Collapse of Mexico City metro bridge on May 3, 2021

Original article in french here.

A span of a Mexico City metro bridge collapsed when a train passed on May 3, 2021.

The purpose of this article is to provide some points of reflection on the causes of this type of accident. This reflection is based solely on the elements available in the press or on the internet. It is therefore limited and scalable. It is not a substitute for an expertise which requires a site visit and the examination of the plans of the structure.

1) GENERAL DESIGN OF THE STRUCTURE

The metro line concerned is line 12, built between 2006 and 2012. The bridge, built between 2008 and 2011, consists of multiple spans resting on reinforced concrete piers.

Each pier is made up of a circular section column surmounted by an inverted T section header, cantilevered on either side of the column.

Each span of the deck is made up of two reformed and welded profiles resting on the pier caps. Each steel beam is plumb with the outer rail of one of the metro tracks. Each span is independent of the neighboring spans. They are obviously beams on 2 isostatic supports.

The reformed and welded profiles used have the particularity of not having the lower and upper footings of the same width. The upper footing is significantly smaller in width than the lower footing. This arrangement is certainly intended to reduce the weight and cost of the steel beams by taking advantage of a connection with the concrete slab. However, in the event of rupture of the connection with the slab, these profiles are particularly vulnerable to spillage due to their geometry. The small width of the upper footing also leaves little room to make the connection with the concrete slab.

Between these beams are positioned spacers intended to prevent the beams from overturning. Taking into account their small section (double angles), these spacers had to have primarily a role of provisional maintenance pending the installation of the concrete slab below. The only large section spacer is on the east side to support a catenary for the south track. In the design of the structure, this spacer obviously does not have the role of stabilizing the steel beams.

On the beams is placed a concrete slab visibly cast on transverse pre-slabs or composed of prefabricated keyed concrete elements. This slab acts as a compression table for two mixed steel-concrete T-beams. Metal connectors (studs) welded to the upper footing of the steel beams provide the steel-concrete connection. The connectors have been welded on the steel beams on the site, before or after the installation of the precast concrete elements.

Principles of design and production of composite steel-concrete decks with prefabricated concrete elements (ENPC presentation extract):

Example of installation of connectors after installation of beams (Stud Welding and Fasteners):

The collapsed span presents a particularity compared to the current spans. It is located at the start of a widening of the deck on the west side which requires an additional steel beam supported on the North steel beam and on the West header.

The photos taken during construction confirm that the connectors have been welded on site after the installation of the steel beams, and, probably, for ease of access and installation, after the installation of the precast concrete elements. They also indicate that the additional metal beam supporting the widening of the deck has been welded to the North steel beam on site, after the installation of the main steel beams.

2) DESIGN OF THE RAILWAYS

The metro's traffic lanes are classic. These are railways comprising 2 standard gauge rails, laid on concrete sleepers. These concrete sleepers are placed on a ballast composed of a bed of stones or gravel. This ballast is placed on the concrete slab of the bridge deck.

The collapsed span supports railway switches which can generate shocks and transverse horizontal forces during lane changes, which the current spans are not subjected to.

3) SITUATION DURING THE COLLAPSE

The collapse occurred during the passage of a train on the south side track, while the train was moving from west to east. The position of the wagons after the collapse indicates that the 2 rear wagons were centered on the collapsed span, the length of both wagons being longer than the span.

In this position, the structural elements of the deck are most stressed in the middle of the span by bending forces. However, during the accident only the south side of the bridge was loaded. The northern route is free. The bridge therefore receives only half of the maximum vertical load to which it can currently be subjected. This load is arranged off-center in relation to the longitudinal axis of the apron, which induces torsional forces in the apron which are added to the bending forces.

4) CHARACTERISTICS OF THE BREAKAGE

Press photos taken shortly after the collapse:

Excerpts from video taken after removing the metro train:

The photos of the collapsed bridge evoke a breakage of the deck in the central part of the span. This type of breakage clearly indicates that it was the load on the subway train that triggered the collapse.

