Building pounding during earthquakes: how ground motion duration affects response

When two buildings stand close enough to collide during an earthquake, the consequences can be severe—and, according to a new study, how long the ground shakes may matter far more than engineers have assumed. The research, published in the Bulletin of Earthquake Engineering, reveals that for certain pairs of adjacent buildings, short-duration earthquakes can produce dramatically larger structural responses than long-duration events, with displacement amplifications exceeding three hundred percent in the most vulnerable configurations. The findings carry direct implications for seismic design codes, particularly in regions where brief, sharp earthquakes are common.

The phenomenon at the heart of the study is known as seismic pounding: the collision between adjacent structures that swing out of phase during ground shaking. Damage from pounding has been documented repeatedly after major earthquakes, from collapsed stairwells and crushed floor slabs to complete structural failures at the interfaces between neighboring buildings. When buildings of different heights or stiffnesses sit side by side, they tend to vibrate at different natural frequencies, and these mismatches in dynamic behavior can cause them to hammer into one another at mid-height or at roof level. Engineers have long recognized the role of the gap width, the relative masses, and the period ratio between the two structures. What has been far less understood is the role of ground motion duration—essentially, how long the shaking lasts.

The research team, led by Mahmoud Miari of Shenyang University of Technology in China, together with Saman Yaghmaei-Sabegh of the University of Tabriz in Iran, Robert Jankowski of Gdańsk University of Technology in Poland, and Chunwei Zhang of Shenyang University of Technology, set out to quantify this neglected variable. Their investigation focused on three pairs of adjacent buildings, each consisting of a fixed four-storey structure paired with a taller neighbor: a five-storey, a seven-storey, and a nine-storey building. These configurations correspond to fundamental period ratios of 1.25, 1.76, and 2.29, respectively—a range that allowed the researchers to systematically explore how the degree of dynamic mismatch between two structures influences their sensitivity to shaking duration.

To do this rigorously, the team assembled an extensive library of ground motion records, dividing them into two ensembles based on a widely used metric called significant duration. This measure, typically defined as the time interval over which the ground motion accumulates between 5 and 95 percent of its total Arias intensity—a measure of energy content—captures the period of strongest shaking. Records with significant durations greater than 25 seconds were classified as long-duration motions, while those below 25 seconds were classified as short-duration motions. In total, 77 ensembles of paired records were used, ensuring that the two categories were broadly comparable in terms of spectral content so that duration itself, rather than amplitude or frequency characteristics, would emerge as the distinguishing variable.

Each building pair was then subjected to both categories of ground motion in detailed numerical simulations, and the responses were evaluated across five key engineering demand parameters: peak floor acceleration, peak displacement, base shear—the total horizontal force transmitted to the foundation—and the pounding force generated at the point of collision. The team also computed fragility estimates, which express the probability that a structure will reach a defined damage state given a certain level of ground shaking intensity, in this case quantified by peak ground acceleration.

For the building pairs with period ratios of 1.25 and 1.76—the 4–5 and 4–7 storey configurations—the influence of duration turned out to be modest. Response quantities showed minor to moderate differences between short- and long-duration inputs, suggesting that when two buildings are reasonably similar in their dynamic characteristics, the length of shaking is not the dominant factor in pounding severity. This result aligns with a substantial body of earlier research suggesting that, for many structures, spectral acceleration at the fundamental period is the primary driver of response, and duration plays a secondary role.

The picture changed dramatically, however, at the highest period ratio examined. For the 4–9 storey configuration, with a period ratio of 2.29, short-duration motions produced drastically higher amplifications than their long-duration counterparts. Peak floor accelerations increased by 103 percent under short-duration shaking relative to long-duration events. More striking still, displacement amplifications jumped by 321 percent, and both base shear and pounding force amplifications rose by 196 percent. In practical terms, the same buildings subjected to an energetic but brief earthquake experienced far more violent collisions and far larger internal forces than when shaken by a prolonged event of comparable intensity.

The fragility analysis reinforced this pattern in probabilistic terms. At a peak ground acceleration of 1.5 g—a level representative of severe earthquake loading—the probability of damage rose by 53 percentage points for the four-storey building and by 41 percentage points for the nine-storey building when they were subjected to short-duration motions compared with long-duration ones. In other words, under identical levels of ground shaking intensity, the risk of significant damage was substantially higher simply because the shaking was brief.

The study identifies a critical threshold between period ratios of 1.76 and 2.29, beyond which the dynamic interaction between adjacent buildings becomes highly sensitive to ground motion duration. Below this threshold, engineers may be justified in treating duration as a secondary consideration; above it, ignoring duration could lead to a serious underestimation of pounding risk. The physical explanation likely lies in the interaction between the timing of collisions and the phase relationship between the two structures. When two buildings are strongly mismatched in their periods, their relative motion can build rapidly out of phase, and short, impulsive ground motions may deliver their energy in a way that maximizes the closing velocity between the structures at the moment of impact—producing sharper, more damaging collisions than the more gradually developing response under prolonged excitation.

The implications for practice are significant. Modern seismic design codes and most conventional ground motion selection procedures do not explicitly distinguish between short- and long-duration records for ordinary building design, relying instead on spectral shape and intensity measures to characterize hazard. The new results suggest that for adjacent buildings with large period mismatches, this approach may be inadequate. In regions prone to short-duration earthquakes—such as areas of shallow crustal faulting where rupture directivity and short source-to-site distances produce brief, high-intensity shaking—the pounding risk between closely spaced structures with dissimilar dynamic properties could be substantially higher than current design procedures indicate. Conversely, in subduction zones and other settings that generate long-duration shaking, the same building pairs may perform better than feared.

The research also adds an important dimension to the growing literature on ground motion duration effects. Previous studies have shown that duration influences the collapse capacity of reinforced concrete frames, the degradation of stiffness and strength under cyclic loading, and the accumulation of damage in masonry structures—effects that stem largely from the increased number of damaging cycles in longer records. This study is distinctive in demonstrating that, for the specific problem of structural pounding, the direction of the duration effect can reverse: rather than long-duration motions being more damaging through cumulative cycling, it is the short-duration records that amplify the interaction between mismatched buildings, a finding that challenges intuitive expectations.

For densely built urban environments around the world—where seismically separated buildings are often constructed with inadequate gaps due to land constraints and property boundaries—the study provides a quantitative basis for reassessing pounding vulnerability. The authors underscore that explicit consideration of duration effects is needed in the seismic design of adjacent buildings with large period mismatches, particularly in regions prone to short-duration earthquakes. As cities continue to densify and earthquake-prone regions continue to urbanize, understanding not just how hard the ground shakes, but for how long, may prove essential to keeping neighboring buildings from becoming inadvertent battering rams against one another.