1 INTRODUCTION
European countries, such as the Netherlands, have a very dense motorways network. Construction of new motorways within the Netherlands has been largely completed (Xing 2004), so rehabilitation of existing pavements has become the primary task for the highway industry. In this regard the full-depth reclamation with Secondary Materials (SM), whether or not in conjunction with self-cementing materials, is one of the most economical ways for rehabilitation (Rens 2005). The scarcity of suitable naturally-occurring aggregates has forced practice to utilize industrial waste products but their use may pose problems. In this connection, utilization of by-products like Blast Furnace Slag (BFS), as ingredients for road construction, needs special attention.
BFS can be considered as a rather heterogeneous material with complex features. It essentially consists of glass with crystalline silicates and aluminosilicates of calcium. Rapid chilling results in preponderance of a glassy (amorphous) phase in the slag (Bohbc 2009). In studying the BFS, the gain of strength was attributed to the hydration at later ages which means the rate of hydration of BFS is initially lower than that of ordinary cement (Menendez 2003). Although BFS base courses are successfully used in the Netherlands for quite some years, unexpected poor structural pavement behavior suddenly occurred on a particular highway which was related to unexpected collapse of the BFS base course.
Characterization of Blast Furnace Slag to be used as Road Base
Material
S. Akbarnejad & O. Copuroglu & L.J.M. Houben & A.A.A. Molenaar
Delft University of TechnologyFaculty Civil Engineering and Geosciences Delft, The Netherlands
ABSTRACT: In congested areas around the world, traffic has significantly grown beyond expectation both in terms of volume and weight. Any hinder to the traffic causes severe delays resulting not only in economic loss but also in extra pollution of the environment. Therefore, maintenance works are desired to be reduced as much as possible. Application of self-cementing materials such as Blast Furnace Slag (BFS) for base courses is one of the methods to reduce the need for maintenance, since such base courses can provide a significant increase of the stiffness and the strength of the pavement structure. However, this type of stabilization appears to be associated with undesired deformations and distresses such as heaves formation and cracking that occurred, e.g. on the A32 motorway in the Netherlands. Different failure mechanisms have been hypothesized. This means, the use of BFS in a road (sub-)base layer requires a good knowledge of its characteristics. This paper presents data on the chemical and mineralogical characteristics of fresh and field aged BFS materials from a case study in the Netherlands. Furthermore a microstructural study was done on samples which have experienced freezing and thawing cycles.
An example of such unexpected behavior was observed on a 10 km long stretch of the motorway A32 (constructed in 1986-1988).
The original structure of the A32 motorway pavement structure consisted of 190 mm asphalt (dense asphalt concrete), 200 mm mixture of slags (granulated BFS sand, steel slag and air-cooled BFS) as a base and a sand sub-base (varying in thickness) on the natural clay subgrade (Houben 2008).
Some 10 years after construction of the pavement, the first transversal heaves occurred at the pavement surface causing unacceptable pavement roughness. Although these initial heaves were milled off and the riding quality was improved, such heaves continued to develop causing regular maintenance operations. Then, in 2008 and 2009 the pavement had to be fully reconstructed over a length of nearly 10 km because of sudden collapse that could be attributed to the BSF base course.
One of the incentives for the present study therefore was to investigate self-cementing secondary materials to see whether the hardening characteristics (hydration reaction) are related to the observed road deformation. Another incentive was from the aspect of quality of BFS materials. It is now widely accepted that secondary materials offer an environment friendly and economically favorable alternative to natural aggregates for use in new pavements. Use of BFS in a motorway applications frequently requires the pre-assessment of future performance, which is often difficult to predict. Therefore, there is an absolute need to explore this type of material in order to be able to estimate its long-term performance.
The first step of exploring the long-term performance of BFS in a highway application requires a better understanding of the microstructural features of BFS when used as (sub-)base material and to discuss its possible effects on pavement performance. Consequently, the influence of freezing conditions, which can be mimicked by means of Freezing Thawing (FT) cycles in the laboratory, as a destructive action on the microstructure of this type of material was investigated.
