Electricity System Performance during snowstorms on 27 and 28 February 2001
A report for the Director General of Electricity Supply for Northern Ireland, on the performance of the Northern Ireland Electricity system during the storms, and the effectiveness of NIE in restoring supplies and communicating with customers
2 Weather conditions
3 Ice accretion
4 Conductor galloping
5 Customer service and communications
5.1 Call handling
5.2 Call handling fault
5.3 Lack of firm information
5.4 Staff communications
6 NIE repair performance
7 NIE system performance
8 Special needs customers
For a period of around 48 hours from the evening of Monday 26 February 2001, parts of Northern Ireland experienced considerable disruption to electricity supplies as a result of damage caused by high winds and snow.
The majority of NIE customers were unaffected. Nevertheless, at the height of the storm, no less than 100,000 customers were off supply.
While 40% of customers were reconnected within 3 hours, and 99% within 48 hours, a few were not reconnected until more than 72 hours had passed.
The purpose of this report is to review the actions NIE took (or did not take) before, during and after the storm, in order to establish
· Why there was such widespread disruption
· Whether NIE’s reaction was timely and sufficient
· What could have been done better
· What NIE did well
· What lessons can be learned to mitigate the effects of future ‘bad weather’ events
2 Weather conditions
Early weather forecasts for the area suggested that high winds and snow were likely. However, the system had withstood heavy snowfall between Christmas and New Year without damage, and the forecast wind speeds were not excessive.
Of particular importance to NIE was a statement in the forecast issued by the Met Office at 14.09 on Monday 26 February, that ice accretion was not expected. Ice accretion is covered in Section 3. Had it been forecast, it is likely that NIE would have staffed up sooner. As it was, no particular difficulties were envisaged.
In practice, the weather turned out to be considerably more severe than forecast.
The gales started in the north of the Province, with driving rain. As the gales moved south, they increased to severe gale, gusting to storm force, while the temperature fell and the rain turned to sleet and snow. The high winds caused severe drifting. The resultant eight-foot snowdrifts were later to hamper NIE crews even reaching some of the damage that was caused to the overhead system.
A second weather warning was issued by the Met Office at 21.49 the same evening to the effect that ice accretion was now expected, with prolonged sleet and snow in the east.
By this time the worsening conditions had already led NIE to open their incident room in order to be able to control the situation, while the first system failure due to ice accretion occurred at 21.53 hours, just four minutes after the Met Office warning.
3 Ice Accretion
Ice accretion is an entirely weather-related problem. It occurs only where three conditions coincide:
· There has to be a high wind
· The temperature has to be between 1° C and -1° C
· It must be snowing
The effect of these three conditions is that the wind blows wet snow onto power lines. The wind-chill factor promptly freezes the snow to the wire. As snow builds up on that side of the conductor facing the wind, its weight twists the conductor round, exposing more conductor to ice build up.
This process continues. The result can be that an overhead conductor substantially less than half an inch in diameter ends up encased in several inches diameter of solid ice.
One of three things then happens.
· The weight of the ice stretches the wire, bringing it to the ground
· The weight of the ice snaps the wire, with similar results
· The force of the wind against a large cross-section of wire, ice and snow topples, or snaps, the pole. Again, the line is brought down
In all three cases, the result is the same. Power has to be cut off, and unless other circuits are available to take over from the damaged circuit, supplies will remain off until repairs can be made.
Ice accretion can occur in any country where the appropriate weather conditions occur. Since the last incident in Northern Ireland (on the Glenshane Pass in 1985) ice accretion has occurred in England and Wales, France and Canada, in each case bringing widespread disruption to power supplies.
4 Conductor galloping
Conductor galloping is another event caused by ice accretion, although the process and results are quite different.
Whereas wood-pole distribution lines of mains voltage, 11,000 volts, or 33,000 volts, use comparatively small conductors, the electricity transmission system uses very much larger (and hence much stronger) conductors at very high voltages, suspended from ‘transmission towers’.
These conductors are often used in pairs, or even fours, in order to increase the power transmission capacity of each circuit. Where two, or four, conductors are suspended from a single insulator, they are held apart by spacers to prevent the mechanical damage that may arise from repeatedly banging together in light winds.
The effect of the spacers is to prevent the wires twisting. Consequently, when ice accretion occurs, the wire cannot twist round, and the snow continues to build on only one side of each conductor.
