After a canal failure at the Rio Seco plant, Pedro, the plant manager, contacted George, a hydropower industry consultant, to ask: "How much discussion of ramping rates should there be in an operating manual?"
George had been following the developments at Rio Seco, having discussed the incident when all parties involved called him for advice shortly after the event.1 The development comprised a 30 year-old irrigation dam where a powerhouse containing three Kaplan units had recently been added.
The embankment dam with an upstream blanket was built at the outlet of a large lake to provide irrigation water during the dry season. The reservoir was drawn down by 20 meters to empty each year. Recently, energy costs had increased to the point where it became economically attractive to add a power facility. This was undertaken with a design-build contract, under the general supervision of consultants engaged by the dam owner.
The facility included a 2 km-long canal, mostly in sandy silt, that carried water to a powerhouse constructed on a rock ridge adjacent to the right abutment of the dam. The portion of the canal that ran through the silt was lined with concrete, and the lining was tied to the clay blanket extending upstream of the storage dam. The Kaplan turbines ceased operation when the reservoir was drawn down by 16 meters to a minimum operating gross head of 14 meters. After this, the reservoir level continued to drop until empty.
The right abutment contained an aquifer in a sandy layer confined by the more impermeable sandy silt. Where this layer met the rock ridge, artesian pressures developed at low reservoir water levels. When the dam was built, a local drainage system was installed below the blanket to relieve the pressure, and seepage flow at the drain outlet was monitored. A similar drainage system was installed below the concrete slabs forming the canal, and the flow from the canal drain was also monitored. The power plant was commissioned and operated successfully for several years.
About halfway through a dry season, the canal failed due to uplift in a few of the concrete slabs. Investigation revealed the failure occurred when the reservoir was drawn down to just over 15 meters, close to the minimum power plant operating level, and shortly after the three turbines were loaded to full output in about 20 seconds. This caused a standing wave to form near the canal entrance, resulting in a substantial drop in the canal water level at the intake. Water pressure in the canal drainage system must have been higher than the weight of the concrete slab and the canal water pressure, resulting in uplift of the slabs. The failure was detected by a sudden increase in flow at the monitored drain outlet, and the units were immediately shut down.
Further investigations revealed there was a restriction on the "ramping rate" — the allowable rate of increasing load on the units — when the reservoir was near the low operating level for the power plant. When the reservoir was within 7 meters of the full supply level, there was no restriction on ramping rate because of the large flow area in the canal and consequent imperceptible head loss. However, below this level, a ramping rate was introduced, and the ramping time increased until it was very long at near full power plant drawdown level. Unfortunately, the consequences of not following the ramping rate were not discussed, and the ramping rules were not prominently outlined in the operating manual.
Consequently, plant operators may not have understood the ramping restrictions, and they may not have been discussed during commissioning, when the reservoir was full and there were no restrictions. Normally the power plant operated on base load, passing a steady flow of irrigation water, hence the ramping rate restrictions were not significant. Unfortunately, the incident occurred when the power plant was returned to service after a short outage on the transmission line and the remote operators called for full load.
Of course, the plant owner immediately instituted a search for the culprit, questioning the consultant, contractor and contractor's consultant. Fortunately, all involved realized trying to assign blame would seriously delay the repair work because the opportunity provided by a low and near-empty reservoir would greatly simplify dewatering the canal for repairs.
Agreement was reached to forget about blame and concentrate on repair work. This was accomplished by constructing a cofferdam at the canal entrance, removing the broken slabs and rebuilding the work to the original design. The operators received detailed instructions on the allowable ramping rates.
However, the plant owner had loss of business insurance and asked for compensation after the plant resumed operation. By then, the losses approached 10 times the repair cost, and this was paid by the insurance company. Then, the insurance company sued all involved for compensation. Fortunately, the case was settled out of court for a fraction of the loss.
There are two lessons in this incident. First, looking for a culprit takes time and is unlikely to be resolved quickly because all involved will consult their lawyers and insurance companies. Meanwhile time is lost, and the value of lost generation will quickly exceed the probable cost of repairs. Second, ramping rate restrictions are very difficult to follow, especially if the rate varies with reservoir level.
A note: For multi-unit power plants, ramping rates should be provided for each unit and for all units operating in unison. Control center operators deal with a large number of power plants, and if there is a ramping restriction on one of them, it is not adequate to have the ramping rules discussed in an operating manual that is rarely consulted by operators. Where a failure is possible from not following ramping rates, there should be a second line of defense, as there is in turbines when the wicket gates fail to close. This can be obtained by incorporating ramping restrictions into software used to operate the power plant, so that an operator cannot exceed the allowable ramping rate when calling for a rapid increase in turbine or power plant load.
— By James L. Gordon, B.Sc., hydropower consultant
1. Gordon, James L., "Ramping Rate Restrictions," HRW-Hydro Review World-wide, Volume 19, No. 3, July-August 2011, pages 46-47.