Engineering Structures 33 (2011) 28882898

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Structural analysis of reinforced concrete chimneys subjected to uncontrolled fire

Ashkan Vaziri a, , Amin Ajdari a , Hosam Ali b , Artemis Agelaridou Twohig b
a b

Department of Mechanical and Industrial Engineering, Northeastern University, Boston, MA 02115, United States FM Global, 1151 Boston-Providence Turnpike, Norwood, MA 02062, United States

article

info

abstract
We studied the behavior and residual structural capacity of reinforced concrete chimneys subjected to an uncontrolled fire. We used a combination of a heat transfer finite element model to obtain the temporal distributions of temperature during the fire event and the structural model of concrete chimney design provided by the American Concrete Institute (ACI 30708). This approach allows estimating the reduction in the vertical (axial) strength and moment strength of the chimney both during a fire and post-fire, and gives a direct estimate of the reduction in the safety factors of the concrete chimney. Using this method, we examined the impact of various design parameters on the residual structural capacity of a concrete chimney subjected to an internal fire. An iterative finite element method was also presented as an alternative to the ACI 307 calculations. Moreover, finite element calculations were used to study the role of thermal stresses on the axial strength of the chimney during fire. Our study provides insight into possible failure mechanisms of concrete chimneys damaged due to fire and could suggest possible approaches for minimizing the risk of chimney failure due to an uncontrolled fire. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 2 December 2010 Received in revised form 12 May 2011 Accepted 9 June 2011 Available online 20 July 2011 Keywords: Concrete chimney Uncontrolled fire Heat transfer Finite element Structural model

1. Introduction Concrete chimneys are used in power plants for venting hot flue gases or smoke to the outside atmosphere. In recent years, the height of power plant concrete chimneys has increased to enhance the draw of air for combustion and to disperse pollutants over a greater area to reduce pollutant concentrations. While the number of reported chimney collapses due to an internal fire is very small, the consequences of chimney damage could be costly in terms of economic and human loss. The popularity of fiberglass reinforced plastic (FRP) liners which are combustible materials has made the risk of a fire in tall chimneys even more relevant in recent years. The FRP liners are used in reinforced concrete chimneys to protect the chimney shell from the effect of hot flue gases. Some of the possible sources of ignition in chimneys with FRP liners can be hot work inside the chimney during FRP installation or maintenance, ignition of stored flammable materials at the base of the chimney, ignition inside the flue gas desulfurization system or other equipment upstream of the chimney. A critical effect of an uncontrolled fire in a chimney is the reduction in concrete strength [16], which leads to a decrease in the chimney load carrying capacity and service life. At an elevated temperature, the concrete experiences a variety of chemical and physical changes. For example, large volume changes resulting

Corresponding author. E-mail address: vaziri@coe.neu.edu (A. Vaziri).

from non-uniform thermal expansion of aggregates and shrinkage of the cement paste, results in cracking and spalling. Spalling which is usually explosive and critical for structural integrity, is induced by mechanical and thermal stresses and pore pressure [7]. Spalling generally occurs at high temperatures, even though it is also observed at temperatures as low as 200 C [8,9]. For spalling to occur, there needs to be a minimum moisture content as well as a temperature gradient of approximately 58 K/mm [6]. Temperature gradients induced by heating or uncontrolled fire depend not only on the heating source temperature but also on the heating rate. In Fig. 1, we have re-plotted the Eurocode 2 [10] data (curve 1) for a concrete with siliceous aggregates which is commonly used in tall chimneys. These data suggest that the concrete compressive strength at 600 C reduces 50% from its strength at room temperature. This reduction in strength is qualitatively similar to the behavior of structural steel, where high temperature could lead to significant weakening and in many cases catastrophic failure of the structure [1113]. However, steel strength almost fully recovers as the steel cools down after the fire, so there is little concern about the behavior of steel structures post-fire. Concrete strength further decreases as it cools down from a high temperature, as the incompatibility between the thermal deformations of the concrete constituents leads to propagations of existing microcracks and formation of new ones [4]. The amount of reduction in the strength of the concrete is a complex function of the cooling method and rate. Sakr and Hakim [4] measured the role of cooling methods on the residual compressive strength

0141-0296/$ see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2011.06.013

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

2889

Fig. 1. Compressive strength of concrete at elevated temperatures (from [6]) and after cooling (from [4]). The cooling of the concrete specimens leads to further reduction in the compressive strength of the concrete. Here, the results are shown for both air cooling and water cooling for samples that are kept at elevated temperatures for 2 and 3 h prior to cooling. Samples are then tested at room temperature.

