Bearing up under the strain17 August 2001
Increased friction, and even the seizure of bearings on dam gates, are growing concerns. A new technique can now measure trunnion bearing friction while the gates are in service
Radial gate arms are normally designed to withstand bending moments from nominal friction on the bearings, but experience shows that lack of lubrication and years of deterioration lead to increased friction and even seizure of the bearings. The bending moments produced by bearing friction imposed on gate arms are beyond the moment capacity, resulting in collapse.
Early dam gates (before 1980) are most commonly designed with trunnion bearings using a carbon steel shaft and bronze radial and axial bearings. The shafts are often chromium plated, and the bearings are usually lubricated manually with grease.
Dam gate bearings are static with relatively few movements during operation. This means that a lubricant, when applied externally, cannot be distributed on the high pressure sector of the bearing. Even when injected to the radial bearing, the lubricant often finds its way to the low pressure side, leaving the high pressure load transformation sector unlubricated. The axial bearings are often connected to the same lubricant pipe as the radial bearing. This gives an evenly distributed load, and the grease escapes through the low pressure side of the radial bearing.
The most important factors in the deterioration of bearings are corrosion and contamination. When water is present an electrolyte can occur with the carbon steel shaft as the anode. This leads to corrosion on the bearing surface of the shaft and increased roughness from pitting and corrosion products. Tight bearing radial clearance worsens this effect.
Radial gate arms are often slender steel structures, dimensioned to withstand bending and buckling from water pressure. Friction forces are often not taken into consideration, or nominal values of friction coefficient are used. Normally a lubricated bronze bearing and carbon steel shaft has a nominal friction coefficient of µ=0.2–0.3.
Deterioration of the bearings leads to increased friction, for which the arms are not dimensioned. When this causes breakage, it is through both fatigue and instantaneous breakage. Cylindrical bearings with a manual lubrication system, combined with an automatic or remotely operated gate, are a particularly vulnerable configuration.
Exposure to failure due to increased bearing friction depends on the design of the arms. The main areas of concern are the transition zones between the trussed beams constituting the arms and the hub accommodating the bearings.
As part of its initial research into the problem, Norconsult analysed the effect of a friction coefficient of µ=0.5 on radial gate arms at Norway’s Lundevann dam. At full water head it imposed a bending moment resulting in a stress of 7MPa in the large cross-section hub. The bending moment is transferred to the more slender trussed beams, constituting the gate arms. In the transition zone between hub and arms, the stress concentration was especially high in this example. A friction coefficient of µ=0.5 resulted in a bending stress in the transition zone of 100MPa. This stress change went from positive to negative when the gate movement changed direction, exposing the transition zone to fatigue.
Trunnion designs vary with regards to vulnerability for bearing seizure. For the common cantilevered trunnion shaft anchored to the abutment, symptoms of bearing seizure are difficult or impossible to detect by visual inspection. High friction moments can be transferred through the shaft without visible indications of increased bearing friction before the collapse occurs.
The trunnions may include a shaft lock, allowing rotation of the shaft in the console, which will remain deformed after a bearing seizure.
Inspection and maintenance of gate trunnion bearings often involves dewatering the dam gate and dismantling the bearing. Dewatering the gate can be accomplished by lowering the reservoir, however this can be time-consuming and costly if production is lost. More commonly, dewatering is achieved by installing stoplogs in front of the gate. This is less expensive than lowering the reservoir, but still more expensive than other methods. Another disadvantage is that during the decommissioning period the discharge capacity of the gate is not available.
Dismantling radial gate trunnion bearings is subsequently not routinely carried out to determine the condition of the bearing, but only if the symptoms of bearing seizure are evident. However, on some radial gates, total collapse has occurred without prior symptoms.
In theory, an increase in lifting force could be detected when bearing friction increases. However, the bearing friction constitutes only a small fraction of the total lifting force. On a typical radial gate with upstream lifting chains, a total lifting force of 156kN is dominated by the gate’s own weight of 126kN (81%). The remaining 30kN (19%) is friction in rubber seals and trunnion bearings. To separate the two friction forces, the friction on the seals must be based on an assumption regarding the friction coefficient between rubber seals and the embedded stainless steel frame.
Assuming µRUBBER-STEEL = 0.9, the friction on the seals constitutes 20kN (13%) of the lifting force. The remaining, 10kN (6%) is the nominal bearing friction based on µBRONZE-STEEL = 0.2. A variance of this relatively small force, compared to the total lifting force, is complicated to detect and is based on unreliable assumptions.
A diagnostic technique has been developed for radial gate bearings without dewatering the gate or dismantling the bearings. The method detects friction on the bearings during operation of the gate. Strain gauges attached to the gate arms can measure the strain caused by bearing friction during movement. The signals from the strain gauges and a gate position transmitter are logged digitally, allow direct presentation of test results. The forces acting on the trunnion bearing can be divided into two classes:
• Perpendicular forces on the bearing surfaces as the result of water pressure, gate weight, operating forces and friction forces from the rubber seals. The resulting force from the water is the dominating force on the trunnion.
• Shear forces parallel to the bearing surface due to friction. Without friction (µ = 0), no shear forces will appear. These forces will give bending movements in the gate arms.
