Abstract
Vandalism, human errors or severe accidents could result in significant fire events on and beneath bridges. The resulting damages often require a reconstruction of the bridge superstructure or, in case of a complete collapse, of the whole bridge. In a research project well-founded and systematic findings regarding the consequences of extreme fire events beneath and on top of bridges were elaborated. The examined scenarios include solid and liquid fires which were investigated with Computational Fluid Dynamics (CFD) calculations. In order to validate the CFD models used, an original-scale bridge fire test was conducted, using a fire scenario with a truck loaded with wooden pallets. In order to answer the question which bridge structures are the most vulnerable to extreme fire events and have to be protected in the future, especially civil engineering aspects should be taken into account with due consideration given to bridge geometry, clearance, material and structure.
Introduction
Bridges and tunnels are key elements of the German federal road network. Due to their connecting function in the road network they create the condition for individual mobility and supply of private households and businesses. Furthermore, they represent a significant macro-economic value. Sustainable protection of the existing structures comprising of currently 240 tunnels and more than 39,000 bridges in Germany is therefore a high priority with regard to the availability of the infrastructure and the welfare of society. The objectives of the completed joint research projects SKRIBT and SKRIBTplus (Protection of Critical Bridges and Tunnels) [1] were to look at possible hazards in terms of current and future threat situations for bridge and tunnel structures on roads and to develop effective protective measures, thus reducing the vulnerability of important infrastructure and their users.
In this context a comprehensive research project on the effects of extreme fire scenarios on the bearing capacity and durability of road bridges [2] was conducted by the Leipzig Institute for Materials Research and Testing (MFPA) on behalf of the Federal Highway Research Institute (BASt). The reason for launching this research project was the fact, that vandalism, human errors or accidents can result in significant fire events on top and beneath bridges. Examples for such fire events in Germany include the Wiehltalbücke (26/08/2004) (Fig. 1) and a bridge on freeway A57 near Dormagen (14/02/2012). Devastating fire events on top and beneath bridges are also documented in countries outside of Europe, e.g. Oakland Freeway Bridge, USA, (29/07/2007), Freeway 70 near Montebello, USA, (15/12/2011). Bridges and bridge parts collapsed partly due to these fire events and it became necessary to rebuild parts of the bridge superstructure and in one case the whole bridge (cf. [3–6]). The high heat development with partly also steep temperature gradients leads on the one hand to a reduced bearing capacity of structural bridge components and on the other hand to material specific structural changes like concrete spalling at concrete bridges or structure changes for steel and composite bridges. Possible protection measures which could help to mitigate the effects of fire scenarios like thermal protection of main bearing elements (e.g. cables) are currently being developed for important, long-span bridges [7].
Accident on Wieltalbrücke (Freeway A4) with damaged steel superstructure due to a fallen down gasoline truck burning beneath the bridge (picture: newspaper “Oberberg-Aktuell”).
The here presented results of the research project consist of different numerical studies taking different fire scenarios and bridge types into account and an original scale fire test which was used to validate and fine tune the numerical models.
The initiated research project deals with consequences of fires on top of bridges and beneath bridges. The investigated scenarios include a burning truck loaded with wooden pallets as well as burning gasoline leaking from a tanker at different release rates. The investigated release rates of the gasoline are 20.6 kg/s and 300 kg/s. Three different scenarios are investigated for the truck loaded with wooden pallets. All scenarios include a truck (40 t) completely loaded with wooden (Euro-) pallets (i.e. 750 pallets with 22 kg each, total fire load 285 GJ). The three scenarios differ in the applied characteristic of the Heat Release Rate (HRR). The gasoline parameters, the burn rates as well as the temporal course of the energy release rate and mass reduction for wood and gasoline fire scenarios were chosen based on [8].
The fire progress and/or the temperature development at the bridge structure are determined by Computational Fluid Dynamics (CFD) with due consideration given to geometrical factors of influence and to wind effects. For the CFD calculations Fire Dynamics Simulator (FDS) was used [9]. The FDS model is a suitable tool for the investigation of fires which has been proofed in many past research projects (e.g. [10, 11]). The use of FDS is also recommended by a guideline of the German Fire Protection Association [12]. The results of the FDS model had been verified by comparing calculations with an independent programming system (openFOAM, http://www.openfoam.com/) within the research project [11]. Additionally sensitivity studies had been performed in order to guarantee that the spatial and temporal discretization was adequately chosen.
As a result of the CFD simulation, all relevant adiabatic surface temperature/time curves are determined which are used as thermal loading for the calculation of the temperature increase behavior by means of transient FEM calculations. In order to increase the significance of the results and to be able to better generalize the latter, different main structural bridge elements are investigated separately from each other.
