PRINCIPAL EARTHQUAKES IN NEW ZEALAND IN 1994

Earthquakes have occurred in New Zealand more frequently in the last five years than in the previous decade. This is not a cause for alarm, but a demonstration of the sporadic way in which they happen. It is clear that during the 1970s and early 1980s we experienced fewer large earthquakes than normal. Earthquakes in 1994 followed the established pattern of an active region from the Bay of Plenty to the northern South Island. We have again been fortunate that large earthquakes have not occurred close to our large cities.

The July 12 shock was also felt as far south as Christchurch, but only at moderate intensities. Effects near the epicentre were not as great as near Wanganui on September 9, because of the different geological basement in the Taupo area.
There were two more earthquakes exceeding magnitude 6 offshore: 140 km north-east of East Cape on November 20, and 40 km north of Tauranga but 285 km deep, on November 16. Neither was felt very strongly onshore.
On January 29 there were two shallow earthquakes within five hours, both about 15 km south of Westport in the area of the lower Buller Gorge. They have been named the Hawk's Crag earthquakes, and were the subject of a field survey and subsequent analysis. Their magnitudes were 5.6 and 5.8. They caused some chimney damage in Westport, Waimangaroa and Inangahua. Goods were displaced off shelves in Reefton, Hokitika and Greymouth. Felt reports have been received as far away as Paraparaumu and Christchurch. There were a few aftershocks, but rather less than expected for main shocks of this magnitude, and this is an item of current investigation. 1

Seismological Observatory DSIR Geology and Geophysics Wellington
Occasional aftershocks of the earthquakes near Weber in Southern Hawkes Bay, in February and May 1990, have continued, all of them small. Aftershock sequences such as this do tend to persist for many months after the main shocks.
On February 15 an earthquake of magnitude 5.5 occurred off the coast from Greymouth. It was felt throughout much of the northern and western South Island, although no damage has been reported. This is at the southwestern end of the Main Seismic Region: between Greymouth and Milford Sound, earthquakes are less frequent despite the presence of the Alpine Fault which runs much of the length of the South Island.
A deep earthquake (106 km) occurred to the south of Patea on 9 June. The magnitude was 5.8, and it was felt from New Plymouth to Greymouth. Earthquakes at that depth are common in the South Taranaki Bight.
Southern Fiordland experienced an earthquake of magnitude 5.0 on September 5. It was centred near the head of Dusky Sound, and was reported felt as far east as Dunedin. Focal depth was 74 km.
Goods were shaken from shelves in Ruatoria on 31 October, when an earthquake of magnitude 5.0 occurred near East Cape. At the same time a swarm of very small earthquakes was in progress near Kaikoura, attracting considerable public attention. This phenomenon of a swarm of small events is one which needs further study in New Zealand. It is simply not known how often these occur, because the sensitive network which can now detect these has not long been in place, a product of the recent upgrading of instrumentation. On such occasions, there is always an outside chance that the swarm will develop into a major earthquake, but the process is not understood. It is fortunate that the most common outcome seems to be that the swarm dies away within a few days, as indeed happened in Kaikoura.

INTRODUCTION
This report has been prepared as part of those prepared by a 3man earthquake reconnaissance team sent to San Francisco by the New Zealand National Society for Earthquake Engineering in the first week of December 1989. The author's brief was to cover the damage to power generation facilities.

DESCRIPTION OF DAMAGE AT THE SUBSTATIONS
Most of the damage occurred at San Mateo, Metcalf and Moss Landing sites. By the time the author visited these sites, the damaged equipment was cleared and repairs were well under way. The locations of these sites are shown on Fig.1.
In this report, the photographs of the damage were taken by E. Matsuda of PG&E, while other photographs were taken by the author.

Damage at San Mateo Substation
San Mateo is a 200/115/66 kV substation located about 30km south of Down Town San Francisco. It was built on flat and relatively soft ground alongside Freeway 101. Its location is about 65km north of the epicentre.
The earthquake intensity experienced at San Mateo has been estimated at MM7. Ground acceleration was estimated at 0.2g.
However, the closest strong motion record was made about 3.2km away in Foster City and showed a peak horizontal acceleration of 0.16g(l).
Director, AC Power Group Ltd Wellington, NZ.
The major damage at San Mateo was to one General Electric (GE) model ATB4 and three GE model ATB4 and three GE model ATB7 live-tank 220 kV circuit breakers. The breaker support insulators were cracked at the base as shown in Figures  2 & 3, in a classic bending failure mode of a weak column supporting a heavy interrupter head at the top. These breakers were replaced with dead tank breakers (Fig.4).
There was other minor damage as described in Ref.1. However, most items of the equipment, including 230 kV bulk oil circuit breakers (Fig.5), withstood the earthquake.

