STAINLESS STEEL SLAGS IN HYDRAULIC BOUND MIXTURES FOR ROAD CONSTRUCTION, TWO CASE STUDIES IN BELGIUM Luc P. De Bock, Hendrik Van den Bergh Belgian Road Research Centre (BRRC), Belgium Abstract This article describes 2 Belgian case studies with the use of stainless steel slag in cement bound mixtures for road construction: in the first case as aggregate in a roller compacted concrete for a rural road pavement and in the second case the slag aggregates are in situ stabilised with cement to form the foundation layer under a large industrial storage area. Slag from stainless steel production is a stony material with good mechanical properties, that can be used as aggregate. Laboratory studies have demonstrated the capability of the slag of being stabilized with cement to produce a lean concrete mix. The slag is available, after breaking and sieving, in two sizes: 0/7 mm and 7/20 mm. Both sizes are used for the mixture, together with an addition of sand. The concrete mixture is fabricated in a conventional concrete plant, laid down, graded and compacted with both steel drum and pneumatic tyre rollers. The in situ stabilized mixture is produced with conventional machinery for soil stabilization and laid in three consecutive layers each 200 mm thick. The results of both projects are satisfactory and demonstrate the possibilities for successful recycling of stainless steel slag as an aggregate in road construction. Keywords : slags, stainless steel, recycling, road, cement 1. INTRODUCTION: ORIGIN AND CHARACTERISTICS OF THE SLAG 1.1. Origin of the raw slag Production of stainless steel also creates slag as a by-product. These slags must be treated as a waste material. The Belgian stainless steel maker ALZ-Ugine uses an electric arc furnace (EAF) for the production of apporoximately 900 000 ton stainless steel a year on its site in the city of Genk, with some 250 000 ton of raw slag. The type of slag can be described as an EAF steel slag, more specific stainless steel slag. The company TRC Recmix valorizes this waste material by processing the slags to get qualitative aggregates suitable for use in construction. The processing consists of crushing and sieving and separation of steel particles from the mineral aggregate and finally maturation of the slag aggregates. The slag aggregates are marketed unther the name STINOX. 1095
1.2. Mineralogical characterisation of the slag Special attention is given to the presence of free calcium oxide and magnesium oxide, as these minerals may undergo expansive reactions resulting in volumetric instability of the slag aggregate. The slag contains a limited amount of free lime: the coarse size fraction 7/20 mm had very little free lime in it ( 0.2 % CaO), and the free lime content of the finer fraction 0/7 mm was rather low ( 0.5 to 1 % CaO). The slag also contained magnesium in different mineral compositions: the total content of magnesium (expressed as MgO) was 7 to 8 %, possibly including some free magnesium. An accelerated swelling test ( steam swelling test ) on the combined fractions 0/7 + 7/20 of the fresh processed slag showed a volume increase of 1.5 % after seven days of testing. The accelerated swelling test performed on the fresh slag according to the BRRC method gave a result of 0.7 to 1.5 % of linear swelling for the size fraction 0/7 and 0.3 to 0.6 % for the fraction 7/20. This result is not dramatic - many applications are still possible, but shows that special attention to this swelling potential is justified, as a completely safe application of this kind of slag is not guaranteed. Ageing the slag by accelerating the hydration and oxidation reactions of the free lime during an extended period of storage had a beneficial effect on potential swelling. This ageing process is slow and takes at least three months. Moistening the slag immediately after crushing and sieving and further moistening during storage was certainly beneficial to ageing: after three months of ageing, the swelling potential of the slag was certainly lower though still present: the BRRC method for accelerated swelling gave a result of approximately 0.5 % of linear swelling for the slag 0/7 and about 0.2 % for the slag 7/20. 