2014 Volume 54 Issue 3 Pages 704-713
The solidification conditions to reduce the porosity of air-cooled blast furnace slag were investigated. From cross-sectinal observation of solidified slag, growth of gas bubble generated in molten slag was estimated to be cause of high porosity. With low thermal conductivity slag, increasing the cooling rate by thin slag casting was effective for reducing the porosity of air-cooled blast furnace slag.
As a method of reducing the porosity of air-cooled blast furnace slag, a process was developed in which the slag is solidified to a plate thickness of 20–30 mm in about 2 minutes by pouring the molten slag in a cast steel mold. When porosity reduced, the abrasion resistance of the slag improved. The possibility of using low porosity slag as aggregate for drainage pavement was confirmed in an experiment with test pavement.
A Japanese Industrial Standard (JIS) for slag aggregate for concrete was established in 1977. However, use of coarse aggregate of air-cooled blast furnace slag in Japan is currently only about 270000 t/year, which is less than 1% of total demand for concrete aggregate in the whole country. As the reason, the water absorption of coarse aggregate made from air-cooled blast furnace slag is 2–4%.1) This is higher than the water absorption of natural aggregates such as macadam. Therefore, pre-wetting that keeps saturated water inside the aggregate is necessary before use in concrete.1,2) Moreover, when the slag is mixed into concrete as coarse aggregate, fresh concrete sometimes becomes clogged in the pipe during transportation by squeeze pump.3,4,5,6) These problems result from the porosity of air-cooled blast furnace slag. On the other hand, it has been confirmed that concrete using experimental blast furnace slag coarse aggregate, which has water absorption of 1.42%, or half that of ordinary slag, has higher compressive strength than concrete using river gravel or macadam as the coarse aggregate.7) Therefore, improvement of the quality of coarse aggregate of blast furnace slag can be expected by decreasing porosity.
Pore generation in air-cooled blast furnace slag is caused by SO2 gas generated by oxidization of sulfur in the slag by air,8,9) or by N2 gas generated by oxidization of dissolved nitrogen in the slag by air and water.10,11) Methods of producing low porosity, high density air-cooled blast furnace slag are the thin and many layer method,11) in which molten slag is poured on a dry sloping yard, and the method of inputting iron oxide into the molten slag.9,10,11) However, these methods are not widely used to produce coarse aggregate because the thin and many layer method requires a large cooling yard, and inputting iron oxide generates SO2 gas by oxidization of sulfur in the slag9) or disintegrates the slag.12)
The objective of the present study is to identify the solidification conditions for producing low porosity, dense blast furnace slag. The pores in air-cooled blast furnace slag, and the influence of slag thickness on water absorption and density were investigated. Based on the results and a heat transfer analysis, the authors proposed a new process for solidifying plate-shaped slag with a thickness of 20–30 mm using a cast steel mold to produce air-cooled blast furnace slag with low water absorption. A pilot plant was designed and constructed, the feasibility of manufacturing low porosity slag was confirmed by a practical-scale experiment, and aggregate was made from this slag and evaluated.
Although both high and low porosity parts exist in air-cooled slag, the process of pore generation during solidification had not been clarified. Moreover, no results of investigation of the pore size distribution and number of pores in air-cooled slag were found in the literature. Figure 1 shows a slag cooling yard. In the lower layer, the slag that is poured previously has little effect on temperature. The upper layer is cooled easily from the surface by radiation and water spraying, while the middle layer is the most difficult to cool, as slag exists on both the upper and lower sides. If it is possible to determine which part contains slag with high density and low water absorption, slag which is suitable for coarse aggregate can be sorted. Therefore, first of all, the height direction distribution in the cooling yard of the density in the absolutely dry condition and the water absorption of the air-cooled blast furnace slag was investigated.
Sampling point in slag cooling yard.