of concrete kept at an elevated temperature for different heating durations. They considered three different cooling methods: air cooling, water cooling and foam cooling. In Fig. 1, we have shown a selected set of results from their study, which show that cooling by water leads to lowest residual compressive strength of the concrete tested at room temperature. In this study, we used the values given by Eurocode 2 [10] to estimate the concrete strength at elevated temperature, which is applicable for heating rates between 2 and 50 K/min. The objective of this study is to evaluate the structural response of reinforced concrete chimneys subjected to uncontrolled fire. Fig. 2 shows the schematic of the scenarios considered here. During the fire, the reinforced concrete shell should be able to resist thermal stresses, its own weight, part of the liners weight, and wind loads. Even if the chimney does not collapse during the fire the concrete will be damaged by the heat. After the fire, the damaged shell should be able to support its weight, the wind load, and possibly the inertial loads of an earthquake. Here, we studied the behavior and residual structural capacity of reinforced concrete chimneys, with and without an opening, subjected to an uncontrolled fire. We combined a heat transfer finite element model to obtain the temporal distributions of temperature during the fire event, with a structural model of concrete chimney design provided by the American Concrete Institute (ACI 30708) [14]. This approach allows estimating the reduction in the vertical (axial) strength and moment strength of the chimney during a fire and post-fire, and gives a direct estimate of the reduction in the safety factors of the concrete chimney. In our study, we used curve 1 in Fig. 1 for estimating the concrete strength at elevated temperatures (e.g. during fire). For estimating the residual concrete strength post-fire, we used the data associated with water cooling since it leads to the greatest reduction in concrete strength (curves 3 and 5 in Fig. 1). The details of the proposed approach are discussed in Section 2. An iterative finite element method is also presented as an alternative to the ACI 307 calculations in Section 3. In Section 4, we examine the impact of various design parameters on the residual structural capacity of concrete chimneys using, as an example, a fire that lasts at most 3 h and causes the average temperature inside the chimney to be about 870 C. In Section 5, we discuss the role of thermal stresses on the behavior of reinforced concrete chimney during a fire. The key conclusions are drawn in Section 6.

Fig. 2. Scenarios considered in structural analysis.

2. Theoretical approach To investigate the effect of fire on the residual structural capacity of a reinforced concrete chimney, we combined finite element heat transfer calculations with the ACI structural approach [14]. The details of the heat transfer model developed for calculating the temperature profile in the chimney shell during the fire event are discussed in Section 2.1. The elevation of the temperature in the chimney shell results in a reduction of the compressive strength of the concrete and thus inherently, a reduction in the axial and moment strengths of the chimney during a fire event and also post-fire. The estimated temperature profile from the heat transfer model was converted to an average reduction in the compressive strength of concrete and axial strength of the chimney, using an approximate method (Section 2.2). Subsequently, the reduction in the vertical strength and moment strength of the chimney was obtained using the structural model provided by the ACI Code as discussed in Section 2.3. In this study, we assumed that the concrete strength is fc = 27.5 MPa (4000 psi), the strength of the reinforcement steel bars is, fy = 220 MPa (32 000 psi), and the density of the concrete is = 2400 kg/m3 . 2.1. Heat transfer model The temperature profile in the chimney shell and its spatial and temporal variations during the fire event is a complex function of fire parameters (e.g. temperature of gases inside the chimney and duration of the fire) and their dynamics, as well as the properties of the chimney shell. For concrete chimney shells with FRP liners, many fire initiation and propagation scenarios are possible. Examples include initiation of the fire at a certain height of the liners followed by vertical propagation both upwards and downwards at different speeds, or the case that the fire results in detachment of the liners from the chimney and their accumulation at ground level inside the chimney. The latter case can result

2890

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

Fig. 3. Heat transfer model of a concrete chimney. (A) Schematic of a one-dimensional heat transfer model used for calculating the temporal variation of the temperature profile in a concrete chimney. (B) Temperature profiles in a concrete chimney with ri = 12.5 m (41 ft), t = 0.46 m (18 in.) during the fire event and the steady state condition.