The effect of this seems to be to create an aerofoil shape, with the result that the wind causes the conductor to lift (in much the same way as the airflow over and under an aircraft’s wings give it lift), and the conductors rise mid-span. The end result can be that conductors rise and fall rhythmically.
There are usually three (groups of) conductors arranged up each side of a transmission tower, and very rarely will all three rise and fall together. The result is that one rises and meets its neighbour above, at which point a high voltage short circuit occurs, and shuts off the power.
It is very rare for this sort of event to bring the lines down, though it can do in extreme conditions. Nevertheless, damage can occur to the conductors due to the burning caused by the electrical arc, and lines will be examined for signs of this once the weather has eased.
Transmission lines can normally be re-energised remotely following flashover, and this is routinely done. However, if galloping continues, failure will recur. After continued tripping, it becomes necessary to leave the line isolated until it can be examined.
The rarity of conductor galloping can be illustrated by the fact that the last significant case of it in Northern Ireland was in 1985, on the Glenshane Pass.
This time around, galloping occurred on several circuits in the eastern half of the province, the result short-circuiting causing several interruptions on the transmission system.
Some transmission circuits became so unreliable that NIE had to download some generation plant, and run more expensive plant elsewhere in the Province, in order to keep the loads on the useable parts of the transmission system within its capability.
That they were successful in doing so is illustrated by the fact that there was only one transmission circuit failure that caused loss of power to customers during the entire event.
Nevertheless, NIE observed that the 275 kV line over the Glenshane Pass, which had been de-spacered and re-tensioned following the 1985 galloping incident, appeared not to suffer from conductor galloping. Previous to 2001, it was the only line where NIE had experienced galloping.
Accordingly, it is recommended that NIE investigate whether this should also be done to other twin-conductor power lines.
5 Customer Service and Communications
Of critical importance in any event of this nature is the service received by customers, both in terms of continuity of (or speed of restoration of) supplies, and in the provision of accurate information about how long repairs may take.
It is an indication of the importance of electricity to modern ‘western’ life that its absence – even for comparatively short periods – can cause major problems. It is a measure of the normal reliability of the system that many customers are unprepared for its absence.
After the Christmas storms in 1997 and 1998, NIE recognised that its customer communications were lacking. There was a clear need to be able to advise customers what was going on, and how long disruption may last.
Customers are not slow to ring in when the power goes off – but NIE cannot expect to cope with 100,000 doing so in a short space of time. To counter this, they have invested in an automated messaging system that recognises the number of the incoming phone, relates that to its geographical area, and plays a message relevant to that area. To do this, the message must be clear, informative and up to date.
Problems do arise. Firstly, and contrary to public belief, NIE do not necessarily know what parts of a distribution system have failed. In common with most electricity distribution companies, they are reliant upon customers calling in to report power failure. The automated messaging service referred to above can capture the phone numbers of incoming calls and associate those with particular geographical areas, whereupon a message relevant to that area is played.
Another computer system is used by NIE staff to look at the areas and spread of faults to predict the most likely cause. This process will be further speeded up over the next few years by the installation of a remote control system for the distribution network (dSCADA), which will give NIE controllers a much better view of which circuits have failed by providing remote control and indication of the operation of the distribution system.
Having established the areas without power, and which circuits have failed, it is usually necessary to go out and find the fault – which is neither the easiest nor quickest task in the middle of the night, during a blizzard, with roads blocked by snowdrifts.
Although field staff normally make enormous efforts to get to faults, until these are found, it is not possible to tell customers how long it may take to restore power. It is also well to bear in mind that there may be more than one fault on a given line, and weather-related mechanical damage can continue to be caused even after the power supplies have failed.
NIE nevertheless received significant criticism about the lack of available information. From a customer’s viewpoint, this is entirely understandable. Work is needed to ensure that customers are aware of what they can expect from NIE – and NIE need to assess how information flow can be further accelerated.
5.1 Call handling
NIE have the scope to take up to 112 calls from customers simultaneously, given enough call handlers. Experience shows that calls last an average of 3 minutes – so the maximum capacity to answer calls is limited to 2,240 per hour.
Much of the information given out by call handlers is repetitive, so an automated messaging system, which in this case handled almost 75,000 calls in one day, is a logical step to take.
Ofreg has held detailed discussions with NIE, and the following points are noted, and recommendations made.
