Preliminary Report on Steel Building Damage from the Darfield (New Zealand) M7.1 Earthquake of September 4, 2010
Steel structures in the Canterbury area suffered little damage. The photo above shows steel bracing with a fractured connection at a warehouse in Rolleston.
MCEER investigators Michel Bruneau (Professor, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo) and Myrto Anagnostopoulou (SEESL Structural Engineer, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo) conducted post-earthquake investigations in New Zealand on behalf of MCEER and EERI's Learning from Earthquakes Program (funded by the National Science Foundation).
The following preliminary report was submitted by Bruneau; Anagnostopoulou; Greg MacRae (Associate Professor in Structural Engineering, Dept. of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand); Charles Clifton (Associate Professor in Structural Engineering, Dept. of Civil Engineering, University of Auckland, Auckland, New Zealand); and Alistair Fussell (Senior Structural Engineer, Steel Construction New Zealand).
Many different interpretations of disaster resilience have been proposed in the literature. In one such concept, which has found broad acceptance, disaster resilience has been described as being a function of four R's, namely the robustness and redundancy of the infrastructure in its ability to limit damage, and rapidity and resourcefulness in returning the affected area to its pre-disaster condition (Bruneau et al. 2003). If anything, the Darfield earthquake showed that robustness of the infrastructure is a cornerstone in achieving disaster resilience that a high resilience level can be achieved by a society when the extent of structural damage is limited. The rate at which a community returns to its pre-disaster condition (i.e. the rapidity dimension of resilience) was again demonstrated by this earthquake to be intrinsically coupled to the ability of the infrastructure to withstand the disaster without debilitating damage (i.e. the robustness dimension of resilience). In the case of the Darfield earthquake, relative to expectations for a Magnitude 7+ earthquake, damage was moderate, making the community quite resilient. In fact, if not for the failures of unreinforced masonry buildings and damage consequent to soil liquefaction, the response and recovery requirements following this earthquake would have been minor.
However, in this case, preliminary investigation suggests that the predominantly good performance of modern engineered buildings of various construction materials and vintage in the affected region (including steel buildings), while partly attributed to the existence of effective seismic design requirements, is attributable to a significant degree on the fact that seismic demands from this earthquake were less than those corresponding to the design level, especially for structures in the Central Business district of Christchurch with as-built first mode periods of under 1.5 seconds. See more on this below. Consistently, steel structures in the Canterbury area have suffered little damage, and this damage consisted of either slight evidence of plastic yielding, damage to concrete elements in lateral load resisting frames made of both steel and concrete, isolated instances of connector failures, and collapse of steel tanks and industrial steel storage ranks. The following provides an overview of this damage. Note that an all inclusive survey of the performance of all steel buildings exposed to severe shaking has not been conducted. Confidential communications reporting evidence of minor inelastic behavior and plastic hinging in buildings started to emerge about week following the main shock, but specific details related to these cases have often not been disclosed in the public domain. These communications advise that none of the inelastic demand is sufficient to warrant repair or replacement of steel members, however some bolts in high strength structural bolted connections may be replaced as a precautionary measure.
Also note that there are significantly fewer medium- to high-rise steel buildings in the affected area than reinforced concrete ones, as a consequence of two factors:
- A strong tradition of seismic design using reinforced concrete and capacity design principles, as the legacy of Professor’s Park and Paulay who developed many of these concepts at the University of Canterbury in Christchurch in the 1970s and 1980s.
- A reticence to build multi-story steel structures following labor disruptions by steel erectors that made steel construction crawl to a rest in the 1970s, and hampered the steel construction industry for the better part of the following two decades. However, that legacy disappeared during the 1990s, resulting in more multi-story steel framed buildings built since around 2000.
Comparison of Earthquake Intensity with Design Intensity
It is possible to make comparison of the earthquake intensity with the design intensity through comparing the 5% damped spectra from strong ground motion stations throughout the region with the elastic design spectrum CZ from NZS 1170.5. Preliminary results from this comparison as of September 13, 2010, show the following:
- There is very considerable variation in intensity throughout the region.