in long-lasting fire leading to locally higher temperatures in the chimney shell. Different scenarios can be assumed depending on the incoming airflow and openings in the chimney shell. If the incoming airflow (from the openings) is adequate during the fire, it is expected that the incoming air pushes the flames towards one side of the chimney causing localized heating of the chimney shell. If the pile of burning FRP rubble causes blockage of the airflow, it is conceivable that less air will flow through the chimney and the concrete shell will be exposed to higher average gas temperatures. That is the scenario considered here, by conservatively assuming that even for restricted airflow conditions, there is enough oxygen for well-ventilated burning, and also having the total inner surface of the chimney shell subjected to the fire (rather than local heating, which leads to a less pronounced reduction in the chimney residual strength). Considering that the maximum bending loads due to the wind also occur at the base of the concrete, the case described above appears to lead to maximum reduction in the residual structural capacity of the chimney shell post-fire. Here, we developed a simplified finite element model to calculate the temperature profile through the thickness of the concrete chimney shell during fire using ABAQUS/CAE 6.9 (SIMULIA, Providence, RI). For the scenario mentioned above, the structure and the fire loading is axisymmetric and thus, the temperature profiles can be estimated using a one-dimensional (through the thickness) axisymmetric finite element model of the chimney shell, Fig. 3A. The model was meshed using 8node quadratic axisymmetric heat transfer elements, and mesh sensitivity analysis was carried out to ensure that the results are not affected by the mesh size. For the concrete, we assumed the coefficient of thermal conductivity, Cc = 1.72 W/m K, specific heat, Cp = 900 J/kg K [14]. In our calculation, the coefficients of heat transmission from the outside surface of the chimney shell to surrounding air, Ko = 68 W/m2 K, and the coefficients of heat transmission from the gas to the inner surface of the chimney shell, Ki = 164.5 W/m2 K, were estimated from data provided in the ACI code. The above value of Ko indirectly accounts for the effect of convection and conduction of heat from the outer surface to the surrounding air, while the Ki coefficient considers conduction and convection along with the radiation of heat from the fire to the inner surface of the chimney shell. It is assumed for this study that as a result of the fire, the temperature at the inner surface of the chimney is 870 C with an outer ambient temperature of 27 C. Fig. 3B shows a set of temperature profiles in a chimney shell with ri = 12.5 m (41 ft) and t = 0.46 m (18 in.) calculated at different fire durations.

2.2. Simple model for axial strength calculation during and post-fire The axial or vertical strength of the chimney, Pn , with average radius r , thickness t and concrete nominal compressive strength, fc , can be estimated from Pn = 2 f c rt , by neglecting the effect of the steel on the axial strength of the chimney. The vertical strength of chimney generally varies through its height since the radius and thickness of the chimney shell are not constant in most designs. To estimate the residual structural capacity of the chimney during fire, first, we calculate the concrete strength through the chimney shell thickness both during the fire event, as well as after the fire, when the chimney has cooled down (i.e. post-fire condition). It was assumed that the strength of the concrete at elevated temperatures depends only on the value of the temperature (i.e. not the rate of temperature change) allowing the concrete strength to be obtained from the curves in Fig. 1. We related each temperature profile to a concrete strength profile through the thickness, both during the fire and after cooling, which was used to calculate the average strength of the concrete, fc . For the post-fire residual strength, we calculated the average temperature at each point across the chimney shell by integrating the transient temperature data at the point over the total fire duration. Then, we used Fig. 1 to find the corresponding strength after cool-down at each point. The integration of concrete strength through its thickness provides an estimate for the reduction in the axial (vertical) strength of the chimney section denoted by = fc /fc . Here, we conservatively considered the water cooling data which will cause the maximum decrease in concrete strength post-fire. Fig. 4A shows the residual axial strength of the concrete chimney with ri = 12.5 m, and t = 0.46 m, and no opening, as a function of fire duration for both during-fire and post-fire conditions. The average reduction in the compressive strength of the chimney for the steady state temperature profile that is plotted in Fig. 3B, and by considering water cooling data as explained above, gives = 0.63. This indicates that the axial strength of the chimney in the case of steady state heat transfer through its thickness is reduced by 37%. Note that the fire duration required for reaching a steady state condition is very long and the strength data is only presented here to show the upper bound in the reduction of concrete axial strength. 2.3. ACI methodology for bending strength calculation From the structural perspective, the stresses acting on a chimney cross section result from bending due to lateral forces (wind and earthquake) as well as normal forces (weight of the reinforced concrete, the liners and the live loads). The structural

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

2891

Fig. 4. Post-fire behavior of a concrete chimney. (A) Reduction in the vertical (axial) strength of the concrete chimney, denoted by , as a function of fire duration during the fire event and post-fire. (B) Normalized nominal moment strength of the chimney versus the normalized vertical load for a fire duration of 3 h. The results are shown for normal conditions (before fire), during an ongoing fire event with the duration of 3 h, and post-fire. (C) Reduction in the nominal moment strength versus the normalized vertical load for the same cases as B. The concrete chimney has ri = 12.5 m (41 ft), t = 0.46 m (18 in.) and no opening.

