The unidirectional (UD) reinforced fabric UT-E500 by GURIT Holding AG was used for the production of the GFRP formwork. UT-E500 consists of aluminoborosilicate (“E glass”) and has an area density of 500 g/sqm. For the thermosetting resin matrix, the epoxy resin Epilox® T19-27 by LEUNA-Harze GmbH was used. By means of manual laminating, UD single layers were made from the UD fabric and the epoxy resin. They had a fibre volume content of 30%. Within this UD single layer, the independent parameters: longitudinal (E1) and transverse (E2) moduli of elasticity, Poisson’s ratio (v₁₂), shear modulus (G₁₂) and the coefficients of linear thermal expansion α₁ and α₂ between +20˚C and +120˚C were determined experimentally. Table 1 shows the results of these tests.

The determined basic parameters of the unidirectional single layer (Table 1) were used for the analytical and numerical calculation of multi-layered formworks, which showed both a balanced symmetrical and an asymmetrical construction. The GFRP formwork with a plate size of 50 × 50 cm² was made of 11 UD single layers with total thickness of 3.3 mm (Figure 1).

After the production of the GFRP bond, the thermosetting matrix was cured in a heating cabinet over a period of 6 hours at a constant temperature of 120˚C. After curing, the GFRP bond were cooled to room temperature (20˚C). Due to this difference in temperature of −100 K, residual stress caused a small curvature of the anisotropic layer structure. Afterwards, high curvatures were caused by external preloading (up to a material load of R = 0.95 after CUNTZE criterion), utilizing the coupling effects that result from the GFRP bond (Figure 2).

The calculations of the coupling effects caused by anisotropy were conducted analytically with the Classical Laminate Theory (CLT) and the First Order Shear Deformation Theory (FSDT). For the experimental verifica-

tion of the anisotropic coupling effects calculated beforehand, selected GFRP bonds segments were produced and tested in the institute’s own structural test bench with the ABD-testing device (Figure 3).

The results were used during the further procedure for the determination of the functional relations of curvatures, process parameters and geometrical parameters. Based on that, the analysis and identification of single and double curved basic forms was conducted. Their combination resulted in a maximum of defined freeform surfaces.

The experimental analysis of the curves was conducted with the aid of the optical forming-analysis-systems ARGUS and ARAMIS by GOM (Figure 4).

For the production of the textile reinforced concrete elements, the textile 3D-fabric “SITgrid” by V. Fraas Solutions in Textile GmbH was used. It is made from alkali-resistant glass (AR glass), with 2400 tex in warp and weft (Figure 5), which had a tensile strength of 978 MPa. The fabric was placed on the formwork elements to show and stabilize the curvatures accurately. Thermosetting and thermoplastic resin systems were used for the stabilization. The resin systems were applied by spraying, rolling or by using of capillary effect of the pile thread on the pre-curved 3D-fabric. By means of the Fourier transform infrared spectroscopy with attenuated total reflection (ATR-FTIR), the impact of the pile thread on capillary suction with the epoxy resin Indufloor-IB1240 of the company SCHOMBURG GmbH could be studied. For this purpose the ATR-FTIR testing apparatus ALPHA-P by BRUKER DALTONIK GmbH was used.

The development of the fine grained concrete was focused on the workability of the fresh concrete as well as the

durability and good bonding between concrete matrix and textile reinforcement. Table 2 shows the qualitative and quantitative composition of the fine grained concrete mix. Apart from white Portland cement type 52.5 R, the fine concrete contained an amorphous aluminosilicate as puzzolanic binder (Table 2). Dolomite sand with a grain size of 0.5 to 1.0 mm was used as aggregate and dolomite powder with an average grain size of 70 µm was used as filler. The alkali-resistant (AR) short glass fibres (16 M.-% ZrO2) were 12 mm long and had a length mass of 45 tex. The high-performance superplasticizers had a polycarboxylate ether (PCE) contentof 30 M.-%. The water-cement-ratio was 0.37.

