Late at night on July 10th, 2006, a car was driving eastbound on Boston’s Interstate 90 connector tunnel, part of a large tunnel expressway system called the “Big Dig.” When the car approached the end of the tunnel, 15 panels from the tunnel’s suspended concrete roof collapsed onto the vehicle, covering it instantly with 26 tons of concrete and steel. The driver of the car survived, fortunately, but the passenger was killed.
What caused the ceiling of the tunnel to suddenly collapse?
Building The “Big Dig” Tunnel System
When the “Big Dig” was conceived in the late 1980s, the designs did not include a suspended ceiling and had no embedded ceiling supports. As construction began, engineers recognized that a suspended ceiling was required for a needed ventilation system, and Gannett Consulting was contracted to design a system of adhesive anchors to support the suspended ceiling. Later sections of the tunnel were designed with steel channels in the concrete roof from which to suspend the ceiling.
The ceiling over the section of the tunnel where the accident occurred was made of concrete panels supported by a steel framework, which was in turn suspended from the concrete tunnel roof. The steel ceiling framework was attached to the roof by stainless steel threaded rods (anchors) inserted into holes drilled in the concrete tunnel roof and fixed into the concrete of the roof by an epoxy adhesive.
For the anchors supporting the suspended ceiling, the consultant Gannett had advised using undercut anchors in the concrete roof of the tunnel. An undercut anchor hole is drilled wider at the base of the hole, deep within the concrete, and the steel anchor expands at the end to brace against the solid concrete. In an undercut anchor, the load is borne by concrete, not by epoxy and the threads of the anchor bolt (as with an adhesive anchor). The managers of the project, however, required Gannett to use adhesive anchors instead.
In 1999, before the first portions of the tunnel opened to traffic, the construction company discovered that some anchors were pulling out of the ceiling after only two months, even though they were designed to support much heavier loads than they were actually supporting. There was an investigation, but “a determination of failure could not be made.” To remedy the problem, they replaced the creeping anchors with ones designed for a heavier load.
Anchors in other areas of the tunnel were found to be pulling out in 2001, but again, the cause was found to be undetermined, and the construction company simply replaced the anchors.
Causal Factors of the Disaster
Wrong epoxy material. An inappropriate epoxy formulation, susceptible to “creep” under constant load, was used for the adhesive anchors of the tunnel sections. The ceiling support anchors pulled free over time, causing a portion of the ceiling to collapse.
Design did not account for anchor creep. Project managers and engineers failed to identify the risks of using the epoxy adhesive selected and did not consider the potential creep in the anchoring system designed for the tunnel. Engineers and builders also didn’t properly identify the cause of anchor creep in other areas of the tunnel.
Failure to monitor for anchor creep. Even after discovering anchor creep problems in other portions of the project, engineers and builders failed to implement a tunnel inspection program that almost certainly would have revealed anchor creep problems and driven correction of the deficiencies before a disaster occurred.
Effects of the Disaster
After the disaster, the National Transportation Safety Board (NTSB) conducted a thorough investigation of the design, construction, and maintenance of the tunnel.
The NTSB investigation showed that prior to the disaster, 13 of 20 anchors in fallen panels had displaced significantly (sometimes up to 80%). Throughout the rest of the tunnel, 161 of the remaining 634 anchors had pulled out to the point where they were in danger of sudden failure. There were measurable voids in the epoxy of almost all anchors and anchor holes that were analyzed, which occur during epoxy injection and compromise the strength of the overall epoxy bond. The void factor was not accounted for in the design calculations even though the NTSB investigation revealed that some voids always occur even if all procedures are followed, as it is a naturally occurring result of overhead installation of adhesive anchors.
Investigators found that during construction, there were significant problems in installing the adhesive anchors. About half of the drilled holes ran into rebar and needed to be patched and re-drilled. Installers sometimes applied epoxy where there was visible moisture, or didn’t allow enough curing time, or drilled anchor holes too deep, or applied inadequate quantities of epoxy, or didn’t clean the anchor holes. These mistakes, combined with cool temperatures and the damp environment of the tunnel, often prevented the epoxy from curing properly.
While load tests completed during construction proved the anchors could hold loads of 6,350 pounds with a safety factor of four, this was in a single test when the requirement should have been for a load supported over time. As the NTSB wrote in its report,
Gannett Fleming engineers should have been aware that …all polymers have the potential to deform under sustained load, [and] the designers included in the contract no specifications regarding the long-term mechanical properties of the adhesive, no requirement for testing of the adhesive for long-term performance, no consideration of the service life of the adhesive anchors in relation to the expected life of the tunnel, and no provision for periodic inspections of the installed anchors.