model of chimney shells provided by the ACI code [14] accounts for the dead loads, thermal stresses and wind and earthquake loads. To quantify the role of the concrete temperature on the moment strength of a chimney, we used the method provided by the ACI code (ACI 30708) for reinforced concrete chimney design. In this method, the nominal moment strength of a cross-section of a concrete chimney is calculated from Mn = Pu r (cos + K2 /K1 ), where Pu is the factored vertical load at that cross-section and K1 and K2 are: K1 = 1.7Q + 2m Ke t Q1 + 2t 1 K2 = 1.7QR + m Ke t Q2 + 2t K where is half of the central angle subtended by the neutral axis, Q is the stress level correction parameter; K , R, Q1 and Q2 are the parameters for nominal moment strength, m is the maximum concrete compressive strain; and 1 are the geometrical parameters for each cross-section of chimney shell; Ke and t are non-dimensional parameters which are defined in Appendix. Considering the ACI recommendation, for a chimney with constant cross-section (average radius r and thickness t ), the axial load of the chimney can be calculated from Pu = 1.4 2 rtH g , where H is the height of the chimney located above the cross-section of interest (e.g. if the calculations are carried out for the bottom cross-section of the chimney, H is equal to the height of the chimney). Thus, the normalized axial load is Pu /rtfc = [2.8 g /fc ] H , where g is gravitational acceleration. However, it should be noted that most chimneys have varying thickness

along their height which should be considered when calculating the normalized axial load applied to the chimney cross-section. For a given chimney with known average radius, r , thickness, t , concrete compressive strength, fc , the angle which denotes the half-opening angle (if any), and the ratio of vertical reinforcement area to total area of concrete, t , the stress envelope for each cross-section of the concrete chimney can be plotted in nondimensional form as Mn /fc rt 2 against Pu /fc rt (see Appendix for details). Examples of the calculated stress envelopes for a concrete chimney with ri = 12.5 m and t = 0.46 m and no opening for a fire duration of 3 h are shown in Fig. 4B. Fig. 4C shows the reduction in the nominal moment strength of the concrete chimney versus the nominal vertical load for the cases considered in Fig. 4B. In Fig. 4, the results associated with the curves labeled during fire are calculated by excluding thermal stresses. The role of thermal stresses on the chimney strength is discussed in Section 5. In general, calculations of the residual structural capacity of chimney shells should be based on the post-fire condition, as the concrete loses additional strength after cooling. Another example is provided in Fig. 5A, where we plotted the stress envelopes, in non-dimensional form, for the same chimney at different times of the fire event. The results are obtained using the heat transfer results shown in Fig. 3B. Using these stress envelopes, we also estimated the reduction in the chimney nominal moment strength denoted by Mn /Mn (Normal condition), as quantified in Fig. 5B. The results are shown for various fire durations, as well as the steady state temperature profile which denotes the upper bound of reduction in the chimney nominal moment strength.

2892

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

Fig. 5. (A) Normalized nominal moment strength of the chimney versus the normalized vertical load in normal conditions and at different fire event durations. (B) Reduction in the nominal moment strength, denoted by Mn /Mn (Normal condition), versus the normalized vertical load for a chimney subject to uncontrolled fire. The results are shown for various fire durations, as well as the steady state temperature profile shown in Fig. 3B.

Fig. 6. Finite element model of the concrete chimney and comparison of the finite element results with the ACI structural model for chimney with no opening.

3. Iterative finite element approach for chimney strength calculation As an alternative to the ACI 307 methodology and to evaluate the validity of our analytical approach, we developed a finite element-based iterative method, a schematic of which is shown in Fig. 6. The method consists of a two-dimensional model of the chimney cross-section where the following assumptions are made (in accordance with the ACI standard code): (i) Plane strain condition across the section, (ii) The tensile strength of the concrete is ignored, (iii) The steel reinforcement in both compression and tension zones is taken into account, and, (iv) The maximum compressive strain in the concrete is 0.003. Assuming an angle for the compressive zone, we eliminate the portion of concrete which is assumed to be in tension as shown in Fig. 6. Then,

the axial load corresponding to the assumed compressive zone is applied and afterwards the bending moment is increased until the strain reaches the maximum allowable compressive strain (m = 0.003). The size of the compressive zone is examined (e.g. the state of stress in the concrete was checked to assure that the portion of concrete we are modeling is in compression). Based on the results, the size of the compressive zone size is changed and a new model is developed for the next iteration. This procedure is continued till convergence is achieved. In Fig. 6, the results obtained from the ACI code are compared with finite element results for a chimney subjected to a wide range of axial loads, showing good agreement between the two approaches. In the calculations, the simulations are repeated until we get the convergence with an error of 1 for the half angle of the compressive zone, .

