The effect of aggregates with high gypsum content on the performance of ultra-high strength concretes and Portland cement mortars

Construction and Building Materials 110 (2016) 346–354

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

The effect of aggregates with high gypsum content on the performance of ultra-high strength concretes and Portland cement mortars Mehmet Gesoglu a,⇑, Erhan Güneyisi a, Ali H. Nahhab a,b, Halit Yazıcı c a

Department of Civil Engineering, Gaziantep University, Gaziantep, Turkey Department of Civil Engineering, Babylon University, Babylon, Iraq c _ Department of Civil Engineering, Dokuz Eylül University, Izmir, Turkey b

h i g h l i g h t s
Finding sand with a normal sulfate level is a major challenge in the Middle East.
Natural sand with a high gypsum content was used in UHSC and PC mortar.
UHSC with a compressive strength of 120–150 MPa was produced.
The strength of steam cured UHSC enhanced with a higher gypsum content.
PC mortar deteriorated with the presence of expansive ettringite.

a r t i c l e

i n f o

Article history: Received 6 January 2016 Received in revised form 15 February 2016 Accepted 16 February 2016

Keywords: Ultra-high strength concrete (UHSC) Sulfates Expansion Compressive strength XRD

a b s t r a c t There is an increasing demand to ultra-high strength cement-based materials in the Middle East despite finding fine aggregates with a normal SO3 level is a major challenge. This study was conducted to investigate the effect of increasing gypsum content of natural river sand on the properties of water and steam cured ultra-high strength concretes (UHSCs) and Portland cement mortars. All concrete and mortars were prepared with a w/c ratio of 0.197 and 0.440, respectively; yielding 28-day compressive strength ranges of 120–142 and 43–70 MPa, respectively. The experimental tests were expansion, compressive and splitting tensile strengths, and X-ray diffraction at varying ages. UHSC and mortar exhibited significant difference in resistance to internal sulfate attack. While UHSCs were not significantly affected by increasing gypsum content of sand, the Portland cement mortars deteriorated as seen by a drop in strength, a significant swelling, and the presence of expansive ettringite. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Aggregates of high gypsum content Most of the aggregates in the Middle East contain high amounts of sulfates, particularly gypsum. For instance, a survey of aggregate in the central and southern Iraq revealed that the most of sulfates in sand took the form of gypsum, which represents 95% of sulfates and the rest are sodium, magnesium, and potassium sulfates [1]. The SO3 content in these aggregates is as high as 0.75–3% by weight. Calcium sulfate, usually gypsum is added to the clinker at an optimum level to control the setting characteristics of cement. Gypsum reacts with C3A to form primary ettringite. This type of ⇑ Corresponding author. E-mail address: [email protected] (M. Gesoglu). http://dx.doi.org/10.1016/j.conbuildmat.2016.02.045 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.

ettringite happens in a plastic matrix and does not cause any damage. However, when the excessive amount of gypsum is present (from modern cement or gypsum-contaminated aggregates) [2], it will still exist within the hardened matrix and during curing of concrete the extra gypsum dissolves under water and SO4 ions release. Then, the gypsum reacts with remaining C3A and monosulfate hydrate according to Eqs. (1) and (2) to form delayed ettringite. The potential deteriorations of concrete caused by these reactions are described as internal sulfate attack.

þ

C3 A tricalcium aluminate

 12 C4 ASH monosulfate

þ

 2 2CSH gypsum

 2 3CSH

þ

gypsum

þ

16H water

26H water

!

!

C6 AS3 H32 ettringite ð1Þ

C6 AS3 H32 ettringite

ð2Þ

347

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

Some international standards specify limits for SO3 content in aggregate. For example, IQS 45 [3] allows 0.5% and 0.1% of SO3 in fine and coarse aggregate, respectively. In the complimentary British Standard to BS EN 206-1 [4], it is reported that the maximum allowable SO3 content in fine aggregate is 1%. However, due to the scarcity of aggregates with low sulfate contents in the Middle East region, many studies have been conducted to investigate the use of aggregate for conventional concrete and mortar with SO3 content above the specified limits given by the international standards. In the study by Crammond [5] mortar made either with ordinary Portland cement or sulfate resistant Portland cement was incorporated with coarsely crystalline gypsum of up to 5% by weight of fine aggregate. According to the results, the allowable gypsum content in aggregates used in ordinary Portland cement mortars would not be much larger than 2.5% by weight of aggregate. For sulfate resistant Portland cement, this limit appeared to lie at a level of about 5%. Al-Rawi et al. [6] found that the effect of internal SO3 on the compressive strength and expansion of concrete depended on the source from which it generated. The SO3 from sand had less effect than that from cement and SO3 generated from coarse aggregate had less effect as compared to that from sand. Kheder and Assi [7] investigated internal sulfate attack resistance of different concrete mixes with compressive strengths of between 15 and 75 MPa. In these mixes, sand with SO3 content ranging from 0.5% to 2% was used. It was found that increasing the compressive strength of concrete resulted in enhancing its resistance to internal sulfate attack, and concrete mixes with a high strength (larger than 45 MPa) had a comparable expansion in water regardless of the sulfate content of sand. Atahan and Dikme [8] found that the use of different types of mineral admixtures such as silica fume, fly ash, and ground granulated blast furnace slag was a remedy for the problem of internal sulfate attack on mortars containing gypsum-contaminated aggregates from Iraq. Rodríguez et al. [9] utilized gypsum-contaminated fine recycled aggregates having 2.9% SO3 in the mortars made with ordinary Portland cement or sulfate resistant Portland cement. Their results indicated that the expansion of both mortar types did not exceed the limit given in the literature, 0.1%. Overall, the previous studies indicate that a certain amount of gypsum present in the aggregate can be tolerated for ordinary concrete and high strength concrete. However, higher gypsum content can be used with high strength concrete due to its lower permeability. This opens up the possibilities of using aggregate from the Middle East in the concrete with a very low permeability, UHSC which is the objective of the present study. 1.2. Ultra-high strength concrete Ultra-high strength concrete (UHSC) is characterized by a higher strength, a much lower permeability and a denser microstructure as compared to the ordinary concrete. This is typically achieved by using silica fume, applying heat treatment, and using superplasticizers so that making possible to use a very low water/cement ratio. Apart from the pozzolanic reaction with lime, silica fume can fill the voids between the next larger class particles, cement leading to a dense material [10,11]. Moreover, the addition of silica fume densifies the packing in the interfacial transition zone, such that the porosity in this region is significantly reduced [12]. Superplasticizer also allows the cement grains to pack more uniformly, reducing the porosity of the paste, and thus improving density [13]. The effect of heat treatment, on the other hand, is to improve cementitious matrix because the pozzolanic reaction of silica fume is activated and the pore size is diminished due to the application of heat treatment [14,15]. Many investigators dealt with the behavior of UHSC after subjecting to various curing regimes such as steam curing and water curing. Yazıcı [16], for

