1 Introduction

Ultra-high performance concrete (UHPC) is known for its extremely high mechanical and durability characteristics [1, 2], which facilitate the use of this class of materials in various infrastructure and rehabilitation applications worldwide [3]. The enhanced mechanical and durability performance of UHPC can further increase the sustainability of infrastructure components [4]. While UHPC applications range from building components and cladding systems to structural repairs, rehabilitation solutions, and energy sector infrastructure, the use of UHPC has gained prevalence in the transportation infrastructure sector, particularly in the construction of bridges [5]. The increasing applications of UHPC in various sectors and the increasing number of commercial UHPC suppliers have led to a pressing need to investigate and define the material behaviors of UHPC-class materials.

The mixture compositions of UHPCs typically make the materials in this class prone to early-age autogenous shrinkage that could cause micro cracking, thereby affecting its long-term durability performance. Also, UHPCs are commonly produced and supplied with the dry constituents preblended and prebagged, thus limiting the knowledge that the user will have regarding the specific details of the constituent types and volumes. Recognizing that accurate determination of early-age autogenous shrinkage is often crucial in ensuring the enhanced durability performance of UHPC-class materials, there is a need for guidance on the relevance and usage of test methods whose results speak to set time and related early age behaviors.

1.1 Background

The current practice by the industry is to evaluate early-age autogenous shrinkage of cementitious materials in accordance with ASTM C1698 [6]. ASTM C1698 requires determining final setting time to evaluate early-age autogenous shrinkage accurately. Additionally, prediction of and field measurement techniques for the assessment of UHPC’s early-age properties, such as setting time, are necessary for production scheduling, material quality control, and life safety during early construction operations. Production scheduling includes formwork removal; removal of formwork before the concrete attains sufficient strength can lead to structural distress caused by the demands on the structure exceeding the still-developing mechanical resistance of the material [7]. For conventional concretes, it is suggested that formwork is removed when the concrete has attained a compressive strength of 6.9 MPa; this strength level is typically attained in 24 h for common concretes, 12 to 15 h for high early strength concrete, and 48 h for concrete with high volume slag or fly ash [8]. However, the development of early-age properties of UHPC-class materials is different than in conventional concretes due to a very low water-to-cementitious materials ratio (w/cm), presence of supplementary cementitious materials (SCMs), high dosages of chemical admixtures, mineral admixtures, and fiber reinforcement.

The initial set represents approximately the time when fresh concrete can no longer be properly mixed, placed, and consolidated [8]. The final set represents the time at which strength begins to develop at a significant rate. The physical mechanisms of cement hydration progression include stiffening, setting, and hardening [9, 10]. Stiffening and setting characteristics are largely dependent on aluminate hydration, whereas hardening is influenced by silicate hydration [8]. The setting phenomena are affected by the aluminate-to-sulfate balance in the solution phase of a hydrated Portland cement paste, where the imbalance in the aluminate-sulfate ratio can result in a quick or flash set [11]. Setting time can also be influenced by the water-to-cementitious ratio and particle size distribution [12, 13].

Commercially available proprietary UHPCs commonly include a well-graded aggregate system of very fine particles such as silica flour and other fillers as well as a combination of cementitious materials such as cement, silica fume, and other supplementary cementitious materials in variable amounts. The specific type and proportions of the above mentioned constituents are unknown, making it challenging to definitively characterize the hydration behavior. However, it is well known that the presence of high amounts of SCMs can delay the hydration of concrete [14, 15], and UHPCs tend to contain high amounts of SCMs [16]. While the enhanced mechanical and durability properties of UHPC-class materials can be attributed to the presence of high amounts of SCMs in the system [17, 18], previous research showed that the setting time of UHPCs—when measured using the penetration test—can range from 6 to 40 h depending on the material compositions, SCMs quantities, and admixture contents [19,20,21]. Even with unknown composition, UHPCs are essentially composite cementitious materials which have the same underlying hydration mechanisms as any other cementitious material. However, to assess the setting time of cement-based materials, Vicat and penetration tests [22, 23] are extensively used; and the use of Vicat and penetration test methods for UHPCs presents some challenges which are explained later in the paper. As such, this study is focused on assessing the setting behavior of UHPC through reactivity and microstructural development based alternate test methods and compared the results to that of standard test methods. As discussed below, alternate test methods can provide greater insight into the setting behavior of some cement-based materials.

1.2 Literature

The setting time of cement-based materials is traditionally measured using the Vicat needle and penetration resistance tests [22,23,24,25], with the use of these standard methods having sometimes been extended to UHPCs. However, the standard test methods may not be ideal due to the arbitrary threshold values chosen in these test methods. The arbitrary values do not mark a specific change in the physical–chemical characteristics but are purely functional points where the initial set defines the limit of handling; the final set defines when the pressure of 27.6 MPa from a flat-headed needle can no longer penetrate through the concrete surface which, in some cases, is the beginning of mechanical strength development [8]. While the Vicat and penetration resistance tests tend to empirically identify initial and final set times for normal-strength concretes, their principles may not be applicable to UHPCs. UHPCs tend to stiffen rapidly such that they appear to be setting. However, the stiffening does not necessarily coincide with mechanical property development for UHPCs. Therefore, these tests might underpredict the set time for UHPCs.

In the available literature, researchers have used non-traditional (and non-standardized) approaches such as isothermal calorimetry, semi-adiabatic calorimetry, chemical shrinkage, autogenous shrinkage, and ultrasonic pulse velocity to assess the setting time of concrete [26,27,28,29]. For normal concrete mixtures, the setting times evaluated using the non-standardized approaches provided accurate correlations with the initial and final setting times assessed using the standard test methods [26,27,28,29].

To predict setting time, isothermal calorimetry uses the evolution of heat release from the cement hydration reactions, while semi-adiabatic calorimetry uses the change in material temperature due to the hydration of cementitious materials [30]. Early-age hydration of cement is a highly exothermic process that includes an initial period, a dormant phase, an acceleratory period, and a deceleratory period. The initial period includes a rapid, instantaneous heat release upon mixing cement with water resulting from the aluminates’ and sulfates’ dissolution [9]. A dormant period follows during which the aluminates’ solubility is restricted by the presence of sulfate, leading to the formation of ettringite [9]. The acceleratory period is of particular focus because it is associated with the hydration of silicates (particularly tricalcium silicate and dicalcium silicate C3S and C2S, respectively, in cement chemistry notation) to form a solid matrix assisting with the development of mechanical properties. The onset of the acceleratory period is commonly associated with the initial set, and the final set occurs closer to the peak of the acceleratory period [31].

In normal-strength concretes, the cement paste retains most of its plasticity prior to the onset of the acceleratory period. At this time, the cement paste starts to solidify, leading to the initial set [32]. The final set is concurrent with complete solidification and occurs before the absolute maxima of the heat in the calorimetry response [32]. Previous research evaluated setting time using isothermal calorimetry based on a derivative approach in which the first derivative of heat evolution identified the final setting times of concretes [33,34,35]. The initial set occurs when the first derivative curve reaches its highest value, where the increase in the rate of heat generation is the fastest [33, 35]. After the point of the initial set, the decrease in the derivative value indicates the decrease of the rate of heat generation and is associated with the final set time [33, 35].