In this study, air-cooled slag manufactured by the cooling yard method was investigated. At the yard studied here, the blast furnace generated 300–400 t of molten slag in one tap. The molten slag was transported to the cooling yard by slag ladles, which hold about 55 t of slag per charge. The slag produced in one tap was transported by 2 or 3 ladles in 3 or 4 trips per ladle. Ladle transportation required about 20–30 min from the blast furnace to the cooling yard. Only molten slag flowed into the cooling yard from the ladle. The temperature of the molten slag was 1643–1673 K at pouring from the ladle, and the slag was solidified in stacked layers in the cooling yard. One cycle was about 3–6 hours. The cooling yard, which was about 2 m in height and had a capacity of approximately 4000 t, was filled in 1.5–3 days. After filling, the slag was cooled for 6 hours without water, and then cooled for 24 hours or more by water spraying. After water cooling and crushing with heavy machinery, lump slag specimens were collected from the upper, middle, and lower levels, as shown in Fig. 1. Lump specimens were collected from five cooling yards.
The specimens were crushed to 40 mm or less, and water absorption and density in the absolutely dry condition were measured. Water absorption and density in the absolutely dry condition were measured in accordance with JIS A 1110-2006, Methods of tests for density and water absorption of coarse aggregates.
Next, slag samples, which were collected by boring from the top surface of the solidified slag in the cooling yard, were examined for porosity and pore size distribution in the depth direction. Boring was performed at three locations in the cooling yard. The boring diameter was 100 mmΦ and the boring depths were 400–800 mm.
In general, pore size distribution is measured with a mercury porosimeter, but this method cannot measure closed pores and large pores with a size of several millimeters, which are frequently seen in slag. When slag aggregate is compared with natural aggregate, the water absorption of the slag aggregate is high, and the workability of the fresh concrete is low. This is estimated to be due to the large influence of the coarse pores in air-cooled slag. Therefore, the pore size distribution was measured by image analysis.
After measuring water absorption, slag specimens were cut out in cubes of about 10 mm. These specimens were embedded in resin and polished, and their cross-sectional photographs were taken with the optical microscope. Figure 2 shows an example of a cross-sectional photograph. This photograph was processed by computer-binarization to distinguish pores from slag-matrix. The number of pores, pore area, and pore diameter were measured from the binarized image by image analysis.
Cross-sectional photograph for image analysis. (Online version in color.)
Because the thermal conductivity of slag is low, the slag thickness is considered to have a large influence on the cooling rate. Therefore, the relationship between the slag thickness and the cooling rate, density in the absolutely dry condition, and water absorption were investigated with molten slag which was generated from a blast furnace.
As shown in Fig. 3, an inclined plate 1.5 m in width and 3 m in length with a slope of 5° was prepared using a SS400 steel plate 9 mm in thickness and set up on a frame. When molten slag was poured on the steel plate, it was observed that the moisture in rust evaporated and generated pores. Hence, a SUS304 plate 6 mm in thickness was pasted on the steel plate of the slope.
Schematic diagram of slag solidification test on slope.
Molten slag generated from the blast furnace was poured on the slope plate from a slag ladle. The temperature of the molten slag flowing from the slag ladle was measured with a radiation thermometer. The slag temperature on the slope was measured with a K type sheathed thermocouple that was inserted in a steel pipe of 15 A so that the tip protruded 10 mm from the pipe. After pouring the molten slag, the steel pipe was set so that the thermocouple tip was in the solidified slag, as shown in Fig. 3.
The solidified slag was thin on the upper part of the slope and became thicker toward the bottom of the slope. Samples of solidified slag weighing several kg were collected from positions with different thicknesses on the slope, and the density in the absolutely dry condition and water absorption of the samples were measured.
To determine the specification of the proposed slag solidification equipment, a laboratory-scale experimental apparatus for solidifying plate-shaped slag was prepared, as shown in Fig. 4. This apparatus comprises a mold-holding table, which can be rotated 180° around the axis by air drive, the mold, a solidified slag recovery box, and a rotation angle meter. This apparatus can reproduce the situation of one mold of the slag solidification equipment. The purposes of the experiment with this apparatus were to confirm the conditions for slag solidification, separability of the slag from the mold, and the mold temperature rise resulting from continuous processing.
Schematic diagram of simulation equipment for slag solidification process.
First, 10 kg of granulated blast furnace slag was melted at 1723 K in a graphite crucible under a N2 atmosphere by an induction furnace. After the molten slag was poured into the mold, the mold-holding table was rotated after the specified time, and the solidified slag was dropped from the mold. The angle at which the slag began to fall from the mold was read from the angle meter by video. The mold was then returned to its original position, and molten slag pouring and mold reversing were repeated three times. The thickness of the slag samples was measured, and the outflow of molten slag from the fracture face was confirmed.