· NIE need to do more to advise their customers what to expect in the event of major system disruption, and the value of the automated messaging system. In particular, there is need to emphasise that the information on the messaging service is just as up to date as that available to human call handlers.
· NIE accept that customers often have useful information they can give to NIE, such as the precise location of damage to poles and wires. They are to investigate means by which customers can leave voice messages with NIE so that this information can be acted upon.
· NIE should review the format of the messages used on the automated system. These messages should include
§ the time they were recorded
§ when they are likely to be updated
§ what is being done, and such information about timescale for restoration as is available
§ assurance that those messages represent the latest available information
NIE should also bear in mind that automated messages can become irritatingly repetitive to the customer. They should consider whether the first sentence should change, and / or a different voice be used, each time the message is changed. This would allow callers to realise immediately that a new message follows.
In this incident, 14% of calls were terminated by the caller in less than 10 seconds, which implies that the caller did not get beyond the first sentence of the message, and could have missed updated information.
The automated messaging system uses the number of the incoming call to select the appropriate message. Where the caller inhibits this facility, or uses a mobile phone, this cannot be done, and a general message will be played. Inevitably, this will not include detail of local areas.
5.2 Call handling fault
An error occurred in the call routing programme which resulted in eighteen incoming lines being connected to desks that were neither active nor manned. The result of this was that callers on lines 13 to 30 inclusive were met with an engaged tone. Callers 1 to 12 and above number 30 were answered by a call handler or by the automated messaging system, but 18 customers at any one time received the engaged tone.
This error occurred when setting up the extra lines and call handler positions early on the morning of Tuesday 27 February. The volume of calls was such that call handlers and the automatic messaging system remained busy and it was only when some callers complained of receiving the engaged tone that the fault was spotted. This fault lasted from 9.15am to 12 noon.
It is a measure of the level of calls that, despite 84 call handlers working and 74,801 calls answered by the messaging system, no less than 53,301 unsuccessful calls were made, many of these being repeat dials following receipt of the ‘engaged’ tone.
NIE have recognised the fault. This report notes NIE’s acknowledgement, and recommends that NIE report back within three months on the steps that they have taken to prevent any possibility of a repeat.
5.3 Lack of firm information
NIE accept that they will be unable to give any clear information until the faults have been found and repair time assessed. Nevertheless, they are aware that customers need better information than this, so that they can make their own arrangements. This report therefore recommends:
a) that NIE seek to accelerate the benefits that introduction of the distribution automation scheme (dSCADA) will bring
b) introduce means of warning customers – possibly via the messaging system – if it becomes likely that they will remain off power overnight. NIE have already advised Ofreg of their intention to do this.
5.4 Staff communication
NIE have acknowledged that their internal radio system is not as flexible as it might be. In particular, field staff have radios fitted to their vehicles, so they have to return to their vehicles to report their findings.
Where staff have had to leave their vehicles and walk over snowdrifts – sometimes for some distance – in order to find faults, this can cause significant delay in reporting back.
There is also the question of staff safety when they are working in atrocious weather, often in the dark, without immediate means of contacting their base.
Delays also arose through the numbers of staff seeking to contact their control rooms, often all trying to speak to the same control engineer.
Accordingly, this report recommends that NIE investigate how their internal communications system could be improved in order to allow staff the use of truly portable radios, and how a system of more local control of repair works could relieve the bottleneck of chanelling all communication from field staff through a single control engineer.
6 NIE repair performance
On the ground, NIE’s performance was generally good.
Their ‘trouble operatives’ already have four-wheel drive vehicles, and there is no doubt that these helped greatly in the bad weather.
NIE also hired in further four-wheel drive vehicles, and three helicopters were used, as soon as weather allowed, to survey lines for damage and to move staff and materials into locations inaccessible by vehicle.
Staff and contractors were also moved in to the worst affected areas from the west of the province, which was unaffected by the storms. This resulted in no less than 300 field and engineering staff being available to the Downpatrick area, the worst affected by the storms.
Use of staff was prioritised and controlled via the ‘TroubleMAN’ computer system which matches available staff skills to jobs, the repairs themselves being prioritised by reference to maximising the number of customers whose supplies could be restored quickly.
When compared with the aftermath of the Boxing Day storms in 1998, when these systems were not available, NIE consider that the time taken to complete the restoration of all supplies was reduced by a whole day. This is to be commended. Nevertheless, this report notes that, on this occasion, the bulk of the damage was in a comparatively small area.
