Hydro-Quebec is carrying out extensive refurbishment of the head gate system at the Beauharnois dam in Canada. This paper outlines the inspection, testing and preliminary engineering which was necessary to define the scope of the subsequent rehabilitation programme. Report by P. Chaput, D. Gawish and M. Limoges*

THE Beauharnois hydropower station, on the St.Lawrence river in Quebec, Canada, has an installed capacity of 1650MW. This is delivered by 36 turbines, each of which have two head gates. The intakes were built in three phases from 1931-1958. In phase one, there are 36 gates, and in phases two and three, there are 24 and 14 respectively. Three gantry cranes are used to operate the phase one and phase two gates, while the phase three gates function through individual drum hoists.

The actual rehabilitation of each head gate system consists mainly of replacement of the existing gates, refurbishment of the gate guides which are affected by corrosion and by the progressive deformation of the concrete due to alkali-aggregate reaction, and installation of individual hoists for the gates of phases one and two.

In 1995, the first part of the programme, covering the refurbishment of the water passages of 19 turbines, including the intake and draft tube gates and guides, was started. It is scheduled for completion in the year 2000 and covers water passages in phases one, two and three, although it will focus predominately on phase one.

In phase one, the gates are of a rivetted construction with an upstream skin plate and are equipped with roller trains. The seals are downstream, except at the sill beam. The roller paths are 22mm-thick carbon steel plates rivetted to the bedded support beam and the gate lateral guide wheels roll on channels rivetted to the steel liner of the gate slot.

In phases two and three, the gates have wheels with bronze bushings and have downstream skin plates and sealing arrangement. The roller paths are respectively 45mm and 20mm-thick; the first is of grade 1040 carbon steel, and the second, of A514 steel. The gate lateral guide wheels roll directly on the steel liner. The gate hoists in phase three are equipped with fan brakes for emergency closure.

Inspection and testing

A number of intake gates and gate guides were inspected in 1992 and 1993. The gates could be raised above the slots by the gantry cranes even while the unit was kept running, so that the inspection was easily accomplished. However, inspection of the gate guides was carried out underwater as no

stoplogs were available at that time.

Inspection of three gates in phase one included ultrasonic thickness measurements and the evaluation of the surface corrosion pits to calculate the effective residual thicknesses. The loss of thickness due to corrosion generally did not exceed five per cent in main structural members and plates with nominal thickness of 12mm or more. Three gates of phases two and three were similarly inspected and the effects of corrosion were generally less significant.

Underwater visual inspections of phase one gate slots had shown that the channels for the guide wheels and angles guiding the roller trains had deteriorated severely. Underwater inspections of the gate guides also included ultrasonic thickness measurements of roller paths and measurements of verticality, linearity and surface roughness.

As mentioned previously, continuous swelling of the concrete caused by an alkali-aggregate reaction has particularly affected the verticality of the gate guides; this phenomenon is also expected to continue in the future. Some of the deviations in verticality are accordingly relatively large and are expected to increase.

Instrumented tests of closure under flow on gates in phases one and two were also carried out. Complete closure was obtained at 25 per cent flow in the first case, and at 12 per cent flow in the second.

Furthermore, it was already known from previous testing that the gates of phase three did not close completely at 50 to 100 per cent of nominal flow. When tested at flows of 35 to 45 per cent of full flow, the phase one gate stopped at a distance of 0.7-1.5m from the sill, with the load oscillating greatly. The downpull is higher on these gates, as the bottom seal is positioned to facilitate closure at full flow; the severe deterioration of the guides is the likely cause of the failure to close completely. The phase two gate stopped at 10-15cm above the sills when tested at flows of 25 to 35 per cent of nominal flow. The calculated friction in the wheel bushings is an important factor in this case.

Flexibility of the existing gates

With the goal of minimizing the extent of the rehabilitation work on the gate guides, flexibility analyses were undertaken on the existing gates to obtain the range of wheel load variations associated with various types of roller path defects. A number of finite element flexibility analyses were carried out; the gates of both phase one and phase two were modelled.

The cases studied first included the effects of the measured difference in verticality of the two roller paths in the upstream-downstream direction. The model of an actual phase two intake gate was equipped with 16 wheels, numbered on either side. Simulation of a verticality difference of 25mm between left and right guides resulted in a change of ten per cent of the maximum value of wheel load resulting from hydrostatic head on the gate.

A number of variations of linearity defects were simulated. Removing the support under wheel five, located at approximately mid-height of the gate, results in overloading the adjacent wheels by 65 to 70 per cent, but the maximum wheel load for this gate, when considering the theoretical load distribution to all wheels under hydrostatic pressure only, is increased by 40 per cent. Linearity defects at the lower end of the roller path can result in more severe wheel load distribution for this gate design. A valley in the guide having the effect of removing the support under the lower wheel causes the load on the second wheel to increase by more than 100 per cent. When it is the second wheel that is left unsupported, the load on the third one increases by 85 per cent. The maximum wheel load, considering all wheels, is increased by approximately 45 per cent.

Figure 6 illustrates the theoretical loading on the wheels with an ideal roller path and the loading obtained when the second wheel is not supported. In the above simulations, the gate end girder deflected 0.4-0.5mm at unsupported wheels.

