1 Introduction

By definition, ultra-high performance concrete (UHPC) is an extremely durable cementitious composite material [1,2,3,4]. Interest is strong and growing in the use of UHPC in infrastructure applications subjected to aggressive environments, such as roadways, dams, and wastewater treatment facilities. The bridge sector has been at the forefront of UHPC engagement, with this material being commonly used for field-cast connections between prefabricated components; the same owners are now beginning to specify the use of UHPC for such applications as bridge deck overlays, link slabs, pretensioned girders, and piles [4,5,6]. With durability performance being a key reason for the selection of UHPC in infrastructure construction and rehabilitation projects, it is imperative that methods exist to assess the durability performance of this class of materials.

UHPC constituents 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 [1]. The dry materials are mixed with low-water-content, steel fibers, and chemical admixtures that control reaction and rheological properties (e.g., yield stress and viscosity) of the system. Typical UHPC mixtures exhibit self-consolidating behavior [7], although thixotropic UHPC mixtures have been developed for such applications as overlays [8]. As the cementitious materials in the UHPC react, a very dense microstructure forms that is typically associated with enhanced strength and durability properties when compared with more-conventional concrete mixtures [9, 10].

Even though standardized test methods exist for evaluation of the mechanical and durability properties of conventional concrete mixtures, applying the same tests to UHPC-class materials has proved more challenging. In the context of durability, that challenge is often a consequence of the extremely low permeability of UHPCs, which prevents these tests from providing meaningful insight on performance. Previous research has demonstrated that UHPC-class materials, when tested with available standards, show very low to negligible rapid chloride ion penetrability, and have good resistance to scaling, abrasion, freeze–thaw degradation, and alkali–silica reaction [10].

The aforementioned tests are reasonable indicators of the durability of many cementitious composite materials, but there has been less focus on the characterization and quantification of the transport properties in UHPC-class materials. Transport properties are important to the longevity of concrete and reinforced concrete structures because transport properties control the rates at which ions, fluids, and gases penetrate the material. Ionic diffusion, gas diffusion, liquid sorption, gas permeability and liquid permeability are associated with some of the most deleterious deterioration mechanisms observed in concrete structures [11]. Accurate determination of transport properties can facilitate service life estimation of a structure and can increase the reliability of life cycle cost analyses.

In recent years, the Federal Highway Administration has been promoting the adoption of electrical-based tests to characterize the transport properties of cementitious materials [12]. This class of tests offers the advantages of being rapid, inexpensive, and reasonably accurate in determining the microstructural quality of concrete materials [12,13,14]. Among the available electrical-based tests, resistivity has shown promise due to its ease of implementation and close relationship to fundamental microstructural properties [15,16,17,18]. The electrical resistivity of a material indicates how much resistance the material offers to the passage of an applied current. Several studies have been able to tie resistivity measurements of concretes to a list of important transport properties, such as the chloride diffusion coefficient [19,20,21] (ASTM C1556 [22]), secondary water absorption rate [19, 23] (ASTM C1585 [24]), water permeability (EN 12390–8 [25]), and rapid chloride migration coefficient [23] (NT Build 492 [26]).

2 Objectives

The main objective of this study is to identify an appropriate test method for characterizing the durability of UHPC-class materials. This study investigates electrical resistivity, water absorption, freeze–thaw, and rapid chloride migration test methods as potential durability test methods for UHPC-class materials. The applicability of each test method is evaluated according to (1) its ability to differentiate among UHPC-class materials, (2) the relevance of associated deterioration mechanisms, and (3) applicability to service life estimations.

3 Materials

3.1 Ultra-high performance concretes

The study used a series of five proprietary, commercially available ultra-high performance concrete materials. The UHPC-class materials are identified herein as U-D, U-E, U-H, U-J, and U-M, which is consistent with the nomenclature used at Turner-Fairbank Highway Research Center (TFHRC) [27]. The UHPC-class materials’ constituents included preblended granular and powdered mixtures consisting of cementitious materials and inert fillers along with chemical admixtures that were mixed following recommendations and proportions from each manufacturer. The type of admixture, admixture dosages, and measured admixtures’ conductivities for the different UHPCs are reported in Table 1. UHPCs U-D, U-E, and U-J include a three-part liquid admixture system containing high-range water reducers (HRWRs), superplasticizers (SPs), and either accelerators (ACCs) or corrosion inhibitors (CIs). U-M includes a two-part liquid admixture system containing an accelerator and a carbon nanofiber (CNF) suspension blended with superplasticizer. U-H consists of a single-part liquid admixture system that is most likely a superplasticizer. The admixture contents were reported with respect to the amount of solids present in the UHPC system and were found to range from 0.45 to 4.30%, similar to what has been reported in the literature [28,29,30,31,32,33]. The solid portion includes any cementitious materials, graded fine aggregates, and fillers that were used in the UHPC preblend. The UHPCs were dosed with 2% steel fiber reinforcement by volume. The steel fiber reinforcement had nominal length of 13 mm and nominal diameter of 0.2 mm. The tensile strength of the fibers, as reported by the manufacturer, was 2850 MPa. The steel fibers meet the standard requirements of ASTM A820: Standard Specification for Steel Fibers for Fiber-Reinforced Concrete [34]. All the UHPC-class materials exhibited a flow in the range of 200–250 mm and had compressive strengths of at least 124 MPa at 28 days.

