Wet Compression Molding of Biocomposites for a Transportation Industry Application

Abstract

The transportation and automotive industries are slowly integrating biocomposite materials into products where the economics make sense; this typically means a short manufacturing cycle time, not using expensive prepreg, and with little waste generated from the process. In a previous investigation into the use of biocomposites for electric bus seats and backs, three different material systems (hemp, flax, and pure cellulosic fibers, each paired with a high-bio-content epoxy) and two manufacturing processes (wet layup followed by compression molding, vacuum-assisted resin transfer molding) were investigated, but neither process proved to be viable. In this paper, a relatively obscure process called Wet Compression Molding (WCM) is considered for economical production of the biocomposite bus seats using the same three material systems. Darcy’s law predictions of full impregnation time for a nominally 3.5 mm thick part using experimentally determined permeability values are all less than 2 s. Furthermore, prepreg is not used, and net-shape parts without excess resin show potential. Important design details of the WCM mold set, used in the manufacturing of flat test panels from each material system, that are generally not discussed in the literature include a high-pressure O-ring seal, and semi-permeable membranes covering injection pins and vacuum vents (evacuates trapped air) to prevent resin ingress. Biocomposite laminate specimens are fabricated using the mold set in a thermal press and a vacuum pump. Part characterization includes fiber volume fraction estimates and measurements of thickness, density, flexural modulus, and outer fiber maximum stress at failure. Due to its rapid impregnation with just enough resin, WCM should be considered for the economical manufacture of parts similar in shape and size to electric bus seats and backs.

1. Introduction

Despite the pervasive use of synthetic materials in manufactured components, the automotive and transportation industries are slowly finding more and more applications for biocomposite materials, primarily for interior components [1]. A recent example considered the use of biocomposites for passenger seats in an electric bus, where structural performance, environmental impact, and cycle time for different material systems and manufacturing processes were compared [2]. Specifically, three natural fiber reinforcements (BioMid cellulosic, hemp, flax), two different epoxies with a high bio-based content, two resin impregnation methods (wet layup, vacuum-assisted resin transfer molding), and two consolidation/curing processes (compression molding, pressure focusing layer tooling) were investigated for their use in manufacturing laminate seats and backs. Despite several material/process combinations yielding acceptable mechanical properties, all cycle times were too long, and significant manufacturing waste was generated. A faster resin infusion process that generates less consumable volume (i.e., materials used to facilitate manufacturing but not incorporated into the final product) is necessary to meet these demanding manufacturing requirements.

In a follow-up study to Ref. [2], a relatively uncommon manufacturing process called Wet Compression Molding (WCM) was considered for the same structural components based on the following material/manufacturing requirements:

• Natural fiber reinforcement;
• High-bio-content thermoset resin as the matrix;
• No prepreg;
• Manufacturing cycle time less than 10 min for a 3–4 mm thick laminate;
• Low use of consumables;
• Simple tooling design.

WCM involves impregnating the dry preform with a layer of liquid resin through its thickness. The primary advantages of WCM (and Compression Resin Transfer Molding or CRTM) over other resin infusion methods such as Resin Transfer Molding (RTM), Light RTM, and Vacuum-Assisted RTM are that infusion times are at least an order of magnitude lower, since the infusion distance (through-thickness vs. in-plane) is significantly reduced [3], and little to no waste is generated. However, WCM differs from CRTM in that the resin is not injected into a cavity formed by a gap between the confined preform and male mold prior to the compression molding step; rather, resin is poured on top of the dry reinforcement preform prior to bringing the two tool halves together during compression molding. Thus, the tooling design is simpler, and less resin waste is possible for WCM, since a resin injection system is not needed. The primary limitation of this process is that it works best for relatively flat or gently curved parts where the deposited resin does not migrate too much [4].

