Keywords

1 Introduction

Textile Reinforced Concrete (TRC) has emerged in recent years as a new and valuable construction material [1]. It can be categorized as a strain-hardening cement composite, made by a continuous textile fabric incorporated into a cementitious matrix consisting of a cement binder and small-size aggregates.

TRC technology is particularly promising for realizing durable and slender concrete structures, thanks to several factors:

  • High corrosion resistance of the non-metallic fibres;

  • Enhanced crack control, characterized by multiple cracking upon loading;

  • Fibres corrosion resistance, allowing thin-walled lightweight structures;

  • Possibility to align the yarns in the direction of expected tensile stresses, enhancing effectiveness in load-carrying capacity;

  • Textile formability, allowing complex freeform geometries.

On the other hand, monitoring of structural behavior can detect anomalies in due time, thus enhancing safety level and enabling maintenance and repair actions to be implemented more efficiently, directly impacting on the reduction of operating costs.

In the present work the TRC technology is implemented and further enhanced by studying textile sensorization, so potentially providing structural self-monitoring capability to the components, enabling its in-service integrity monitoring from the interior.

Fiber Optic Sensors (FOS), serving as both sensor and data conduit, are particularly interesting for the long-term health monitoring of civil structures [2]. Fiber Optic sensing can be implemented with two types of approach in terms of spatial distribution of measures, namely discrete and distributed measurement. While discrete measurement approach is usually preferable in terms of accuracy, resolution and dynamic behavior, distributed measurement is the cheapest and winning strategy when one does not know in advance where critical conditions can take place. Nevertheless, in distributed sensing the main challenge is to achieve an acceptable spatial resolution on sufficiently long distances [3]. The solution investigated in the present work is a truly-distributed sensing based on Brillouin scattering, which allow monitoring on several kilometers length, and on the application of a post-processing algorithm for the enhancement of spatial resolution.

Proper procedures for fastening the sensor to the textile before its insertion as panel reinforcement, as well as for protecting the sensor during and after concrete casting were preliminarily investigated. Two samples were then realized, one reinforced with traditional steel mesh, one reinforced with sensorized textile mesh. FOS and reference electrical strain gages were applied to both samples.

The panels were tested in flexure up to failure, investigating the effect of mesh substitution and of strain monitoring by means of the embedded distributed FOS.

The obtained results confirm the feasibility of the proposed set-up, though evidencing the need of few enhancements to the sensor embedding technique.

2 Textile, Sensor and Integration Procedure Definition

The first stage of the work led to the definition of the most suitable textile material and FOS and of a suitable procedure for integrating these two components.

Based on TRC literature analysis, specific requirements were set for the choice of the fiber material (in terms of alkali resistance) and the textile structure (in terms of tensile strength, fabric structure, mesh stability and size, yarn straightness, degree of orthotropy, finishing). A set of suitable commercial textile products were then selected and compared on the basis of the trade-off they can offer on such set of requirements (Table 1). This process eventually led to select the AR 640 TEXTILE produced by Kast, made of AR-glass fibers, sewed with a leno weave structure and coated with SBR finishing. The textile presented a tensile strength of 58 × 65 (kN/m) × (kN/m) and a mesh size of 3.45 × 3.45 cm × cm (the two figures refer to longitudinal and transverse directions).

Table 1 Compared textile products
Full size table

With reference to the sensor cables, the choice was made on the basis of their robustness (to survive concrete casting) and capability to provide adequate shear strain transfer from concrete. On this basis, distributed sensors produced by Solifos AG (hereinafter referred to as SOL sensors) and by Smartec SA (hereinafter referred to as SMA sensor) were selected, based on single-mode fiber and allowing two different temperature compensation strategies. More in detail, the SOL system consists of two different cables, a strain-sensitive one (trade name BRUsens DSS 3.2 mm V9 grip) and a strain insensitive one (trade name BRUsens DTS STL PA), while the SMA systems is a single multi-fiber cable containing both strain insensitive and strain sensitive fibers (trade name SMARTProfile). The SOL sensors have round-shaped cross-section with radius of 3.2 and 3.8 mm, respectively, while the SMA sensor cross-section is rectangular, with dimensions 8 × 4 mm. In terms of external surface, all the sensors present a polymeric sheat, which is roughened in the case of the SOL strain sensor in order to guarantee a more effective strain transfer.

Different solutions (see Fig. 1) for integrating the sensor inside the textile were evaluated and tested, including gluing, stitching and weaving (i.e. embedding during the textile production). Stitching was successfully implemented but eventually discarded because, in order to properly house the cable onto the textile warp, it was necessary to insert a ribbon between the sensor and the textile, which could, in principle, induce some bonding defects to the concrete-textile-sensor chain. On the other hand, proper weaving resulted unfeasible due to thickness and stiffness of the selected sensor cables. Continuous gluing by means of hot melt EVA was tested to quantify the shear adhesion strength between the sensors and a 15 cm portion of a single textile warp. An average adhesion strength of 2.1 and 3.1 kg/cm was found, respectively with SMA and strain-sensitive SOL sensor, which eventually led to select such solution for integrating the sensor.