Accordingly, it is recommended that NIE consider the likely outcome had the events of February 2001 been as widespread as the storms of Boxing Day 1998, and whether their systems and manpower could have coped with the resultant workload.
Staff ‘on the ground’ performed extremely well in difficult and tiring conditions. Their effort and commitment is to be applauded.
7 NIE system performance
Following the 1997 and 1998 storms, NIE introduced a major refurbishment programme on their overhead distribution network. No damage resulted from the heavy snowstorm in late December 2000 (when the temperature was well below the range in which ice accretion occurs) and it appeared as though the system was far better able to withstand high winds than before.
This time around damage was almost exclusively confined to conductors. Such pole damage as did occur resulted from the unbalanced forces on the poles following the snapping of conductors.
In 1997 and 1998 many poles broke off at ground level, often revealing high levels of rot. This time around, no poles broke at their bases.
It is also noted that those lines that had been fully refurbished appear to have performed much better than unrefurbished lines.
There is one particular cause for concern on system performance, and that is the number of 25 square mm conductors that snapped under ice loading. This report agrees with NIE’s intention to investigate whether 25 square mm conductors are suitable for use in areas of high wind speed, or areas prone to ice accretion.
It is also recommended that NIE investigate the use of insulated overhead conductors on 11 kV and 33 kV lines in areas known to be prone to icing, as these conductors appear to be more resistant to ice accretion than their uninsulated equivalents.
8 Special needs customers
NIE have put great effort into considering the needs of ‘special needs’ customers, providing them with up to date information and, in many cases, providing mobile generators where these were needed to run medical machinery.
This is commendable, though it takes up a significant staff resource to do it, and can place NIE staff in the position of having to make decisions (for example, about provision of generators in response to health needs) that they are not qualified to make.
There needs to be liaison between NIE and the health and welfare services in order to assist NIE to draw up a priority list for such customers, based upon medical assessment of need.
NIE should also reassess its policy of contacting all special needs customers within four hours of a power failure. Over 600 were involved in this incident, which would have needed a call rate of 150 per hour. More widespread interruption could double or treble this rate.
Perhaps there is scope to use an automated messaging system here. Could this in fact provide a better service, especially if the customer has a facility to call back?
It is clear that the NIE system performed better in February 2001 than it did in December 1998, albeit in different weather conditions.
It is also evident that NIE’s response to the emergency was better structured, making much better use of available resources than hitherto.
The problem of communicating effectively with very large numbers of customers has been addressed, although there is still room for improvement here.
It is noted that NIE were more proactive in this incident than on previous occasions. The work by the transmission system operator to spread generation around the system, as a precaution against further loss of major circuits, and the Network Operations decision to man up the incident room as the weather started to look worse than forecast, are commendable.
Inevitably, not everything went to plan. Major difficulties are listed in this report, and recommended actions are summarised here.
à NIE to investigate whether despacering or retensioning of conductors may improve resistance to conductor galloping
à NIE to consider the future use of 25 mm square conductors, as these seem particularly prone to ice damage
à NIE to investigate the use of insulated overhead conductors on 11 kV and 33 kV lines as a means of improving resistance to ice accretion
à NIE to consider whether their reactions would have been adequate had the spread of the storm been greater, affecting more of the province
à NIE to consider how the benefits of dSCADA can be introduced, in particular in linking information flow from dSCADA to TroubleMAN
à NIE to consider provision of portable mobile radios, that operatives can carry with them rather than leave in their vehicles
à NIE to investigate using a greater number of call handlers. Up to 112 is possible.
à NIE to investigate their telephone line connection systems, so that it becomes impossible to connect calls to unused phone lines
à NIE to reconsider the scripts used on the automated messaging service, so that it is clear to customers that the information is up to date
à NIE to consider means by which customers can be warned if their supplies are likely to be off overnight
à NIE to provide advice to customers on the levels of information they can reasonably expect during a system emergency
à NIE to provide a dedicated line by which customers may convey damage information to the company
à NIE to discuss with the Met Office why the ‘ice accretion’ warning came so late – in fact, well after the phenomenon had started. Does the Met Office use the correct parameters to predict it? Is allowance made for the altitude difference – and consequent temperature difference – between the Met Office measuring station and the overhead lines when assessing the likelihood of ice accretion?