Prior research has reported on RTM applied to biocomposite materials—for example, hemp and kenaf mat in a polyester matrix [5,6]—but not using WCM or CRTM. In this paper, the authors investigate whether WCM can satisfy the demanding requirements for this transportation industry application involving gently curving parts with a strong sustainability focus. In addition, the paper (1) formalizes a seven-step procedure for designing WCM tooling using simple analytical models, (2) describes the first reported application of WCM to biocomposite materials, and (2) discusses a novel method of preventing resin ingress into mold features.

2. Literature Review

The two relevant technologies covered in this literature review are CRTM and WCM. Reference to CRTM can be traced back to the mid-1990s with the work of Young and Chiu [7]. While the literature on CRTM is extensive, the most highly cited papers have focused on analytical and numerical modeling of the process [8,9,10].

Reference to WCM demonstrations in the literature is relatively recent (since 2016); this is despite claims by the Molded Fiber Glass Company of having used it since the late 1940s [4]. Bockelmann [11] demonstrated a process refinement to WCM called Carrier-Integration Pressing, where resin is stored in small, uniformly distributed wells formed in an elastomeric sheet and positioned beneath the bottom mold and dry reinforcement preform. During pressing, the resin in each well locally impregnates the preform from below and reportedly provides more homogeneous filling and fluid control than conventional WCM. Fels et al. [12] performed an experimental comparison of WCM to High-Pressure Injection RTM (HPiRTM) using the same composite system (carbon/epoxy), layup, part shape (C-channel), and tooling. Although cycle times were not reported, the flexural and tensile properties of the parts from both processes were similar, while shear properties for the WCM part were 15% higher due to a lower void content. Keller et al. [13] computationally and experimentally compared the manufacturing of a benchmark part—specifically, a 20-ply woven carbon/glass preform impregnated with a fast-cure epoxy composite part—by CRTM and two variants of WCM. Their major finding is that when impregnation time is a limited factor, WCM is the better option because of its shorter cycle time. Fial et al. [14] demonstrated process innovation by creating an embroidered carbon fiber preform formed and impregnated with an epoxy resin using WCM processing that involved a rigid, heated lower tool and a rubber-coated upper matching tool. In their review paper, Henning et al. [15] describe two variants of WCM that are used: (1) viscous draping, where the preform is sprayed with resin outside the molds then moved inside prior to forming; and (2) dry draping (pioneered by Huntsman Corporation), where the resin is sprayed on the preform situated inside the bottom tool half. Lee et al. [16] measured the impact properties of biaxial non-crimp fabric preforms impregnated with epoxy and made using viscous draping WCM, and found their mechanical properties to be comparable to conventional autoclaved parts. Recently, Ayatollahi [17] showed that for dry draping WCM of carbon preforms with a fast-cure epoxy, mold temperature, resin temperature, and resin set time had significant effects on the final mechanical properties of the part. She also found no correlation between initial resin distribution during pressing and the final mechanical properties.

Some research groups investigated fluid flow in a WCM tool using interesting flow visualization methods. Muthuvel et al. [18] demonstrated an experimental method to visualize resin flow and measure the in-mold pressure profile at discrete flow front locations for WCM by compressing mineral oil (resin rheological analog) between flat transparent plates. This is similar to the method that Chang et al. [19] used for investigating vacuum-assisted CRTM. The Muthuvel group also performed WCM process parameter studies on a 10-ply carbon UD preform using this apparatus [20]. Poppe et al. [21,22] used experimentally determined Darcy’s law permeability values (woven carbon fabric preform impregnated with silicone oil) to demonstrate a 3D modeling approach for viscous draping WCM.

At least three manufacturers have reported either using or investigating the use of WCM for commercial applications [4,23,24]. For example, Bergmann et al. [25,26] describe the data analysis of the process used to make several carbon/epoxy frame parts for BMW’s 7 Series cars in an automated manufacturing cell.

3. Materials and Methods

3.1. WCM Process

As mentioned previously, the WCM is investigated for transportation application. The steps in the WCM process (see Figure 1) are as follows.