Similar to isothermal calorimetry, semi-adiabatic calorimetry can be used to predict the setting behavior of cement-based materials [36]. ASTM C1753 approximates initial and final set times using thermal profiles [36]. The hydration kinetics associated with isothermal calorimetry can also be associated with semi-adiabatic calorimetry. However, when comparing the heat of hydration curve from isothermal calorimetry to the temperature change curve from semi-adiabatic calorimetry, the onset in the semi-adiabatic curve can be attributed to the heat exchange between the specimens and the test environment. In other words, some portion of heat is lost to the environment during semi-adiabatic calorimetry, as opposed to the highly insulated environment of isothermal calorimetry. According to ASTM C1753 [36], the initial and final set times determined by ASTM C403 [23] occur at 20 and 50% fractions, respectively, of the main temperature peak response. However, these fractions of the temperature response are subject to change depending on the material [36], as in the case of UHPCs.

Autogenous shrinkage can be described as the volume change in the material due to a drop in internal relative humidity caused by hydration reactions and their accompanying chemical shrinkage. Autogenous shrinkage is generated because of the self-desiccation process that takes place in the paste component of concrete materials [37]. Autogenous shrinkage is more significant in concretes with a low w/cm (e.g., UHPCs) [38]. Autogenous shrinkage occurs after the final set and is a major factor causing early-age cracking and premature damage in concrete structures [39]. Autogenous shrinkage testing, as measured in ASTM C1698, uses a time-zero value, which occurs around the final set, to differentiate between the shrinkage strains developed during the liquid phase and the solid phase. Time-zero is defined as the moment at which the material is able to resist the internal tensile stresses, occurring immediately after the transition of the material from liquid to solid state [40]. To accurately determine autogenous shrinkage, researchers have considered several approaches to determine this time-zero phase transition point [28, 39,40,41,42]. The transition point, which is referred to as the inflection point throughout this manuscript and in other literature, is when the autogenous strain rate significantly changes [40]. The several approaches used in literature to determine time-zero include identifying an inflection point in the autogenous shrinkage measurements using a corrugated tube [41, 43], obtaining the peak of the rate of deformation from the total shrinkage strain measurements [39], determining the final set using the Vicat needle and penetration test [42], and assessing the deviation of chemical shrinkage from autogenous shrinkage strains [28, 43].

Chemical shrinkage is the absolute internal volume change due to the hydration of cementitious materials. Chemical shrinkage occurs while the system is still in its liquid phase (prior to the final set) and is not yet at risk of early-age cracking [38]. Previous research demonstrated that, prior to the set, chemical shrinkage and autogenous strain were similar [44, 45]. As the material transitions from a liquid to a solid phase, a rigid skeleton forms in the cement paste that resists shrinkage, resulting in a divergence of autogenous and chemical shrinkage strains [28].

As described in the previous paragraph, the solid network formed because of hardening in the cementitious systems can resist internal tensile stresses. In other words, the development of a solid network (as a result of the final set) can be related to the ability of the material to generate stresses [46]. Previous research demonstrated that residual stress development implies the development of finite elastic properties and a solid skeleton that resists the stress in a cementitious system [47, 48]. The development of stresses in the solid network can be monitored using the restrained ring or dual ring test (DRT) in accordance with AASHTO T 363 [49]. While there was no guidance on determining a set time from DRT, previous literature revealed a significant change in stress around the final set time [46].

For mixtures with w/cm ratios ranging from 0.36 to 0.60, literature correlated standard and non-standardized setting time evaluation techniques [26, 28, 39, 42]. However, these same correlations for materials with low w/cm (< 0.25), unknown constituents, and mixture proportions of commercially available proprietary UHPCs have not yet been tested. Applying the same correlations to UHPCs observed in normal concrete materials may result in improper identification of the early-age properties, such as setting time and shrinkage development.

2 Objectives

The research presented herein has two objectives. The first objective is to identify accurate test method(s) to assess the setting behavior of UHPC-class materials. Seven different test methods were assessed, including standard and non-standardized approaches, to evaluate the setting time of UHPC. The standard approaches include Vicat needle and penetration tests, while the non-standardized approaches include isothermal calorimetry, semi-adiabatic calorimetry, autogenous shrinkage, chemical shrinkage, and DRT. Upon identifying the test method(s) to accurately determine the setting time of UHPCs, the second objective is to compare the standard and non-standardized testing techniques and to provide guidance on the use of setting time approaches in laboratory and field applications.

3 Materials

Three proprietary, commercially available UHPC materials were used in this study. The UHPC materials are identified herein as U-D, U-H, and U-J; consistent with the nomenclature used at Turner-Fairbank Highway Research Center [5]. The UHPCs consisted of preblended, prebagged granular and powdered mixtures containing cementitious components, inert fillers, chemical admixtures, and steel microfiber reinforcements. They were mixed following recommendations and proportions from each manufacturer. The steel microfibers were dosed at 2% by volume. The steel microfibers had a manufacturer-reported tensile strength of 2,850 MPa, with a nominal length of 13 mm and a nominal diameter of 0.2 mm. The materials exhibited a static flow measured in accordance with ASTM C1856 [50] in the range of 216–254 mm. Electrical resistivities measured in accordance with AASHTO TP 119 [51] were greater than the established threshold of 1500 Ω·m [52]. Compressive strengths were determined in accordance with ASTM C1856 [50] using cylindrical specimens measuring 76 mm in diameter and 152 mm in height and measuring at least 124 MPa at 28 days. Mixture proportions and properties of each of the commercially available proprietary UHPCs tested are provided below in Table 1.

Table 1 Mixture proportions and properties of the commercially available proprietary UHPCs used in this study
Full size table

The three UHPC materials tested as a part of this study were chosen to represent the variety of commercially available UHPC products on the market. Each UHPC material represents a different anticipated time of set, classified as fast, medium, and slow based on the reactivity. Reactivity, using isothermal calorimetry, was assessed, which is explained further in the results and discussion section below. Despite fibers being a typical component of UHPCs, the presence of fibers could provide additional challenges for determining set times when using specific tests requiring small material quantities. Therefore, fibers were not included in the specimens evaluated for the time of set for any of the testing described herein.

4 Experimental methods

Testing techniques to assess the time of set are classified into three categories: mechanical, reactivity, and microstructural development. As described in the objectives, the time of the set was evaluated through standardized and non-standardized approaches using standard test methods. Mechanical testing involved the use of Vicat and penetration tests [22, 23]. Reactivity testing was conducted using isothermal [53] and semi-adiabatic calorimetry [36]. Microstructural development testing was conducted using chemical shrinkage [54], autogenous shrinkage [6], and DRTs [49]. Mechanical testing involved the use of standard approach to assess time of set while reactivity and microstructural development involved the use of non-standard approaches which are explained further under each test method.

