1. Introduction
Ultra-High-Performance Concrete (UHPC) is characterised by a compressive strength above 150 MPa and high durability. These properties are based on a high packing density, a low water/binder ratio < 0.25, and the use of reactive fillers [
1].
In contrast to its high compressive strength, the tensile and flexural strength of UHPC are comparatively low. Furthermore, the failure behaviour of UHPC is brittle and abrupt due to its high packing density [
1,
2,
3]. To improve these characteristics, reinforcement in the form of fibres or rebars (or a combination of both) is added to the concrete [
4,
5,
6]. Fibre reinforcement is used to introduce a more ductile failure behaviour, while a rebar reinforcement strongly improves the tensile and flexural strength of the concrete [
7,
8,
9,
10,
11,
12]. However, buildings with reinforced concrete must adhere to different regulations, e.g., the DIN Norm in Germany. These standards control the manufacturing of building components to ensure sufficient mechanical properties for the use case. Compliance with these regulations sometimes counteracts the environmental approach of using CO
2-friendly building materials and, therefore, is not favourable in terms of life cycle assessment. The possibility of manufacturing material-saving components with sufficient mechanical properties is a task that has been examined for a long time now.
Current research in the fields of UHPC addresses these problems in different manners. The use of waste and recycled materials reduces the CO
2 output of UHPC while maintaining sufficient mechanical properties [
13,
14,
15]. Another approach is the optimisation of the mechanical properties by combination of different materials and using novel methods for applications [
16,
17]. One of these approaches is the use of prestressed high-strength steel rebars for reinforcement in concrete [
10]. These rebars introduce a compressive stress into the concrete that counteracts the achieved tensile stresses in the application. This technique has been known for quite a long time. P.H. Jackson registered a patent for prestressed floor plates in 1886, and the first prestressed bridge was built in 1938 in Germany [
18]. Today, prestressed concrete has reached nearly all parts of constructive engineering. Most prestressed concrete is used in bridge and building construction [
10]. Prestressed concrete allows an increase in the spans of reinforced concrete beams, as well as reducing material components due to the enhanced mechanical properties.
These advantages are based on a system that counteracts the resulting stresses in the application of a component. Therefore, steel rebars are placed in the area where the component will receive the highest tensile stresses. Subsequently, the rebars are elastically pre-tensioned with hydraulic devices and fixed externally on the concrete formwork while the concrete is poured around it. In the case of a bond-free prestress application, cladding tubes are used, and the concrete is prestressed due to an external anchorage [
10]. After sufficient hardening of the concrete, the rebars are detached from the anchorage device and, therefore, try to contract due to their elasticity. The bonding with the surrounding concrete and the anchorage prevent this contraction and transfer the resulting stresses into the concrete (
Figure 1).
The prestress counteracts the appearing stresses due to loads in the application. This increases the serviceability and, thus, increases the life cycle of the component. These advantages are mainly based on minimizing crack formation, less deformation of the component, and less susceptibility to vibration and fatigue [
10].
In contrast to the named advantages, prestressed concrete also has some disadvantages. The main disadvantage is that the manufacturing process is very time-intensive and requires a complex setup to enable the prestraining of the rebars. Furthermore, shrinkage and creep of the concrete as well as relaxation of the prestressing steel can lead to a decrease in the prestress level [
10]. Stress corrosion cracking is also a problem, as constantly stressed steel is much more susceptible to corrosion than regular steel [
19]. Therefore, prestressed structures need to be monitored over time [
10,
20,
21].
Recent research has shown the feasibility of using shape memory alloys (SMAs) as prestressing elements in concrete [
22,
23,
24,
25,
26,
27]. SMAs have the ability to return to their original imprinted shape after large deformations. Depending on the inherent properties of the SMA, the shape recovery is initiated upon removal of applied stress or upon heating of the SMA. The according shape memory phenomena are called superelasticity (SE) and the shape memory effect (SME) [
28]. These unique properties are based on a reversible solid-state phase transformation between a martensitic low-temperature phase and an austenitic high-temperature phase and vice versa [
29]. The SME was discovered in the late 1940s by Kurdjumov and Khandros [
30], but the commercial breakthrough of SMAs was achieved with the development of the Ni-Ti alloy published in 1963 by Buehler et al. [
31]. Ni-Ti is still the most commonly used SMA; however, its high costs resulting from expensive alloying elements and its challenging machining process [
32] impede extended use, especially for material-intensive applications. With regard to applications in civil engineering structures, low-cost iron-based shape memory alloys have recently been the focus of research [
33,
34,
35,
36,
37,
38].
