High-temperature behavior of zeolite-containing mortar. Experiment and modelling

High temperature experiments were carried out on cylindrical-shaped specimens of mortar containing different concentrations of either cement-replacing zeolite or aggregate-replacing zeolite. The behavior was compared with that of plain mortar. Furthermore, finite-element models were designed, based on material characteristics obtained on the real specimens, to test the temperature exposure effect when similar conditions were applied.


Introduction
The construction industry has a considerable impact on both available resources and the environment.There is a growing interest in finding valid alternatives for the usual construction materials.Zeolites have received a lot of attention due to their pozzolanic activity [1], which stems from high concentrations of Al2O3 and SiO2 [2].They are naturally occurring, volcanic, rocks that can be also synthetized, and have been used as either aggregate [3] or cement replacement [4,5].
Using zeolites in concrete and mortar has been shown to have an impact on many properties of the resulting material, with the benefits or disadvantages depending on the concentration [1].Compressive strength [6], freeze-thaw resistance [7], increased resistance to chloride and sulphate attacks [8] are some of the problems tackled by using zeolites.
This study focused on thermal behavior, which was tested to be, generally, better for mortars containing zeolites [9].On the other hand, Uzal and Turanli [10] experimentally showed that cements containing clinoptilolite tuff have less pores larger than 50nm when compared to ordinary Portland cement specimens, which means a higher probability for the appearance of exfoliation.Nevertheless, similar to other properties, there are many parameters that can influence, directly or indirectly, the performance at elevated temperatures, such as: purity, aluminum content, the type and structure of the zeolites.
Testing a new material to fire resistance can be an expensive endeavor.Getting as much information as possible from the material, rather than the element, characterizations might prove important from an economic point of view.An in-house developed experimental stand was used to measure and record the heat transfer through mortar samples subjected to elevated temperature.Based on the recorded temperature data, the numerical models were calibrated to obtain a good correlation between simulations and experiments, in order to validate the models as instruments for high-temperature exposure studies.Based on the validated numerical models, a comparison of previously tested mortar specimens, only by using simulations, in a controlled and error free environment, was conducted.This is believed to be a valid approach for the preliminary stage of larger studies on the thermal behavior of new materials subjected to elevated temperatures.

Materials and methods
Reference mortar specimens were cast using CEM II/B-M (S-LL) 42.5R rapid hardening Portland cement, (0-4) mm sand as fine aggregate and tap water.The cement had a 65-79% clinker amount, with 21-35% blast furnace slag and limestone.The reference specimen was referred to as "Ref.".The fine aggregate and the cement were replaced in various percentages, and designated as follows:  AZx = recipes with aggregates replaced by "x" percentage of zeolite aggregates (x = 10, 20, 30). PZx = recipes with cement replaced by "x" percentage of zeolite powder (x = 10, 20, 30).In all cases, the percentage denotes a replacement by volume.This was translated into mass quantities to be used in the actual mixing procedure.Nevertheless, in the case of zeolite aggregates, the density of particles is similar to those of sand, thus the replacement was performed by mass from the beginning.For the PZ specimens, the densities used in computing the mass values were: ρcement = 3150kg/m 3 and ρzeolite_powder = 2377kg/m 3 .
Reference water-to-cement ratio (w/c) was 0.6.Based on the existing literature, it is expected that using zeolite, in either form (aggregate or powder), will change the water needed for the mix to have good workability.Thus, when zeolite aggregate was used, the w/c was increased alongside the percentage replacement in order to balance its water absorption.When replacing the cement with lower density zeolite powder, to achieve the same water-to-binder ratio (w/b) as the reference would imply lowering the water quantity.However, for the same reason described above, the same amount of water was employed, resulting in an increased w/b.Quantities used for a reference amount of 500g of cement are given in table 1.The zeolite used in this study was Clinoptilolite (87-90%) and has a composition (both powder and aggregate) provided by seller and presented in table 2. It is important to note that Clinoptilolite is stable up to 450°C and has a pH of 8.7.The maximum particles size for the zeolite aggregates was 4mm, and the powder consisted of particles with a mean diameter of 29µm.The specimens were cast in cylindrical steel molds with a diameter of 70mm and a height of 30mm.Every recipe was used to create 12 cylinders, that are used in the current research.Provided that existing results indicate a good amount of consistency across the specimens of the same mix, and given the focus of this study, only one specimen of each recipe was used.Density and thermal transfer properties were assessed close before the high-temperature testing.
The study consisted of two parts.First part was the experimental evaluation of the high-temperature effect on the specimens.The second part was the finite element simulation aimed at both calibration and comparison.