4.1 4.1. Mechanical testing

4.1.1 Vicat test

Vicat is a standardized test method used for measuring the setting times of cement pastes and mortars which assumes the limit of handling to be the initial set and the beginning of mechanical strength development to be the final set [55]. UHPCs were evaluated for setting time following method B of ASTM C191 [22] using an automatic Vicat needle apparatus. Upon mixing the UHPCs in accordance with the manufacturer’s recommendations, fresh material was placed into the specimen mold of the Vicat needle apparatus. The specimen mold consists of a removable base plate and a conical ring with an internal diameter of 60 ± 0.5 mm at the top, 70 ± 0.5 mm at the bottom, and a height of 40 ± 1 mm. The needle used for measuring penetration is 1 ± 0.05 mm in diameter and of length at least 50 mm. The UHPC material was placed into the specimen mold, and a flat trowel was used to finish the test surface. The excess material was carefully removed from around the mold.

Due to the longer setting times reported in literature for UHPC-class materials [19,20,21], the time to first penetration was delayed by at least 60 min, and the corresponding penetration measurements were taken at 15-min intervals. The test was performed at a temperature of 23 ± 4 °C and a relative humidity of at least 50%. Data obtained from the automatic Vicat apparatus were used to determine the initial and final set times. The timeframe of the test was then corrected to account for the delayed start time to the first penetration. The initial set time was then determined using Eq. 1 [22].

$$\text{Initial set} = \left(\left(\frac{H-E}{C-D}\right)\times \left(C-25\right)\right)+E$$
(1)

where H is time to the first penetration where the needle penetrates less than 25 mm into the sample (in minutes); E is time to the last penetration where the needle penetrates more than 25 mm; C is the penetration reading at time E; and D is the penetration reading at time H.

The initial set was reported to the nearest one minute. The final set was determined by the first penetration measurement that did not mark the specimen surface with a complete circular impression and was reported to the nearest five minutes. Due to the limitations of the testing equipment, only one specimen was evaluated initially. Results obtained from each of the UHPCs were later verified using a second set of specimens.

4.1.2 Penetration test

Similar to the Vicat test, the penetration test is another standardized mechanical test used to determine the time of the setting of concrete mixtures; it is completed in accordance with ASTM C403 [23]. While the penetration test is relatively simple to perform, it uses an arbitrary method to define setting time. The test measures the fresh material’s mechanical resistance to the penetration of a loaded pin. ASTM C403 defines the initial and final sets as the times when penetration resistance equals 3.5 MPa and 27.6 MPa, respectively [23].

In preparation for the test, freshly mixed UHPC material is placed in a single layer into two rigid, watertight, non-absorptive cylindrical containers 152 mm in both diameter and height. Upon placing the UHPC in both containers, a tamping rod (16 mm in diameter and 600 mm in length) with a hemispherical tip was used to rod the specimens to remove any entrapped air. After rodding, the sides of the containers were gently tapped using the tamping rod to close any voids left by the tamping rod and to level the surface of the specimens further. The UHPC surface was maintained 10 mm below the top edge of the cylindrical containers, as recommended in ASTM C403 [23]. The containers were then covered using a plastic sheet until the first penetration measurement to avoid the loss of moisture due to evaporation. The specimens were stored at a temperature of 23 ± 2 °C.

A standard set of six needles with bearing areas 645, 323, 161, 65, 32, and 16 mm2 were used on the loading apparatus to determine penetration resistance. To determine the depth to which the needle can penetrate the specimen, the needle shank consists of a circumferential scribe mark at a distance of 25 mm from the bearing area. Depending on the degree of setting of the material and the time elapsed since mixing, an appropriate needle size was selected to determine the first penetration resistance measurement. The distance between the penetration measurements and the side of the container was at least 25 mm but not more than 50 mm. The appropriate needle was inserted into the loading apparatus, and the needle’s bearing surface was brought into contact with the surface of the specimen. A uniform force was gradually applied to the specimen until the needle penetrated a depth of 25 mm, as indicated by the scribe mark. The force and time of measurement were recorded, and the penetration resistance was calculated by dividing the force by the bearing area of the needle used. The distance between the subsequent measurements was at least twice the diameter of the needle size used but not less than 15 mm. Testing continued until the penetration resistance was equal to or greater than 27.6 MPa.

Using linear regression analysis, a log–log graph of elapsed time versus penetration resistance was plotted to obtain the initial and final setting times. The regression constants and the equation obtained from the curve fitting were used to determine the initial and final set times corresponding to penetration resistances of 3.5 MPa and 27.6 MPa, respectively. Additionally, for verification, non-logarithmic data of elapsed time versus penetration resistance were plotted, and setting times were determined based on visual readings.

4.2 Reactivity testing

4.2.1 Isothermal calorimetry

Isothermal calorimetry is most frequently used for studying cement hydration and can be used to quantify hydration kinetics [56]. Heat release from the hydration reactions captured by isothermal calorimetry can be used to obtain the heat flow and cumulative heat release from the hydration reaction, which are further used to characterize the early-age hydration phases. Heat flow of UHPC materials was evaluated using an eight-channel isothermal calorimeter in accordance with ASTM C1679 [53].

Before mixing UHPC, the calorimeter was set to a temperature of 23 °C. A baseline heat was measured on graded standard sand, which was used as the inert reference specimen to reduce noise in the measurements, and the empty mass of each of the two glass ampoules was recorded. The time of initial contact with water and UHPC was recorded. The ampoules were then filled with approximately 15 to 20 g of freshly mixed UHPC material using a small funnel to ensure no material touched the upper sides of the ampoule. The mass of each ampoule filled with UHPC was measured to determine the weight of the material in the ampoule. The ampoules were then sealed tight with a metal cap to prevent moisture loss so that the endothermic heat of evaporation would not influence the measurements. The heat flow measurements began recording immediately after lowering the ampoules into the calorimeter. After 3 days of recording heat flow data, the test was terminated, and the data were collected and processed to obtain the rate of heat evaluation. The data were corrected to account for the signal correction time of 45 min of the equipment and the initial mixing time. As mentioned in literature, a derivative-based approach was used to determine setting times of UHPCs where the initial set was found to occur when the first derivative curve reaches its highest value (i.e., where the increase in the rate of heat generation is fastest) [33, 57]. After the point of the initial set, the decrease in the derivative value indicates a deceleration of the rate of heat generation and is associated with the final set time [33].

