Alkali-reactive and inert Fillers in Concrete. Rheology of fresh Mixtures and expansive Reactions

• Main Author: Pedersen, Bård Magne
• Format: Doctoral Thesis
• Language: English
• Published: Norges teknisk-naturvitenskapelige universitet, Institutt for konstruksjonsteknikk 2004
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-231; http://nbn-resolving.de/urn:isbn:82-471-6401-9

Due to the rather limited obtainable resources of natural aggregates suitable for concrete purposes, the technology of crushed aggregates becomes more important. The production of crushed aggregates generates large amounts of fines of fillers, presently to some extent considered to be leftovers. From an environmental point of view, as well as from an economic point of view, it is important to be able to utilize these fines. Because of the large surface area of the filler fraction of the aggregate, the addition of filler may modify the rheological properties of fresh concrete to a great extent. Recently, the need for good fillers has increased due to the development of self-compacting concrete. To attain the much higher flowability of self-compacting concrete compared to ordinary vibrated concrete, the volume of the fluent phase, i.e. the matrix phase, has to be increased. At the same time, the self-compacting concrete has to be stable. The addition of filler may then be beneficial from a technical point of view. In addition, the use of fillers may be more cost efficient than other possible solutions. There are examples in the literature that fillers may modify the properties of the hardened state as well as the properties of the fresh state of concrete. Fillers have been reported to accelerate the cement hydration in some cases. Examples of increased compressive strength also exist. This is believed to be due to a general filler effect, i.e. that the cement hydration products may grow faster and become more evenly distributed in the presence of small mineral particles. In addition to the general filler effect, there might be chemical effects, in some cases pozzolanic reactions. The present study has focused on filler classified as alkali-reactive. The alkali-silica reaction in concrete is known to result in cracking and overall expansion of structural elements. There are some examples in the literature indicating that the finest particles of alkali-reactive aggregates should not be considered dangerous in concrete. Some researchers have reported that filler particles below a critical limit, which has been reported to be in order of 50 μm for some rocks, may give pozzolanic reactions, and consequently be beneficial. However, there have been reported cases where particles smaller than 20-30 μm gave very fast and deleterious reactions. In the present study, alkali-reactive fillers from two Norwegian cataclastic rocks have been investigated. The study has included fillers of Icelandic glassy rhyolite and crushed bottle glass. Non-reactive reference fillers were included in the study, as well as silica fume and fly ash known to mitigate alkali-silica reactions. The direct pozzolanic reactivity of the fillers has been quantifies by mixing calcium hydroxide, filler and artificial pore water. The loss of calcium hydroxide over time measured by TGA is then a direct measure of the pozzolanic reactivity. When testing the 0-20 μm fractions of the different fillers at 20°C, the materials could be divided into two distinct classes with respect to pozzolanity: • The pozzolanic reactivity of fly ash, glass and rhyolite filler was distinct •The pozzolanic reactivity of mylonite, cataclasite and quartz fillers was insignificant at the age of 28 days All the materials being highly pozzolanic have a distinct amorphous phase, while the silica phase of the non-pozzolanic materials is well crystalline quartz. The known deformation and sub-grain development due to cataclasis of the tested reactive Norwegian rocks does not seem to increase the pozzolanic reactivity much. The mylonite filler has also been tested at curing temperatures of 38°C and 80°C. The pozzolanic reactivity was very low also at 38°C. However, at a temperature of 80°C, corresponding to the temperature used by the accelerated mortar bar test, the pozzolanic reactivity was significant. Non-reactive granite/gneiss filler of glacioflucial origin was also pozzolanic at 80°C. It may then be suggested that all fillers of rocks containing silica will be more or less pozzolanic as such high temperatures as 80°C. Based on testing by the concrete prism method, the fillers could be divided into two distinct classed with respect to their effect on alkali-silica reactions in concrete: • Fly ash, silica fume, glass filler and rhyolite filler significantly reduced the expansions compared to the reference concrete • Mylonite and cataclasite filler had no effect or gave increased expansions compared to the reference concrete These results are based on experiments by the concrete prism test, which is believed to provide a realistic picture of the real behaviour in field conditions. Micro structural analyses, using optical microscopy and electron probe micro analyser, have given additional information regarding the performance of glass filler, mylonite filler and rhyolite filler compared to the reference concrete, and confirmed the expansion results of the concrete prism test. The effect of the tested fillers with respect to alkali-silica reactions matched their pozzolanic reactivity. Fillers being highly pozzolanic reduced the expansions due to ASR, while non-pozzolanic alkali-reactive fillers gave in most cases increased expansion compared to the reference mix. Consequently, such alkali-reactive fillers should be treated as potentially deleterious in concrete. The accelerated mortar bar test has been widely used around the world to test the effect of different additives, such as silica fume, fly ash and slag. Other additional materials, such as crushed bottle glass, have also been tested using this method. Some studies have indicated a rather strong correlation between the