Journal of Structural Engineering

The collapse of conventional/nonhardened structures was a concern of the U.S. Department of Defense (DoD) for years before the collapse of the World Trade Center (WTC) towers during the terrorist attacks on September 11, 2011 (9-11), owing to the bombings of the Murrah Federal Building in Oklahoma City, the U.S. embassies in Africa, and the U.S. Marine barracks in Lebanon. Since 9-11, motivated by the lack of any meaningful U.S. progressive collapse design requirements, DoD has worked with the civilian community on a number of significant efforts to improve the design of buildings to resist disproportionate collapse. The DoD efforts have included laboratory and field experiments, numerical simulations, and development of design requirements. Synergy and coordination with the civilian community resulted in combined programs with the General Services Administration, guidance and feedback provided by the ASCE Structural Engineering Institute (SEI) Committee on Disproportionate Collapse Standards and Guidance (DCSG) and its members, and adoption of some European civilian approaches to progressive collapse design. A significant result of the DoD effort was the creation of Unified Facilities Criteria (UFC) 4-023-03, Design of Buildings to Resist Progressive Collapse. The approaches employed in UFC 4-023-03 are currently being evaluated and modified for civilian applications by the SEI DCSG committee. The development and underlying approaches used in UFC 4-023-03 are briefly summarized in this paper, as are the previous DoD laboratory and field tests and numerical simulations.

This paper presents an experimental and computational assessment of the performance of steel and reinforced concrete beam-column assemblies under monotonic vertical displacement of a center column, simulating a column removal scenario. The assemblies represent portions of structural framing systems designed as intermediate moment frames (IMFs) and special moment frames (SMFs) for Seismic Design Categories C and D, respectively. The steel IMF and SMF assemblies were designed in accordance with ANSI/AISC 341-02 by using prequalified moment connections specified in FEMA 350. The concrete IMF and SMF assemblies were designed and detailed in accordance with ACI 318-02 requirements. Each full-scale assembly comprises two beam spans and three columns, and downward displacements of the center column are imposed until failure. The study provides insight into the behavior and failure modes of the assemblies, including the development of catenary action. Both detailed and reduced finite-element models are developed, which capture the primary response characteristics and failure modes. Analyses with the reduced models can be executed rapidly without loss of accuracy, facilitating implementation in models of entire structural systems.

Progressive collapse resistance of an actual 11-story reinforced concrete structure following severe initial damage is studied experimentally and analytically. The initial damage was caused by simultaneous explosion (removal) of four first-floor neighboring columns and two second-floor perimeter deep beam segments. The structure resisted progressive collapse with a maximum permanent vertical displacement at the top of the exploded columns of only about 56 mm (2.2 in.). The response of the structure is evaluated analytically using different modeling methods. Beam growth and, in turn, the development of the beam axial compressive force are modeled and discussed. It is demonstrated that such axial compressive force can significantly affect progressive collapse resistance of the structure. The shortcomings of nonlinear modeling with commonly used plastic hinges are quantified and discussed. It is shown that such a modeling method ignores axial and flexural interaction in beams and can underestimate the resisting element internal forces and in turn progressive collapse resistance of the structure. By using fiber hinges in an analytical model, such interaction is accounted for and the local and global experimental data are closely estimated. The progressive collapse-resisting mechanisms primarily include the axial-flexural action of the second-floor deep beams and Vierendeel action of the flat plate system in floors above.

The vast majority of past research on blast-resistant structural design focuses on buildings, with limited attention directed specifically towards bridges. Although many of the same principles apply, bridges pose unique challenges that are not often encountered when designing buildings for blast resistance. Specifically, establishing standoff with bridges is difficult because they are intended to provide open access to the traveling public, and structural components are directly loaded rather than having loads transferred to them through a facade system. Thus, relative to buildings, bridge components may be exposed to large blast threats that can be in close proximity to the potential target. To address these unique challenges, experimental and computational research was carried out, through support from the National Cooperative Highway Research Program (NCHRP), to understand the behavior of blast-loaded concrete bridge members. Although spalling of concrete cover off the back of reinforced concrete walls subjected to blast loads is a well-understood phenomenon, specimens experimentally tested for the current research exhibited spalling of side-cover concrete, which previously has not been reported in the research literature. Using detailed finite-element models, this paper explains the cross-sectional response mechanisms that cause spalling of side-cover concrete in blast-loaded slender reinforced concrete members by numerically reproducing the behavior observed during the experimental testing program.

