Comparison between cold incremental and stretch forming of flax fiber-reinforced polypropylene composites

The composite laminates used in this study (area of 200 × 150 mm², thickness of 2.2 mm) were manufactured using neat PP films (supplied by GDC S.r.l.; thickness of 0.5 mm and density of 0.92 g/cm³) and a woven fabric of flax, supplied by FIDIA S.r.l. - Technical Global Services. The fabric had a mass per unit area of 320 g/cm², a tex number of 324 g/km, and was safely stored in a polymeric bag under vacuum at 20 °C and 45% humidity. As previously mentioned, it was not subjected to any previous chemical or surface treatment. Before the molding process, it was dried at 60 °C for 12 h to eliminate any trace of humidity. A figure of the fabric and a magnification of a single yarn constituted by the winding of filaments are shown in [53], while their main properties are summarized in [51].

The laminates were produced using a conventional compression molding press, considering the stacking sequence (five layers with a symmetric layup; the first two and the last two layers were PP films, while the central layer was flax fabric) and following the operations schematized in Fig. 1. Specifically, the molding temperature was 200 °C, and the total molding time was 300 s, with the first 120 s being the dwelling time, a waiting period after which the plates were closed, applying a pressure of 4 MPa for the remaining 180 s. The choice of a woven fabric with a large mesh size and the process parameters mentioned above proved to be an effective solution for the manufacture of flax and hemp fiber-reinforced PP composites, even with a different stacking sequence; compared to unreinforced PP, these composites showed notable improvements in tensile and bending properties, as well as higher bearing capacities and service temperatures [53]. Tensile tests according to ASTM D3039 standard were performed using an MTS Alliance RT/50 universal testing machine, equipped with a 5 kN load cell and an MTS 634.31 F-24 extensometer (five replications for both reinforced and unreinforced laminates). The reinforced laminates showed increased tensile properties, with strength and elastic modulus equal to 37.3 MPa and 2.3 GPa (the corresponding values for the unreinforced PP were 27.1 MPa and 1.2 GPa, respectively). The laminates exhibited bilinear behaviour [54], which is attributed to the complex structure of the flax fibres. These fibers consist of cellulose microfibrils helicoidally wound along the fibre axis and embedded within an amorphous hemicellulose matrix. Under loading, the microfibrils tend to realign in the direction of the applied load, while the surrounding matrix exhibits a viscous response. As a result of this structural reorganization, combined with the inherent PP matrix viscoplasticity, the composite displays a similarly viscous and nonlinear mechanical response [55].

Spherical caps (see the schematization in Fig. 2a, where a and θ denote the base radius and the polar angle, respectively) were manufactured using cold NIF and SF processes (see Figs. 2b and 3 for a schematization and the actual equipment of the tests); the forming tests were conducted using a C.B. Ferrari high-speed four-axis vertical machining center.

The laminates were secured using a clamping frame with a square working area of 100 × 100 mm². To reduce the sheet bending defect close to the base of the cap, the tests were also conducted using a hollow cylinder as a partial counter die; differently from a full counter die, it only supports the essential areas of the sheet, allowing to manufacture components with some similarities and preserving the process flexibility. Additionally, the probability of failures and defects was reduced by carrying out the tests under lubricated conditions, using Boelube 70,104 (100 A) synthetic lubricant, developed by Boeing and supplied by Orelube.

The incremental and stretch forming tests without partial counter die are labelled as NIF₀ and SF₀, respectively, while the corresponding ones with partial counter die as NIF_PD and SF_PD, respectively. Two tests for each different case were performed.

For the NIF tests (see Fig. 3a), a non-rotating stainless-steel stylus with a hemispherical head, 10 mm in diameter, was driven by the CNC machine at a nominal speed v_NIF = 1000 mm/min to impose progressive deformation on the laminate. The tool followed a path with helical turns that alternated in anticlockwise and clockwise directions. This approach, based on observations from metal [12] and polycarbonate ISF parts [56], significantly reduced the probability of twisting, a defect caused by the uncontrolled pivoting of the formed component around the clamping frame due to in-plane forces exerted by the tool. Notably, the twist generated in one turn is almost entirely recovered in the subsequent one [57]. Figure 4 shows a not-to-scale representation of some turns of the toolpath; θs = 1° is the angular step down, i.e., the angular distance described after one complete turn.

For the SF tests (see Fig. 3b), the machine simply imposed on the die a vertical displacement, equal to the cap height (equal to 18.65 mm), at a nominal speed v_SF = 60 mm/min.

To evaluate the geometrical accuracy of the processes, the shape of the caps was measured using a Zeiss DuraMax coordinate measuring machine (measurement accuracy of 2.4 μm) and Calypso software and then compared to the target geometry. A ruby sphere stylus with a diameter of 3.0 mm was used for the measurements. Each measurement involved 450 individual points evenly distributed across the diagonals AC and BD (see Fig. 2a).

The wall thickness was measured by a 0.01 mm resolution micrometer (five measurements for each case), while observations of failure zones were made by a Hirox RX-100 digital microscope.

To estimate the magnitude of the forming loads, F_X, F_Y, and F_Z forces were acquired at 50 Hz by the K-MCS10 multicomponent sensor (fixed between the clamping fixture and the base plate of the CNC machine, see Fig. 3), equipped with the QuantumX MX840B data acquisition system and the Catman Easy AP software. From their combination, the magnitudes of the force in the XY plane and of the total forming force (labelled as F_XY and F_TOT, respectively) were also obtained.

