The 8-foot-deep University Club of Boston (UClub) trusses are vital parts of the building’s structure. They support the roof deck structure, roof-mounted mechanical equipment, and the weight of hanging items inside. But are they robust enough to support the proposed additional loads? The facility is undertaking a significant renovation from top to bottom.
Our report analyzes the trusses and the design of the 1925 construction and current conditions. The report concludes that the 8-foot-deep trusses are structurally adequate to support the proposed loads, but what is the limit?
This report is a “leave behind” for the next generation of engineers who may be tasked with adding more load to the roof of the UClub. It provides a detailed analysis of the truss and its limitations, and it is hoped that this information will be helpful for the next proposed renovation designers.
Background
The University Club of Boston, located at 426 Stuart Street in Boston, is a premier social and athletic club located in the Back Bay neighborhood. The UClub was constructed in 1925 as one of many transitional buildings being built at that time, as a connecting link between the eight-story 40 Trinity Place Conference Center and Hotel building and the 14-story YWCA building, the three of which were one of the first commercial developments in Boston’s Back Bay. Since 1925, the Trinity Place building was rebuilt as a 40-story residential and commercial tower replacing the Conference Center and Hotel. The UClub facility is a four story steel frame and masonry building. It is topped at the roof level with six steel trusses, spanning approximately 72 feet supporting a reinforced concrete ribbed deck. It is believed that the current UClub served as a recreation function for the hotel in the past. The steel elements of the trusses are encased with concrete, presumably placed for fireproofing. Although one would expect that trusses with similar 72-foot spans, spacing and loading would be the same, we found in the field that all the trusses varied in their makeup and even in their parallelism. The front portion of the building roof is comprised of a reinforced concrete ribbed deck supported by steel beams encased in concrete. Refer to the SKS 1, Roof Plan Schematic and SKS 2, Truss Elevation for orientation.
The architectural design provided for the replacement of a hanging utility personnel walkway with an observation bridge utilizing glass railings located on the fourth floor (aka mezzanine) to provide a unique view of the recreation courts below. With uninterrupted sightlines being the main goal of the design, the placement of vertical Hollow Structural Steel (HSS) hangers from the trusses and the spans of horizontal members were optimized for viewing purposes. The new catwalk, observation bridge, and food preparation kitchen are connected to Truss 1, Truss 2, Truss 4, and Truss 5 using hangers. The bridge also serves as access to a UClub fit-out center in the newly constructed 40 Trinity Place building.
Evaluation of the Roof System’s Capacity to Support Additional Loads
In 2020, SOCOTEC, formerly CBI Consulting LLC, was assigned to evaluate and analyze the existing roof system and determine whether the truss components and concrete rib portions are structurally adequate to support the additional dead and live loads from a proposed rooftop terrace-pergola, 32,000 lbs. of mechanical equipment, the suspended catwalk, observation bridge, food preparation platform, and railing system being constructed at the mezzanine (aka 4th floor) level. From historical research, a study undertaken in 2011 concluded that the existing roof could support one additional floor level.
Determination of Loading
A 2011 analysis determined that structural system additional loading could sustain an additional live load of 100 psf in addition to dead loads, if properly suspended and positioned to the existing deck level framing members. It was decided to confine the proposed use program for the roof to one of a social gathering terrace function for a quarter of the roof surface adjacent to Stuart Street and the balance of the roof to the support of mechanical equipment and the support of the expansive catwalk/observation gallery.
The mechanical consultant provided the equipment loads, and SOCOTEC engineers, working with the architect, developed hanger loads from the expanded pedestrian and service areas to be hung from the trusses. There is a paucity of design data for snow loads for adjacencies with relatively short north-south fetch distances and high-rise site conditions like the YWCA and 40 Trinity Place buildings. Given the UClub’s structural performance to endure without distress, the historic east coast blizzard of 1978 when Boston received up to 30-inches of snow in one storm, and the 110-inch snow accumulation in the Boston winter season of 2015/2016, it was concluded the snow on the UClub roof from the adjacent abutting structures simply equates to the code-prescribed ground snow load. In the easterly direction, wind drifting snow against a newly constructed Stair #2 penthouse extension above the UClub roof was considered in the design.