Assumptions must necessarily be made when the collapse response of structures is investigated using simulation models. The type and extent of modeling assumptions depend on the computational resources available, modeling expertise, and results sought. Modeling choices that are commonly made include planar versus three-dimensional (3D) representation, simplification of member response for modeling purposes, and the use of macroelements to mimic behavior instead of using elements that are based on fundamental constitutive relationships. Using four different types of models, this paper sheds light on the effect of some commonly employed approximations in collapse modeling. The models represent a 10-story seismically designed steel building and encompass computationally expedient planar and 3D macromodels as well as continuum models of individual frames and the full 3D structural system. After a validation exercise, the simulation models are exercised to investigate system collapse response when columns are forcibly removed and to highlight the effects of the various modeling approaches. The simulation studies show that the floor system contributes significantly to collapse response. It is also shown that well calibrated macromodels can be relied on for accuracy when modeling progressive collapse and that the results of planar analyses cannot always be viewed as conservative.

This paper investigates the robustness of moment-resisting steel frames that are typical of building construction in seismic regions before the 1994 Northridge earthquake against progressive (disproportionate) collapse. Uncertainties in the collapse demands and the resisting capacities of the connections in the frames are modeled probabilistically. The dominant connection failure mode, which involves fracture of the weld connecting the beam and column flanges under scenarios involving sudden column loss, is developed using a J-integral formulation of fracture demand and is characterized probabilistically. The connection behavior model is validated using connection test data from the SAC Project on steel frames conducted following the Northridge earthquake. The robustness of two three-story steel frames designed in the SAC Project is assessed by utilizing (a) the requirements in the new Unified Facilities Criteria (UFC), and (b) a system reliability analysis. This analysis reveals that steel moment frames with connections similar to those found in pre-Northridge building construction may not meet the UFC requirements for general structural integrity following notional column removal.

The redundancy of a structure refers to the extent of degradation the structure can suffer without losing some specified elements of its functionality. However, because future structural degradation is unknown during design and analysis, it is evident that structural redundancy is related to robustness against uncertainty. This paper proposes a quantitative and widely applicable concept of strong redundancy and shows its relation to the info-gap robustness of the structure. In particular, one of this paper’s propositions establishes general conditions in which the strong redundancy is equivalent to the robustness. This paper also defines a concept of weak redundancy and presents propositions that relate it to the strong redundancy and the robustness. Results are illustrated with several heuristic and engineering examples.

This paper describes damage patterns in sprayed fire resistive material (SFRM) on steel moment frame beam-column assemblages owing to a strong seismic event, and the thermal consequences of this damage when exposed to postearthquake fire. Large-scale experimental tests were performed to examine the bond of SFRM to steel in the three-dimensional configuration of a moment frame beam-column connection region. Two beam-column assemblages treated with SFRM were subject to quasi-static cyclic loading, resulting in large deformations and plastic hinges in the beam and local damage to the SFRM on the beam. Heat transfer finite-element analyses were performed to compare beam-column connections with damaged SFRM and fully insulated connections (with no damage) under the action of both standard and natural fires. Results of the heat transfer analyses show that SFRM damage on the beams adjacent to the column causes an increase in heat transferred into the column and elevated temperatures in the column. The elevated temperatures in the column cause important reductions in strength and stiffness of the steel.

This paper presents the results of experimental investigations conducted to determine the fundamental behavior of steel members under fire loading. A total of eleven full-scale steel members were tested under combined thermal and structural loading. First, five A992 steel beam-columns ($\mathrm{W}10\times 68$) were tested to determine their fundamental moment-curvature responses at elevated temperatures and different axial load levels. The experimental approach involved the use of radiant heating and control equipment to apply the thermal loading, and close-range photogrammetry combined with digital image processing techniques to measure the deformations (curvature) in the heated zone. Next, six A992 steel wide-flange ($\mathrm{W}8\times 35$ and $\mathrm{W}14\times 53$) columns were tested to determine their inelastic buckling behavior and axial load-displacement responses at elevated temperatures. A self-reacting test frame was designed to subject the column specimens to axial loading and heating. The thermal loading was applied by using the same type of radiant heating and control equipment as the beam-column specimens. The measured behaviors (and strengths) of the tested beam-column and columns specimens are presented and then compared with those obtained from detailed 3D finite-element analyses. The experimental investigations showed that the fundamental behavior and strength of steel members is governed mostly by the steel surface temperature, and the strength and stiffness of steel columns decreases significantly with increasing temperatures, particularly in the range from 500–600°C. The elevated temperature behavior of steel members can be predicted reasonably by using detailed 3D finite-element models. These verified models are recommended for conducting analytical parametric studies. The experimental approaches are recommended for evaluating the fire behavior of other structural members and loading conditions.