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

2893

Fig. 7. Effect of chimney thickness. (A) Temperature profiles in a concrete chimney for various thicknesses after 3 h of fire. (B) Reduction in the vertical (axial) strength of the concrete chimney with different thickness (post-fire) as a function of fire duration. (C) Normalized nominal moment strength of the chimneys with various thicknesses (post-fire) versus the normalized vertical load for a 3 h fire event. The plot for the normal condition is also shown. (D) Reduction in the nominal moment strength versus the normalized vertical load for post-fire behavior of the chimney with different thicknesses after a 3 h fire event. The inner radius of the concrete chimney considered here was ri = 12.5 m (41 ft) and the section examined was assumed to have no opening. The temperature at the inner surface of the chimney was 870 C.

4. Parametric study and discussions In this section, we examine the impact of various design parameters on the residual structural capacity of concrete chimneys using as an example, a fire that lasts 3 h and causes the average temperature inside the chimney to be about 870 C. The results are presented in the form of stress envelopes and reduction in the chimney nominal axial and moment strengths. Residual strength maps were constructed to give an estimate of the reduction in the safety factors of the chimney as a function of chimney height, thickness and steel relative density, as well as the number and size of the opening in the chimney shell. Fig. 7 shows the effect of chimney shell thickness, t , on the stress envelopes of a chimney with ri = 12.5 m (41 ft) and no openings. Fig. 7A displays the temperature profiles in a concrete chimney with different thicknesses after 3 h of fire. For chimneys with smaller thickness, the average temperature would be higher, which results in lower concrete strength, and therefore lower axial strength and moment strength of the chimney. Fig. 7B shows the average reduction in the concrete strength, which is equal to the reduction in the vertical strength of the chimney, as a function of fire duration for different shell thicknesses. For relatively thick chimney shells, the reduction in axial strength of the chimney is sensitive to chimney thickness, as well as to fire duration. These results are extended in Fig. 7C and D, by calculating the normalized nominal moment strength of the chimneys with

various thicknesses versus the normalized vertical load for postfire behavior after a 3 h fire event. The plots for the normal condition are also shown. In Fig. 8, we look at the role of steel relative density, t , on the structural capacity of the chimney. We considered four different relative densities starting as low as 0.5% up to 2.0%, which is the normal range considered in designing concrete chimneys. The results for chimneys as built and after a 3-h fire are presented in Fig. 8A and B, respectively. The reduction in the nominal moment strength, which suggests a small difference for different relative densities of steel reinforcement, is plotted in Fig. 8C. In this set of calculations, the chimney has no opening, and has an inner radius, ri = 12.5 m (41 ft) and thickness, t = 0.46 m (18 in.). Fig. 9 presents the results for reinforced concrete chimneys with one opening. A schematic of the chimney cross-section with one opening is shown in Fig. 9A. The role of opening angle on the stress envelopes of a concrete chimney with ri = 12.5 m, and t = 0.46 m is quantified in Fig. 9B for both normal conditions and post-fire conditions (3-h fire event). The reduction in nominal moment strength for chimneys with one opening angle after 3 h of fire exposure (Fig. 9C) shows that, for a cross-section with wider opening, the average moment reduction is higher due to the overall weakness of the cross-section compared to the cross-section with no opening. The results are extended in Fig. 10 for chimneys with two openings, where denotes the half-opening angle and denotes

2894

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

Fig. 8. Effect of steel reinforcement relative density. (A & B) Normalized nominal moment strength of the chimneys with different relative steel densities versus the normalized vertical load. The results are shown for normal conditions (A) and after a 3 h fire event (B). (C) Reduction in the nominal moment strength versus the normalized vertical load for post-fire behavior of the chimneys with different relative steel densities for post-fire behavior after a 3 h fire event. The concrete chimneys have ri = 12.5 m (41 ft) and t = 0.46 m (18 in.) and no opening.

