141

WELLINGTON CENTRAL POLICE STATION

BASE ISOLATION MAINTENANCE

C. R. Smart1

SUMMARY

Wellington Central Police Station was built in the late 1980s. Its 10-storey tower block featured
robust X-bracing around the perimeter and strong floor diaphragms to distribute earthquake actions
amongst the perimeter frames, thus making it extremely stiff. To provide the ductility required by the
1984 Loadings Standard, the tower block was supported on a base isolation system that allowed up to
400 millimetres of movement in any horizontal direction.

This paper describes investigations into the continuing serviceability of the base isolation parts, and
the remedial maintenance required.

EARTHQUAKE DEFENCE: BASE ISOLATION

Figure 1: Tower block, view from Victoria Street. Pile types 1, 2, and 3 have spherical bearings of slightly different designs
near their upper ends. Plan is an excerpt from MOWD drawing 5/1132/1 7002 207/5.

The tower block features very robust diagonal X braces in the of an earthquake the tower block can move as a single unit. In
perimeter frame and strong floor diaphragms. Their purpose is an extreme earthquake, the block is designed to move up to
to provide great stiffness to the structure, so that in the event 400 millimetres in any horizontal direction.

1
Waterfront Engineering Ltd., Wellington, NZ.

BULLETIN OF THE NEW ZEALAND SOCIETY FOR EARTHQUAKE ENGINEERING, Vol. 46, No. 3, September 2013
142

Figure 2: Left - type 1 column, showing spherical bearing in painted sleeve; Right - type 3 column in which spherical bearing is
located below basement floor level near the top of the hollow casing.

The movement can occur because the block is supported on The piles also act as springs which limit the horizontal
reinforced concrete piles which pass down through hollow movements and provide a restoring force. If the massive tower
casings to the bedrock 15 metres below. The piles are founded block were supported on the spring-like piles with no other
in and most are rigidly fixed to the bedrock. At their upper restraining devices, the effect would be rather like a heavy
ends the piles are free to bend sideways in their casings. The rubber-tyred digger travelling on a highway: a small bump in
rotation at their upper ends is prevented from being the road causes the digger to bounce on its tyres through many
transmitted to the building by spherical bearings as shown in cycles of oscillation. Cars and trucks do not bounce because
Figure 2. Steel tendons pass through some of the piles to take they have energy absorbing devices known as shock absorbers
uplift forces and so prevent the block from “jumping off”. mounted alongside their springs.

As well as carrying the weight of the building, the piles allow In this building, the energy absorbing devices near the springs
horizontal movements. In the event of a large earthquake, the look rather like the telescopic shock absorbers in a road
tower block is designed to move bodily sideways, the piles vehicle. Instead of absorbing energy by forcing oil through
bending into a banana-like shape under it. Several cycles of holes, those under the tower block absorb the energy
oscillating motion are expected to occur. generated by earthquake motions by extruding solid lead
through a central streamlined orifice.
 143

Figure 3: Left - plan positions of lead extrusion dampers; right - arrangement in elevation. Details from MOWD drawing
5/1132/1 7002 205/3.

Figure 4: Lead extrusion dampers "on guard", and as they would be at their maximum excursion. The pair shown in the
drawing at right are compressed to their shortest length, while the one that disappears on the left of the drawing is
extended to its greatest length. The abutment shown in the photograph is midway between columns B3 and B4. Image
at right from DSIR drawing.