1.3. Mechanical and technical properties of the slag The slag aggregate is available in two size fractions: a sand fraction 0/7 mm and a stone fraction 7/20 mm. The fraction 0/7 can not be considered as totally sand, as it also contains particles with dimensions > 2 mm. The fraction 7/20 contains 1.3 % of fines (smaller then 0.063 µm), which is OK for application in road bases following the Flemish standards for road construction but slightly exceeds the maximum of 1 % of fines for use in pavement wearing courses. With an absolute density of 3.25 g/cm³, the slag is substantially heavier than conventional natural stones and sand. This calls for attention when calculating mix designs, because of the differences between volumetric and mass compositions. The intrinsic characteristics of the slag are good: resistance to abrasion and crushing is good, with Los Angeles and Micro-Deval resulting in a class A qualification individually but in a lower class B if taken together. The result of the static compression test also slightly exceeded the maximum limit for class A qualification. The polished-stone value was difficult to quantify, because the type of stones required for the test flat and smooth surface was not representative of the average slag as such. BRRC measured the PSV twice, with results 46 and 50. The minimum requirement for application in wearing courses of road surfacings is 50. So the slag was, generally speaking, of a good quality, but some characteristics did not always comply with the requirements for use in pavement wearing courses. On the other hand, the material did meet the requirements for road bases. 1096
1.4. Environmental quality of the slag aggregate The raw slag contains high concentrations of the heavy metals chromium and nickel, as weel as high concentrations of fluorine ion, specific for the stainless steel making process. Flemish environmental legislation limites the leaching of heavy metals like chromium and nickel. Because the leaching of chromium and nickel exceeds the limits for use as unbound aggregate, Flemish legislation accepts the use of the stainless steel slag aggregate only in applications where the aggregates are bound by a hydraulic or bituminous binder, like in concrete or asphalt mixtures. 2. USE OF THE SLAG AGGREGATE IN ROAD BASE LAYER 2.1. Design of the mixture A cement-bound mixture of the stainless steel slag aggregates for application in a road base layer was tested in the laboratory. For this kind of mixture important parameters are: a continuous particle size distribution 0/20 mm; a cement content of 4 to 6 % by mass of the dry aggregates; a water content equal to the normal Proctor optimum. The proposed mixture composition is (for 1 m³): - stone: stainless steel slag Stinox 0/20 : 78 % ~ 1,950 kg, - sand : round natural sand : 18 % ~ 450 kg, - cement : CEM III/A 32.5 LA : 4 % ~ 100 kg, - water: : 8 % ~ 200 kg. The dry density of a lean concrete mixture with 100 % of slag aggregates is approximately 2.53 g/cm³ at the normal Proctor optimum, at an optimum water content of 8 %. CBR (California Bearing Ratio) of this mixture with 5 % cement and 8 % water reaches a value of 251. A compaction rate of 95 % of this optimum, that is, a dry density of around 2.4 g/cm³, is the target on site; the target for the cement content of the mixture is 100 kg/m³. This amounts to 20 kg of cement per m² for a 20 cm thick layer and 30 kg of cement per m² for a 30 cm thick layer. 2.2. Execution of a pilot work: in-situ mixing with cement As the company who processes the slag, TRC-Recmix, needed to build a new site for storage of the slag aggregates, this work was used as a pilot work for evaluation of the use of the slag aggregates in road construction. The base layer for the pavement of the site consists of a cementbound mixture of the stainless steel slag aggregates and natural sand, executed by in-situ mixing. The site will be paved over an area of approximately 13,500 m² (135 m x 100 m). The structure of the pavement includes: a watertight protective bottom course in the form of a geomembrane, a subbase in sand fitted with a drainage system, a 60 cm thick road base in cement-bound slag, and an asphalt surfacing consisting of a base course and a wearing course. The road base uses stainless steel slag as part of the aggregates in a cement-stabilised mixture. The road base is constructed in three consecutive layers each about 20 cm thick after compaction. The components of the mixture were not mixed in a central plant, but in situ with mobile machinery, in this case a cold milling machine. Each layer was completed in the following consecutive steps: spreading and levelling a layer of 17 cm of slag STINOX 0/20), spreading and levelling a layer of 4 cm of natural round sand, spreading 20 kg/m² of cement on top of the sand, mixing the slag, sand and cement with water, levelling the mixture, compacting the layer. The equipment used for this job was conventional road construction equipment; mixing was performed by one pass of milling machine (type Wirtgen W 2200). 1097
