Collapse and Connections

I-W35 Bridge Collapse

Consider the failure of the I-35W Bridge from several years ago:

During the wreckage recovery, investigators discovered that gusset plates at eight different joint locations in the main center span were fractured. The Board, with assistance from the FHWA, conducted a thorough review of the design of the bridge, with an emphasis on the design of the gusset plates. This review discovered that the original design process of the I-35W bridge led to a serious error in sizing some of the gusset plates in the main truss…On November 13, 2008, the NTSB released the findings of its investigation. The primary cause was the under-sized gusset plates, at 0.5 inches (13 mm) thick.

Or the infamous Hyatt Regency walkway collapse:

…someone looking at the original details of the connection must have said he had a better idea or an easier way to hang one skywalk beneath the other…But no matter how much more convenient to assemble, the new rod configuration effectively doubled the push of the washer on the box beam supporting the upper walkway’s floor, and this made the already under-designed skywalks barely able to support their own weight.

Or the partial collapse of the Centergy parking deck in Atlanta:

First, the connection holding the fourth floor spandrel beam to the column broke. This caused the beam to slide away from the column. The beam moved away from the garage far enough that the T-beams composing the main floor of the parking deck were bearing on a very thin edge on the ledge of the beam. The concrete of the ledge spalled and as the t-beam fell it pushed the spandrel off of the structure. The fourth floor of the deck fell on top of the floor beneath. The weight of the falling floor overwhelmed the capacity of the floor beneath, and initiated a progressive collapse of the entire bay beneath.

Unless they’ve been catastrophically damaged, structural failure in the US tends to occur at the connections between members.

And this isn’t a modern phenomenon either. Here’s a passage from a “Wood Construction: Principles, Practice, and Details” from 1929:

The most important and, at the same time, the most difficult phase of wood design is the proper detailing of truss joints. The truss members themselves are easily computed, and it is seldom that the main sections of a truss fail. A case where truss details are amply sufficient for the stresses that may come on them, and the main members are insufficient, rarely, or never, occurs. The connections and details, however, are often neglected, and practically all the failures that occur are directly attributable to faulty joint details.

There are a variety of reasons for this. One is warning time before collapse occurs. Some types of failures, such as bending, deflect and crack significantly before ultimately failing. These outward signs provide time to address the problem before more serious damage can occur. Connections, however, will often fail in shear, which happens suddenly and without warning.

Another reason is that the connection is where all the engineering takes place. As mentioned above, member design is usually pretty simple: there simply aren’t very many decisions to make. All an engineer has to do is pick the member size and material, both of which are heavily standardized. Charts giving member capacity make this job even easier.

Not shown: the million other ways to connect wood together

Connections, on the other hand, require a great deal of design work. To start, there are many, many options for how to connect members together – wood members alone can be connected with shear rings, anchor bolts, nails, screws, pins, hangers, epoxy, and a host of other options. There are also many criteria to consider to make when specifying a connection – material availability, cost, ease of installation, installation time, structural effects, degree of fixity. Whereas a normal member has only 2 or 3 possible failure modes, a connection can have many, some non-obvious. And while there’s some degree of standardization, it’s not unusual to have to invent a brand new connection for an unusual situation.

All this means that there are many decisions that go into designing a connection. And the more decisions, the more likely a mistake will be made. It’s simply not very likely that an engineer will correctly design a complicated connection, and whiff on the much simpler task of sizing the member. Thus, structural collapse tends to mean connection failure, in this country at least.

Outside the US, however, this no longer holds true. For example, the recent garment factory collapse in Bangladesh:

…Ali Ahmed Khan, said that the upper four floors had been built without a permit. Rana Plaza’s architect, Massood Reza, said the building was planned for shops and offices – but not factories. Other architects stressed the risks involved in placing factories inside a building designed only for shops and offices, noting the structure was potentially not strong enough to bear the weight and vibration of heavy machinery.

Bangladeshi news media reported that inspectors had discovered cracks in the building the day before and had requested evacuation and closure. The shops and the bank on the lower floors immediately closed, but garment workers were forced to return the following day, their supervisors declaring the building to be safe.

Or the collapse of a tenement in Mumbai:

The building collapse appeared to be the result of poor quality building material and having been “weakly built”, according to Police inspector Digamber Jangale and Police commissioner K.P. Raghuvanshi. Lawful building construction in Thane district requires that blueprints are filed and approved by municipal agencies and permits are obtained to connect electricity, water and sewage services. The builders did not file the blueprints in this case.

Or the 1995 department store collapse in Seoul:

In April 1995, cracks began to appear in the ceiling of the south wing’s fifth floor. During this period, the only response by Lee and his management staff involved moving merchandise and stores from the top floor to the basement.

On the morning of June 29, the number of cracks in the area increased dramatically, prompting managers to close the top floor and shut the air conditioning off. The store management failed to shut the building down or issue formal evacuation orders, as the number of customers in the building was unusually high, and they did not want to lose the day’s revenue. However, the executives themselves left the premises as a precaution.

Outside the US, there are frequently weak institutions for building code enforcement. Buildings are often built illegally, with low quality materials. Plans are not submitted and building officials get bribed. Orders to evacuate are unheeded.

However, in the US we (for better or for worse) have quite strong institutions. Anyone with experience in construction will have a host of tales of jobsite screw ups. But these stories always end with “…and the contractor had to pay xxx dollars to fix it.” In short, while there are still construction problems, they’re almost always addressed before they become a safety issue. Engineers and inspectors are listened to. Repair and evacuation orders are followed. The threat of legal action is a strong motivator. As such, here it’s exceptionally rare for a building to fail due to poor quality materials or workmanship, or for danger warnings to go unheeded. The failures we get are egregious failures of engineering.


Holtman, Dudley. “Wood Construction: Principles, Practice, Details”. 1929

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s