The spans on both sides are intact, which makes it unlikely that piers failed.

This is probably a failure of the deck under bending and torsional stress.

However, it is observed that the eastern part of the deck simply tipped while being little degraded and remaining supported on the eastern pier, while the western part escaped the western pier and was completely dislocated. The rupture phenomenon could therefore have originated in the western half of the span.

The position of the elements after collapse seems to indicate that there was no complete and immediate breakage in two sections of the steel beams in their middle but :

• a spillage of the steel beams under the bending and torsional forces in the central part, west side, and a rupture of the concrete slab or of the steel-concrete connectors in this zone,
• a bending of the spilled steel beams, having lost their vertical inertia, under the effect of the bending force (this is equivalent to the accidental formation of a plastic ball joint in the span which makes the structure unstable),
• a sliding of the West ends of the steel beams on their supports, due to their deformation in the span, and a tilting of the East ends which remained on their supports,
• an exit of the ends of the steel beams out of their western supports leading to the fall of the apron on the western side.

5) HYPOTHESIS ABOUT THE CAUSES OF THE BREAKAGE

The structural weakness that caused the collapse may be in an element of the deck or of a bearing of the deck on the pier caps.

• Strength defect of the concrete compression table.
• Resistance failure of the connectors between the steel beams and the concrete compression table.
• Lack of strength of steel beams, in particular at the welds
• Failure of a support devices (breakage, settlement) causing excessive stresses in the deck and its breakage in a chain.

Abnormal loads linked to the laying of the tracks or their operation, as well as forces linked to phenomena such as an earthquake may have played a role.

In view of the shape and brutality of the breakage, which simultaneously affects the entire width of the deck at mid span, the hypothesis of a resistance defect of the concrete slab acting as a compression table for T-beams or connectors of steel beams seems most likely at first glance. The breakage of the compression table and / or the connectors causes the tilting or the rupture of the steel beams which are not designed to resist the forces on their own.

The analysis below will be limited to apparent disorders and defects in photos. A complete diagnosis would require on-site verifications with destructive soundings, verifications by calculations on the basis of plans and surveys, and material quality analyzes.

5.1) WELDING, DEFORMATION AND CRACKING AT MID-SPAN

Examination of photos of the underside of the deck (2011 to 2019) at mid-span shows in the same alignment:

• Traces of infiltration at an open joint between pre-slabs or precast concrete elements, not visible in 2011.
• The presence of transverse welds in the lower footings of the steel beams.
• The weld of the 3rd beam, supporting the widening of the deck, on the North main steel beam.
• The lateral deformation (buckling), visible since 2011, of an angle spacer.
• Perhaps a lateral deformation of the upper footing of the South steel beam (check that this is not a deformation related to the Google Street View software or to the lens of the device). This deformation was not visible in 2011.
• A clear paint mark on the South steel beam core which could be a fault report following inspection.

There seems to be, at mid-span (most stressed section), a set of delicate points to be treated in the metal framework, one or more deformations of the metal framework and degradation of the concrete slab (defect of the sealing, cracking).

The deformation (buckling) of the spacer during construction indicates that it is compressed. Its section is however clearly insufficient to work in compression. Its tensile work is also compromised since it would take a significant lateral displacement of the steel beam lower footings for it to be stretched. This spacer seems insufficient in view of the forces brought by the 3rd beam on the North steel beams in this section.

The deformation of the upper footing of the South steel beam (if it is confirmed) could be the sign of the start of a spill due, in particular, to a connection fault with the slab and/or to the ineffectiveness of the spacer.

The comparison of the photos of the underside of the mid-span deck from 2011 to 2019 indicates that what may at first appear to be the opening of a joint between prefabricated elements could be the gradual opening of a crack linked to the setting in tension of the lower fiber of the concrete slab.

The deck functioning as a composite steel-concrete beam in TT, the tensile setting of the lower fiber of the slab shows a major dysfunction of the structure. In normal operation, the compression table of a composite T-beam must be in compression over its entire height, which excludes any cracking by tensioning. The tensioning of the compression table in this type of beam can be considered as a starting point of breakage.