To summarise, the tower block is designed to move HISTORY

horizontally 300 millimetres during an earthquake with a
return period of 450 years, and 355 millimetres for a 1,000- Base isolation as a method of earthquake defence appeared in
year earthquake. Even more severe earthquakes would cause an American patent in 1930: this illustrated a six-storey
the tower block to hit its stops at deflections of 400 building without ductility founded on a tank filled with ball
millimetres. The X braces on the outside together with the bearings. The idea was revived in more practical form by Dr
floor diaphragms constrain the block to move as a massive R. Ivan Skinner at the Physics and Engineering Laboratory of
rigid unit. The piles in their hollow casings support the weight the Department of Scientific and Industrial Research in the
but allow horizontal movement: they act as horizontal springs late 1960s. During the 1970s several projects presented
in a mass- spring- damper system. The lead extrusion devices themselves as candidates for base isolation and the author of
absorb the energy of the shaking: they act as dampers. This this paper took part in the experimental development of
system of earthquake defence is of the type known as base several different types of energy-absorbing dampers. PEL’s
isolation. Dr Bill Robinson became interested in the technology from the
materials science viewpoint and invented the “lead-rubber
The purpose of base isolation is to give the building “a soft bearing” which provided a simpler and more compact form of
ride” during moderate and major earthquakes. No structural base isolation. Lead-rubber bearings are on display in the
damage leading to collapse is expected even if the oscillations “Quake Breaker” exhibition at Te Papa. One of PEL’s
are large. The oscillations are expected to be less violent than mechanical workshop technicians was Chris Gannon who
would occur with a similarly strong and stiff building rigidly helped to build the test rigs that simulated the earthquakes.
fixed to the ground. Unrestrained furniture and office
equipment may topple or slide off desks, but the structure is Wellington Central Police Station was designed by the
designed to be operational immediately after the earthquake. Ministry of Works and Development with Andrew Charleson
144

and Peter Wright playing prominent roles. Dr Skinner one of many topics discussed. That discussion recalled the
provided the base isolation expertise. These three presented a work done 22 years earlier, and the author started thinking
paper [1] describing the design to the Pacific Conference on about the solid lubrication system applied to the bearings at
Earthquake Engineering in August 1987. At the time of the the tops of the piles.
Conference the design was not quite finalised, so some of the
as-built details differ from those shown in the paper. Could the lubrication system have deteriorated so that instead
of acting as a lubricant it had become an adhesive?
Some years later Skinner, Robinson, and their colleague Dr
Graeme McVerry published their classic design manual on If the mating surfaces had somehow bonded together, thus
base isolation [2]. “seizing” the bearings solid, what would happen in the event
of an earthquake? There was no reason to suspect that this
BASE ISOLATION DETAILS: SPHERICAL BEARINGS had happened, but nor was there a reason to be confident that
it had not.
Near the upper ends of the base isolation columns are
spherical journals (upper part) and bearings (lower part), made Preliminary calculations
as steel castings, shown in Figures 2 and 3. In order to spread
the weight of the building as evenly as possible on the bearing To get a feel for the magnitude of the problem the author
surface, the castings were required to fit precisely together made some preliminary calculations based on the theory of
with a smooth surface finish. This metal cutting task was classical elasticity. They are not adequate for structural design,
accomplished in PEL’s mechanical workshop using but give an indication of the effect of a lubricant in the bearing
techniques originally developed for making spherical surfaces turning into an adhesive.
of tools used for grinding optical lenses and mirrors.
The first point was to calculate the relative movements of each
The task of lubricating the surfaces followed. From other work half of the bearing. If a 15 metre cantilever pile moves 400
at PEL, it was known that liquid oil, no matter how viscous, millimetres at its free end, the bending rotation is 2.3 degrees.
would slowly be squeezed out. This was known from As the radius of the spherical surface is 400 millimetres, this
theoretical work and simple demonstrations. Grease, which is angular rotation requires 16 millimetres of sliding movement
an emulsion of oil in soap, would fare no better. on the spherical surface. The central stop in the bearing allows
a little more than this.
The designers, including the present author, considered a
system of grooves into which high pressure oil would be If a spherical bearing seized solid, the effect can be idealised
injected at intervals of a few years. The designers rejected this by assuming the cantilever becomes a column of twice the
on the grounds that no-one would take seriously a building length, built-in at the top as well as the bottom. The central
maintenance instruction to oil the foundations each 29 deflection of the double-length column is then taken as 400
February. (Since the earthquakes in and around Christchurch, millimetres, the same as the cantilever. Elastic beam theory
a similar instruction today may have more credence.) shows that the force required to do this is eight times the force
required to deflect the cantilever. This means that the
A literature search led the designers to choose a solid lubricant sideways spring stiffness is eight times as great as designed.
system. The surfaces were to be coated with a tightly-adherent
layer of the manganese phosphate crystals well known to The natural frequency at which a building shakes has a strong
military armourers as a preservative and lubricant for gun effect on how much energy it will extract from an earthquake.
parts. A layer of molybdenum disulphide, also a solid
lubricant, in a greasy vehicle, was to be spread over the The natural frequency is given by
phosphate crystals, filling the spaces between them. Finally, a
ω1 = √(k1/m)
third solid lubricant, graphite, was to be deposited from an
aerosol. If the stiffness increases by eight times, the natural frequency
increases to
The present author arranged to do this work at the army base
at Trentham, using the vats of chemicals used by the ω2 = √(8k1/m)
armourers. The chemistry was found to be out of balance, and
the author’s colleague Dr Graham McNaughton used his skills ω2= 2.8√(k1/m)
to correct the mixture. The army provided the facilities; PEL
provided the chemicals and skills; no money changed hands. where ω is the natural frequency in radians per second, k1 is
The author personally put all the journals and bearings through the lateral stiffness of the cantilever piles in newtons per
the trichloroethylene cleaning tanks, the phosphating baths, metre, and m is the mass of the tower block in kilograms.
and while the parts were still hot applied the molybdenum
disulphide and graphite layers. Hence an eight-fold increase in stiffness gives 2.8 times the
natural frequency. In structural engineering it is customary to
The present problem think in terms of period, the reciprocal of frequency. Hence a
factor of 2.8 in frequency means a factor of 1/2.8 = 0.36 on
Late in 2011 the author attended a conference of the period.
Maintenance Engineering Society where solid lubrication was
 145