The slag thickness, mold material, and mold initial temperature affect the cooling rate in the slag. The temperatures in the slag interior and at the surface and the back of the mold were also measured with this apparatus to understand their influences. Figure 5 shows the positions where the thermocouples were set. 3.5 kg of granulated blast furnace slag was melted at 1723 K, after which the molten slag was poured into the mold. During cooling of the slag, the interior temperature in the slag was measured with a R type sheathed thermocouple, and the mold temperatures on the back and on the slag contact face were measured with K type thermocouples. The surface temperature of the slag was measured with the radiation thermometer. The emissivity of the slag was assumed to be 0.92. When measuring the temperature in the slag, the slag was poured only once. Two types of mold materials were used, these being SS400 and SUS304 which have different thermal conductivities. In the experiment to confirm the influence of the mold initial temperature, the mold was preheated with a burner.
Method for measurement of slag and mold temperature.
Based on the results of the laboratory-scale experiments described in 2.3, pilot-scale equipment for plate-shaped slag solidification was constructed. The pilot experimental equipment comprised three molds 2 m in width × 1 m in length × 20 mm in depth. The outline of the pilot experimental equipment is shown in Fig. 6, and the equipment specification is shown in Table 1. This equipment can drop the solidified slag off the mold by horizontally moving and reversing the mold. Assuming practical use, molten slag can be poured into the moving molds in this equipment directly from a slag ladle.
Schematic diagram of pilot plant for slag solidification process.
An experiment was performed with the pilot experiment equipment by the following procedure. After the molten slag was poured into the moving molds, it was cooled for 60–120 sec without water, after which the solidified slag was dropped by rotating the molds.
The slag thickness was controlled by the mold movement speed and ladle rotation speed. After the slag was dropped, the molds were moved and cooled by water spray from a lower nozzle for 20–40 s. These processes were repeated a maximum of six times. Control of the slag thickness, cooling time, slag dropping conditions, conditions for reducing mold deformation, and product quality were confirmed by this experiment.
After the dropped slag was cooled without water spraying, it was collected the next day. The thickness and the apparent porosity of the specimen were measured.
Using this equipment, 40 t of plate-shaped slag was manufactured. After crushing the slag, aggregates for a pavement test were prepared. Density in the absolutely dry condition and water absorption were measured for each aggregate gradation. Abrasion loss was measured in accordance with JIS A 1121-2007, Method of test for resistance to abrasion of coarse aggregate by use of the Los Angeles machine.13)
Figure 7 shows the density and water absorption of the different layers in the cooling yard. Table 2 shows the average chemical composition of the slag samples. The densities and water absorptions varied widely. Water absorption displayed a range of 1.5% to 4% and an average of about 3%. The density of the upper layer was higher than that of the lower layer. The N classification of coarse aggregate in JISA5011-1, Slag aggregate for concrete – Part 1: Blast furnace slag aggregate, prescribes density in the absolutely dry condition of 2400 kg/m3 or more and water absorption of 4.0% or less. From the results of these measurements, air-cooled slag for coarse aggregate should be selected from the upper layer in the cooling yard. However, slag with water absorption of 1% or less, that is, on the same level as natural macadam, does not exist in the cooling yard.
Density (a) and water absorption (b) of different layers in slag yard.
Although both open pores and closed pores exist in air-cooled slag, only open pores influence water absorption. The relationship between all pores in air-cooled slag and water absorption was examined. Total porosity Pt [%], which is the sum of the open pores and closed pores, and maximum possible water absorption Qt [%], which assumes all pores are saturated with water, are expressed by the relationships in Eqs. (1) and (2) using true density Dt [kg/m3] and density in the absolutely dry condition Dd [kg/m3]. Here, ρw [kg/m3] is the density of water.
Figure 8 shows the relationship between the water absorption and density of air-cooled blast furnace slag. The straight line in this figure shows the maximum possible water absorption Qt, which is calculated from Eqs. (1) and (2). Water-absorbing pores were 35–65% of all pores, the average being about 50%. Because water absorption decreased with increasing density in the absolutely dry condition, increasing density by reducing the total porosity Pt is effective for reducing water absorption.
Relationship between water absorption and density in absolutely dry condition.
