the plant

The original schedule specified completion of the waterways and closure of the intake gate and the draft tube gate by the end of April 2000, within the safe dry season. However, the works were delayed for a number of reasons. By the end of May 2000, the intake gate was set, the draft tube gates ready to be set within a few days. However, on the 3 June 2000 exceptional night rains in the distant mountains raised the water level of the downstream river course, overflowing the small tailrace cofferdam and flooding the powerhouse through the open draft tube to a level just below the generator winding. Consequently, divers were engaged, the draft tube sills were cleared, clearance confirmed, the gates were set and pumps started to attempt the dewatering of what was thought to be an hydraulically isolated power house. The measures were not successful and the water level inside the powerhouse did not drop, instead it moved with the variations of the tailwater level. After a careful study of all pipe connection between the power house and the tailwater, an open connection on that cause could be excluded. The attention then focussed on the draft tube gates. By precise measurement of the assumed closed position of the two draft tube gates, with a tape and reference dimensions taken from the drawings, it was concluded that the gates did not touch the sill, and that both gates were suspended askew. Divers were called in and discovered obstructions across the sill. After cleaning of the sill, the gates were set again.

Within one day, the power house was dewatered and normal work could be resumed.

Uprating and increase of efficiency

The new stator winding with modern insulation material permitted an increase of the copper section, a reduction of the copper losses, a reduction of the temperature rise, and an increase of the potential capacity. Thus, the efficiency in the upper part and full load range at rated power factor was increased from about 96.8% to 97.1%. The full load continuous operation could be raised from 18.3MVA to well above 21MVA.

The old turbine runner was still in good shape and practically without any cavitation marks. However, it was decided to install new runner blades of improved efficiency, designed and optimised for the future operating range. The additional cost was thought to be compensated by the additional energy output. For similar reasons the worn out mild steel guide vanes were replaced by stainless steel blades of modern design.

The turbine efficiency tests of the old and the new runner and the comparison of the two efficiency curves confirmed that the decision taken was correct.

Time schedule

The complete rehabilitation project took about three years from the determination of the required scope of the works to the physical completion of the work. The construction time at site originally was scheduled eight months from January to August, 2000. However, due to the inundation of the plant, non-availability of the network during commissioning and the overoptimistic planning of reconditioning activities for old equipment, 10 months were required. Energy was not lost in completion, since the energy remained stored in the reservoir.

The rehabilitated unit now operates at 4% higher efficiency at the rated operating point, and more than 7% in the part-load range. The unit can now be computer-operated from the keyboard and screen by one operator, thus reducing the number of operating staff and manpower cost and maximising the energy output.


Main technical data