Authors: Barbara Lane, director of fire engineering, Arup and Susan Lamont, associate director of fire engineering, Arup. Additional modelling team members: Graeme Flint, senior engineer, Arup and Allan Jowsey, fire engineering manager, Akzonobel. Thanks also to University of Edinburgh fire research centre Barbara Lane is one of the speakers at the upcoming CPD seminar on ‘New Developments and Challenges in Fire Safety’ (26 March). Presentations will address the new I.S. 3218 Fire Detection and Alarm Systems standard and the new Building Control (Amendment) Regulations. There will be a presentation on the latest developments in fire suppression systems and Lane will examine structural failure on 9/11. 

What about tall buildings that are treated as a higher risk for fire, as opposed to a building the size of Plantation Place? More specifically, what structural fire behaviours can be beneficial or, in fact, create intrinsic weak response to fire? Tall building fires are not limited to the events of the World Trade Center (WTC), which makes it even more important to learn from these events. We have therefore spent several years analysing structural response to fire, in order to develop new design techniques for structures in real single-floor or multiple-floor fires. This work was then expanded after 9/11 to try and understand the structural responses observed during the events of that day. As part of these studies, we have analysed WTC-type structural designs in various severe fire scenarios. In addition, we have been investigating the behaviour of long-span cellular beams in fire – the most popular form of construction in London at this time. The Windsor Tower fire, in a Madrid skyscraper fire in 2005, showed that concrete frames appear to be very robust in multiple floor fires, but is this true of all concrete frames? The Madrid fire highlighted the risk of fire spread floor to floor via breaches in compartmentation and via the facade. This resulted in the structure having to cope with multiple floor fires, even though this is not assumed as a basis for design in prescriptive regulations. The tall building studies presented here must not be viewed as a forensic investigation of the WTC buildings. Nor is that what we want to achieve – for we must be able to translate any new understanding to all different forms of construction. So, we are therefore carrying out a series of parametric studies to understand structural response to fire. Using real events to confirm or validate model assumptions is a critical way to determine confidence in the model. Our aim is to be in a position where we can understand if there are any specific progressive collapse mechanisms in tall structures that are not known or not understood in the fire-limit state. The goal is to develop better solutions for fire, without total reliance on passive fire protection, or on single element behaviour. That way we can take advantage of intrinsic design strengths, and attempt to design out any intrinsic design weakness, in the future. This is not an easy goal, and we consider the impetus from National Institute of Standards and Technology’s (NIST’s) work to be a major stepping-stone along the road to the profession achieving this. The specific aim of our structural fire research is to understand

• Whole frame response to multiple floor fires;
• Whether fire protection is effective;
• The collapse mechanism in WTC 1, 2 and 7 construction;
• If there is achievable strengthening measures that could limit such collapses in buildings in the future;
• If there are intrinsic weaknesses in specific construction forms or geometries;
• What NIST’s final recommendations are, what are they based on, how they impact design.

In particular, we want to understand if there are any specific progressive collapse mechanisms in tall structures not known or understood as a result of fire. The WTC towers behaved very well following impact and in response to multiple-floor fires, indicating that it was a robust system. The NIST report appeared to rely on dislodged fire protection. Our main concern with this conclusion is that thermal expansion can swamp all other behaviours. We believe it should be included in a thermo-mechanical analysis to predict the response of any structure to fire, particularly when determining a probable collapse mechanism. Protected structures – especially slender elements like truss diagonals – heat and deform in a fire. Fire protection is not a shield; it only delays heating. So, in a global structural system, it is our view that fire proofing to structural steelwork does not imply that collapse cannot occur. [caption id="attachment_12506" align="alignright" width="1220"] Mid-span deflection response of a beam analysed by our computer models. The deflected shape is very different if the thermal expansion is omitted from the analysis[/caption] This graph (right) shows the mid-span deflection response of a beam analysed by a non-linear finite element analysis. The deflected shape is very different if the thermal expansion is omitted from the analysis. [caption id="attachment_12510" align="alignright" width="592"] Modelling work carried out looking at the response of long-span truss floors in fire[/caption] The image (left) is an example of the modelling work we have carried out looking at the response of long-span truss floors in fire. We are analysing the results of our models and arriving at collapse mechanisms, which can be caused by thermal expansion. In ambient design, the column has a particular buckling mode based on an effective length between each floor. In a multiple-floor fire scenario, that buckling mode can be changed. The columns are initially pushed out as the floors expand in response to the fires. As the floors increase in length and buckle as a result of expansion, they provide less support to the columns. In addition, the floor stiffness decreases as a result of material degradation. There is then potential for the external columns to buckle over their increased length. [caption id="attachment_12513" align="alignright" width="1226"] An example of the data output from our structural-thermo analyses, showing the support available to the columns from truss floors at different steel truss temperatures[/caption] The graph (right) is an example of the data output from our thermomenchanical analyses and shows the support available to the columns from truss floors at different steel truss temperatures. It forms one part of the basis of our understanding of restraint to columns in fire. It demonstrates that even at very high temperatures, the truss floors can provide restraint. This could explain the time gap between the column inward bowing shown at approximately 18 minutes and the structural survival in that state to collapse several minutes later. However, this requires some detailed forensic examination and quantification before a formal statement could be made. The point of all this work is to one day provide advice to structural engineers about secondary systems that can be introduced to a tall-building design to support columns over multiple floors in a fire. This is a structural solution to a fire problem that we consider to be more robust than solely applying passive fire protection.

