Seismic Double Stress

Before any of the engineering, this is a hard one to write. The toll from the earthquakes that struck Venezuela on the evening of 24 June 2026 is still moving: early official counts had already confirmed hundreds of deaths, thousands of injuries, and many people still missing or trapped as these words go down. The people of Venezuela, and the large Venezuelan community far beyond its borders, are carrying a loss that no technical account can lighten. This piece is written in the belief that understanding why these particular buildings failed is part of how the next ones are kept standing.

Two earthquakes hit northern Venezuela that evening: a magnitude 7.2 shock, followed thirty-nine seconds later by a 7.5. Many of the buildings that collapsed were likely brought down not by the first event but by the second, which arrived before anything could be done about the damage the first had already inflicted. Seismologists call this pairing a doublet. For a structural engineer, it exposes an assumption sitting quietly at the centre of almost every seismic code we use: that the design earthquake happens once.

The raw energy was never in doubt. The 7.5 was Venezuela's largest earthquake in more than a century, and its waves radiated out through the whole planet. Seismometers across Europe, an ocean away, registered the wavefront as it crossed the continent. Visualisations of large earthquakes render each recording station as a dot that rises and falls as the surface waves pass through; the schematic below illustrates that idea rather than plotting the actual station records. But the destruction in Venezuela was never a simple function of that energy. It was decided building by building, by the things engineers actually control: ductility, detailing, soft storeys, soil response, residual drift, and the capacity left when the second shock arrived.

Buildings survive by breaking

It is worth being clear about what seismic design actually promises. Outside of hospitals and a handful of critical facilities, we do not design ordinary buildings to ride out a major earthquake undamaged. We design them not to collapse while taking heavy, deliberate damage. The structure is detailed so that specific regions, the ends of beams, the bases of columns, the reinforcement itself, yield and deform well beyond their elastic limit, absorbing the earthquake's energy by bending rather than tearing apart. Cracked concrete, spalled cover and yielded steel are not failures of the design. They are the design working.

This is the logic of ductility, and it runs on a finite budget. A ductile frame can absorb a certain quantity of inelastic deformation before its capacity is exhausted. The design earthquake is calibrated to consume a large share of that budget and leave a margin against collapse. The unspoken premise is that once the shaking stops, the building has time to be inspected, shored, evacuated, condemned, or repaired, before it is ever asked to do that work again.

A doublet removes the time.

What the second shock meets

Thirty-nine seconds is not an inspection window. It is barely long enough for people to reach a doorway. When the second rupture followed, the structures in its path were no longer the structures the engineers had designed. The first event had already cracked concrete, opened beam-column joints, yielded reinforcement, and in many cases left buildings with a permanent lean: residual drift that does not recover when the ground stops moving.

Each of those changes lowers the building's defences for the event still to come. Stiffness drops, so the frame deflects further under the same demand. Strength drops, as confined concrete crushes and longitudinal bars, stripped of their cover, begin to buckle. Where the first shock left a residual lean, gravity turns against the structure: the weight of the floors now acts through a displaced geometry, and the resulting P-delta moments amplify the demand and erode whatever collapse margin remained. Reinforcement that survived the first set of large strain reversals can fracture under the second through low-cycle fatigue. Steel does not care that the two events were technically separate earthquakes; it counts cycles.

So the second 7.5 did not strike a population of healthy buildings. In many cases it struck buildings that were already part-way through failure, with much of their energy-dissipating capacity spent and no chance to recover it. A full forensic timeline, establishing which shock finished which building, will take months; but the structural problem was immediate, and it is a large part of why the casualty figures, still changing as rescue work continues, rose so fast.

What the failures looked like

The footage from Caracas and La Guaira appears consistent with failure arriving in more than one form, and the differences are not cosmetic. Each pattern points to a different weakness, though assigning a mechanism to any single building will take formal assessment rather than video.

Some of the footage appears consistent with damage concentrating in a single storey, almost always the lowest one. Where a ground floor is left open for parking or shops while the floors above are stiffened by infill walls, that open level is the weak link: it has far less stiffness and strength than everything above it, so the whole earthquake's drift demand is forced into one storey. The columns there reach their limit, the level crushes, and the building above drops more or less whole onto what used to be the ground floor. This soft-storey collapse is one of the most dangerous and recurrent patterns in reinforced-concrete buildings, and it is fast, which is part of why so little warning preceded the worst of it.

Other footage appears to show failure at the top. A tall, slender frame does not sway as a rigid stick; its upper storeys can whip under the higher modes of vibration, and where those floors are lighter, added later, or simply less well tied together, the top section can shear and topple while the lower building stands. From the street it reads as the building losing its head.