Damage at Metcalf Substation
Metcalf is a 500/230/115 kV substation located about 5 km south of Down Town San Jose. It was built on flat but harder ground than that at San Mateo. Its location is about 30 km north of the epicentre.
The earthquake intensity experienced at Metcalf has been estimated at MM 7. Ground acceleration was estimated at 0.3g.
However, records obtained about 11 km southeast and 11 km west of the substation showed peak horizontal accelerations of 0.28g and 0.19g respectively (1).
Damage at Metcalf substation mainly occurred in the 500 kV switchyard. The damaged equipment included 500 kV live-tank circuit breakers, current transformers, lightning arresters and a transformer. However, items in the 230 kV switchyard and the control building withstood the earthquake without any damage. A 500 kV bushing on transformer 11 C shifted on its support which put it out of service (Fig.6). This transformer was replaced by the spare unit.

ii)
Severe oil leaks at lhe flange joining the top header of the readiator to the main tank of transformer 12C (Figures 7 & 8). This occurred as the radiators swayed during the earthquake.
iii) Two out of seven 230 kV transformer-mounted lightning arresters associated with bank 11 cracked at the base and fell to the ground ( Fig. 9).
They were replaced by the units from Bank 12.

kV live-tank Circuit Breakers
All three of the air blast WE circuit breakers were damaged at the base as shown in Figures 10 & 11. The hold down clamps also failed and allowed the circuit breakers to shift up to 250 mm horixontally (Fig. I I).
There were three columns per circuit breaker. Each consisted of four porcelain cylinders, stacked on top of each other, separated by rubber gaskets (I) and an interrupter head at the top. The whole assembly was held together and secured to the support base by post-tensioned fibreglass rods. These rods replaced the original wooden rods as an attempt to strengthen the circuit breakers (Fig.12). Unfortunately, this did not prevent the porcelain bushing from extensive cracking despite the fact that the rods were not damaged.
It should be noted that the Hitachi 500 kV dead tank breaker ( Fig.13), which was tested to 0.5g ground acceleration using the sinewave method, was undamaged.        Damage at Moss Landing substation mainly occurred in the 500 kV switchyard. The damaged equipment included 500 kV livelank circuit' breakers, current transformers, disconnectors, buswork, CCVT (coupling capacitor voltage transformers) and linetraps. Other items in the same switchyard such as power transformers, the control room and the control building withstood the earthquake undamaged. In the 230 kV switchyard, the major damage was to the disconnector switches supported on the gantry structure.

Linctrap and CCVT
One line trap out of four and all four CCVT's associated with the linetrap failed when the support porcelain bushings broke at their base (Fig. 19).   Out of a total of I 4 three-phase disconnector switches units, 12 (three-phase units) were damaged. (Figures 20 & 21).
The two three-phase units connected to the surviving dead-tank circuit breaker did not experience any damage. This suggested the damaged units were pulled down by the damaged CT's and circuit breakers connected to them.

kV buswo1·k
A Jong span of the 500 kV aluminum busbar came down during the earthquake (Fig. 22). This took place because the support insulator broke at the base and the fitting attaching the bus to the top of the insulator also failed.
The buswork was repaired within two days.

kV disconuector switches
Most of 230 kV disconnector switches supported on gantries were found to be misaligned after the earthquake. On one bay (out of 10), a few disconnector switches were damaged. The support porcelain broke at the base (Figures 23 & 24).

Moss L'lnding Power Station
Moss Landing Power Station is next to the switchyards, damage or which has been described earlier. It is an oil and gas-fired power station. There are 7 units : the 2x750 MW units were built in the late 1970's, the 3xll0 MW and 2xl20 MW units were built in the early 1950's.
The 2x750 MW units are Units 6 and 7.
Unit 7 was out of service at the time but Unit 6 was in operation. It was put out of service by the earthquake. It was believed that the shutdown done manually. Other units did not experience any significant damage.
The damaged equipment included water tanks, chimney ties, pipe hanger rods, feed heater supports and air preheater bearings.
In general, the damage was not considered serious. The cost of damage was estimated at $US 5 millions.

Water Tanks
One 1,135,500 litres (300,000 gallon) tank of approximately 16 m in diameter and 12 m high, fabricated from steel plates, was badly damaged. The damage occurred at the base, which was corroded, and allowed water to escape. This created a drop in pressure inside the tank and caused the tank plate to buckle (figures 24 & 25). The internal roof structure was also damaged.
In itself, the damage of this free-standing tank was not serious. However, as it stored water for fire fighting, the consequence of its damage would have been much more serious if fire had broken out at the site.
It should be noted that there were other tanks at the site which survived the earthquake. The large fuel oil tanks, for example, did not experience any damage.

Chimney ties
The 152 m tall chimneys for Units 6 and 7 consisted of a concrete stack with a steel liner inside. The liner was supported against the concrete outer stack by tie rods at JO m intervals up the stack. The relative movement between the liner and the stack caused the rods to break at the anchor points, both at the liner and the stack ends. However, the majority of the damage was at the liner end.
The damage to various degrees occurred at both chimneys, but did not render them inoperable.
The repairs involved the replacement of the tie rods with a new design, using a bumper system, which allowed for relative movement during earthquakes. This is in line with the current USA design practice.