example, showed that the compressive strength of UHSC was higher under steam curing compared to water curing. However, no attention was given to the effect of curing methods on the strength of UHSC made with aggregate of high gypsum content which is the main source of aggregate in the Middle East. 1.3. Research significance The present study aims at investigating the effects of using a fine aggregate with gypsum content higher than the normal limits in UHSC mixtures having a 28-day compressive strength exceeding 120 MPa. The other important parameter in this study is the influence of curing method, namely water curing and steam curing on the properties of UHSC with and without additional gypsum. The UHSC samples were tested for compressive strength, splitting tensile strength and expansion to monitor the composition-induced internal sulfate attack. The experimental results were supported by XRD analyses. In addition, the comparison was made between UHSCs and Portland cement mortars with respect to compressive strength, swelling, and XRD analysis because both of the materials contain no coarse aggregate. This would assist in a better understanding of the behavior of the UHSCs. 2. Experimental 2.1. Materials Natural river sand (0–4 mm) and commercial quartz sand (0.6–1.2 mm and 1.2– 2.5 mm) with a specific gravity of 2.66 and 2.65, respectively were used as fine aggregates. The natural crushed gypsum with the SO3 content of nearly 38% was used as a partial substitution of the natural river sand to raise the original gypsum content of 0% to the desired contents of 1.68%, 3.66%, 7.61%, and 11.55%. Both crushed gypsum and natural river sand had a similar grading. A type F polycarboxylate-based superplasticizer (SP) in accordance with ASTM C494 [17] was used. Portland cement (CEM I 42.5 R) with C3A content of 8.8% was used. Silica fume was also used in the present work. The properties of the cement and silica fume are presented in Table 1. 2.2. Mixture proportioning of UHSCs and Portland cement mortars UHSC is typically characterized by a high level of silica fume and a very low water/cement (w/c) ratio [16]. The UHSC mixtures were designed with a constant w/c ratio of 0.197 and a constant silica fume content of 13% by weight of Portland cement. The high amount of silica fume is essential to optimize the filling performance and increase the compacted density of UHSC [15]. However, the hydration reaction of cement in UHSC is incomplete due to a very low w/c ratio and so the

Table 1 Properties of Portland cement and silica fume.

a b

Item

Portland cement (PC)

Silica fume (SF)

SiO2 (%) Al2O3 (%) Fe2O3 (%) CaO (%) MgO (%) SO3 (%) K2O (%) Na2O (%) Cl (%) Loss on ignition (%) Insoluble residue (%) Free CaO (%) Specific surface (m2/kg) Specific gravity Compounds C3S (%) C2S (%) C3A (%) C4AF (%)

19.69 5.16 2.88 62.12 1.17 2.63 0.88 0.17 0.0093 2.99 0.16 1.91 394a 3.15

90.36 0.71 1.31 0.45 – 0.41 1.52 0.45 – 3.11 – – 21,080b 2.2

Blaine specific surface area. BET specific surface area.

56.9 13.8 8.8 8.8

348

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

amount of incorporated silica fume is more than required by the pozzolanic reaction [15]. All mixes were designed at a high flow of 270 ± 10 mm to overcome challenges during casting of UHSC in practice. The preliminary trial mixes indicated that the use of quartz sand with the same amount as that of natural river sand was necessary to obtain ultra-high strength and high workability. Five UHSC mixes were made with five different gypsum contents of 0%, 1.68%, 3.66%, 7.61%, and 11.55% by weight of the natural river sand, yielding SO3/cement ratios of 2.76%, 3.24%, 3.8%, 4.91%, and 6.02%. Three Portland cement mortar mixes were prepared with 0% gypsum and the two largest gypsum contents of 7.61% and 11.55%, yielding SO3/cement ratios of 2.78%, 6.75%, and 8.82%. The sand/cement and w/c ratios for mortars were 2.75 and 0.44, respectively. Details of both UHSC and Portland cement mortar mixes are given in Table 2. The designation assigned to the mixes was UH-x, and M-x which UH and M denote the UHSC and Portland cement mortar, respectively while x is the gypsum content of natural sand.