the half central angle between the two openings. The stress envelopes of a chimney with two openings, ri = 12.5 m and t = 0.46 m and total opening angle of 2 = 20, in normal conditions and following a 3 h fire event, are shown in Fig. 10B. Fig. 10C shows the reduction in the nominal moment strength of a chimney subjected to a 3-h fire. Our study indicates that for a constant opening angle, the variation of central angle does not much affect the structural performance of the chimney. In Fig. 11, we show contour plots of the reduction in nominal moment strength for different design parameters with respect to the height and thickness of the chimney cross-section. In the calculations, the vertical load applied to the chimney cross-section is simply calculated by estimating the weight of the concrete above the cross-section. The value of the vertical load is then used to obtain the nominal strength of the concrete chimney using the stress envelope curves corresponding to post-fire conditions after a 3-h fire event. Fig. 11A shows the results for a chimney cross-section with no opening and 1% and 2% relative density of reinforced steel. Fig. 11B and C compare the results for a crosssection with no opening, with cases for one opening, = 20 and two openings, = 10 and = 40. Note that the total opening angle is kept constant in Fig. 11C. These results can be used for estimating the reduction in the moment strength of concrete chimneys.

5. Effect of thermal stresses In the calculations provided above, the role of thermal stresses in calculating the axial and moment strengths of chimneys during fire was not considered. This aspect of the problem is considered here. It should be noted that the study includes only the effect of thermal stresses on the chimneys axial strength, since it is not very likely that the chimney will experience high design lateral loads during the internal fire [14]. Thus, calculation of the moment strength was considered less useful in this case. The ACI code [14] provides a method to calculate the stresses induced by the steady state (service) temperature profile. The ACI code accounts for the effect of the thermal stresses, by reducing the specified compressive strength of concrete according to fc = fc 1.20fCTV , where fCTV is the maximum vertical stress in concrete occurring inside the chimney shell due to temperature variation through its thickness. This method cannot be applied for the thermal stresses induced in the reinforced concrete chimney during an internal fire, since the temperature profiles across the thickness of the chimney vary significantly with time. To calculate thermal stresses, we have carried out a coupled heat transferstructural analysis using axisymmetric models of the concrete chimney shell. In the simulations, the concrete was taken to have different tensile and compressive responses. The behavior of the concrete was taken as elasticplastic under both

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

2895

Fig. 9. Residual structural capacity of concrete chimney with one opening. (A) Schematic view of the cross-section of a chimney with one opening. denotes the halfopening angle. (B) Normalized nominal moment strength of the chimneys with different opening angles versus the normalized vertical load for the normal condition and after a 3 h fire event. (C) Reduction in the nominal moment strength versus the normalized vertical load for post-fire behavior of the chimneys with different opening angles after a 3 h fire event. The concrete chimneys have ri = 12.5 m (41 ft), t = 0.46 m (18 in.), and t = 0.01.

tension and compression, however, with different yield strengths. Under compression, the yield strength was set equal to fc , while under tension, the yield strength of 3 MPa was assumed in the numerical simulations [15]. The steel was taken as an elastic material with Es = 200 GPa. Each simulation required several iterations similar to the finite elementbase iterative method explained in Section 3. In each iteration, we created a geometrical model of the chimney shell by assuming which portions of the concrete shell are under compression and tension. Material properties were assigned accordingly to each portion and to the steel. The finite element calculations gave an estimate of both the temperature profile during the fire, as well as the induced thermal stresses. We examined the stress profile across the thickness of the chimney shell and modified the compression and tension portions accordingly. This procedure continued until the assumed configuration was consistent with the induced stress profile. We used our method to analyze a chimney with a thickness of 18 in. (0.46 m) and 0.01 steel reinforcement subjected to an internal fire with temperature 1000 C for a duration of 50 min. The profile of the thermal stress induced by the internal fire after 50 min is shown in Fig. 12, where we have also shown the concrete strength at elevated temperatures simply calculated by relating the concrete strength to the local temperature at each point along the chimney thickness, Fig. 1. The calculated thermal stress profile was then used to estimate the residual axial strength of the chimney shell by accounting for the reduction in the concrete strength due to thermal stresses. In our calculations, we used the method

Table 1 Reduction in axial strength of the concrete chimney subjected to an internal fire with temperature 1000 C. The results are presented for a fire duration of 50 min. Effect of thermal stresses N (%) Reduction in concrete strength at elevated temperature N Y 100 88.52 Y (%) 86.14 83.56

provided in the ACI code to select the thickness of the chimney that is under compression for the first iteration. In Table 1, we have summarized the results of our calculations for the reduction in the axial strength of the concrete chimney during fire for four different cases (i.e. no material degradationno thermal stresses [NN], no material degradation with thermal stresses [NY], material degradationno thermal stresses [YN], and material degradation with thermal stresses [YY]). It should be mentioned that for the YY case, we assumed the compressive strength of the concrete to be fc (27.5 MPa), rather than varying it along the thickness according to fc , to simplify the calculations. However, this assumption did not change the steady state results compared to the ACI method and will not change the results of other cases considerably, since it would slightly change the location of zero stress along the thickness. The results show that if the effect of thermal stresses is neglected (case [YN]), the estimated axial strength of the chimney at elevated temperatures is 0.885 of chimney axial strength in normal conditions. On the other hand, by considering the thermal