Deflection Monitoring System
Out of an abundance of caution and curiosity, the authors recommended installing a deflection monitoring system along the length of the two most heavily loaded trusses, Truss 1 and Truss 5. This was done after discovering each truss is unique, dead and live loads can only be approximated with so much accuracy, and other impacts, including temperature changes, can affect the truss performance. Additionally, real-world deflections would serve as an invaluable cross-check of the calculated deflections (both by hand and with various analysis software).
Deflection charts obtained from the monitoring system for Truss 5 are attached below, with milestone dates and deflections highlighted. When reviewing the data in Table 1, note that:
Green highlight indicates the date when the observation bridge was installed.
Orange highlight indicates the date when the catwalk was installed.
Blue highlight indicates the date when the steel work was completed at the catwalk and deck level.
Red highlight indicates the date when concrete was placed at the observation deck and catwalk.
Analysis
We needed to compare its performance to our idealized engineering models. It is generally accepted that nearly all structural connections have some amount of fixity, but the extent of fixity provided by an archaic riveted gusset connection is unknown. It is also unknown whether the original truss designers counted on this fixity. Since the original designers left no reports, drawings, or memoirs, our modern team of engineers had to rely on the existing structure to answer these questions. To this end, we conducted various exploratory probes throughout the early and mid-phases of the project’s design. These probes revealed that the internal truss connections varied but generally consisted of large steel gusset plates and multi-riveted connections to the diagonal, vertical, and horizontal members.
Vierendeel Truss Action vs. Common Triangulated Truss
A Vierendeel truss is a type of truss made up of rigid rectangular frames with no diagonal members. In contrast, other common trusses are typically triangulated frames. Since a true Vierendeel truss does not utilize diagonal members to transfer forces, the UClub does not have a true Vierendeel truss configuration and instead uses modified Warren pin joints which have vertical and diagonal elements that are the only contributors to engage in stress transfer through moment (rigid frame) joints. An interesting analogy that structural engineers can appreciate is that top and bottom chords act as full span beams and the vertical and diagonal members functioning as horizontal shear (VQ/It) transfer members. (That equation is a bit of gallows humor that only a structural engineer can appreciate, as the “Q” is an item most students have spent considerable time with). “Q” is defined as the first moment of the steel area between the location where the shear stress is being calculated and the location where the shear stress is zero about the neutral axis.) In contrast to Vierendeel trusses, common triangulated trusses transfer loads as axial forces through diagonal as well vertical elements using tension and compression (PL/AE) (another bit of gallows humor for the structural engineer).
The As-Built Specifics
The truss consists mostly of double-angle diagonal and vertical members and double-angle top and bottom chords. Approximately two-thirds of the top chord length of Truss 6 consists of layered plates, riveted together and changing in length as one might expect to find in a turn-of-the-century long span beam. Little by little, the exposing of various locations, in various trusses revealed an illogical (haphazard) process, the reason for which was (is) left to one’s imagination. There are no as-built records.
The diagonal members were connected to the top and bottom chords using steel gusset plates. Rivets were used to make the connections. Refer to SKS 3 for an appreciation of a typical top chord/diagonals’ connection. The truss is configured as a Warren Truss as it consists of diagonal/vertical members.
All members are encased in approximately 12-inch to 18-inch thick concrete. The hefty concrete encasement was an interesting sight to see for the younger engineers, as it is seemingly counterintuitive to add so much dead weight to a long-span truss. It was concluded that the concrete encasement is merely fireproofing, however the thought lingers that encasement might also (although unreinforced) function as tension / compression bending member reinforcement.
