Nội dung

Construction and Building Materials 44 (2013) 317–328 Contents lists available at SciVerse ScienceDirect Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat High performance concrete under elevated temperatures Abdullah Huzeyfe Akca, Nilüfer Özyurt Zihnioğlu ⇑ _ Department of Civil Engineering, Boğaziçi University, Istanbul, Turkey h i g h l i g h t s  Performance of HPCs under elevated temperatures.  Use of PP fibers with air entraining admixture to decrease damage.  Decreased spalling and increased residual strength.  Complete disintegration of dense matrix under very high temperatures.  Microstructural examination of cement paste–aggregate interface. a r t i c l e i n f o Article history: Received 4 February 2013 Received in revised form 28 February 2013 Accepted 2 March 2013 Available online 10 April 2013 Keywords: High performance concrete Elevated temperatures Polypropylene fibers Air entraining admixture ESEM a b s t r a c t In this study, PP fibers and air entraining admixture (AEA) were used together in an high performance concrete (HPC) mix so as to create interconnected reservoirs in concrete and to improve fire performance of HPC. For this reason, nine mixes of HPC incorporating blast furnace slag with 0.24 water-to-binder ratio and various PP and AEA contents were produced. Specimens were cast in two different sizes in order to see the effect of size and 18 series of specimens were obtained. These series subjected to elevated temperatures (300 °C, 600 °C and 900 °C) with a heating rate of 10 °C/min and after air cooling, residual mass and compressive strength of specimens were determined. The heated specimens were observed both at macro and micro scales to investigate the color changes, cracking and spalling of HPC at various temperatures. Also, thermogravimetric analyses were performed on powder samples from each nine mixes. Results showed that addition of AEA diminished the decrease in residual strength but this result was found to be irregular after 300 °C for thick specimens. The collaboration of AEA and PP fibers decreased the risk of spalling of HPC. Also, size of specimen was found to be important in deterioration of HPC. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction Concrete with increased strength and durability has been primarily used in special constructions such as high rise buildings, infrastructures and nuclear power plants since it became commercially available [1]. Thenceforth, some of these HPC structures exposed to severe fire conditions have exhibited poor performance. The main reason of this insufficiency of HPC at high temperatures is a result of the changes made in the composition of concrete mixes. Decrease in water to cementitious ratio, use of supplementary cementitious materials and plasticizers lead to impressive improvements such as strength, rheology of fresh concrete, impermeability and compactness. On the other hand, in most cases these changes may lead to a decrease in fire performance of HPC [2]. _ ßaat Mühendisliği ⇑ Corresponding author. Address: Boğaziçi Üniversitesi, Ins _ Bölümü, 34342 Bebek, Istanbul, Turkey. Tel.: +90 212 359 70 39, Mobile: +90 533 690 22 44; fax: +90 212 287 24 57. E-mail addresses: abdullah.akca@boun.edu.tr (A.H. Akca), Nilufer.ozyurt@boun. edu.tr (N. Özyurt Zihnioğlu). 0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2013.03.005 Lower water to cementitious materials ratio leads to lower porosity and this decreases permeability of concrete. With the increase in temperature, water in the pores of concrete evaporates and consequently pressure within the cement paste increases. Reduced permeability of HPC limits the diffusion of water vapor from the concrete pores and therefore pore pressure continues to increase until the internal stresses reach the tensile strength of concrete and eventually causes spalling [3]. Free water and moisture gradients influence the behavior of concrete at elevated temperatures and according to Hertz, they must be regarded as main reasons of spalling [4]. Meyer-Ottens treats that tensile stresses caused by steam in the closed pores of normal concrete can reach the tensile strength of concrete with more than 3% moisture by weight [5]. Hertz concluded traditional concrete with less than 3% moisture by weight will not spall and in the range of 3–4% moisture by weight has a risk of spalling, on the other hand, concrete with a dense microstructure (most of the HPC) may spall even when the moisture content is zero [4]. Due to the increased impermeability, only the crystal water arisen from 318 