Dynamic Properties of High Volume Fly Ash Nanosilica HVFANS Concrete Subjected to Combined Effect of High Strain Rate and Temperature

The study aims to determine the dynamic properties of high volume fly ash nanosilica HVFANS concrete exposed to strain rates between 30.12 to 101.42 s‐1 and temperatures of 25, 400, and 700 oC by using split Hopkinson pressure bar SHPB machine. The static and dynamic compressive strengths of HVFANS concrete were slightly lower than plain concrete PC at room temperature, while its values were higher at 400 and 700 oC. The results proved that the CEB model of dynamic increase factor is more reliable to estimate the behaviour of HVFANS concrete at studied temperatures. The toughness, critical strain, and damage of HVFANS concrete recorded a superior performance than PC under studied strain rates and temperatures that would reflect the possibility of use HVFANS concrete in structures to improve its resistant of fire and impact loads, as well as to decrease the demand on Portland cement which could lead to restrict the risks of liberated gases during cement production. Furthermore, equations were proposed to estimate the dynamic increase factor, toughness, and critical strain of both concretes under investigated conditions.


INTRODUCTION
Since early decades of 20th century, numerous studies were conducted to assess the concrete properties under elevated temperature to ensure the safety of engineering structures 1-3 .The extensive use of high strength concrete HSC in modern construction motivated several scholars to study its thermal properties.The results were contradictory and indicated that the strength of HSC might deteriorate a higher or lesser than normal strength concrete NSC at elevated temperature 4-7 .Recently, several alternatives binders were widely used in HSC as a partial cement replacement to reduce the risks of CO 2 emissions liberated during cement production 8 .
Fly ash FA was broadly utilized as a Portland cement replacement owing to it's a great performance in concrete and low price.FA is a by-product of the pulverized coal ignition in thermal power plants.Since the 1980s, high volume fly ash HVFA concrete started to use in Canada via replacing the Portland cement by 55 to 60% of FA 9 .There is no a unified definition for HVFA concrete until now.Sivasundaram et al. 10 stated that replacement of Portland cement with 30% of FA may define as HVFA concrete.However, other standards and researchers declared that FA replacement percentage should be above 40 or 50% 11-15 .
The main disadvantage of using HVFA in concrete is decreasing of compressive strength at early ages due to the slow pozzolanic reaction of FA 16-20 .Hence, nanosilica NS were extensively used in HVFA concrete due to its high pozzolanic reactivity with CH liberated from cement hydration to produce calcium silicate hydrate CSH which greatly enhances the concrete strength 21-24 .Several studies were carried out on HVFA concrete with NS to investigate its thermal properties.Ibrahim 25,26 reported that HVFANS concrete can be used as a fire proof material under temperatures reach up to 700 o C via replacing the Portland cement by 2.5 and 52.5% of NS and FA, respectively.Whereas the HVFANS concrete maintained about 94.54% of its room compressive strength after exposing to a temperature of 700 o C. In the same context, Radzi 27 used the same above proposed mixture of Latin American Journal of Solids and Structures, 2018, 15  Table 2 showed the properties of colloidal nanosilica NS type Cembinder W8 provided by AkzoNobel Company, which was used in the concrete mixture to improve the early compressive strength of concrete that contained a high volume of FA.The fine aggregate was natural river sand passed through a 4.75 mm sieve size with a fineness modulus of 2.98 and bulk specific gravity of 2. 53 40, 41 .The loose and compacted bulk density of fine aggregate were 1510.18 and 1721.83kg/m 3 , respectively 42 .The coarse aggregate was crushed granite provided from local sources with a maximum size of 10 mm and bulk specific gravity of 2.07.
Superplasticiser SP type Darex Super 20 was utilized to acquire the desired workability of concrete mixture and increase the compressive strength by decreasing the required volume of mixing water based on naphthalene sulphonate.Fibrillated polypropylene PP fibres manufactured by Timuran Company with a length of 12 mm and specific gravity of 0.9 had utilized to increase the ductility of concrete and eliminate the surface spalling 43-47 .The dosage of SP was equal to 1% of cementitious materials weight and PP was equal to 1 kg/m 3 of concrete volume.

