Generators have been the workhorses of the hydro industry since the nineteenth century – and some of those early machines are still around, providing good service. Bill Moore, of US-based National Electric Coil, says the company rewound two 100-year old machines a couple of years ago. ‘They were designed conservatively and they weren’t pushed to their design limits,’ he says, ‘so they were still giving good service.’ And that plant is not unique: ‘Oil and paper, asphalt-mica, polyester-mica-paper, epoxy resin and mica paper; all these forms of transformer are still out there running,’ Moore says.

But after 100 years of conservative operation, the world for hydroelectric plants has turned around. Now they are on the front line of a privatised electricity industry, and major revenue-earners for smaller generating companies who need to respond quickly to the market.

Blake Lloyd, of Canada’s Iris Engineering, puts it succinctly: ‘Previously hydro was the poor man on the block, but now hydro is where the gravy is.’

That high profile also makes hydro vulnerable. Operators can increase the revenue from their hydro plants, but only by pushing generators and other components closer to their design limits. That might mean losing reliability, as Ian Slocombe of Adwel in the UK explains. ‘Increasingly, when generators are overhauled, they are rewound with more modern insulation. This is thinner, so it allows you to increase the copper and therefore the power – as has been done at Kariba North and South recently where the generators have been upgraded. But it means there is less margin for mechanical damage. If the insulation degrades or fretting affects the thickness, the reality is that there is a higher risk of failure.’ And that means losing money. ‘It was okay when electricity companies were state-owned and they had other generating capacity,’ Slocombe says. ‘After privatisation you want to have as few generators as possible and run them for as long as possible.’

The high cost of failure

In markets where electricity is directly traded, the cost of failing to meet a contract is high. The key is being able to predict, minimise and manage outages. Blake Lloyd says the industry is investing in technology that will help maintain reliability. ‘The length and frequency of outages is becoming very important,’ he said. ‘For example with partial discharge tests: instead of shutting down and doing extended inspection, we will put continuous condition monitoring on every generator and automate it. We are developing better monitoring and diagnostics for on line flux measurements. That will provide information about rotor pulls or shorted turns.’

‘Hydro plants are getting rid of bodies,’ Lloyd adds, ‘so they are getting involved with automation. Previously hydro plants would have to go down overnight to do an impedance test – this has been the method for a century – but now there are no personnel to do the tests, and companies do not want to take the units off line. The second step is using diagnostic systems.’

Maintenance planning is changing to fit better with these new operating constraints. Speaking about the UK, Ian Slocombe says: ‘The old way to do it was to take a plant off line for regular maintenance, say every five years. The time intervals were set for old styles of insulation- it’s very rare to find a failure within five years. The failure mechanisms vary from machine to machine, depending on the duty cycles, the environment and so on, so it is difficult to predict when the generator will fail.’

Similarly, in the US, the authorities currently require a maintenance outage every five years. However, epri is doing a risk assessment on extending the time between outages for large hydro plants, aiming to stretch the maintenance interval, or introduce an element of condition-based maintenance.

Slocombe gives an example of how monitoring is helping owners plan their outages.

‘Partial discharge monitoring costs around £10-15,000 to install, with a portable analyser that takes a few minutes to upload the results,’ he says. ‘But if it means you can delay the outage of a large hydro plant it can have a very rapid payback. Having the information gives engineers the confidence to keep machines running beyond the normal maintenance period. That increases revenue. And because you can see what is happening over time you can plan in maintenance and choose a time that suits you.’

With this information, he says operators can pick and choose when to operate. ‘You want to run at the best times, when there is lots of water behind the dam or during peak demand.

The worst thing is to have an unexpected shutdown. If you monitor, you will get a warning when the generator is starting to need attention.’

Dealing With The Data

It is not just baseload operation that makes money. Increasingly, hydro plants are being asked to load follow, or to undergo far more start/stop cycles than were expected when they were designed and built. This can place new stresses on the generator, along with the other components in the power house.

Bill Moore explains: ‘Load following does not increase stresses in the steel parts because the speed does not change. What does accelerate the stresses is if the machines are brought on and off line more frequently. This introduces fatigue on the rotor spider arms. They are more subject to bending and torsion strains and this can increase the possibility of fatigue cracking. This can happen if the unit is rewound and upgraded – then there can be a higher torque.

‘But load following will place a stress on the winding because of temperature variations as the voltage is changing. That can accelerate degradation of the winding insulation. Ramping up of load, increasing operating temperatures – these things can cause premature thermal ageing. In the past hydro had a long life but recently design limits have been pushed and this has shortened operating life. Failures in the strand or turn insulation, for example, can cause shorts.’