- Modification from the underlying soils is significant.
- In the Central Business District:
To the northwest/west/southwest of the city (i.e. near the epicentral region) the intensity was up to 100% ULS (see Figure 1b).
To the northeast and east of the city the intensity was under 50% ULS but there was very significant soil instability and ground movement in these regions (see Figure 1c).
A significant aftershock on September 8, 2010 (four days after the main shock), caused higher spectral accelerations in the city, for some periods, than did the main shock.
- ground conditions were stable
- the intensity was approximately 60-70% of design ultimate limit state (ULS) for periods of under 1.5 seconds and increased to 100% ULS for periods over 2 seconds, especially in the north-south direction (see Figure 1a for the clearest example of this)
- there was a noticeably stronger north-south component
Figure 1a. From Christchurch hospital (soil type D)
Figure 1b. From Lincoln crop and Food research (soil type D)
Figure 1c. From North New Brighton (North east of city [soil type D/E])
Figure 1a-c. Five percent damped spectra from various locations with design ULS spectra for the relevant soil types added (Generated by Charles Clifton based on Geonet data)
Behavior of Eccentrically Braced Frames
A number of eccentrically braced frames (EBF) were recently constructed in Christchurch. Given the limited seismic demands during this earthquake on low- to medium-rise buildings, they generally performed well, however this was also the case for the tallest such building, a 22-story EBF, which was subjected to greater than 100% of ULS design earthquake loading in the north-south direction and also performed with no externally visible damage. This is partially attributed to the building being designed for a lower level of structural ductility demand than is typical for an EBF due to its height and plan dimensions. The behavior of a few of these frames is discussed below.
Typical EBF in the three-level parking garage of a shopping mall are shown (for a given story) in Figure 2a. Note that two tiers of bracing were used at each level in this structure (Figure 2b), which required the addition of channels along the EBF mid-height beams to provide lateral bracing of the links (Figure 2c). None of the EBF links showed evidence of yielding. Unrelated to the performance of the EBF, a few shear failures were observed at the corners of precast units on steel supports (Figure 1d), and steel plates tying precast panels suggested slight slippage at those ties locations, in embedment plates that fastened into more than one precast panel and were subject to failure as the panels slid relative to the embedment plate (Figure 2e). For perspective, although a few URM buildings suffered damage a block away, all other surrounding buildings were also free of visible structural damage.
Figure 2a. Global elevation (Photo by Michel Bruneau)
Figure 2b. Story elevation (Photo by Myrto Anagnostopoulou)
Figure 2c. Lateral bracing of EBF link using channels (Photo by Michel Bruneau)
Figure 2d. Shear failure at support of precast panel (Photo by Michel Bruneau)
Figure 2e. Failed embedment plate into precast concrete wall panels (Photo by Charles Clifton)
Figures 2a-e. Shopping mall on Dilworrth St and Clarence St, Christchurch (430 31’ 53” S - 1720 36’ 05” E)
The EBFs used in a hospital parking garage also performed well (Figure 3a). These differ from the previous ones in that concrete columns are used together with the steel beams and braces. The frames throughout the garage did not have evidence of inelastic action, except in a ramp built at the top level to accommodate the future addition of additional stories. The EBF only supported the east end of that ramp, and the reinforced concrete columns at the ramp expansion joint west of that EBF suffered shear failures as shown in Figure 3b. Slight flaking of the paint on that EBF link also suggests it underwent some limited yielding (Figure 3c). A few other links exhibited similar instances of flaked paint. Some of the steel plates used to laterally brace the links were observed to be bent, suggesting that the link started to laterally-torsionally buckle until restrained (Figure 3d). Note that these restraints will provide effective lateral restraint to the top flange of the EBF active link only.