The tensioning of the lower fiber of the compression slab may be due to vertical deflection and / or horizontal deformation of the deck. It may be due to a weakness in the metal structure, a fault in the steel-concrete connectors, or a quality defect in the concrete (keying, concrete poured on a pre-slab). The concomitance of horizontal and vertical deformation would be a sign of reformed and welded profile spillage.

Such an evolving crack should have alerted the inspector during periodic visits. The presence of a transverse joint between prefabricated elements at mid-span could have misled the operators as to the seriousness of the phenomenon if they did not know the initial state of this area.

These disorders could have led to the breakage of the apron.

5.2) STEEL BEAMS - CONCRETE SLAB CONNECTIONS

On 2011 photos taken from the north side, it is visible series of regular traces, along the steel beams, which could correspond to connection orifices of pre-slabs or prefabricated elements intended for sealing the connectors of the beams. A fault in the prefabrication of the concrete elements or a fault in the layout of the steel beams may have led to these reservations not being aligned with the connectors. In principle, the orifices should be located on the upper footing of the steel beams and therefore not visible on the underside of the apron.

If the connectors have been welded on the beams on site, after installation of precast concrete elements, new orifices had to be created in the prefabricated elements to allow their installation and keying.

If the connectors have been welded on the beams on the site, before the installation of the precast concrete elements, this defect could have several consequences depending on the operating mode:

• Connectors could have been cut to allow the installation of pre-slabs or prefabricated elements despite this defect without being reconstituted afterwards.
• Connectors could have been cut to allow the installation of the pre-slabs and then replaced by connectors welded on the site, after the installation of the pre-slabs and the opening of new orifices.
• The connectors could have been kept and the orifices in the concrete elements modified on site to allow installation.

Welds made on site could suffer from a defect in quality or protection against corrosion.

It appears that new orifices have been drilled to allow for the placement and keying of the connectors. Traces of holes are visible in places in the concrete along the edges of the steel beams footings. However, it seems to be very basic openings, the keying of which is uncertain.

Cutting, a fault in welding or sealing the connectors, would be particularly serious execution errors because they would considerably reduce the resistance of the beam (the compression table no longer playing its role). They can, on their own, cause the tilting and rupture of the deck (concrete slab and reformed and welded profiles) by bending and torsion.

In all cases, the resistance to the shearing forces of the zone of openings straddling the upper footing of the steel beams and the vacuum could be reduced. There is also the question of the compatibility of the reinforcement of the prefabricated elements with the offset of their supports.

The hypothesis of a resistance defect of the connectors seems corroborated by the lateral sliding of the South steel beam under the concrete slab, without significant damage to the underside of the slab during the rupture or during the fall of the deck. The connectors, if present, were sheared at their solder on the steel beam footing.

The deformation of the steel beams, if it is not due to the impact with the ground, could result from a spill due to a connection fault with the slab.

5.3) EXCENTREMENTS OF LOADS AND STRUCTURAL ELEMENTS

Examination of the 2011 photos of the deck soffit shows that the axis of the steel beams is not parallel to the longitudinal axis of the concrete slab. The distance from the South steel beam to the South edge of the slab varies from 1 to 1.4 from West to East. While in aerial view, the southern edge and the northern edge (up to the widening) are parallel, in the extension of the rectilinear spans on the eastern side.

This observation confirms a fault in the implantation of the steel beams (and possibly of the West support), mentioned above concerning the offset of the orifices in the prefabricated slab elements with respect to the axis of the steel beams.

The gaps between the orifices of the prefabricated elements of the concrete slab and the upper footings of the steel beams are visible in the photos taken during the construction. The misalignments of the axes of the steel beams of the collapsed span with the axes of the steel beams of the neighboring spans are also visible in these photos.