Experimental assessment
The experiments were carried on a stand created for this purpose.A sketch the arrangement is presented in figure 1.The heat source was an air-heating gun capable of reaching temperatures up to 550°C, as per the producer's technical data.The specimens were positioned 1cm away from the gun.Before starting the high-temperature exposure, a K-type thermocouple was employed to measure the temperature of the hot air, in order to start the experiment from a known state.This would also help comparing results with data obtained by simulations.Temperature was monitored on the side opposite to the heated one, in the center point, by means of a FLIR BCam thermal imaging camera.

Simulations assessment
Simulations were performed using the LISA 8.0 FEM analysis software.It provides ample analysis options even in the free version, that allows 1300 nodes.For the present purpose, the number of nodes and, subsequently, the number of elements associated to it was enough for a thorough assessment.
Each specimen was modeled starting from an initial circular surface mesh consisting of 96 elements and 107 nodes (figure 2a), and of appropriate radius.The nodes were arranged in a concentrical manner.Furthermore, the extrusion process to the needed thickness was employed with a number of 10 subdivisions (figure 2b).This resulted in a total of 960 elements and 1177 nodes for each specimen.Temperature was applied on the central surface, on a central zone of d=17.5mmdiameter.This was chosen to be in agreement with the experimental setup.The heat source was modeled as a part of the body, extruded from the center of one of the faces (named "hot surface").This part was kept constant at the desired high-temperature (figure 3).For the opposite circular surface (named "cold surface") and the lateral surface we applied the property of convection with the outside air, using the following parameters: Tair = 22°C, transfer coefficient = 5 W/(m 2 •K).

Results and discussions
In the first instance, specimens' dimensions, mass and thermal properties (thermal conductivity and specific heat) were measured.Table 3 holds the values obtained for the particular specimen used from each recipe.These values were used as input parameters for the simulations.The density of the zeolite containing specimens is lower, than the reference mix, as expected.The influence is greater for zeolite aggregates than for zeolite powder, since the porosity should be higher.In a similar fashion, the thermal conductivity decreases linearly with increasing the zeolite concentration.Replacing aggregates results in a higher decrease in thermal conductivity than replacing cement (figure 4a)).It can be noted that this could also be an indicator of the lack of water inside the pores of zeolites.
The AZ specimens show an increased need of energy for increasing temperature when the replacement percentage is higher.Nevertheless, the AZ10 specimens is the only mix that has a lower specific heat than the reference one.The AZ20 presents a similar value to the reference and AZ30 a higher one.In the case of PZ mixes, the behavior is not linear.While PZ10 and PZ30 show similar, though lower than the reference, values of the specific heat, the PZ20 spikes above that (figure 4b)).The porous structure of zeolite aggregates could play an important role in the thermal behavior of the mortars using them.A small amount of zeolite aggregate might result in lower porosity than the reference mix, following that the water in the zeolite pores will lead to a higher amount of hydration products.A bigger replacement percentage could result in a higher porosity, since the cement quantity is fixed in a certain specimen.With a specific heat value of 1000 J/(kg•K), the air is harder to heat up than mortar, thus more air results in a higher specific heat.
On the other hand, zeolite powder might produce less porous mortars than its aggregate counterpart, based on the density comparison.In this case, the above attempted explanation, on how the specific heat is influenced by entrapped air, holds for PZ10 and PZ30.While the exception showed by PZ20 requires further study in the microstructure of the specimens in order to be explained, this is not the aim of this paper.