Fig. 1
figure 1

Tested solutions for sensor-textile integration: gluing (left), stitching (center), weaving (right)

Full size image

3 Samples Design and Preparation

Two middle-scale sensorized concrete samples were designed and produced, both in panel form with overall dimensions 250 × 90 × 10 cm and equipped with lateral curbs made of steel bars: Sample A (reinforced with traditional welded steel mesh), used as reference, and Sensor B (reinforced with textile mesh). The reinforcing meshes were placed on the lower (tensile) edge. The two panels were designed in order to have the same flexural stiffness, applying the semi-competitive method to the Limit States in compliance with the current Standards and in agreement with the current Italian guidelines [4]. Both the samples were equipped with FOS and different types of reference electrical strain gages, as shown in Fig. 2.

Fig. 2
figure 2

Overview of the FOS and electrical strain gages and of their relative positions (drawing not to scale)

Full size image

Sensors were glued to the textile and their egress points from concrete were protected by including them in corrugated plastic ducts (Fig. 3). Textile reinforcement straightness during casting was guaranteed by clamping it on a perimetral steel rebar, serving also to fasten four steel hooks for panel demolding and handling. It’s worth mentioning that sample B showed a crack already before testing, likely due to an impact during handling.

Fig. 3
figure 3

Some steps of samples preparation

Full size image

After sensors installation and concrete casting, a bracket with standard optical connectors was applied to both ends of each cable, followed by connection and cable integrity testing by means of light source and power meter. Testing showed that all the FOS sensors survived both the processes of sensor fastening to the net and of concrete panel production, with acceptable optical losses.

4 Mechanical Testing

4.1 Test Set-up

The loading scheme is that of a simply supported beam loaded to flexure in one point, implemented as shown in Fig. 4.

Fig. 4
figure 4

Experimental set-up

Full size image

Neoprene pads were interposed between the panel and the supporting structures, in order to allow free rotation without stress concentration. Load was applied quasi-statically by means of two hydraulic jacks, anchored to ground and pulling down an IPE steel beam, able to evenly distribute load along the panel width. The jacks were controlled in parallel by a hydraulic pump.

The goal of the mechanical tests was two-fold, namely (a) Check reliability of the strain-measuring system by means of FOS and (b) Assess the structural performances of the textile reinforcement.

The used Data Acquisition Unit (FEBUS G1-R) implements, for the determination of the Brillouin Frequency shift, the so called “parabolic fitting” algorithm. Such algorithm is generally the fastest and most accurate way to determine the Brillouin shift but it can lead to poor results in zones of very high strain gradient, due to the lack of spatial resolution (1 m). Therefore, Febus provides an alternative algorithm for it, namely the “Spatial Resolution Enhancement” one, which allows obtaining an enhanced spatial resolution value of 25 cm.

4.2 Results

4.2.1 Assessment of the Measuring Behavior

General considerations

Figures 5 and 6 report the strain curves for Samples A and B, respectively, measured by the FOS, at increasing load. It’s worth recalling that the sensors are all connected in a row along the same channel. Though all the sensors were able to read data up to samples failure, the SMA sensors presented more irregularity at higher loads, likely due to sliding of the cable inside concrete due to its smooth surface.

Fig. 5
figure 5

Data obtained along the single FOS channel—Sample A

Full size image
Fig. 6
figure 6

Data obtained along the single FOS channel—Sample B

Full size image

Moreover, the strain-insensitive sensors, at higher loads, show strain values which are too high for being due to only thermal effects. This is likely due to fusion splicing, which provokes some small tension on the fibre and which, on the relatively small length of the tested sample, does not vanish. Hence, thermal compensation is unreliable in such situation; for this reason, taking also into account that temperature is nearly constant in a short time test and uniform along the panel, no thermal compensation was carried out in the present data analysis. It’s worth stressing, however, that, in real applications, where much higher lengths are monitored, the possible small tension due to splicing should vanish along the length, so that thermal compensation can be correctly implemented.

Strain profile from FOS at increasing loads

Data from sensors M-SOL-S1 (for Sample A) and S-SOL-S (for Sample B) are reported in Figs. 7 and 8, respectively. It can be noted that the curves at low load are quite irregular, showing also unphysical negative values, which probably means that the sensor system is still settling, with the sensors recovering some initial compression state. This finding leads to the suggestion to apply a slight pre-tension to the FOS. On the other hand, at the highest load values the strain pattern is again irregular, likely due to cracking.