The temperature load for fire scenarios on the bridge is determined for a suspension bridge with a standard cross section of RQ 15.5 (15.5 m width) and for an arch bridge (RQ 10.5). The evaluation is focused on the determination of the temperature load on all supporting structural bridge elements above the roadway (cables, hangers, pylons, arch, gusset plates, etc.).
For the evaluation of fire events beneath bridges, the following bridges are taken into consideration: Bridges with a flat undersurface (e.g. bridges with slab cross-section, composite bridges with internal steel sections, prestressed concrete box girder bridges, bridges with orthotropic roadway slab or composite reinforced concrete roadway slab with low steel girder heights) and bridges with T-beam-like bridge cross-section and/or T-beam-like roadway (double-girder prestressed concrete bridges and/or reinforced concrete bridges, steel/concrete composite bridges with open or closed steel section beneath the roadway). The geometry of the investigated bridges is exemplarily shown in Fig. 2. A total of 38 different scenarios for fires on bridges and 39 different scenarios for fires beneath bridges are investigated.
Geometry of the investigated bridges with slab and T-beam cross-section, fire beneath bridges.
Test setup and realization
Up to date no specific experimental studies at an original-scale for fires beneath bridges are known. Hence, in order to validate the numerical models used, MFPA Leipzig GmbH erected an original-scale bridge fire test stand in order to carry out a truck test fire (Fig. 3). The test stand consists of a 50 cm thick and 10 meters long abutment wall and two reinforced concrete circular columns, d = 75 cm, and a composite superstructure with a “bridge deck” area of 100 m2. The composite superstructure consists of two steel girders (IPE 360) connected with a 20 cm thick poured concrete slab by 4 shear studs 19×125 mm2 per meter. The concrete slab is reinforced with 12 mm thick steel bars every 15 cm in both directions and in two layers (2 cm concrete cover). The distance between the steel girders is 6 m. The clearance height is 4.50 m.
Test setup for the original-scale fire test (left), truck during full fire phase 13 minutes after ignition (right) (pictures: MFPA Leipzig).
A fire of a truck loaded with wooden pallets, with an admissible total weight of 7.5 t was simulated. For the different pallet fire scenarios to be investigated in the research project it is assumed that the truck (e.g. due to a technical defect) caught fire during its movement and the vehicle with the cargo on fire finally breaks down just beneath the bridge. Hence, the first part of the fire development phase did not occur beneath the bridge.
In order to simulate this scenario, a rapid fire development was required in the test with a full fire phase occurring as quickly as possible. For this purpose Isopropanol was used, the ignition of which resulted in a rapid spread of the fire. Thus, the fire load consisted of 3000 kg wood pallets (European coniferous wood) and 230 kg Isopropanol as well as the fire load of the truck. The truck has a total weight of 4268 kg without additional load. After the fire a residual mass of the truck of 3582 kg was determined by weighing. Thus, 686 kg of different materials in and/or at the truck were burned. The measuring program consisted of in total 382 points for gas and structure temperature measurements.
Figure 4 shows the maximum measured gas temperatures 5 cm beneath the concrete slab as a contour plot (top view). The displayed area is in the center between the steel girders, whose tensioning direction is referred to as “Länge”, and in the center between abutment and concrete columns. The measured maximum gas temperature below the concrete slab was 960° Celsius.
Measured maximum gas temperatures 5 cm beneath the concrete slab (bridge superstructure).
Figure 5 shows the curve of the adiabatic component surface temperature in the middle of the right steel girder (cf. Fig. 3) of the composite superstructure which was determined by comparing the calculation data from the FEM analysis and measuring data from the fire test. Due to the good consistency with the measured gas temperatures shown in Fig. 4, the determined curve (Fig. 5) could be applied for the whole red area in Fig. 4.
Evaluated curve of the adiabatic component surface temperatures at the steel girder.
Also when comparing the temperature measurement points in different depths inside the concrete slab (area above the steel girder without concrete spalling) with the numerical results a good consistency could be stated (Figs. 6 and 7). The determined curve of the adiabatic surface temperature could be proofed with good accuracy (cf. Fig. 5). Additionally it could be stated that the used FE-models are a good picture of reality.
Numerically determined temperatures in the concrete slab depending on the time (“Zeit”) and the distance to the fire exposed surface (“Betontiefe”) based on the adiabatic surface temperature curve according to Figure 5.
During the fire test measured temperatures in different depths of the concrete slab depending on the time (“Zeit”) and the distance to the fire exposed surface in the area above the steel girder (“Messstellen” corresponding to “Betontiefe” in Fig. 6).