Experimental results of high-temperature exposure
For every specimen in the study, the model was calibrated around the experimental results.The heating of the specimens was done at a constant distance, but the device used to obtain the high temperature was prone to errors.The aim of the study was to perform a comparison of the involved materials even in the presence of such an inconvenience.In figure 5, the results of high-temperature exposure are presented.The AZ specimens show an increasingly insulating behavior with increased replacement percentage (figure 5a)).This is to be expected, taken into account the values for thermal conductivity and specific heat of the mixes.For the PZ recipes, the maximum temperatures reached after 10 minutes of heating are closely related to the specific heat properties shown above (figure 5b)).
This data gave us an insight into the behavior at high-temperature for the specimens, and it also provides an incentive to rectify, if needed, the comparison charts with the reference.We wanted to show that, even if a somewhat unprecise experimental setup is used, valuable and reliable data can be obtained with additional fine-tuning.Each curve presented in figure 5 was used as a basis for comparison with simulation results.

Simulation results of high-temperature exposure
The simulations have been performed for every specimen that was previously experimented on.Using the precise characteristic data (density, thermal conductivity and specific heat), models were created in order to demonstrate that the experiments were correct, albeit with an exposure temperature that was not exactly known.This step in the study was important for validating the process rather than the data.In order to achieve this, several exposure temperatures in the 400ºC -500ºC interval were used for each specimen, so that, in the end, to find a value that will result in a temperature curve as close as possible to the experimentally determined one.
Simulations results that are best fitting the measured behavior are presented in figure 6.It can be observed that there is great correlation between results in every case.This was achieved for exposure temperatures listed in table 4. Variation of temperature values presented in table 4 can be explained either by errors in the airheating gun, but they can also represent the result of inhomogeneities in the mortar materials.The temperature was applied on a surface, while it was measured in a point.The material that this point was representing could have induced a lower or a higher speed of temperature modification.
The final process modeling all the specimens with the same applied temperature to them (denoted as fcs -for comparison simulations).The models were the identical to those used for the previous simulations, with only the applied temperature being changed to 500ºC in all cases.Thus, we sought an accurate comparison between the specimens.Time of exposure was increased to 15 minutes in order to highlight differences in curves of similar slopes.
As it can be observed from figure 7a), the AZ specimens show a behavior similar to the one captured by experiment, in both the relation between them, as well as the one with the reference.Even if the calibrating temperatures were quite different, the relationship was not fundamentally changed.This might be due to the big differences among the recipes as far as the thermal properties are concerned.
The PZ recipes (figure 7b)) show a higher deviation from experiment.The order of the biggest temperature that is reached after 10 minutes exposure, between the replacement percentage, is the same.Nevertheless, the actual difference between them is altered.The PZ10 and PZ30 demonstrate similar performances and PZ20 is the least affected, in agreement with the experiment.The simulations demonstrated that the reference is the most impacted by the exposure, when compared to PZ, contrasting the experiment.The thermal parameters of PZ10 and PZ30 were found to be more closely related to each other and also to the reference, thus a change in calibrating temperature is shown to be important.

Conclusions
The study focused on demonstrating the viability of using a free simulation software for assessing the performance of small-scale specimens when exposed to high temperature.By coupling the modeling results with experimental data, we showed that a less precise setup gives enough information about the process, so that simulation can be further used for an accurate characterization of involved materials.This type of analysis should prove useful as a first step in the researching of new materials and their thermal load performance.Bridging the gap between testing small-scale and large-scale specimens would be beneficial with respect to many types of involved resources.It is noteworthy that such simulations, based on only one specimen from a particular recipe, can be of great importance given that inhomogeneities are not influencing the result.With respect to the materials used in the study, it can be concluded that zeolites, particularly in aggregate form, represent a valuable replacement in increasing high-temperature performance.The best result indicated a decrease of almost 30% in the temperature achieved after 15 minutes of exposure.

Figure 3 .
Figure 3. Model of the specimen with heat source.

5Figure 4 .
Figure 4. a) Thermal conductivity and b) specific heat variations of mortar specimens.

Figure 5 .
Figure 5. Experimental results for a) AZ and b) PZ high-temperature exposure.

Figure 7 .
Figure 7. Same temperature exposure (500ºC) simulations of a) AZ and b) PZ for 15 minutes.

Table 1 .
Mix proportions for 500g of cement.

Table 2 .
Chemical composition of the clinoptilolite.

Table 3 .
Measurement values of parameters used for simulations.

Table 4 .
Temperature values of calibrated simulations.

Figure 6 .
Experiment vs. simulation at calibrated temperatures.