Table 1 Total admixture contents and conductivities observed in UHPCs used in this study
Full size table

3.2 Concretes

Four different concretes were tested as part of this study to compare against the performance of UHPC-class materials: an A4 concrete, an A5 concrete, a latex-modified concrete (LMC), and a high-performance concrete (HPC). The A4, A5, and LMC concretes were prepared using Virginia Department of Transportation (VDOT) specifications [35] and the HPC mixture developed at TFHRC. All four mixtures used a type I/II portland cement with a Blaine fineness of 382 m2/kg and a density of 3070 kg/m3. The supplementary cementitious materials (SCMs) replacements reported were based on cement replacement percentages by mass. A quartzite river sand with an absorption capacity of 0.93% was the fine aggregate. No. 57 coarse aggregate, a crushed limestone with an absorption capacity of 0.44%, served as the coarse aggregate in the A4, A5, and HPC concretes. The LMC used a No. 8 stone, a crushed limestone with an absorption capacity of 0.49%, as the coarse aggregate. A commercially available latex admixture was added by weight of cement to develope the LMC concrete. The aggregates used in the study meet the standard requirements of ASTM C33: Standard Specification for Concrete Aggregates [36]. Table 2 displays the mixture design parameters and properties of the conventional concretes used in this study.

Table 2 Conventional concretes used in this study and their properties
Full size table

4 Experimental methods

4.1 Durability testing: electrical-based tests

4.1.1 Electrical resistivity

Electrical resistivity is a rapid, nondestructive test that has shown strong correlations to different concrete durability properties and is fundamentally linked to the diffusion coefficient through its relationship with the formation factor and the Nernst-Einstein equation [19,20,21, 23, 38]. The UHPCs and concretes were evaluated for uniaxial resistivity in accordance with AASHTO TP 119-22: Standard Method of Test for Electrical Resistivity of a Concrete Cylinder Tested in a Uniaxial Resistance Test [39]. The UHPCs were evaluated for electrical resistivity by using 75-mm-diameter by 150-mm-tall cylinders, and the conventional concretes were evaluated for electrical resistivity by using 100-mm-diameter by 200-mm-tall cylinders. The UHPC specimens were fabricated in accordance with ASTM C1856: Standard Practice for Fabricating and Testing Specimens of Ultra-High Performance Concrete [40]. Larger-sized specimens, fabricated in accordance with ASTM C192: Standard Practice for Making and Curing Concrete Test Specimens in the Laboratory [41], were used for conventional concretes to accommodate for the larger aggregate sizes. A set of three cylinders was tested for each mixture. The UHPC specimens were mixed and cast following manufacturer recommendations. However, fibers were excluded from the UHPC specimens on which electrical resistivity was performed because the inclusion of steel fibers could affect electrical-based tests. All specimens were demolded within 48 h from the time of mixing, at which time the temperatures were measured using a noncontact infrared thermometer, the masses were recorded, and conditioning began. The specimens were conditioned using one of two conditioning methods. Conditioning method 1 was in accordance with the AASHTO TP 119–22 [39] default conditioning option, and the conditioning method 2 was in accordance with Sect. 7 of AASHTO M 201-21: Standard Specification for Mixing Rooms, Moist Cabinets, Moist Rooms, and Water Storage Tanks Used in the Testing of Hydraulic Cements and Concretes [42]. The two conditioning methods differ based on the type of solution in which the specimens are submerged before testing. The solution used in method 1 is an alkali-concentrated conditioning solution—also referred to as an artificial pore solution—consisting of potassium hydroxide, sodium hydroxide, and calcium hydroxide; and the solution from method 2 is a saturated calcium hydroxide solution, also referred to as limewater. A 19-L bucket was used to condition the specimens. Total volume of conditioning solution used for concrete was 12 L and 5 L for UHPC. Each set of specimens was submerged in the conditioning solution, with the depth of the solution 38 mm above the top of the specimen. The AASHTO TP 119-22 [39] testing for each specimen was conducted at ages of 1, 7, 28, 56, 91, 180, and 365 days.

Prior to testing, the specimens were removed from the conditioning solution, brought to a surface-dry condition, and tested for mass and temperature. The uniaxial resistance was then tested using an alternating current (AC) resistance meter. Resistivity was obtained from the measured resistance value by using Eq. (1),

$$\rho = R \cdot k$$
(1)

where \(\rho\) is the electrical resistivity (ohm·meters, Ω·m), R is the measured resistance (Ω), and k is the geometry correction factor (m). The geometry correction factor is the ratio of a cylinder’s cross-sectional area divided by length; that ratio equals 0.03 m for a 75-mm-diameter by 150-mm-tall cylinder and 0.04 m for a 100-mm-diameter by 200-mm-tall cylinder. All temperature measurements were found to be between 21 and 25 °C, as such, temperature corrections were not applied (Fig. 1).

Fig. 1
figure 1

Source: Federal Highway Administration

Experimental test setup to determine electrical resistivity in a uniaxial configuration [12].

Full size image

4.1.2 Rapid chloride migration

The rapid chloride migration test determined the non-steady-state (NSS) chloride migration coefficients of UHPCs and concretes (A4, A5, LMC, and HPC) in accordance with NT BUILD 492: Concrete, Mortar and Cement-Based Repair Materials: Chloride Migration Coefficient from Non-Steady-State Migration Experiments [26], as shown in Fig. 2. The NSS chloride migration coefficient can be used in corrosion-based service life estimations when combined with Fick’s second law [43]. Compared with conventional cementitious materials, Fick’s second law of diffusion can be challenging to apply to cementitious composites with extremely low permeability such as UHPC. For comparison purposes, it was assumed that the same relationships apply to UHPCs as to conventional concretes. A higher chloride migration coefficient suggests chlorides can penetrate through the material faster, thereby increasing the risk of corrosion initiation of the embedded reinforcement. The UHPCs and concretes were evaluated for chloride migration by using 100-mm-diameter by 200-mm-tall cylinders. A set of two cylinders was cast for each mixture. Steel fibers were excluded from all UHPCs to avoid possible interaction with the applied voltage potential.

Fig. 2
figure 2

Chloride migration test setup in accordance with NT BUILD 492 [26]. Source: Federal Highway Administration

Full size image

The test cylinders were sealed inside the molds and cured for 28 days in an environmental chamber at 23 ± 2 °C and 50 ± 3% relative humidity. After 28 days, the cylinders were demolded, and two 50-mm-thick discs were cut from the center of each cylinder. The disc specimens were vacuum saturated with deionized water at an absolute pressure of 6 Torr for 24 h. The specimens were then tested for chloride migration in accordance with NT BUILD 492 [26]. The test parameters are summarized in Table 3.