• Pour a mixed-liquid thermoset resin on top of a dry composite laminate preform and lay it within a temperature-controlled female mold (viscous draping), or pour resin on top of the dry preform inside the mold (dry draping). The female mold typically requires a venting arrangement to draw vacuum.
• Use a matching temperature-controlled male mold with a resin sealing system mounted on a high-tonnage press to rapidly push the resin through the dry preform’s thickness.
• Cure the impregnated laminate in place at an elevated temperature.
• Eject the consolidated and cured composite part.

3.2. Bio-Based Materials

The industrial collaborator was interested in having three natural fiber reinforcements and a resin with a high bio-content investigated for the WCM study. The reinforcements include the following: (a) hemp fiber yarn in a plain weave made by Hem Mills, Inc. in Concord, NC, USA [27]; (b) EKOA flax fiber yarn in a 2 × 2 twill weave fabric made by Lingrove, Inc. in San Francisco, CA, USA [28]; and (c) BioMid continuous filaments made of highly crystalline cellulose [29] in a plain weave fabric by Absecon Mills, Inc. in Cologne, NJ, USA [30]. These materials will be referred to as hemp, flax, and BioMid, respectively. The matrix is SR GreenPoxy 33 resin (35% bio-content) mixed with SD 4775 hardener in a 100/27 weight ratio as prescribed by the manufacturer, Sicomin Epoxy Systems (Châteauneuf-les-Martigues, France) [31].

3.3. Process and Tooling Design

For a given reinforcement, ply layup, part shape, and resin matrix, the general WCM process and tooling design steps followed are as given below:

• Choose the resin catalyst and curing temperature to achieve the desired cure time.
• Measure the resin viscosity or obtain it from the manufacturer’s product data sheets.
• Experimentally determine the reinforcement/resin permeability.
• Estimate resin impregnation time, t_imp, for a particular molding pressure and resin viscosity.
• Pick the specific WCM approach—viscous draping or dry draping—and estimate the minimum amount of resin required per part.
• Estimate the total cycle time, t_c.
• Design tooling to include resin sealing features and strategically located vacuum ports and part ejection features protected from resin ingress by semi-permeable membranes.

Details discussed in the following sub-sections include the following: (a) how these steps are applied to process and tooling design for this investigation, (b) descriptions of novel tooling features, and (c) simple analytical models to aid in tooling and process design.

Square 127 × 127 mm laminate specimens consisting of the three reinforcements impregnated with the bioresin were chosen for testing. Since one of the manufacturing requirements is a target laminate thicknesses between 3 and 4 mm, the dry preform layups for the three textile reinforcements considered are [0/90₄] for hemp, [0/90₅] for flax, and [0/90₁₈] for BioMid.

• Steps 1 and 2—Resin choice and viscosity

The cure time for GreenPoxy 33 with its fastest hardener (SD 4775) is 30 min at 100 °C. The upper limit for thermoset resin cure temperature is around 200 °C for cellulosic fibers, since above it, either the cellulose or natural gums that bind the fibrils together start to degrade [32]. A snap-cure resin (e.g., 5 min cure time) is required to meet the <10 min manufacturing cycle time; unfortunately, there were no high-bio-content snap-cure resins commercially available at the time the investigation was conducted. Regardless, the viscosity (μ_r) of the GreenPoxy 33 resin at 27 °C was measured using a Brookfield DV-II+ Pro Viscometer for high strain rates associated with WCM and was found to be relatively constant at 1.30 Pa·s.

• Step 3—Permeability Measurement

To measure permeability, Darcy’s law, which mathematically describes a porous medium’s (reinforcement, in this case) opposition to fluid flow (resin), is used. The general form for 1D flow at a flowrate Q (m³/s) is

$$Q=vA={\displaystyle \frac{KAP}{{\mu }_{r}x}}$$

(1)

where v = flow front average velocity (m/s)

A = cross-sectional area of flow (m²)

K = permeability of porous media (m²)

P = pressure drop across fluid flow (Pa)

x = distance flow front has travelled (m).