Figure 3a. Typical bent (Photo by Mytro Anagnostopoulou)
Figure 3b. Damaged reinforced concrete column (Photo by Mytro Anagnostopoulou)
Figure 3c. Evidence of EBF link yielding (Photo by Michel Bruneau)
Figure 3d-e. Bent lateral restraints (Photos by Myrto Anagnostopoulou)
Figure 3a-e. Parking garage on St Asaph Street and Antigua Street,
Christchurch (430 32’ 10” S - 1720 37’ 41” E)
A 12–story building, whose construction was completed less than a year before the earthquake, also relied on EBF as its lateral load resisting system. The EBF were used on three sides of a stair/elevator core located on the west edge of the building. Beyond slight cracking of non-structural partitions, cracks were visible at the top of the concrete slabs parallel to the collector beams leading to the EBF (a composite metal deck slab with 80mm deep trapezoidal profile, with approximate 150mm overall slab thickness). These cracks, wider than hairline, suggested evidence of shear transfer in the concrete slab to the steel collector beams. The brittle intumescent paint coating used on the steel frames flaked in some of the EBF links, providing evidence of minor inelastic behavior during the earthquake. The links were free of visible residual distortions. Interestingly, cracking patterns in the slab were observed at the fixed end of a segment of the floor cantilevering on one side of the building (a feature present only over two stories for architectural effect); these cracks are likely to have been produced as a results of vibration modes excited by vertical ground excitations.
Some warehouses in Rolleston, three miles from the eastern tip of the surface faulting, suffered limited damage. Although exposed to greater ground accelerations that all buildings in Christchurch, these industrial facilities have light roofs and are designed to resist high wind forces—with typical average design wind pressures (external+internal) of 0.8 to 1.6 kPa on exterior cladding of single-story buildings and snow loadings of at least 0.5 kPa.
Most of these warehouses rely on concentrically braced frames built with slender steel rod braces for their lateral load resistance (Figure 4a). Braces are connected to the columns using one of various types of turn-buckle systems. One such system often used in these warehouses is a proprietary New Zealand system, which is sold as a kit and used a particular banana end fitting, as shown in Figure 4b. These are rated for earthquake loading following testing by the manufacturer, with a requirement of the test that failure occurs outside the connection region. Brace bars are typically pre-tensioned to 25% of their ultimate load using this system. A number of these braces were found to be sagging after the earthquake (by as much as 200mm according to some technicians involved in the post-earthquake inspection and refitting work), both in the vertical braced bays and roof diaphragm braced bays. The loose braces were simply re-straightened by retorquing the nuts at the end fitting after the earthquake. When tightening nuts remained in place throughout the earthquake, the presence of sag was indicative of effective axial plastic elongation of the braces.
Figure 4a. Elevation of concentrically braced bay (Photo by Myrto Anagnostopoulou)
Figure 4b. Banana end of proprietary brace connector (Photo by Myrto Anagnostopoulou)
Figure 4a-e. Warehouses in Rolleston
In one instance, fractures of banana bars near their pin end was observed in a roof braced bay (Figure 4c); it was alleged that this fracture occurred because the banana ends were oriented in a way that induced bending in their plates during vertical sagging of the bars, and that this would not have been the case if they had been oriented to allow rotation about their pin under gravity loads (i.e. by orienting the connector at 90 degree from how it had been installed). It was also suspected that some of these connectors were installed without pretensioning the nuts on both sides of bar fitting lug on the banana end, which could also explain the observed unsatisfactory behavior; the absence of pretension creates impacts of the nuts to the connector under the reversed cycling loading, which can push the nuts away from the connector upon repeated impacts, and result in loss of bar pre-tension, followed by loss of bracing action (some nuts were reportedly found to have been displaced by as much as 200mm from their original position). Finally, in one case, the grooves of the special purpose threaded bars used in this system were stripped within the connection itself, releasing the brace from its anchorage (Figure 4d), and in a few cases, the gusset plates to which the braces were connected suffered bearing failures.
Figure 4c. Bracing with fractured connection (Photo by Alistair Fussell)
Figure 4d. Stripped threaded bar (Photo by Michel Bruneau)
Figure 4e. Damaged garage door (Photo by Michel Bruneau)
None of the warehouses suffered damage beyond cosmetic cracking as a result of these behaviors.