4.2.2 Semi-adiabatic calorimetry

While isothermal calorimetry measures heat release, the semi-adiabatic calorimeter measures the change in the temperature of the specimens due to hydration reactions. Thermal profiles evaluated using semi-adiabatic calorimetry can be used to assess the early-age hydration performance of cementitious materials in accordance with ASTM C1753 [36]. The thermal profile of each UHPC material was evaluated using a four-channel semi-adiabatic calorimeter. Each channel can fit cylindrical specimens of dimensions 102 mm in diameter and 203 mm in height. Two cylindrical molds were filled with inert material of graded standard sand and were placed in the calorimeter. Prior to mixing UHPC, a baseline temperature of the inert material was measured for 30 min. The UHPC was mixed according to the manufacturer’s instructions, and the time of initial contact with water was recorded. Upon completion of mixing, the initial temperature of the UHPC was recorded, and the material was placed in two layers into two cylindrical molds 102 mm in diameter and 203 mm in height. Each layer was rodded to remove any entrapped air using a tamping rod (16 mm in diameter and 600 mm) in length with a hemispherical tip. The specimens were lowered into the semi-adiabatic calorimeter and the thermal profiles were recorded for up to 48 h using the equipment’s data collection software. After 48 h, the test was terminated, and the data were extracted for processing. The obtained data were corrected to account for the mixing time. Corrected temperature data were plotted against time, and the initial and final set times were obtained in accordance with the guidance provided in ASTM C1753 [36]. According to ASTM C1753 [36], initial and final set times determined in accordance with ASTM C403 [23] occur at 20 and 50% fractions, respectively, of the main temperature peak response.

4.3 Microstructural development testing

Some of the approaches used in this paper to determine time-zero include identifying (1) the inflection point in the autogenous strain measurements from the corrugated tube, and (2) the point of divergence between chemical and autogenous strains.

4.3.1 Autogenous shrinkage

As described in literature, the inflection point resulting from the change in the rate of autogenous strain due to the solidification of the cementitious matrix is a set time indicator [28]. To evaluate UHPC set times, autogenous shrinkage deformations were assessed in accordance with ASTM C1698 [6]. As seen in Fig. 1, three sealed, corrugated plastic tubes of length 420 ± 5 mm and diameter 29 ± 0.5 mm were used to evaluate autogenous shrinkage deformations. The corrugates minimized restraint in the longitudinal direction. The distance between the corrugates was 5.8 ± 0.2 mm. The tubes were prepared by sealing one end using an end plug of length 19 ± 0.5 mm and a diameter tapered from 21 ± 0.1 mm to 22.4 ± 0.1 mm. The prepared tube was then placed in a plastic support and transferred to a vibrating table to be filled with UHPC. Freshly mixed UHPC was slowly poured into the corrugated tube while undergoing constant vibrating, tapping, and rodding to remove any entrapped air. After filling each tube, the end was sealed using an end plug. Both ends of the tubes were sealed using aluminum tape to avoid moisture loss. The initial mass and length of each tube were measured, and the filled tubes were moved to the automated test setup where each end of the tube rested on a spring-loaded linear variable differential transformer (LVDT). The LVDTs were connected to a data acquisition system that recorded the tube deformations every five minutes for 3 days. Testing period of 3 days was selected to account for all the deformations in the materials as they transitioned from a fresh to hardened state. Testing was conducted in an environmental chamber with a temperature of 23 ± 1 °C and a relative humidity of 50 ± 5% to maintain isothermal conditions. After 3 days, the test was terminated, and the final mass and length of each specimen were measured. The data extracted from the data acquisition system were processed to convert deformations into strain using Eqs. 2–4. The strains were plotted against time, and the point of inflection was determined to predict the setting time of UHPC.

$${L}_{ref}={L}_{i}-2\bullet {L}_{plug}$$
(2)
$$L\left(t\right)={L}_{ref}+{LVDT}_{1}(t)+{LVDT}_{2}(t)$$
(3)
$${\varepsilon }_{autogenous}=\frac{L\left(t\right)-L({t}_{i})}{L({t}_{i})}\bullet {10}^{6}\mu m/m$$
(4)

where Lref is the reference length in mm, Li is the initial length measured prior to recording deformations data in mm, Lplug is the length of the end plugs in mm, L(t) is the length of the specimens at a given time in mm, LVDT1(t) is the LVDT 1 reading at a given time in mm, LVDT2(t) is the LVDT 2 reading at a given time in mm, L(ti) is the length of the specimens at the beginning of the test in mm, εautogenous(t) is the autogenous strain at time t expressed as µm/m.

Fig. 1
figure 1

Source: FHWA

Corrugated plastic tubes to assess autogenous shrinkage in accordance with ASTM C1698 [6].

Full size image

4.3.2 Chemical shrinkage

Chemical shrinkage is the reduction of internal volume change due to the hydration of cement prior to the time of set [58]. Prior to setting, chemical, and autogenous shrinkage strains are similar [28, 44, 45]. Upon setting, autogenous volume reductions in the cementitious matrix are restricted by the formation of a rigid microstructure [59]. As chemical reactions keep occurring between the cementitious materials and water, the related volume reduction of the solid fraction causes the formation of vapor-filled space within the pores of the solid matrix [45, 60] resulting in a divergence of autogenous and chemical shrinkage strains. The time of divergence between the recorded chemical and autogenous shrinkage values can be used to determine the final set time [28, 46]. As such, UHPCs were evaluated for chemical shrinkage in accordance with procedure A of ASTM C1608 [54].

Upon mixing the UHPC, the glass vial was filled with a small amount of material up to a height of 5 to 10 mm. The UHPC in the glass vial was consolidated using a small vibrating table to get rid of any entrapped air. The glass vials are 22 mm in diameter and 55 mm in height. The glass vial was filled with water to the top and was tightly fitted with a rubber stopper that was attached to a graduated capillary tube with graduations of 0.01 mL. The capillary tube was filled with water from the glass vial. The water level was adjusted such that it was closer to the top mark of the graduated capillary tube. Using a long syringe, a few drops of paraffin oil were added on top of the water to prevent moisture loss during testing and to monitor the volume change occurring within the UHPC during the testing period. Two replicates were tested for chemical shrinkage in accordance with ASTM C1608 [54]. Test specimens were transferred to a constant temperature water bath at 23 °C such that the tops of the vials were above the water level in the bath. An initial measurement of the testing time and height of the water level in the graduated tube was recorded, followed by consecutive measurements of time and height of the water level over 48 h. Around 48 h, the test was terminated after collecting final measurements. The data extracted from the testing was processed to obtain chemical shrinkage using Eqs. 5–7.

$${M}_{sample}={M}_{vial+paste}-{M}_{empty vial}$$
(5)
$$CS\left(t\right)=\frac{h\left(t\right)-{h}_{initial}}{{M}_{sample}}$$
(6)
$${\varepsilon }_{CS}\left(t\right)=CS\left(t\right)\bullet {10}^{6}\frac{{cm}^{3}}{{m}^{3}\bullet g sample}$$
(7)

where Msample is the mass of sample in the vial in g, Mvial+paste is the mass of the glass vial with the added sample in g, Mempty vial is the mass of empty vial in g, CS(t) is the chemical shrinkage at time t in (mL/g sample), h(t) is the water level in capillary tube at time t in mL, hinitial is the initial water level in capillary tube at the beginning of the test in mL, and εcs(t) is the chemical shrinkage strain at time t expressed as cm3/m3⋅g sample. Chemical shrinkage was determined as volumetric strain and, assuming isotropic volume change, volumetric strain was converted to linear strain by dividing by a factor of three to be compared against autogenous strain [61]. Strains obtained from chemical shrinkage and autogenous shrinkage tests were plotted against time, and the point of divergence between the two strain curves was used to predict setting time [28, 46].