results obtained by concrete prism testing and results obtained by accelerated mortar bar testing. In the present study, extensive testing of fillers by the accelerated mortar bar test was carried out to give a preliminary screening of the materials. Testing of rhyolite filler, glass filler, fly ash and silica fume reduced the expansions significantly compared to the reference mortar when tested by the accelerated mortar bar test. This is in accordance with the results obtained for the same fillers by the concrete prism test. However, the accelerated mortar bar test also predicted that Norwegian reactive rock fillers to inhibit expansions due to alkali-silica reactions. This contradicts the predicted effect of these fillers by the concrete prism test. Testing of non-reactive limestone fillers gave no effect at all on mortar bar expansion. This indicates that the effects of the Norwegian reactive rock fillers by this method is due to chemical, and not physical effects. Due to the high temperature used by the accelerated mortar bar test (80°), the quartz in these rock fillers are believed to react pozzolanic. Methods such as the accelerated mortar bar test, or other methods using very high temperatures, should consequently not be used to evaluate the effect of rock fillers containing silica, unless their pozzolanic reactivity are evident also at lower temperatures. The pozzolanic materials (fly ash, silica fume, rhyolite and glass filler) gave a significant increase in compressive strength. This is believed to be due to their pozzolanic reactivity. Ni significant effect on the compressive strength of any of the Norwegian rock fillers (mylonite, cataclasite and granite/gneiss filler) was notified at normal filler additional levels. The present study has given valuable information concerning the practical implications of using alkali-reactive fillers. The similarity between the alkali-silica reaction and the pozzolanic reaction has been highlighted. However, some of the more fundamental issues concerning the paradox of the alkali-silica reaction and the pozzolanic reaction are still far from being fully understood, and it is clear that more basic research is needed in this area. Testing of the effects of fillers on the rheological properties of fresh concrete was done by matrix testing within the present study. The matrix refers to the fluent phase of the particle-matrix model. It consists of the cement paste and all powders of particle size < 0.125 mm, including the aggregate filler. Some of the limitations of the particle-matrix model with respect to self-compacting concrete have been treated in the present study. The characterisation of the matrix phase by simple flow viscometers are believed to be insensitive to the small, but significant, changes in yield stress. A more fundamental characterisation of the matrix phase has been introduced. By using the Physica rheometer with parallel plate configuration, the fundamental measures yield stress and plastic viscosity could be obtained. The effect of fillers on the flow resistance ratio of the matrix has benn tested. As expected, addition of filler increased the flow resistance ratio. The effect of the different fillers varied much, to a large extent due to the variations in particle size distribution. The granite/gneiss filler, which is coarse compared to the crushed rock fillers, gave the lowest flow resistance ratio. Fly ash, which has a particle size distribution similar to cement, gave the highest flow resistance ratio. Replacement of cement by reduced the flow resistance ratio for most of the fillers. The exception was fly ash and glass filler, the mineralogy seems to have some influence. In this respect, limestone filler gave rather low flow resistance ratios in relation to its fine particle size distribution. A laboratory program using the Physica rheometer to give a more fundamental characterisation of the effects of filler on the matrix has been carried out. The plastic viscosity obtained from testing by the Physica rheometer is more or less an equivalent measure to the empirical flow resistance ratio obtained by the FlowCyl testing. Consequently, the effect of the fillers with respect to plastic viscosity was basically equal to their effect on flow resistance ratio. The replacement of cement by filler has been shown to alter the rheological properties significantly. Generally, fillers have lower yield stress and plastic viscosity than equal volumes of cement. An increase in plastic viscosity was generally followed by an increase in yield stress. However, it has been shown that the ratio between yield stress and plastic viscosity is highly dependent on the type and dosage of plasticizer. The new co.polymers generally reduced the yield stress to a much higher extent than lignosulphonate or naphthalene. Further, the co-polymer seems to level out the large difference due to different fillers, which may be apparent when using other types of plasticizers. Also the cement type has been shown to influence the ratio between yield stress ad plastic viscosity largely. A study of the relationship between the rheological properties of the matrix phase and self-compacting concrete has indicated that the yield stress of the matrix phase has a crucial influence on the empirical slump-flow measure. No straightforward correlation between the rheological properties of self-compacting concrete and the corresponding matrix phase was found. Studies of rheological properties of the matrix phase should be considered useful to gain fundamental knowledge regarding the effects of different constituents. Matrix testing may to some extent be useful to predict the effects in self-compacting concrete, but the limitations should be kept in mind when using matrix results to predict the behaviour of a given constituent in concrete. The study had confirmed the basic principles of the particle-matrix model for self-compacting concrete. However, further work is needed to go deeper into the very complex relationship between the concrete rheology and the matrix rheology.