2.3. Checking the characteristics of the materials and mixtures used The quality of construction of the road base was monitored by performing tests on the constituents and on the mixed materials. The results of the various tests and measurements made are presented hereafter. volumetric stability (swelling potential) The volumetric stability of the slag was evaluated from an accelerated swelling test, the socalled German steam test, following the method EN 1744-? (version 200?). From the slag material stabilized in situ samples were taken at three different places on the work site. As a departure from the test method, the samples (with a particle size 0/20 mm) were not regraded to match the Füller curve, but tested as such. The results on 6 samples indicate a volume increase of 0.4 % after 1 day and 1.8 to 2.9 % after seven days of testing, with an average value of 2.3 %. water content Water content was measured on the samples taken for the grading test. The results (average 6.2 % for layer 1, 6.4 % for layer 2 and 7.1 % for the top layer 3) are lower than the optimum of 8 % predicted in the theoretical design of the mixture. Cement content of the stabilized mixture The rate of spread of cement on top of the slag and sand was checked by placing a small container on the surface before the cement distributor arrived. After the cement distributor had passed, the cement collected in this container was weighed and divided by the surface area of the container. The result obtained 22.3 kg cement per m² - is in line with the designed cement content of 20 kg/m², equivalent to 4 % by mass. grading Samples of slag-sand-cement-water mixture were taken to check its grading. Sugar was added immediately to inhibit the setting process of the cement. The particle size distributions found indicate good continuous grading of the mixture. dry density after compaction (Modified Proctor procedure) Samples of the mixture of slag, sand, cement and water, as manufactured in situ by the milling machine, were taken in two different places from every layer and manually compacted in the field laboratory according to the Modified Proctor procedure. The results are given in table 1. Values for dry density range from 2.33 to 2.51 g/cm³, with an average of 2.47 g/cm³. The values for bulk density of the wet samples vary from 2.48 to 2.71 g/cm³, with an average of 2.63 g/cm³. Especially the sample taken from layer 2 has a lower density. This may be due to failure to achieve optimum mixture composition, resulting from a different proportion of stone to sand. Table 1: Densities of Proctor-compacted samples of stabilized road base mixture. Modified Proctor compaction Bulk density (g/cm 3 ) Water content (%) Dry density (g/cm 3 ) Layer 1 Layer 2 Layer 3 Average Test Test Test Test Test Test Test Test 1 2 1 2 1 2 3 4 2.644 2.707 2.484 2.588 2.670 2.639 2.695 2.649 2.64 6.2 7.7 6.7 6.4 6.8 7.0 7.4 7.2 7.0 2.489 2.513 2.332 2.432 2.500 2.468 2.509 2.471 2.47 1098
This result for the mixture as manufactured in situ is lower than the value achieved in the laboratory with the optimum lean concrete mixture (which was 2.54 g/cm³ of dry density at 8 % of water content = 2.74 g/cm³ wet density). This comparison, however, falls short, as the laboratory mixture contained a sand fraction made of the heavy slag, whereas the site mixture has a sand fraction made of lighter natural sand. The average value for water content reaches 7.0 %. unconfined compressive strength of in laboratory compacted and moulded specimens Four different samples of the mixture of slag, sand, cement and water, as manufactured in situ by the milling machine, were taken from the top layer and manually compacted in the field laboratory according to the Standard Proctor procedure. After twenty-eight days the unconfined compressive strength was determined. The bulk density of the in situ-stabilized mixture with stainless steel slag, measured after twenty-eight days of curing, averages 2.43 g/cm³. The results of the compressive strength tests are reviewed in table 2. Table 2: Compressive strength of moulded specimens of stabilized road base mixture Series No. 1 2 3 4 Compressive strength after 28 days at 20 C (N/mm²) Sample 1 Sample 2 Sample 3 Average 8.73 6.51 8.52 9.51 8.27 4.29 7.83 8.68 8.05 4.58 7.25 7.61 8.4 5.1 7.9 8.6 Although these values are rather widely dispersed ranging from 4.3 to 9.5 