The shear forces on the bearings can not be measured directly without modifying the gate structure. Instead the mechanical stress variation is measured in the gate arm, near the trunnion bearing, using strain gauges. One strain gauge is attached at the upper side of the gate arm, and one is attached at the lower side of the gate arm. Each strain gauge measures the surface mechanical stress in parallel to the main stress direction in the gate arm.
Superimposed axial and bending forces cause stress. By connecting the strain gauges in a Wheatstone half-bridge circuit the two measured stresses are subtracted and the output only indicates active bending moment in the gate arm. By using this circuit, superimposed axial (normal) strain is compensated, and the strain caused by thermal changes is compensated to a high degree.
The measured stress creates a hysteresis curve superimposed on a mean stress variation. The variation in mean bending stress is mainly caused by variation in how the flow forces act on the open gate. These forces are dependent only on the gate opening, not the direction of the movement. The friction forces in the bearing always act against the direction of the gate motion. When the motion changes direction the measured bending stress, due to friction forces in the bearing, also changes direction. In addition the friction force varies with the gate opening, as the water load on the gate alters.
Dependent on the hoist design, the friction forces from the gate rubber seal have some influence on the total measured friction force. The upstream chain hoist has no influence. In this actual design the rubber seal friction gives less than 5% effect on the measured bending stress when µRUBBER-STEEL = 0.9 (worst case) is assumed.
The friction torque at the bearing trunnion is calculated based on the measured bending stress due to friction. This calculation is done as a function of moment of inertia in the gate arm at the strain gauge location and radial distance from the strain gauges to the bearing centre line.
The water load on the gate is calculated as load from the static water pressure at the wet surface of the gate. The accuracy of this calculated load is high when the gate was closed. At increased gate opening, larger areas were exposed to flow velocity, which decreases the static pressure.
The method over-estimates the load when the gate opening increased, but according to pressure distribution calculation on a gate, the error is limited to approximately 15% at 25% gate opening. This calculation gives the main input to the radial load at the bearings transmitted through the gate arms.
Dependent on the design, the gate arm axial force can be split into two components, one acting on the axial bearings and the second on the radial bearings. This split is dependent on the gate design. When both radial load and friction torque on the trunnion bearings are known, we can calculate the average friction co-efficient on the bearings.
Since Norconsult developed the method in 1998, a total of 30 bearings on 15 radial gates in six different dams have been measured in this way. On one of the dams where increased bearing friction was detected (Lundevann dam) repeated measurements were carried out.
Lundevann dam impounds the reservoir for the Åna-Sira hydro plant. In the outlet of Lundevann a concrete dam was constructed with two spillways and surface radial gates. Measurements were carried out on the bearings under three different conditions:
• Original bearing without prior lubrication
• Original bearing thoroughly lubricated
• New bearing with self-lubricating bushing
Measuring the original bearings
The first measurements on the four bearings, two on each gate, were carried out in 1998. Prior to the measurements, the gates had been stationary for one year without maintenance or lubrication. The measurements revealed increased friction values in all bearings. The friction on the bearings of the most frequently used gate 1 was especially high. On the first opening movement, the friction coefficient was 0.75, dropping to left 0.60/ right 0.50 in the second opening. The coefficient of friction remained considerably higher than expected for the material properties of the bearings.
In 1999, Norconsult repeated the measurements after the bearings had been thoroughly lubricated and moved repeatedly to distribute the grease. The friction coefficient then fell by 25% to 0.45 and 0.35 on the same bearings. It looked like the bearings had suffered permanent damage and the company advised dewatering and replacement.
In 2000, the gates were duly dewatered and the bearings were dismantled. The inspection revealed traces of seizure on the bronze surfaces and seizure and corrosion on the surfaces of the cast carbon steel shaft. It was also evident that the manual lubrication system had not functioned as intended, partly due to clogged grease channels and the fact that the grease escapes from the bearing on the low pressure side.
The bearings were completely replaced. New shafts were manufactured in stainless steel, SIS 2387. For the radial and axial bearing, a non-metallic material was chosen – Orkot, a brand from Busak&Shamban. Orkot TLM marine grades are non-asbestos composite materials incorporating woven fabric reinforcement and solid lubricants within a thermosetting resin matrix. The manufacturer gives them a friction coefficient between 0.10 and 0.15 against stainless steel.
After assembly and commissioning of the new bearings, the friction on bearings on gate 1 was measured. The results show friction coefficients within the values given by the manufacturer. The measured friction coefficient was 0.09 and 0.13 for the left and right bearings respectively. The results from the measurements of the new bearings, together with the known properties and characteristics of the bearing materials, verified the diagnostic method.
The initial (static) and dynamic friction of the original unlubricated bearings was almost constant. When lubricated the original bearing had almost the same static friction as unlubricated, while the dynamic friction dropped by 25%.
The method of measuring bearing friction by means of strain gauges provides dam owners with a better diagnostic technique. Experience has shown that the method will detect bearing failure at its bending moment capacity. The influence from dynamic conditions and the friction on the seals is insignificant. Measurements are carried out without dewatering or decommissioning the gates. The method gives objective and precise verification of measuring results, minimising the possibility of subjective and wrong assessments.