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 1000 900 800 700 600 500 400 300 200 100 0 0 60 120 180 240 300 360 420 480 Time (Min) Fig. 1. Heating cycles. dehydration of hydrates at high temperatures may cause the spalling of concrete. Recently, many studies have focused on the contribution of different materials such as fibers and mineral admixtures to improve fire resistance of HPC [2,6,7]. Addition of PP fibers into HPC was found as an efficient way to avoid spalling of concrete. Because, PP fibers melt in concrete above 170 °C and leave micro channels in concrete and these channels form a network more permeable than cement matrix which contributes to outward migration of gases and water vapor and result in the reduction of pore pressure [7–10]. As a mineral admixture, inclusion of silica fume caused reduction in residual strength and spalling of concrete by densifiying microstructure [11,12]. On the contrary, addition of fly ash or slag showed better performance and also in some studies, strength gain observed at temperatures ranged from 200 °C to 300 °C because of tobermorite formation [13]. Furthermore, rapid heating of concrete is another factor which causes a high temperature difference between the deeper zone and the surface of a specimen and therefore explosive spalling may occur during heating [14]. Anderberg stated that during his Table 1 Properties of cement, slag and polypropylene fibers. Cement and slag Polypropylene fibers Chemical composition Cement (%) Slag (%) SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O Loss on ignition Specific gravity (g/cm3) Specific surface (cm2/g) 20.17 4.91 3.41 64.28 1.18 2.84 0.13 0.96 1.61 3.14 3910 38.37 11.89 1.05 37.25 8.13 0.38 0.28 1.28 0 2.93 4320 Length (mm) Diameter (lm) Specific gravity (g/cm3) Specific surface (cm2/g) Fiber number (fibers/kg) Tensile strength (MPa) Modulus of elasticity (GPa) Melting point (°C) 12 32 0.91 1340 110 Million 250 3.5 165 Table 2 Mix proportions. a b c Series W/B Cement (kg/m3) Slag (kg/m3) Water (kg/m3) Sand (kg/m3) SPa (kg/m3) PP (dm3/m3) AEAb (g/BA)c F0A0 F0A0.5 F0A1 F8A0 F8A0.5 F8A1 F16A0 F16A0.5 F16A1 0.24 0.24 0.24 0.24 0.24 0.24 0.24 0.24 0.24 515 515 515 515 515 515 515 515 515 435 435 435 435 435 435 435 435 435 227 227 227 227 227 227 227 227 227 1175 1175 1175 1175 1175 1175 1175 1175 1175 15 15 15 15 15 15 15 15 15 0 0 0 8 8 8 16 16 16 0 5 10 0 5 10 0 5 10 SP stands for superplasticizer. AEA stands for air entraining admixture. BA stands for binder amount of 1 m3 concrete mixture in kilogram. Table 3 The number of specimens spalled at different temperatures (out of 3 for each series). Series Maximum temperature 300 °C T5F0A0 T5F0A0.5 T5F0A1 T5F8A0 T5F8A0.5 T5F8A1 T5F16A0 T5F16A0.5 T5F16A1 a 0 0 0 0 0 0 0 0 0 600 °C 3 3 3 2 0 0 0 0 0 Series 900 °C a N.E. N.E. N.E. 2 0 0 0 0 0 T10F0A0 T10F0A0.5 T10F0A1 T10F8A0 T10F8A0.5 T10F8A1 T10F16A0 T10F16A0.5 T10F16A1 N.E. – these specimens were not exposed to 900 °C since equivalent specimens spalled at 600 °C. Maximum temperature 300 °C 600 °C 900 °C 0 0 0 0 0 0 0 0 0 3 3 3 2 0 0 0 0 0 N.E. N.E. N.E. 1 0 0 0 0 0 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 Fig. 2. (a) A PP fiber passing through an air void, (b) a micro-channel formed after heating (a part of a melted PP fiber reaches to an entrained air void creating a micro-channel). 319 experiments only rapid fires have given rise to spalling [15]. Related to this, size and shape of concrete can be considered as factors which directly affect the heating rate of concrete. When a structural member subjected to high temperatures, the heat level of thin parts increases rapidly and this may cause spalling due to rapid heating [4]. On the other hand, although they are widely used especially in HPC mixes, there are a limited number of studies on the effect of chemical admixtures on concrete under fire conditions. For example, as a chemical admixture, air entraining admixtures (AEAs) are used for producing air bubbles in concrete to improve resistance of concrete to damage caused by freezing and thawing situations. Moreover, the entrained air enhances the workability and may reduce bleeding and segregation of concrete mixtures at fresh state and with the increase in air voids in the concrete thermal conductivity of the concrete decreases at the hardened state [16]. As Riley stated the surface of concrete with low thermal conductivity which acted as a refractory material would effectively produce an insulation layer to inner parts of concrete [17]. In a study conducted by Seçer, AEA was used to entrain air into