In 1987 the designers estimated the building to have a period As a result of the literature search, the present author
of about three seconds. Since 3 *0 .36 = 1.08, the increase in concluded that:
stiffness would reduce the period to about one second.
1. There was no evidence from the literature that
Inspection of the spectral shape factor curves from NZS suggested the lubricant in the spherical bearings
1170.5 [3] shows that reducing the period from three seconds could have deteriorated with age.
to one second on the soil type now thought to exist at the site 2. There was no evidence from the literature that the
[4] would increase the earthquake forces about three-fold. possibility of the lubricant in the spherical bearings
changing into an adhesive had ever been
These preliminary calculations show that if the lubricant in all investigated.
the spherical bearings turned into an adhesive the tower block
would experience earthquake forces about three times as great The author recommended that a 50 millimetre square be
as it was designed for. removed from the painted rubber sleeve shown in Figure 2 (or
one like it) and a sample of lubricant be removed for
LITERATURE SEARCH examination. Evidence of changes in the lubricant adjacent to
the steel surfaces would be looked for. At the time the client
The author consulted materials and surface coatings engineer declined to accept this recommendation, but since the
Willie Mandeno at Opus International Consultants, and earthquakes of July and August 2013 has accepted it.
structures manager Robert Davey. Opus had inherited the
Ministry of Works archives, and Mr Davey arranged for the Hence the question of deterioration remained open. The client
relevant drawings to be retrieved – most of the figures in this agreed that the investigation could be publicised in the
paper are extracts from the MOWD drawings. Mr Mandeno engineering community with a view to discovering evidence
arranged to use the Opus library facilities to search the of solid lubricant deterioration in other systems. This has been
literature. done through the Maintenance Engineering Society, with no
new evidence yet appearing. The present paper serves to
Out of 40 papers discovered, none reported deterioration of notify the Society for Earthquake Engineering of the question.
molybdenum disulphide (MoS2) solid lubricants with age. Mr
Mandeno was not surprised by this, as before starting the
search he had noted that molybdenum disulphide was a
naturally-occurring mineral that would not be expected to alter BASE ISOLATION DETAILS: LEAD EXTRUSION
with age. DAMPERS