In the porosity of the boring samples, high porosity parts and low porosity parts existed at a cycle of several 100 mm, varying from 2% up to 30%. The results of measurements of the pore size of these boring specimens by image analysis are shown in Figs. 9 and 10. Figure 9 shows the relationship between the average pore diameter and the ratio of the pore total area in the measured sectional area. The average pore diameter increases as the ratio of the pore area increases. That is, there are many coarse pores in high porosity slag. The relationship between the number of pores and the average pore diameter is shown in Fig. 10. The number of pores decreased as the average pore diameter increased.
Relationship between pore diameter and pore area.
Relationship between pore number and pore diameter.
The following may be conjectured from Figs. 9 and 10: When gas is generated in slag, first, many small bubbles are generated. As the amount of evolved gas increases over time, the bubbles grow and unite, and as a result, the number of bubbles decreases. When the slag solidifies, these bubbles remain as pores in the slag. Therefore, quenching the molten slag, thereby suppressing growth and coalescence of bubbles due to gas evolution, is considered effective for reducing porosity.
Figure 11 shows the influence of the solidified slag thickness on density and water absorption in the experiment in which molten slag was poured and solidified on the sloping plate. Density increases and water absorption decreases as the slag thickness decreases. This is basically the same tendency as that reported by Futamura et al.,11) in which density in the absolutely dry condition of 2400 kg/m3 or more was obtained when the slag layer thickness was 60 mm or less. From the results of the present experiment, slag with density in the absolutely dry condition of 2650 kg/m3 or more and the water absorption of 1.5% or less could be obtained by reducing the slag thickness to 25 mm or less. Low porosity slag is not obtained in ordinary cooling yards. Figure 12 shows the relationship between the slag thickness and the initial cooling rate calculated from the solidified slag thickness and the measured temperature in the plate-shaped slag. The plots in this figure are cooling rates that were obtained from measurements of the temperature in the slag at 1–2 minutes after pouring. The respective lines show the cooling rates of the surface, center, and mold contact face and the average for the thickness direction calculated by an unsteady one-dimensional heat transfer analysis model, which will be described later. The cooling rate increases with decreasing slag thickness. The cooling rate calculated by the actual measured temperature was about 75 K/min at the thickness of 25 mm.
Dependency of dry slag density and water absorption on slag thickness.
Effect of slag solidification thickness on slag cooling rate. (Online version in color.)
To suppress growth and coalescence of bubbles due to gas generation, it is necessary to clarify the retention time for solidification of all the slag in the mold. Therefore, whether the slag was unsolidified or solidified was judged by the plate-shaped slag solidification experiment. That is, after falling from the mold, it was judged that molten slag that flowed from the fracture face was unsolidified, and the other slag was solidified. All slag in the mold solidified in a retention time of 90 sec when the thickness was 24 mm or less and in 120 sec when the thickness was 30 mm or less.
In this experiment, the procedure was to pour molten slag into the mold, solidify the slag, and then rotate the mold to drop the slag. When this operation was repeated three times, the slag did not fall from the mold even when the mold was rotated 180°. At that time, the back surface of the mold had been red heat. Figure 13 shows the relationship between the temperature of the mold back surface and the slag dropping angle from the mold. When the mold was reversed at a mold back temperature 925 K or less, the slag fell from the mold after rotation to an angle of 60–120°. However, when the mold back temperature exceeded 967 K, the slag did not fall from the mold even when inclined 180°. With passing time, the mold temperature decreased and the attached slag fell from the mold. Observation of the cross section of the plate-shaped slag after cooling revealed a glassy phase about 1 mm in thickness from mold contact face. When the slag does not drop from the mold, it is estimated that the mold contact face temperature of the slag is 1000 K or more. Adhesion of slag to the mold is attributed to the fact that this glassy layer is still in a softened condition.
Slag dropping condition from mold.
The influences of thermal conductivity and initial temperature of the mold on the cooling rate of slag were examined. Two kinds of materials, i.e., SUS304 and SS400, were used in the mold. The SUS304 mold was also used for the preheating experiment. Figure 14 shows the measured temperatures in the slag and the mold at 90 s after pouring the molten slag on mold. When the mold material is changed to one with a different thermal conductivity, the temperature inside the slag is almost the same, even though the temperature in the mold is different. The temperature in the slag was also virtually unchanged when the mold surface was preheated to 464 K, even though the mold back temperature increased about 200 K. Because the thermal conductivity of the slag is low, the mold material and the initial temperature of the mold have no influence on the cooling rate of the inner slag.