During a fire, columns on the perimeter of a building will be subject to moments induced by both a thermal gradient and the restraint of axial expansion by adjacent heated beams, which themselves develop axial load. These members thus act as beam-columns because they are then subject to a combination of axial load plus moment caused by a combination of gravity plus thermal loading. This paper presents a two-pronged procedure to predict the behavior of the perimeter column as a beam-column, considering both the individual member response (including thermal gradients) and the global response (including the interactions of adjacent members). All methods discussed in the paper are closed-form (i.e., they require no iteration) and can therefore be solved by using a spreadsheet or simple mathematical algorithm. The framework is sufficiently simple for use in codified structural-fire design and could be included in a reference of performance-based analysis methods for steel structures. Although this paper specifically addresses the performance of columns on the perimeter of buildings, the proposed framework can be a blueprint for the performance-based analysis of other beam-columns, such as floor beams.

This paper is concerned with the ultimate behavior of lightly reinforced concrete floor slabs under extreme loading conditions. Particular emphasis is given to examining the failure conditions of idealized composite slabs which become lightly reinforced in a fire situation as a result of the early loss of the steel deck. An experimental study is described which focuses on the response of two-way spanning floor slabs with various materials and geometric configurations. The tests enable direct assessment of the influence of a number of key parameters such as the reinforcement type, properties, and ratio on the ultimate response. The results also permit the development of simplified expressions that capture the influence of salient factors such as bond characteristics and reinforcement properties for predicting the ductility of lightly reinforced floor slabs. The companion paper complements the experimental observations with detailed numerical assessments of the ultimate response and proposes analytical models that predict failure of slab members by either reinforcement fracture or compressive crushing of concrete.

This paper describes numerical and analytical assessments of the ultimate response of floor slabs. Simplified analytical models and finite-element simulations are described and validated against the experimental results presented in the companion paper. The simplified analytical model accounts for membrane action and the underlying mechanisms related to failure of floor slabs by either reinforcement rupture or compressive crushing of the concrete. In this respect, the significant influence of material properties, including bond strength, is considered in the model and described in detail. A detailed nonlinear finite-element model is also employed to provide further verification of the simplified approach and to facilitate further understanding of the overall response. The results and observations of this study offer an insight into the key factors that govern the ultimate behavior. Finally, the models are applied under elevated temperature conditions to demonstrate their general applicability and reliability.

This paper presents the behavior observations, results, and findings from experimental investigations of the structural behavior of thin composite floor systems subjected to combined gravity loads and fire loading. The $3.96\times 4.57\mathrm{m}$ floor systems consisted of A992 steel $\mathrm{W}10\times 15$ interior beams and $\mathrm{W}12\times 16$ girders acting composite with a 38.1-mm-deep ribbed steel deck with 63.5 mm of lightweight concrete on top. Three composite floor-assembly specimens were tested with two different shear connection types (welded-bolted shear tab and all-bolted double-angle connection), two different fire scenarios (realistic fires with standard heating and uncontrolled or controlled cooling paths), and two different fire protection scenarios (i.e., interior beams with or without fire protection). The experimental results indicate that removal of fireproofing from the interior beams causes them to heat, deflect, and fail more rapidly. The beams and girders have similar deflection-versus-temperature behaviors irrespective of the fireproofing on the interior beams. The thin lightweight composite slab used in these tests contributes significantly to the load transfer from the interior beams to the girders, but it does not seem to be able to support the interior beams once they started failing. Removing the fire protection from the interior beams of thin lightweight composite slabs such as those tested in this paper is not recommended, unless better behavior can be demonstrated through future tests that include effects of neighboring floor systems.