2.4.3. Splitting tensile strength test The splitting tensile strengths of UHSCs were determined in accordance with BS 1881-117 [25] using 50 mm cube samples at a rating load of 0.2 kN/s. The test was conducted on three samples for each mix at three different ages, namely 28, 90, and 180 days. 2.4.4. Expansion test Expansions in water for both UHSCs and mortars were measured on three (25  25  285 mm) bars on which the initial comparator readings were made prior to the permanent storage in water by means of a standard length comparator with a digital display accurate to 0.001 mm. Additional comparator readings on the prismatic bars were made periodically up to 365 days. 2.4.5. XRD test X-ray diffraction (XRD) was performed on powder samples at the end of the curing period of 365 days via SHIMADZU XRD-6000 diffractometer using Cu Ka radiation with a wave length of 1.5405 Å. Similar to Ramlochan et al. [26], the XRD sample were prepared by crushing the sample to a fine powder with a mortar and pestle and thereafter passing the specimen through a 90 lm sieve to minimize the amount of quartz aggregate.

2.3. Sample preparation and curing Mortars and UHSCs were mixed in a Hobart mixer. Mortars were mixed according to ASTM C305 [18]. The mixing procedure of UHSC was as follows: dry ingredients of aggregate, PC, and SF were mixed at a low speed of 140 ± 5 rpm for 5 min. Water was added thereafter, and mixing resumed at the low speed for another 5 min. Superplasticizer was added to a premix and mixed at the same low speed for 5 min. Finally, the mixing was resumed for 2 min at a medium speed of 285 ± 10 rpm. The mixtures were then poured into the molds and compacted by vibration. The prepared specimens were covered with polyethylene sheets and then left in the laboratory environment for about 24 h before demolding. After demolding, the UHSC samples were subjected to either water curing or steam curing. In water curing condition, the specimens were stored in water at 22 ± 2 °C until testing. In steam curing condition, the specimens were kept in water for 24 h and then subjected to heat curing at 80 °C for 48 h at a heating rate of almost 11 °C/min. In other words, heat treatment was not started until the second day after casting according the suggestions of Richard and Cheyrezy [19]. Then, steam-cured samples were also stored in water at 22 ± 2 °C until the testing age. Heating concrete to a high temperature (without steam) accelerates hydration reactions, but it leads to the evaporation of water which is also necessary for the hydration process. So, one can conclude that heating gypsum-incorporated concrete may decompose gypsum to hemihydrite (CaSO4 0.5H2O) and then to soluble anhydrite (CaSO4) due to the evaporation of chemical water of gypsum. However, this is not the case when concrete is subjected to steam curing because instead of evaporation of water, vapor-saturated atmosphere can supply water to the system [20]. Thus, steam curing cannot dehydrate the gypsum but can increase the rate of reaction of gypsum with C3A to form further ettringite than that formed at normal temperature [21]. The Portland cement mortars were cured under water only.

3. Experimental results and discussions 3.1. Expansion As mentioned before, in the region of the Middle East it is difficult to find aggregate with an allowable SO3 content. The excess sulfates in the aggregate may cause an undesirable expansion and hence deterioration of concrete. Fig. 1a–c show the expansion in water over a period of 365 days for water-cured UHSCs, steam cured UHSCs, and Portland cement mortars, respectively. As seen in Fig. 1a, all the water cured UHSC samples showed an increase in expansion during the first 11– 14 days, but thereafter their longitudinal expansion started to drop, most likely due to self-desiccation, and then they underwent shrinkage except for mix with the highest gypsum content which showed a net increase in volume during the entire curing period. In other words, the shrinkage of the UHSC mix with 11.55% gypsum was wholly compensated by expansion. As shown in Fig. 1b, the expansion of steam-cured prismatic bars developed rapidly in the early ages, but after that the expansion rate decreased, and then stabilized. In contrast to the water-cured samples, none of the steam-cured samples experienced a fall in expansion thus confirming that the application of heat treatment ignored the selfdesiccation observed in the normally cured samples. One possible explanation of this finding may be that most of the evaporable water, which was necessary for self-desiccation, may have been consumed due to the increase in hydration reactions through steam curing. It can be seen from Fig. 1c that the mortar mix with 7.61% gypsum (M-7.61) experienced a steep development in the expansion in the first 28 days beyond which it showed a slowdown in the rate of expansion up to 210 days followed by a gradual gain in expansion. On the other hand, the 11.55 mortar mix (M-11.55) showed

2.4. Testing methods 2.4.1. Slump flow test The flow of UHSC was measured using the mini- slump cone. The truncated mini cone had a lower diameter, upper diameter, and height of 100, 70, and 60 mm, respectively. After pouring fresh mixtures in the cone to full capacity, the cone was lifted straight upwards to allow free flow on a plate. Actually, the slump flow test is a simple and widely used test method for highly flowable concrete [22,23]. Through the use of an adequate amount of superplasticizer, all the UHSC mixtures were designed to reach a flow value of 270 ± 10 mm. 2.4.2. Compressive strength test Compressive strengths of the specimens were measured on 50 mm cubes at a rating load of 0.9 kN/s according to ASTM C109 [24] using a digital testing machine of 3000 kN capacity. The compressive strengths of UHSCs and mortars were investigated at the ages of 28, 90, 180, and 365 days. Each result is the average of three samples. Table 2 Mix proportions of UHSCs and Portland cement mortars. Type