2 MATERIALS AND METHODS 2.1 Materials
In order to ensure that within this study a range of potential failure mechanisms would be covered, fresh material consisting of air-cooled blast furnace slag, steel slag and granulated blast furnace slag (GBFS) sand was obtained from a well-known producer. The selected materials were clean and free from detrimental levels of chemical impurities and harmful constituents.
Also actual field aged material was needed and based on previous research (Houben 2008), it was decided to collect actual field samples from the base layer of the A32 motorway near Wolvega, in the province of Friesland in the north of the Netherlands. As mentioned above, the A32 motorway experienced numerous heaves formation and finally complete failure. The role of BFS material in this failure was not fully clear. Thus it was arranged (with the Ministry of Transport, Public Works and Water Management) to collect base materials at different locations of the A32. Several slabs (maximum size was ~1×1.5 m) and about 50 kg loose granular materials were collected from the Eastern carriageway during its complete reconstruction. In general, the field aged BFS was found to be coarse, porous and rough. The particle size distribution curves of the A32 and the fresh materials are presented in Figure 1.
In order to obtain a homogeneous and representative sample of the aggregate population, a quartering method was used as suggested by Goodsall (1970). The samples of 50 kg of each type of aggregate (fresh and crushed field aged) were reduced to 5 kg samples using this method.
The physical properties of the materials are given in Table 1. The apparent relative density values are determined after 3 trials with a gas pycnometer device. The tests are done according to the ASTM D5550 procedure. The water absorption of the aggregates was determined based on the principle of water saturation and in accordance to EN1097-6.
Figure 1. Grading curves of the materials used in this study. Table 1. Physical properties of materials.
Apparent relative density
(103 kg/m3) Water absorption (%)
Air cooled BFS 2.654 2.985
Steel slag 2.899 4.137
GBFS sand 2.901 4.867
A32 material 2.827 3.824
2.2 Experimental details of microstructural evaluation
The available materials were examined through ray diffractometry (XRD) techniques and X-ray florescence (XRF) spectrometry to establish their chemical as well as mineralogical composition.
Samples were grinded into very fine powder. Then they were sieved to obtain particles smaller than 150 µm for further analyses. The XRD analysis was performed using a Siemens D5005 diffractometer, with an incident beam monochromator and a position sensitive detector, allowing determination of the mineralogical phases within the constituents. Diffraction patterns were recorded in the 2θ angular range 5-70°, with 5 s/step. For the XRD analysis, a search/match procedure was used to identify the crystalline phases in the diffraction patterns. It uses the database from the International Centre for Diffraction Data to compare the sample diffraction patterns with patterns of pure phases.
Subsequently the major elements (Si, Ti, Al, Fe, Mn, Mg, Ca, Na and K in % by mass) were determined by means of XRF using a Philips PW2400 Rh 60kV spectrometer.
Additional cores were drilled from the acquired field slabs (in addition to the collected granulated materials) and consequently standard thin sections from epoxy impregnated samples were prepared. Standard petrographic thin sections were studied with a Leica DMRXP polarized light microscope (PLM) equipped with excitation and blocking filters for fluorescence microscopy. Photomicrographs were acquired with a Leica DFC425 digital microscope camera at 2592×1944 interpolated resolution and exported as JPG format.
The optical investigation was performed according to ASTM C856. The microscopic analysis was performed by means of PLM on standard petrographic thin sections prepared from the cores which were drilled at different locations. The cores (100 mm diameter) were taken with the lowest possible amount of cooling liquid to avoid any undesirable damage. The cores without any visible cracking were used for thin section studies.
In order to prepare a thin section, a small block was sawn in the longitudinal direction (Figure 2) from each core, glued to an object glass, and vacuum dried and impregnated with an epoxy resin containing a fluorescent dye (Epodye from Struers). After hardening, one thin section with
a surface area of approximately 30 × 50 mm and a thickness of ~ 30 µm was prepared from each block for the analysis.