Some reported deterioration with oxidation, but noted that the The dampers are installed in eight sets of three, giving 24 in
oxidation products were not abrasive. As air has been all, located as shown in Figure 3. Earthquake motions can
excluded from the spherical load-bearing surfaces of the force them to extend or contract up to 400 millimetres in any
bearings by grease which itself contains MoS2, as well as a horizontal direction, including misalignment as shown in
tightly-fitting rubber sleeve, oxidation is unlikely to have Figure 4. The abutments where three are attached are fixed to
occurred. the basement and thence to the ground. The ends where one
or two are attached are fixed to the underside of the tower
Deterioration was reported as a result of moisture. .The present block and hence are free to move with the tower but to restrain
author was aware of this danger when he applied the MoS 2 its movements. A small amount of vertical movement is
layer, and took care to do this work while the steel castings allowed to occur through SKF plain spherical bearings in the
remained at high temperature (about 60 - 70°C) from the ends of the dampers.
phosphating tank, thus ensuring that no moisture would
condense on the surfaces. Because of the MoS2 grease and the The dampers carry no weight and in the absence of earthquake
rubber sleeve, moisture is unlikely to have reached into the or wind do nothing. When the wind blows or an earthquake
load-bearing surfaces during the period of about 24 years in occurs, the dampers resist the horizontal forces applied to the
service. An inspection in October 2012 showed that the tower block. Sufficient force has to be applied before they will
basement area was dry, with no sign of leaks from the building move: the design force required to initiate extension or
services nor rainwater leaking in from the outside. contraction of one damper is ±250 kilonewtons. At forces
below the 250 kN yield force each damper behaves as a near-
There were reports of deterioration with temperatures of the rigid link, as shown by the central part of the trace in Figure 5.
order of 300°C. This is important when lubricating machinery,
but in the absence of fire such temperatures are inconceivable
in this basement.
146

Figure 5: Dr Jim Cousins recorded these hysteresis loops for damper Number 10 in 1989. The test rig used at that time could
deliver a stroke of ± 250 mm, with the initial force rising to 300 kN before dropping to the design value of 250 kN on
subsequent cycles.

Dr Bill Robinson and his colleague Dr Jim Cousins described the parallel project on the spherical bearings at the top of the
several forms of damper at the 1987 Pacific Earthquake piles. By extension, the author thought about the lead
Engineering Conference. Their paper included a report of tests extrusion dampers.
done in 1976 and 1986 on one of the smaller 140 kN dampers
at ± 120 mm stroke – the ten-year interval resulted in If the lead had somehow become bonded to the steel cylinder,
negligible change in performance, the desired result. thus “seizing” the damper solid, what would happen in the
event of an earthquake? There was no reason to suspect that
The only other occasion on which a retest had been attempted this had happened, but nor was there a reason to be confident
occurred in 2005. Mr Bob Stephenson of Opus Central that it had not.
Laboratories (a successor organisation to Ministry of Works)
tested a damper from a bridge. His equipment was able to The author approached Mr Gannon of RSL to enquire if the
apply only a “one way” motion, and the result showed “spare” damper still existed, and if so, could the RSL test
resistance without movement to about 40 kN, then a rise to equipment be used to conduct a new test. Answers were
100 kN at 20 mm deflection. affirmative, provided the test rig was modified to make it
capable of performing the required tests. The author designed
The present problem suitable parts, and Mr Gannon added a device that would
protect his load cell and hydraulics in the event of the damper
As recounted above, the 2011 conference of the Maintenance “seizing” and causing the force to rise catastrophically out of
Engineering Society discussed solid lubrication. That control. The author let a sub-contract to Acme Engineering
discussion recalled the work done 22 years earlier, and led to Limited, Petone, to make the parts.
 147

Figure 6: U-shaped adaptor frame supported fixed end of damper in the existing test machine at Robinson Seismic Limited.

Tests on damper No. 10 the test damper. A great effort was required to operate the
grease gun, resulting in some grease oozing from the nipple
The load cell was calibrated on a Baldwin test machine at fitting, but little if any getting into the damper. The
Quest Integrity Ltd. on 10 October 2012. The calibration significance of this is discussed later.
showed that its error was less than 0.5%, a value acceptable
for these tests. The initial test consisted of six extension and compression
cycles each ±150 millimetres with the instruments
The LVDT (displacement measuring device) built into the automatically recording force and movement. The machine
hydraulic ram was found to be unreliable, so was replaced cycled less rapidly than would be expected of a large
with a similar instrument mounted externally between the load earthquake and corrections for this are discussed later.
cell and the fixed part of the machine frame.
At the end of this series of cycles the central portion of the
An attempt was made to pump grease containing molybdenum damper, where the energy is absorbed, was warm to the touch.
disulphide and graphite into each of the six grease nipples on
148

Figure 7: Maximum extension (left) and maximum compression (right) on first test cycle. Figure 8 shows that the forces just
before and after the photograph at left was taken were substantially greater than those designed for.