Rearranging Equation (1) yields

$$v={\displaystyle \frac{Q}{A}}={\displaystyle \frac{KP}{{\mu }_{r}x}}$$

(2)

If we use the differential form of velocity, i.e., v = dx/dt, assume a constant cross-sectional area A for fluid flow, and integrate both sides, we obtain an expression for (a) estimating fill times for very simple resin infusion mold geometries (i.e., a rectangular prism) and (b) deriving K based on experimental observations of a fluid that travels a distance x in time t:

$$t={\int }_0^{t}dt={\int }_0^x{\displaystyle \frac{{\mu }_{r}x}{KP}}dx={\displaystyle \frac{{\mu }_r{x}^2}{2KP}}$$

(3)

The permeability for any particular reinforcement can be experimentally determined for a particular fluid viscosity and constant pressure drop by isolating K in Equation (3) and measuring flow front position vs. time:

$$K=\left({\displaystyle \frac{{\mu }_r}{2P}}\right)\left({\displaystyle \frac{x^2}{t}}\right)$$

(4)

To determine permeability values for use in process analysis, three separate infusion experiments were conducted for each reinforcement—hemp, flax, and cellulose—to measure K in the thickness direction for GreenPoxy 33. The experimental setup, shown in Figure 2 for the BioMid layup, consisted of a flat steel plate to serve as a base, a bottom layer of vacuum bag material, two layers of flow mesh surrounded on all four sides with sealant tape, a 5 × 5 × 2 cm high stack of dry reinforcement plies, and a top layer of vacuum bag material sealed to the bottom layer using the tape. Two holes were cut in the top vacuum bag—one for a tube supplying resin to the flow mesh and another for a tube that applies vacuum to the top of the reinforcement stack. Both the hose openings were sealed using sealing tape. Once vacuum was applied and resin started flowing, the times for the resin flow front to rise to thickness levels of 0.5, 1, 1.5, and 2 cm were recorded (

Table 1. Time vs. position data and measured pressure drop for resin infusion experiments, resin viscosity measurement, and derived permeabilities.

	t (s)
x (m)	BioMid	Hemp	Flax
0.005	125	135	55
0.010	285	285	95
0.015	615	510	140
0.020	975	930	255
P (Pa)	$8.4\times {10}^4$	$8.0\times {10}^4$	$8.2\times {10}^4$
μr @ 27 °C (Pa·s)	1.30
K (m2)	$3.1\times {10}^{-12}$	$3.3\times {10}^{-12}$	$7.9\times {10}^{-12}$

).

Experimental values of permeability are determined from the data in Table 1. If viscosity, μ_r, and vacuum pressure, P, are measured or provided by the manufacturer, then the first collection of terms in Equation (4) is known. It should be noted that resin viscosity behavior in Equations (1)–(4) is assumed to be Newtonian (i.e., constant). In actuality, mixed-thermoset resins exhibit viscosities that are both strain-rate- and temperature-dependent, so complete rheology work-ups may be required for thick and/or complicated parts infused using WCM. By plotting x² vs. t on a graph (Figure 3), the slope of the straight-line curve fit (representing the second collection of terms in Equation (4)) is substituted into the equation to solve for K. The measured vacuum pressure (via a pressure gauge), viscosity, and calculated permeability values are all provided in Table 1. The flax dry preform infused the fastest of the three candidate reinforcements for the same thickness and driving pressure, and thus has the highest permeability.

• Step 4—Resin Impregnation Time

The 1D version of Darcy’s law expressed in Equation (3) is used to provide a 1st-order prediction of the infusion time for each type of reinforcement, which is acceptable for process design purposes if the part length-to-thickness ratio is high. For the viscosity measured in Step 2 and an assumed WCM pressure of P = 3.45 MPa and nominal preform thickness of x = 3.5 mm, the infusion times for BioMid, hemp, and flax preforms are 0.74, 0.69, and 0.29 s, respectively. The flax twill weave is more open than the hemp and BioMid weaves, so a higher permeability and shorter impregnation times are expected. The reader should note that these impregnation times are negligible compared to the 10 min cycle time requirement; thus, resin cure time is the process bottleneck in this case.