Selected representative freeform surfaces, double-curved with different radii of curvature, were produced in the production tests for the system structure GFRP formworks/concrete-lightweight-elements. The adjustment of the curves is carried out via the new flexible GFRP formwork elements. The basic proceeding when conducting the production tests comprised the production of the flexible formwork with reference curvature or preload curvature. After the production of the formwork, the spacer (Figure 6—left) of the reinforcement was positioned and the spacer fabric was installed, fixed and stabilized (Figure 6—middle ). After positioning and stabilizing the textile 3D reinforcement structure from Figure 5, the fine concrete was laminated (Figure 6—right), using the adapted concrete composition displayed in Table 2.

The theoretical major curves (for calculation approach cf. [12] ) of an asymmetrical layer structure with 0˚- and 90˚-layer content (90_n/0_m) are exemplarily depicted for two loading cases in Figure 8, dependent on the 0˚-layer content. The curves increased with increasing layer content accompanied by increasing anisotropy, both in the 1- and in the 2-axis (Figure 7).

The highest anisotropy was found with a relative 0˚-layer content of about 50 percent. This caused the largest

curvature around both axes (Figure 7 and Figure 8). A further increase of the 0˚-layer content caused a decrease of curvature, because the anisotropy of the GFRP layered bond structure decreased.

Due to external preload forces, the tensile load caused an increase of curvature around the 2-axis. At the same time, curvature around the 1-axis decreased, due to traction.

Figure 9 displays the theoretical and experimentally verified major curves of the asymmetrical layer structure (90_n/0_m), depending on 0˚-layer content with a pure temperature load of (ΔT = −100 K). The major curves around the 1-axis that were determined experimentally agreed qualitatively and approximately also quantitatively with the major curve that was calculated previously (Figure 9 and Figure 10). The minor quantitative differences between the calculated and the experimentally verified major curves around the 1-axis were caused by physico-chemical reactions of the thermosetting matrices. In the calculations, these matrices could be included only insufficiently. Apart from chemical shrinkage, they included residual stress caused by swelling after increased water absorption in or among the molecular chains of the thermosetting matrices [17] .

In contrast to that, there are greater differences between the calculated and experimentally verified major curves around the 2-axis (Figure 9 and Figure 10). In reality however, only one of the major curves prevails, due to stability problems (problems with the transmission/distribution of forces). In this case, this is the curvature around the 1-axis.

Technique and technology from textile manufacturing had an influence on shape through coating with a thermosetting resin system. Mechanical characteristics were adapted to the variety of shapes and drapery. This resulted in an exact contour adapting (Figure 11—left). Reverse deformation was less than 5 percent one day after the stabilization (Figure 11—right). Thus, an exact depiction of the curvature state could be ensured.

Further production studies aimed on an additional stabilization with the cold-setting epoxy resin Indufloor-IB1240 by using of the capillary effect of the pile threads. However, it was not possible to prove an increase in stiffness of the 3D-fabric, due to the insufficient capillary effect of the spacer threads (Figure 12). That

can be recognized on the sample “pile thread without epoxy resin contact”, which had no common quantitative characteristic peaks with the sample “epoxy resin”. As opposed to this, the sample “pile thread with epoxy resin contact” had a high quantitative and qualitative peak alignment with the sample “epoxy resin” (Figure 12).

Table 3 shows the characteristics of fresh concrete and hardened concrete after 28 days. In the fresh concrete, an air content of 2.5% by volume and a geometrical density of 2.32 g/cm³ were found by the use of an air entrainment meter. The total shrinkage deformation, measured with a shrinkage drain, was 0.71 mm/m. The compressive strength and 3-point bending tensile strength were 109.3 MPa respectively 14.74 MPa. For the textile reinforced concrete, a 3-point bending tensile strength of 41.51 MPa was measured.

Of special importance during the production tests was to ensure the evenness of the concrete layer thickness, consistently good surface quality, sufficient stability of the GFRP formwork and to avoid critical cracks both in the concrete and in the formwork system. Also essential were good sheeting qualities (Figure 13).