Police LED No 10 Period - 65 sec Temp 23c

50

40

30

20
Force Tonne

10

-10 Series1

-20

-30

-40

-50
-200 -150 -100 -50 0 50 100 150 200
Displacement mm

Figure 8: Hysteresis loops for the first six cycles. The trace starts at the zero - zero position and then circulates in a clockwise
direction. The first cycle shows larger than design forces (+32 and -44 tonnes) but on the second cycle the forces fall
to design (about 25 tonnes). The force decreases on subsequent cycles as the lead increases in temperature and
becomes softer. (Note that 25 tonne force is almost identical to 250 kN.)

Twenty four minutes after the first six cycles were completed undesirable sudden spike in force as the damper moved past
the test rig was set to give an increased stroke of ± 175 its previous limit.
millimetres. The purpose of this was to look for evidence of an
 149

Police LED No 10 Period - 75 sec Temp 23c

50

40

30

20
Force Tonne

10

-10 Series1

-20

-30

-40

-50
-200 -150 -100 -50 0 50 100 150 200
Displacement mm

Figure 9: Hysteresis loops from a second series of tests. The behaviour shown is excellent, with no spikes in force, but a
reduced maximum (20 tonnes) as would be expected from the lead core having a higher initial temperature.

Significance of rate of test cycles and allowed the lead to form some sort of bond with the
surrounding steel cylinder. To break this supposed bond, an
Although the test rig could exert forces greater than necessary, attempt was made to inject new grease at high pressure,
it had insufficient power to oscillate the damper at rates slightly expand the cylinder, and so allow grease to spread
comparable to those expected in an earthquake. This problem between the lead and the surrounding steel. It is probable that
was anticipated during development of the lead extrusion this attempt failed.
dampers.
Mr Bob Stevenson of Opus Central Laboratories reported
The expression for rate dependence is somewhat similar problems when he re-tested a smaller lead
extrusion damper in 2005. On that occasion he found that lead
p = avb was protruding through small holes under the grease nipples,
and that it was necessary to drill out this lead before grease
Where p is the extrusion pressure (proportional to force), v is
could be injected – he then managed between 20 and 30
the speed of extrusion, and a and b are constants. For speeds in
strokes at each nipple [6].
the test and earthquake range, b has a value of 0.03 see
reference [5]. The fact that the initial half cycle required a force substantially
greater than that designed was consistent with the idea that
With the spherical bearings and dampers working as designed,
some form of bond had formed between the lead and the steel.
the tower block is expected to oscillate with a period of
The second half of the first cycle, and all subsequent cycles,
approximately 3 seconds. The cycles reported in figure 8 had a
occurred at about the design force, as did the initial tests 24
period of 65 seconds. The rate dependence formula suggests
years earlier.
that the forces at earthquake rates would be about 10% greater
than those at the test rate. Significance of high force during first half cycle I
Hence the maximum force in the first cycle of earthquake The end fittings of the dampers are made of a material known
shaking would be expected to increase from about 440 kN to as spheroidal graphite cast iron. Like all cast irons SG iron
480 kN. includes graphite in its metallurgical structure. In common
grey cast iron the graphite is in the form of flakes, sometimes
Significance of failure to inject grease
likened to small “cabbage leaves” that make it very brittle. In
Dr Cousins’ laboratory test records from 1988 show that all SG iron the graphite flakes are made to coalesce into tiny
the dampers were greased before their initial tests. The retest spheres that remove the brittleness of grey iron. Hence SG
in 2012 was prompted by the thought that during the iron possesses some ductility, much more than grey iron but
intervening 24 years the grease layer may have deteriorated not as much as steel. When the author designed these parts in
150

1986 he was keen for directional solidification of the castings sectional area from one end to the other. To check the design
to occur to minimise the possibility of defects peculiar to he conducted a finite element stress analysis to predict the
castings. To do this he detailed the parts to increase in cross behaviour during overload [7].

Figure 10: Stress analysis of end fitting shown in figure 4. Left image shows stresses at design load of 250 kN; Image at right
shows stresses at 492 kN. Darkest areas around eye in both images show material predicted to yield.