2.3 Freezing and Thawing (FT) Action
The porous structure of BFS has raised questions about the performance of this type of material under freezing and thawing action.
To investigate the effect of freezing and thawing, cylindrical samples of fresh material having a diameter of 100 mm and a height of 180 mm were produced. The specimens were prepared according to the method proposed for making triaxial samples (Van Niekerk 2002) and accordingly the three types of slags were mixed to obtain a Fuller curve gradation using a power of 0.45 with a maximum grain size of 22.4 mm. This mixture consisted of 86% air-cooled BFS, 9% steel slag and 5% GBFS sand.
Metal base and top plates were used at each end of the sample for the purpose of sealing (keeping optimum moisture content constant which is about 7.5%). Before the FT test, the sealed samples were stored for 28 days in a temperature controlled room (22 + 2°C). During this period the weight of the samples was measured in order to control the quality of sealing and avoid moisture loss.
The samples were preheated for 5 hours to ensure that the whole sample (including its core) had a temperature of 30°C (starting air temperature).
During the FT cycles the specimens were wrapped in two Polychloroprene membranes (high resistance to temperature variation) in order to ensure that the moisture content stayed constant during the cycles. All FT samples were made at optimum moisture content.
The freezing and thawing procedure was following the RILEM TC 176 recommendations and each test specimen was cooled from 30 to -10 °C in air with a cooling and warming rate of 4.0 °C/h (rate is meant the temperature change of air in the climate chamber). The cooling and warming procedure was repeated for 4, 8, 12 and 16 cycles. The number of FT cycles was selected based on the Fagerlund (1997) study and it was assumed that for this type of material a maximum of 16 FT cycles is sufficient to possibly cause cracking and damage. Reference samples were also made in order to be compared with the other samples which were to undergo FT cycles.
For optical microscopic studies on the specimens which have experienced the FT cycles and on the reference samples, sampling was done by cutting slices from the middle portion of the specimen. The observation area was arranged to be parallel to the side surface (perpendicular to the compaction surface, Figure 2) with a depth of ~ 5 mm. Two plane sections with dimensions of 20×40×15 mm were prepared for each sample in accordance with the procedure developed by Gran (1995). The optical microscopy investigation was performed mainly according to ASTM C856. The microscopic analysis was performed by means of PLM on standard petrographic thin sections prepared from different specimens.
Figure 2. Thin sections orientation was perpendicular to the compaction surface of the sample.
3 RESULTS AND DISCUSSION 3.1 Mineralogy and chemical composition
The chemical compositions of the fresh and field aged BFS samples were obtained using XRF analysis. The composition of both types of material is plotted as a ternary phase diagram for the system CaO-Al2O3-SiO2 and SiO2-Al2O3-MgO (Figure 3). All materials were found to be at the corner of the triangle, revealing an almost pure CaO composition. Additionally it shows a rich
composition in SiO2. Also in one aged sample (A32-chainage 33.38 km) a relatively richer composition in CaO was detected when compared to the other samples.
The XRF results showed that for both types of samples, there was no significant variation in oxide contents in comparison with recommended limits (EN 197–1). Given this, a further detailed analysis was undertaken in order to investigate the mineral composition of the fresh and field aged BFS. Chemical analysis of the data obtained for the fresh material (Figure 4) shows a predominance of Akermanite peaks, followed by SiO2 peak. In further studies CaCO3 was also detected that it is expected to be a product of the hydration of free lime which subsequently carbonates.
Figures 4 presents the XRD patterns of the fresh and aged BFS. The only major mineralogical distinction between the fresh and aged samples was identified as hydration products such as ettringite (Figure 4).
Figure 5 shows a variation in the chemical and mineralogical composition of the aged slag in different locations. This variation took place mainly due to hydration reactions (different concentration of ettringite). Peak identification for the aged materials was complicated because of the overlapping of phases.
Figure 4. XRD patterns of aged sample (A32-chainage km 33.36 ) and fresh BFS aggregates.