concrete in the ratios of 4%, 6% and 8% in volume and these concretes subjected to temperatures of 300 °C, 500 °C and 700 °C. Results of the study showed that as the air content of concrete increased, reduction in the strengths of concretes subjected to high temperatures diminished. In this study, PP fibers and AEA were used together in HPC so as to form a more permeable network consisted of micro channels of melted PP fibers and entrained air voids in concrete at elevated temperatures. Thus, it was aimed to permit the evacuation of gases and water vapor appeared due to heating. To the authors’ knowledge this is the first study to discuss the combined effects of polypropylene fibers and air entraining admixture on the properties of high performance concrete under elevated temperatures. Considering the moisture and size factors which influence the effect of temperature, moisture contents of HPC specimens were adjusted Fig. 3. Outer surfaces of the specimens of T5F16A1 series. (a) specimen kept at room temperature (20 °C), specimens heated to (b) 300 °C, (c) 600 °C, (d) 900 °C (2X magnification). 320 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 content becomes more significant since HPC is much denser than normal concrete. In this study, moisture contents of HPC specimens were aimed to adjust approximately to 3% by weight to simulate the most negative condition. Therefore, moisture contents of one specimen from each series were determined in accordance with BS1353 before exposure to elevated temperatures [20]. All the specimens were found to have moisture content in the range of 2.7–4.4% which can be considered as in the critical region. 2.2. Heating procedure An electrical furnace that was capable to operate up to 1250 °C was used. After the curing period, moisture content of specimens reached the desired value and specimens of each series were exposed to 300 °C, 600 °C and 900 °C temperatures for an hour in the furnace. The heating rate was set to 10 °C/min which can be considered the same as the average heating rate of standard fire curve (ISO-834) for the first 90 min. It should be emphasized that this heating rate is detrimental because of the thermal gradients between the outer part and the inner core of the specimens which cause to additional internal stresses in concrete and initiates spalling [4]. At the end of the set exposure time, the hot concrete specimens were not taken out until the furnace cooled down to 100 °C with a cooling rate of 3 °C/min. Fig. 1 represents the heating cycles. 2.3. Test procedures Mass measurements and compressive strength tests were performed on both unheated and heated concrete specimens. Also, specimens were observed at both macro and micro scales. 2.3.1. Macroscopic and microscopic observation Assessment of fire-damaged concrete begins with visual observation of color change, cracking and spalling of concrete. Changes on visual appearance of concrete give information about the temperature which concrete has been exposed. In the scope of this study, occurrence of spalling, crack patterns and color changes were examined at macro scale. Additionally, changes on surface and interior part of the heated specimens were observed microscopically. Fig. 4. (a) Popouts (8X magnification) on sand particles heated to 900 °C (b) aggregate cracking (6X magnification) on the specimen (T5F16A1) heated to 900 °C. approximately to 3% by weight to simulate the most negative condition before heating and concrete specimens with two different sizes were prepared to see the effect of size at high temperatures. 2. Experimental study 2.1. Materials and mix design CEM I Type 42.5R Portland cement, ground granulated blast furnace slag (GGBS) from Karabük, river sand, multifilament polypropylene (PP) fiber, a high-range water reducing admixture based on chains of modified polycarboxylate ether and an air entraining admixture based on oil alcohol and ammonium salt were used in the production of concrete. It should be emphasized that no coarse aggregate was used in the mixture. The maximum size of the sand used was 1 mm. The properties of cement, slag and PP fiber are presented in Table 1. Nine mixes of HPC with 0.24 water-to-binder ratio and various PP and AEA contents were produced. Mix proportions are shown in Table 2. F0A0 is control group and represents concrete with no PP fiber and AEA. F8, F16 specimens contain PP fiber at 8‰ and 16‰ of the volume of the concrete amount, respectively. A0.5 and A1 specimens have AEA to binder ratio of 0.5‰ and 1‰ respectively. Mixes were cast into 10  10  50 cm