Table 3 Chloride migration test parameters in accordance with NT BUILD 492 [26]
Full size table

After completion of testing, the discs were removed from the test setup and the surfaces were rinsed off. The discs were split in half, exposed surfaces were sprayed with a 0.1 N silver nitrate solution, and specimens were left for 15 min to allow the silver nitrate to react with the chlorides, resulting in a change in color of the chloride-penetrated area. The chloride penetration depths were measured at 10-mm increments along the diameter of each disc. The NSS migration coefficient was obtained based on the measured specimen thickness and the average chloride penetration depths by using Eq. (2) as listed in NT BUILD 492 [26].

$${D}_{\text{nssm}}=\frac{0.0239\left(273+T\right)L}{\left(U-2\right)t}\left({x}_{d}-0.0238\sqrt{\frac{(273+T)L\cdot {x}_{d}}{U-2}}\right)$$
(2)

where Dnnsm = NSS migration coefficient, × 10–12 m2/s; U = absolute value of the applied voltage, V; T = average value of the initial and final temperatures in the anolyte solution, °C; L = thickness of the specimen, mm; xd = average value of the chloride penetration depths, mm; and t = test duration, hour.

4.2 Durability testing: alternative tests

4.2.1 Rate of water absorption

ASTM C1585: Standard Test Method for Measurement of Rate of Absorption of Water by Hydraulic-Cement Concretes was used to assess the rate of fluid ingress in UHPC and concrete specimens by capillary action [24]. The water absorption rate is a durability indicator that has been used to predict the service life of structural elements in terms of freeze–thaw damage [44]. Faster rate of water absorption in concrete is indicative of a more permeable microstructure [45, 46]. Furthermore, concretes with higher rates of water absorption will reach a critical degree of saturation faster and will have greater risk of damage during freeze–thaw events [47,48,49]. Previous research that looked into the effects of sample conditioning, found that the environmental humidity played a significant role during the conditioning period on the outcome of the water absorption tests [50,51,52]. Because of this, it was proposed to modify the conditioning procedure of ASTM C1585 [24] by equilibrating the test samples at a constant relative humidity, before measuring the rate of water absorption [51]. However, the applicability of this modified method to UHPCs was found to be particularly challenging due to the extended conditioning times; laboratory tests indicated that a duration of more than 2 years could be necessary. As a result, the standard ASTM C1585 [24] was used to evaluate the rate of water absorption in all the materials tested as a part of this study. For the water absorption testing, standard 100-mm-diameter by 200-mm-tall cylinders were used. A set of two cylinders was cast for each mixture. The cylinders were sealed and cured for 28 days in an environmental chamber at 23 ± 2 °C and 50 ± 3% relative humidity. After 28 days, the cylinders were demolded, and three 50-mm-thick discs were cut from each of the cylinders. The specimens were then placed in an environmental chamber to condition for 3 days at 50 ± 2 °C and 80 ± 3% relative humidity. After 3 days, the specimens were transferred into a sealable container. The sealed containers were stored at 23 ± 2 °C for 15 days. After the conditioning period, the specimens were weighed and then wrapped with aluminum tape around the lateral surface, and the top surface was covered with plastic to prevent excess drying. Each specimen was weighed once again to determine initial mass. The exposed surface was placed into water, and mass measurements were conducted at a series of preselected time intervals.

4.2.2 Freeze–thaw degradation resistance

Freezing and thawing performance of the UHPCs, HPC, and A4 concrete was evaluated in accordance with procedure A of ASTM C666: Standard Test Method for Resistance of Concrete to Rapid Freezing and Thawing [53]. State DOTs use freeze–thaw resistance as a performance requirement when specifying UHPC-class materials [54, 55]. This accelerated durability test helps determine the ability of concrete materials to resist the expansive forces caused by ingress and subsequent freezing of water.

The UHPCs and concretes were evaluated for freeze–thaw by using standard 76-mm-wide by 102-mm-tall by 406-mm-long prisms [53]. A set of three prisms was cast for each mixture. After 24 h of initial curing at room temperature, the prisms were demolded and cured in a saturated limewater tank for 14 days. After 14 days, the prisms were moved into a tempering tank at 4.4 °C for an initial test measurement. After the initial mass and transverse resonant frequency were determined, the prisms were then subjected to a total of 600 cycles at specified freezing and thawing temperature values of −18 °C and 4.4 °C, respectively, with periodic mass and transverse resonant frequency testing after every 36 cycles.

The transverse resonant frequency of each test specimen was determined in accordance with ASTM C215: Standard Test Method for Fundamental Transverse, Longitudinal, and Torsional Resonant Frequencies of Concrete Specimens [56]. At the end of each interval, the specimens were visually inspected and the mass and transverse resonant frequency were recorded to monitor any mass loss or change in the relative dynamic modulus of elasticity. The relative dynamic modulus of elasticity is calculated using the transverse resonant frequency, as shown in Eq. (3),

$$P_{c} = \left( {{\raise0.7ex\hbox{${n_{c}^{2} }$} \!\mathord{\left/ {\vphantom {{n_{c}^{2} } {n^{2} }}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{${n^{2} }$}}} \right) \cdot 100$$
(3)

where Pc is the relative dynamic modulus of elasticity after c cycles of freezing and thawing (%); n is the fundamental transverse resonant frequency at 0 cycles of freezing and thawing (Hz); and nc is the fundamental transverse resonant frequency after c cycles of freezing and thawing (Hz).

5 Results and discussion

The results and discussion focus on durability testing by using electrical-based tests, alternative durability tests, correlations among the electrical-based tests, and service life estimations.

