Figure 1: Truck at beam failure location; from WJE Final Investigation Report
The David L. Lawrence Convention Center represents Pittsburgh moving forward, breaking from its steel and smog and creating a new, more sustainable future. However, from the beginning of construction, it has encountered problems in its design and construction that have led to unrest in the Pittsburgh community. These problems started at the foundations with shifting caissons. They expanded to more dangerous and fatal failures, starting with the truss collapse during installation of the roof. Even after completion, two connection failures occurred, leading the public to question the structural safety of the building. Small overlooked details led to the bulk of these more dangerous failures; these are discussed within the context of this case study.
Summary
Figure 2: Expansion joint connection details; from Pittsburgh SEA
Beginning on February 12, 2002, roof trusses were being installed that would eventually support a massive span of roof over the showroom floor. The trusses were cantilevered and made out of cables; t he top cables were to be tensioned after truss installation to bend up and support the roof. During installation of the 13th truss, the connection failed and sent an iron worker who was tied to the truss falling to the floor below. The collapse resulted in his death, along with two other injured workers. After inspection of the failure, it was determined that one-inch diameter nuts instead of two-inch diameter nuts were being used to connect the cables, ultimately causing collapse.
On February 5, 2007, a tractor trailer fell through the floor of a second floor loading dock on the second floor of the David L. Lawrence Convention Center in Pittsburgh, Pennsylvania, as seen in Figure 1. The collapse happened in a structural bay immediately adjacent to the primary building expansion joint, with the outdoor temperature at about negative three degrees Fahrenheit. Wiss, Janney, Elster Associates, Inc. was hired to investigate possible causes of the failure. There was much speculation about the connection to the expansion joint of the beam supporting the floor, and it was eventually determined that this was, in fact, the root of the problem. The connection relied on slotted holes in the beam web to accomodate movement at the expansion joint. An inability for the bolts to slide freely within the slotted holes (coupled with other extenuating circumstannces) proved to be the cause of failure. A graphic of the failure can be seen in Figure 2.
Causes
Figure 3: Photo of the trusses; from Pittsburgh SEA
The truss collapse of 2002 occurred during construction, when roof trusses were being installed prior to the final roof material. It was triggered by installed 1-inch diameter nuts that did not meet the required 2-inch diameter strength as specified (Barnes 2002). The nuts were not able to hold the load applied by the truss at the support. They popped off the ends of the bolts, sending the truss falling to the road below. The failed truss's position can be seen as installed in Figure 3. At first glance, this mistake indicates a problem with communication between parties. However, it was never determined where the error originated; ironworkers insisted they installed the nuts they were told to install, while contractors said they specified the correct nut size (Barnes 2002). Upon further investigation, it was determined that there were multiple errors that led up to this collapse. The steel fabricator, Williams Form Engineering Corporation, began using 1-inch instead of 0.5-inch nuts as "jam nuts" to hold the 2-inch diameter nuts in place; these nuts were never intended to replace the 2-inch nuts. However, this was not specified by the fabricator, and also not put on the contractors' drawings (Franken 2002). Nuts were also never hand-tightened after placing of the truss, in turn not allowing the connection to reach capacity. In addition, the project was reportedly rushed, as the roof was expected to be completed for a show the next week (Barnes 2002).
The failure in 2007 did not result in injuries but proved to be more publicized and criticized. A 30-foot by 60-foot bay collapsed under the weight of a delivery truck on a cold day in February. The temperature was estimated to be around 7 degrees Fahrenheit at the time of the failure, but the structural steel could have been as cold as -3 degrees Fahrenheit from the frigid temperatures earlier in the day (WJE 2008). The initial cause of collapse was the weight of the truck bearing on an expansion joint, coupled with restrained shrinkage of the structure due to the low temperatures. The double-angle connection at the expansion joint was designed to allow the bolts to slide in their slots when the beam expanded and contracted. However, multiple errors in design, fabrication, and workmanship ultimately rendered this connection ineffective at allowing free movement. The final investigation report outlines these errors as follows (WJE 2008):
Figure 4: Double-angles showing deformation at welds; from WJE Final Investigation Report
The design of the connection did not factor in the horizontal forces caused by temperature.
On design drawings, bolt threads were not prohibited from bearing on the slots.
The steel of the double angle strength was A992, not A36 as designed, causing the weld to fracture before the angles yielded (see Figure 4).