Mixture Preparation and Curing Process
American concrete institute standard ACI 211.4R-9348 was applied to select the mixture proportions of PC and HVFANS concrete samples with water to cementitious materials ratio w/c p equal to 0.29 and targeted static compressive strength of 60 MPa as shown in Table 3.The HVFANS concrete samples were prepared via replacing the cement with 52.5% of FA and 2.5% of NS based on the recommendations of Ibrahim's study conducted in Universiti Kebangsaan Malaysia UKM 25, 26 .During the study, the serving sequence of materials according to ASTM C192 49 caused cement balls particularly for HVFANS concrete because the mixture was quite hot and stiff.To avoid this problem, the fine and a portion of coarse aggregate with cement and FA were firstly mixed via horizontal pan mixer.Then, the colloidal NS and water were added to the mixture taking into consideration the water content of NS particles.Finally, the rest of coarse aggregate and PP were added to break up any noodles of the mixture 50 .The slump test was performed before adding SP and kept between 25 to 50 mm.The mixture was cast into cylindrical and cubic steel moulds with dimensions of 50 Ø50 mm and 150 150 150 mm, respectively.The samples were de-moulded after 24 h, except the HVFANS concrete specimens which stayed in moulds for 48 h 26 .Then, all the samples were placed into a water tank and cured for 28 days.

Heating Process
The static and dynamic properties of PC and HVFANS concrete samples were determined at high temperatures of 400 and 700 o C, these temperatures were selected according to prior studies performed by Ibrahim 25,26 .The heating process followed ISO 834 fire curve 51 and carried out by using an electric furnace with the capacity of 3 concrete cubes and heating rate of 9 o C/min.After 2h of heating at the mentioned target temperatures, the heating process was stopped and then the samples were removed and cooled down to the ambient temperature.

Static Test
The static test was performed based on British standard BS EN 12390-3 52 by using an automatic compression test machine produced by Unit Test Scientific Company UTS with a loading capacity of 5000 kN.The average compressive strength values of the three specimens were recorded under 25, 400 and 700 o C at curing age of 28 days.

Dynamic Test
The dynamic test was conducted by using split Hopkinson pressure bar SHPB machine available in UKM laboratory as shown in Fig. 1a and 1b 53, 54 .The SHPB machine consisted of striker bar moving inside launcher cylinder, incident bar and transmitter bar.All the bars are fabricated from stainless steel with Young modulus of 210 GPa and wave velocity of 5190 m/s.The bars designed with a large ratio of length to diameter in order to ensure uniaxial wave propagation through the bars.The bar diameter of 50 mm was found to be the most appropriate to achieve the minimum friction between bars and supports to reduce the waves dispersion which significantly improved the recorded signals of strain gauges as shown in Fig. 1c 53 .A good contact between bars and concrete samples was achieved to minimize friction by smoothing the sample surface with a hand grinder and sandpaper.In addition, the bars were well coaxial by using laser alignment within an accuracy up to 0.01 mm to satisfy the force equilibrium condition as shown in Fig. 1d 53 .
A total of 150 cylindrical concrete samples with 50 Ø50 mm were tested within an average of five samples for each applied strain rate under different thermal conditions for PC and HVFANS concrete.Lindholm 55 stated that the samples have to be short enough to rapidly achieve a uniform stress state along the length of concrete sample during loading.However, using of very short samples may lead to increase the friction between the bars and sample ends, hence an apparent rise in strength could occur 56 .The specimen diameter also requires to be small comparative to the wavelength of the applied load pulse to successfully achieve the theory of one dimensional wave without the complication of Pochhammer-Chree radial oscillations 57 .In most cases, the optimum ratio of length to diameter L/D is 0.5 according to the relationship suggested by Davies and Hunter 57 to prevent wave dispersion.Nevertheless, the experimental results of Davies and Hunter did not observe any obvious effect of specimen size on stress-strain curves within L/D ratios range between 0.31 to 1.55 57 .Consequently, the L/D ratio was chosen to be 1 by using specimens with diameter of 50 mm and length of 50 mm.Similar L/D ratio was used by other researchers 53, 54 .
During the test, the cylindrical concrete sample is sandwiched between the incident and transmitter bars, and then the striker bar was launched at a known velocity toward the incident bar to achieve the desired strain rate.Newton's second law applied to determine the striker's acceleration according to the pressure exerted on the striker bar 58, 59 .A compression stress pulse was generated into the incident bar during the impact and travelled in a uniaxial direction toward the interface between the incident bar and concrete specimen.Its amplitude mainly relied on the striker bar velocity 60, 61 .At this point, a part of the pulse is reflected back along the incident bar as a tensile pulse, and the remaining part is transmitted from the concrete sample to the transmitter bar as a compressive pulse and then recorded by a strain gauge.Two Omega strain gauges' model SGD-3/350-LY11 were used to mimic the material elongation of incident and transmitter bars.The change in the resistance of strain gauge was too small due to the low values of strain and the only way to detect its change is by using a voltage measurement.Therefore, the strain gauge was quarterly connected to Omega bridge sensors model OM2-163 which considers as a complete signal conditioning system and consist of high performance instrumentation amplifier.The bridge sensors were connected to Omega data acquisition module OMB-DAQ-3000 which used OMB-DaqView-XL computer software to convert the analogy voltage signals to digital data that could be stored and analysed.MATLAB software was used to analyse the digital data and obtained the stress-strain diagrams of concrete samples.The strain gauge of the incident bar recorded the incident pulse and reflected pulse , while the strain gauge on the transmitter bar measured the transmitted pulse 62 .Once all the pulses are recorded, the stress , strain rate t , and strain histories of the sample can be computed, respectively, by using the following equations 54, 62, 63 : where , , are the cross-sectional area, the Young's modulus and the wave velocity of the bar material, and, , are the length and cross-sectional area of the specimen.