prisms and the specimens were kept in laboratory environment for 24 h. After demoulding, the specimens were labeled and then they were cured in a water tank at 20 °C for 10 days. After 10 days of curing period 10  10  50 cm specimens were cut with a diamond blade in order to obtain 10  10  10 cm cubes (represented by T10) and 10  10  5 cm prisms (represented by T5) and then these specimens were kept in laboratory environment for 3 months. At elevated temperatures, it is known that the extent of damage increases with an increase in the moisture content and 3–4% moisture by weight was found to be critical for normal concrete by various researchers [4,18,19]. For HPC moisture 2.3.2. Measuring mass loss Mass of each specimen was measured before heating test and measured again after the heated specimens cooled down to room temperature. Moreover, thermogravitmetric analyses were performed on powder samples obtained from each concrete type by using a TA Instruments Q50 thermal analyzer. The thermal analyzer heated the sample to 863 °C (which is the maximum heating capacity of the used testing machine) with a constant rate of 10 °C/min and simultaneously measured the mass of the sample. Finally, thermogravimetric (TG) and differential thermogravimetric (DTG) curves of samples were drawn. 2.3.3. Residual strength measurement After exposure to high temperatures, three specimens of each series were subjected to compression test in accordance with BS 12390 to measure the residual compressive strength [21]. The specimens with dimensions of 10  10  10 cm were tested as they were. On the other hand, all the 10  10  5 cm specimens were cut using a diamond blade in order to obtain 5  5  5 cm cube specimens for compression test. These dimensions were considered representative of the material since concrete produced in this study contained only fine aggregates with a maximum size of 1 mm and entrained air voids whose maximum and average diameter were 500 lm and 100 lm, respectively. The fiber length (12 mm) was also smaller than 1/3 of the smallest dimension of the 5  5  5 cm specimen. This test method consisted of applying a compressive axial load to cube specimens at a constant rate of 0.2 MPa/s until failure occurred. 3. Results and discussion The experimental test results obtained from compression tests and mass loss measurements and visual observations are discussed in this section. Summary of the results are given in the form of tables and figures. 3.1. Spalling Explosive spalling of some specimens were observed when the specimens exposed to 600 °C and spalling began at approximately 500 °C (This statement is done based on the sound of explosive spalling) [22]. Partial spalling such as corner spalling and surface layer delamination was not observed in this study. Table 3 shows IDs and number of specimens destroyed due to explosive spalling. A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 321 Fig. 5. Inner surfaces of specimens of T5F16A1 series (a) the specimen kept at room temperature (20 °C), specimens heated to (b) 300 °C, (c) 600 °C, (d) 900 °C As is seen in Table 3 none of the specimens spalled at 300 °C, all the non-fibrous specimens were exploded at 600 °C and therefore specimens from these series were not exposed to 900 °C. As a similar result to the findings of Han, no explosive spalling was observed when PP fibers were used except some specimens with IDs T5F8A0 and T10F8A0 [23]. Explosive spalling was observed in these fibrous specimens and this phenomenon can be explained by dense microstructure of HPC. According to Peng (based on experiments done using HPC with a compressive strength of 80 MPa), regardless of the interconnected channel system formed by PP fiber melted above 170 °C, concrete at 0.24 water-to-binder ratio was so dense that it could still keep the water pressure high enough to result in explosive spalling [14]. It should be noted that while the specimens with 8‰ fibers and no air entrainment (T5F8A0 and T10F8A0) exploded, the air entrained specimens of the same series (T5F8A0.5, T10F8A0.5, T5F8A1 and T10F8A1) were not exploded as can be seen in Table 3. This result could be attributed to the effect of air entrainment. It is hypothesized that, the contribution of air entrainment to resist against explosive spalling began with PP fiber addition. Entrapped and/or entrained air voids in concrete are almost closed and as Hertz stated; if water vapor cannot escape from these closed pores it causes increase in pore pressure and increases the risk of spalling [4]. In the air entrained PP fiber reinforced