Mix code

Gypsum% by weight of natural sand

w/c

Material (kg/m3) PC

SF

0–4 mm natural sand

0.6–1.2 mm quartz sand

1.2–2.5 mm quartz sand

Water

SP

Ultra-high strength concrete

UH-0 UH-1.68 UH-3.66 UH-7.61 UH-11.55

0 1.68 3.66 7.61 11.55

0.197

824

107

611

270

341

162

50 50 50 52 54

Portland cement mortar

M-0 M-7.61 M-11.55

0 7.61 11.55

0.440

545.9

0

750.6

331.7

418.9

240.2

0 0 0

349

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354 0.35

0.03

Maximum Expansion (%)

(a)

Expansion (%)

0.02

0.01

0

-0.01

Water-Cured UHSCs

Steam-Cured UHSCs

Cement Mortars

0.3 0.25 0.2 0.15 Safe Margin

0.1 0.05 0

-0.02

Mix Code

-0.03 0

40

80

120

160

200

240

280

320

360

Fig. 2. Maximum expansion of water cured and steam cured UHSCs with and without additional gypsum in comparison to that of Portland cement mortars.

Age (days)

(b)

0.025

Expansion (%)

0.02

0.015

0.01

0.005

0 0

40

80

120

160

200

240

280

320

360

240

280

320

360

Age (days)

(c)

0.3

Expansion (%)

0.25

0.2

0.15

0.1

0.05

0 0

40

80

120

160

200

Age (days) Fig. 1. Expansion versus age data for (a) water-cured UHSCs, (b) steam-cured UHSCs, and (c) Portland cement mortars.

a continuous increase in expansion without a decreasing tendency. Contrary to the water-cured UHSCs, none of the mortar bars showed any contraction though the control mix showed a fluctuation in the measured expansion. In conventional concretes and mortars, self- desiccation typically occurs in sealed specimens as a result of a reduction in internal relative humidity in the pores during the hydration process. However, the phenomenon may happen in the moist-cured mixtures with a very low w/c ratio, which is the case with UHSC, since such mixtures become almost impermeable after a certain curing time, allowing little or no penetration of external water [27]. In the UHSC mixes with additional gypsum, the reaction between C3A in Portland cement and SO3 in gypsum might have led to further expansion because of ettringite

formation. However, this reaction was not complete due to the lack of free water from outside and hence the sufficient expansion was not available to compensate for self-desiccation. The maximum expansion for the water-cured UHSCs, steamcured UHSCs, and Portland cement mortars is comparatively shown in Fig. 2. For 365 days of storage in water, the maximum expansion of water cured and steam cured UHSCs did not exceed 0.023% and 0.024%, respectively. On the other hand, the expansion of 7.61% and 11.55% mortars reached 0.098% and 0.287%, respectively. It is generally accepted that the formation of expansive phases such as ettringite and loss of strength are the main factors responsible for the degradation of concrete attacked by sulfates [28]. However, the mechanism of expansion is not fully understood. In this context, two different hypotheses are available namely crystal growth pressure and uniform expansion of paste. The former theory has been considered as the most cited hypothesis. Müllauer et al. [29] states that the formation of ettringite in small pores (10–50 nm) is responsible for the expansion and damage of Portland cement mortars exposed to sulfate attack. Ettringite formation in larger pores does not contribute to the expansion and deterioration because a sufficient pressure is not developed. Different expansion limits were reported in the literature. Samarai [30] suggests an expansion limit of 0.1% as a safe margin for determining the maximum SO3 percentage that can be induced into a given mixture without causing any significant deterioration. ASTM C1038 [31], however, proposes that the amount of SO3 in the cement and pozzolan, and hence in concrete can be increased by any level as long as the expansion after 14 days of storage in water is less than 0.02%. For moderate resistance, expansion limit of 0.1% after 6 months is generally accepted [2]. When the 0.1% expansion was considered as a safe margin, it could be concluded that the Portland cement mortar with 11.55% gypsum was the only mix that underwent a serious durability related problem which expanded by 0.287%, though the mortar mix with 7.61% gypsum also showed a significant expansion of 0.098% which was very close to the safe limit. The lower expansion of UHSCs was probably due to that the lower w/c ratio of UHSCs, 0.197 made the composite less permeable than Portland cement mortars of the 0.44 w/c ratio. One more possible explanation was that a higher tensile strength of UHSCs as compared to that of Portland cement mortars restrained the swelling caused by the ettringite formation.