The left image in figure 10 shows the stresses at the design into the test damper then installed in the test rig at
load of 250 kN: the fitting is almost entirely elastic with only Robinson Seismic.
tiny areas of material (shown in a dark brown colour) that are 2. Wavefront Engineering, with assistance from Police
required to become ductile. The right image shows the contractor Transfield Services, should devise a
situation that would have occurred if the test had been method to remove one of the installed dampers,
conducted at earthquake speeds: the substantial dark (brown) grease it, repeat the tests described above, and
areas show material predicted to yield into the ductile state. reinstall it in the Wellington Central Police Station.

This is the situation that concerned the author when he The remainder of this paper describes how these
brought the matter to the attention of the Police. When the recommendations were carried out. To save costs, the
damper was new the maximum force was a little over 300 kN. dampers were exchanged on the same day, and the tests
During the 24 years of waiting “on guard” some sort of bond performed later.
had developed between the lead and the steel and this required
a force of 440 kN (480 kN predicted at earthquake rates) to Greasing
break the bond. The stress analysis shown on the right of
Figure 10 suggests that only a little more force would be Each damper is fitted with six grease nipples, designed to
required to yield the remaining elastic material. Fracture allow a molybdenum disulphide grease to be injected into a
would then be expected to occur. space created by the injection pressure between the lead and
the steel containment cylinder. A greasing operation had been
Significance of high force during first half cycle II done before the acceptance tests of 1989. An attempt to inject
new grease before the 2012 retest almost certainly failed.
The 250 kN design force on each damper keeps the tower Three reasons were discovered for this failure:
block in place during wind storms and small earthquakes. If
this threshold force did not exist, the building would move  The 1989 testing forced lead into the cavity under
about during wind storms and small earthquakes, an effect that each grease nipple, preventing new grease from
would be disconcerting to passers-by who would see that the getting in.
building had shifted its position in relation to the footpath, and  The non-return valve in each grease nipple had
to occupants who would sense unusual movements. become stuck in the closed position during the past
Conversely, if the threshold were too high, the base isolation 24 years.
system would be inactive during moderate earthquakes, thus  The original coupler that connected the grease gun
removing the benefits of the desired “soft ride”. hose to the nipple was made as a zinc die casting.
This was insufficiently robust to take the required
It was likely that the partial bonding demonstrated on the test pressure, and cracked. Leaking grease obscured the
damper also existed on all 24 installed under the building, and crack.
the author assumed that this was the situation. After
discussions with the Police, the author agreed to remove, Remedies for each of these problems were discovered:
grease, and retest one of the installed dampers to demonstrate
the value, or otherwise, of renewing the grease film in all 24.  Each nipple was unscrewed in turn, and the passage
under it cleared with a 2.5 mm drill. Care had to be
Damper exchange taken to prevent the drill advancing too quickly into
the soft lead and becoming stuck.
Following the tests reported above the Police accepted the  The non-return valves needed to be pressed inwards
following recommendations: a few times against their springs. A pin punch
proved to be a good tool. This was best done with
1. The author (trading as Wavefront Engineering)
the nipples away from the dampers so that free flow
should discover a method of injecting new grease
 151

of grease could be checked. Some nipples were Before attempting the exchange, the centre distance between
found to be incapable of either passing grease the connection pins was carefully measured with a steel tape,
inwards or preventing it exuding outwards: these and the No. 10 damper, still in the test rig at Robinson Seismic
were assumed to be internally corroded and were Limited, was set to this length. The length was then
replaced. deliberately set about 1 mm short, as it would have been
 A new coupler machined from solid brass worked possible to slacken the clevis fittings to get the pins in, and
much better, allowing grease to be successfully then tighten them again. An overlength error would have been
injected. impossible to correct on site.

Damper Exchange Entry to the base isolation area was gained through a
convenient hatch. A small crane of the type used for lifting
As each damper weighed 410 kg, too great for manhandling, engines out of motor vehicles was used to lift the damper up to
mechanical handling equipment was necessary. The author the height of the hatch.
and Transfield Services Ltd devised the broad outlines of a
plausible scheme and subsequently passed responsibility for The exchange, executed on 7 February 2013, is best described
the details to L&K Builders Limited. Police allowed the “sally pictorially.
port”, an outer door normally used to transfer prisoners, to be
used, thus eliminating the problem of getting the dampers up
and down staircases.