4.3.3 Dual ring test (DRT)

DRT was developed to evaluate the shrinkage cracking propensity of cementitious materials [62] and is an extension to the commonly used single ring test described in ASTM C1581 [63]. As described in literature, upon setting, stresses develop in cementitious materials due to the formation of a rigid micro-skeleton which restrains volume change [28]. As such, for this study, DRT was used to evaluate early-age stress development in UHPC materials using guidance provided in AASHTO T 363 [49]. As seen in Fig. 2, the test setup consists of a heavily insulated chamber with two volumetrically stable restraining rings made of low thermal expansion iron-nickel alloy placed on a non-absorbent base formwork. The rings were designed to provide a degree of restraint of approximately 72%. UHPC material was cast between the inner and outer rings, which were placed at an equal distance of 38 mm. The rings were instrumented with four strain gages that were equally spaced from each other to measure the strain developed in the rings. Two thermocouples were connected to each of the rings, and one was embedded in the UHPC material. The strain gages and the thermocouples were connected to a data acquisition system that recorded the strains and temperatures every 200 s. The temperature of the setup was maintained at 23 ± 1 °C throughout the test using copper tubing inside the insulated box setup that connected to an external water bath consisting of a water/ethyl-glycol mixture. The strains were recorded for a minimum of 48 h and a maximum of 96 h, depending on the anticipated time of set. The recorded strains were used to calculate the stress development in the sample in accordance with the guidance provided in AASHTO T 363 [49]. Stresses were plotted against time, and setting times were determined as the time at which stresses began to develop significantly.

Fig. 2
figure 2

Source: FHWA

DRT test setup.

Full size image

5 Results

5.1 Mechanical testing

5.1.1 Vicat test

Results presented in Table 2 provide information about setting times of commercially available UHPCs (U-D, U-H, and U-J) determined using the Vicat needle test run in accordance with ASTM C191 [22]. According to the Vicat needle test, U-H and U-J exhibit an initial set of 4 h and a final set of around 7 h, while U-D exhibits an initial set of around 6.5 h and a final set of around 9.5 h.

Table 2 Setting times of UHPCs determined in accordance with ASTM C191 [22]
Full size table

5.1.2 Penetration test

Figure 3 presents the log–log plot of penetration resistance versus elapsed time determined using the penetration test run in accordance with ASTM C403 [23]. The best-fit equations representing the collected data were determined through regression analysis, used to obtain the initial and final setting times of U-D, U-H, and U-J, and are reported in in Table 3. According to the penetration test, UHPCs U-D, U-H, and U-J exhibited initial set times of 8, 3.5, and 6.5 h, respectively, and final set times around 10, 6, and 14 h, respectively. For U-J, the penetration resistance at 11 h was found to be 13.8 MPa. Penetration resistance obtained up to 11 h was plotted on a logarithmic scale against time, and the trendline was used to obtain a final set time of 14 h. When evaluated at 24 h, U-J exceeded the penetration resistance of 27.6 MPa indicating that the final set occurred between 11 and 24 h.

Fig. 3
figure 3

Source: FHWA

Log–log plot of penetration resistance of UHPCs over time measured in accordance with ASTM C403 [23].

Full size image
Table 3 Setting time of UHPCs determined in accordance with ASTM C403 [23]
Full size table

5.2 Reactivity testing

5.2.1 Isothermal calorimetry

UHPC reactivity was evaluated using isothermal calorimetry in accordance with ASTM C1679 [53]. Figure 4a presents the heat flow over time in all UHPCs tested. For normal-strength concretes, the dormant period lasts a few hours before the setting begins [64]. For construction practices, the dormant period is crucial as it controls the time in which the concrete is workable and placeable [65]. As seen in Fig. 4a, the dormant period lasts 9, 5, and 20 h from the time of mixing for U-D, U-H, and U-J, respectively. Based on the observations from the dormant period of isothermal calorimetry, the UHPCs used in this study are classified as fast (U-H), medium (U-D), and slow reacting (U-J). Note that the dormant period observed in the UHPCs is longer than that of conventional concrete and the delayed dormant phase of UHPCs can be attributed to material compositions, SCM quantities, and admixture contents. The end of the dormant period and the onset of accelerated hydration is associated with the initial setting of concrete where the concrete is no longer workable [32]. As such, the initial set times of U-D, U-H, and U-J from isothermal calorimetry are 9, 5, and 20 h, respectively. The initial set times of U-D, U-H, and U-J are marked on Fig. 4a.

Fig. 4
figure 4

Source: FHWA

Reactivity of UHPCs measured using isothermal calorimetry in accordance with ASTM C1679 [53], showing (a) heat flow over time along with the initial and set times determined using the derivative method and (b) cumulative heat release over time.

Full size image

Figure 4b represents cumulative heat release in the UHPCs tested; cumulative heat release prior to the initial set as determined using isothermal calorimetry was very low, followed by an increase in heat upon the initial set. The increase in cumulative heat is significant between 12 to 18 h for U-D, 6 to 12 h for U-H, and 30 to 36 h for U-J, which might be a result of the material reaching the final set. The final set times of U-D, U-H, and U-J are marked on Fig. 4a. However, the approach used to determine final set times is discussed later in this section.

Literature describes the final set as occurring during the acceleratory phase of isothermal calorimetry [8], and it is associated with the hydration of silicates (particularly C3S) to form a solid matrix assisting with the development of mechanical properties. Figure 5 represents the derivative of the heat flow curve for U-D, U-H, and U-J. As defined in literature, the final set time is when the first derivative drops to zero, indicating a decrease in the rate of heat generation. For U-D, the final set was determined to be 16.5 h. The same approach was applied to U-H and U-J, and the final set times were determined to be 10.5 and 32.5 h, respectively.

Fig. 5
figure 5

Source: FHWA

Derivative of heat evolution curves for (a) U-D, (b) U-H, and (c) U-J.

Full size image

5.2.2 Semi-adiabatic calorimetry

UHPCs were evaluated for reactivity using semi-adiabatic calorimetry in accordance with ASTM C1753 [36]. Figure 6 presents the change in temperature of U-D, U-H, and U-J over 48 h and the thermal profiles used to assess setting times. According to ASTM C1753, initial and final set times occur at 20 and 50% of the main temperature peak response and were arbitrarily established using data obtained from penetration resistance testing in accordance with ASTM C403 [23]. Based on ASTM C1753, U-D exhibited an initial set time of 9.5 h and a final set time of 11.5 h; U-H exhibited an initial set time of 5.5 h and a final set time of 6.5 h; U-J exhibited an initial set time of 24.0 h and a final set time of 27.0 h. The setting times obtained using semi-adiabatic calorimetry differed from the setting times obtained using isothermal calorimetry. The differences between the setting times obtained in the reactivity testing is further explained in the Discussion section.

Fig. 6
figure 6

Source: FHWA

Thermal profiles of UHPCs evaluated using semi-adiabatic calorimetry in accordance with ASTM C1753 [36] and the initial and set times determined using the guidance provided in Appendix of ASTM C1753 [36].