Slots in the angles were fabricated with a bump on the surface that bore the bolts, adding resisistance to sliding of the bolts in the connection.
Washer plates were mistakenly added during construction, increasing restraint forces on the slotted connection.
This restraint of the slotted connection was found to be the root of the problem. An ideal expansion joint connection would have as little friction as possible to the assembly to move freely with expansion and contraction of the beam. What friction is anticipated should be accounted for in the design of the structure. It was noted by an engineer from AISC that these slotted connections are not ideal for use at expansion joints; recommendations for expansion joints in the AISC Manual include a sliding seat connection or building the structure as two separate structures at the joint (Delatte 2008).
The 2008 WJE report is the only publicaly released opinions for the cause of the 2007 collapse; all sources found with research agree with their analysis and conclusions outlined above. There were other investigations done but no information from these could be found. There was no formal report for the truss collapse or the other connection failure in 2005, discussed in more detail below.
Prevention
The truss collapse during construction could have been prevented if warning signs were heeded and if there was a method of consistent communications on the job site. Before the accident, an inspector from Solar Testing Labs noticed that the nuts connecting the trusses to the frame were not tightened; upon inquisition to his supervisor, he was told that the trusses were being compressed by the weight of the building and therefore the bolts were not critical at the time for the connection (Franken 2002). However, the roof cables would later be tensioned and would apply the opposite pressure at the connection; fortunately, this tensioning did not happen before the collapse. In addition, the manufacturer did not explain its procedure for the jam nuts to the contractors, and this was also not specified on the drawings. There was no communication to the ironworkers that they were supposed to use the 1-inch nuts only as jam nuts to back up the 2-inch nuts.
After the backlash surrounding the 2007 slab and beam collapse, it became public opinion that a sliding connection is not the correct way to connect to an expansion joint (Post 2007). To prevent this, the design team should have consulted the AISC for common practices for designing expansion joint details and connections. The errors in design and fabrication could have been avoided through more specific details on the drawings; a statement prohibiting threads from bearing on slots was needed. Proper design procedure and more thorough inspections and quality control of the connections could have eliminated the steel yield strength issues, bumps, and addition of washer plates. These mistakes were almost a repeat of those in atrium failure in the Hyatt Regency; if designers had learned from the past, this could possibly have been prevented (Houser and Ritchie 2007).
However, the most important thing that could have been done to prevent this collapse was to heed warnings from the structure two years before. In 2005, a different beam fell from its connection position in a similar fashion, but only dropped 2.5 inches before it was caught by a column (Belko 2007). Unfortunately, the design team explained that the problem was only caused by an over-torqued bolt, and required a spot check of other connections to make sure they were not experiencing a similar problem. No further investigation was performed. The instance was not reported to the mayor or other authorities as the design team advised the situation was not a major issue. The fix outlined for the failed 2005 beam was strikingly similar to that prescribed for all other beams after the 2007 failure. A Teflon-coated bearing plate assebly was installed between the beam bottom flange and a new stiffened seat at each beam along the expansion joint. The existing slotted bolt connection was effectively abandoned. If the warnings from two years earlier had been looked at in more detail, this collapse would probably not have occurred.
Consequences & Developments
Figure 5: Teflon-coated seats by Thornton Tomasetti; from WJE Final Investigation Report
As a result of the 2002 truss collapse, the inspectors at Solar Testing Labs required a thorough inspection of all connections on jobs, not just a random selection (Franken 2002). This ensures that a problem is not overlooked. A very fatal consequence of the 2002 event was the death of an ironworker who was tied off to the truss at the time of collapse, as well as two other serious injuries. This death resulted in homicide charges to be brought against Dick Corporation (Barnes 2003). According to Barnes, "public officials tend to target employers when workers get hurt or die." Later, the district attorney threw out the charges; he felt the workers themselves were to blame for not making sure the connections were correct before tying off to the steel (2003). This displays a very unfortunate consequence of construction failures.
After the 2007 beam and slab collapse, Thornton Tomasetti (retained by the design team) recommended low-friction bearing plates be installed on new seats under each of the beams at the expansion joint. The existing bolts were removed from the slotted angle connections, and each beam now bears on the Teflon-coated seats. Twenty-five beams were fixed in this manner; they are now free to slide back and forth under thermal loads (Delatte 2008). The implemented solution is shown in Figure 5.