To utilize SMAs for prestressing applications in UHPC, the thermally induced SME is used. The procedure can be divided into three steps. The first one is prestraining of the SMA tendon. After subsequent unloading, the second step is the application to the construction element that has to be prestressed. Referring to the example of the prestressed concrete beam in
Figure 1, the SMA rebar is placed into the mould in the area where the prestress is needed in the application. Finally, after pouring and sufficient hardening of the concrete, the activation of the SMA rebar is carried out. Therefore, the complete concrete element, including the SMA rebar, is heated up above the alloy-specific activation temperature. Due to the thermally induced phase transformation, the SMA rebar aims to reshape itself into its original imprinted (non-prestrained) shape. As it is bonded with the surrounding concrete, no strain recovery is possible and, therefore, a compression stress is imposed on the concrete beam. The thermal activation can be accomplished by convective heating with a heat source, conductive heating with electric current, or inductive heating.
Compared to conventional prestressing steels, the use of SMAs has several advantages. First, no anchorages and no hydraulic tensioning devices are necessary for applying prestresses to the concrete, which significantly simplifies the manufacturing process. Furthermore, this offers additional design opportunities, as the SMA rebars can also be placed in a curved geometry into the concrete, which is not possible with the conventional prestressing steel, if a prestressing in a fixed bond is envisaged.
The Fe-Mn-Al-Ni SMA used in this work is quite new. Vollmer et al. [
35] demonstrated the prestressing potential of Fe-Mn-Al-Ni for the first time in 2021. The functional properties of Fe-Mn-Al-Ni are strongly related to nano-sized precipitates being present after solution annealing and subsequent quenching. The transformation temperatures are strongly related to the size of these precipitates. By applying an aging heat treatment at temperatures around 200 °C, the size of the precipitates increases, leading to a shift in transformation temperatures to lower values. Therefore, the alloy can show the SME in the initial state and the SE after aging. The modification of functional properties can be very beneficial for prestressing applications. The activation of the SMA prestressing element in the temperature regime causing precipitate growth enables multi-stage, tailorable prestress levels. Moreover, the change in functional properties from SME to SE can reduce prestress losses. The prestressing level can be tailored by the maximum heating temperature. Moreover, lower heating temperatures can be compensated by longer dwell times of heating. The maximum prestress level shown by Vollmer et al. for Fe-Mn-Al-Ni was about 485 MPa and is, therefore, slightly lower than but in the general range of regular prestressing steels [
39]. However, it should be mentioned that the functional properties in Fe-Mn-Al-Ni are strongly influenced by the microstructure present, which requires precise temperature control during heat treatment and quenching.
This study investigates the use of the Fe-Mn-Al-Ni SMA as a prestress reinforcement in UHPC. The high strengths of the UHPC as well as the high amount of binder result in an enhanced bonding between rebar and cementitious matrix [
40] and, therefore, are most suitable for the prestress application. The aim of this study is to generate a proof of concept for the general suitability of the Fe-Mn-Al-Ni SMA as a prestress element for UHPC. To have a general comparison to reference material, additional investigations with steel rebars were also carried out. To quantify the bond strength, rebar pullout tests out of a UHPC were conducted. The influence of the activation of the SMA was determined by flexural bending tests of concrete prisms reinforced with SMA rebars as well as steel rebars with different diameters and in different setups.
4. Discussion
The focus of this study was on the capability of Fe-Mn-Al-Ni shape memory alloys to be used as tendons within UHPC. The aim was to generate a proof of concept for the general usability of the Fe-Mn-Al-Ni SMA as a prestressing tool in UHPC. This was investigated by quantifying the influence of the thermal activation of the prestrained SMA rebars and the generated prestress. Flexural bending tests of UHPC prisms reinforced with SMA and steel rebars in different test setups were carried out. Furthermore, rebar pullout tests were conducted to describe the influence of the activation process on the bond between rebar and cementitious matrix.
The microstructural investigations of the Fe-Mn-Al-Ni SMA revealed an unfavourable metallic microstructure for an application as a tendon. This resulted in a recovery stress that is significantly lower than regular prestressing steels but also is similar to iron-based shape memory alloys, which are often in range of about 250 MPa to 600 MPa [
26,
34,
49], or the well-known NiTi-SMA, with recovery stresses of about 250 MPa to 500 MPa [
50,
51]. Thus, a quantitative comparison of the generated results is not directly suitable, but a proof of concept for the use of Fe-Mn-Al-Ni shape memory alloys as a prestressing element in UHPC was achieved.