3.2. Compressive strength The variations of compressive strength with the gypsum content for water-cured UHSCs, steam-cured UHSCs and Portland cement mortars at different ages are given in Fig. 3a–c, respectively. As seen in Fig. 3a, the application of gypsum did not affect

350

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

(a)

(b)

(c)

3.6–8.5% with respect to the control UHSC. The observed maximum increase in compressive strength at gypsum level of 7.61% under water curing and 7.61–11.55% under steam curing might be related to the rule of optimum gypsum content. It is accepted that there is an optimum gypsum content for each cement at which the compressive strength is maximum and this content is higher with increasing a curing temperature [32]. If higher gypsum content than 11.55% is used, higher amounts of sulfates are expected to react particularly at very early ages at which the curing water can penetrate the pore system. Under the water curing condition, this may lead to further expansion and decrease in compressive strength. Since UHSC is a high-quality concrete, a gypsum content of 11.55% may be a safe limit at which there is only a slight reduction in the strength of UHSC caused by the additional gypsum. However, under the steam curing condition, further research is needed to evaluate a limit of gypsum content as the investigated UHSC was not negatively affected by the addition of even 11.55% of gypsum. On the other hand, as observed in Fig. 3c, the compressive strength of conventional mortar was lower with higher gypsum contents in natural sand at all ages because of the compositioninduced internal sulfate attack. At 90 days, for example, a decrease of 10.4% and 39.1% was recorded in mortar mixes with gypsum contents of 7.61% and 11.55%, respectively as compared to the reference mix. The compressive strengths of water cured and steam cured UHSCs are comparatively given in Fig. 4. The water cured and steam cured mixes recorded a compressive strength varied between 120.8–142.1 MPa and 131.2–147.9 MPa, respectively. For a given gypsum content, the strength was generally higher under steam curing compared to water curing at all ages as a result of accelerated pozzolanic reaction between silica fume and lime as well as due to the refinement in the microstructure of UHSC. However, the effect of increasing curing temperature was more obvious at relatively early ages and at higher gypsum contents. At 28 days, applying steam curing at 80 °C led to enhance the compressive strength of UHSC by 4.8–17.8% compared to that of water curing. As compared to the normal concrete, UHSC behaves differently when exposed to similar steam curing temperatures. It is generally reported that the application of steam curing a few hours after casting leads to enhance the early strength but adversely affects the long-term strength of normal concrete. The possible explanation of the reduction in the strength at later ages may be due to the presence of microcracks in the cement paste caused by the expansion of air bubbles and the development of tensile stresses in the surrounding cement paste [20]. However, a prolonged delay period before steam curing can reduce the deleterious effects of air bubbles because the tensile strength of concrete increases during

Fig. 3. Effect of gypsum content of natural sand on compressive strength of (a) water-cured UHSCs, (b) steam-cured UHSCs, and (c) Portland cement mortars.

Compressive Strength (MPa)

the compressive strength of water-cured UHSC appreciably though the best and the lowest strengths were achieved for mixes with 7.61% and 11.55% gypsum, respectively. The 28-day compressive strengths of mixes with gypsum contents of 7.61% and 11.55% were 2% higher and 3.5% lower, respectively, in comparison to the control mix. As seen in Fig. 3b, increasing the gypsum content had a positive impact on the compressive strength of steam-cured UHSCs at all ages. The mix with the highest gypsum content, 11.55%, showed the highest compressive strength of 142.3 MPa at 28 days, beyond which the best strength was recorded for the mix with 7.61% gypsum. Under steam curing condition, the 28-day compressive strength of gypsum-incorporated UHSC showed an increase of

155 Water Curing

Steam Curing

150 145 140 135 28 days 130

90 days

125

180 days 365 days

120

Mix Code Fig. 4. Effect of curing method on compressive strength of UHSCs.

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

the delay period [20]. In the present work, the delay was 48 h during which the UHSC developed high tensile strength. This favored the development of mechanical properties at early and later ages. The better performance of UHSC at later ages as compared to normal concrete was, therefore, due to the higher tensile strength of UHSC and the longer delay period prior to steam curing. The compressive strength of normally cured samples shown in Fig. 3a increased as the curing time prolonged up to 180 days beyond which the strength marginally diminished most likely due to self-desiccation. In the study on the long-term behavior of high strength concrete with silica fume, Hooton [33] also observed a small reduction in strength at later ages and reported that this reduction appeared to be within the usual variation in compressive strengths when there was no further increase in strength. On the other hand, Fig. 3b indicated that the compressive strength of steam-cured UHSC developed as the curing period increased from 28 to 90 days, beyond which the samples did not experience any appreciable gain in strength. In contrast to UHSC, the Portland cement mortar continued to gain strength with time during the entire curing period as shown in Fig. 3c. This was due to that the curing water could ingress into the pores from outside the mortar which led to a continuous increase in the amount of the desirable hydration product, C–S–H phases. Recently, the authors [34] have successfully developed ultrahigh performance fiber reinforced cementitious composites (UHPFRCCs) made with aggregate of high gypsum content. Two groups of mixes were prepared with binary and ternary blends of Portland cement, silica fume, and/or ground granulated blast furnace slag. All mixes were reinforced with 2% by volume micro steel fibers. For a given gypsum content, the compressive strength of UHPFRCC was higher than that of UHSC. Moreover, in contrast to UHSC, UHPFRCC with 11.55% gypsum did not show any decrease in compressive strength as compared to reference UHPFRCC, with 0% gypsum. The difference between the results of both studies may be related to the presence of micro steel fibers in UHPFRCC which bridge tensile cracks and delay their propagation. 3.3. Splitting tensile strength The measurement of tensile strength of concrete is important since tensile stresses may develop in hardened concrete because of many reasons such as drying shrinkage and temperature gradients. Splitting tensile strength is the most frequently applied method to measure the tensile strength of concrete because of the simplicity and properly of the test. Fig. 5a and b summarizes the splitting tensile strength results of water cured and steam cured UHSC, respectively. The splitting tensile strength results generally showed a trend similar to that observed in the compressive strength of the corresponding composites. The effect of adding gypsum to the UHSC mixtures on the splitting tensile strength was found to be dependent on the gypsum content in natural sand and curing regime. Of all mixes cured under water, the highest and lowest splitting tensile strengths were observed at 7.61% and 11.55% gypsum, respectively (Fig. 5a). The 7.61% mix recorded an increase of 3.4%, 3.8%, and 2.8%, compared to the control mix at the ages of 28, 90, and 180 days, respectively. The mix with 11.55% gypsum, on the other hand, experienced a reduction in splitting tensile strength of 2.8%, 1.6%, and 0.33% compared to reference mix at the ages of 28, 90, and 180 days, respectively. As seen in Fig. 5b, the splitting tensile strengths under steam curing were generally higher for mixes with additional gypsum and the best strengths were observed for mixes with 7.61–11.55% gypsum. There was an improvement of up to 6.2% in the splitting tensile strength of gypsum-incorporated mixes cured at elevated temperature over the corresponding mix not containing additional gypsum. As with the compressive strength,