This lack of installation would also be liable to have other consequences, in particular a variable eccentricity of the vertical axis of the rails and sleepers with respect to the vertical axis of the reformed and welded profile. This eccentricity having the second consequence, when loading the South track:

• The increase in the transverse bending moment in the concrete slab between the 2 steel beams, west side, can lead to its breakage.
• The increase in the torsional moment around the axis of the South steel beam, resulting from the deformation of the slab, on the west side, which can lead to tensile forces in the connectors (if still in place), which can lead to the buckling or the rupture of the steel beams.
• The increase in horizontal and vertical forces and a parasitic torsional moment in the North steel beam, transmitted by the small section spacers punctually connecting the 2 steel beams, which could lead to its overturning or its breakage.

This phenomenon could explain that the western half of the span (totally dislocated on the ground) is the most damaged while the eastern half has simply tilted while remaining in one piece.

Also, on the east side, a large section spacer, intended for a catenary support for the south track, connects the 2 steel beams at the level of the upper footings. This spacer may have opposed the spillage or torsional forces, helping to maintain the integrity of this part of the deck. The western part does not have a large cross-section spacer that can compensate for a break or an absence of connectors.

This phenomenon could also explain that the breakage occurred during traffic on the South track, since on the North side, the offset is reduced and compensated by the 3rd beam supporting the widening of the deck.

It is therefore possible that an implantation error of the western support or the steel beams resulted in 2 major faults for the stability of the span:

• A lack of connection between steel beams and concrete slab due to the eccentricity of the openings of the prefabricated slabs with the connectors. This defect can significantly reduce the bearing capacity of the beams and lead to their ruin by tilting or breakage.
• An increase in the eccentricity of the southern metro track and its rolling loads compared to the South steel beam, on the west side. This defect leading to increased bending forces in the slab, tensile forces in the connectors (if still in place), and an increased torsional force not foreseen by the calculations in the South steel beam, which could lead to the failure of the slab. or connectors, and steel beams spillage or breakage.

5.4) EXECUTION AND MAINTENANCE OF RAILWAYS

The design of the railways, the quality of their execution and their maintenance, can have consequences on the stability of the bridge:

• The settlement and displacement of the ballast can have consequences on the nature and position of the operating loads applied to the structure (vibrations, shocks, horizontal forces linked to the deformation of the rails, change of position of the loads in relation to the elements. structure).
• The thickness, nature and density of the ballast can have consequences on the permanent loads applied to the structure.
• The quality of stormwater drainage in and under the ballast can lead to a variation in the overloads linked to the presence of this water on the structure.
• There are also interactions between the behavior of the bridge structure and the state of the ballast: the deflection of the structure can cause displacements and deformations of the ballast and water drainage faults in the ballast.

In the case of non-concerted or poorly controlled interventions by the services responsible for the tracks and the services responsible for the civil engineering structures, there may be a deleterious chain of assessment errors leading to the overloading of the structure.

For example: Excessive deflection of the undiagnosed deck can cause ballast displacement or settlement. This displacement or this settlement can be compensated by adding ballast on the span without considering the structure of the bridge. This addition of ballast can then, in turn, increase deck deflection, then again displacement and addition of ballast etc, until the ballast overload causes the span to break.

It is reported that line 12 has been the subject of multiple defects residing in particular in vibrations when the trains pass. These defects would have been compensated by resuming the laying of the tracks and their ballast. It would therefore be useful to take an interest in these interventions and their coordination with the control of the structure.

5.5) POSSIBLE ROLE OF THE 2017 EARTHQUAKE

Defects in the structure can find their cause in a construction defect, in a lack of maintenance or be the delayed consequence of an external event such as an earthquake. All or several of these causes can also be combined.

Mexico City was hit by a strong 7.1 magnitude earthquake in 2017. Parts of the bridge were damaged, mainly the piers which were repaired. However, the collapsed span does not seem to have suffered any apparent damage during the earthquake.

However, an earthquake can weaken the structure without causing immediate breakage. The accumulation over time of stresses by dynamic loads (earthquake, train traffic) can lead to fatigue breakage in an accelerated manner.

Author: Jean-Luc VIANES - Structural Engineer - Saretec Expert

Photos: Getty Images - AFP - Google Maps - EFE 

Original article in french here.