Bibliography
Barnes, Jonathan (July 29, 2002). “Steel collapse causes cited.” Engineering News-Record, 249(5).
This article explains the causes of the truss collapse on February 12, 2002. It documents statements from the contractor and other parties involved with respect to the nuts that were found to be the wrong size, initiating the failure of the truss.
Barnes, Jonathan (September 15, 2003). “District Attorney won’t charge Dick Corp. in Pittsburgh death.” Engineering News-Record, 251(11).
In this article, a different perspective is brought to light surrounding the truss collapse at the convention center. It highlights the problems with responsibility associated with construction site deaths, as well as gives a basic overview of the problems leading up to the collapse.
Belko, Mark. (February 22, 2007). “Convention center beam slipped in 2005.” Pittsburgh Post-Gazette. <http://www.post-gazette.com/pg/07053/764100-53.stm> (September 24, 2009).
Pictures and animation are included in this newspaper article to demonstrate the failure mechanism of the connection. In addition, the article reveals that there was a previous problem that was covered up by authorities that could possibly have prevented the floor collapse in February 2007.
Delatte, Norbert J. (2008). “Beyond Failure: Forensic Case Studies for Civil Engineers.” Pittsburgh Convention Center Expansion Joint Failure. ASCE Publications. 206-211.
Brief overview of collapse of Pittsburgh convention center.
Franken, Stephanie. (June 26, 2002). "Incorrect use of nuts blamed for collapse." Pittsburgh Tribune-Review. <http://www.pittsburghlive.com/x/pittsburghtrib/news/cityregion/s_78245.html> (October 8, 2009.)
This article includes statements from the inspector as well as from the fabricator about the size of the nuts used at the connections.
Houser, Mark and Ritchie, Jim. (February 22, 2007). “Convention Center collapse blamed on bolt connection.” Pittsburgh Tribune-Review. <http://www.pittsburghlive.com/x/pittsburghtrib/news/s_494416.html> (September 24, 2009).
Convention center collapse report compares connection to the Hyatt Regency atrium collapse. The same problems were found in both situations, and although the Hyatt Regency happened in 1979, professionals did not learn from the mistakes.
Post, Nadine M. (March 5, 2007). “Engineers slam failed joint detail, fear that slotted connection behind Pittsburgh collapse may be ‘locked up’ in other structures.” Engineering News-Record, 258(9).
This article provides comments from many engineers explaining the problems with the slotted connection involved in the February 5, 2007 collapse of a section of the convention center floor. The article goes into details of issues with the specific and general connection details.
Wiss, Janney, Elstner Associates, Inc. “David L. Lawrence Convention Center: Investigation of the 5 February 2007 Collapse.” WJE Final Investigation Report. October 8, 2009.
The final investigation report on the causes and possible solutions of the beam collapse, with many pictures for reference.
Additional References
Post, Nadine M. (February 18, 2002). “Fatal collapse in Pittsburgh.” Engineering News-Record, 248(6).
This article explains the layout and connection detailing for the roof trusses.
Staff. (February 11, 2008). “Multiple errors led to beam failure.” Engineering News-Record, 260(5).
This article was released after the final investigation report provided by WJE. Testimonies from engineers include those from WJE as well as from Thornton Tomasetti, who was responsible for fixing the connections at all locations.
DAVID L. LAWRENCE CONVENTION CENTER
Table of Contents
Erika L. Bonfanti, BAE/MAE, Penn State, 2009
Keywords
Expansion joint, sliding connection, temperature, truss, connection, nuts, communication, construction error, fabrication error, double angle, frictionAbstract
The David L. Lawrence Convention Center represents Pittsburgh moving forward, breaking from its steel and smog and creating a new, more sustainable future. However, from the beginning of construction, it has encountered problems in its design and construction that have led to unrest in the Pittsburgh community. These problems started at the foundations with shifting caissons. They expanded to more dangerous and fatal failures, starting with the truss collapse during installation of the roof. Even after completion, two connection failures occurred, leading the public to question the structural safety of the building. Small overlooked details led to the bulk of these more dangerous failures; these are discussed within the context of this case study.