351

(a)

(b)

Fig. 5. Effect of gypsum content of natural sand on splitting tensile strength of (a) water-cured UHSCs and (b) steam-cured UHSCs.

the rule of optimum gypsum content was also valid here such that there was a range of gypsum content at which the splitting tensile strength was maximum. The splitting tensile strengths of normally cured and steam cured UHSCs are comparatively presented in Fig. 6. For a given gypsum content, the effect of increasing curing temperature was to enhance the splitting tensile strength at all ages. This was due to that the pozzolanic reaction was activated and pore size was reduced through the application of steam curing. At 28 days, for example, the indirect tensile strength of steam-cured UHSC was 5.6–16.1% higher than that of water-cured UHSC depending on the gypsum content. Neville [20] states that the splitting tensile strength and compressive strength are closely related though the ratio of the two strengths relies on the strength level such that as the compressive strength rises, the indirect tensile strength also raises but at a declining rate. The results revealed that the ratio of splitting tensile strength to compressive strength for UHSC was 0.063–0.069 for mixes with and without additional gypsum. Arıoglu et al. [35] investigated the relationship between compressive strength and splitting tensile strength for concretes with the different cylinder compressive strength of between 4 and 120 MPa. They found that the splitting tensile strength was about 10% of the compressive strength at low strength levels while the ratio decreased to almost 5% at extremely higher strength levels. The relationship between the compressive strength and splitting tensile strength UHSC is given in Fig. 7. In order to show how well the data fit the statistical model, the coefficient of correlation, R2, is also given in the figures. The high coefficient of correlation (R2 = 0.7949) indicated a good correlation between the two phenomena.

352

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

Splitting Tensile Strength (MPa)

11

Water Curing

Steam Curing

10.5 10 9.5 9 8.5

28 days

8

90 days

7.5

180 days

7

Mix Code Fig. 6. Effect of curing method on splitting tensile strength of UHSCs.

Splitting Tensile Strength (MPa)

10 y = 0.0008x2 - 0.15x + 15.434 R² = 0.7949 Gypsum Content

9.5

From the foregoing discussion, it can be concluded that the UHSC with additional gypsum has a performance as high as that of UHSC without additional gypsum. So, this concrete can be used in many civil engineering applications where conventional UHSC is used. These are ultra-high strength columns, coupling beam in tall buildings, and precast units. Moreover, the thickness of concrete elements can be decreased which leads to increase the usable floor in tall buildings and decrease the dead loads. Furthermore, in bridge applications, UHSC allows girder span to increase while maintaining smaller or similar cross sections. Increasing span lengths results in fewer support structures. However, UHSC with additional gypsum may be in contact with ground water or sea water which contains sulfate ions. These ions may react with hardened cement leading to disintegration of concrete due to external sulfate attack. It is known that three conditions are essential for the external sulfate attack to take place. These are: high permeability of concrete, sulfate rich environment, and the presence of moisture. Due to the very low permeability of UHSC, the sulfate ions are not expected to ingress into the pores from outside. However, this should be established experimentally and so further research is needed in this area.

0% 1.68% 3.66%

9

3.4. XRD analysis

7.61% 11.55%

8.5

8 110

120

130

140

150

160

Compressive Strength (MPa) Fig. 7. Relationship between compressive strength and splitting tensile strength of UHSCs.

The crystalline phases detected by the XRD for the reference mixes and mixes with the largest gypsum content are shown in Fig. 8. The peaks of gypsum (G), ettringite (E), portlandite (P), larnite (L), quartz (Q), and calcite (C) are highlighted in the plot. In both Portland cement mortars and UHSCs with 11.55% gypsum, the only observed sulfoaluminate phase was ettringite (Ca6Al2(SO4)3(OH)1226H2O), detected at the 2H of 9.09°, 15.78°, and 22.94°. Generally, the more obvious peaks of ettringite were observed for Portland cement mortar mixes, remarkably at the

Fig. 8. XRD patterns of UHSCs and Portland cement mortars with and without additional gypsum. E, ettringite; G, gypsum; P, Portlandite; Q, quartz; L, larnite; C, calcite.