concrete, most probably micro-channels were formed due to melting of PP fibers at above 170 °C and some of the closed pores connected to each other by these micro-channels. On the other hand, in non-fibrous air entrained concrete, absence of fibers limited the ability of water vapor to escape from the entrained air voids in HPC and thus spalling occurred. To examine this effect, microstructures of concrete specimens were examined by using an environmental scanning electron microscope (Philips XL30 ESEM-FEG/EDAX) and formed micro-channels were observed. In Fig. 2a, PP fiber passes through an air void and in Fig. 2b, a melted PP fiber creates a micro-channel by reaching an entrained air void. Consequently, this result shows that the existence of both PP fibers and entrained air voids in HPC may reduce the risk of explosive spalling. 3.2. Color and cracking observation on the outer surfaces of the specimens In all cases red discoloration was observed at 300 °C, gray discoloration was observed at 600 °C on the outer surface of the specimens and at 900 °C the surface colors of the specimens were changed to whitish gray. Ingham stated that red color change is a result of hydrated iron oxides present mostly in siliceous aggregates, pink to red discoloration is very important and has a structural significance because it means that temperature around 300 °C where the reduction in concrete strength mostly began was attained [24]. The deterioration of a structural member exposed to 300 °C can be repairable. On the other hand, whitening of a structural member indicates that temperature has exceeded 600 °C and it corresponds to a serious loss in compressive strength. After this amount of strength loss, concrete cannot be repairable anymore and it cannot withstand service loads. Outer surfaces of the specimens were examined visually and by using a stereomicroscope (Nikon SMZ1500) (Fig. 3). Almost no cracks were observed on the outer surfaces of the specimens that were heated until 300 °C. On the other hand, cracks with an opening going up to 0.2 mm were observed on the surfaces of the specimens heated to 600 °C. Surface cracks formed on the specimens exposed to 900 °C were larger and the crack widths were in the range of 0.3–0.4 mm. Moreover, siliceous sand particles were separately heated to 900 °C and cracks and popouts on the surfaces 322 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 Fig. 6. Decomposition of a specimen heated to 900 °C. were observed due to the volume change of sand particles (Fig. 4a). The effect of volume change of sand particles can also be seen in Fig. 4b. 3.3. Color and cracking observation on the inner surfaces of the specimens On the other hand, color of the inner surface of specimens is less influenced than that of outer surface at the same temperature level. Color of cement paste in inner surfaces is prominently darker than in outer surfaces at 300 °C and 600 °C, Fig. 5b and c. Moreover, reddened fine aggregates in the inner surface are lighter than outer surface at these temperatures. This could be due to the fact that the exact same temperature may not be reached in the inner sections of concrete. Furthermore, no cracking was observed on the inner surfaces of the specimens which were heated up to 600 °C. Inner cracks both in aggregates and cement paste were only visible in the specimens which were heated to 900 °C as seen in the Fig. 5d. Moreover, widths of the cracks in the inner zone of the concrete were smaller than that of the outer zone and were around 0.04 mm. The specimens heated to 900 °C decomposed and decomposition did not take place immediately. One or two large cracks (0.3–0.4 mm) were observed in the first day (following removal of the specimens from the furnace, Fig. 6a), then, these cracks turned into spider web-like cracks in the 2nd day (Fig. 6b) and finally complete disintegration of specimens were observed on the Fig. 7. Average mass losses of specimens. 