As a result of inadequate specifications for mechanical properties or monitoring of the epoxy over time, the construction company selected an epoxy that was significantly more susceptible to creep than other epoxy options.
The NTSB acknowledged that if Gannett’s specifications had included long-term performance and monitoring, “the construction contractor would have at least been made aware of the potential for anchor creep so that it could have specifically considered this factor when selecting the anchor adhesive.” In fact, the anchor installation proposal submitted by the construction company did note that the epoxy specified was approved for short-term loads only, but this was either overlooked or ignored by Gannett when they reviewed it for adequacy.
Investigators also found fault with the failure of the construction company to react to instances in 1999 and 2001 when anchors were discovered to be pulling out of the tunnel roof, noting that after those discoveries, designers…
…should have instituted a program to monitor anchor performance to ensure that the actions taken in response to the displacement were effective. Had these organizations taken such action, they likely would have found that anchor creep was occurring, and they might have taken measures that would have prevented this accident.
The investigation showed how ill-suited adhesive anchors were for the application of supporting the suspended ceiling. As a result, the companies responsible for designing and building the collapsed tunnel faced both criminal and civil charges. They settled for $458m with the State of Massachusetts in a criminal case, $16m with the City of Boston to avoid manslaughter charges, and $28m with the family of the victim in the civil case.
There is something inherently wrong with joints that depend on shear of epoxy. The article mentions an expanding anchor that would fill a larger undercut hole, mechanically locking. A much better long term approach!
Gannett recommended undercut anchors, but were overruled by “project managers.” Seems to me the burden of guilt should lie mostly with those “project managers.”
But then, having been told to use adhesive anchors, Gannett should have employed due diligence and made sure shortcomings of adhesive anchors were well-known to all involved and should have paid attention to warnings from company furnishing adhesive.
So, did the Massachusetts Board of Registration of Professional Engineers and Land Surveyors take action against any of the engineers on the project? I can’t find anything on the ‘net regarding this.
My last firm was an independent testing laboratory. Part of my job was performing tensile pull-out load testing on epoxy embedded anchors….which in my expertise are NEVER used for suspended ceilings. ACI did offer a course to become certified on the physical inspection while holes were being prepared for adhesive injection. It doesn’t seem that a QA / QC inspection program was incorporated for this project at all
The report says Gannett recommended an undercut anchor. Why would the owner go against this given the alternative, epoxy, even if the right epoxy was used would strap the owner with a lifetime of monitoring. It sounds like epoxy application in an overhead installation is not even practical given the potential for voids. Seems like the owner shares the blame.
The use of epoxy in tension was an error in concept even if the correct epoxy is used, and should never have been allowed. I can’t imagine anyone at any point in the design thinking that an overhead ventilation system not be required, whether suspended ceiling or some for of suspended large duct. Attachements to the sides of the tunnel supporting some form of cross brace would seem to have been what should have been used so that the embedded bolts, treaded rods, or whatever would be in shear, not tension. An undercut hole and enlarged base anchor should have been the absolute minimum supporting method. Much preferred would have been a rod cast in place with the overhead concrete with bends so that it would not be dependent solely upon bond to stay in place. Lack of complete contact and bonding between epoxy and insert and wall is simply a normal happening with anything less than 100% threat of unemployment level inspection of the installation. All the issues concerning inspection or lack thereof would have been simply processes to reduce the disaster potential of a bad concept.
They never identify “the managers of the project” who required Gannet to use the wrong type of anchors. This was a Bechtel – Parsons Brinckerhoff JV. They do a good job with the settlements, but failed to note that this one project came very close to putting Parsons Brinkerhoff out of business. PB senior management then used this incident to justify the sale of the 100+ year old firm to Balfour Betty (Balfour sold to WSP later). There was so much more to the story.
Seems odd that “the firms responsible for building and constructing and building the collapsed tunnel” are omitted from the article. The Big Dig contract was given to Bechtel– Parsons Brinckerhoff JV.
The last paragraph says a lot in that the use of epoxy anchors are ill suited for support of the suspended (concrete) ceiling. I have dealt with a client who in their General Specs allows the use of epoxy anchors in the material section, however in the installation section does not prohibit their use in overhead applications. As an Engineer when reviewing large pipe support details, I had always made sure that epoxy anchors were not used overhead and if anyone ever asked, I just sited this case to them on what can go wrong. I don’t see the epoxy anchor manufacturer / supplier mentioned, however in reading other articles on this failure it seems that they were scapegoated. One item that I recall is that the Contractor just requested “epoxy anchors” from the vendor and did not specify the application. The supplier indicated that they had supplied standard epoxy, however if informed of the application they would have provided rapid cure epoxy which would have been better suited, however I don’t know if it would have made a difference.