2896

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

Fig. 10. Residual structural capacity of a concrete chimney with two openings. (A) Schematic view of the cross-section of a chimney with two openings with equal opening angle, 2 . (B) Normalized nominal moment strength of the chimneys with different versus the normalized vertical load for normal conditions and after a 3 h fire event. Note that the results for = = 10, are very close to the results shown in Fig. 6 for a chimney with one opening and = 20. (C) Reduction in the nominal moment strength versus the normalized vertical load for post-fire behavior of the chimneys with different after a 3 h fire event. The concrete chimneys have ri = 12.5 m (41 ft), t = 0.46 m (18 in.), and t = 0.01.

stresses and neglecting the reduction in the compressive strength of the concrete at elevated temperatures (case [NY]), the axial strength of the chimney is 0.861 of chimney axial strength in normal conditions. Considering the effects of both thermal stresses and reduction in the concrete strength at elevated temperatures (case [YY]) gives a reduction factor of 0.836. The residual axial strength of the chimney shell for post-fire conditions is 0.885 for the same fire conditions. 6. Conclusions Reduction in the axial strength and moment strength of a reinforced concrete chimney, due to a fire event, was investigated for a range of design and geometrical parameters. The temporal variation in the concrete chimney during the fire was calculated using an axisymmetric heat transfer finite element analysis. The reduction of the concrete strength due to the fire was calculated using experimental data available in the literature, which gives a direct estimate of the reduction in the axial strength of the chimney at different cross sections. The moment strength of the damaged concrete was calculated using the standard method of ACI. Based on the proposed approach, the reduction in the axial strength depends only on the thickness and fire duration. On the other hand, the reduction in the moment strength is a complex function of chimney design parameters (e.g. thickness, height, steel reinforcement and opening(s) size and location(s)) and fire duration. The ACI code [14] provides a method to calculate the

stresses induced only by the steady state temperature profile and thus, cannot be used directly to estimate the reduction in the chimney strength due to the thermal stresses during the fire. We used finite element simulation to study the role of thermal stresses on the behavior of reinforced concrete chimney during fire. The results show that considering thermal stresses does not significantly affect the estimated reductions in chimneys axial strength. Our study showed that the detrimental effect of uncontrolled fire on the residual strength of the chimney is stronger for tall chimneys with thinner shells. Openings in the chimney shell result in further reduction in the residual strength of the chimney. Thus, shell openings and their exact geometry should be considered when evaluating the residual strength of concrete chimneys. Acknowledgments The authors thank Franco Tamanini and Louis Gritzo of FM Global for their valuable comments. This work was supported in part by FM Global, in part by the US Department of Homeland Security and in part by the Department of Mechanical and Industrial Engineering at Northeastern University. This material is based upon work supported by the US Department of Homeland Security under Award Number 2008-ST-061-ED0001. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the US Department of Homeland Security.

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

2897

Fig. 11. Reduction in the nominal moment strength for post-fire behavior of chimneys with a range of heights and thicknesses. The results are presented for fire temperature 870 C and duration 3 h and considering the post-fire condition. The numbers of the plots show the corresponding value of Mn /Mn (normal condition). (A) Reduction in the nominal moment strength for two chimneys with 1% and 2% vertical steel reinforcement. Solid and dashed lines are plotted for chimneys with t = 0.01 and t = 0.02, respectively. (B) Reduction in the nominal moment strength for two chimneys with and without opening. Solid and dashed lines are plotted for chimneys with no opening and with one opening with = 20, respectively. (C) Reduction in the nominal moment strength for two chimneys with one opening and two openings. The concrete chimneys had ri = 12.5 m (41 ft), t = 0.46 m (18 in.), and t = 0.01. Solid and dashed lines are plotted for chimneys with one opening with = 20 and with two openings with = 10 for each opening and = 40.

Appendix For a given chimney with known radius, r , shell thickness, t , concrete compressive strength, fc , the angle which denotes the half-opening angle (if any), and the ratio of total vertical reinforcement to total area of concrete, t : Pu /rtfc = K1 = 1.7Q + 2m Ke t Q1 + 2t 1 where

= n1 sin sin ( ) cos Q1 = 1 cos 1 = +

Fig. 12. Induced thermal stresses and the calculated compressive strength of concrete at elevated temperatures through the chimney shell thickness. The stresses and strength are normalized by the concrete compressive strength.

