A 100m3 tank at Cookie Time Templeton (24km from epicentre) survived the 'quake.

A 100m3 tank at Cookie Time Templeton (24km from epicentre) survived the 'quake.

After the Darfield earthquake - standard 1170 is not enough!

Monday, 29 November 2010

The Canterbury earthquake of September 4 has provided an opportunity for the manufacturer of timber water storage tanks in New Zealand, Timbertank Enterprises Ltd., to study the impact this unique event has had on their tanks used in fire systems, town supply and industrial installations.

Timbertanks' managing director Justin Jordan and the company's consultant engineer Waldo Granwal, met with engineers from Holmes Consulting Group and Canterbury University College of Engineering, to discuss the impact on tanks and timber structures of the earthquake.

There were numerous reports of damage and collapse to water storage tanks of all makes and while the study will be ongoing, some initial findings and conclusions were reached. A major conclusion is that tanks in the Christchurch region will in future require special design.

"The earthquake was very localised and characterised by the underlying very deep alluvial shale layer creating an unusual action of a strong pressure wave followed by a long period of low frequency secondary shear waves that caused most of the damage in water tanks," said Waldo. "It was well beyond any previous expectations specified by the Code at the time for the design and construction of such tanks. It exceeded it by a large margin - three to four times more than would have been expected.

"We achieve the code standard as a minimum. If tanks were built to exceed the standard to withstand such a quake, the cost factor would outweigh the economic benefit - double the cost. Perfection is attainable, but it comes at a cost."

Timbertanks inspected a total of 25 tanks that it built over the last 30 years in the region affected by the earthquake. Of these, seven tanks lost their water as a result of the 'quake. The remaining 18 performed as expected - shuffling approximately 20mm as base isolated tanks are designed to do.

Four out of the seven affected tanks had base isolation compromised by rigid pipe attachments or illegal ground obstructions, leaving three tanks that the company "would have liked to perform better." Certain size diameter tanks only were affected, others around the epicentre that performed perfectly had a different diameter.

The locations of all the tanks are shown on the attached map of the region, showing their proximity to the epicentre and to those areas where liquefaction affected structures the most. Three were in the Hornby/Islington area.

It took until a month out from the event to firstly inspect tanks, then start understanding what had happened and work through the issues to try and find answers. It evolved with different inputs being received almost every day. Subsequent after shocks also exacerbated the situation at a couple of sites.

"Of course we would have preferred the tanks to have performed better, but with the magnitude of the 'quake being more than anyone expected and some owners not maintaining the requirement of a base isolated system... quite frankly at this level we would almost expect failure," said Justin.

"We were caught out by the extent of the sloshing - the wave velocity acceleration which caused the tank to move around," he added. "In all of this, we're learning new things and it takes a while to understand what happened and then translate into what can we do better - for us to set up our tanks so they can slide freely and control the wave slosh.

"Everyone is trying to react quickly to this earthquake but we also have to ensure we move correctly - because we may have to change our design as we learn more."

An initial report by AON New Zealand - the Sprinkler Certification Authority - on tanks inspected after the event, noted however that the current method of piping Timbertanks via a buried pipe rising into an anti-vortex plate in the centre of the tank 'appeared sound'. The consequence of the pipe work not being articulated only caused further damage.

"We agree with the report from AON regarding sprinkler system performance, of the need to introduce flexibility into piping design where deflection due to seismic movement is anticipated. In the Christchurch situation we need to ensure these flexible knuckles and similar critical features are built in," he said.

The team spent six days collecting information. All the damaged tanks were fire tanks, all of a similar size over 10-13 metres across, which appears to equate to the sloshing frequency that occurred with the water over periods of about three seconds. (Steel tanks of the same proportions had the same problems).

"We acknowledged that the integrity of fire protection systems is essential for such as major hospitals, CD headquarters, prisons or airports. We are working with the Fire Protection industry to re-establish those tanks as soon as possible, within the required code."

So is there an answer to this whole earthquake problem, given that the law requires that a fire tank be designed to retain its contents and be useable for purpose?

"Anyone who claims that a tank designed to earthquake code 1170.5 would have survived this rogue event does not understand the whole picture!" says Justin. "Structures susceptible to these large secondary waves will need special attention beyond code 1170. In the past, base isolation has neutralised tank slosh - so in Christchurch a tank with a diameter at risk will have dampening baffles installed."

"Also in the past we have found it very difficult to go to a customer and say we believe it's a very good idea that you spend some money to make your tank earthquake ready. In the 30 years since it was built, if a tank has not been maintained in a satisfactory manner to meet the current code, you can expect problems. And they occurred. This has highlighted the need for more than normal maintenance to preserve a tank and make it earthquake ready.

"We will look at better signage around tanks to ensure they are kept free of debris, earth, structures built up against them etc., which would inhibit the base isolation from performing as it should.

"We will strongly recommend that companies have a maintenance contract which will provide a warrant of fitness that the tank meets all the codes for performance in earthquake according to the existing standards, that they meet the designer's and manufacturer's specifications for purpose."

Could this happen to other large Timbertanks around the country - for example large timber Council reservoirs?

"No we don't believe so," said Justin. "We had a tank at Templeton where the highest reading of the earthquake was taken and it is working today. The phenomenon that amplified the 'quake in Christchurch is not to our current knowledge present in any other location."

As the chaos of Canterbury recedes, a responsible review of existing clients and their tanks will be a key aspect of Timbertanks' future work with the Holmes Consulting Group to assure earthquake readiness of the product. Their design analysis may evolve to indicate a change to code specifications.

"Any designer is expected to adjust upwardly if present knowledge has emerged which suggests otherwise - we are obliged to do just that."

In a recent statement, United States company Caldwell, which has been building water tanks since 1887 stated: '...only a tank manufacturer fully understands the structure's design and ongoing needs that ensure optimal life. Full-service tank manufacturers now offer extended warranties on new tanks, provided that the tanks are regularly cleaned, inspected and repaired. Safety improvements and upgrades are also critical to the tank's long-term usefulness. Life cycle cost comparisons based on a 40 year time frame fail to recognise that a well designed tank can be expected to provide useful service for more than a century.'

Justin adds - "there is a significant cost to upgrade a tank to meet new conditions and we will be tackling it on two fronts. We are determined to hold on to our base isolation."

Why did 18 tanks withstand the 'quake? Those that came through with minimal or no damage related to their locale e.g. they were on the hillside; they were base isolated and fit for purpose. Large Timbertanks' water storage tanks in the region at West Moreland and Springhill, also one in Ashburton, suffered no damage or loss of function as a result of the earthquake.


Timbertank performance during Darfield 2010 earthquake

- comments from a report by Trevor Kelly - Holmes Consulting Group

'The Christchurch earthquake motion at the sites where tanks failed had a combination of high energy content in the long period range, presumably due to underlying deep alluvium, and high vertical accelerations in the short period range. The long period motions cause high surface waves in the tanks which apply large convective forces to the side of the tank. As these forces are applied high up the staves, they tend to cause overturning rather than sliding. This mode of response is undesirable as it permits the line to extrude beneath the staves where it is damaged.'




'Tank evaluation procedures were developed in the 1990s partly using observations from the most severe earthquake to which they had been subjected up to that time - the 1987 Edgecumbe earthquake. The Christchurch sites at which tank failure occurred had much greater demands than Edgecumbe in some respects.' - Holmes Consulting Group

Earthquake Location Map (PDF) showing location of water tanks which failed and those which survived


A 400m3 tank at located at Northland Mall Papanui (35kms from epicentre) moved off its site and ruptured the liner.