Summary
Beginning on February 12, 2002, roof trusses were being installed that would eventually support a massive span of roof over the showroom floor. The trusses were cantilevered and made out of cables; t
he top cables were to be tensioned after truss installation to bend up and support the roof. During installation of the 13th truss, the connection failed and sent an iron worker who was tied to the truss falling to the floor below. The collapse resulted in his death, along with two other injured workers. After inspection of the failure, it was determined that one-inch diameter nuts instead of two-inch diameter nuts were being used to connect the cables, ultimately causing collapse.
On February 5, 2007, a tractor trailer fell through the floor of a second floor loading dock on the second floor of the David L. Lawrence Convention Center in Pittsburgh, Pennsylvania, as seen in Figure 1. The collapse happened in a structural bay immediately adjacent to the primary building expansion joint, with the outdoor temperature at about negative three degrees Fahrenheit. Wiss, Janney, Elster Associates, Inc. was hired to investigate possible causes of the failure. There was much speculation about the connection to the expansion joint of the beam supporting the floor, and it was eventually determined that this was, in fact, the root of the problem. The connection relied on slotted holes in the beam web to accomodate movement at the expansion joint. An inability for the bolts to slide freely within the slotted holes (coupled with other extenuating circumstannces) proved to be the cause of failure. A graphic of the failure can be seen in Figure 2.
Causes
The failure in 2007 did not result in injuries but proved to be more publicized and criticized. A 30-foot by 60-foot bay collapsed under the weight of a delivery truck on a cold day in February. The temperature was estimated to be around 7 degrees Fahrenheit at the time of the failure, but the structural steel could have been as cold as -3 degrees Fahrenheit from the frigid temperatures earlier in the day (WJE 2008). The initial cause of collapse was the weight of the truck bearing on an expansion joint, coupled with restrained shrinkage of the structure due to the low temperatures. The double-angle connection at the expansion joint was designed to allow the bolts to slide in their slots when the beam expanded and contracted. However, multiple errors in design, fabrication, and workmanship ultimately rendered this connection ineffective at allowing free movement. The final investigation report outlines these errors as follows (WJE 2008):
This restraint of the slotted connection was found to be the root of the problem. An ideal expansion joint connection would have as little friction as possible to the assembly to move freely with expansion and contraction of the beam. What friction is anticipated should be accounted for in the design of the structure. It was noted by an engineer from AISC that these slotted connections are not ideal for use at expansion joints; recommendations for expansion joints in the AISC Manual include a sliding seat connection or building the structure as two separate structures at the joint (Delatte 2008).
The 2008 WJE report is the only publicaly released opinions for the cause of the 2007 collapse; all sources found with research agree with their analysis and conclusions outlined above. There were other investigations done but no information from these could be found. There was no formal report for the truss collapse or the other connection failure in 2005, discussed in more detail below.
Prevention
The truss collapse during construction could have been prevented if warning signs were heeded and if there was a method of consistent communications on the job site. Before the accident, an inspector from Solar Testing Labs noticed that the nuts connecting the trusses to the frame were not tightened; upon inquisition to his supervisor, he was told that the trusses were being compressed by the weight of the building and therefore the bolts were not critical at the time for the connection (Franken 2002). However, the roof cables would later be tensioned and would apply the opposite pressure at the connection; fortunately, this tensioning did not happen before the collapse. In addition, the manufacturer did not explain its procedure for the jam nuts to the contractors, and this was also not specified on the drawings. There was no communication to the ironworkers that they were supposed to use the 1-inch nuts only as jam nuts to back up the 2-inch nuts.
After the backlash surrounding the 2007 slab and beam collapse, it became public opinion that a sliding connection is not the correct way to connect to an expansion joint (Post 2007). To prevent this, the design team should have consulted the AISC for common practices for designing expansion joint details and connections. The errors in design and fabrication could have been avoided through more specific details on the drawings; a statement prohibiting threads from bearing on slots was needed. Proper design procedure and more thorough inspections and quality control of the connections could have eliminated the steel yield strength issues, bumps, and addition of washer plates. These mistakes were almost a repeat of those in atrium failure in the Hyatt Regency; if designers had learned from the past, this could possibly have been prevented (Houser and Ritchie 2007).