The flexural bending tests revealed the prestress potential of the novel Fe-Mn-Al-Ni SMA that was proposed by Vollmer et al. [
35]. In all setups, prestraining and thermal activation of the samples led to an increase in maximum flexural strength compared to the untreated SMA rebars. This is supported by the fact that heating of samples with unprestrained SMA rebars did not lead to a significant increase in maximum flexural strength. Therefore, this increase can be attributed directly to the activation of the shape memory effect within the prestrained SMA rebars. However, this effect was mainly visible in the samples containing rebars with a diameter of 5 mm. The samples with the 3 mm rebar reinforcement did not show a significant improvement by thermal activation of the rebars in either tested setup. Furthermore, heating of the samples reinforced with 3 mm steel rebars, revealed a significant increase in maximum flexural strength. This is supposed to be due to the generally lower bond strength between steel and the UHPC matrix. The heating shows a more prominent influence on the bond strength compared to the SMA rebars. In the case of the SMA rebars, a sufficient bond is gained by the rough surface (
Figure 8), which superimposes the effects of heating on bond strength. Due to the low-volume fraction of reinforcement in these samples, the shape memory effect is not that distinct and did not lead to a significant improvement regarding maximum flexural strength.
A direct quantitative comparison with the steel rebars is not suitable, as the surface structure and the mechanical properties of the materials are significantly different. As both characteristics have distinct influence on the bond strength [
12,
40,
52] and, therefore, on the performance as reinforcement, the results regarding steel rebars have to be interpreted in a more qualitative than quantitative manner. The SMA rebars with a diameter of 3 mm led to a similar flexural strength after prestraining and thermal activation compared to the roughened and thermally treated steel rebar samples. Due to relatively low maximum tensile strength combined with the good bond due to the rough surface of the SMA material, most of the rebars tore in the testing process. The failure could also be promoted by the microstructure of the rebars, as the surfaces also revealed cracks and depressions (see
Figure 8). More cracks inside the rebars, therefore, cannot be excluded. These defects lead to no existent post-failure behaviour for the samples with SMA reinforcement, and, thus, a quantitative comparison with the steel rebars is not suitable. However, the samples with 5 mm SMA rebars reveal a worse post-failure behaviour compared to the 3 mm steel rebars. This is in accordance with the rebar pullout tests, as the steel rebars led to a slightly better post-failure behaviour compared to the SMA rebars. Thus, a direct relationship of pullout behaviour and post-failure behaviour cannot be assumed, which is also shown in literature [
12,
53]. To achieve a better comparability with the steel samples, the Fe-Mn-Al-Ni alloy must be designed with similar mechanical, microstructural, and topographic characteristics. This would counteract tearing of the rebars during the tests and should allow much higher resulting flexural strengths of the reinforced concrete. Possible ways to achieve this could include the addition of a fifth element into the alloy system as well as different quenching procedures in the manufacturing process, as these parameters have a significant influence on the mechanical and microstructural characteristics of the material [
54].
Sawaguchi et al. investigated the prestress potential of Fe-Mn-Si SMA [
55]. They reinforced prisms (20 mm × 20 mm × 80 mm) made of a high-strength mortar with prestrained SMA wires with a diameter of 2 mm. A recovery stress of about 250 MPa was determined for these wires. By activating the wires, the flexural strength of the mortar was increased in a range of about 15% to 40%, depending on the activation temperature and duration. Shahverdi et al. revealed an even higher increase in flexural strength using Fe-Mn-Si-Cr-Ni SMAs as reinforcement bars inside a normal-strength concrete (compressive strength: 53 MPa) [
56]. The rebars were placed inside grooves that were cut out of the concrete and were afterwards filled with a cement-based grout. Prior to testing, the rebars with a recovery stress of 220 MPa to 250 MPa were thermally activated. The activation led to an increase in maximum bearing load of about 80% in relation to the non-activated samples. Compared to the results of this study, where a maximum increase in flexural strength was about 26%, the more distinct influence could be assigned to the higher recovery stresses of the used reinforcement rebars. This is supported by the fact that Vollmer et al. revealed generated prestresses of 485 MPa with the same alloy system that was used in this study [
35]. The comparably low prestress in this study can be assigned to the high abundances of martensite in the initial state of the rebars. Therefore, by limiting the martensite percentage in this state, higher prestress levels can be assumed [
35].
Kromoser et al. examined the feasibility of an iron-based SMA-grid as prestressing reinforcement in UHPC [
23]. The reinforcement in the form of a grid has the advantage of a much better bond between the SMA and the matrix, as the cross braces of the grid serve as anchorage inside the UHPC. Therefore, they tested reference and reinforced UHPC beams. The activation of the SMA led only to a slight increase in the bending strength of the beams compared to the non-activated samples. They found that the activation led to an increase in the flexural rigidity of the beams, as the samples with the activated SMA revealed a much higher displacement at maximum force compared to the non-activated samples. This is in accordance with the results of this study, as the displacement at maximum flexural strength also increases mostly with the activation of the SMA rebars.