5.1 Durability testing: electrical-based tests

5.1.1 Electrical resistivity

Figure 3a and b display uniaxial resistivity measured on the specimens at ages of 1, 7, 28, 56, 91, 180, and 365 days when tested in accordance with AASHTO TP 119-22 [39], with the test specimens conditioned: (1) in accordance with the AASHTO TP 119-22 [39], default conditioning option, also referred to as artificial pore solution, and (2) alternatively using Sect. 7 of AASHTO M 201-21 [42]. The two different conditioning methods were used to achieve understanding of the effect of conditioning solutions on resistivity development over time. Figure 3a represents the data of the specimens conditioned according to method 1, and Fig. 3b represents the data of the specimens conditioned according to method 2. At each age, the bars in Fig. 3 are organized from left to right in accordance with the legend. Due to the wide range of resistivity values observed among the materials at each testing age, the data are presented on a logarithmic scale to better identify and differentiate among the data points. The first five bars from the left are commercially available products marketed as UHPC. Though all UHPC specimens were cast without fiber reinforcement, the fifth UHPC (U-M) included CNF, which could not be removed from the mixture because it is an integral part of the admixtures system (Table 1). The last four bars represent the data obtained for the four concrete systems: A4, A5, LMC, and HPC. The UHPC-class materials evaluated represent many of the commercially available preblended UHPCs in the U.S. market, and the values are representative of what might be encountered during the use of UHPC-class materials.

Fig. 3
figure 3

Source: Federal Highway Administration

Development of AASHTO TP 119–22 [39] resistivity as a function of curing time for UHPCs and conventional concrete specimens conditioned in accordance with (a) the AASHTO TP 119–22 [39] default conditioning option and (b) Sect. 7 of AASHTO M 201–21 [42].

Full size image

As seen in Fig. 3a, the resistivity of the first four UHPCs (U-D, U-E, U-H, and U-J) conditioned in accordance with the AASHTO TP 119–22 [39] default conditioning option increased by two orders of magnitude from 1 to 28 days. However, the resistivity of the U-M increased by one order of magnitude during the 28-day period. The resistivity of conventional concretes A4, A5, and LMC increased by a factor of 5 from 1 to 28 days, and HPC’s resistivity increased by a factor of 16. The resistivity values of U-M and HPC were similar at 28 days. From 28 to 365 days, resistivity values increased on average by a factor of 4 for U-D, U-E, U-H, and U-J. For U-M, resistivity values increased by a factor of 3 from 28 to 365 days. The average resistivity value of U-D, U-E, U-H, and U-J was one order of magnitude higher than U-M at 365 days. The resistivity of conventional concretes A4 and A5 increased by a factor of 8 from 28 to 365 days, and LMC and HPC resistivity increased by a factor of 2.

Similar trends were observed for specimens conditioned in accordance with Sect. 7 of AASHTO M 201-21 [42], as seen in Fig. 3b. In particular, from 1 to 28 days, resistivity values increased by an average factor of 80 for four commercially available UHPCs (U-D, U-E, U-H, and U-J). U-M resistivity values increased by a factor of 9, resulting in the average resistivity value of U-D, U-E, U-H and U-J being an order of magnitude higher than that of U-M at 28 days. The resistivity of conventional concretes (A4, A5, and LMC) increased by a factor of 5 from 1 to 28 days, and HPC resistivity increased by a factor of 12. The resistivity values of U-M and HPC were similar at 28 days. From 28 to 365 days, the resistivity values further increased on average by a factor of 5 for U-D, U-E, U-H, and U-J. In the case of U-M, resistivity increased by a factor of 3, causing its resistivity value at 365 days to be approximately one order of magnitude lower than U-D, U-E, U-H, and U-J. The resistivity of conventional concretes A4 and A5 increased by a factor of 6 from 28 to 365 days, and the resistivity of LMC and HPC increased by a factor of only 2.

The lower resistivity measured for U-M may have been a result of the CNF, a different microstructure, or both. Previous research performed on straight ordinary portland cement pastes that included CNF reported a decrease in resistivity when CNF with a diameter of 0.1 μm and a dosage of 0.5% by volume were included in the system [57]. However, the same study also reported that the observed decrease in resistivity was less significant than when larger (15- μm) carbon fibers were included [57]. Similar findings were reported in other studies [58]. Because of this, alternative test methods, which are discussed in the following sections, might be appropriate for use as durability performance indicators for U-M and other, similar UHPCs containing CNF or conductive inclusions.

Figure 4 demonstrates the role of conditioning solution in resistivity development for UHPCs and concretes tested during this study. The x-axis represents the materials tested, and the y-axis represents the ratio between resistivity values of specimens conditioning in limewater and artificial pore solution. The role of conditioning solutions was studied to address industry’s and agencies’ hesitancy to adopt one particular curing solution for resistivity testing. Limewater is commonly used in the construction industry for conditioning concrete specimens; artificial pore solution as described in AASHTO TP 119-22 [39] is a novel approach designed to mimic the pore solution of typical cementitious materials so as to help minimize alkali leaching and improve resistivity measurement reliability.

Fig. 4
figure 4

Source: Federal Highway Administration

Role of conditioning solution on resistivity development.

Full size image

The results demonstrate that the conditioning solution may significantly affect measured test specimen resistivity. For UHPCs, the effect of conditioning solution was often found to be significant at early ages—that is, up to 91 days. At ages beyond 91 days, the effect of conditioning solution is generally less significant. As seen in Fig. 4, the ratio of measured resistivities of all UHPCs (U-D to U-M) conditioned in limewater and artificial pore solution were found to be 1 ± 0.1 beyond 91 days. In the case of concretes, the effect of conditioning solution was found to be inconsistent. For example, the resistivity of A4 and A5 concretes tested up to 365 days is consistently higher in specimens conditioning in artificial pore solutions compared with specimens conditioning in limewater. For LMC and HPC, the role of conditioning solution was similar to that of UHPCs—that is, the effect of conditioning solution was found to be more significant at early ages, up to 91 days. At ages beyond 91 days, the effect of conditioning solution was negligible, with the ratio of LMC and HPC resistivities conditioned in limewater and artificial pore solution equaling 1 ± 0.1. The differences found in A4 and A5 are attributed to high incorporations of class F fly ash. An increase in fly ash reactivity was observed in high-pH-solution environments, as in the case of the AASHTO TP 119-22 [39] default conditioning option [59]. It has been reported that the chemical composition and the type of glassy phases in the fly ash can lead to changes in the fly ash’s early-age reactivity when exposed to solutions with different alkaline levels [60].