Full size image

5.3 Microstructural development testing

5.3.1 Autogenous shrinkage

UHPCs were evaluated for autogenous shrinkage in accordance with ASTM C1698 [6]. While the data was measured over 72 h, the deformations were plotted over 48 h to be able to clearly visualize the inflection points. Figure 7 presents the total deformation evaluated in U-D, U-H, and U-J over 48 h, and the collected data were used to evaluate set time (also referred to as time-zero). Time-zero is defined as the moment in time when the material is able to resist tensile stresses that occur right after the transition of the material from a liquid to a solid state [28, 45, 66, 67]. Autogenous shrinkage occurs after the final set [39], and therefore time-zero is used to assess concrete’s autogenous shrinkage. Though several approaches have been used in literature to determine time-zero, this study uses the inflection point from the total deformation measurements obtained using the corrugated tubes [41, 43].

Fig. 7
figure 7

Source: FHWA

Total deformations obtained using autogenous shrinkage testing in accordance with ASTM C1698 [6] for UHPCs (a) U-D, (b) U-H, and (c) U-J. Dashed lines in the figures are tangents drawn to evaluate inflection points.

Full size image

As stated in the literature, inflection point is a point on the autogenous deformations curve when the strain rate changes significantly or in other words when the slope of the curve changes abruptly. In this case, an inflection point occurs at the maximum change in slope of the total measured deformation due to the transition of material from a liquid to a solid state [41, 43]. As observed in Fig. 7, the inflection points with respect to time occurred at 16 h for U-D and 10 h for U-H. While U-D and U-H exhibited one inflection point, U-J exhibited two inflection points with the first at 9.5 h and the second at 32.5 h. Two inflection points displayed by U-J could potentially be a result of different phases of the cementitious system reacting at different times.

5.3.2 Chemical shrinkage

To confirm the time-zeroes obtained using the point of inflection approach from the autogenous shrinkage testing, UHPCs were further evaluated for chemical shrinkage to be compared against autogenous shrinkage using the point of divergence approach as described in the literature [28]. UHPCs were evaluated for chemical shrinkage in accordance with ASTM C1608 [54]. Because the chemical shrinkage data cannot be used alone to determine set times, the chemical shrinkage data were plotted against autogenous shrinkage data and used to determine set time, as seen in Fig. 8. As described in the literature, point of divergence is another approach used to determine set time. Autogenous and chemical shrinkage are equivalent while the paste is fluid. As the system begins to set, the formation of a rigid skeleton restricts autogenous shrinkage, though the chemical shrinkage continues to increase in the form of vapor-filled space in the concrete matrix [28]. The times at which chemical and autogenous shrinkage measurements deviate from each other were 16 h for U-D, 10 h for U-H, and 9.5 h for U-J. In case of U-D, upon obtaining the point of divergence, chemical shrinkage data collection was terminated at 33 h.

Fig. 8
figure 8

Source: FHWA

Deviation between chemical and autogenous shrinkage deformations of UHPCs (a) U-D, (b) U-H, and (c) U-J.

Full size image

5.3.3 DRT

DRT is traditionally evaluated in accordance with AASHTO T 363 [49] and was developed to evaluate the shrinkage cracking propensity of cementitious materials [62]. This study employed the use of DRT to evaluate early-age stress development in UHPC materials. Upon setting, stresses develop in cementitious materials due to the formation of a rigid micro-skeleton that restrains shrinkage. Data from the DRT were used to assess the time at which internal stress starts to develop in the UHPC mixtures and compare it to the setting time estimated through the other techniques. Figure 9 displays the residual stress development (secondary y-axis) in U-D, U-H, and U-J obtained from the DRT and plotted against their autogenous shrinkage measurements (primary y-axis) to validate the setting times obtained from the inflection point and point of deviation approaches. The horizontal gray line in Fig. 9 represents 0 MPa in stress, and the vertical arrow represents the time at which stresses began to change significantly. Additionally, the time at which stresses began to change significantly was found to coincide with the inflection points identified using the autogenous shrinkage test. As seen in Fig. 9, residual stresses began to change significantly beyond 13 h for U-D and beyond 10 h for U-H. For U-J, the stresses began to decrease at 32 h indicating expansion followed by increasing stresses beginning at 37 h. Further discussion on validating the setting times based on DRT is provided in the next section.

Fig. 9
figure 9

Source: FHWA

Evolution of stress measured using DRT compared to autogenous shrinkage in UHPCs (a) U-D, (b) U-H, and (c) U-J.

Full size image

6 Discussion

Based on the results of this study, Table 4 summarizes the initial set times determined for each material, and Table 5 summarizes the final set times for each material. Discussion on the initial and final set times obtained using standard and non-standardized approaches is provided thereafter.

Table 4 Summary of initial set times determined for each UHPC
Full size table
Table 5 Summary of final set times determined for each UHPC
Full size table

Mechanical testing involved the use of Vicat and penetration test methods. As previously stated, setting times were measured using the Vicat needle test in accordance with ASTM C191 [22]. In the case of UHPCs, U-H and U-J, set times determined using the Vicat needle test were found to be similar to those of a Portland cement concrete where the initial set typically occurs 45 min after mixing and the final set occurs no later than 6.5 h after mixing [68]. However, U-D displayed longer setting times than U-H and U-J.

Literature reported that the final setting time of UHPCs determined that using penetration tests ranged from 6 to 40 h depending on the mixture composition [19,20,21]. The setting times determined using the penetration test in accordance with ASTM C403 [23] were found to be different than the setting times determined using the Vicat needle test, and the set times observed in the UHPCs during this study were in the lower range of those observed in literature. Similar trends were observed in the final set times. Low w/cm ratio and high cementitious contents in UHPCs can lead to rapid stiffening. Additionally, the formation of a hardened layer on the surface with an elephant skin-like texture was commonly observed in UHPCs [69]. This hardened surface layer can prevent the Vicat needle from penetrating the UHPC and can incorrectly suggest that the materials have been set. The size and shape of needles used in the penetration test also can make the needles less sensitive to penetration compared to the Vicat needle. As such, the differences observed in the final set times between the Vicat, and penetration tests can be attributed to the sensitivity of the testing equipment. Among the traditional test methods, data obtained using the penetration test compare better to the alternate test methods described in the following paragraphs.

Reactivity testing involved the use of isothermal and semi-adiabatic calorimetry test methods. Isothermal calorimetry was performed in accordance with ASTM C1679 [53]. Initial set times observed from isothermal calorimetry for fast (U-H), medium (U-D), and slow reacting (U-J) UHPCs were 5, 9, and 20 h, respectively, while the final set times determined using penetration resistance test for fast (U-H), medium (U-D), and slow reacting (U-J) UHPCs were 6, 10, and 14 h, respectively. Visual observations of UHPC revealed that the material appears to lose workability and begins to stiffen rapidly within an hour from mixing, which was hypothesized to be a result of rapid consumption of water resulting in a change in rheological properties of UHPCs. However, the cause of the false or flash set was not explored further due to the unknown compositions of the proprietary materials used in this study. Additionally, early stiffening of UHPCs could have caused the material to withstand the resistance of the needles used in the penetration test resulting in the final set times presented by the penetration test in this study.

