Ep 23 Burnaby Supermarket Roof Collapse

Engineering NewsMexico City Subway Overpass Collapse (1:49).

This week`s episode is about the Burnaby Supermarket Roof Collapse (4:40). An undersized beam and added weight to the roof deck ultimately caused the beam to buckle and collapse (11:51). This change resulted in several important and necessary changes to the structural design of major projects in British Columbia (16:30).


Engineering News

Burnaby Supermarket Roof Collapse


Hi and welcome to Failurology; a podcast about engineering failures. I’m your host, Nicole, and I’m from Calgary, Alberta.

Did you know that Failurology has a brand new website? I wasn’t happy with the way it used to look; I was fairly limited with the templates that came as part of the podcast host subscriptions. So I decided to make a new site. Please check it out and let me know what you think. It’s still www.failurology.ca or there is a link in the show notes.

For every episode, you can listen right on the website, and there are photos for more information on the failure, all of my research sources for the episode and a transcript. I’ve been including phots of each failure since the beginning. I think a picture is worth a thousand words. And honestly, some of the failures are hard to explain using only words. It can be a bit tricky to follow how different components tie together, and that is often integral to how the failure occurred. I also want to mention that I only started including transcripts for later episodes. Should I have been including transcripts since the beginning? Absolutely. But I am here now and am working on adding them for all episodes; so please stay tuned for that. I do want to say, going back and listening to the first few episodes has been a bit of a trip. I’d like to think I’ve come a long way in recording and editing. And as tempting as it is to go back and tweak those episodes, I think it’s part of the podcasting journey to leave them there. It’s always fun to see the evolution of a show.

This week’s engineering news is about the subway train overpass that collapsed in Mexico City on Monday May 3rd. Yes, this past May 3rd. As of recording this episode, at least 24 people have died and about 80 people were injured. This is tragic. These people were riding public transit, many of them probably on their way home from work, when the overpass collapsed. The Mexico City mayor has said it appears a girder gave way on the overpass which was inspected the previous year.

The collapsed section was part of an addition that finished less than a decade ago. In 2014, two years after opening, several stations that were part of this addition were closed for structural repairs. In 2017, a powerful earthquake may have caused some damage to the support columns. Some people living nearby saw the support structures of the overpass shake when trains crossed. And during construction there were warnings of humid soil unfit for major construction. There have been allegations of corruption and structural weakness over the course of the project; although at this time those are just allegations. A full investigation of the failure will take place; I assume they’ve already started. And I look forward to reading it once complete.

I am covering this as a news piece because it is far too soon to confirm one way or another what the cause of the failure was. Based on the failures I have researched to date, the investigative reports are typically issued somewhere between 6 and 18 months after the failure occurs. Any conclusions before the investigation is complete would be speculation at best. There are far too many moving parts in this type of investigation, to be able to determine the cause of the failure. Not to mention, several components have to be reviewed in situ and then removed for additional testing, the original design drawings and construction documents have to be reviewed, and several people involved in the project have to be interviewed.

I do want to mention though, except for a few failures that had a settlement and/or non disclosure agreements, I have been able to find all of the investigation reports online. Even the ones from 30 or 40 years ago. They are almost always public knowledge. For the failures that I have covered that have full investigative reports, you can find links on the website. And if you are interested in reading them, pro tip, start with the conclusions or recommendations; these sections often paint a really good picture of the overall investigation. From there you can determine which sections you want to read in more detail.

I will be keeping my eye out for the Mexico City overpass collapse report, and depending on the cause of the failure, if it’s related to the design and engineering, I might cover it on a future episode. So stay tuned for that.

Now on to this week’s engineering failure; the Burnaby supermarket roof collapse in British Columbia. Burnaby is in the lower mainland neighbouring the City of Vancouver to the west. I always think of greater Vancouver as just Vancouver, but it is actually a grouping of several separate cities that make up the greater Vancouver area. Similar to Toronto. We do it a little differently in Calgary, we basically absorb any towns that we expand to border against and they all become part of the City of Calgary. I’m sure each way has its advantages and disadvantages, but I like that we all become part of one City. Seems like a more efficient way to operate, rather than having a bunch of separate municipalities all trying to work together while doing things differently. Although I’m curious what will happen when we expand to Airdrie to Okotoks, and whether they amalgamate into Calgary. Alright that’s enough of a civics lesson, lets get on with the supermarket roof collapse.