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

2H of 22.94°. A peak of gypsum (CaSO4 2H2O) was also observed in the XRD patterns of the Portland cement mortar and UHSC mixes with additional gypsum. This was probably related to the primary dissolution of small gypsum particles since the large gypsum particles (2–4 mm) dissolved slowly and would not react. Moreover, the gypsum peaks were greater in the UHSCs compared to Portland cement mortars. The lower w/c ratio of the UHSC than mortar caused the former more resistant to the water penetration. The lack of free water limited the reaction of gypsum with the calcium aluminates of cement. Consequently, lower ettringite peaks and higher peaks of the unreacted gypsum were observed in the UHSCs compared to Portland cement mortars. Portlandite (Ca(OH)2) was detected at the 2H of 18° in all of the investigated mixes though its intensity varied between the different mixes. As expected, the less pronounced portlandite peaks were observed for UHSC mixes because of the pozzolanic reaction of silica fume with lime in these mixes. The peaks of portlandite were larger for mixtures with 11.55% gypsum as compared to the control ones, particularly in the case of mortar mixtures. Increasing the quantity of calcium sulfate in pore solution leads to decrease calcium hydroxide [36]. However, when the crystalline gypsum is used as aggregate in concrete or mortar (as in the present study), gypsum particles interact with the alkalis of cement to form plates of Ca(OH)2 which initially concentrate around the boundaries of gypsum particles and then larger portlandite crystals grow towards the centers of particles of gypsum [5]. In other words, the presence of gypsum particles leads to the further formation of portlandite. Hussen [37] also found that the gypsum particles present in Bahrain sands reacted within concrete to form Ca (OH)2 and ettringite. Gesoglu et al. [34] observed plates of portlandite around the boundaries of undissolved gypsum grains in ultra-high performance fiber reinforced cementitious composites prepared with gypsum-contaminated aggregates (Fig. 9). Expectedly, the XRD diffractograms confirmed peaks due to larnite (Ca2SiO4) suggesting that a significant amount of cement remained unhydrated. However, the patterns of the two types of curing methods suggested that the peak of larnite was lower in the case of steam curing. Contrary to the UHSCs, there were approximately no peaks corresponding to the larnite in the case of Portland cement mortars due to their high w/c ratio. It is generally believed that, there is no unhydrated cement to be left in water-cured mixes with a w/c ratio of higher than 0.36 [20]. Peaks of calcite (CaCO3) were also identified by the XRD analysis. Some reaction may have occurred with the atmospheric CO2 during the length measurements leading to formation of calcite

Fig. 9. Micrograph of ultra high performace fiber reinforced cementitious composite showing a plate-shped calcium hydroxide around the boundaries of gypsum particle [34].

353

on the surface of the bars which was perhaps ground with the UHSC and Portland cement mortar during sample preparation. Calcite may also have formed during grinding of the samples as a result of carbonation [38].

4. Conclusions The following conclusions may be drawn from the results of this research:
Under the condition of water curing, the addition of gypsum did not have a significant effect on the compressive and splitting tensile strengths of UHSCs; however, there was a slight reduction in strengths at the largest gypsum content of 11.55%. Under the condition of steam curing, on the other hand, both types of strengths improved with a higher gypsum content of natural sand.
The steam-cured UHSCs performed better under compression and splitting tension, compared to water-cured UHSCs.
Irrespective of the curing regime, the maximum expansion of gypsum-incorporated UHSC was very low even after one year of storage in water.
Contrary to the UHSC, the Portland cement mortar samples with additional gypsum showed a poor resistance to internal sulfate attack as confirmed by a great tendency to increase in expansion and decrease in compressive strength.
XRD plots of UHSCs and Portland cement mortars with additional gypsum revealed peaks corresponding to ettringite and gypsum; however, higher peaks of ettringite and lower peaks of gypsum were observed for mortars.
XRD analysis, particularly for Portland cement mortars, also showed that the peaks of portlandite were more obvious for mixes with additional gypsum compared to the corresponding control mixes.

References [1] National Center for Construction Laboratories of Iraq. Suggestion for the determination of sulfate content in sand. In Arabic, 1981. [2] J. Skalny, J. Marchand, I. Odler, Sulfate Attack on Concrete, Modern Concrete Technology: Spon Press, 2002. [3] IQS 45. Iraqi standard specification for aggregate from natural sources for concrete and building construction. ICOSQC, Baghdad, Iraq, 1984. [4] BS 8500-2. Concrete, complementary british standard to BS EN 206-1, specification for constituent materials and concrete, British Standards Institution, 2006. [5] N.J. Crammond, Examination of mortar bars containing varying percentages of coarsely crystaline gypsum as aggregate, Cem. Concr. Res. 14 (2) (1984) 225– 230. [6] R.S. Al-Rawi, R.A. Al-Salihi, M.H. Ali, Effective sulfate content in concrete ingredients. Challenges of Concrete Construction, Vol. 6, Concrete for Extreme Conditions, 2002. pp. 499–506. [7] G.F. Kheder, D.K. Assi, Limiting total internal sulphates in 15–75 MPa concrete in accordance to its mix proportions, Mater. Struct. 43 (2010) 273–281. [8] H.N. Atahan, D. Dikme, Use of mineral admixtures for enhanced resistance against sulfate attack, Constr. Build. Mater. 25 (2011) 3450–3457. [9] G.T. Rodríguez, M. Barra, S. Pialarissi, D. Aponte, E. Vázquez, Expansion of mortars with gypsum contaminated fine recycled aggregates, Constr. Build. Mater. 38 (2013) 1211–1220. [10] P.C. Aitcin, The durability characteristics of high performance concrete: a review, Cem. Concr. Compos. 25 (2003) 409–420. [11] G. Long, X. Wang, Y. Xie, Very high-performance concrete with ultrafine powders, Cem. Concr. Res. 32 (2002) 601–605. [12] K.L. Scrivener, A.K. Crumbie, P. Laugesen, The Interfacial transition zone (ITZ) between cement paste and aggregate in concrete, Interface Sci. 12 (2004) 411– 421. [13] V.Y.G. Yanni, Multi-scale investigation of tensile creep of ultra-high performance concrete for bridge applications (PhD thesis), Georgia Institute of Technology, 2009. [14] M. Ipek, K. Yilmaz, M. Sümer, M. Saribiyik, Effect of pre-setting pressure applied to mechanical behaviours of reactive powder concrete during setting phase, Constr. Build. Mater. 25 (2011) 61–68.