MPa, which may be due to differences in the composition of the mixture (more or less sand), the average result of 7.5 MPa is certainly good, as we remember that a low dose (only 4 %) of low strength cement is used. It was not the intention to manufacture a lean concrete mixture, but rather a stabilized aggregate road base. The relation between the bulk density of the specimens after manufacture and compaction and their compressive strength is shown in figure 1. Figure 1: Bulk density and compressive strength after twenty-eight days hardening of stabilized slag samples compacted in laboratory moulds. Samples compacted in laboratory moulds 10 9 Compressive strength (MPa) 8 7 6 5 4 3 2 1 0 2,3 2,35 2,4 2,45 2,5 2,55 B ulk density (g/cm ³) 1099
in-situ compacted density, tested with nuclear gauge After the completion of the mixed-in-place base layer a great number of measurements with a nuclear density gauge was done to check the compaction of the total construction. The results of reached density are summarized in table 3. Tabel 3: density by nuclear densitometer Number of Volume mass (Mg/m³) measurements taken Average Standard deviation % of the reference value Layer 1 35 2,56 0,08 97,3 Layer 2 38 2,47 0,11 93,9 Layer 3 82 2,61 0,08 99,2 The lower layer 1 and especially top layer 3 are very well compacted (97 to 99 % compared to the average compaction of 2.64 g/cm³ realized in laboratory mould following Modified Proctor procedure, which is taken as reference value). The intermediate layer 2 is less well compacted, probably due to sub-optimal mixture composition and low water content. bearing capacity of the road base The bearing capacity of the compacted road base was evaluated from a plate-bearing test, using a plate of 200 cm² in area. The Belgian standard tender specifications require a modulus of compressibility M 1 110 MPa to be measured within max. 72 h after completion of the course. Five out of the eight results meet this criterion, but the other three are lower then 110 MPa. Further compaction is necessary on those locations. The results are presented in table 4. Table 4: Results of plate-bearing tests on the compacted road base Results Location number First cycle, Second cycle, ratio M 1 /M 2 M 1 (MPa) M 2 (MPa) 1 64 399 6 2 80 798 10 3 200 1,596 8 4 177 798 4 5 532 1,596 3 6 73 399 5 7 266 1,596 6 8 160 798 5 compressive strength Samples taken from layer 3 were compacted in the laboratory in cilindrical moulds (diameter 100 mm and height 120 mm) and tested on unconfined compressive strength after 28 days curing at 20 C and 90 % relative humidity. The results vary from 5.0 to 9.5 MPa, with an average of 7.0 MPa. 1100
3. USE OF THE SLAG AS AGGREGATE FOR ROLLER COMPACTED CONCRETE FOR RURAL ROAD PAVEMENT The objective of this part of the research was to use the stainless steel slag aggregates as aggregate for concrete for road pavement, more specific in roller compacted concrete for pavement of a rural road with low trafic (agricultural machines and bicycles). 3.1. Design of the concrete mixture The stainless steel slag aggregate was available in a 0/30 mm calibre. To reach a good continuous grading and good workability a natural round sand 0/4 mm calibre was added, so the composition of the mineral aggregates was 85 % slag 0/30 + 15 % natural sand 0/4. To reach the objective of 30 N/mm² concrete strength after 90 days, a cement dose of 300 kg/m³ was used, this means around 16 % in mass of the aggregates. Following the Modified Proctor procedure the optimum water content of the mixture was found to be 5,5 %, with optimum density reaching 2.58 g/cm³. Load bearing capacity is very good: 250 % CBR at optimum water content, quickly deteriorating to 80 % CBR at a water content of 8 %. The theoretical mixture composition resulting from this design is thus (for 1 m³ of roller compacted concrete): 1938 kg STINOX 0/30 stainless steel slag 342 kg natural round sand 0/4 300 kg cement type CEM III/A 32,5 142 kg water. The addition of 142 liter of water is calculated from the optimum water content found in the Proctor study (2.58 g/cm³ x 5.5 % = 142 kg/m³), and is valid for oven dry aggregates in the laboratory. In practice do aggregates, which are stored in open air, already contain some water. The moisture content of the aggregates used, as measured by the contractor on the day of production, was 5.0 % in the slag and 5.3 % in the natural sand. Also, not all the water added to the mix is available for reaction with the cement, e.g. because the water is absorbed by the aggregate in fine