At the Portsmouth Naval Shipyard (PNSY) in Kittery, Maine, the Naval Sea Systems Command repairs and upgrades the U.S. Navy’s fleet of nuclear-powered submarines. To facilitate this work, the PNSY owns and maintains a system of removable submarine covers (RSCs) that provide protection and access to the submarine exterior during maintenance operations. The RSC system is comprised of modular wall-stacks and stair-towers lined along each side of the submarine, roof covers spanning over the top, and closure units on the ends.
Four to five individual steel-framed and steel-clad modules stacked atop each other comprise a typical modular wall stack. The column base plates of each upper module rest on the column cap plates of the module below. At each interface, 12 single-bolt connections with perpendicular slotted holes serve as the only module to-module connection.
A typical module wall stack is approximately 50 feet high when fully assembled. Each module within a wall stack is rectangular in plan, roughly 6.5 feet wide by 47.5 feet long with three approximately equal bays. In contrast, a typical stair-tower consists of a single continuous module, approximately 15.5 feet square in plan and 55 feet high. Framing for both the wall-stacks and stair-towers consists of vertical and horizontal steel tubes connected at fully welded joints.
A windy day
During 2013 and 2014, the RSCs were undergoing restoration and modernization. Several assembled RSC structures were standing vertically near the shoreline at the PNSY when a strong wind gust caused a 50-foot-tall wall-stack to overturn. As the wall-stack fell, it impacted a stair-tower, which then also overturned and fell onto a small escarpment and a concrete barrier.
The force of the impact damaged and distorted portions of the steel framing and cladding of both structures. All bolts connecting the individual modules in the wall-stack failed as a result of the impact. The contractor responsible for the RSC restoration work retained Simpson Gumpertz & Heger Inc. (SGH) to perform a structural condition assessment and design repairs to the damaged modules, with the unique challenge of verifying that the repaired modules could be properly aligned, assembled, and returned to service as part of the full RSC system.
SGH personnel visited the site and evaluated members and all visible portions of joints on each module — both in the zone of impact and away from the zone of impact — for the purpose of identifying twist and sweep in members, verticality and straightness of columns, evidence of impact damage, distressed paint over welds that could indicate damage to connections, and global twist, sweep, and racking of each module.
SGH retained the services of a registered surveyor to perform a comprehensive total station survey of each damaged module. The survey data was vital for accurately identifying unacceptable distortion and misalignment of members (local damage) and of an entire module (global damage). SGH specified multiple survey points along most primary structural members, at the ends of the slotted holes in the connection plates, and at joints. At the time of the survey, the 55-foot-tall stair-tower rested horizontally on leveled timber cribbing at every other story. Each of the wall-stack modules rested upright on leveled timber cribbing at each corner.
Proper inspection of all welded joints required full removal of paint covering each joint — a potentially costly and time-consuming proposition given the large number of joints in each module. To help reduce the number of inspected joints while maintaining a high degree of confidence that damaged welds would be identified, SGH developed a sampling program for inspecting welded joints using the statistical approach presented in American National Standard Institute (ANSI)/American Society for Quality (ASQ) Z1.4-2008.
SGH applied a tightened inspection level (Level III) to joints in the impact zone, which were more likely to have suffered damage, and a reduced inspection level (Level I) to joints away from the impact zone. The inspection level determines the relationship between the total number of joints and the number of joints to be tested, as well as the acceptance quality limits.
SGH selected which joints to test up to the minimum numbers determined using ANSI/ASQ Z1.4-2008, excluding welds with visually apparent cracks, as it was clear that those joints must be repaired. An AWS-certified welding inspector tested each weld in each selected joint using magnetic particle inspection to detect surface and near-surface cracks.