Figure 11: Left, damper No. 10 being rolled in through hatch. Right, damper turned parallel to installation position. In both
pictures the damper is supported on wooden blocks fastened to floor jacks.

Figure 12: Left, damper being transferred to trolley. Right, damper returned to floor jacks ready to be lifted to correct height
and rolled into position.
152

Figure 13: Left, equipment improvised to extract pin at right hand end from No. 21 damper. Right, No. 10 "spare" damper in
place previously occupied by No. 21.

Extracting and inserting the pins was worthy of note. The pins With the left pin out, there could be no residual force on the
made for the 2012 tests could be tapped in and out with light damper, and it was expected the right pin would move easily.
hammer blows with no load on the damper, but this was not This proved not to be the case. At the right hand end the jack
expected to be possible for a damper with 24 years’ service had to be on the edge of the beam above, and when a small
history for two reasons: piece of concrete spalled away the crew found it necessary to
improvise the system shown on the left side of Figure 13. A
 Experience suggested light corrosion would have piece of flat steel plate about 10 mm thick protected the
partially bonded the pins to the clevises. concrete edge and provided a small overhang to give a secure
 Small earthquakes known to have occurred may position for a truck-type jack. With this arrangement the pin
have left the dampers under loads approaching the moved in a series of jerks, each requiring 100 - 200 kN, and
“trigger” level of 250 kN. These loads may have each accompanied by a loud ‘bang’ which caused some
decreased through creep, but there was no way of anxiety to staff on the floor above.
knowing.
The crew found it impossible to insert the old pins into damper
The author considered building a reaction frame to surround No 10. Hence they substituted the new pins made for the 2012
the clevis and a hydraulic jack. Preliminary design showed tests – these featured a taper at their lower ends, and were able
that this would cost 3,000-4,000 dollars, and would require to be inserted with light taps of a hammer. Copious quantities
someone to spend an uncomfortably long time close to the of WD 40 penetrating oil helped.
dampers while it was fitted and removed. If the crew were
unlucky enough to experience a big earthquake while close to Tests on damper No. 21
a moving damper (or column), they would be likely to suffer
fatal injuries. Hence the author wished to minimise the time The author and Mr Gannon fitted the “service” damper, No.
dangerously close to the moving parts. 21, into the test rig at Robinson Seismic in February 2013. The
arrangement was identical to that used for “spare” damper No.
Building Maintenance Officer Jack McAllister suggested that 10 in 2012.
a jack placed between a pin and the beam above would
eliminate the need for the reaction frame and would minimise Three test runs were performed:
the time in the dangerous position. This appeared to be a good
 A stroke of ± 150 mm, extension first, with force
suggestion, and the author decided to adopt it.
governed by the damper
At the left hand end this worked very well. An Enerpac RC-  A stroke of ± 175 mm, extension first, with force
55 hydraulic jack had sufficient stroke to remove a pin in one governed by the damper, thus simulating a severe
push, a desirable state of affairs as it eliminated the time need aftershock
to insert packers of successively longer lengths. The pin  A stroke of ± 10 mm, with force governed by the
moved in jerks, each jerk occurring at a force of about 40 kN. damper, thus simulating a small aftershock.
 153

Figure 14: Damper No. 21 on test in 2013. Left: 150 mm extension on first test stroke. Right: 150 mm compression on first test
stroke.

Figure 15: Force-displacement hysteresis test loops recorded during acceptance tests on 9 December 1988. Stroke was
± 250 mm.
154

Figure 16: Force-displacement hysteresis loops from first test in 2013 on damper No. 21, showing "as-new" performance. The
damper heats up during cycling, thus softening the lead, and so each successive loop occurs at a slightly lower force.

Figure 17: Force-displacement hysteresis loops from second test on damper No. 21. The test machine was set to change
direction at ± 175 mm for this test, thus simulating a severe aftershock.
 155

Figure 18: Force-displacement hysteresis loops from third test on damper No. 21. The test machine was set to change direction
at ± 10 mm but was not “re-zeroed” for this test. This test simulates an early but small aftershock when the building
has been left slightly displaced (about ½ mm), thus leaving about 90 kN (~9 tonne) compression in the damper.