Temperature of slag and mold at 90 sec after pouring molten slag in mold.
In the pilot experiment, plate-shaped slag with an average thickness of 23.5 mm was obtained. To investigate the time necessary for solidification of the slag, the slag that fell from the mold to the floor was observed for evidence that the melt flowed from the fracture face and broke into two pieces on impact with the floor. Figure 15 shows the relationship between the slag thickness, the time from pouring into mold to falling, and the solidification of the slag. In this figure, the open circles (○) show fully solidified slag and the closed circles (●) show the unsolidified center of the slag. From Fig. 15, it can be stated that a cooling time on the mold of 90 sec is required when the slag thickness is 25 mm.
Solidification condition at pilot plant.
After pouring into the molds 4–6 times, the plate-shaped slag dropped from the mold, piled up, and kept a high temperature for a long time. Because the cooling rate of the surface slag on the piled slag was fast, a glassy part with a thickness of about 1 mm remained on the mold contact face side, and this was easily crushed by dropping from the height of about 1 m. On the other hand, the piled internal slag was slowly cooled from the red-hot state, and the glassy part changed to crystalline. No cracks occurred, and a hard material which gave a sound of porcelain when struck was obtained.
About 40 t of the plate-shaped slag produced by the pilot plant was crushed with an impact crusher, and aggregate was made from the crushed slag. Figure 16 shows the appearances of the plate-shaped slag and the crushed aggregate with a size of 13–5 mm. Unlike ordinary air-cooled blast furnace slag, few large pores exist in the crushed aggregate, and its appearance is similar to that of crushed stone.
Appearance of plate-like solidified slag (a) and crushed aggregate (b) made by pilot plant.
Table 3 summarizes the quality of the slag aggregate made by the pilot experiments in comparison with air-cooled slag and tight sand. The quality items are density in the absolutely dry condition, water absorption, and abrasion loss. As the grain size of the crushed aggregate increases, its density increases and water absorption decreases. This result is attributed to preferential reduction of the size of highly porous, weak parts by crushing. The water absorption of the aggregate with the grain diameter of 13–5 mm was 0.66%, and its abrasion loss was 15%. The quality of the slag aggregate produced by the pilot plant improved remarkably from that of conventional air-cooled blast furnace slag and was on the same level as that of the tight sand used as aggregate in asphalt concrete.
The cooling rate in the slag was examined by heat transfer analysis. Because slag is solidified in a plate shape in this process, the unsteady one-dimensional heat conduction of a simple plate during cooling can be applied to the temperature in the slag. The basic equation is Eq. (3).
As shown in Fig. 17, the temperatures of the slag and mold were calculated by dividing the slag into ten sections and dividing the mold into five. The heat transfer coefficient of the interface between the atmosphere and the slag surface hs [W/(m2·K)], the heat transfer coefficient of the interface between the atmosphere and the mold surface hm [W/(m2·K)], and the thermal resistance interface between the slag and the mold R [m2K/W] are given as parameters of this calculation model. The values of these three parameters were determined so as to obtain agreement between the actual measured temperature in the laboratory-scale experiments and the calculated results. Because the initial slag surface temperature is at least 1300 K, thermal radiation is considered. The influences of convection and pores inside of the slag are neglected. It is assumed that the atmospheric temperature Ta is constant at 293 K and Ta does not rise. The explicit calculation method is applied with assuming Δt=0.5 s.
Heat transfer calculation model of slag and mold temperature distribution.
Previous research on the thermal conductivity of slag has demonstrated that the value is small and displays a maximum at around 1400 K. Nagata and Goto measured the thermal conductivity of blast-furnace slag,14) but because that value is extremely small, being 0.2 W/mK at 1600 K or more, there is virtually no decrease in the temperature in the central part if this is used in the calculation, and the calculated result does not agree with the measured temperature. Therefore, in the present calculation, thermal conductivity λ [W/(m·K)] was calculated by the following Eqs. (4) and (5). Equation (4) is based on measurements of CaO–Al2O3–SiO2 system in the temperature region over 1400 K by Kan and Morita,15) and Eq. (5) is based on measurements of blast furnace slag in the region 1400 K and under by Nagata and Goto.14)
The specific heat of slag is based on the work by Ogino and Nishiwaki16) and is Cp=1.039 J/(g·K) at T<1443 K, Cp= 2.2425 J/(g·K) at 1443 K≦T<1673 K, and Cp=1.326 J/(g·K) at 1673 K≦T<1773 K.