However, the most important thing that could have been done to prevent this collapse was to heed warnings from the structure two years before. In 2005, a different beam fell from its connection position in a similar fashion, but only dropped 2.5 inches before it was caught by a column (Belko 2007). Unfortunately, the design team explained that the problem was only caused by an over-torqued bolt, and required a spot check of other connections to make sure they were not experiencing a similar problem. No further investigation was performed. The instance was not reported to the mayor or other authorities as the design team advised the situation was not a major issue. The fix outlined for the failed 2005 beam was strikingly similar to that prescribed for all other beams after the 2007 failure. A Teflon-coated bearing plate assebly was installed between the beam bottom flange and a new stiffened seat at each beam along the expansion joint. The existing slotted bolt connection was effectively abandoned. If the warnings from two years earlier had been looked at in more detail, this collapse would probably not have occurred.
Consequences & Developments
After the 2007 beam and slab collapse, Thornton Tomasetti (retained by the design team) recommended low-friction bearing plates be installed on new seats under each of the beams at the expansion joint. The existing bolts were removed from the slotted angle connections, and each beam now bears on the Teflon-coated seats. Twenty-five beams were fixed in this manner; they are now free to slide back and forth under thermal loads (Delatte 2008). The implemented solution is shown in Figure 5.
Bibliography
Barnes, Jonathan (July 29, 2002). “Steel collapse causes cited.” Engineering News-Record, 249(5).
This article explains the causes of the truss collapse on February 12, 2002. It documents statements from the contractor and other parties involved with respect to the nuts that were found to be the wrong size, initiating the failure of the truss.
Barnes, Jonathan (September 15, 2003). “District Attorney won’t charge Dick Corp. in Pittsburgh death.” Engineering News-Record, 251(11).
In this article, a different perspective is brought to light surrounding the truss collapse at the convention center. It highlights the problems with responsibility associated with construction site deaths, as well as gives a basic overview of the problems leading up to the collapse.
Belko, Mark. (February 22, 2007). “Convention center beam slipped in 2005.” Pittsburgh Post-Gazette. <http://www.post-gazette.com/pg/07053/764100-53.stm> (September 24, 2009).
Pictures and animation are included in this newspaper article to demonstrate the failure mechanism of the connection. In addition, the article reveals that there was a previous problem that was covered up by authorities that could possibly have prevented the floor collapse in February 2007.
Delatte, Norbert J. (2008). “Beyond Failure: Forensic Case Studies for Civil Engineers.” Pittsburgh Convention Center Expansion Joint Failure. ASCE Publications. 206-211.
Brief overview of collapse of Pittsburgh convention center.
Franken, Stephanie. (June 26, 2002). "Incorrect use of nuts blamed for collapse." Pittsburgh Tribune-Review. <http://www.pittsburghlive.com/x/pittsburghtrib/news/cityregion/s_78245.html> (October 8, 2009.)
This article includes statements from the inspector as well as from the fabricator about the size of the nuts used at the connections.
Houser, Mark and Ritchie, Jim. (February 22, 2007). “Convention Center collapse blamed on bolt connection.” Pittsburgh Tribune-Review. <http://www.pittsburghlive.com/x/pittsburghtrib/news/s_494416.html> (September 24, 2009).
Convention center collapse report compares connection to the Hyatt Regency atrium collapse. The same problems were found in both situations, and although the Hyatt Regency happened in 1979, professionals did not learn from the mistakes.
Post, Nadine M. (March 5, 2007). “Engineers slam failed joint detail, fear that slotted connection behind Pittsburgh collapse may be ‘locked up’ in other structures.” Engineering News-Record, 258(9).
This article provides comments from many engineers explaining the problems with the slotted connection involved in the February 5, 2007 collapse of a section of the convention center floor. The article goes into details of issues with the specific and general connection details.
Wiss, Janney, Elstner Associates, Inc. “David L. Lawrence Convention Center: Investigation of the 5 February 2007 Collapse.” WJE Final Investigation Report. October 8, 2009.
The final investigation report on the causes and possible solutions of the beam collapse, with many pictures for reference.
Additional References
Post, Nadine M. (February 18, 2002). “Fatal collapse in Pittsburgh.” Engineering News-Record, 248(6).
This article explains the layout and connection detailing for the roof trusses.
Staff. (February 11, 2008). “Multiple errors led to beam failure.” Engineering News-Record, 260(5).
This article was released after the final investigation report provided by WJE. Testimonies from engineers include those from WJE as well as from Thornton Tomasetti, who was responsible for fixing the connections at all locations.