The Woodlands at Clemson clubhouse floor collapsed in the early Sunday morning hours of October 20, 2018, injuring several individuals who had gathered for an annual homecoming weekend party. Several engineers and media outlets concluded that the rhythmic jumping (i.e., dancing) by a significant number of individuals was the cause of the floor collapse. However, two structural engineers who investigated the floor collapse found evidence that improper metal plate connected wood truss design and fabrication were more likely the proximate cause. The rhythmic dancing on the floor exposed a truss deficiency that prevented the truss system from supporting code-prescribed loads.
Building Code
The clubhouse was a wood-framed, two-level structure identified as an A-3 Assembly on the Certificate of Occupancy. The applicable building code for the building was the 2000 International Building Code (IBC). The IBC specifically associated dance halls to the A-3 Assembly and prescribed a 100 pound per square foot (psf) design live load. The “old-time” dance halls probably did not consider rhythmic jumping; however, the prescribed 100 psf static load appeared reasonable for a community building in an apartment building complex.
The local Fire Marshal limited the maximum number of occupants to 135 for the main floor level based on documents made available during the investigation. The IBC assigned a maximum of 5 square feet per occupant for standing space in an Assembly space. The collapsed floor area measured approximately 20 feet x 30 feet to imply at least 120 individuals could congregate in the area.
Collapsed Floor Observations
The collapsed floor framing consisted of 24-inch-deep metal plate connected wood trusses spaced 16 inches on center with a measured length of approximately 20 feet-9 inches. The chord and web lumber grades were noted, and plate sizes were measured. The truss lumber failure locations were predominately limited to the truss end that intersected the stair wall (Figure 1). On the opposite end, the floor framing rotated and fell against the exterior wall with the broken ends resting on the lower-level floor. The truss bottom chord (tension member) in the center one-third of the span did not appear to be damaged and the metal truss plates were fully embedded. The truss evaluation was made to address three specific questions:
Was the truss designed for the code-prescribed 100 psf live load?
Is a 100 psf live load sufficient to support a rhythmic dancing dynamic load?
Why did the floor collapse?
The Truss Plate Institute’s 1995 Edition of the National Design Standard for Metal Plate Connected Wood Truss Construction (TPI) as referenced in the 2000 IBC was used to collect data and perform an analysis of the floor truss component.
Initial Truss Analysis
The wood truss design drawings were requested but not available. The truss midspan bottom chord tension stress was calculated for a 115 psf total load and compared to the published allowable tension design value (Ft) for the noted truss chord’s lumber grade. (The 115 psf total load is the sum of a 100 psf live load and the historical truss industry common practice of a total 15 psf top and bottom chord dead load). The wood chord member was found to be structurally sufficient for the calculated tension force. Additionally, metal connector plates at truss panel points were found to have sufficient coverage for the 115 psf total load. Therefore, it was concluded that the truss was originally designed for the code-prescribed 100 psf live load.
Dynamic Truss Analysis
There are no wood truss industry design guidelines available to evaluate dynamic loads for a group of individuals who perform rhythmic dancing or jumping on a floor. Research revealed studies that reported an amplification factor on a person’s body weight as they ascended or descended a staircase. A 1998 study by Stuart C. Kerr found an enhancement factor of 4 times the body weight for a large group (greater than 25 people) ascending or descending a flexible staircase. With this research information, the wood truss ultimate tension chord capacity was used to evaluate the code-prescribed 100 psf design load for dynamic loads.
The published Ft was increased by the 2.1 general adjustment factor to approximate the lumber ultimate tension strength value. (The 2.1 factor includes a 1.6 factor for duration of load the author considered acceptable for rhythmic dancing or jumping). It was found the truss could support a 255 psf maximum uniform static live load before the truss bottom chord might break. Additionally, the dynamic floor capacity was investigated using the 255 psf live load and 20 feet x 30 feet floor area. It was determined that 170 individuals could be on the 20 feet x 30 feet floor area using a 225 pound per person weight and enhancement/amplification factor of 4. (The 170 occupants exceeds the 120 people determined using the 5 square feet per person IBC limit.) Based on the analysis, a wood truss floor system designed to a code prescribed 100 psf static live load is anticipated to perform without failure/collapse when subjected to dynamic loads by the occupants.