Figure 18 shows the heat transfer calculation result of the solidification experiment for plate-shaped slag using the SUS304 mold. This figure shows a comparison of the calculated results and measured results at 90 s and 120 s after pouring molten slag with a temperature of 1673 K into the mold. The calculated results are in good agreement with the measured results when hs=30 W/(m2·K), hm=10 W/(m2·K), and R=9×10–4 m2K/W are used. The thermal resistance interface between the slag and the mold R=9×10–4 m2K/W is twice the value of the thermal resistance interface between mold flux and a copper mold R=4–5×10–4 m2K/W measured by Watanabe et al.17)
Heat transfer calculation result of slag and mold temperature distribution.
Applying these values of hs, hm, and R to the unsteady one-dimension heat transfer analysis model, the cooling rates of the surface, center, and mold contact face and the average of the slag are calculated at 120 s after pouring the molten slag into the mold. The calculation results are shown in Fig. 12, together with the measured data. The average cooling rates have been estimated from the average temperature of 11 points in the slag thickness direction based on the calculated results at 120 s after pouring the molten slag into the mold. When the thickness becomes 25 mm or less, the cooling rates of any position increase with a decrease in thickness, whereas, when the thickness is 25 mm or more, the cooling rate of the surface and the mold contact face become constant, and when the thickness is 40 mm or more, the cooling rate of the center is nearly zero. On the other hand when the thickness is 25 mm or more, the average cooling rate decreases gradually with an increase in thickness. Because the thermocouple position is not at the center of thickness, the cooling rates calculated by the measured temperature are higher than the calculated results for the center.
Next, the unsteady one-dimensional heat transfer analysis model is applied to the conditions of the pilot experiment, and the temperature of the center is calculated from the thickness and retention time in the mold. The calculation results are also shown in Fig. 15. Judging solidified or unsolidified from the fracture surface, the boundary temperature is 1633 K and not 1473 K as solidus. 1633 K is the temperature where viscosity begins to rise rapidly,20) even though the slag still has not solidified completely and the solid phase ratio of the slag is about 20%.
A method for improving the cooling rate of low thermal conductivity slag was examined based on this computation model. First, assuming the surface of the slag and the back of the mold are cooled by air, the temperatures of the slag and mold are calculated after 90 s and 120 s in the range of the heat transfer coefficient hs=hm=40–1000 W/(m2·K). Figure 19(a) shows the calculation results. The slag surface temperature decreases greatly with an increase in the heat transfer coefficient of the slag and mold, but the temperature of the center decreases gradually and the temperature of the mold contact face hardly changes.
Effect of heat transfer coefficient (a) and slag thickness (b) on slag temperature by heat transfer calculation. (Online version in color.)
Figure 19(b) shows the results of a similar calculation of the influence of thickness. When the slag thickness is 25 mm or less, the temperature of the center decreases remarkably with decreases in thickness. The slag surface temperature also decreases when the slag thickness is 25 mm or less. However, there is virtually no change in the mold contact surface temperature.
From these calculation results, because the thermal conductivity of blast furnace slag is low, at about 1 W/(m·K), producing thin slag is effective for increasing the cooling rate in the slag interior.
From the results of measurements of the pore diameter and pore number of the slag specimens collected in the cooling yard, it was found that porosity increases by pore growth, and porosity can be decreased by decreasing the slag thickness. This shows that an increase in the cooling rate of the slag is effective for decreasing porosity.
The center temperature of the plate-shaped slag differs greatly from the surface temperature, as Figs. 12 and 19(b) show, and the influence of the slag thickness on the cooling rate of the slag surface, mold contact face, and center is negligible. Although slag cools and solidifies from the slag surface and the mold contact face, the solidification ratio influences porosity. Therefore, because the temperature distribution in the thickness direction is thought to show the solidification ratio of the slag, porosity was arranged by the average cooling rate calculated from this temperature distribution. Since the initial cooling rate is considered to have a large influence on pore generation and growth, the average cooling rate for 120 s was predicted from the calculated value of the temperature distribution 120 s after the molten slag was poured. Figure 20 shows the relationship between total porosity and the average cooling rate. Here, total porosity Pt was calculated from the density in the absolutely dry condition in the slope test in Fig. 11 and the pilot experiment. From Fig. 20, it is understood that total porosity decreases as the cooling rate is increased up to an average cooling rate of 3 K/s, and at cooling rates over 3 K/s, total porosity becomes constant at approximately 5%.