Free-Body Diagrams for Use by Engineering Methods
Nevertheless, the truss can also be analyzed as a modified Vierendeel truss since it has large gusset plates and riveted connections at the panel points. Those connections provide fixity at the diagonal members. Referring to SKS 4 and 5 as examples, various approaches (degrees of fixity) were used to analyze one of the trusses: pinned joints for the Warren truss model and fixed joints for the Vierendeel truss model.
The trusses were evaluated using TEDDS Structural Analysis software using the Vierendeel truss model approach, which assumes fixity at the panel points. The truss was assumed to have vertical members only, keeping the same properties. The top and bottom chords of the truss were assumed to be a continuous beam, and the vertical (web) members were assumed to transfer the VQ/It horizontal shear stress. The digital truss did not respond well, as the deflection computed was significantly higher than what is evident in the real world (aka UClub). Note that various conditions were ignored/simplified in the analysis to provide for a clean model for engineering analysis. For example, the American Institute of Steel Construction’s (AISC) criteria for slenderness, width/thickness (w/t) ratios, etc., was not considered in the TEDDS model, nor was the improvement in those characteristics imparted by the concrete confinement.
For comparison, the trusses were again evaluated using the TEDDS software using a modified Vierendeel approach which also assumes fixity at the panel points. In this model, the trusses were assumed to have vertical and diagonal members, as in the actual truss. The top and bottom chords of the beam were assumed to be continuous members, and the vertical and diagonal members were assumed to transfer the bending stress. This would be similar to the case if we treated the truss as a deep beam with similar properties. See SKS 4.
For another comparison, the trusses were evaluated using TEDDS software as Warren Truss, assuming the panel points are pinned. From the deflection analysis, it could be noted that there was no significant change in the deflection, whether the diagonals were pinned or fixed. This might be due to the rotational stiffness of the chord member being significantly higher than the web members. See SKS 5. In a university setting, this would be an interesting research topic to delve into as to how and why slender members do not attract moment even though they are constructed with fixed connections.
For a fourth comparison, the truss was assumed to be a deep beam (say a W-section), and the deflection was calculated as such.
For a fifth comparison, the truss was assumed to be a composite deep beam with concrete encasement at the top flange. The transformed section properties of steel with the concrete encasement were determined, and the new section of the steel beam was used to determine the deflection of the beam (truss). Refer to SKS 6.
For completeness and precedence, recalling the collapse of the Hartford Coliseum in 1978 when one of the elements of the investigation was the “pressure cooker” effect on that structure resulting from a change of the building environment (aka heating/cooling/moisture) of the internal atmosphere, we used the thermal cycle that the mechanical systems of the UClub create for comparison with the sensor grades and found no troubling relationship. This was certainly not a scientific study but one more effort to “close the loop.”
To replicate such a study for the UClub, the authors discussed the internal condition of the building regarding heating, cooling, and moisture, with the UClub maintenance engineer to form an understanding of how the roof would move or “breathe” as the internal environment changed. Based on this conversation, it was determined that influences of internal breathing would not be applied as the maintenance engineer consistently maintains an ambient temperature of 72 degrees Fahrenheit and does not adjust the humidity (a little warm for playing sports but great for observing sports activities from an observation deck).
Conclusions and Results
The deflection of the truss with pinned diagonals and fixed diagonals appeared to have the same amount of deflection. This deflection is higher than the deflection calculated for any other simulation representing the 8-foot-deep truss. The deflection that was determined assuming the truss to be a deep beam with a composite section, appeared to be the smallest. The sketch that shows the comparison between the deflections for various cross sections is shown in SKS 7. Considering that no firsthand knowledge was available of how the trusses were fabricated and erected, the authors had to make assumptions as to the site constraints, transportation conditions in the city of Boston in 1925, fabrication and erection techniques, and the erection equipment and tools that were available for a project of this magnitude.