Subsequently, the XRD technique provided a semi-quantitative analysis. Semi-quantitative analysis was done with the aid of the reference intensity ratio method. X-ray diffraction analysis was done on three samples and the average concentration of each phase is presented in Table 2. Table 2 shows there is a certain decrease of hydration products including ettringite near failed spots on the A32 motorway.
Table 2. Mean concentration of the main phases identified in the field aged BFS materials (in % by mass).
Ak erman ite Geh len ite Ettrin gite Th au masite C-S-H Details
Location A 53 41 6 - - A32 Chainage km 33.36 failure spot)(sample close to a Location B 41 47 12 - - A32 Chainage km 33.37 Location C 27 22 25 20 6 A32 Chainage km 33.38
3.2 Macroscopic evaluation
The slag particles and their hydrated phases present in the field aged slabs were studied by optical microscopy. The microscopy images show that the investigated region consists of unhydrated crystalline slag particles, slag hydration rims, C–S–H products with ettringite phases and fine sands (Figure 6). No thaumasite was observed.
The hydration products of granulated blast furnace slag are formed at interfaces around slag particles and inside particles as well. The optical images indicated a presence of potentially reactive aggregates and crystalline grains. Figure 6 shows the hydration products of slag form as rims around the slag particles. The rims have formed inward and outward of the slag particles (Figure 6). These rims of hydration are often segmented in two or more unknown layers.
Further optical photomicrographs show the total observed hydration phases could be divided into more than one hydration zone. It was also identified that some parts of the original BFS have never been hydrated (after approximately 22 years).
Figure 6. Typical photomicrographs of a BFS mixture exhibiting different zones (in plane polarized). Image width [↔ L=0.35 mm]. A: crystalline slag particles.
The presence of ettringite was investigated using both X-ray diffraction and optical microscopy since there was the initial hypothesis that the volume changes that happened in the BFS base of the A32 motorway were the result of massive ettringite formation.
From a chemical point of view, the mechanism of ettringite formation is well understood. Ettringite [Ca6[Al(OH)6]2.(SO4)3.26H2O] precipitates in environments with high pH and its stability in the system is controlled by pH, temperature, ion activities and concentrations of dissolved CO2 and H2O (Little 2005).
A
A
A
A definitive relationship is not yet established between the amount of ettringite precipitated and the associated volume changes in the matrix. Expansion may vary among pavement layers due to differences in construction methods, availability of water, ion migration and the void structure within the pavement material (Little 2005). In this study, regardless of the mechanism of expansion associated with ettringite formation, the objective was to see in the field aged materials whether the ettringite is formed in massive scale or not.
Ettringite phases formed radial needle-like crystals (Figure 7). Their size and shape as well as other optical properties such as characteristic birefringence and parallel extinction made it possible to distinguish them among other phases. Conspicuous formation of ettringite was observed mainly in voids. The amount of damage that may be caused by ettringite formation depends on two main factors a) the strength of the matrix and b) the spatial arrangement of the ettringite crystals. In this case ettringite mainly grows in voids and it seems that it managed to accommodate its growth without any substantial micro-cracking around it. Additionally, there is no sign of ettringite crystals having grown within the dense phase. After carrying out optical microscopy investigations on different samples it became clear that the ettringite formation in hardened BFS materials cannot be the dominant damage mechanism.
Figure 7. Photomicrographs of the formation of ettringite (in plane polarized). Image width [↔ L=0.35 mm]. A: crystalline slag particle. B: Ettringite crystals, right image shows needle-like ettringite crystals with a higher magnification.
Additionally a porous structure and pores communicating with each other through channels have been observed. These interconnected channels transport water which can penetrate into voids (Figure 8). The process of water penetration may be accompanied by volume changes as a result of freeze-thaw action. Thus, in order to prevent such volume changes the BFS material should not get saturated.
Figure 8. Interconnected channels and voids within BFS system. Left image in in plane polarized and right one in fluorescence light (in black & white mode). A: crystalline slag particle. Image width [↔ L=2.87 mm].