And some buildings appear to have gone down all at once. Once any single storey loses its columns, the floor above falls onto the floor below, and the impact of that falling mass overloads the next level, which fails in turn. The collapse runs down the building floor by floor, faster than anything can arrest it, until the whole structure is a stack of slabs a few metres high. This is progressive, or pancake, collapse, and it is consistent with the total losses reported, including a twenty-two-storey building in Altamira said to have collapsed entirely.

On the coast at La Guaira, a further mechanism appears to have been at work. Rows of housing were left visibly shifted off their foundations, displaced sideways as whole units. The likely explanation is ground failure: soft, saturated coastal soils amplify shaking and can briefly lose strength under it, a behaviour related to liquefaction, so the ground itself moves and carries the structure with it, or stops holding the footings in place. First-order hazard assessments flagged high liquefaction and landslide potential for this event, though the exact effects of the rapid double shock remain uncertain. Where this happens, a building can be left more or less intact and still be lost, because what failed was underneath it.

Venezuela's older lesson

There is a reason the most experienced engineers in the region were not surprised by where the damage concentrated. Venezuela has taught a version of this lesson before, with a far smaller earthquake.

In July 1967 a quake of about magnitude 6.5 struck with its epicentre tens of kilometres from Caracas. By magnitude it was unremarkable. Yet it collapsed four ten- to twelve-storey apartment buildings in the Los Palos Grandes and Altamira districts and killed somewhere between 225 and 300 people. The reason was geology. Caracas sits on a deep alluvial basin, and that basin behaves like an amplifier, taking the bedrock motion and magnifying it, by a factor of three or more in the studies that followed, within a particular band of periods. The damage was not random. It tracked the depth of the sediment, and, more tellingly, the height of the worst-hit buildings tracked it too: the tall towers came down where the basin was deepest, because their natural period matched the period the ground was amplifying. The buildings and the soil were tuned to the same note.

That earthquake became one of the founding case studies in site-effect engineering, precisely because it proved that the number on the magnitude scale is a poor predictor of what a structure actually experiences. What matters is the demand delivered to the building, shaped by the soil beneath it, the duration of shaking, and the direction of rupture, set against the real capacity of the building, which is governed by its detailing far more than by its drawings.

Two ways to double the stress

Seen side by side, 1967 and 2026 are two versions of the same problem. In 1967 a moderate earthquake was made devastating because the ground doubled the demand before it ever reached the buildings. In 2026 large earthquakes were made more devastating because time collapsed two demands into one, and the structures were never allowed to recover between them. One doubles the load through geology; the other doubles it through timing. In both, the headline magnitude understates the stress the structure was actually made to carry.

What pushes either scenario from damage into collapse is the condition of the building stock. Non-ductile reinforced-concrete frames, with inadequate confinement at the joints and columns weaker than the beams they support; soft ground storeys left open beneath stiff infilled floors above; unreinforced masonry that has no ductility to spend in the first place; and, across a country under prolonged economic strain, a decade of deferred maintenance on all of it. These are not exotic conditions. They describe a large share of the mid-rise stock in many rapidly urbanising cities, including across East Africa, where seismic provisions are unevenly adopted and enforcement is often limited, and where the single-event assumption is just as embedded in everyday practice.

The assumption worth revisiting

The practical takeaway for the profession is narrow and concrete. Our codes size structures against a design event treated as a single occurrence, and our post-earthquake procedures, tagging a building safe, unsafe, or restricted, assume there will be hours or days to make that judgement before the next major shock. A doublet voids both assumptions at once. It is rare enough that designing every ordinary building against back-to-back design events would be neither affordable nor proportionate. But for critical facilities, for regions sitting on fault systems known to produce paired ruptures, and for the rapid-assessment protocols we lean on after any large earthquake, the Venezuela sequence is a reminder that the second shock is a real load case, and that a structure's capacity to survive it is only ever whatever the first shock left behind.

Sources: earthquake parameters for the June 2026 sequence are per the U.S. Geological Survey, which classified the events as a doublet: a magnitude 7.2 foreshock followed 39 seconds later by a magnitude 7.5 mainshock. Casualty figures changed quickly after the first official counts and are expected to keep changing while rescue and recovery work continues. The 1967 Caracas analysis draws on the USGS ground-amplification field studies by Espinosa and Algermissen and the structural and soil-response work of Seed and Whitman, later formalised in Venezuela's seismic microzonation programmes. Descriptions of how individual buildings failed are preliminary, drawn from early footage and first-order hazard assessments; formal post-earthquake evaluation will take months. The observation on seismic-code adoption and enforcement across the region reflects published reviews of African seismic design codes.