From that, it was surmised that the longest structural members available for shipment would have been no longer than 30 to 35 feet and that the individual truss elements were assembled by workers standing on a shoring system that provided a platform for landing steel pieces to be connected onsite using the hot rivet process by other workman standing on the same platform, and manhandling individual steel pieces raised to the working level using winches and cranes.
Using laser levels, we were able to determine that the trusses were, for all practical purposes, built dead level. Observing the form lines in the concrete and the concrete surfaces of the trusses has confirmed that standard wood construction was used to create forms to contain the concrete applied for fireproofing purposes. It is just a guess, but because of the levelness, the builders of 1925 using this technique realized that the fireproofing was not only serving the fire protection purpose but that the concrete was imparting significant stiffness to the overall truss. Taking that speculation further, this contributed to the development of a fiber wrap alternative for strengthening the two compression diagonals by turning them into concrete columns. Computer analysis of the trusses indicated that the steel elements, for the most part, had the requisite load capacity and that, more than likely, the original builders were not utilizing the concrete for strength purposes. Early on, using this information directed our attention to understanding how the connections functioned [steel connected with rivets], and removing concrete at these connections to evaluate capacity and supplement rivet capacities with welding, when necessary.
Given that the trusses were not all parallel and that misalignments were inconsistent, it was unsurprising that the south end of Truss #1 had a fabricated cantilever bracket extension as part of the top chord utilizing plates and angles that were supported on a fabricated header connected between two columns. For the duration of the project, the contractor called this end of the truss a “Wonky Pork Chop.” At all other bearing locations, the truss reaction load was delivered from the truss to large, rolled steel columns [pile section size] riveted to plates and, in turn, riveted to the columns. Although the roof members are not directly supporting drywall or plaster ceilings, Table 1604.3 from the 2015 International Building Code (IBC) is an exemplar for deflection criteria more familiar to the casual reader. The calculated and observed deflection was viewed for comparison with these suggested limits. See SKS 6 for deflection calculations.
While structural stiffness and strength are not necessarily sympathetic, viewing both with visual and computational means provides a level of comfort that the final product satisfies reasonable safety for the intended use. If, in the future, there develops a need to modify the existing structure and/or to change the imposed loading, one should proceed with caution as the UClub roof structure is a complicated beast.
The actual grades recorded by Feldman Associates, a digital deflection monitoring firm, contain unusual “blips” and gaps. While they can’t all be explained, the authors have assessed there is no cause for alarm. Gaps with no record probably resulted from disruption from disturbances during the construction of surrounding elements or workers traversing the bottom cord.
To make the study as accurate as practical, a steel piece was removed from Truss #1 that was serving a nonstructural purpose and was connected to the truss with rivets so that a coupon test could be conducted to determine the strength of the typical truss angle and the shear capacity of the field-driven rivets. The rivets tested to 17 kips per square inch in the elastic range and steel to the American Society for Testing and Materials (ASTM) A36, Standard Specification for Carbon Structural Steel criteria.
The inclusion of deflection results from a simulated standard beam consisting of top and bottom flanges that represent imaginary beams with an 8-foot depth, a uniform top and bottom flange cross section and an imaginary 3/8-inch-thick web. Considering how the truss was erected and how the concrete ribbed structural surface embedded the top chord, a top chord transformed section was created for comparison with other truss and beam deflections. ■
Note: Included in this article is a sample of the digital grades recorded for Truss 5. For a full report of both trusses and a description of site conditions and photographs that show construction interruptions that have created gaps in the recording process, interested parties should contact Angela Joshi at angela.joshi@socotec.us
About the Authors
Angela Joshi, B.E., M.S., EIT; Craig Barnes, B.S., M.S., MBA, PE, SE; Timothy Cella-Mowatt, B.S., PE, LEED AP; Jibreel Mustafa B.S., EIT, are structural engineers in the Boston office of SOCOTEC US.