Based on results from the commercially available UHPCs, two different threshold resistivity values—one for each conditioning solution—were proposed for use when assessing performance via tests completed at 28 days [12]. The proposed threshold can be used as a minimum resistivity target for UHPC-class materials during mixture qualification operations. At 28 days, the minimum resistivity values the UHPCs displayed were 2,150 Ω·m and 1,920 Ω·m when conditioned in accordance with (1) the AASHTO TP 119-22 default conditioning option [39] and (2) Sect. 7 of AASHTO M 201-21 [42]. A multioperator coefficient of variation of 13.2% as described in AASHTO TP 119-22 was used to calculate the standard deviation [39]. For UHPCs conditioned in accordance with the AASHTO TP 119-22 [39] default conditioning option, a rounded value of 1500 Ω·m—which represents the 95-percent confidence interval—is proposed as the threshold for 28-day resistivity measurements. Similarly, for UHPCs conditioned in accordance with Sect. 7 of AASHTO M 201-21 [42], a rounded value of 1,400 Ω·m is proposed as the threshold for 28-day resistivity measurements. The thresholds are based on a statistical analysis that included UHPCs U-D through U-J. U-M, which displayed a resistivity at 28 days that was one order of magnitude lower than that of the other commercially available UHPCs, was excluded from the analysis.

5.1.2 Rapid chloride migration

NSS chloride migration behavior was assessed in accordance with NT BUILD 492 [26]. This test was selected for ionic penetrability evaluation because it is quicker than the traditional chloride ponding test [22], thanks to the applied electrical voltage, which accelerates the ingress of chlorides into the concrete specimen. The analyzed test data allow for the calculation of a chloride migration coefficient, which is an indication of how effectively chloride ions can penetrate through concrete and can be used for service life estimations. Figure 5 presents calculated NSS migration coefficients for the materials evaluated in this study. Due to the wide range of migration coefficients observed among the materials tested, the y-axis is presented on a logarithmic scale to facilitate differentiation between data points.

Fig. 5
figure 5

Source: Federal Highway Administration

NSS chloride migration in UHPCs and conventional concrete specimens tested in accordance with NT BUILD 492 [26].

Full size image

As shown in Fig. 5, similar to electrical resistivity, rapid chloride migration was also able to differentiate among the UHPC-class materials. The NSS migration coefficients obtained for UHPCs U-D, U-E, U-H, and U-J range from 0.07 × 10–12 to 0.3 × 10–12 m2/s, with U-E being the lowest and U-H being the highest. U-M and HPC displayed NSS migration coefficients of approximately 2.0 × 10–12 m2/s, which is one order of magnitude higher than the other UHPCs. NSS migration coefficients obtained for the conventional concretes (A4, A5, and LMC) range from 6.6 × 10–12 to 9.2 × 10–12 m2/s, with LMC being the lowest and A4 being the highest. Those values are approximately four times higher than the migration coefficient of HPC.

Research conducted in Pennsylvania on 68 different concrete mixtures with water/cementitious ratios (w/cm) ranging from 0.30 to 0.65 and with SCM replacement levels ranging from 0 to 50% resulted in NSS migration coefficients that varied from 0.5 × 10–12 m2/s for a concrete with 59.4% type I cement, 35% slag, and 5.5% silica fume to 30.6 × 10–12 m2/s for a concrete with 100% type I cement [61]. The average migration coefficient for 68 concrete mixtures tested at 28 days equaled 9.5 × 10–12 m2/s [61]. The average migration coefficients measured on A4, A5, and LMC in this study are similar to the average migration coefficient values from the 68 mixtures in the referenced study. Additionally, similar to what was observed in previous research, the decrease in migration coefficients among the conventional concretes can be attributed to the decreasing w/cm contents—A4 at 0.43, A5 at 0.4, LMC at 0.38, and HPC at 0.33—and the presence of SCMs. In the case of UHPCs, similar to what was observed in concretes, the further decrease in migration coefficients can likely be attributed to the presence of high amounts of SCMs and very low w/cm (i.e., the typical w/cm of UHPCs are 0.16 or lower [25]). The UHPCs were tested at an applied voltage of 60 V for 96 h, and most displayed an average chloride penetration depth of 5 mm or less. The only exception was U-M, which exhibited an average chloride penetration depth of 10 mm (i.e., twice as much as the other UHPCs). Because the same voltage was applied to all the UHPCs, the higher chloride penetration exhibited by U-M might have been a consequence of the CNF, a different microstructure, or both.

5.2 Durability testing: alternative test methods

Non-electrical-based durability tests were performed to investigate their applicability to UHPC-class materials.

5.2.1 Rate of water absorption

Rate of water absorption, measured in accordance with ASTM C1585 [24], is a durability test the construction industry uses to assess the rate of fluid penetration into a concrete structure. The fluid penetration rate can be associated with the resistance of the structure to external sulfate attack and freeze–thaw damage [44]. The results of this test relate directly to the microstructural characteristics of concrete because total pore volume and pore connectivity affect the rate at which the material absorbs fluids [45, 46]. Because no electrical field is generated during the test and the fluid transport is governed by capillary suction, the results should not be affected by the presence of conductive inclusions within the concrete. Because of this, ASTM C1585 [24] may be considered as an alternative durability test whenever there is concern about conductive inclusions’ affecting the resistivity measurements.