Static Test Results
Table 4 and Figure 2 indicated that the compressive strength of HVFANS concrete at room temperature was marginally decreased by 4.21% as compared with PC due to the high quantity of fly ash FA class F which contained a low amount of cement-like properties CaO 64 .HVFANS concrete appeared a clear increase in compressive strength at exposing temperature of 400 o C because the heating process significantly enhances the NS reactivity as well as increase the hydration process via producing a high-density calcium silicate hydrate structure which greatly improves the concrete strength 64 .At temperature of 700 o C, a dramatic reduction in compressive strength of HVFANS and PC concrete was observed due to the extreme accumulation of vapour pressure inside the specimens which caused a large number of cracks, besides the dehydrate of cement paste binder products at this temperature.Nevertheless, HVFANS concrete revealed a higher residual compressive strength by 85.13% from their original strength at room temperature as compared with PC which maintained only 59.44%, this behaviour may attribute to the filler effect of NS in the mixture which increased the calcium silica hydrate content for HVFANS concrete.The relative error of each compressive strength group was less than 7.78% which revealed that the experimental records were fairly reliable.

Compressive Strength and dynamic increase factor DIF
The ultimate dynamic compressive strength represents the peak stress determined from stress-strain curves of PC and HVFANS concrete specimens subjected to five average strain rates and temperatures of 25, 400, and 700 o C as shown in Fig. 3.
The dynamic stress-strain curves appeared a close relationship with the increasing of average strain rate for both concrete types and the descending part of curves was pointed to the samples destruction and strain relaxing after ultimate stress 65 .The dynamic compressive strength increased approximately linearly with the increase of average strain rate for both concrete types under temperatures of 25, 400, and 700 o C as shown in Fig. 4. Small variations in compressive strength were observed at 400 o C for both concrete types; however, a rapid reduction was detected at 700 o C particularly for PC samples.The HVFANS concrete showed a good resistant for dynamic loads at exposing temperatures of 400 and 700 o C than PC in all the investigated cases of strain rate, that reflected the great effect of NS on concrete strength which works as a filler for small voids that may increase the calcium silica hydrate content in mixture.
Dynamic increase factor DIF broadly used to evaluate the effect of strain rate on the compressive strength and defined as the ratio between dynamic to static compressive strength 3, 34 .The results proved that DIF mainly depended on the strain rate and increased almost linearly with the increase of strain rate for PC and HVFANS concrete samples at all the investigated temperatures as shown in Fig. 5. Fig. 5. DIF of PC and HVFANS concrete at various strain rates and temperatures.
The response surface methodology RSM available in the Design Expert Software version 11 66 was used for the statistical design and analysis of experimental data.The main goal was to derive an equation that could significantly represent the relationship of DIF with the exposing temperature T and dynamic strain rate .The linear and 2FI solver models with the inverse transformation were the most accurate to fit the experimental data of DIF for PC and HVFANS concrete, respectively, as follows: For PC: where: T is the exposing temperature between 25 to 700 o C, and is the dynamic strain rate within a range of 30.12 to 101.42 s -1 .