Fig. 7
figure 7

Strain profile from M-SOL-S-1 at different load values (in kg)—Sample A

Full size image
Fig. 8
figure 8

Strain profile from S-SOL-S at different load values (in kg)—Sample B

Full size image

Strain comparison between FOS and electrical strain gages

M-SOL-S-1 and S-SOL-S are considered, respectively for Sample A and Sample B. As these sensors were bonded, respectively, to steel bar and to longitudinal textile thread, the corresponding strain gages considered for benchmark are the E2 ones, which are bonded onto the same elements. Results are shown in Table 2, where strain values at low load levels are not reported due to their unreliability, as explained in previous section.

Table 2 Strain comparison between FOS and electrical Strain Gages (SG)
Full size table

The above results confirm that FOS and strain gages provide strain values which are comparable, as order of magnitude. Nevertheless, a not negligible results scattering is observable. This finding must be interpreted taking into account, first of all, that the two sensors have very different gage length: about 1 cm for the strain gages, 25 cm for the FOS (spatial resolution). This means that the two values are not fully comparable, in principle. Moreover, at high loads concrete is cracked, then tension (hence strain) is released in the cracked concrete, while it is increased in the reinforcement, so assuming a very localized behavior. As a consequence, local values measured by strain gages are already scattered on their own. In this framework, it was hard to expect much closer results than observed.

4.3 Assessment of the Structural Behavior

Figure 9 reports the load-deflection curves obtained in the flexure tests, which allows us taking some conclusions in terms of global stiffness, ductility and ultimate load of the two systems.

Fig. 9
figure 9

Load-deflection curves

Full size image

In terms of stiffness, two different regions can be identified:

  • Region I: Different initial stiffness, likely due to the pre-existing crack in Sample B, which makes it less stiff than Sample A;

  • Region II: Same stiffness, according to design, where both panels are in post-cracking stage.

After Region II the stiffness is still comparable but not perfectly superimposable, due to diffused cracking.

In terms of ductility, Sample B shows a significant enhancement (ultimate deflection 56 mm vs. 43 mm of the Sample A).

Table 3 reports the ultimate load of the two specimens, in terms of design and experimental values.

Table 3 Ultimate loads of the tested samples
Full size table

It can be observed that Sample B attains a significantly higher (39% experimental, 25% theoretical) ultimate load with respect to Sample A, when the panels are designed for the same stiffness. In terms of difference between theoretical and experimental values, a slight increase is obtained for both samples, as expectable taking into account that design models are precautionary.

Finally, cracks appear more diffused in Sample B (Fig. 10), though it is hard to discriminate whether this is due to the different reinforcing materials (glass instead of steel) or to the different spatial arrangement of the reinforcing nets (more closely spaced in textile).

Fig. 10
figure 10

Crack patterns of the tested samples (details on the right are referred to top and bottom of Sample B)

Full size image

5 Conclusions and Acknowledgements

A study on design, implementation and testing of TRC panels sensorized with two different types of distributed FOS is presented in this paper. The potential of such a solution relies in the easiness of realizing a structural health monitoring system on site, namely by simply connecting each component to the adjacent one by means of fiber optic patch cords and by plugging the last one to the data acquisition unit. Moreover, this solution can enable in-service integrity monitoring from the interior of the component.

Two samples, one of which reinforced with traditional steel mesh, used as reference, were realized and tested in flexure, investigating the FOS performances in strain monitoring and the structural performances of TRC.

The main conclusions drawn from the research are reported in the following.

Regarding the FOS measuring capability:

  • All the FOS sensors survived both sensor fastening to the net and concrete panel production;

  • The sensors were able to read data up to advanced cracking stage;

  • The SMA tested sensors evidenced some sliding at higher loads, unlike SOL strain sensors;

  • Fusion splicing introduces a slight tension on the optical fibre, which makes unreliable thermal compensation at higher load level in short-length tests;

  • It is important to apply a slight pre-tension to the FOS at installation stage;

  • At high load levels FOS and strain gages provided comparable strain values, in the limits of the inherent differences between the two sensor types.

Regarding the load-carrying capability:

  • The two samples exhibited the same stiffness in the post-cracked phase, according to the design;

  • The TRC sample showed a more diffused crack pattern and enhanced ductility and ultimate load with respect to the steel-reinforced panel designed for the same stiffness.

The work was carried out in the framework of the Research Project EnDurCrete, (New Environmental friendly and Durable conCrete, integrating industrial by-products and hybrid systems, for civil, industrial and offshore applications) funded by the EU’s H2020 Framework Programme, under G.A. n. 760639. The Project is still ongoing and currently in the field demonstration phase.