Another important result of the fire test was the occurrence of concrete spallings in relatively large areas of the superstructure (concrete slab) and parts of the abutment. A tachometric survey of the actual spalling depth was not possible due to the large (lasting) deformation of the slab after the fire test. The depth of spalling has thus to be estimated via engineering assumptions by means of the measuring values obtained from the various temperature sensors during the fire test. The resulting spatial distribution of the spalling depth obtained from this evaluation is shown in Fig. 8 (left). The maximum spalling depth is 3.5 cm. This contour plot corresponds relatively well with the contour of the visual observation during the fire test (Fig. 8 right) and the location of exposed reinforcement (concrete cover completely spalled).
Evaluated spalling depths at the bottom side of the superstructure (left). Area with spalled concrete and exposed reinforcement near the end of the fire test (45 min., right) (picture: MFPA Leipzig).
Based on the scenarios described in chapter 2 the temperature loads and the specific temperature/time curves for the different bridge components are determined. Due to the huge amount of investigated scenarios (see chapter 2) only some selected results could be presented in this Paper.
During fires on bridges, the temperature acting on the supporting structure is somewhat lower in general than during fires beneath bridges. On the one hand, the location with the absolutely maximum temperature (identification maxT) and on the other hand the location with the absolutely maximum average temperature (identification maxE) is used to determine the temperature/time curves. As an example Fig. 9 shows the determined relationships for the fire scenario with the maximum total energy release of all gasoline fire scenarios on the bridge. Hence the temperature/time curve for to the inside oriented surfaces of vertical bridge components (e.g. phylons or hangers) is the most severe for liquid fires on bridges.
Fire on the bridge: By CFD calculation determined adiabatic temperature/time curves for locations with the highest temperatures (maxT) and the highest average temperatures (maxE) depending on the orientation of the bridge component surface (red: vertical inside; black: horizontal bottom; green: horizontal top; blue: vertical outside) in 5 m height over the roadway for the fire scenario: “constant slope, 20.6 kg/s release, one gutter and no wind”.
Figure 10 shows the trend of the adiabatic temperature/time curves in the event of a fire beneath a bridge with slab cross-section for the location with the maximum temperature (identification maxT) and/or the maximum average temperature (identification maxE) for the example of the liquid fire scenario with a release rate of 20.6 kg/s. Based on the comparison of different fire scenarios and/or the temperature load resulting from the fire scenarios, one can determine that very high temperatures are reached also over a longer period of time. The temperatures correspond to those which were measured and/or calculated for comparable fire scenarios in tunnels.
Fire beneath the bridge: By CFD calculation determined adiabatic temperature/time curves in the centre of the slab.
The measured maximum values of the adiabatic component surface temperatures can be compared with the values of the German national temperature/time curve (so called “ZTV-ING temperature/time curve”) for tunnels (Fig. 11) [13]. Also the duration of the full fire phase for liquid fires with release rates of 20.6 kg/s corresponds to the ZTV-ING temperature/time curve. Thus, the gasoline fire scenarios beneath the bridge differ clearly from those on the bridge (cf. Fig. 9).
ZTV-ING temperature/time curve [13].
The determined temperature/time curves (cf. chapter 4) are evaluated regarding their effect on temperature load and the material behavior, especially regarding concrete spalling. The evaluated temperature load is used to calculate the temperature depending behavior of different materials and cross section geometries. These results are used later to evaluate the bearing capacity of different bridge components. Finally general design aids are developed which could be directly applied to many different bridge types.
Evaluation of the concrete spalling risk
Concrete spalling occurs typically during the initial 10 – 25 minutes of a fire and/or a fire test. It is promoted by high temperatures and rapid temperature increase. Concrete spalling can be expected for usual concrete recipes without special design for high temperature resistance (i.e. concrete without PP-fibres). However, they can only be forecast with difficulty because this forecast depends decisively on the actual temperature/time curve, the high-temperature behavior of the materials used (e.g. aggregates in the concrete) and thus on the individual case. Additionally, on the basis of experiences gained from the original-scale fire test, spalling in the range from 5 to 10 cm has to be expected which can definitely be exceeded locally. Spalling of concrete bridge components which are specifically worthy to be protected can be prevented in order to limit heating of the reinforcement and thus to ensure the load-bearing capacity of the bridge during and after the fire. Concrete spalling can be limited and/or prevented only by appropriate concrete recipes (e.g. by adding PP-fibers).
Determination of the time depending temperature increase behavior of bridge components
The determined temperature/time curves and/or the evaluated temperature/time curves are used as initial variables for calculating the time-depending temperature increase behavior by means of Finite Element Models (FEM). The temperature-depending material parameters for concrete have been considered according to DIN EN 1992-1-2 (2010) [14] and for structural steel (carbon steel) according to DIN EN 1993-1-2 (2010) [15]. An appropriate concrete recipe has been assumed for all calculations for reinforced and prestressed concrete cross-sections. Thus, spalling is not taken into account when calculating the temperature field.