Results shown in Fig. 6 display the water absorption measurements for UHPCs and conventional concretes. Four of the UHPCs (U-D, U-E, U-H, and U-J) did not exhibit the distinctive bilinear behavior typically observed in concrete mixtures when absorption data are plotted as a function of the square root of time [62]. Pore structure of conventional concrete is made up of gel pores, small capillary pores, large capillary pores, and air voids. In case of conventional concretes, initial absorption is a result of the gel pores and the small capillary pores being filled due to high capillary suction [51, 63]. And secondary absorption is a result of the large capillary pores and air voids being filled [51, 63]. Slope of initial absorption is higher as the initial absorption process is very quick and happens within the first few hours (6 h). Slope of secondary absorption is lower as the process is slow which happens beyond 24 h. The initial and secondary sorptivities can be used to evaluate the connectivity of the pore network [51]. Four UHPCs (U-D, U-E, U-H, and U-J) did not exhibit significant differences in water absorption rates among one another, and the lack of initial and secondary absorption slopes in UHPCs is likely related to reduced pore connectivity and the presence of a pore structure not commonly found in conventional concretes [46]. As seen in Fig. 6, among the UHPCs, U-M exhibited bilinear behavior likely due to a pore structure different from that of the remaining UHPCs. Independently from the presence of the bilinear pattern, secondary water absorption rates were calculated for UHPCs and conventional concretes as the slope of the measurements performed from 24 h to 14 days. Conventional concretes showed higher water absorption rates and total water absorptions compared to the UHPCs, with A5 being the highest, followed by LMC, A4, and HPC. This behavior can be attributed to the increased porosity and pore connectivity observed in conventional concretes compared with UHPCs.

Fig. 6
figure 6

Source: Federal Highway Administration

Water absorption tested in accordance with ASTM C1585 [24] as a function of square root of time in UHPCs and conventional concrete specimens.

Full size image

Table 4 provides a summary of secondary water absorption rates in the UHPCs and conventional concretes tested. Based on the results presented in Table 4, four UHPCs (U-D, U-E, U-H, and U-J) have an average secondary water absorption rate of approximately 5 × 10–3 mm3/mm2√h, and U-M, A4, and HPC have an average secondary water absorption rate of approximately 16 × 10–3 mm3/mm2√h. The findings agree with the resistivity and migration data, which indicated that U-M has a more-permeable microstructure compared with other UHPCs. A5 and LMC exhibited similar secondary water absorption rates.

Table 4 Summary of the secondary water absorption rates in UHPCs and conventional concretes
Full size table

As seen in Fig. 6, compared with the other concretes, the lower water absorption observed in HPC may result from the low w/cm of 0.33 and the presence of high SCM contents, which likely led to reduced total pore volume of the system. However, comparing the secondary water absorption rates between the A4 and A5 concretes, the SCM contents (35% by mass for A4 and 21% by mass for A5) affected water absorption rates more than the w/cm (0.43 for A4 and 0.40 for A5) did. Unlike electrical resistivity and rapid chloride migration tests, ASTM C1585 [24] was not able to differentiate among UHPC-class materials. However, a water absorption test performed in accordance with ASTM C1585 [24] proved to be helpful in differentiating between UHPCs and conventional concretes.

5.2.2 Freeze–thaw degradation resistance

Freeze–thaw degradation resistance is a durability test that is widely performed or required by infrastructure owners to assess the resistance of concrete to freeze–thaw cycles. Similar to ASTM C1585 [24], the outcome of this test is not affected by the presence of conductive inclusions in the same way as the electrical-based tests. In general, the presence of sufficient entrained air and the absence of faulty aggregates are critical attributes in the freeze–thaw resistance of concrete [64, 65]. The rate of test specimen water absorption is also important because no matter the air content, the concrete will start deteriorating upon reaching critical degree of saturation [49]. For very low-permeability materials, such as UHPC, the absence of permeating water can significantly restrict the opportunity for freeze–thaw degradation.

During this study, all specimens were tested up to 600 cycles. After every 36 cycles, a visual examination of the specimens was conducted. The visual examinations indicated that U-D, U-E, U-H, and U-J experienced no degradation. However, U-M exhibited a small material loss along the surface as the cycling progressed. The HPC and A4 also exhibited material loss. However, material loss in A4 was found to be significantly higher than that observed in HPC and U-M, with A4 showing exposed coarse aggregate on all sides of the prisms, as seen in Fig. 7.

Fig. 7
figure 7

Material loss on the A4 concrete prismatic specimen, leading to an exposed aggregate surface. Source: Federal Highway Administration

Full size image

Results shown in Fig. 8a display the change in relative dynamic modulus of UHPCs calculated using Eq. (3). A4 and HPC were tested along with the UHPCs for comparison purposes. As seen in Fig. 8a, the relatively dynamic modulus of all UHPCs increased as the testing progressed up to 600 cycles. Among the conventional concretes tested, HPC exhibited an increase in relative dynamic modulus similar to that of the UHPCs. After 600 cycles, the A4 concrete exhibited a 31% decrease in relative dynamic modulus.

Fig. 8
figure 8

Source: Federal Highway Administration

(a) Change in relative dynamic modulus of UHPCs and conventional concretes tested in accordance with ASTM C666 [53] and (b) change in mass of UHPCs and conventional concretes tested for freeze–thaw degradation resistance.

Full size image

Mass change data are presented in Fig. 8b. After every 36 cycles, the mass change of the prismatic specimens was measured in a surface dry condition. Based on the results shown in Fig. 8b, minimal mass change occurred in the UHPCs and HPC specimens over time. However, A4 exhibited a mass loss of 5% and significant material loss, which led to exposed coarse aggregates on all sides of the prisms (Fig. 7). The incongruency between the low mass loss and the severe surface scaling may be a result of the specimens’ water absorption’s offsetting the material loss.