3rd day (Fig. 6c). In literature this phenomenon is explained by the decomposition of hydrates which starts at 400 °C and almost ends at 900 °C. Therefore, most of the dehydration takes place and concrete loses almost all of its initial strength and stability at 900 °C due to loss of all chemical water [25]. Culfik and Ozturan 323 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 TGA Instrument: TGA Q50 V6.7 Build 203 0.05 100 0.04 Deriv. Weight (%/°C) 98 Weight (%) 96 0.03 94 0.02 92 0.01 90 88 0 200 400 600 Temperature (°C) 0.00 1000 800 Universal V4.7A TA Instruments Fig. 8. TGA and DTG curves (F0A0). Table 4 Average initial and residual compressive strength values of T5 series (the results given are the average compressive strength values calculated by using the data obtained from 3 specimens). The residual strength values given for the specimens exposed to 900 °C should be carefully evaluated based on the explanation given in Section 3.5 Mixes T5F0A0 T5F0A0.5 T5F0A1 T5F8A0 T5F8A0.5 T5F8A1 T5F16A0 T5F16A0.5 T5F16A1 a b Control strength Residual strength 20 °C 300 °C (MPa) (MPa) (%) (MPa) (%) (MPa) (%) 121.10 93.60 91.37 114.70 86.23 92.63 109.47 92.13 81.03 101.40 112.07 104.99 71.27 75.28 78.33 69.03 71.18 72.04 83.7 119.7 114.9 62.1 87.3 84.6 63.1 77.3 88.9 E.S.a E.S. E.S. 57.41 48.53 55.69 52.81 48.22 46.97 – – – 50.1 56.3 60.1 48.2 52.3 58.0 E.S. E.S. E.S. C.D.b C.D. 18.05 C.D. 18.59 C.D. – – – – – 19.5 – 20.2 – 600 °C 900 °C ES stands for explosive spalling. CD stands for complete disintegration. Table 5 Average initial and residual compressive strength values of T10 series (the results given are the average compressive strength values calculated by using the data obtained from 3 specimens). The residual strength values given for the specimens exposed to 900 °C should be carefully evaluated based on the explanation given in Section 3.5. Mixes T10F0A0 T10F0A0.5 T10F0A1 T10F8A0 T10F8A0.5 T10F8A1 T10F16A0 T10F16A0.5 T10F16A1 a Control strength Residual strength 20 °C 300 °C (MPa) (MPa) (%) (MPa) (%) (MPa) (%) 133.97 98.33 80.76 125.17 97.70 95.74 124.27 99.65 94.34 121.57 94.54 77.72 110.33 85.51 91.16 120.13 78.81 80.58 90.7 96.1 96.2 88.1 87.5 95.2 96.7 79.1 85.4 E.S.a E.S. E.S. 85.09 55.29 59.55 75.71 58.76 51.89 – – – 68.0 56.6 62.2 60.9 59.0 55.0 E.S. E.S. E.S. 42.08 22.61 24.81 37.49 21.27 23.48 – – – 33.6 23.1 25.9 30.2 21.3 24.9 600 °C 900 °C ES stands for explosive spalling. exposed their specimens to 900 °C. They also reported hair-like cracks in first day and then complete disintegration of concrete specimens 1 day after cooling period [26]. Only six specimens of two series (T5F8A1 and T5F16A0.5) did not decompose. These specimens were exposed to water during the cutting process and most probably were healed when came contact with water [27–29]. This recovery can be attributed to regeneration of some of C–S–H bonds on rehydration [28,29]. Detailed information about disintegration of these specimens will be given in Section 3.5. 3.4. Mass losses After exposure to high temperatures mass losses of the specimens, resulting mainly from water and carbon dioxide transport and loss, were recorded. Residual masses of each series can be seen in Fig. 7 except the spalled series. Average mass losses of specimens exposed to 300 °C, 600 °C and 900 °C were 5.2%, 9.8% and 12.9%, respectively. Results of thermogravimetric analyses were similar. Average mass losses of powder samples at 300 °C, 600 °C and 863 °C were 324 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 Fig. 9. Residual strengths of the specimens with increasing air entrainment. Fig. 11. Residual strengths of the specimens with increasing specimen size. est and appears between approximately 550 °C and 700 °C and corresponds to the loss of carbon dioxide ensuing from the carbonation products and the loss of water ensuing from the decomposition of calcium silicate hydrates (C–S–H) [24,25,30–32]. During the last peak mass of the samples reduced approximately 4%. This explains the serious loss of strength after 600 °C. TGA and DTG curves obtained for the other samples were not published here since they were very similar to the one given below for F0A0 sample. 3.5. Residual strength Fig. 10. Residual strengths of the specimens with increasing PP fiber content. 4.6%, 7.9% and 11.6%, respectively. These results were smaller than the residual mass values found by weighing the full-size specimens. This is expected since the full-size specimens included closed pores which may entrap extra moisture when compared to powder samples. Moreover, the specimens heated in the furnace were held at their maximum temperatures for an hour before they left for cooling, while the powder samples exposed to TGA test were cooled down immediately. Fig. 8 represents the TGA and Differential Thermo Gravimetry (DTG) curves. As shown in DTG curve, there are three peaks on the graph and these peaks represent instantaneous mass losses with temperature. The first peak is wider than others and starts with the beginning of the heating (approximately 20 °C) and continues until the complete evaporation of moisture of powder samples (approximately 150 °C). Average mass loss of the samples is 3.6% around