Subsequent to our initial dynamic floor analysis, M.A. Broers et al. (2021) published Residential Floor Failures from Dynamic Occupant Loading that gives a procedure to evaluate the dynamic impact of jumping on a wood floor. The study found amplification factors ranged between 1.55 and 3.2 and recommended that a 2.5 factor be applied to the static code-prescribed live load. The author’s analysis and conclusions were consistent with the Broers study. Therefore, a wood truss floor system designed for a 100 psf static live load should not have failed when subjected to dynamic loading.
Improper Truss Bearing Design Caused Floor Truss Failure
The investigation focused on the truss end bearing condition when it became evident a floor truss system designed to a 100 psf live load should be able to support dynamic loads. The source of the truss failure appeared to originate at the interior stair wall. The 10-inch-wide double stud wall consisted of a 2×6 stud adjacent to the stairs, a 1-inch air space, and 2×4 stud wall that was positioned within the first bottom chord truss panel (Figure 2). As load was applied, the primary truss bearing support became the 2×4 plates/wall instead of the 2×6 wall located below the double vertical and diagonal web at the truss end. The absence of a truss web element above the 2×4 wall changed the intended load path and caused the bottom chord to bend until breaking which precipitated the floor collapse when dynamically loaded. The metal connector plate buckling or tooth “back-out” was a result of high localized forces (Figure 3). The splintered top edge of the truss bottom chord at the inside edge of the 2×4 wall identified the de facto primary truss support locations (Figure 4). The buckled metal connector plate and wood fracture at the interior face of the wall are signs of improper design of the truss bearing.
The truss industry publishes typical floor truss bottom chord bearing details and four examples are depicted in Figure 5. In each condition, a minimum of one vertical web and a portion of the metal connector plate is located immediately above the bearing wall. For this truss failure, the webs and metal connector plates were absent above the 2×4 wall that became the primary truss support even though the 2×6 wall was solely intended to serve that purpose.
The truss bottom chord capacity at the inside edge of the 2×4 wall was investigated to determine the approximate maximum uniform design load the truss could support. The bending stress was calculated and compared to the published allowable stress design bending design value (Fb) that was increased for a ten-minute load duration factor (CD = 1.6), flat use factor (Cfu = 1.1), and repetitive member factor (Cr = 1.15). The tension component that would have reduced the truss design capacity a minimal amount was neglected. The truss uniform design live load was calculated to be 50 psf or half of the original 100 psf live load. Therefore, the wall positioned below the first bottom chord truss panel where webs were absent was the reason the truss failed. The rhythmic dancing exposed the truss’s weak point from the improper design. At least one web should have been located over the 2×4 portion of the double stud wall. All webs were located over the 2×6 portion of the double stud wall which permitted the bottom chord to fail in bending as the truss was loaded to less than design capacity.
Conclusion
The code prescribed design live load of 100 psf was found to be sufficient for static and dynamic live loads when metal plate connected wood trusses are designed and fabricated properly. The rhythmic jumping (i.e., dancing) by a significant number of individuals as cited by engineers and media outlets was not the primary reason for the Woodlands at Clemson clubhouse floor collapse. A truss bearing design error reduced the floor truss capacity by 50% and the dynamic load exposed the truss design deficiency. The absence of a truss web above the inner truss bearing location created by a double wall placed the bottom chord in bending and tension. This configuration changed the load path which exceeded the lumber capacity and resulted in failure when subjected to a less-than-code-prescribed load. ■
About the Author
Scott D. Coffman, PE is a Senior Engineer with REI Engineers, Inc. in Westminster, South Carolina, that provides forensic engineering services. He can be reached at scoffman@reiengineers.com.
References
- Stuart C. Kerr. 1998. Human Induced Loading on Staircases
- M. A. Broers et al. 2021. Residential Floor Failures from Dynamic Occupant Loading. American Society of Civil Engineers (ASCE)