354

M. Gesoglu et al. / Construction and Building Materials 110 (2016) 346–354

[15] P. Richard, M. Cheyrezy, Composition of reactive powder concretes, Cem. Concr. Res. 25 (1995) 1501–1511. [16] H. Yazıcı, The effect of curing conditions on compressive strength of ultra high strength concrete with high volume mineral admixtures, Build. Environ. 42 (2007) 2083–2089. [17] ASTM C494/C494M. Standard specification for chemical admixtures for concrete. ASTM International, West Conshohocken, PA, 2013. [18] ASTM C305. Standard practice for mechanical mixing of hydraulic cement paste and mortars of plastic consistency. ASTM International, West Conshohocken, PA, 2012. [19] P. Richard, M.H. Cheyrezy, Reactive powder concretes with high ductility and 200–800 MPa compressive strength, ACI SP 144–24 (1994) 507–518. [20] A.M. Neville, Properties of Concrete, Longman, New York, 1995. [21] R.S. Al-Rawi, Gypsum content of cements used in concrete cured by accelerated methods, J. Test. Eval. 5 (3) (1977) 231–237. [22] K. Wille, A.E. Naaman, S. El-Tawil, G.J. Parra-Montesinos, Ultra-high performance concrete and fiber reinforced concrete: achieving strength and ductility without heat curing, Mater. Struct. 45 (2012) 309–324. [23] R. Yu, P. Tang, P. Spiesz, H.J.H. Brouwers, A study of multiple effects of nanosilica and hybrid fibres on the properties of ultra-high performance fibre reinforced concrete (UHPFRC) incorporating waste bottom ash (WBA), Constr. Build. Mater. 60 (2014) 98–110. [24] ASTM C109/C109M. Standard test method for compressive strength of hydraulic cement mortars (using 2-in. or [50-mm] cube specimens). American Society for Testing and Materials. West Conshohocken, PA, 2005. [25] BS 1881: Part 117. Testing concrete. Method for the determination of tensile splitting strength. British Standard Institutions, UK, 1983. [26] T. Ramlochan, M.D.A. Thomas, R.D. Hooton, The effect of pozzolans and slag on the expansion of mortars cured at elevated temperature Part II: Microstructural and microchemical investigations, Cem. Concr. Res. 34 (2004) 1341–1356.

[27] P. Acker, Swelling, shrinkage and creep: a mechanical approach to cement hydration, Mater. Struct. 37 (2004) 237–243. [28] W. Müllauer, R.E. Beddoe, D. Heinz, Sulfate attack on concrete-solution concentration and phase stability, in: Concrete in aggressive aqueous environments, performance, testing and modeling. 3–5 June, Toulouse, France, 2009. [29] W. Müllauer, R.E. Beddoe, D. Heinz, Sulfate attack expansion mechanism, Cem. Concr. Res. 52 (2013) 208–215. [30] M.A. Samarai, The disintegration of concrete containing sulphatecontaminated aggregates, Mag. Concr. Res. 28 (96) (1976) 130–142. [31] ASTM C1038. Standard test method for expansion of hydraulic cement mortar bars stored in water. ASTM International, West Conshohocken, PA, 2004. [32] ASTM C563. Standard test method for approximation of optimum SO3 in hydraulic cement using compressive strength. ASTM International, West Conshohocken, PA, 2013. [33] R.D. Hooton, Influence of silica fume replacement of cement on physical properties and resistance to sulfate attack, freezing and thawing, and alkalisilica reactivity, ACI Mater. J. 90 (2) (1993) 143–151. [34] M. Gesoglu, E. Güneyisi, A.H. Nahhab, H. Yazıcı, Properties of ultra-high performance fiber reinforced cementitious composites made with gypsumcontaminated aggregates and cured at normal and elevated temperatures, Constr. Build. Mater. 93 (2015) 427–438. [35] N. Arıoglu, Z.C. Girgin, E. Arıoglu, Evaluation of ratio between splitting tensile strength and compressive strength for concretes up to 120 MPa and its application in strength criterion, ACI Mater. J. 103 (1) (2006) 18–24. [36] F.M. Lee, The Chemistry of Cement and Concrete, Chemical Publishing Co., New York, 1971. [37] T.K. Hussen, The natural aggregate materials of Bahrain and their influence on concrete durabihty (PhD thesis), University of London, UK, 1975. [38] T. Ramlochan, The effect of pozzolans and slag on the expansion of mortars and concrete cured at elevated temperatures (Ph.D. Thesis), University of Toronto, 2003.