The forces also enabled the measurement of power (P) and energy consumption (E). Unlike the evaluation of the electrical energy, the use of forces allows for the estimation of the actual energy required for the process, which represents only a small fraction of the total energy consumed, as most of the energy demand is associated with auxiliary functions of the equipment [58].

For the NIF tests, P was obtained by only considering the contribution of F_XY, according to the following equation:

This simplification is possible because, for the NIF processes, v_NIF can be approximated with the speed in the XY plane, due to the low θs value that makes the path almost horizontal, while the speed along the Z axis is very low and does not significantly contribute to the total power and energy. This was observed in a previous study by the authors on NIF of cones and spherical caps starting from laminates of flax and hemp fiber-reinforced PP composites [59]; the simplification resulted in an underestimation in terms of energy of less than 2.5% in the worst case.

For the SF tests, P was determined by:

E trends were determined as time-integrals of the P curves. The Riemann integral was used, with a regular partition of the time interval equal to 0.02 s, i.e. the period of acquisition of the forces.

Spherical caps with a = 40 mm and θ = 50° were manufactured using NIF and SF processes, with and without a partial die (with internal and external diameters of 80 and 100 mm, respectively); an NIF_PD cap is reported in Fig. 5.

In all cases, the parts were sound and had good surface quality. Figure 6 shows the actual and the target cap shapes (for easy reading, only half of the experimental profiles are reported). Two features were evaluated to estimate the geometrical accuracy of the forming processes, i.e. the difference in maximum height (dh₁) and the gap in correspondence of the intersection between the base of the cap and the flange (dh₂); Table 1 summarizes these values.

The NIF process guaranteed a more accurate geometrical quality due to the incremental and localized nature of the deformation mechanism, and the use of the partial die improved it, particularly near the clamped zone. Moreover, non-severe working conditions in terms of forces and moments and limited thinning were predictable, in view of the lack of instabilities and wrinkling [60]. Additionally, the selected toolpath strategy proved effective in preventing twisting. Concerning the thinning, from wall thickness measures it ranged between 0.02 mm on the top of the caps and 0.09 mm, in line with the above.

In contrast, the SF process proved to be completely ineffective, in both the variants. The stretching mechanism proved to be unsuitable for these geometries when starting from cold laminates, resulting primarily inelastic deformations that were almost entirely recovered after processing. The low formability efficiency of the SF processes was further confirmed by more severe forming tests to obtain spherical caps with a = 20 mm and θ = 70° (see Fig. 7). While the NIF tests were concluded without incurring failures (Fig. 7a), the SF tests failed, as highlighted in Fig. 7b; the laminates experienced localized cracking, as a result of excessive matrix loading after fibers breakage [61].

Figure 8 reports the trends of F_TOT. Concerning the NIF tests (Fig. 8a), the fluctuations of the trends reflected the alternating nature of the toolpath. The initial part of the NIF_PD curve had a higher slope, due to the contact of the laminate with both the tool and the counter die, which made the system highly stiff. This resulted in reduced dh₁ and dh₂ values, as the flange acted as a weak constraint in NIF₀, compared to the more effective action provided by the counter die in NIF_PD. The curves reached their maximum value, corresponding to the condition of maximum tool/sheet contact, during which the processes exhibited their most effective incremental deformation of the laminates. The final part of the curves showed a decreasing trend, as a consequence of the less severe working conditions encountered when the tool reached the top of the caps, due to the decreasing wall angle. Figure 8b reports F_TOT trends when using the SF processes. The forces continuously increased, because of the increasing contact area between the die and the laminate as the tool displacement increased. But they did not result effective in terms of formability, as observed by Fig. 6.

Figure 9 reports the power curves. They obviously followed the same trend of the forming forces, both for NIF (Fig. 9a) and SF tests (Fig. 9b), since they were obtained by multiplying them and constant values of velocity.

Finally, Fig. 10 reports the energy curves. They showed an increasing trend, because they were obtained by integrating the power curves.

Table 2 summarizes the results of this subsection, reporting the maximum values of F_TOT, P and E. In all cases, higher values were recorded when using the partial counter die, because of a higher stiffness of the system. The low forming forces for NIF processes confirmed the above predicted non-severe working conditions and the usability of non-dedicated tools and machines. This was not true for SF processes, for which the higher values of forces reached did not guarantee forming efficiency. Both NIF and SF processes required low and similar power levels, while the energy for NIF processes, compared to SF, was higher due to the high process time but guaranteed a good formability.

The forming test shown in Fig. 7b was also considered as a penetration test to compare the mechanical performance of undeformed and postformed laminates. For the second ones, NIF_PD caps were clamped with the concave side up. Figure 11 clearly highlights the increased strength of the spherical caps, with a penetration force (the maximum value of F_TOT) more than doubled, compared to the undeformed laminate.

Figure 12 reports the 30× magnification failure surfaces from a penetration test on a postformed laminate (but they are similar for the undeformed ones). The failures at the end of the tests were perpendicular to the yarns (Fig. 12, up); then, the fibers played the role of PP reinforcement. The figure highlights that stress transfer was primarily governed by mechanical interlocking, promoted by the large mesh size, rather than by fiber/matrix adhesion. The latter is weak in this composite system (Fig. 12, down) due to the contrasting chemical natures of flax fiber and the PP matrix, hydrophilic and hydrophobic respectively [62], especially in the absence of fiber treatments and coupling agents [43].