pores or cracks. This amount of bound water is thought to be 2 % of the total mass of the slag aggregate; the rest of the moisture (3 %) is considered to be free water, available for reaction with the cement. In the natural sand all of the water added is thought to be available for cement. The amount of free and available water in the aggregates does not count for the calculation of the water to be added extra in the mixer. So, the composition of the mixture based on not-dried aggregates is (per m 3 of roller compacted concrete): 1996 kg stainless steel slag aggregate 0/30 (= 1900 kg oven dry slags + 38 kg bound water + 58 kg "free" water) + 360 kg natural sand (= 342 kg oven dry sand + 18 kg "free" water) + 300 kg cement + 66 kg extra added water. The total water content in the mixture is then 179 l (= 66 l extra added water + 18 l in the wet sand + 38 l bound water in the slag + 57 l free water in the wet slag), which will result in a value of 7.0 %-m measured after oven drying at 105 C. With this mixture composition samples were fabricated in the laboratory for testing the simple compressive strength. The results are given in table 5. 1101
Table 5 Compressive strength of in laboratory fabricated and compacted specimens Specimen N Compressive strength (MPa) After 7 days hardening Average of 3 results Compressive strength (MPa) After 14 days hardening Average of 3 results Compressive strength (MPa) 1 26.2 36.6 37.9 2 28.1 26.7 29.3 31.1 39.7 3 25.7 27.2 33.5 After 28 days hardening Average of 3 results 3.2. Production of the concrete mix This composition served for the fabrication of the concrete mixture by the mixing plant. The workability and the compactability of the mixture on the workspace was considered to be satisfactory well. An attempt to increase the water added with 5 to 10 liter per m³ was not successful; the mix was too wet and needed extra time before it could be compacted. At the end of the day, with hot sunny weather, the concrete mixtures arriving at the workplace seemed rather dry, probably due to desiccation during transport. At the end of the day a change in the mix composition was tried in the hope to counter the former problems of segregation of the mix with more sand and less slag: the amount of natural sand was increased from 360 kg before to 470 kg per m³ of concrete now, equivalent to 20 % of the total mass of aggregates instead of 15 % before, and the amount of slag was decreased from 1990 to 1880 kg per m³. Samples of the produced mix were taken immediately at the concrete mixing plant to undergo some tests for quality control. Therefore specimens were fabricated in Proctor moulds and compacted in three different layers using the energy of the Modified proctor procedure. Equally spread over the duration of the production of concrete, 3 samples of concrete were taken to fabricate 3 test specimens each. The results of these tests are given in table 6. The compressive strength reaches an average value of almost 29 MPa after 7 days, increasing to 31,6 MPa after 32 days and finally reaching an average value of 36,2 MPa after 90 days. These results are very good, and consistent with the values found for the mixtures fabricated in the laboratory during the design phase (see 3.1.). Table 6: characteristics of the concrete fabricated industrially and then compacted in laboratory Specimen N 1 2 3 4 5 6 7 8 9 Water content (%) 7,0 9,5 6,7 Volume mass wet (g/cm 3 ) 2,76 2,78 2,76 2,78 2,77 2,72 2,76 2,80 2,72 Compressive strength after 7 days curing (MPa) 9,2 * 25,8 31,9 Compressive strength after 32 days curing (MPa) 28,8 30,6 35,3 Compressive strength after 90 days curing (MPa) 43,2 32,1 33,4 * specimen damaged during the demoulding action 37.1 1102
3.3. Paving the roller compacted concrete pavement of the demonstration road track This composition served for the fabrication of the concrete mixture at the mixing plant. The fresh concrete mixture was transported to the work site in open trucks, spread out, leveled with a grader and finally compacted with a roller combining vibration and tyre pressure compaction. The workability and the compactability of the mixture on the workspace was considered to be satisfactory well. Control of the compaction by means of a nuclear gamma-source density gauge gave results of almost 100 % degree of compaction, compared to the optimum densities reached in laboratory. The roller compacted concrete does not have a smooth surface finishing, due to the compacting with a tyre-fitted roller. This is done on purpose because it is believed to make the road pavement better fitting (more accepted) in the rural environment and make the road less attractive to potential car traffic. At some places the pavement surface shows aggregates not fully bound by the cement mortar. This is due to a lack of mortar in that piece of concrete composition, possibly due to segregation in the aggregate mixture. Eventually this can lead to surface damage in the pavement, resulting in loss of concrete particles. This defective aspect occurred especially at the end of the production day, when due to the sunny weather less water was available in the fresh concrete mixture. 3.4. Quality control of the concrete pavement The quality of the concrete pavement was tested with simple compressive strength test after hardening of the concrete, on cores drilled out of the pavement layer. At 5 places equally spread over the length of the road repeatedly were taken 3 cores by drilling in the pavement layer, after 7, 28 and 90 days of hardening. By removing top and bottom edges by sawing, out of every core a cylindrical test specimen of 100 mm height and 113 mm diameter was taken for testing the compressive strength. The results are given in table 7. Table 7: Compressive strength of roller compacted concrete pavement cores (in N/mm²) Age of concrete 7 (+ 4 days curing under water in the lab) 27 (+ 5 days curing under water in the lab) 90 Coring L 1 L 2 L 3 L 4 L 5 L 1 L 2 L 3 L 4 L 5 L 1 L 2 L 3 L 4 L 5 location Core n 1 34 27 28 21 23 38 38 36 44 5 46 48 49 37 11 Core n 2 37 28 39* 25 24 36 39 39 38 6 53 44 41 33 7 Core n 3 40 31 40* 29 30 46 34 ** 25 5 52 48 29 24 13 Average strength 37 29-25 26 40 37 37 35 5 50 47 40 31 10 * : These cores have a 2 days greater age, because they could not be tested on the same day as the others, due to a breakdown of the test equipment. ** : This core could not be tested because it partly disintegrated. From these results it can be observed that the compressive strength of the concrete cores shows a large spread, e.g. the values of strength after 11 days of hardening vary from 21 to 40 MPa, giving an average of 30 MPa with a standard variation of 6 MPa, and after 90 days of hardening. The cores taken after 90days of hardening of the concrete pavement show an even wider spread, with values for compressive strength ranging from 24 to 53 MPa giving an average of 42 and a standard variation of 9 MPa. 1103
In these last calculations the results of the cores taken at coring location L 5 are not included because they are far outliers. There seems to be no logical explanation for the abnormally weak strength of the concrete at this location; perhaps there was a problem with the composition of the mixture, e.g. a problem in the cement distributor resulting in a temporary absence of binder in the mix. This location is the one where at the end of the day an alternative mixture composition was tried (see 3.2.). The cores taken at this location showed lack of cohesion; at certain depths in the pavement layer loose stones were present and during the core-drilling part of the mortar was washed away. It is however strange that the 3 cores taken at the same location after 7 days hardening did not deviate substantially form the average on the other locations. The third core taken at location L3 at the age of 27 days could not be tested because it partly disintegrated due to a pop-out -phenomenon. This is that a piece of about a few cm³ broke off from the rest of the core at a place where a larger crack was situated. This crack was situated around a piece of hydrated free lime of about 1 cm³ in dimensions. Due to the extra curing under water during 5 days between coring and testing this CaO probably reacted with the water present, giving a volume increase of the CaO molecules to Ca(OH) 2 molecules, and this swelling pressure resulted in the crack and disintegration of the concrete core sample. This adverse phenomenon did not show up on any other moment during the research. 4. CONCLUSIONS The stainless steel slag STINOX investigated here is a stone of good mechanical characteristics, suitable for application as aggregate in hydraulic bound mixtures for road construction, as demonstrated in the 2 case studies here mentioned. In-situ mixing of a mixture composed of 78 % slag 0/20, 18 % of natural round sand and 4 % of cement plus 8 % water gave good results for use as a base layer for an industrial store yard. The application as aggregate in roller compacted concrete for pavement of a low trafficked rural road was also successful. 1104