Design aids for reinforced concrete and prestressed concrete cross sections
In order to improve the clarity of the research results in terms of fires beneath bridges and thus to ensure the transferability of the results also to other bridge structures, the temperature/time curves determined for the different boundary conditions by CFD calculation are simplified and summarized, evaluated and set into relation to the ZTV-ING (Fig. 11) and/or extended national temperature/time curve (i.e. 55 minutes of full fire phase). Figure 12 shows exemplarily the evaluated, summarized and simplified temperature/time curves for the three wooden pallet fire scenarios beneath concrete bridges (slab or T-beam cross-section) in relation to the ZTV-ING and extended ZTV-ING temperature/time curve. Evaluated temperature/time curves for gasoline fire scenarios are generally less decisive regarding the temperature increase in the concrete structures. Only the evaluated temperature/time curve LE1 (Fig. 12) could also be applied for the gasoline fire scenario with release rates of 20.6 kg/s (cf. Fig. 10).
Evaluated temperature/time curve for wooden pallet fire scenarios (LE1 until LE3) and ZTV-ING and extended ZTV-ING temperature/time curve.
Design aids were developed both for reinforced concrete and/or prestressed concrete cross-sections and for typical steel cross-sections such as hanger with diameter 80 mm, 90 mm and 100 mm or flat steel and/or steel plates with different thicknesses with one-side and two-side temperature load.
Figure 13 shows exemplarily the reduction of the concrete steel strength depending on the web width b and the center distance of the reinforcement to the fire exposed surface a due to the ZTV-ING temperature/time curve (Fig. 11).
Mean reduction of the concrete steel strength depending on the web width b and the centre distance of the reinforcement to the fire exposed surface a due to the ZTV-ING temperature/time curve.
In the research project findings about the consequences of extreme fire events beneath and on top of bridges, above all in terms of the load-bearing capacity and durability were obtained. However, it is important to note that this was done for very low frequency scenarios with effects far above the design scenario for bridges in Germany and Europe.
As a result of the CFD calculations, which can also be transferred to other bridge types, the effect of fires beneath bridges has to be differentiated from the effect of fires on bridges. Due to the hindered convection by the bridge superstructure and the reflecting effects of abutment and superstructure, temperatures can occur during fires beneath bridges which are known from tunnel fires. This depends of course on the geometrical boundary conditions of the bridge (mainly bridge clearance height and bridge width). The temperature increase in the phase of fire development is also very steep. Thus, the fire curve can be approximated well by the German national temperature/time curve (ZTV-ING) for road tunnels. Additionally during fires beneath concrete or prestressed concrete bridges, one has to expect enormous concrete spalling in the range of several centimeters unless appropriate and/or suitable concrete recipes are used. In the event of fire one has to fear a total loss of the bridge at least by the scenarios investigated in this research project.
During fires on bridges, the temperature acting on the supporting structure is a bit lower in general than during fires beneath bridges. It is suggested to use the time temperature curve as design fire scenario related to and/or favored by the German ZTV-ING temperature/time curve. While the qualitative curve remains unaffected, the maximum temperatures for the design temperature/time curve to be taken into consideration can be reduced accordingly.
Using the described CFD simulations, which have been validated regarding the temperature load by the original-scale large fire test, the determination of the adiabatic component surface temperatures is possible for different supporting bridge components and/or for differently orientated bridge surfaces. The differences among the fire scenarios are investigated and the factors of influence are elaborated such as the wind velocity during fires on bridges and the clearance height during fire beneath bridges. The determined time temperature curves were evaluated both in terms of the load and the material behavior to be expected considering the concrete spalling behavior. The temperature increase behavior is calculated for different materials and cross-section shapes.
These include, in particular: Slabs and T-beam cross sections of reinforced concrete with different concrete covers, Hangers of steel with diameter 80 mm, 90 mm and 100 mm and Flat steel and/or steel plates with one-side and two-side temperature loading with plate thicknesses of 10 mm, 20 mm, 30 mm and 40 mm.
For round bars, a differentiation has to be made according to the size of the temperature-loaded surface. Besides the case that round bars are exposed to the temperature over the whole circumference, also cases are studied where only 1/4, 1/2 and 3/4 of the surface is exposed to the (full) temperature load. The temperature load can be reduced by increasing the distance between the fire and the construction.
On the basis of these calculations, generalized design diagrams for extreme fire scenarios are developed which can also be applied directly to other bridges. In this way, both the findings of the research project in terms of temperature effects and regarding design can be applied directly to real bridge structures.
As already mentioned, the investigation concerns very unlikely scenarios that go far beyond the design scenarios. Therefore, no direct need for action can be derived from the results. However, the results show very well possible impacts of extreme fire scenarios and give a first indication which bridge designs and under which conditions could be affected. Where necessary the research results can be taken into account for the design of new bridges.