The lack of damage in the UHPC specimens after 600 freeze–thaw cycles, especially in the absence of entrained air, is associated with both the slow rate of water penetration into the sample and to the predominant presence of gel pores over larger capillary pores within the UHPC cementitious matrix [66, 67]. The freezing temperature of water in a porous system is inversely proportional to the size of the pore in which the water is trapped [68]. Reducing or removing the larger capillary pores from the system makes it technically possible to lower the freezing point of water enough so that no phase change occurs during a freezing cycle. The absence of larger capillary pores would prevent damage from occurring and reduce system susceptibility to the critical degree of saturation theory [49, 68]. The increase in relative dynamic modulus over time in the UHPCs and HPC, other than indicating the lack of damage, is potentially linked to a continuation in the reaction of cementitious materials over time. In absence of available capillary pore spaces, where new reaction products can precipitate, the occurrence of localized microcracking and the continuous supply of water might have provided the required additional space and moisture for the continuation of the cementitious materials reactions over time [10, 68]. In summary, although the ASTM C666 [53] test is likely to provide a good indication of concrete mixtures’ durability with regard to freeze–thaw damage, it may not provide useful insight on the difference in durability performance between a UHPC-class material and any concrete having an adequate air-void system [69, 70].

5.3 Correlation between electrical-based durability test methods

Figure 9 plots the electrical resistivity determined in accordance with AASHTO TP 119-22 [39] against the NSS migration coefficient. The established correlation agrees with previous research that reported an inverse proportionality between electrical resistivity and NSS migration coefficient for concretes using only ordinary portland cement and concretes with different SCMs containing water-to-binder ratios ranging from 0.23 to 0.84 [71].

Fig. 9
figure 9

Source: Federal Highway Administration

Correlation between electrical resistivity and NSS migration coefficient of UHPCs and conventional concretes tested at 28 days.

Full size image

The UHPCs (U-D, U-E, U-H, and U-J), which displayed electrical resistivity of 1,500 Ω m or higher at 28 days, exhibited migration coefficients lower than 0.5 × 10–12 m2/s. All conventional concretes, which were characterized by lower electrical resistivity, also displayed higher migration coefficients. The inverse relationship between electrical resistivity and migration coefficient reveals that the UHPCs with higher resistivity will likely experience less chloride ingress. Migration coefficients obtained from the rapid NSS migration test can be further used as input data in predictive models for service life estimations.

The inverse relationship observed in the study with a high degree of correlation (R2 of 0.89) follows a trend as expected between electrical resistivity and NSS migration coefficient and as defined by the Nernst-Einstein equation [19,20,21, 23, 38]. However, the established relationship in this study does not account for the differences in initial conditioning between the two test methods, the pore solution electrical properties, and the binding capacity of proprietary UHPC-class materials; therefore, the relationship remains empirical in nature. Only after accounting for these variables may electrical resistivity be used to estimate an apparent diffusion coefficient of the system and to, based on this correlated coefficient, perform service life estimations.

5.4 Service life estimations

The service life of any structural element is dependent on the element’s ability to withstand dominant deterioration mechanisms during the predicted time period without undue loss of serviceability [72]. Based on the test results obtained in this study, the dominant deterioration mechanism considered for service life estimations of UHPCs is chloride-induced corrosion. The fib Model Code for Service Life Design [43] takes full-probabilistic, semi probabilistic, deemed-to-satisfy, and avoidance-of-deterioration approaches to service life design. Of those methods, a full-probabilistic approach is desirable for large infrastructure projects [72]. The NSS migration coefficients obtained during this study can be used in the full-probabilistic approach to determine the service life of a UHPC-based structural element.

According to fib Model Code 34, service life is defined as the time elapsed for the reinforcement to reach a certain chloride concentration level at which depassivation and corrosion initiation occur [43]. This study used a full-probabilistic design for the modeling of chloride-induced corrosion in uncracked concrete based on Fick’s second law of diffusion and the guidance available in fib Model Code 34 [43]. As stated previously, it is recognized that Fick’s second law of diffusion can be challenging to apply to composites with extremely low permeability, such as UHPC. For comparison purposes, it is assumed here that the same relationships apply to both UHPCs and ordinary concretes. The service life estimations of the UHPC-class materials used in this study assume that the materials are used in a bridge deck, with the top reinforcement consisting of 12.7-mm-diameter rebar and a concrete cover of 32 mm. The serviceability limit state equation for reinforcement depassivation is given in Eq. (4) [43]:

$${C}_{\mathrm{crit}}=C\left(x=a,t\right)={C}_{0}+({C}_{s,\Delta x}-{C}_{0})\cdot \left[1-erf\frac{a-\Delta x}{2\cdot \sqrt{{D}_{app,C}\cdot t}}\right]$$
(4)

where Ccrit is the critical chloride content in weight-percent/cement; C(x, t) is the content of chlorides in the concrete at a depth x and at time t in weight-percent/cement; C0 is the initial chloride content of the concrete in weight-percent/cement; Cs,∆x is the chloride content at depth ∆x at a certain point of time t in weight-percent/cement; x is the depth with a corresponding chloride content C(x, t) in mm; erf is the error function; a is the concrete cover in mm; ∆x is the depth of the convection zone in mm; Dapp,C is the apparent chloride diffusion coefficient in m2/s; and t is time in seconds.

The apparent diffusion coefficient in Eq. (4) can be determined using Eq. (5) [43]:

$${D}_{\mathrm{app}, C}={k}_{e}\cdot {D}_{RCM, 0}\cdot {k}_{t}\cdot A(t)$$
(5)

where DRCM,0 is the chloride migration coefficient in m2/s; kt is the transfer parameter, which is taken as 1 [43]; ke is the environmental transfer variable, which can be determined using Eq. (6) [43]; and A(t) is the subfunction considering the aging, which can be determined using Eq. (7) [43]:

$${k}_{e}=\mathrm{exp}\left({b}_{e}\left(\frac{1}{{T}_{\mathrm{ref}}}-\frac{1}{{T}_{\mathrm{real}}}\right)\right)$$
(6)
$$A\left(t\right)={\left(\frac{{t}_{0}}{t}\right)}^{{a}_{e}}$$
(7)

where be is the regression variable in Kelvin, Tref is the standard test temperature in Kelvin, Treal is the temperature of the structural element in Kelvin, t0 is the reference point of time in seconds, and ae is the aging exponent.