Moore explains how these effects are monitored. ‘We tend to increase monitoring on air temperatures on the inlet to the cooler,’ he says. ‘As the machine is pushed harder you must examine the air temperature. Slot resistance temperature detectors are used to monitor the coil temperature in the slot. Partial discharge monitoring is popular on rewinds, and long term trending monitors the deterioration of the winding insulation over time.’

Blake Lloyd says some operators have seen some core shifting. Vibration in the end windings can work on the lashing, and it may break loose. Rotors that have non-shrink rims may see them coming loose: when it starts to get oval instead of circular you can get air gap problems.

It is not just the windings that age. Bill Moore says as machines age owners just think about replacing windings: they don’t consider the stator core as much, but they should evaluate it carefully. It could alter the slot, or the laminations may shift and cut into the coil. Hot spots may develop. The stator core should be requalified and repaired or replaced.

‘As the winding ages the support brackets wear,’ he said. ‘Then vibration in the windings can accelerate further, and that can lead to shorts or it can wear the ground wall insulation. But vibration monitoring is not used so much in hydro, except on the large rated machines when the design is pushed to the limit. Most hydro is small, with little overhang and a lower speed so vibration problems are smaller. The important area is the vibration of the rotor at the bearings.’

More and more, the generator is monitored as it operates, but what happens to all the data that is produced? As Lloyd explains, in some cases it can be linked with data from other parts of the utility system to provide powerful information to the operation and maintenance teams.

‘New York Power Authority has 16 hydro units on the St Lawrence river,’ he says. ‘In its automation project it is installing new turbine runners to increase output. It is also adding advanced diagnostics – air gap monitoring, vibration monitoring, continuous partial discharge monitoring and new SCADA systems. Project Hydro takes information from all these sources and examines it periodically. The end users are the operations and maintenance departments.

‘The system will examine trends and longer term operation. For example it can compare this run up to one done a week ago, or this outage to the previous one. And instead of looking at simple emergency levels it examines the trends.

‘For operators this provides smarter diagnostics and alarming. For example the stator temperature may be acceptable, but if we compare what it should be with what it is, it may be two degrees hotter than last time.

‘Where previously the temperature would just hit or not hit an alarm, the new systems will compare with previous records, and suggest some reasons for the trend to the operator. We know the temperature is higher: now we can suggest why. Similar systems are already used for turbines.’

Experienced operators

Where does this leave the operator? Might ‘expert’ systems mean operators know less about the machines they work with? Blake Lloyd disagrees: ‘The operators are knowledgeable, but they have more units to manage and less time to find the information they need. They still have to respond to alarms or danger points,’ he said. ‘The benefit with expert systems is that you can load them with background information. Normally, for example, operators are not generator experts, but we can bring manufacturing and maintenance experience into the system.’

Slocombe points out that in some cases data cannot be interpreted with simplistic tools and experience is required. Partial discharge monitoring is an example. ‘Some operators want to feed the results of the monitoring into an automatic condition monitoring system – developing countries are interested in this because they may not have skilled people available,’ he comments. ‘There are systems coming out that are permanently installed, and in an ideal world you would have an expert system, but at the moment you need an experienced person to interpret the results.’

Watching Old Machines

New technology is always considered when operators are upgrading or uprating their generators. But what about the old machines – are they candidates for new monitoring techniques?

Slocombe puts a case for monitoring older machines. Generators often add condition monitoring when they carry out rewinds. The trend with some customers is that after they have put them on new machines they think of putting them on older machines – because if they can get another year before they have to do a rewind then it is worth the money.

For example Scottish & Southern have been doing major overhauls on their hydro plants in Scotland. The programme has been around for three years and will have several more years to run, and some generators are very near the end of their life. So they have found it is worth putting instrumentation on old machines and watching so they can select which generators to repair next. This way they can plan the overhauls and it gives them confidence in the generators that are still in operation.

Trends in the hydro industry are currently seeing many plants switched to remote operation. ‘Scottish & Southern has over 100 stations averaging 15MW. Only two are manned: the rest are operated remotely and visited periodically,’ says Slocombe. ‘Now you take readings for a few minutes every three months.’

Generators have been serving hydro operators well for over a hundred years, but that does not mean they are out of date. With new monitoring systems they are ready to meet the needs of the new hydroelectricity marketplace.