Intake gates - upstream or downstream skinplates?16 November 2001
Before the 1960s, all intake gates had downstream skinplates. After 1960, roller wheeled gates with upstream skinplates were introduced when it was discovered that such gates eliminated one wall in the intake and had a lower downpull at part gate openings. But, as J L Gordon explains, the increased risk and installation costs associated with upstream skinplated gates were overlooked
GATES with downstream skinplates are installed in a gate well upstream of the air vent well, and require a concrete wall to separate the two wells. The gates are generally equipped with rubber seals on the downstream face, which are forced against the sealing path by water pressure. As the gate deflects downstream under water pressure, the deflection increases the compression force on the seals, increasing the sealing pressure and reducing leakage. In the gate bedded parts, the roller path and the seal face are beside each other. This means that only one set of anchors on the heavy roller path need adjustment, which facilitates installation. With a separate air vent, the cold exterior air can be kept away from the gate well which can be heated in cold climates to prevent freezing. However, there is a major detraction in that downpull forces at near-closed openings are high, depending on the configuration of the bottom lip.
A downstream skinplated gate has several advantages. These include:
• Gate deflection increases sealing force.
• The gate well can be easily heated.
• Alignment of bedded parts is considered relatively simple.
However, it does have disadvantages, namely that a concrete wall is needed between gate well and air vent well, and downpull at part gate openings is high, requiring a stronger hoist.
Gates with upstream skinplates, however, are installed in the same well as the air vent, thus saving the expense of a wall between the gate well and the air vent well. With gate and vent in the same well, heating of the water surface is more difficult in cold climates. The seals are located on the upstream face and are pulled away from the sealing surface as the gate deflects downstream under water pressure. Hence gate deflection decreases sealing force and the seals need to have more flexibility to overcome this deflection. Due to gate deflection, there is an increased risk of top seal 'rollover' under pressure or on lowering the gate, resulting in damage to the seal. Alignment of the top seal is more difficult, and often the seal has to be shimmed in the centre to reduce leakage to acceptable levels. In the bedded parts within the gate check, the roller path is on the heavy downstream face and the seal path is on the lighter upstream face, separated by the diameter of the rollers. Alignment to within a fraction of a millimetre is more difficult, requiring the adjustment of two sets of anchors: those on the heavy downstream roller path and those on the lighter upstream seal path. On the other hand, downpull at part gate openings is negligible. As such, an upstream skinplate gate has the reverse of the advantages/disadvantages of a downstream skinplate gate.
From a design standpoint, it is evident that an upstream skinplate gate is the preferred choice due to the lower cost of the installation. But what about operating aspects?
When the intake gate is located at the head of a long pipeline or tunnel, there are clear indications that a downstream skinplate gate is the preferred alternative.
Over the past few years both anecdotal and hard evidence has been accumulated on several incidents of inappropriate gate operation, resulting in damage to facilities but fortunately no loss of life. All incidents have resulted from accidental rapid opening of a gate when the pipe or tunnel was empty, or only partially full. All gates were equipped with controls to prevent such an operation but somehow the controls failed, or the operating instructions were overlooked.
Normal pipe filling is accomplished by lifting the gate about 10cm to a 'prime' position, and then waiting until the pipe is full before lifting the gate to the full open position. This operation takes from several minutes to several hours, depending on the volume of the empty pipe. This length of time allows for the steady release of air in the pipe being displaced by the water. Problems arise when the gate is lifted fully, without pausing at the 'prime' position. In such an event, the rapid flow of water down the pipe entrains a large volume of air, and the air/water mixture traps more air within the pipe to produce a large bubble of compressed air at the bottom of the pipe. As more water enters to fill the pipe, the downward water velocity reduces below the rate of air bubble rise, and the compressed air bubble/water mixture starts to rapidly ascend the pipe to emerge through the air vent with a highly explosive force.
Two of the following incidents occurred at low level sluice gates, when the upstream emergency gate was opened with the downstream gate closed, a situation equivalent to the opening of an intake gate on an empty penstock.
The intake in this example has a downstream skinplate gate, heated air vent house at deck level, and a wire rope hoist within a housing which is supported on a steel tower high enough to permit lifting the gate clear of the deck for painting and inspection. Downstream there is a long buried steel pipeline to the power house. The gate was originally operated from the power plant.
A technician was working on the gate controls in the hoist house and inadvertently left a jumper cable to disable the opening halt at the prime position. On activating the open gate control in the power house, the gate opened to the full open position, despite all attempts to halt the operation. The air vent house was demolished in the ensuing expulsion of an air/water mixture. Gate opening controls were subsequently transferred to the hoist house, where a master switch could be opened, cutting off electrical power to the hoist in the event of control malfunction.
The intake has an upstream skinplate and is located at the head of a long pipeline. As in the first incident, the gate was inadvertently opened to 900mm with about 200m of the pipe being empty, before opening could be stopped by the local operator. Normal gate prime position is at 76mm. The water and air/water mixture lifted grating covering the air vent. It also lifted heavy gate chamber steel covers, damaged some water level recording equipment and lifted the roof off the concrete gate control house located above the intake deck. The operator in the control house, although completely submerged in water, suffered only minor grazes. The water was estimated to have risen about 3m above deck level to cascade out over the walls of the control house. The deck is about 7m above full reservoir level.