On April 23, 1988, minutes after the new supermarket’s grand opening, the roof top parking deck collapsed into the store. This wasn’t a recent failure, but it’s definitely an important one. Firstly, the collapse took four and half minutes from when the steel started to deform until the parking deck collapsed, which was enough time for most of the shoppers and employees to exit. Although 21 people were injured, no injuries were fatal. Had their not been a delay in the collapse, I’m sure it would have been a different story. And second, this failure instigated some significant, important, and necessary changes to the British Columbia Building Code and the regulations of the Association of Professional Engineers and Geoscientists of British Columbia. The failures that I cover on this show are often tragic, costly and embarrassing; but there is so much to learn from them. And they are integral to improving and shaping the integrity of future infrastructure.

As I said, about 10 minutes after the store opened, a 590m2 section of the roof top parking deck, including 20 cars, collapsed into the produce section below. When the steel started to deform, before the full collapse, it burst an overhead pipe, spraying water all over the store. This let staff know there was a problem and they were able to issue an evacuation alert through the PA system. There were about 600 customers and 307 employees in the store at the time.

The grocery store was part of a larger development that began construction in 1985. Ironically, the roof top parking was proposed by one of the contractors during tender or pricing of the project. The project architects and engineers reviewed the possibility of adding rooftop parking and recommended changes to the design of the structure, as well as adding a walkway to improve access to the parking area. These recommendations were implemented into the project design.

The steel beam and column that failed were part of four structural bays that make up a 27x23m section of the roof structure. The beam was undersized for its intended purpose of supporting rooftop parking and did not have the lateral bracing required to resist buckling. Even though the structure that failed was designed by an experienced engineer, detailed by an experienced detailing firm, fabricated and erected by an experienced steel contractor under a pre-qualified general contractor; as well as independently reviewed, questioned and accepted by an experienced structural engineer, the failure still occured. This can happen to anyone. We are all human and we all make mistakes.

I am going to go into more depth about how the failure occurred, but first I want to talk a bit about structural design. The structural engineer calculates the stresses created by both the dead and lives loads or the loads of the structure itself as well as the loads placed on the structure by occupants, weather, etc. They also calculate the capacity of each structural member. But its important to note that these are not necessarily precise, there are probabilities. They’re not looking at a floor plan and saying ok I have a couch here and a table there so the load is x. No, they look at the overall use of the space and then apply a uniform design capacity across that area. At the time of the supermarket construction the British Columbia building code assigned parking decks a live load of 245kg/m2, which is the most probable maximum load. Even though over years of parked cars, the actual load may not exceed 100kg/m2.

They then calculate the strength of each structural member. They try to account for and mitigate small errors in construction, slight changes to dimensions, and environmental conditions such as temperature variations and foundation movements. Preventing all of these is pretty much impossible, so they often include a safety factor which slightly overdesigns the structure to cover any of these anomalies.

The trick is to design the structure so the capacity of each member is greater than the loads placed on it. Which seems pretty straightforward to me, even having never designed a structure myself. Unfortunately that was not the case here.

The thickness of the roof beam that failed was reduced by 27% during design. I haven’t really been able to find the reason why. At this point, its anyone's guess. In November 1987, during construction, a beam deflected in another part of the store. This was so significant that the residential structure above tilted 50mm from vertical. A second structural engineer was brought in to review and they recommended additional columns, stronger beams and concrete reinforcing. The original structural engineer agreed with the recommendations and they were completed.

In March 1988 the second structural engineer was hired to review the entire supermarket project. They found two beams had insufficient capacity and recommended corrections. But, and this is really so interesting, the original engineer advised the next day that the steel mill who fabricated the beams provided additional data of the actual capacity of the beam. The data showed that the actual capacity of the beam was greater than what they calculated during design. And both engineers agreed that based on this information, the beams were satisfactory. The steel mill test information, which showed the beam as 25% stronger than expected, was taken as accurate. But, and here comes the interesting part, the test is done with a sample cut, which by nature of the cut, generates a higher test strength than the beam itself, and the steel and manufacturing fluctuations result is varying strengths. So they probably never should have trusted the mill test in the first place and instead relied on their industry manuals which had much more accurate information on different types of beams. The report does not outright state that the beam that failed was one of the two beams in question, but I think it’s a safe assumption that the steel mill data provided a bit of a false confidence in the strength of the structure.