, , are geometrical parameters for a cross-section that are

discussed in [14], and

2898

A. Vaziri et al. / Engineering Structures 33 (2011) 28882898

cos = 1 1 (1 cos ) cos = cos cos = cos +

fy

1.0 m Es 1 cos fy 1.0 m Es

1 cos

Q2 =

( ) 1 + 2 cos2 + (1/2) (4 sin 2 + sin 2 sin 2) 4 cos (sin + sin sin )

(1 cos ) K = sin + sin + ( ) cos

R = sin ( n1 ) cos (n1 /2) [sin ( + ) sin ( )] where is the one-half angle between center lines of two openings (For a cross-section with no opening, n1 = = = 0; for a crosssection with one opening in the compression zone, n1 = 1, = 0; for two openings in the compression zone, n1 = 2). References
[1] Arioz O. Effects of elevated temperatures on properties of concrete. Kidlington, ROYAUME-UNI: Elsevier; 2007. [2] Mohamedbhai GTG. Effect of exposure time and rates of heating and cooling on residual strength of heated concrete. Mag Conc Res 1986;38:8. [3] Nassif A. Postfire full stressstrain response of fire-damaged concrete. Fire Mater 2006;30:32332. [4] Sakr K, El-Hakim E. Effect of high temperature or fire on heavy weight concrete properties. Cement Concr Res 2005;35:5906. [5] Xiao J, Falkner H. On residual strength of high-performance concrete with and without polypropylene fibres at elevated temperatures. Fire Safety J 2006;41: 11521. [6] Fletcher I, Borg A, Hitchen N, Welch S. Performance of concrete in fire: a review of the state of the art, with a case study of the Windsor tower fire.in: Proceedings of the 4th international workshop in structures in fire. 2006. p. 77990. [7] Tenchev R, Purnell P. An application of a damage constitutive model to concrete at high temperature and prediction of spalling. Int J Solids Struct 2005;42:655065. [8] Canisius TDG, Waleed N, Mattews SL. A preliminary study of the Cardington Concrete Building under a compartment fire. In: Proc structural faults & repair conf. 2003. [9] Both C, van de Haar P, Tan G, Wolsink G. Evaluation of passive fire protection measures for concrete tunnel linings. In: Proc int conf on tunnel fires and escape from tunnels. 1999. [10] Eurocode 2: Design of concrete structuresGeneral rules, Structural fire design (BS EN 1992-1-2:2004). British-Adopted European Standard; 2005. [11] Ali HM, Senseny PE, Alpert RL. Lateral displacement and collapse of singlestory steel frames in uncontrolled fires. Eng Struct 2004;26:593607. [12] Ali HM, Senseny PE, Alpert RL. Clearance between single-story steel frames and firewalls. Eng Struct 2005;131:13. [13] Wang Y. Steel and composite structures: behavior and design for fire safety. New York, NY: Taylor and Francis; 2002. [14] Design and construction of reinforced concrete chimneys (ACI 307-08). American Concrete Institute. [15] Hillerborg A, Moder M, Petersson PE. Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements. Cement Concr Res 1976;6:77381.

where Ke = Es /fy , t = t fy /fc , n1 is the number of openings entirely in the compression zone (the maximum allowed base on the ACI code is two openings), is one-half the central angle subtended by the neutral axis, and,

1 = 0.85 for fc 4000 psi 1 = 0.85 0.05(fc 4000)/1000 0.65 for fc 4000 psi m = 0.07 (1 cos ) / (1 + cos ) 0.003.
The nominal moment strength of a cross-section, Mn can be estimated from Mn = Pu rK3 , where K3 = cos + K2 /K1 and K2 = 1.7QR + m Ke t Q2 + 2t K . For 5

0.523 + 0.181 0.0154 2 + 41.3 13.2 + 1.32 2 (t /r ) . For 5 < 10 Q = 0.154 + 0.01773 0.00249 2 + (16.42 1.980 + 0.0674 2 (t /r ) . For 10 < 17 Q = (0.488 + 0.076) + (9.758 0.640 ) (t /r ) . For 17 < 25 Q = 1.345 + 0.2108 0.004434 2 + (15.83 1.676 + 0.03994 2 (t /r ) . For 25 < 35 Q = (0.993 0.00258 ) + (3.27 + 0.0862) (t /r ) . For > 35 Q = 0.89.
Q =

Other parameters are calculated from the following equations: