The Potential of Lime and Grand Granulated Blast Furnace Slag (GGBFS) Mixture for Stabilisation of Desert

This study describes experimental results achieved on the use of Grand Granulated Blast Furnace Slag (GGBFS) and Lime in stabilising desert silty sand for possible use in geotechnical engineering applications, especially for roadways and railways constructions. The GGBFS and lime were added in percentages of 5, 10 and 15% and 1, 3, and 5% respectively, by dry weight of sand. Different laboratory tests such as mechanical aggregation test, hydrometer analysis, liquid -plastic limit , pH value test, compaction, unconfined compressive strength (UCS), California bearing rat io test CBR , were performed on samples to understand the engineering characteristic of soil and influences of mixtures on the silty sand soil. The study results demonstrate significant improvements in unconfined compressive strength and Californ ia bearing ratio strength. Moreover the swelling behaviour of mixtures was decreased effectively. Thus mixture of GGBFS and lime can be suggested to improve engineering characteristic of desert silty sands.


Introduction
Nowadays faster transportation and saving more energy has undeniable role on development of societies. In countries with large desert areas, expansion of roadways and railways and preparing suitable construction materials is one of the main technical and economical engineering challenges.
Desert sands are usually fine-grained and poorly graded materials with small amounts of silt [1]. Desert sands are not suitable for support of structures and roads, because they are loose and vulnerable to co llapse upon wetting [2]. Low bearing capacity, strength, stiffness and high porosity of this type of soils cusses excessive settlement and severe damages to roadways and railways constructions. Also preparing and transporting proper construction material fro m other areas forces excessive costs on project and is not economical.
There are several methods for improving the strength of so ils an d o n e o f th e mo st effect iv e meth ods is so il stabilisation. Various methods of soil stabilisation, such as,

Background
Incineration of MunicipalSo lid Waste (MSW) is a common practice to reduce the volume of the waste to be disposed in a landfill [1].Regarding to existence of numerous iron and steel smelting factories in Iran, one of the materials which incineration process produces are different kinds of slag like (GGBFS) that can fu rther be utilized in construction activities. Blast Furnace Slag (BFS) is a non-metallic co-product, produced in the process of iron and steel production. Different forms of slag product are produced depending on the method used to cool the molten slag. These products include Air-Cooled Blast Furnace Slag (A CBFS), expanded or foamed slag, pelletized slag, and granulated blast furnace slag [16]. GGBFS fo rm produces by cooling the mo lten slag using high pressure water jets to cool it rapidly .This method of cooling, results in produces of granular product [17], and formation of sand size (o r frit-like) frag ments, usually with some friable clin ker like material. The physical structure and gradation of granulated slag depends on the chemical co mposition of the slag, its temperature at the time o f water quenching, and the method of production. When crushed or milled to very fine cement-sized particles, GGBFS has cementit ious properties. It primarily consists of silicates, alu mina-silicates, and calciu m alu mina-silicates [16]. GBFS used as a stabiliser has latent hydraulic properties. This means that, similar topozzolanic materials, the slag can form strength enhancing products with calciu m hydroxide (Ca [OH] 2 ). The difference is that the slag contains rather more react ive lime. However, the reaction rate of the slag itself is so slow as to be negligible. So me form of activators is therefore necessary [18]. Higgins [19],observed that GGBFS on its own has only mild cementitious properties and lime (calciu m hydroxide) can provide the necessary alkali for activation. The most commonly used activators to activate GGBFS are lime, alkalis [20], calciu m hydro xide, calciu m sulphate, ordinary Portland cement, sodium hydro xide, sodium carbonate and sodium sulphate [21]. The use of GGBFS is well established in many applications where it provides good durability, high resistance to chloride penetration, resistance to sulphate attack and protection against alkali silica reaction [17]. For instance In South Africa, GGBFS act ivated by lime, is a common ly used binder for soil stabilisation [22], also blends of lime and GGBFS are frequently used in Australia [23,24].
The earliest work in modern t imes on the use of lime in road construction goes back to 1925 in the American state of Missouri [25]. Lime stabilisation is one of the most commonly applied soil strength improvement techniques. Generally, addit ion lime to clayey soil increases the soil strength to a certain limit, however adding excess lime tends to decrease the strength [26]. This technique is widely used in the sub-grade, sub-base and base layers of road construction [27].
Lime is produced by burning limestone and it can be used to treat soils in the form of, quicklime (CaO), hydrated lime (Ca [OH] 2 ), o r lime slurry. Qu icklime is manufactured by chemically transforming calciu m carbonate (CaCO3) into calciu m o xide. Hydrated lime is created when quicklime chemically reacts with water, the hydrate can be reconverted to quicklime by removing the water by heating it. It hydrated lime that reacts with soil part icles and permanently transforms them into a strong cementitious matrix [16]. The reaction formu la of quick lime and water is shown as below: This reaction generates heat and the pH value increases to approximately 12.5. It is a suitable condition for the subsequent pozzolan icreactions [18]. Similar studies by Mallela [28], showed above fact which results to soil stabilisation. The pozzo lanic reactions occur between silica and alumina with in the clay structure with lime and water to form calciu m silicate hydrate (C-S-H) and calciu m alu minates hydrate gels (C-A-H) which subsequently crystallise to bind the structure together [29].
The most important reactions of lime with soil can divided into four groups; (a) cation exchange; (b) flocculation and agglomerat ion; (c) carbonation; and (d) pozzolanic reactions [30,31,32], and the following changes are observed in the soil in short term [27]: optimu m water content values increase, proctor densities decrease, plasticity indices reduce, proctor curve levels out, unconfined compressive strength, and CBR values increase. The use of lime stabilisation for road constructions reduces the thickness of the upper layers due to high CBR values and makes the overall construction more economical [27]. In studies conducted by Kavak [33], and Kavak and Bay kal [34] pure bentonite and kao lin ite clays were lime -stabilised and unconfined co mpressive strengths were increased significantly. Based on the studies conducted by Thompson[35] andNewbauer and Tho mpson [36], they have found changes in the water content-density relationships as a result of the reactions between the lime and the soil. They have also found that the optimu m water contents of the lime-stabilised soils are higher when compared to that of the natural soils.
Lime stabilisation has a detrimental effect on soil behaviour if adequate amounts of sulphate are present in soil [37]. Su lphates can do reaction with lime and causes serious consequences such as swelling, heave, and damages [38][39][40][41]. Excessive sulphate in the soil will lead to ettringite formation. Ettringite will lead to excessive heaving or swelling due to its needle like shape [27]. Regarding this case, using Pozzolan ic and semi Pozzo lanic materials are considered to decrease such a problem. W ild [42], and Veith [43], stated that slag at predetermined percentages will decrease this effect and if the sulphate content is less than 1%, sulphate will not have any effect on swelling. So b lends of lime and GGBFS might be resistant to swelling caused by sulphate [44]. In addition, laboratory tests have shown a previously undemonstrated advantage where the incorporation of GGBFS co mbats the deleterious swelling which can occur when sulphate-containing soils are stabilised with cement or lime [45]. Higgins [19] showed that GGBFS was completely successful in reducing swelling caused by sulphate. They also found that substitution of GGBFS for lime could significantly reduce swelling and heave in the presence of sulphates. Higher percentages of replacement of lime with GGBFS, with only sufficient lime to activate the GGBFS, are the most effective in preventing sulphate attack [45]. The addition of GGBFS reduces the permeab ility of stabilised soil significantly wh ich has high permeab ility in natural state. The addition of GGBFS can reduce the coefficient of permeability to 10 -6 cm/s, which satisfied the requirements for water retaining structures [46]. The investigation showed two major reactions when GGBS and lime were added to the soil (especially clay soils), hydration of GGBS act ivated by lime to produce calciu m alu mina silicate hydrate gel (C-A-S-H) and hydrotalcite type phase, and the clay-lime reaction to produce (C-S-H), (C-A-H) and (C-A-S-H) [17]. The addit ion of GGBS provides additional alu mina, calciu m, silica and magnesia to the mixtures depending on the type and amount of GGBS replacement [47].
A successful stabilisation method depends on many factors such as: (a) soil type and properties; (b) s tabilising agent; (c) stabiliser content; (d) potential use of the stabilised soil; (e) field mixing method; and (f) economical considerations, such as choosing type of additive considering its price per litre or per kilogram. For a given soil and a given stabiliser, the field mixing method and the economic factors will control the success of the stabilisation process [1]. It should be noted that the strength-enhancing reactions that occur during stabilisation with GGBFS are h ighly temperature-sensitive. Higher temperatures normally increase the reaction rate and hence the strength [18]. Gupta and Seehra [20], studied the effect of lime -GGBS on the strength of soil. They found that lime -GGBS soil stabilised mixes with and without addition of gypsum, or containing partial replacement of GGBS by fly ash produced high UCS and CBR in compare with plain soil. More informat ion and detailed records can be found in relevant PhD Theses [43,48,49].
The above background and review of available literatures shows that the main thrust of the research on soil stabilisation have been focused on use of lime alone or mixture of incineration process produces like GGBFS and lime as an activator especially in soils which contain considerable amounts of clay and it seems that fewer res earches have been done on desert silty sand soils wh ich do not have clays or have few amounts of clays. Thus, the research presented in this paper aims to contribute to this important issue.

Desert Sands
The sand used in this study was obtained from the desert area of JANDA GH-GA RM SAR which is located in central desert of Iran. The engineering plan was to constructing 230 kilo metre road and railway between these two cities. Figure1 illustrates the study area. The silty sand soil in this region starts at 0.5-1 m depth below the g round level (bgl) and extends down to about 3 mbgl. Due to excessive absorption of salt minerals over the past decades, the soil strength is very poor, it has porous shape and contains of large amounts sulphates. Figure 2 shows the borehole and surface of untreated soil in the study area.

Li me
The lime used in this study is a fine ground calciu m hydroxide (Ca [OH] 2 ) provided fro m Iran-Qo m limestone factory. Lime part icles were finer than sieve No.60 (0.250mm).

Grand Granul ated Blast Furnace Slag (GGB FS)
The slag used in this study is Grand Granulated Blast Furnace Slag (GGBFS) obtained fro m Iron smelting factories (Iran-JAJROOD) in process of producing ST-37. GGBFS part icles used in this study are milled and finer than Sieve No.200 (0.075mm).

Treatment Procedure
In this experimental study, several nu mbers of specimens fro m the untreated silty sand soil and mixture of GGBFS-lime-soil were investigated.
At the room temperature (25±°C), GGBFS was added in percentages of 5, 10, and 15% and lime was added in percentages of 1, 3, and 5%, by dry weight of the soil. The sand, GGBFS and lime were mixed thoroughly by hand until homogeneity and a uniform colour were reached. Water was added as needed to facilitate the mixing and compaction processes. In each case, modified proctor test performed to determine optimu m water content and dry unit weight of untreated soil and mixtures. Co mpaction was performed with optimu m water content determined in the compaction tests, just immediately after mixing, since the delay decreases the unconfined compressive strength [50], and have negative effect on CBR strength. The compacted specimens were cured for 7 and 28 days in tied plastic package to prevent los s of mo isture content.Finally all samples were tested after curing time.

Mineralogical and Micro Structural Tests
The mineralogy of the silty sand, GGBFS and Lime used in this research were identified by the X-Ray Diffraction technique (XRD). X-Ray powder diffraction analysis is a powerful method by which X-Rays of a known wavelength are passed through a sample in order to identify the crystal structure. Peak positions occur where the X-ray beam has been diffracted by the crystal lattice [51].
Specific percentages of soil-lime and soil-GGBFS-lime mixtu res were prepared and analysed under Scanning Electron Microscope (SEM) with 200 to 7,500 times magnificat ion. SEM is used to generate images of the surface and the subsurface of specimen at magnifications in the range 20 x-20000 x. It can be used to examine the micro -structure of specimens and to determine particle crystallinity. SEM may also be used to characterize and identify particu lar phases and their shape and forms [17].

Geotechnical Tests
Various geotechnical experiments performed in this research such as, the grain size analysis, specific grav ity of soil, the atterberg limits tests and the standard proctor compaction test. Unconfined comp ressive tests were strain-controlled. The rate of strain was maintained at 1 mm/ min . In this test specimens were co mpacted using Harvard co mpaction hammer, in 5 layers by 25 hammer blows in each layer. Samp les were made with 31mm diameter and 75mm length. Also CBR test were performed. In this test, the moulds were filled in five equal layers, and each layer was compacted by 10, 30 and 65 hammer blows (represented by N), then 2.26 kilograms overhead load was placed on the specimen to represent the weight of pavement layers. Moreover pH values and soil swelling potentials were evaluated.

General Soils Specificati ons
The mechanical aggregation test was done by using wet method and therefore minerals and salt were d issolved in water while washing the soil [52]. Hydro meter test was conducted based on the ASTM standard [52]. The grain size distribution of untreated soil sample has indicated that the soil is composed of 67.7% sand, 25% silt and 2% clay, and According to the Unified Soil Classification System (USCS), the sand can be classified as fine grained, silty sand (SM).It should be noted that about 30% of soil weight was found to be minerals and salts. Figure 3.shows the grain size distribution of the usedsilty sand soil. Sand has a specific gravity of 2.52 [53]. The atterberg limits were conducted based on ASTM standard [54], and liquid -plastic limit values were measured. Cohesion of the soil is so poor due to lo w clay content. Therefore, the soil was classified as N.P soils (Non Plastic).

X-Ray Di ffraction Analysis and Chemical
Composition of Materials X-ray diffraction (XRD) test was performed on soil, GGBFS and lime. Figure 4 illustrates X-Ray pattern of GGBFS.   [55], and presence of sulphate in the soil can cause heave problems by reaction with lime and water.
The predominant compounds in the GGBFS is silicon dio xide (SIO 2 ) 25.86%, Iron (III) o xide or ferric o xide (Fe 2 O 3 ) 24.34% and Calciu m o xide (CaO) 18.77%.SIO 2 is most commonly found in nature as sand or quartz, as well as in the cell walls of diato ms [56,57], and Iron (III) o xide is the feedstock of the steel and iron industries, e.g. the production of iron, steel, and many alloys [58].
The predominant compound in the hydrate lime is calciu m oxide common ly known as quicklime or burnt lime (CaO) 51.64%.

PH Values
In order to determine the optimu m content of lime required fo r stabilisation, the pH value tests were conducted by using Eades& Grim method [59]. The p H values of untreated soil and soil-GGBFS mixtures were equal and found to be 7.6. So change in percentage of GGBFS does not make any change on the pH values. Then the pH tests were performed on the mixtures of soil-lime and soli-GGBFS-lime. Figure 5 shows the results of this analysis.
Regarding to above results,the optimu m amount of lime belongs to the sample which contains 10 % GGBFS and 1% lime. Maximu m p H values were found in samples which contains 4% lime and different percentages of GGBFS. It should be noted that in general the lime addition increased the pH value of the samples. It was also observed that the pH value decreases in all mixtures when the lime content reaches to 5%.

Compacti on Tests
Co mpaction test is usually performed to re-arrange soil particles by mixing water with the soil [60].Themodified Proctor tests were perfo rmed in accordance with ASTM standard [61], for both untreated soil and soil-GGBFS-lime mixtu res. The maximu m dry unit weight of untreated soil was 20.1 (kN/ m 3 ) while the optimu m water content was 9.96%. The results of the compaction tests for the various mixtu res are shown in Table 4. and 5% GGBFS, and the optimu m water content is increased fro m 9.96% for untreated soil sample to 13.66% for a samp le which contains 15% GGBFS and 5% lime. Also lime addition is reduced the maximu m dry unit weight and so increased the optimu m water content. The main reason for the increase in optimu m water content is that the larger quantities of water is required to hydrate the increased amount of (Cao) in the lime, and reduction of maximu m dry unit weight is result of flocculation and agglomerat ion produced by immediate reactions between lime and soil. These results were in parallel with previous researches like [62,63].
Addition of GGBFS increased the optimu m water content and the maximu m dry unite weight of mixtures slightly. It seems that fine GGBFS powder were filled the voids between soil particles. This result is in parallel with previous researches i.e. [64,65].

Unconfined Compressive Strength (UCS) Test
Unconfined compressive strength (UCS) tests were performed based on ASTM standard [66]. The unconfined compressive strength of untreated soil was measured 160 (kN/ m 2 ). Th ree samp les were prepared for each mixture and each curing time and they were cured for 7 and 28 days . The average values of every three samples were determined as results of the UCS tests. Results of performed UCS tests are presented in Figure 6. The UCS results of mixtures showed that, in general, as the lime content increased the unconfined compressive strengths of mixtu res were increased too.
The least increase of UCS is found to be 3 times for 1% lime and 5% GGBFS and the ut most increaseis 24.5 times for 3% lime and 15%GGBFS in co mpare with untreated soil. Addition of lime was produced more calciu m hydroxide to react with GGBFS and so increased the strength of the mixtu res.
GGBFS addition was increased the unconfined compressive strength of samples slightly, but generally by increasing GGBFS content in mixtures, higher dose of lime is required to activate it. Presence of silica was caused producing of more solid part icles and so more cementation bonds were formed at the contact points between the solid particles.
Extending curing time fro m 7 to 28 days had considerable effect on increase unconfined strength of samples. The least increase of UCS is found to be 2.7 t imes fo r 1% lime and 10% GGBFS and the ut most increaseis 5.2 t imes for 3% lime and 15% GGBFS. The increase of UCS strength in parallel with increase of cu ring t ime is main ly due to the pozzolan ic reaction, hydration and crystallisation of the products which cussed to forming cementitious structure of the materials. This result is in agreement with previous researches i.e. [67].
The optimu m content of lime depends primarily on the type of soil and curing conditions [19].Previous engineering test results by other researchers have found that the optimu m lime-GGBFS rat io to achieve maximu m UCS is 1: 5 [ 17]. It was also suggested that this ratio o f a lime-GGBFS mixtu re is enough to activate GGBFS [45]. As shown in Figure 6, maximu m USC value is obtained in the sample which contains of 3% lime and 15% GGBFS.

CBR Values
California bearing ratio (CBR) tests were conducted in accordance with ASTM standard [68]. Wet condition was prepared for soil sample and mixtures by soaking 7 and 28 days cured samples in water for 4 days (96 hours).
The results of CBR tests on untreated soil are shown in Table 5 and Figure 7. The un-soaked CBR values found to be higher than soaked CBR values.  Twelve samples were made for each soil-GGBFS-lime mixtu re, six of them fo r 7 days curing time and the other six for 28 days curing time. Every six samp les were div ided to two trine samples for testing in un-soaked and soaked conditions. Figure 8, 9 and 10 show the CBR values obtained fro m un-soaked condition tests.
Co mparing results of CBR test on untreated sand ( Figure  7) and treated sand (Figure 8, 9 and 10); it is clear that treating the material using GGBFS and lime has greatly increased the CBR values. In most cases it has been observed Slag (GGBFS) M ixture for Stabilisation of Desert Silty Sands that 28 days cured samples had more CBR values in compare with 7 days cured samples. These samples had enough time for po zzo lanic reactions, hydration and crystallis ation to gain higher strength and CBR values.
Also it was found that the CBR value of 7 days cured sample which contains 5% lime and 5% GGBFS and compacted by 10 b lows, is 3.6 t imes higher than the untreated one. This result is the minimu m gro wth rate of CBR values among all other un-soaked samples. Moreover the CBR value of 7 days cured sample which contains 1% lime and 10% GGBFS and co mpacted by 30 b lows, is 5.6 times higher than the untreated one. This result is the maximu m growth rate of CBR values among all other un-soaked samples.
7 days cured samp les which contains 10% GGBFS, show reduction in the CBR values by increase of the lime content, so excess lime has decreased CBR strength of the soil.  The results showed that, GGBFS addit ion has increased the CBR values of mixtures, especially in presence of 3% and 5% lime. Ho wever by increasing of GGBFS content, more lime is needed to activate it, and presence of 1% lime was not enough to activate 15% GGBFS and therefore CBR values of these mixtures is found to be less than expected value.
In the next stage the soaked CBR tests were conducted.    Results fro m perfo rmed tests in soaked condition are very similar to un-soaked condition. Furthermore the following results were also observed.
The CBR value of 7 days cured sample wh ich contains 5% lime and 5% GGBFS and co mpacted by 10 blo ws, is 5.3 times higher than the untreated one. This result is the minimu m growth rate of CBR values among all other soaked samples. Moreover the CBR value of 28 days cured sample which contains 3% lime and 5% GGBFS and co mpacted by 30 b lows, is 5.7 t imes higher than the untreated one. This result is the maximu m growth rate of CBR values among all other soaked samples.
Maximu m CBR values obtained in d ifferent co mpaction blows are associated to lime -GGBFS ratio of 1: 5.
It was observed that the soaked CBR values are slightly smaller than the un-soaked CBR values, however much mo re reduction in CBR values were expected. Perhaps while sample is soaked for 96 hours, water penetrates into the sample, lubricates the soil part icles and reduces the samples strength subsequently.

Swelling Values
Swelling potential was measured during 4 days (96 hours), while samp les were soaked in water.
As shown in Figure 14, for the soil-lime mixtures, maximu m swelling values were observed in presence of 1% lime and other mixtures wh ich contain 3% and 5% lime shows less swelling values.    In general, it was found that samples compacted by 65 blows in each layer show the min imu m swelling values than other samples which have been co mpacted by 10 and 30 blows, so there is a d irect relat ionship between minimu m porosity obtained from compaction and decrease of swelling values.

Scanning Electronic Microscopy
In this set of experiment, in order to evaluate interaction between soil, GGBFS and lime, three samp les were prepared and cured for 7 and 28 days . One sample contained 0% GGBFS and 5% lime and other two samples contained 1% lime and 10% GGBFS and 3% lime and 15% GGBFS. SEM analyses were conducted on samples with 200 to 7500 times magnificat ions. Figure 18, 19, 20, 21 and 22 present these results.
The SEM analysis of sample with 5% lime and w ithout GGBFS is shown in Figure 18. It can be seen that the soil particles were slightly coated and surrounded with lime and minerals had spherical shape.
It can be found fro m Figure 19, 20, 21 and 22 that curing time can play an important role in stabilising of samples.Regard ing to 7 and 28 days cured samples, it can be seen that soil has become denser in a time dependent manner and there are less voids available after 28 days curing. It seems that producing of cemented materials is because of pozzo lanic reactions between soil, GGBFS and lime. Most soil particles were covered by silica and alu mina hydrate gels which cussed forming cementitious structure of the materials and subsequent crystallisation to bind the structure together. The voids became smaller, so pore spaces have reduced significantly and a denser structure obtained and this event can reduce the permeab ility of the samples. These results were in parallel with above presented CBR and UCS tests results. In Figure 22, the angular shape of particles can be seen clearly.

Conclusions
In this study mixture o f GGBFS and lime were utilized as a soil stabiliser to improve engineering properties of desert silty sands. The following conclusions were derived fro m this experimental research: Adding lime to the silty sand has increased the pH value of the samples, but generally addition of GGBFS has no effect on pH values. Lime addition has reduced the maximu m dry unit weight and has increased the optimu m water content of silty sand.Also addition of GGBFS has increased the maximu m dry density and optimu m water content of samples.
Generally as the lime content increased the unconfined compressive strengths of mixtures were increased too and GGBFS addition has increased the unconfined comp ressive strength of mixtures.
Results of CBR tests showed that when untreated soil has mixed with various percentages of GGBFS and lime, the un-soaked and soaked CBR values of samples have increased significantly, but soaked samples shows lower CBR values in co mpare with un-soaked samples.
GGBFS addition has increased the CBR values of mixtu res, especially in presence of 3% and 5% lime.
In general, samples compacted by 30 b lows in each layer shows higher CBR values than other samples co mpacted by 10 and 65 blo ws.
Increasing curing time fro m 7 to 28 days had a considerable effect on increasing the unconfined compressive strength and CBR values.
Addition of GGBFS has significantly reduced the swelling ratio of mixtures.
It was seen that mixtures co mpacted by 65 blo ws in each layer show the minimu m swelling values than other mixtures compacted by 10 and 30 blows.
 the optimu m lime -GGBFS ratio to achieve the maximu m unconfined strength and CBR values is 1: 5 and the maximu m measured UCS and CBR values are for samp le with 3% lime and 15% GGBFS.
Due to the large volume of GGBFS wh ich is produced as a waste material in the world, GGBFS can be considered as an economical and valuable material with lots of positive effects to increase the engineering properties of soils.