These analyses do not take into consideration the local deformation in the wheel and the roller path due to the very high contact stresses. For wheels such as those on the gates modelled, these elastic deformations, calculated from the centreline of the wheel to a plane at the middle of the roller path, total 0.25mm when a wheel is subjected to the average load. This value is then quite significant compared to the calculated deflection of the end girders and would reduce the overload determined by the analysis.

It can be concluded from the above that the structure of the existing phase two gates is sufficiently flexible to allow linearity defects greater than 0.5mm within the wheel spacing, depending on the location, without exceeding acceptable overload to some wheels. Verticality defects in the upstream direction, even when projected to double over the next 50 years, will be a major concern as long as the new or refurbished gates are at least as flexible as the existing ones. As the phase two and phase three gates are of similar designs, this applies to both types.

The criteria was then established that the intake gates, whether they be refurbished or new ones, must be sufficiently flexible to withstand doubling of actual verticality defects of the roller paths without causing excessive over-loading to some of the wheels. Also, to reduce the cost of refurbishment, the linearity tolerance for the rehabilitated roller paths in phases two and three was increased to a range of 0.8mm over 3m; but more significantly, flexibility analysis was made a requirement to demonstrate that the design would not result in overload exceeding 50 per cent of the maximum theoretical wheel load.

Basic concepts

After due consideration of the measured corrosion rates, the structural soundness of the existing intake gates was found to be adequate for at least another 25 years. However, the modifications to the equipment required to retrofit individual hoists and ensure closure under flow added to the cost of reconditioning the whole gate, rendered the rehabilitation of the phase one gates, at least, economically unattractive. Thus, following detailed evaluation, the phase one gates, operated on roller trains, were to be replaced by new fixed-wheel gates, and the gates in phases two and three were to be modified and refurbished. All the gates were to be equipped with roller bearings to ensure emergency closure,

and in phases one and two, individual drum-type hoists with fan brakes were to be installed.

Rehabilitation

Following evaluation of the different proposals submitted by tenderers, it was decided that new head gates would be installed in all three phases because of the low differential cost between new and refurbished gates, and the greater uncertainty with regards to the final cost and quality associated with refurbishment. The proposal that was retained included concepts for refurbishment of the gate guides that generally followed those developed for the technical specifications.

The new head gates have downstream skin plates and are fabricated in two sections; the end girders being hinged near mid-height. This design feature has the advantage of minimizing installation time and also provides more flexibility for warping of the whole gate as the two roller paths may not be truly parallel. The end girders have a reduced cross-section; however the skin plate is much heavier than on the existing gates. The new design also features 22 wheels whereas the existing fixed-wheel gates have 16; thus resulting in a lower nominal load on each wheel.

Since the specified installation tolerances for the refurbished roller paths have a broader range in phases two and three, where local surface grinding only had been foreseen, the more significant results from the inspection and refurbishment work on the guides in a phase two and a phase three intake are summarized hereafter.

For the phase two gate guides, the measured values of residual thickness, verticality, linearity and surface roughness, prior to refurbishing, generally fell within the expected ranges. However, the linearity defects did exceed locally the predicted value of 1.6mm to reach a maximum of 2-2.5mm over spans of less than 1m.

Considering the positions of the gate wheels, rectification of the roller path surfaces had to be defined over part of their effective length. Since the important parameter to attain in this case was the maximum wheel load, the contractor performed a flexibility analysis of the gate design to confirm which part of the roller path needed rectification; this turned out to cover 15 to 20 per cent of the effective length. This work was carried out with a rotary sander and flexible discs. The surface roughness of the seal faces was almost entirely eliminated by grinding an average of less than 1mm in depth, using the rotary sander.

In the case of the phase three roller paths, where the original plate was made of A514 material, inspection by the contractor revealed that a number of grooves up to 3mm deep had formed under the wheels over a length corresponding approximately to the height the gates are raised to when filling the water passages. These grooves were deeper at the elevations of the lower gate wheels. Since the original design of these roller paths allowed for the simple replacement of the original plate by a new one, it was decided the roller paths should be replaced.

The new roller paths were surveyed after installation and corresponding wheel reactions were calculated. The contractor then proceeded with sanding the roller path peaks in the areas surrounding the wheels

where unacceptably high reactions had been identified.

Final survey

A final survey was done and a set of reactions were calculated to ensure the adequacy of the sanding (see figure 5). These results clearly show that acceptable wheel loads were obtained although the specified linearity tolerance was exceeded.

When it is feasible to carry out the correction of roller path surfaces by grinding with a rotary sander, and this depends on the design and the state of deterioration of the original equipment and components, the investment in engineering analysis is worthwhile as it reduces the scope of the refurbishment work to the roller paths and eliminates the need to install new sills and lintels.

At the Beauharnois dam, it also eliminated the need to install new light roller paths in the upper part of the gate guides. Significant cost reductions can be achieved without compromising the reliability of the equipment when all aspects of this approach are thoroughly analyzed and carefully executed throughout the installation phase.
Related Articles
Spotlight on… Canada