Non-structural damage in these structures was substantially more significant. For example, many warehouse doors unhinged themselves, moving substantial distances out-of-plane (yellow paint marks on the door shown in Figure 4e provided evidence that the door hit the yellow post during the earthquake). This was costly damage given that some of these doors cost up to $75,000, and that some individual warehouses suffered up to a dozen such door failures. There was also failure to cold formed steel storage systems inside some of these buildings, as shown in Figures 5a and 5b.
Figure 5a. Global view (Photo by Myrto Anagnostopoulou)
Figure 5b. Close–up view (Photo by Myrto Anagnostopoulou)
Figure 5a-b. Damaged industrial storage racks in Rolleston
Damage to Heritage Structure (Steel and URM)
Figure 6a. Elevation (Photo by Myrto Anagnostopoulou)
Figure 6b. Damaged front piers (Photo by Myrto Anagnostopoulou)
Figure 6a-b. Heritage building on Manchester St and Hereford St, Christchurch (430 31’ 56” S - 1720 38’ 23” E)
The tallest unreinforced masonry (URM) building in Christchurch is shown in Figures 6a and 6b. Built in 1905-061, construction is as follows:
- bottom two stories are reinforced concrete
- top stories comprise timber floors supported on an internal steel gravity frame
- the perimeter comprises lateral load resisting piers of URM probably with embedded steel columns in the thinner piers which connect into the internal frame
At the time of writing this paper the future of the building is under debate. Some engineers consider that the building is basically sound and can be repaired provided that all the following are met:
- there is a regular spacing of steel columns embedded in the perimeter wall piers
- these columns match the spacing of the internal gravity columns so that the overall steel frame layout is regular and extends to the external walls
- there are steel beams connecting the grid of columns
- the steel beam to column connections will allow for up to 2.5% rotation to occur without loss of vertical load carrying capacity and crippling of the columns
- tither the steelwork runs through the bottom two stories of reinforced concrete or, if it starts from the top of these two stories, these stories are robust enough to avoid major damage and the interconnection between the top of the concrete and the steel frame to the upper levels is sound; and finally
- a survey of the out-of-plumbness of the external walls of the building show that vertical out of plumb developed by the building is not more than 0.3% (this should be done on the four corners of the building)
The building has been stable under the regime of aftershocks up to the date of writing these parts of the paper (September 21) and is the first high-rise building built in the city. It is considered by many to be of great historical value.
Collapse of Steel Tanks
Many steel tanks have collapsed during this earthquake. The failures are similar to what was observed in the 1987 Edgecumbe earthquake in which the most significant damage was to thin walled tanks and silos. In most instances the failures are due to one or more of:
- rotational or bearing failure of short columns supporting rigid tanks due to soft story action
- failure of bracing units supporting the tanks
- tearing failure at the attachment of the supporting structural steel frame into the thin walled tank itself
- foundation instability due to each supporting column being on an individual pad footing with no interconnection between the plates
A design document produced by the New Zealand Earthquake Commission (1990) following that 1987 earthquake recommended proper base support and details for tanks and silos. Figure 7e shows a silo with what appears to be a detail designed to that publication.
Figure 7a. Example collapse (Photo by Myrto Anagnostopoulou)
Figure 7b. Close-up view (Photo by Michel Bruneau)
Figure 7c. Tanks with brace legs
Figure 7d. Close-up view of collapse (Photo by Myrto Anagnostopoulou)
Figure 7e. Tank with continuous strip footing (Photo by Myrto Anagnostopoulou)
Figure 7a-e. Farm tanks in Darfield
Light Steel Framed Housing
Timber framing has been traditionally used for housing in New Zealand. However, the use of light steel frames for housing is a growing industry. Most are typically clad with brick veneer, consisting of 70mm thick bricks supported laterally by the steel frame.
There were approximately 40 houses built with light steel frames in the strongly shaken region. They all performed well where the underlying ground remained stable, with the worst reported damage in these cases being hairline cracks in the gypsum board lining of some internal walls. In one instance, a brick of the external cladding became loose, as shown in Figure 8.
A few houses situated where the ground slumped and spread underneath the house suffered greater damage.
Figure 8. Light steel frame house from epicentral region (Photo Courtesy of Graham Rundle)