There were two types of beam failures that occurred during the roof collapse. The likely prime cause of the failure was the beam bending and buckling at the lower flange due to its unsupported length, and there was also a buckling of the beam-column assembly. The beam that failed was a W24x76 beam which is 24 inch or 600mm deep wide flange steel beam, weighing 76lb/ft or 113kg/m. The beams ran east-west and were supported by columns spaced 9-13m apart. Each alternate beam extended over its two support columns and cantilevered by about 2m in both directions. Shorter beams were installed in between the longer beams and supported by those cantilevered ends. This is referred to as a Gerber System or Cantilever Suspended Span; it’s fairly common for roofs and is generally a safe structure and more economical than simple column to column spans. The columns were 300x300mm hollow steel on concrete footings; the footings were not a factor in the collapse.

In addition to the beams and columns, there was a series of open web steel joists or trusses, spaced 2m apart, that spanned north-south from beam to beam, and the top of each joist was welded to the corresponding beam. The joists themselves were not a factor in the collapse. However, the question of whether or not lateral bracing should have been installed with the joists came up during the shop drawing review process. But there was a misunderstanding of who determined the need for lateral bracing. The steel contractor didn’t provide bracing because it wasn’t noted on the drawings and the engineer thought the decision to provide bracing was by the steel contractors detailer. It seems the capacity of the beams were reliant on the added support of lateral bracing, even though it was never installed. The mill test I mentioned earlier might have played a factor in how important the engineers thought the stiffeners were to the beam strength. They may have believed the beams to be strong enough for the stiffeners to be less critical. Although the investigation found that failure would have occurred even with bracing, but perhaps might have taken longer to do so.

And lastly, there was the roof deck. It was made up of corrugated metal, called Q-deck, a waterproofing membrane, a structural concrete layer which extended 65mm above the Q-deck corrugations, a 100mm light Styrofoam insulation layer and a 75mm topping to act as pavement for the cars parked on the roof. The topping was supposed to be 50mm thick and was increased to 75mm during construction. While the roof joists were revised to accommodate the extra inch of topping, the beams were not. There was also a walkway in this area that was originally designed to be 1.5m wide, but revised to 3.4m wide during construction. It was supposed to have a 250mm layer of light Styrofoam insulation and a 150mm topping layer. The contractor skipped the insulation under the walkway and filled the depth entirely with concrete, which added a bunch of weight to the structure that went unaccounted for. While the roof deck structure itself was not a factor in the failure, its added weight certainly was.

Under the added weight of the roof deck, the already undersize beam buckled and rotated 90 degrees from vertical above the supporting column. Essentially what used to be a 600mm tall beam, was now laying on its side. When it did this, the rest of the structure acted as a shallow “suspension bridge” and held for four and a half minutes before fully collapsing, which was enough time for most of the shoppers and staff to exit. There are a couple really good pictures on the Failurology website that show the build up of the roof deck, how the structural members were tied together, and what happened when they failed.

A full investigation into the cause took place. The actual load was found to be about one and a half times the capacity of the beam. Some of that was due to construction changes like changing topping thickness and adding concrete instead of insulation; both of which the structural engineer should have noted during their site reviews. But there was also some calculation errors by both engineers on the loads placed on the beam. The failure was inevitable, it was just a matter of when.

From the investigation, a number of important and necessary changes were recommended for structural designs of major projects. The Commissioner of Inquiry noted three common industry practices that were cause for concern.

First, there is fast-tracking, which means to obtain permits and commence construction on a portion of the project before the design is complete. Sometimes it’s unavoidable, you are designing an underground parkade for multiple towers, but then only building one tower at a time. Mechanically, you have to try to plan the layout of the future towers so that slab openings can be blocked out when the slab is poured rather than trying to cut holes later. And structurally, you have to design your foundation for all of those future structures without a full picture of what those structures are. And then sometimes, you are still designing the tower while they are building the below grade structure. You have to try to future proof the below grade design so you are prepared for any changes that occur. Fast tracking can involve significant changes during construction, it creates confusion and delay, as well as increasing opportunities for error. You can do things fast, cheap or good; and you only get to pick two.