As mentioned before the microscopic analysis was carried out to investigate the degradation of the BFS caused by FT test. FT cycles often result in different levels and configurations of cracks.
B
B
B
A
A
In this part of the study, the resin impregnation technique in combination with the optical fluorescence microscopy and a computer image analysis technique were used to provide qualitative and quantitative determination of the crack system. The observation was carried out by means of an optical microscope at a magnification of 10× in order to detect fine cracks. On the generated photos a quantitative analysis was performed.
Two levels of damage were observed after the FT test. At the macroscopic level, large cracks were identified on the surface of the samples. In some cases, especially after 8 cycles, the
samples were seriously damaged. On the microstructural level, the cracks were mainly observed
in the samples that were subjected to more than 8 cycles. The crack formations were classified
with the cracks density, Dcr, which was estimated by the point counting method (ASTM C457-Procedure A).
In the point counting method, a two-dimensional grid was used and each crack which falls under a grid point was counted. 200 points for each sample were analyzed and the crack density assessment took into account cracks in the slag particles as well as in the paste area. Dcr is defined as follows:
D
cr=
Cracks pointsTotal points×100% (1)
The first step of this part was to quantified the cracks of a sample (reference sample) before exposure to the FT action. As it is shown in Figure 9 for a sample made with fresh BFS
material, the higher the number of FT cycles, the higher the crack density was. This figure
gives the crack characteristics of the samples sawn from the 100 mm cylinders subjected to the different number of FT cycles.
Figure 9. Estimation of crack density by means of the point counting method.
The BFS samples after 8 FT cycles showed a relatively high crack density. The average crack width was measured in different samples which had undergone the same number of FT cycles. The results are presented in Table 3 (in this part three samples were studied).
Table 3. Results of crack width measurement.
Number of FT cycles Crack width range (mm)
4 0.055-0.065 8 0.07-0.085 12 0.065-0.1
Figure 10 gives the crack characteristics of the samples sawn from a sample subjected to the different FT cycles. It appeared that the cracks (yellowish line) are randomly oriented and there is not a definite orientation of the cracks. The cracks mainly propagate through the paste (hydrated BFS aggregate) and around the boundary of slag aggregates.
In several samples generally two types of cracks were observed in the BFS mixtures after FT cycles (Figure 10), a) cohesive cracking, in which the texture of the mixture is disintegrated into several small cracks, b) adhesive cracking in which loss of bond between coarse slag particles and paste took place and peripheral cracks formed around the slag particles. The observed crack pattern can be due to the low strength (~ 4 MPa) and the highly porous nature of the BFS
0 5 10 15 Reference 4 8 12 16 C rack densi ty (%) Number of FT cycles
system. The distance between the cracks generally varies according to the size of the coarse aggregate which cracks pass.
Generally, the BFS system has a slow hydration rate in comparison with normal concrete so that in freezing conditions the hydration probably will be much slower, thus this type of material can be susceptible to freezing and volumetric expansion of water in its pore system.
More optical microcopy test results after FT cycles are presented in Figure 10. Cracks are present around coarse slag particles and through the paste (left image). Label A is crystalline slag particle and B is hydration products.
As shown in Figure 10 (middle and right), cracks are mainly parallel to the surface of slag particles. As can be seen clearly in the right figure cracking (yellowish line) in the mixture is close to the boundary between the dark (crystalline slag particle) and brownish zone (hydration products). No changes to the slag particles can be observed and the main damages were restricted to a series of cracks in the paste.
Further research showed that the observed damage of the samples was the result of hydraulic pressure caused by water migration through the porous system.
Figure 10. Photo of different BFS mixtures which have been exposed to FT cycles [in plane polarized ↔ L=2.87 mm].
4 CONCLUDING REMARKS
The chemical composition and microstructure of the available fresh BFS and field aged BFS (after ~ 22 years) and also the microstructure of BFS material under freezing condition were investigated. The major conclusions that follow from this study are as follows:
1) The A32 base layer mixture was originally designed as an unbound mixture but during service life, BFS materials have mostly transformed into a bound material and test results clearly showed that hydration reactions occurred.