The inspection program required that in the event that a cracked weld was discovered and confirmed during testing of joints in the impact zone, all other joints in the impact zone would be tested. If a cracked weld was discovered and confirmed during testing of the joints away from the impact zone, then the inspection level would be increased to Level II, resulting in additional joints that would need to be tested. In the end, the inspection did not reveal any nonvisible weld cracks but did expose original fabrication imperfections (lack of fusion and incomplete welds).
Geometric acceptance criteria
All modules contained some limited amount of distortion and misalignment prior to the fall due to the limitations of the original fabrication and construction processes. SGH established acceptance criteria to evaluate the damaged modules that accounted for the permissible variations defined in industry standards for steel construction, including permissible variations in sweep of beam and column members as specified in ASTM A500, and the AISC Code of Standard Practice for Steel Buildings and Bridges (COSP), which provided the permissible variations for column verticality (length/500) and, by extension, permissible limits for global racking, twist, and sweep of the individual modules.
Where necessary, SGH adjusted the acceptance criteria to account for the accuracy of the total station survey. To evaluate the alignment of the module-to-module connections in the stacked modules, SGH considered the requirements of the Research Council on Structural Connections Specification for Structural Joints Using High-Strength Bolts (RCSC Specification), which limits the slope of the surface of plates in a connection to 1:20 with respect to a plane normal to the bolt axis. Lastly, SGH visually evaluated local puncture damage of steel members and damage to wall panels, guardrails, and other nonstructural elements to determine the need for repair or replacement.
Modeling and analysis
SGH converted the survey results into three-dimensional (3D) points and then leveled the data points by aligning survey points at the lower corners of each module onto a common plane to minimize the effects of levelness of the support points. SGH established the centerline of each surveyed member and created AutoCAD 3D models of each module that also included the locations of the slotted holes in the module-to-module connections.
The 3D models of the surveyed modules allowed SGH to confirm visual observations and to quantify local and global deformations in each damaged module and compare those values to the established acceptance criteria. This evaluation revealed that seven tube members exceeded the permissible limit for local straightness and two modules exceeded the permissible limit for global straightness (i.e., the end bays had rotated in plan relative to the middle bays). More than one module exhibited evidence of global twist about their horizontal longitudinal axis; however, under closer scrutiny, it appeared that the observed twist was due to limited and uneven support points during the survey.
To confirm, SGH studied the raw survey data, quantified the difference in elevation of the support points, and then performed a finite element analysis (FEA) of the affected modules with the uneven supports. The FEA showed that the anticipated deflection of the modules under self-weight with the supports provided at the time of the survey would result in a twist comparable to that reported by the survey. Therefore, the observed twist was not a result of permanent damage to the modules.
Finally, SGH combined the individual 3D models of the four wall-stack modules to evaluate the alignment of the slotted holes in the module-to-module connection plates and to confirm adequate alignment of the stacked modules. All but four of the 36 pairs of connection plates were within the permissible limits provided by the RCSC Specification. The misaligned connections occurred only at interfaces involving one of the modules exhibiting unacceptable global sweep; therefore, the entire stack could be properly assembled without issue once the contractor realigned those modules.
Repairs and realignment
The modeling and analysis procedures showed that all damaged modules could be repaired and salvaged. SGH developed repair drawings for the contractor that illustrated local repairs (member replacement, weld repairs, nonstructural repairs) and global repairs, including a realignment sequence to detach the misaligned end bays and reattach them in a position that satisfied the construction tolerances in the AISC COSP.
Similarly, for the stair-tower, SGH developed a procedure for installing temporary supports and bracing elements to allow the contractor to replace structural members without damaging or distorting the remainder of the module. After completing all the repairs, the contractor erected the stacked modules and was able to achieve proper alignment between modules and make all module-to-module connections.
Diego Arabbo, P.E., is a senior staff II – Structures, and Matt Gilbertson, S.E., P.E., is a senior staff I – Structures, both with Simpson Gumpertz & Heger Inc. (www.sgh.com). They may be contacted at email@example.com and firstname.lastname@example.org, respectively.