this region. The second peak appears between approximately 400 °C and 450 °C and corresponds to the loss of water from portlandite [24,25,30–33]. The instantaneous mass loss of the samples during second peak was 0.7%. Finally, the third peak is the high- Compressive strength tests were conducted on three specimens of the control series before beginning heating cycles. Compressive strength values ranged from 80 MPa to 130 MPa at ambient temperature for different mixes. After exposure to high temperatures, the residual compressive strengths of specimens were measured. Residual strength measurements could not be conducted for non-fibrous specimens after exposure to 600 °C, since all the specimens were ruined due to explosive spalling. The overall results of the residual compression test show that the concrete specimens exposed to lower temperatures are stronger than the specimens exposed to higher temperatures as expected. Tables 4 and 5 show the original and residual strength of all series. An important note should be given here. Tables 4 and 5 show some residual strength values for the specimens heated up to 900 °C. However, these values may be misleading since all of these specimens were found to disintegrate couple of days after testing. The testing program was prepared such that specimens of T10 series were tested before T5 series. All the specimens of T10 series were tested 1 day after heating. The specimens showed some residual strength as is seen in Table 5. Unfortunately, after couple of days the tested specimens were found completely disintegrated. Having that in mind, it was decided to test 2 series of the 5 cm thick specimens immediately, while keeping the rest of them in the laboratory environment for 3 days to check if the same phenomenon will occur. The specimens that were spared to be tested immediately were first cut to obtain 5  5  5 cm cubes and then tested. The specimens of these 2 series (T5F8A1 and T5F16A0.5) showed some residual strength as can be seen in Table 4, while the specimens kept in the laboratory environment were completely disintegrated in 3 days. The specimens that were not A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 (a) 325 (b) (c) Fig. 12. ESEM pictures of the specimens kept at room temperature (20 °C). (a–b) matrix aggregate interface, (c) hydration products. disintegrated during and after the test were exposed to water during the cutting process and most probably were healed when came contact with water [27–29] as explained before in the end of Section 3.3. This recovery can be attributed to regeneration of some of C–S–H bonds on rehydration [28,29]. The parts of these specimens (T5F8A1 and T5F16A0.5) further observed for couple of months after the test and no disintegration was observed. This is an important information and may be the subject of another project. Table 4 shows that residual strength increases when air entraining admixture was used for all the specimens with a thickness of 5 cm when the specimens were heated to 300 °C and 600 °C, respectively. A similar comment cannot be made for the specimens heated to 900 °C since most of them were completely disintegrated. This result is also valid for some of the 10 cm thick specimens. Residual compressive strengths (%) of specimens at various temperatures with different air entrainment are shown in Fig. 9. According to the results, air entrainment affects the residual compressive strength of concrete. However, air entrained non-fibrous specimens exploded above 600 °C. With the absence of microchannels formed by melted PP fibers, this situation can be explained by the increased pore pressure in the closed air voids of concrete. Air entrained specimens lost strength less than others (except T10F16 series) at 300 °C for all specimens and residual strengths percentages of T5 series increased with air entrainment at 600 °C. However, the positive effect of air entrainment was not clear on the specimens of T10 series after 300 °C. Also, there is not a prominent difference between the residual strength percentages of A0.5 and A1 series. All the fibrous specimens withstood heating cycles and they were subjected to compressive strength test. Residual compressive strengths (%) of specimens at various temperatures with different PP fiber ratios are shown in Fig. 10. Although PP fibers prevented spalling of specimens, their existence in material adversely affected the residual compressive strength of HPC. Adding more PP fiber into HPC mixes had negative effect on the residual strength of the specimens and these results