The rebar pullout tests revealed a much higher bond strength of the SMA rebars compared to the steel rebars. This was expected due to the much rougher SMA surface (see
Section 3.3) due to corrosion products [
12,
52]. This difference in bond strength was eliminated by roughening of the steel rebars. Compared to NiTi-SMA, the bond strength is significantly higher for the Fe-Mn-Al-Ni SMA [
40,
57]. Other studies mentioned a lower bond strength of a Fe-Mn-Si-Cr-Ni SMA rebar compared to regular steel rebars, which contrasts with the results of this study [
58].
The prestraining of the SMA rebars and the heating of the specimens had a slight negative influence on the bond stress, but not in a significant manner. This is in accordance with the literature, where it is stated that activation temperatures of ≤200 °C only led to slight changes in bond strength [
59,
60]. Regarding the results in this study, the slight decrease could be due to the expansion of the rebars during activation. When the rebars are prestrained, the diameter is reduced slightly. When retaining its original shape due to thermal activation within the UHPC matrix, the SMA rebars transfer a stress onto the surrounding concrete, analogical to the Hoyer effect [
61]. When these stresses are too large, cracking around the rebar could occur, resulting in loss of bond strength. However, as regular prestressing steels mostly have a rippled surface structure, this could also be applied onto SMA rebars, and, therefore, those would have a much more prominent impact on the bond compared to the differences regarding rebar material [
10].
Due to the low tensile strength of the SMA rebars, some samples tore during testing. This applies to the rebar pullout tests as well as to the flexural bending tests. The latter is especially unfavourable, as no post-failure behaviour could be achieved due to rupture of the rebars. This was mainly found in the setups using the thinner rebars. In these setups, the steel rebars led to better performance factors, as much higher stresses could be received after failure of the UHPC matrix compared to the SMA rebars. Furthermore, due to the lower tensile strength of the SMA rebars, the maximum tensile strength achieved is just about in the range of the roughened steel rebars. Therefore, to achieve even better values compared to regular steel rebars, the alloy has to be adjusted in terms of tensile strength, as a higher tensile strength of the reinforcement increases the mechanical properties of the component [
10].
In general, the ability of Fe-Mn-Al-Ni SMA to be used as prestressing tendon in UHPC was shown within this work. As a next step, the SMA has to be adjusted in order to improve its mechanical properties and also to improve the stability of the system, i.e., to ensure the necessary microstructure. Therefore, the manufacturing process has to be further investigated, and the influence of the processing routes (annealing temperature and time, quenching temperature and time, etc.) on the microstructure and the transformation temperatures has to be characterised [
35,
62,
63,
64]. Another aspect resulting from the processing route is the chemical composition of the alloy. The alloy composition of the used material differs from known and functional samples [
35] by up to 13 at.% in regards to the four main elements of the Fe-Mn-Al-Ni SMA. This is, of course, related to the composition of the base material prior to the manufacturing of the rebars; however, the annealing and quenching processes can also lead to differences in chemical composition and microstructure [
35,
65,
66]. As the rebars used in this work were annealed at 1250 °C together with Mn pellets within steel foil, the Mn pellets could have led to a slightly higher Mn concentration in the SMA samples. Mn is known to be a stabilizer for the γ-phase [
47]. That could explain the high amounts of γ-phase in the SMA samples of this study.
In terms of application in civil engineering, different aspects have to be investigated in further research. As mentioned before, the influence of stronger rebars has to be examined. With stronger rebars, a stronger bond can also be used to create even higher prestresses. Furthermore, long-term investigations must be carried out. One of the most beneficial properties of the Fe-Mn-Al-Ni SMA is that they can hold the prestress over long strains [
35,
67,
68,
69]. Therefore, this ability could counteract prestress losses due to shrinkage and creep of the concrete, which is a problem for existing prestressing concepts [
70,
71]. A combination of fibre and SMA rebar reinforcement could also be used in order to increase the effectiveness of the shape memory effect. A stronger bond between rebar and UHPC due to fibre reinforcement could lead to a more effective implementation of the prestress [
72]. Alignment of the fibres could improve this effect even further [
5]. However, this improvement in bond strength must be in accordance with the tensile strength of the used SMA rebars.
These aspects will be investigated in future research. The proposed method reveals a huge potential, and a proof of concept is shown. The used Fe-Mn-Al-Ni alloy has unique characteristics that allow novel applications and will counteract known issues of regular prestressing systems. These novel applications must be investigated further, and the results shown in this study are the basis for upcoming research. However, quantitative comparison with other methods is not applicable here, as too many open questions are still existent.