Following the successful demonstration of the efficacy of Hence all pins within tolerance will have interference fits with
greasing damper No. 21, the Police accepted the author’s all bearings within tolerance.
recommendation to leave spare damper No. 10 in position, to
keep “used” damper No. 21 at Robinson Seismic, and to The on-line catalogue gives two choices for pin tolerance:
grease the other 23 dampers whilst they remained in place.
 m6 for loads of all kinds with an interference fit,
The author wrote a work instruction for L&K Builders Ltd and with a suggestion of the even tighter interference n6
reiterated the safety precautions needed to work close to the for very heavy loads
columns and dampers that would move during an earthquake.  h6 for loads of all kinds with a clearance or
transition fit.
L&K Builders performed the greasing task and reported its
successful completion, including the previously undiscovered It is clear that the author’s 1989 choice of m6 for the pin was
need to replace some of the grease nipples. influenced by SKF’s suggestion of tighter interferences for
heavy loads. No records have survived on how the pins were
The author recommended to the Police that if no earthquake fitted (and the author does not recall), but a technique used at
violent enough to move the dampers occurred in the interim, the time was to shrink the pin in liquid nitrogen before
the greasing should be repeated in ten years’ time. dropping it into its hole. As it warmed it expanded to create
the interference.
Notes on pin fits
Following the original design, the pins for the 2012 tests on
The original manufacturing drawing for the end connection damper No. 10 were dimensioned with the oversize m6
pins specified a diameter of 50 mm with tolerance m6. Charts tolerance. The manufacturer of the test equipment (Acme
for the International Standards Organisation system of limits Engineering Ltd) noted that this would require an oversize bar
and fits show that m6 on a 50 mm diameter allows a to be machined down to size. They thought this a needless
maximum diameter of 50.025 and a minimum of 50.009 mm. expense, and suggested making the pins from a standard bar
which would have a maximum size of 50 mm. The author
These pins go through SKF spherical plain bearings
accepted their suggestion.
designated GEH 50 ES-2RS. The SKF on-line catalogue
gives the deviation of the mean bore diameter Δdmp as a The pins extracted with difficulty from damper No. 21 were
maximum of 50.000 and a minimum of 49.985 mm. given a taper and had a very light skim taken off their
diameters. They then fitted easily and performed well under
test.
156

If new pins are ever to be made, it would be best to accept

SKF’s second recommendation and specify an h6 tolerance.
This would mean a maximum diameter of 50.000 mm and a
minimum of 49.981 mm

POSTSCRIPT

On Sunday 21 July 2013 an earthquake magnitude Mw 6.6

occurred in Cook Strait, off the Marlborough town of Seddon.
This caused alarm, a few injuries, and minor damage to some
buildings in Wellington.

Inspection of the base isolation system showed that the

dampers had moved a few millimetres, probably for the first
time in service, and the entire building had rotated clockwise
about 10 millimetres at its periphery.

The columns responded elastically, with no evidence that the

spherical bearings tilted. It is probable that the movements
were insufficient to cause the bearings to break free of their
friction, but the possibility remains that a seizure, of the type
feared late in 2011, has occurred. Investigation has been
recommended.

REFERENCES

1. Charleson, A.W., Wright, P.D. and Skinner, R.I.,

“Wellington Central Police Station – Base Isolation of
an Essential Facility”, Pacific Conference on Earthquake
Engineering, 1988, Vol. 2, pp377-388, NZSEE.

2. Skinner, Robinson and McVerry: An Introduction to

Seismic Isolation, Wiley, 1993.

3. New Zealand Standard, (2004), NZS 1170.5, Structural

Design Actions , Part 5 Earthquake actions, Figure 3.1.

4. Semmens, S., Perrin, N.D., Dellow, G.D. & Van Dissen,

R. “NZS 1170.5:2004 Site Classification of Wellington
City”, Figure 3B, Ninth Pacific Conference on
Earthquake Engineering, April 2011, Auckland, NZ,
NZSEE.

5. Robinson, W. and Cousins, J., Pacific Conference on

Earthquake Engineering, Wairakei, August 1987,
NZSEE.

6. Stevenson, R.B., “Testing of a Lead Damper” Opus

Laboratory Report, June 2005.

7. Sandford and Smart, C., “Stress analysis of 250 kN

extrusion damper’s long end fitting”, Physics and
Engineering Laboratory Report S10151, November
1986.