confirm the findings given in literature [10,34]. Residual compressive strengths (%) of specimens at various temperatures with different specimen sizes are shown in Fig. 11. Size of the specimens affected the compressive strength of HPC. The specimens with 10 cm height showed better performance at elevated temperatures and this effect is clearer when temperature was increased up to 600 °C (The specimens heated up to 900 °C were assumed to represent no residual strength since they were disintegrated 3 days after the test). Specimens with 5 cm height retained 77% and 54% and 20% of their ambient strength at 300 °C, 600 °C and 900 °C, respectively. On the other hand, specimens with 10 cm height retained 89% and 60% of their original strength for 300 °C, 600 °C. The residual strength results measured are in agreement with the values given in Eurocode 2 for concrete with siliceous aggregates which retains 85%, 45% and 8% of its initial strength at 300 °C, 600 °C and 900 °C, respectively [35]. The difference in residual strengths between the specimens with 5 cm height and the specimens with 10 cm heights can be a result of rapid heating of specimens with smaller size as mentioned by Hertz [4]. 3.6. ESEM observations Selected specimens were examined by using an environmental scanning electron microscope for better evaluating the effects of high temperatures on the microstructures of the specimens. The regions which are considered as matrix–aggregate interface were especially chosen to be examined since hydration products are supposed to develop in these regions [36]. Energy dispersive Xray spectroscopy (EDX) analyses were carried out to detect CSH phases. Phase regions with a Ca/Si ratio between 0.8 and 2.1 are considered to be CSH regions [36,37]. The regions examined in the scope of this study had a Ca/Si ratio around 1, 7. This result 326 A.H. Akca, N. Özyurt Zihnioğlu / Construction and Building Materials 44 (2013) 317–328 is similar to the average Ca/Si atomic ratio reported by Djaknoun [36]. Fig. 12 represents ESEM pictures taken on a specimen kept at room temperature (20 °C). Fig. 12a and b shows aggregate–matrix interface. Aggregate–matrix interface of the specimens kept at the room temperature represents a continuous structure with no pores and cracks. Cement hydration products have well defined crystal structure, with portlandite and CSH (Fig. 12c). Fig. 13a and b represent ESEM pictures of a specimen heated to 300 °C. Fig. 13a shows matrix aggregate interface. Fig. 13b shows a magnified view of the cement matrix given in Fig. 13a. Cement paste still has crystal structure characteristics, however is not as distinctive as it was for unheated specimen. Fig. 14 shows pictures taken on the specimens heated up to 600 °C. Fig. 14a again shows aggregate–matrix interface. Structure of the hydration products is amorphous. Fig. 14b shows the cracks on the surface. As can be seen on the figure, cracks are spread all over the surface of the specimen due to the important amount of water loss from the structure. Fig. 15 shows pictures of a specimen heated up until 900 °C. As mentioned before, almost all of these specimens disintegrated couple of days after being removed from the electrical furnace (the specimen shown below is one of the specimens which was not disintegrated). Fig. 15a shows porous cement–aggregate interface. Fig. 15b represent, the cement structure which has amorphous structure, Fig. 15c shows a cracked sand particle and finally Fig. 15d shows the cracks spread all over the specimen. (a) (b) (a) Fig. 14. ESEM pictures of the specimens heated to 600 °C. (a) matrix aggregate interface, (b) distributed cracks on the surface of the specimen. ESEM pictures show the effect of high temperatures on the microstructure of HPC. Concrete microstructure is highly damaged with an increased temperature leading to failure of the specimens. 4. Conclusions Behavior of HPC under high temperatures is different than normal concrete due to very dense microstructure. Precautions should be taken to decrease the damage occur when HPC exposed to high temperature. In this study air entraining admixture was used together with polypropylene fibers to create channels for evacuating water vapor. Following conclusions were drawn as the result of this study. (b) Fig. 13. ESEM pictures of the specimens heated to 300 °C. (a) matrix aggregate interface, (b) cement paste.  Spalling of HPC seems to be dependent on presence of PP fiber in concrete. Explosive spalling was observed especially in nonfibrous specimens and began after 500 °C. For other HPC