The setting times obtained using semi-adiabatic calorimetry, according to the guidance provided in ASTM C1753 (i.e., based on the relation with ASTM C403) for fast (U-H), medium (U-D), and slow reacting (U-J) UHPCs, differed from the setting times obtained using the penetration resistance test. As such, guidance provided in ASTM C1753 to evaluate the setting times may not be appropriate for UHPC-class materials. Peak temperature responses for U-D, U-H, and U-J occurred at 16, 10, and 33 h, respectively, and were found to coincide with the final set times obtained using isothermal calorimetry for UHPCs U-D, U-H, and U-J, which were 16.5, 10.5, and 32.5 h, respectively. as seen in Fig. 10. Additionally, the initial set times determined from isothermal calorimetry occurred at the onset of the thermal peak. However, the setting times obtained using reactivity testing were further evaluated using the methods employed in microstructural development.

Fig. 10
figure 10

Source: FHWA

Peak temperature responses of UHPCs evaluated using semi-adiabatic calorimetry.

Full size image

UHPCs were evaluated for autogenous shrinkage in accordance with ASTM C1698 [6]. As described in the previous section, the inflection point approach was used to determine time-zero. In comparing the inflection point data to the reactivity testing data, for U-D and U-H, the time at which the inflection points occur coincides with the final set times determined using isothermal calorimetry and the peak temperature response of semi-adiabatic calorimetry. In the case of U-J, the time at which the second inflection point occurs (i.e., 32.5 h) coincides with the final set time determined using isothermal calorimetry and the peak temperature response of semi-adiabatic calorimetry.

Two inflection points displayed by U-J could potentially be a result of different phases of the cementitious system reacting at different times. In comparing U-J data from autogenous shrinkage testing to isothermal calorimetry, the time at which the first inflection point occurs is at 9.5 h, which was during the dormant period as displayed in the isothermal calorimetry plot of U-J. Additionally, in the case of U-J, the evolution of heat between 9.5 h and 32.5 h indicated a slower degree of reactivity. After 32.5 h, the evolution of heat increased due to the hydration of certain cementitious phases, resulting in a more rigid cementitious matrix. The autogenous shrinkage data for U-J displayed a change in slope between 9.5 and 32.5 h, which was higher than the change in slope beyond 32.5 h. Literature revealed that upon set time or time-zero, the change in autogenous strain rate decreases significantly due to the solid matrix resisting the developing shrinkage [28].

To further investigate this result, a visual inspection and compressive strength test (performed in accordance with ASTM C1856 [50]) were performed on U-J. U-J was cast into a cylinder 76 mm in diameter and 152 mm high. The cylinder was demolded at 24 h for visual inspection and appeared hardened. However, when tested for compressive strength immediately after the visual inspection, the material immediately crumbled and achieved a compressive strength of less than 3.45 MPa. Based on the above observations, the set time or time-zero for U-J is considered the time at which the second inflection point occurs, i.e., 32.5 h. Due to the use of proprietary UHPC materials in this study, the cause of the two inflection points was not explored further.

To confirm the time-zeroes obtained using the point of inflection approach from autogenous shrinkage testing, UHPCs were further evaluated for chemical shrinkage in accordance with ASTM C1608 [54] to be compared against autogenous shrinkage using the point of divergence approach as described in the literature [28]. For U-D and U-H, the time at the point of deviation agrees with the final set times determined using isothermal calorimetry and time-zero as determined using the inflection point from autogenous shrinkage. However, the time at the point of deviation of 9.5 h for U-J was identified closer to the first inflection point obtained from the autogenous shrinkage data, which occurred many hours prior to the secondary inflection point in which strength development was limited. Results from the point of deviation between chemical and autogenous shrinkage data for U-J revealed microstructural development began from the point of deviation. Previous research reported that the point of deviation happened upon the final set [28]. With U-J being a proprietary material like other UHPCs used in the study, the phenomenon behind the observed differences in the time of set from the point of deviation compared to the other tests (isothermal calorimetry, semi-adiabatic calorimetry, inflection point from autogenous shrinkage) is challenging to explain. The differences can, however, be attributed to the material composition due to the presence of different SCM types, SCM contents, SCM reactivity, and admixture contents. Further research is necessary to confirm the effect of material composition on microstructural development in slow-setting materials.

Based on the calorimetry data, for U-J, the heat release at 9.5 h is found to be very low. Isothermal calorimetry displayed a final set of 32.5 h, which coincides with the peak temperature response of semi-adiabatic calorimetry and the second point of inflection from autogenous shrinkage testing. Additionally, the compressive strength of U-J at 24 h displayed minimal strength (less than 3.45 MPa) revealing that the material did not reach a final set prior to 24 h. Due to the delayed reactivity of U-J compared to other UHPCs, the point of inflection may result from the initial set in U-J in contrast to the final set as observed in U-D and U-H. Therefore, the first point of inflection observed in U-J may be a result of the initial setting and the second point of inflection may be a result of the final setting. Further research is necessary to confirm the initial and final setting behaviors of U-J. The differences observed between the inflection point and point of deflection approaches when estimating the setting time were further evaluated using DRT.

DRT was evaluated in accordance with AASHTO T 363 [49], and as shown in Fig. 9a, stresses began to increase at 13 h. Based on both the inflection point and point of deviation approaches, U-D exhibited a final set time of 16 h. The strain rate obtained from autogenous shrinkage testing is plotted against time in Fig. 11 along with a shaded region ranging from 13 to 16 h. The shaded region is based on the results obtained in the microstructural development testing. Rate of strain data revealed an inflection point at 13 h, as seen in Fig. 11; a significant change in strain beginning at 13 h and lasting until 16 h (as seen in the shaded region) beyond which the strain rate reduced. Additionally, the rate of heat flow obtained using isothermal calorimetry for U-D revealed that maximum reactivity occurred between 13 and 18 h with the final set at 16 h. Considering the data obtained using alternate test methods, the increase in stresses starting at 13 h in U-D can be attributed to microstructural development due to hydration reactions. The final set time for U-D is thus considered to be 16 h.

Fig. 11
figure 11

Source: FHWA

Autogenous shrinkage strain rate versus time in U-D.

Full size image

Similar to U-D, U-H exhibited a final set time of 10 h based on both the inflection point and point of deviation approaches. Stress development, as shown in Fig. 9b, also began to change at 10 h. Additionally, as shown in Fig. 12, a significant change in autogenous shrinkage strain between 9.5 and 11 h (as shown in the shaded region) beyond which the rate of change was reduced, and the rate of heat flow obtained using isothermal calorimetry for U-H revealed that maximum reactivity occurred between 10 and 11 h. From the data obtained using alternate test methods, the final set time for U-H was determined to be 10.5 h.

Fig. 12
figure 12

Source: FHWA

Autogenous shrinkage strain rate versus time in U-H.