The results of variance analysis ANOVA were used to evaluate the accuracy of the adopted models as shown in Table 5.The outcomes revealed that the current models can be efficiently used to predict the values of DIF for both types of concrete within the mentioned ranges of temperature and strain rate.The quality of the fit polynomial model was expressed by the value of correlation coefficient R 2 which should be close to 1 with a minimum value of 0.8 67, 68 .Moreover, the differences between "Predicted R 2 " and "Adjusted R 2 " should be less than 0.2 and the Adequate Precision AP which measures the signal to noise ratio of the models should be greater than 4   Several experimental tests had extensively used the DIF to evaluate the effects of dynamic strain rate on concrete in tension or compression 70, 71 .Based on to their studies, expressions were recommended to determine the DIF of concrete in compression as follows: a In the 1993s, CEB 72 proposed an equation to calculate the DIF of concrete in compression under high strain rates reach up to 300 s -1 : where is dynamic strain rate, is static strain rate which equal to 30  In the present study, the dynamic increase factor curves of the experimental test were extended behind the strain rate of 30 s -1 until the static condition when the DIF is equal to 1.The experimental dynamic increase factors of both concrete types under temperatures of 25, 400, and 700 o C were compared with the results of the recommended expressions of dynamic increase factor as described in Fig. 6.The outcomes proved that the CEB model expression may efficiently estimate the dynamic increase factor of both concrete types under room temperature.However, the differences between the experimental and CEB results clearly increased at elevated temperatures particularly in case of PC under 400 and 700 o C that may attribute to the great loss in concrete strength.Nevertheless, the CEB model was more reliable to predict the behaviour of HVFANS concrete under mentioned temperatures that reflect the excellent role of nanosilica material which significantly enhances the strength of concrete under high temperature.
It can be concluded that the CEB expression could successfully use to obtain a full range of dynamic increase factor versus strain rate relationship for HVFANS concrete in compression under temperatures of 25, 400, and 700 o C which could significantly utilize as input data for the nonlinear concrete material model in numerical analysis of structure constructed by using this type of concrete and subjected to fire and blast loads.
where A, E, c are the cross sectional area, Young's modulus, and elastic wave speed of bars, respectively, and , are the length and cross sectional area of the specimen: , , are the incident pulse, reflect pulse and transmit pulse, respectively; T is the complete failure moment of specimen.concrete specimens increased linearly with increasing of strain rate, similar behaviour was observed by researchers on concrete material tested by using SHPB machine under room or elevated temperatures 76, 77 .Small variances were noted in toughness of both concrete types at room temperature.However, the increase of temperature until 400 o C led to reducing the toughness of PC, while the same temperature caused an increase in toughness of HVFANS concrete by 9.79% under average strain rate of 100.67 s -1 as compared with the toughness at room temperature.Fig. 7.The toughness of PC and HVFANS concrete at various strain rates and temperatures.
On the other hand, a notable reduction was observed in the toughness of PC and HVFANS at a temperature of 700 o C by 84.02 and 28.83% under average strain rates of 100.81 and 100.54 s -1 , respectively, as compared with the toughness at room temperature.Design Expert software was used to suggest the relationship of SEA with dynamic strain rate and exposing temperature T by using quadratic solver model with square root and natural algorithm transformations for PC and HVFANS concrete, respectively, with a high accuracy as shown in Table 6.For PC:

