Madrid Windsor fire: the Arup view

The remains of the Windsor tower.
View from the south-west side after the blaze

On the evening of Saturday February 12 2005 a fire broke out in the Windsor tower in Madrid, Spain. The building is located in the heart of Madrid's commercial and banking centre and is one of the tallest in the capital.

The Madrid fire has increased interest in the real response of structures to fire, coming so soon after major building fires in New York, Chicago and London, amongst others.

The following is an Arup view based upon what is known about the fire event in conjunction with our structural fire design and analysis experience.

It has been prepared based upon information in the public domain only and will be updated as further information becomes available.

Structure and fire safety measures

The building had a concrete central core with two rows of reinforced concrete columns in the north-south direction, aligned with the core side walls. The structure above ground was characterised by two transition floors at 3rd and 17th Floor levels, which housed plant and services.

The typical floor slab construction was reinforced concrete bi-directional ribbed slabs, spanning onto composite steel beams in the east-west direction. The slabs were supported along the perimeter by steel columns, supplemented by RC columns on two sides below 17th Floor level.

The transition floors were formed with solid RC slabs and deep beams. The original facade mullions and transoms were fixed to the steel perimeter columns, and a new facade structure had been added to outside of old facade. The perimeter columns in turn were supported by transition structures at 17th and 3rd Floor levels.

Spanish codes, in common with most continental European codes, place more emphasis on passive control measures than on active measures. The Madrid regional code does not require sprinkler protection for buildings with an evacuation height of less than 100m so active measures were limited to automatic detection and alarm, fire hose reels and a dry riser system.

The building was being refurbished at the time of the fire so we do not know the state of the passive measures in place but from information available it would appear that Edificio Windsor had the following passive protection:

Fire brigade call out

The control room in the basement of the building registered a fire signal from the 21st Floor at 23:05 on the night of Saturday 12 February 2005.

A time of 16 minutes elapsed between this signal and the call being made to the Fire Brigade - who arrived on site at 23:25. Security personnel claim that a time of 10 – 15 minutes elapsed from Fire Brigade arrival to first fire attack i.e. 30 – 35 minutes after the initial alarm was registered in the control room.

Development of the fire

It appears that the fire fighters used the stairs to reach the 21st Floor. Nothing has been said of their efforts to connect to the dry riser system or its effectiveness in relaying water to the 21st Floor (about 75m above ground). By the time the firemen intervened, fire fighters were not able to control the fire from the inside. By 01:00 they had retreated and adopted a defensive mode; wetting down the facades of adjacent buildings to avoid fire spread.

By about 01:15 the fire had spread to most of the floors above the 21st Floor, resulting in a 10-storey blaze. Soon afterwards the first chunks of facade started falling off, taking the perimeter bay of the RC slab with it in places. The spread of fire downwards was gradual at first, probably due to burning embers dropping through services penetrations, through slab edge openings and through other openings in the concrete slabs caused by core wall expansion.

The fire led to the collapse of virtually all the slab edge bay above 17th Floor as well as one internal bay on the north side. The transition floor resisted the impact of the partial collapses. Below this level there was substantial structural damage and deformation, but no significant collapse.

Extensive damage

The long delay between detection and fire brigade intervention played an important role in allowing the unsprinklered fire to grow out of control. In addition the rapid spread of the fire above the 21st Floor appears to be due to failure of the compartmentation measures between the facade detail and the floor which is intended to prevent vertical fire spread.

Fire safety design in many countries relies heavily on sprinkler protection to prevent fire growth and thereby limit possibilities for fire spread via the facade. The lack of sprinklers, along with the failure of compartmentation, appears to be an important factor in this case.

Although there is a requirement to fire stop the gap between the slab edge and the inside of the curtain wall, most codes do not address the tie-back connection of the curtain wall to the structure. Therefore a light facade structural element can heat up quickly and the resulting expansion can produce an outward bulging away from the slab edge, which can create internal flues if it happens before the facade glazing breaks. In other words by not considering the thermo-mechanical response of the system, there are no provisions to prevent such damage in Building Codes worldwide.

An added complication in the case of Edificio Windsor was that the curtain wall facade had recently been replaced and it appears that a new support structure had been fixed onto the outside of the original mullion and transom arrangement. This means that there would have been a double-layered gap that needed to be fire stopped, complicating this detail still further.

Structural response

View from the south-west side after the blaze as shown above.

The steel perimeter columns, even if they had been protected, or even concrete columns, would not necessarily be expected to survive the effects of such a 10-storey blaze.

A multiple floor fire can result in lack of restraint of columns over a number of floors due to heating which could trigger a collapse mechanism. However structural fire analysis of this structural form is required before such a conclusion can be drawn.

What is clear however in the case of the Windsor building, is that the concrete perimeter slab could evidently not remain in the place without this support.

The reason the collapse of one internal bay on the North side supported by concrete columns is not immediately apparent but could be attributable to any of a number of factors: less concrete cover, more intense heat exposure, and the increased likelihood of concrete spalling this would create and/or momentum of adjacent bay collapse, amongst others.

The presence of a robust transition floor halfway up the tower is not a standard fire protection measure but undoubtedly played a role in limiting the extent of the collapse.

Using a fire engineered approach for the design of the structure, features such as the transition floor could be taken advantage of in the design. Similarly, weaknesses such as the lack of a tie between the curtain wall and the structure would be apparent and could be addressed at the design stage.

The reduced damage below the 17th Floor can be attributed to a function of the more robust structure below this level, the existence of fire protection measures and a less intense fire due to the more gradual fire spread.

Lessons to be learned

  • Procedures to ensure early call out to the Fire Brigade
  • Provisions for speedy access to the fire floor via protected fire fighting lifts and use of wet risers
  • Effective compartmentation measures, including sprinkler protection for high-rise buildings
  • Good coordination of fire safety measures with refurbishment works and programme, especially in an occupied building
  • Structural fire full frame analysis, rather than single element small-scale fire tests, as a basis for design

The central concrete core appeared to perform well in the fire and on initial observations seems to have played a major role in ensuring the stability of the building throughout the incident. The role of cores in multiple floor fires is now an immediate area of study required for the industry, and Arup have commenced investigating this issue.

A thermo-mechanical assessment of this structural design, an understanding of why the structure performed as it did and why total collapse did not occur would provide valuable information for future structural fire analysis in design.

It would assist in the strong move now towards structural fire engineered buildings, and therefore help with the move away from reliance on Building Code based single element testing and associated fire proofing techniques which do not address real and structural behaviour in real fires.

See also

Other Arup Links