Effect of cooling rate on total porosity of slag.
When 300 g of blast furnace slag was melted under a N2 gas atmosphere in a small furnace and cooled in an air atmosphere, the pores which formed were similar to those of ordinary air-cooled blast furnace slag cooled at 2 K/min or less.21) To adjust porosity to 10% or less, the average cooling rate in the slope test must be 150 K/min or more, while a cooling rate of 10 K/min or more was necessary with the small furnace.
Whether the origin of pores is oxidation of sulfur in the slag or oxidation of nitrogen dissolved in the slag, these gases are generated in by contact with oxygen in the air. Based on the facts that the influence of the cooling rate was different in the small furnace experiment21) and the slope test, and density increased in higher parts of the cooling bed, it is estimated that gas generation in slag is greatly influenced by agitation in the slag and by the contact time between the slag and the air, and porosity is determined by these factors. Thus, it was found that regardless of the reason for pore generation, an increase in the cooling rate is effective for reducing the porosity of blast furnace slag.
Figure 21 shows the relationship between the density in the absolutely dry condition and the abrasion loss of air-cooled blast furnace slag. Abrasion loss decreases as density in the absolutely dry condition increases. The micro Vickers hardness was measured in order to clarify whether the improvement of abrasion resistance was due to a change in the hardness of the slag itself or due to a change in the pore structure. The hardness of the pilot slag was Hv=177–581, and the average was Hv=302. No effect of slag thickness on hardness was found. On the other hand, the hardness of tight sand used as aggregate in asphalt mixtures is Hv=63–986, with an average of Hv=345. Although the hardness of the pilot slag was slightly smaller than that of tight sand, the slag contained neither an extremely hard part nor a weak part. This hardness is substantially equivalent to the measured Hv=151–348 of air-cooled blast furnace slag reported by Koshida et al.22) Therefore, the decrease in abrasion loss is attributable to a decrease in pores, and not an increase in hardness.
Effect of density on abrasion resistance of slag aggregate.
The relationship between abrasion loss Ra [%] and total porosity Pt (%) is shown by the following Eq. (8).
Because the abrasion resistance of the pilot slag was high, application of this slag to aggregate for an asphalt mixture of drainage pavement was examined. The test pavement was constructed at JFE Steel Corporation, East Japan Works (Chiba District). The pavement had no cracks and no rutting after 10 months. The result of test pavement confirmed that the crushed aggregate of plate-shaped slag can be used for drainage pavement.
The conditions for pore generation in air-cooled blast furnace slag were investigated, and the solidification conditions for reducing porosity were examined. In addition, a manufacturing process for low porosity air-cooled blast furnace slag was developed. As a result, the following conclusions were obtained.
(1) Pores in air-cooled blast furnace slag increase in size and decrease in number as porosity increases. In order to reduce porosity, it is necessary to control pore growth.
(2) Because the thermal conductivity of blast-furnace slag is low, the cooling rate in the interior of the slag is slow. Producing thin slag is effective for increasing the cooling rate in slag.
(3) The porosity of air-cooled blast furnace slag can be decreased by increasing the average cooling rate.
(4) The abrasion resistance of air-cooled blast furnace slag is improved when porosity decreases and density increases.
(5) A new process for solidifying plate-shaped slag was developed in order to produce air-cooled blast furnace slag with low water absorption. In this process, molten slag is poured into cast steel molds, and complete solidification of plate-shaped slag with a thickness of 20–30 mm is achieved in a short time of about 120 s. A dense aggregate with water absorption of 1% or less and high abrasion resistance of 15% was obtained by crushing and classifying the plate-shaped slag produced by this new process.
(6) An experiment with test pavements confirmed that it is possible to use blast furnace slag aggregate, which was produced by crushing plate-shaped slag with an average solidification thickness of 23.5 mm, in drainage pavement.