Full size image

In the case of U-J, as shown in Fig. 9-c, stress at 9.5 h (first point of inflection) was 0.14 MPa followed by a minimal change in stress until 32 h. A second inflection point occurred at 32.5 h, after which the stresses were observed to decrease indicating an expansion of the material. The expansion phase lasted until 37 h beyond which the material exhibited shrinkage. As described in the literature, the ability of the material to generate stresses can be related to the development of a solid network (as a result of the final set) [46], and the change in stress starting at 32 h is indicative of solid network formation. Figure 13a displays the change in autogenous strain versus time. The vertical lines in Fig. 13-a represent time markers at 9.5 and 32 h. As shown in Fig. 13a, there was an initial increase in autogenous shrinkage strain rate until 12 h followed by no change until 30 h. The time axis was scaled down in Fig. 13b to closely visualize the strain rate between 24 and 48 h. As seen in the shaded region of Fig. 13b, the strain rate approached zero between 30 and 34 h. Separately, the rate of heat flow obtained using isothermal calorimetry for U-J revealed that the maximum reactivity occurred between 32 and 34 h with the final set suggested to occur at 32.5 h. As such, based on the alternate test data, the final set time of U-J was determined to be 32 h. A summary of the approaches used in each test method to determine set time is provided in Table 6.

Fig. 13
figure 13

Source: FHWA

Autogenous shrinkage strain rate versus time in (a) U-J from 0 to 48 h, and (b) U-J during 24 to 48 h.

Full size image
Table 6 Summary of the approaches used in each test method to determine set time
Full size table

7 Conclusions

Three commercially available UHPCs were evaluated using standard and non-standardized testing techniques to assess setting behavior. The standard approaches included Vicat needle and penetration test methods, while the non-standardized approaches included isothermal calorimetry, semi-adiabatic calorimetry, autogenous shrinkage, chemical shrinkage, and DRT methods. The initial reactivity of UHPCs was assessed using isothermal calorimetry; of the three UHPCs, U-H, U-D, and U-J were classified as fast, medium, and slow reacting, respectively. Based on the results obtained, the following conclusions can be drawn:

  • Due to the proprietary nature of the UHPCs and the differences in mixture compositions between individual UHPCs, the setting behavior is expected to widely vary, and the use of a single test method to evaluate the setting behavior of UHPCs can be challenging. The findings from this study suggest use of a combination of test methods depending on the purpose of the setting time assessment and the practicality of applying a particular test method. Further discussion is provided in the recommendations section.

  • The use of isothermal calorimetry was successful in providing insights into the reactivity-based setting behavior of UHPCs. When using isothermal calorimetry, a derivative-based approach can be used to determine the setting time of UHPC.

  • Final set times determined using isothermal calorimetry were found to coincide with the peak temperature response observed through semi-adiabatic calorimetry. Based on these findings, the final set time of UHPCs can be estimated as the time of the peak temperature as measured through semi-adiabatic calorimetry.

  • The guidance provided in ASTM C1753 [36] was established based on the correlations between peak temperature response from semi-adiabatic calorimetry and penetration resistance tests for normal strength concretes. Based on the test results obtained in this study testing UHPCs, no correlation was found between the penetration resistance and semi-adiabatic calorimetry test methods, and the ASTM C1753 [36] prediction was found to be inaccurate. As such, the guidance provided in ASTM C1753 [36] to estimate the setting times from semi-adiabatic calorimetry may not be appropriate for UHPCs.

  • The findings from reactivity-based testing were further evaluated using microstructural development testing. These tests, including autogenous shrinkage, chemical shrinkage, and DRT methods, can be used to assess the fundamental response of the cementitious system during the setting process.

    1. o

      The setting times obtained using microstructural development testing for fast (U-H) and medium (U-D) reacting UHPCs were found to correlate well with the setting time determined in accordance with isothermal and semi-adiabatic calorimetry.

    2. o

      When evaluating the slow-reacting UHPC (U-J), the use of a single test method to determine the setting time was found to provide insufficient information. Different setting times were observed depending on the test method employed. When using autogenous shrinkage measurements, the point of inflection approach was successful in providing insights into the setting behavior; however, the change in autogenous strain over time should also be considered more broadly. The time at which the autogenous strain rate approaches zero was found to coincide with the last inflection point on the autogenous shrinkage curve. As such, the time of set was taken as the time corresponding to the last inflection point.

    3. o

      While the point of divergence approach between autogenous and chemical shrinkage curves provided insights on the setting behavior of certain UHPCs (U-D and U-H), when used by itself, the approach might not be accurate in determining the setting behavior of other UHPCs (e.g., U-J) without further validation using other approaches like DRT.

  • The DRT results captured and evaluated in accordance with AASHTO T 363 [49] were found to be relevant for assessing the setting behavior of UHPCs. While this test method does not offer a quantitative process for determining set time, the time corresponding to a significant change in stress coincides with the setting time determined using isothermal calorimetry, semi-adiabatic calorimetry, and the point of inflection approach from autogenous shrinkage tests. As such, the time of set was taken as the time at which stresses began to significantly develop.

  • The use of standard test methods (Vicat needle and penetration test) to determine the setting time of UHPCs may not be appropriate. Test results indicated differences between the final setting times obtained using Vicat and penetration test methods. The final set times observed using the mechanical tests can be attributed to the rapid hardening due to the change in rheological properties within the first few hours after mixing as caused by rapid consumption of available water and high SCM contents present in the UHPCs. These standard test methods failed to capture the actual setting behavior of UHPCs.

8 Recommendations

The non-standardized approaches investigated in this paper were able to provide an estimation of UHPC set times. Based on the findings herein, the setting time of UHPCs can be evaluated using alternate test methods such as isothermal calorimetry, semi-adiabatic calorimetry, autogenous shrinkage, chemical shrinkage, and DRT. The recommendations provided herein are based on the data obtained from this study. The end user should be cognizant of the nuances of individual test methods and consider alternative methods to confirm the validity of the setting time data obtained from an individual test method. While different alternate test methods were used to obtain the setting times of UHPCs, each of them can be used for different applications as follows:

  • For field measurements of UHPC set time, the change in temperature of the UHPC material can be monitored for 48 h using a semi-adiabatic calorimeter in accordance with ASTM C1753 [36]. This test method is relatively simple to execute, and the equipment is generally available. The final setting time can be determined as the time corresponding to the peak temperature of the semi-adiabatic calorimetry response. Peak temperature response from semi-adiabatic calorimetry can be further used in scheduling operations such as formwork removal. Note however that, given the limited number of different UHPCs assessed through this study, further testing to verify the broad applicability of this recommendation to field applications is needed.

  • To identify time-zero in the context of shrinkage strains that result in tensile stress development in the UHPC, the autogenous shrinkage strain point of inflection method should be considered, accompanied by a broader assessment of the change in autogenous strain over time. The time at which the strain rate approaches zero was found to coincide with the last inflection point on the autogenous shrinkage response curve. As such, time-zero can be taken as the time corresponding to the last point of inflection.