Steam pipe support damage
There were significant lateral movements in steam piping, especially in the main steam pipe. This caused numerous damage to support hangers, lagging and guides. Spring loaded supports experienced damage to the bolts securing the spring base to the anchor. Some springs actually broke.
There was also damage to boiler insulation and leakage from boiler tubes.

Feed heater support
The tension-only braces of the three horizontal feed heater's steel support stands were buckled. Concrete support pedestals of other feed heaters were cracked.

Other damage
One low pressure (LP) turbine bearing was damaged when the unit was shut down after the earthquake. This was caused by the loss of AC power which was required to supply the lubricating oil to the bearing.
One air preheater shaft was shifted out of its pedestal.
The steel brace joining the steel structures of Units 6 and 7 was buckled and the base of the cooling water plant crane was shifted.
Overall, the damage was not considered serious.

Potrero Power Staiion
Potrero power station has lx220 MW steam turbine (oil/gas) and 3x50 MW gas turbines. The steam unit was commissioned in 1965 and the gas turbines in 1976. The loss of steam caused the steam drum temperature to increase to the point where a differential temperature between the feed water condensate and the steam drum was too high to allow the boiler to be restarted. Fortunately, a US Navy ship was in port at the time and it was used to provide hot steam to heat up the feed water. It took nearly 52 hours before the boiler was fired up again.

Hunter Point Power Station
Hunter Point power station has 2x110 MW and lxl70 MW steam turbines and lx50 MW gas turbine. The steam units were commissioned in 1949 and 1958 respectively. The gas turbine was a recent addition in 1976.
At the time of the earthquake, 2xl 10 MW units were operating and, at about 1hr 40 mins after the earthquake, there was a malfunction of relays which caused an overload of the house unit and a subsequent loss of auxiliary power supplied to the main generator. The whole power station lost power completely. Power was not restored until one hour later. The back-up generator failed to start as the vibration monitor did not switch to DC. The inverter failed and no AC power was available.

DAMAGE TO OTHER NON-PG&E POWER STATIONS
At the University of California, Santa Cruz, the 2.5 MW diesel plant survived the earthquake without any damage. The ground acceleration was measured at 0.4 g. Problems occurred when the unit could not keep up with the load when PG&E power supply was lost.
The 50 MW Cardinal Co-generation plant near Stanford University was designed to withstand 0.5 g ground acceleration.
It was built between two and three years ago and withstood the earthquake successfully. Ground acceleration was measured at 0.3 g.
The 120 MW combined cycle (50 MW gas, 70 MW steam) at Gilroy was built in 1987 and withstood the earthquake without any damage. There was some leakage from the 115 kV substation which caused a forced outage for two hours. The ground acceleration was measured at 0.4 g in a school house at about one kilometre from the plant.

CONCLUSIONS
While the damage, as illustrated by the photographs, appeared extensive, it should be viewed in perspective.
Within the area where intensities MM 7 and greater were felt, there were an estimated total of 100 or so substations of various voltages, 1000 or so kilometres of high voltage overhead transmission lines and ten or so power stations. Out of this total system, it would be difficult to refer to the damage as described above as extensive.
As experienced in past earthquakes : 2.
The damage to outdoor substation equipment was more extensive and severe than to power station equipment.

3.
Live tank circuit breakers were more susceptible to earthquake damage than dead tank circuit breakers.

4.
Equipment of 230 kV and 500 kV voltages were more likely to be damaged by earthquakes than those at 115 kV and below. In fact, there was no damage to equipment in this lower voltage rating.

5.
The damage was mostly to older equipment of at least 25 years old. Equipment and facilities built to modem seismic design practice suffered little or no damage.

RECOMNIENDATION
The Loma Prieta earthquake has shown that power systems, in general, are quite fragile and it would not take much damage to put them out of operation.
It would be impractical, if not impossible, to ensure that a power system with a mixture of old and new equipment like that of PG&E to remain operational during earthquakes of Loma Prieta magnitude.
It would be more realistic for power system owners to develop an earthquake protection programme based on seismic performance criteria set, not only by earthquake design loads, but also by an outage time, say 48 hours. Once these criteria are selected, the owners would formulate an implementation programme to ensure that their system is designed, strengthened and carries adequate redundancy or spares so that power can be restored within this outage time.
To formulate an implementation programme, a risk assessment of the power system taking into account the structural vulnerability of each component in the system, its importance in the network and the earthquake hazard including local soil conditions where the components are located should be carried out.
7. ™PLICATIONS FOR NEW ZEALAND 7.1 Our power system equipment are rated at 230 kV AC and below, which are structurally less vulnerable to earthquake damage than 500 kV equipment at Moss Landing and Metcalf substations.
7.2 Our current seismic design requirements for power system equipment, which have been applied since 1971, are on the same level if not higher than those specified by PG&E.

7.3
If a Loma Prieta-magnitude earthquake were to occur in Wellington, we would expect a similar if not higher level of damage to our power system. This is because the vulnerable items are older transformers on wheels which would take a longer time and more costly to repair if they come off the rails (2).