The following assumptions were used in the determination of the service life of the materials investigated in this study. First, a surface chloride concentration of 3% by weight of binder was assumed based on AASHTO T 259: Standard Method of Test for Resistance of Concrete to Chloride Ion Penetration, which uses a 3-percent chloride solution to mimic the surface chloride concentration observed on surfaces that are not continually submerged under solutions with salts [73]. Next, the background chloride concentration was assumed to be 0.1% by weight of binder [74]. Based on fib Model Code 34, the critical chloride content (Ccrit) considered for corrosion to occur was set at 0.6% by weight of binder [43]. All materials used in this study have high SCM contents. Therefore, an average age factor (ae) of 0.5 was assumed based on the available literature [43, 74]. Because it is recognized that a change in temperature can influence the diffusion of chlorides and alter corrosion initiation [75], the model assumes that the diffusion coefficients are estimated at a standard test temperature (Tref) of 293 Kelvin and an average real-time temperature (Treal) of 296 Kelvin. The regression variable (be) is taken as 4,800 Kelvin. The diffusion coefficients were determined in accordance with fib Model Code 34 by using the measured NSS migration coefficients [43]. The diffusion coefficients at 10 years determined the total service life of the uncracked structural element when made with each of the tested materials. A 10-year timeframe was considered to account for different environmental and material parameters for the structural elements. Based on those assumptions, estimated service life/time to corrosion initiation for each material is provided in Table 5.

Table 5 Service life estimations of ultra-high performance concretes and concretes
Full size table

As shown in Table 5, all UHPCs except U-M exhibited predicted times to corrosion initiation greater than 100 years. HPC and U-M are projected to have a time to corrosion initiation of approximately 50 to 55 years followed by A4, A5, and LMC, at 12, 14, and 17 years, respectively. The service life estimations provided are based on the listed assumptions and are subject to change in consideration of real-time field data, including concrete properties, steel properties, and exposure conditions.

6 Conclusions

This study investigated the relevance and effectiveness of various durability performance test methods for commercially available materials marketed as UHPC class. In recognition of that—and with the possible exception of the freeze–thaw test—these test methods only indirectly assess the potential durability of a material, the results revealing differences between the durability performance indicators of the tested UHPCs. It can be inferred from the results that materials that meet the mechanical performance requirements for UHPC may not exhibit the same durability performance. Based on the obtained results and analysis contained herein, the following conclusions can be made:

  • Electrical resistivity is a reasonable tool to characterize the anticipated durability performance of different UHPC-class materials when the materials are tested in the absence of fiber reinforcement. Based on the measured data and statistical analysis for the two different conditioning solutions investigated, two different resistivity thresholds are proposed to qualify the durability of UHPC-class materials tested at 28 days. A threshold value of 1500 Ω m is proposed for specimens conditioned according to Sect. 10.2—Default Conditioning—of AASHTO TP 119–22 [39]—and a threshold value of 1400 Ω m is proposed instead for specimens conditioned in saturated limewater in accordance with Sect. 7 of AASHTO M 201 [42].

  • Resistivity test results agree with NSS chloride migration testing, a rapid test that can determine the rate at which chloride ions travel through concrete. Four of the UHPCs, with the exception of U-M, achieved migration coefficients of less than 0.5 × 10–12 m2/s. The conventional concretes tested within this study displayed migration coefficients ranging from 6.5 × 10–12 to 10 × 10–12 m2/s.

  • For materials that may contain conductive inclusions, electrical-based test methods may not be appropriate. Alternative test methods such as ASTM C1585 [24] and ASTM C666 [53] may provide insight on the performance of these materials.

  • Rates of water absorption, tested in accordance with ASTM C1585 [24], were used to distinguish between the UHPCs and the conventional concretes. However, the test was not able to individually differentiate between four of the five UHPC systems. The results from this test generally agree with the electrical resistivity and NSS migration, with just a few exceptions. Additionally, UHPC-class materials with very low water absorption showed an absence of the commonly observed bilinear response, likely attributable to the UHPC’s microstructure.

  • When tested for freeze–thaw performance in accordance with ASTM C666 [53] procedure A, neither the UHPC-class materials nor the HPC concrete exhibited deterioration of any significance. The absence of damage is attributed to the absence of capillary porosity and the low water absorption rates of UHPCs and HPC. Because of that, ASTM C666 [53] may not provide useful insight into the difference in durability performance between a UHPC-class material and any other concrete having low penetrability and an adequate air-void system.

  • The electrical-based tests used in this study were able to successfully differentiate the anticipated durability performance among the UHPC-class materials. Furthermore, electrical resistivity at 28 days displayed a strong, inverse correlation with NSS rapid chloride migration coefficients. The inverse relationship between electrical resistivity and migration coefficient reveals that across the range of materials tested, cementitious composite materials with lower resistivity will likely experience deeper chloride ingress.

  • Time-to-corrosion-initiation estimations were developed in accordance with fib Model Code 34 [43]. Calculations for the majority of the UHPC-class materials yielded an expected service life greater than 100 years, and U-M yielded an expected service life of 55 years. The expected corrosion initiation timeframe for the conventional concrete mixtures was less than 20 years, with the exception of HPC, which yielded an estimated service life of 50 years. The service life estimations are based on the listed assumptions and are subject to change when real-time field data is considered. The reported service life values cannot be directly associated with the materials tested without consideration of real-time field performance.