Critical strain
Critical strain defined as the strain of concrete corresponding to peak stress, which used to describe the ductility of PC and HVFANS concrete samples exposed to high strain rates and temperatures of 25, 400, and 700 o C as shown in Fig. 8.The critical strain increased approximately linearly with increasing of strain rate and recorded small values at temperatures of 25 and 400 o C for both types of concrete ranged between 0.005 to 0.0075 due to the existent of PP fibres which considerably enhance the ductility of concrete and provide escape channels for vapour pressure inside the voids at elevated temperatures 78-80 .Fig. 8. Critical strain of PC and HVFANS concrete at various strain rates and temperatures.
The smallest values of critical strain were observed for HVFANS400 that may attribute to the high reactivity of NS in the mixture which increased at a heating temperature of 400 o C. Otherwise, the critical strain obviously increased at a temperature of 700 o C for both concrete types and the highest increment of critical strain was noted for PC700 by 47.75% as compared with PC25 under average strain rate of 100.81 s -1 .The relationship of critical strain ε with dynamic strain rate and temperature T was proposed by using Design Expert software with a quadratic solver model and inverse transformation for both concrete types within a good accuracy as shown in Several studies had used the continuous damage mechanics theory to describe the critical damage of concrete under dynamic loads according to the differences in stress-strain curves with the increase of strain rate 81-83 .The critical damage variable was presented and defined as a scalar: where is the peak stress, is the critical strain and is the initial modulus of elasticity.The given expression defined the critical damage as a kinematic variable that mainly relies on the density of voids or cracks in a specific cross section which may result in the gradual damage of concrete 84, 85 .The case of 1 means that the concrete is not able to resist the applying loads.Figure 9 indicated that the critical damage increased almost linearly with the increase of strain rate and proved that the damage of HVFANS concrete at room temperature was slightly higher than PC under similar conditions of strain rate, that may attribute to the high volume of fly ash in the mixture.However, the HVFANS concrete appeared lower values of critical damage at temperatures of 400 and 700 o C by 6.26 and 9.14% under average strain rates of 100.67 and 100.54 s -1 , respectively, as compared with PC under similar range of strain rate.Fig. 9. Critical damage of PC and HVFANS concrete at various strain rates and temperatures.
During experimental study, the influences of strain rate and temperature on the damage of PC and HVFANS concrete were visible.The failure of concrete samples was so violent and caused severe destruction of cylindrical test samples into small pieces at high strain rates as shown in Fig. 10a.The pattern of shear failure could only observe at the range of strain rate between 30.12 to 70.91 s -1 under studied cases of room and elevated temperatures except for the case of PC under 700 o C which was totally destructed at an average strain rate of 70.17 s -1 .The variation of concrete failure with increasing of strain rates was shown in Fig. 10b.
It can be noted that the shear failure of concrete samples could pass through three stages according to the increase of strain rate.In the first stage when the average strain rate between 30.12 to 50.91 s -1 , the crack's path was passed throughout the concrete mortar only.For average strain rates of 70.17 to 71.32 s -1 , the path of cracks was more straight and passed through a rough fracture surface which had a more broken aggregate that may attribute to the rapid increase of stress in samples which was quite sufficient to break the aggregate and concrete mortar zones along the path of less resistance.In the final stage of strain rates  85.17 s -1 , the cracks amount was greatly increased and resulted in a severe fracturing of concrete samples into small fragments to dissipate the internal energy.In the same context, numerous researchers were obtained similar patterns of shear failure for cement-based materials according to the increase of strain rate 75, 86-88 .

CONCLUSION
The results can be summarized as follows: 1.The static compressive strength of PC was higher by 4.21% than HVFANS concrete at room temperature.A dramatic reduction in compressive strength was observed at 700 o C for both types of concrete; nevertheless, the HVFANS concrete recorded a high residual strength of 85.13% as compared with PC which maintained 59.44%.2. HVFANS concrete achieved an excellent performance than PC in terms of dynamic compressive strength, toughness, critical strain and damage under elevated temperatures of 400 and 700 o C subjected to average strain rate between 30.12 to 101.42 s -1 .3. The CEB model was the most appropriate to predict the dynamic increase factors of HVFANS concrete under the studied temperatures.4. Equations were proposed to predict the dynamic increase factor, toughness, and critical strain of both concretes within an exposing temperature range of 25 to 700 o C and average strain rate of 30.12 to 101.42 s -1 .The mentioned results proved a good possibility to use HVFANS concrete in structures to improve its resistant to fire and dynamic loads, besides its eco-friendly properties.

Fig. 1 .
Fig. 1. a SHPB machine view b Components of SHPB mm c Friction effect on incident wave d Recorded signals of incident and transmitted bars during the test

Fig. 2 .
Fig. 2. Behaviour of PC and HVFANS concrete at different temperatures

Fig. 6 .
Fig. 6.Comparison between DIF of test results and existing expressions at different temperatures Figure 7 indicated that the SEA of PC and HVFANS Dynamic Properties of High Volume Fly Ash Nanosilica HVFANS Concrete Subjected to Combined Effect of High Strain Rate and Temperature Latin American Journal of Solids and Structures, 2018, 15 1 , e06 12/19

Fig. 10 .
Fig. 10.Failure modes of PC and HVFANS concrete at various strain rates.

Table 1 .
Chemical compositions of Portland cement and fly ash.

Table 3 .
Mixture materials kg/m 3 of PC and HVFANS concrete.

Table 4 .
Static compressive strength of PC and HVFANS concrete at various temperatures.
69 .Dynamic Properties of High Volume Fly Ash Nanosilica HVFANS Concrete Subjected to Combined Effect of High Strain Rate and Temperature Latin American Journal of Solids and Structures, 2018, 15 1 , e06 10/19

Table 5 .
The results of ANOVA analysis for DIF models of PC and HVFANS concrete.
Tedesco et al. 73derived an equation to determine the DIF of concrete, whereas the transition from low strain to high strain rates was occurred at 63.1 s -1 :

Table 6 .
The results of ANOVA analysis for SEA models of PC and HVFANS concrete