2) Hydration products in BFS samples are observed a) at the slag grains and mixture interface and b) within slag grains.
3) The major failure on the A32 motorway was originated from the BFS layer and it was influenced by a combination of many factors. Microcracks in the BFS base layer were formed as a result of volume changes during hydration and furthermore FT cycles may have caused further deformations. The formation of cracks due to the hydration over a long period of time may have contributed to the failure and this process possibly was exacerbated by FT action.
4) Ettringite was mainly formed in cavities which suggests that ettringite formation is not a key factor for expansion and further damage.
5) The overall structure of this type of material is porous with a well-interconnected capillary structure, which makes them sensitive to harmful water ingression and thermal variation (FT actions). Both the crack density and the crack width were formed to increase with the number of FT cycles.
6) Temperature may be an important experimental variable for promoting the accelerated aging of BFS mixtures and the results show that mechanical and physical performance of BFS materials can be affected by temperature and time.
Further research into complex relations between self-cementing secondary material (e.g. slags) and environmental conditions which can cause a developmental and destructive actions is necessary. Their mechanical performance and physical integrity can be affected by the environment and can be changed by during the time.
B
A
A
5 REFERENCES
American Society for Testing and Materials. 1994. Standard Test Method for Specific Gravity of Soil Solids by Gas Pycnometer, ASTM Standard D5550 – 06, Philadelphia, USA.
American Society for Testing and Materials. 2011. Microscopical Determination of Parameters of the Air-Void System in Hardened Concrete, ASTM Standard C457/C457M – 11, Pennsylvania, USA. American Society for Testing and Materials.2004. Standard Practice for Petrographic Examination of
Hardened Concrete, ASTM Standard C856-04, Philadelphia, USA.
Bohbc M. and Miroslava G., 2009. The influence of blast-furnace slag hydration products on microcracking of concrete, Materials Characterization, volume 60, pp. 729-34.
Fagerlund, G. 1997. Internal frost attack – State of the art, Proc. Rilem Workshop on Resistance of
concrete to freezing and thawing with or without de-icing chemicals, Essen.
Goodsall G.D., Mathews D.H. 1970. Sampling of road surfacing materials. Journal of Applied Chemistry, pp. 361–366.
Gran Hans Chr. 1995. Fluorescent liquid replacement technique. A means of crack detection and water: Binder ratio determination in high strength concretes, Cement and Concrete Research, pp. 1063-74. Houben L.J.M. 2008. Long term mechanical behavior of hydraulic blast furnace slags road base, 7th
International Conference Sustainable aggregates, asphalt technology and pavement engineering (LJMU), Liverpool, 2008, pp. 1 - 10.
Little D.N., Herbert B. and Kunagalli S.N. 2005. Ettringite Formation in Lime-Treated Soils: Establishing thermodynamic foundations for engineering practice. Transportation Research Record, No. 1936, 2005, pp. 51-59.
Menendez G., Bonavetti V. and Irassar E. F. 2003. Strength development of ternary blended cement with limestone filler and blast furnace slag, Cement and concrete Composites, pp. 61-67.
Methods of testing cement, 2000. Part 1 : Composition, specifications and conformity criteria for common cements. EN 197–1, European Committee for Standardization (CEN): Brussels.
Rens L. 2005. Recycling by cement stabilization in Belgian road construction, 2nd
International symposium on treatment and recycling of materials (TREMTI), Paris, vol. 86, 2005, pp. 2148-2153.
Tests for mechanical and physical properties of aggregates, 2000. Part 6 : Determination of particle density and water absorption, EN 1097-6, European Committee for Standardization (CEN): Brussels. Xing W. 2004. Quality improvement of granular secondary building materials by separation and
cleaning techniques, Delft University of Technology press, ISBN 90-9017982-1, The Netherlands, pp.