COLLAPSE MECHANISM [caption id="attachment_12522" align="alignright" width="774"] Geometry of collapse mechanism proposed by NIST[/caption] The basis of NIST’s collapse theory is also column behaviour in fire. However, we believe that a considerable difference in downward displacement between the core and perimeter columns, much greater than the 300mm proposed, is required for the collapse theory to hold true. Upward expansion of the column would act against the mechanical shortening. Crude initial calculations indicate that the elastic downward deflection at half the modulus (say at approximately 500°C) will be roughly 38mm. Assuming plastic strains, a maximum yielding of approximately 190mm is possible. If the downward displacement is 300mm, as assumed, the rotation at the perimeter connection would be 300mm vertical over an 18000mm span – extremely small. The floor elongation must be less than 2.5mm to generate tensile pulling forces on the exterior columns as a result of the column shortening in the core. Thermal expansion of the floor truss would be 65mm at 300°C over a length of 18000mm. Therefore, the 2.5mm is swamped by thermal expansion and the core columns cannot pull the exterior columns in via the floor simply as a result of column shortening. The NIST collapse theory also states: “Floors weakened and sagged from the fires, pulling inward on the perimeter columns. Floor sagging and exposure to high temperatures caused the perimeter columns to bow inward and buckle—a process that spread across the faces of the buildings. Collapse then ensued.” This is similar to some of our collapse proposals, but no mention of thermal expansion is made; the floor buckling and lack of support to the columns seems to be entirely due to loss in strength and stiffness in their view, which we would consider to be only part of the story. With regard to the influence of the hat truss on the buildling’s performance, we have analysed models with and without a hat truss at the top of a tall building. We found that a hat truss significantly improves stability in multiple floor fires. In the image above, the hat truss shows clear redistribution from outer columns to the core (primarily the outer core columns). NIST have also observed load transfer via the hat truss. Such issues could become the basis for future fire-related structural design guidance.

CONTINUING RESEARCH AND DEVELOPMENT [caption id="attachment_12528" align="alignright" width="1080"] We have analysed models with and without the hat truss at the top of the building and found that it significantly improves stability. The hat truss shows clear redistribution from outer columns to the core (primarily the outer core columns)[/caption] Arup will continue this study, as it is important to our understanding of how buildings work. We will also continue to follow NIST’s work on the collapse of the WTC because of our interest in tall-building response to fire and tall-building design. We are committed to the belief that building design needs to address thermal expansion effects and the treatment of fire as a load on the structure. Our aim now and the reason for our continuing research and development in this field is to introduce quantified secondary structural systems to help the structure cope with the loads induced in fire. On a final note, structural fire engineering continues to be important. We have noticed an interesting step change in the approvals process with specific requests now, even for fully code compliant buildings in terms of structural fire proofing, for global structural responses be quantified and justified, in order to obtain structural design approval. Events will happen and we have to address concerns but we need to make sure that our response is measured and beneficial in many ways. Our goal is to deliver the design vision for our architects and clients and all the key stakeholders in a project, safely. Therefore, we are recommending threat and risk assessments to determine design solutions, innovative evacuation strategies that address real human response and imminent catastrophic events, and whole frame structural analysis to be employed, as required, on tall buildings in the future.