Fortunately, the gate has a hydraulic hoist and the rigidity of the operating cylinder and piston was sufficient to hold the gate in place against the upwelling air/water in the gate well. If the gate had been equipped with a wire rope hoist, there is a strong possibility that the gate would have been ejected from the gate well. The cause was found to be a faulty float switch incorrectly signaling a pipe full condition, thus allowing the gate to open past the prime position.
The intake has a downstream skinplate, with a wire rope hoist on a tower and separate vent house. The water-full pipeline had been shut down for several weeks, with water held against the turbine rotary valves. Unfortunately, one of the rotary valve bypass valves was leaking and water slowly drained from the pipe. The operator was instructed to determine the water level in the intake air vent before initiating the opening sequence. Not being able to see the water surface, he poured some coffee from his thermos into the air vent and listened for the impact.
Assured that he had heard the coffee striking a water surface, he climbed the intake tower, pressed the 'prime bypass' and opened the gate. The air/water mixture emerged from the air vent with such explosive force that the air vent building was totally demolished, and the hand railing around the house was completely flattened (see above). Anyone standing in the vicinity would have been severely injured.
Fortunately, the operator had remained on the intake tower deck, so escaped any injury. An estimate of the leakage flow indicated that the pipe was between two-thirds and completely full, with the water level at the intake well below sill level. The operator had heard the coffee striking a shallow pool of water on the concrete floor. Water level determination procedures were changed to dropping a float on a tape.
The 11m high intake gate has an upstream skinplate. Normal pipe filling was through a small bypass valve and required a considerable time to fill the pipe. The operator wanted to speed up the process and instead cracked open the gate. Towards the end of the pipe filling, the gate catapulted upwards by about 12m. On dropping back onto the sill, the bottom 4m of the gate was destroyed along with the sill sealing plate.
At another dam with low level outlets, there are downstream tainter gates and upstream bulkhead gates with upstream seals. With the conduit between the gates empty, the by-pass valves were being opened when the gantry crane operators inadvertently cracked open the 3.6m high bulkhead gate. The gate catapulted upwards by 76m to deck level, to rest canted sideways, with 75% of the gate out of the guides. Fortunately there were no injuries.
This project has low level outlets equipped with regulating gates and, further upstream, emergency wheeled closure gates. The emergency gates have upstream skinplates and operate under a head of 80m. With both gates closed, and the conduit empty, the emergency gate was opened and catapulted upwards by about 3m to destroy the top seal on closure.
An upstream seal slide gate catapulted under a head of only 21m.
At this project there is a downstream skinplate gate in a tightly fitting well. The air vent comprises a small pipe about a metre downstream of the gate. Penstock filling is accomplished by cracking open the gate by just under 10mm, to avoid excessive velocities in the vent. In this incident, the gate opening limit switch failed and the gate continued to open. Calculations indicated that the penstock filled when the gate had opened to about 60mm, and the ensuing waterhammer was sufficient to catapult the gate to the deck where a pickup truck had been parked partially over the gate well. The pickup and well cover were thrown into the reservoir. A steel screen over the air vent on the downstream face of the dam was recovered about 400m away.
Model studies undertaken in the US have shown that catapulting of upstream skinplated gates can occur when the area of the gate shaft is smaller than the gate opening. Normally, the gate shaft on most intakes has an area of about 25-50% of the gate flow area.
From these incidents, it is evident that there is a basic difference in the operating risks associated with upstream and downstream skinplate gates. With a downstream skinplate, the upwelling air/water mixture encounters a smooth gate surface and consequently cannot lift the gate. On the other hand, with an upstream skinplate, the air/water mixture encounters the horizontal beams supporting the gate pressure face and the full area of the bottom beam. These beams provide sufficient purchase for the air/water mixture to lift the gate and 'catapult' the gate upwards, perhaps out of the gate slots if there is sufficient force. This scenario should be taken into account in any assessment of operating risks.
From an operating standpoint, the preferred gate is one with a downstream skinplate, since there is no risk of the gate being damaged by catapulting or ejected from the slots during a control malfunction, provided the air vent is generously sized. With a downstream skinplate the risk of damage to the air vent house can be minimised by appropriate design measures such as reinforced concrete walls on three sides, with a heavy roof and an expendable 'blow-off' side located where no personnel would be at risk.
This presents the gate designer with a dilemma. A downstream skinplate gate is more expensive but has less operational risks than an upstream skinplate. The only recourse is to discuss advantages and disadvantages with the project owner.
With fewer operating staff, and more reliance being placed on electronic controls, automatic opening of intake gates is becoming more common. Under such conditions, it is suggested that downstream skinplates offer the lower risk. This opinion is based on the minimal extra cost of the additional concrete wall in the intake - added concrete only costs about half of the unit price if the concept is shown on the initial drawings. As for the hoist, at about double the capacity of that for an equivalent upstream skinplate gate, the incremental cost is also minimal due to the current low cost of materials. Hoist costs are mainly a function of the manufacturing man-hours, not material cost, resulting in a doubling of capacity only increasing hoist cost by about 30%. The hoist only represents about 35% of the total gate, hoist and bedded part cost.
With existing intake gates, the gate opening controls should be arranged so that a timer prevents the gate from opening past the prime position until the pipe is full. If this time is lengthy, on long pipelines, the controls could be enhanced to include two float switches wired in parallel. If one or both of the switches indicates that the pipe is not full then the timer would be energised to delay opening until the pipe filled.