Second, the commissioner took issue with the bidding process for professional services. They looked at all of the fees that were submitted, and just based on the numbers, all of the firms who bid on engineering services “did not intend to provide” the proper level of professional service. In this case, the structural engineering fee was $17,000 on a $5.4 million project. It was recommended that the Association of Engineers and Geoscientists of British Columbia establish and reinforce a minimum fee schedule for major projects.

And lastly, there was a fragmentation of contractual responsibility. It is extremely important for the success of a project, to generate a good working relationship between the consultant and construction team. Creating a dialogue between these two parties, and working together as a team, is the best way to ensure things aren't missed.

The Commissioner of Inquiry also had some recommendations for the British Columbia Building Code and Engineers Act. Municipalities should require structural calculations and drawings be submitted during the building permit stage. A detailed review should be conducted of a random sampling of projects. Where warranted designs should be submitted for detailed review to the governing body. And those review costs should be born by municipal levees on building permits. As well, the report recommended that structural engineers pass a special examination before becoming professionally qualified and required to carry specified limits of professional liability insurance.

Regarding professional levels of assurance the British Columbia Building Code should provide province wide standard Letters of Assurance templates. Today, registered professionals, which includes architects and engineers, provide formal documentation to the municipality which certifies that, to the best of their knowledge, the design conforms to building codes, and that construction conforms to the design documents and codes. I can’t even tell you how important this is. First of all it simplifies what each jurisdictions wants to see from the consultant teams during occupancy. And second, it clarifies the responsibility of the registered professional. In Alberta, you submit A and B schedules when you apply for building permits and then at occupancy you submit a C schedule. Not all jurisdictions have this, and it complicates an already hectic occupancy process.

In addition to this, the report also recommended that engineering firms, in addition to individual engineers, be registered with the engineering association. And that steel industry construction manuals be revised to provide more accurate assistance to engineers.

In 1989, four engineers from both of the structural engineering firms were found guilty of incompetence, negligence and professional misconduct. This was based on inadequate design of the roof structure, including the support beams, failure to inspect the construction of the roof assembly, and failure to direct or check the work on the project of the engineer in training who was directly managing the project. The engineers code of ethics was breached by a partner signing and stamping the plans and specifications that he hadn’t prepared or supervised. This is a big no-no in engineering. You either need to do the work, or directly oversee it if you are stamping designs.

After the collapse, several lawsuits were filed. By the developers against the general contractor, architect and structural engineer. By the developers and the consultants insurance companies. And by the owners of the cars that were destroyed in the collapse.

The supermarket and parking structure were ultimately repaired, re-opened and operated until 2012. Then it was demolished as part of a new complex. The supermarket re-opened in 2015 in this development. Which is kind of cool, because you don’t often see a retailer return to a complex following such major reconstruction.

So there you have it, the Burnaby supermarket roof collapse. The outcome of this failure shaped the British Columbia Building Code for decades and is the reason they have some of their current policies and procedures in place today. While sometimes jumping through extra hoops can seem like a nuisance, in this case, those checks and balances can save lives and livelihoods and are a necessary part of the building construction process. While it was preventable, I am going to take this as an overall win for structural engineers and the construction industry because of all of the changes this failure instigated.

For photos, sources and transcripts from this week’s episode, check out the show notes or head to Failurology.ca. If you’re enjoying what you’re hearing, please rate, review and subscribe to failurology, so more people can find it. If you want to chat with me, my twitter handle is @failurology, you can email me at thefailurologypodcast@gmail.com, or you can connect with me on Linked In. Check out the show notes for links to all of these.

Thanks everyone for listening. And tune in to the next episode of Failurology where I will cover the Eindhoven Airport parking garage collapse. The project was constructed using a bubble deck structure with is a really interesting slab method that I haven’t personally seen in North America. But more on that next time. Bye everyone, talk soon!