• Step 5—WCM Approachand Resin Volume Required

The WCM dry draping approach shown in Figure 1 is chosen for this application, since it eliminates the need to transport a wet preform into a bottom mold. Other applications might favor viscous draping to avoid, for example, resin overspray getting on the mold. An approximation of the minimum amount of resin volume required for each preform, V_r, is given as follows:

$$V_r=A_{t}h\left(1-V_f\right)+w_{gap}{l}_p$$

(5)

where A_t = surface area of a dry ply within the mold (127 mm square with radiused corners in our case), h = laminate thickness, V_f = target fiber volume fraction, w_gap = gap between ply edge and die (e.g., (133 − 127)/2 = 3 mm in our case), and l_p = perimeter distance of the die (e.g., $4\left[\left(133-2\left(10.4\right)\right)+{\displaystyle \raisebox{1ex}{$\pi $}\!\left/ \!\raisebox{-1ex}{$4$}\right.}\left(10.4\right)\right]=481\mathrm{mm}$ in our case). Ideally, w_gap should be zero to avoid wasting resin. Excess resin in the laminate is also undesirable, since it will reduce the laminate’s fiber volume fraction. However, excess resin is oftentimes necessary to ensure full preform impregnation, so a good rule of thumb is to increase V_r by 2–3%.

• Step 6—Total cycle time

Total manufacturing cycle time, t_c, for WCM dry draping is estimated using the following equation:

$$t_c=t_{ins}+t_{rp}+{t_v+t}_{imp}+t_{cure}+t_{ej}$$

(6)

where t_ins = time to insert a preform stack into the bottom mold, t_rp = time to pour resin over the inserted preform, t_v is the time required to pull maximum vacuum, t_imp comes from Step 4, t_cure is specified by the manufacturer, and t_ej = time to eject the cured laminate from the mold. Since t_cure dominates the cycle time in our case, other times were not calculated or estimated for this study.

• Step 7—Tooling design

Male and female WCM molds were designed in CAD and fabricated to infuse and cure 133 × 133 mm flat biocomposite laminates with 10.4 mm radiused corners. Engineering drawings for and the actual machined versions of both molds are shown in Figure 4 and Figure 5, respectively. The male mold consists of a machined body with a step around its bottom edge. A laser-cut sheet metal part is fastened to the main body to form an O-ring groove around the perimeter. A silicone O-ring fit into this groove maintains pressure and prevents resin leakage during impregnation. Important female mold design features include the following: (a) two manual ejector pins for expelling the cured part and (b) a vacuum hole in the center of the mold to remove compressed air during resin infusion. Both the ejector pins and the vacuum hole were covered with a semi-porous peel ply seal to prevent resin ingress. If needed, a specific gap within the mold cavity can be maintained by positioning shims in the flange area between the two mold halves during compression molding.

3.4. Experimental Plan

Laminate panels for each material system are fabricated using the WCM tooling placed within a Carver Laboratory Thermal Press (Wabash, IN, USA), as shown in Figure 6. First, dry preform plies are cut slightly smaller (127 × 127 mm square) than the mold cavity using a custom steel rule die (Apple Die, Milwaukee, WI, USA) and a 25-ton clicker press (Cutting Atom Model MF9.4, Vigevano, Italy). Liquid release agent is thoroughly applied to both the mold halves to avoid resin attachment and build up, and then peel ply seals are placed over the center vacuum hole and ejection pins. The dry preforms are placed in the female mold, and the appropriate volume of mixed resin is poured from a beaker on top. The male mold is inserted into the lower mold, and the entire assembly is transferred to a thermal press with platen temperatures adjusted to maintain 100 °C mold temperatures. After the hose from a vacuum pump is attached to the female mold, 0.8 bar vacuum is applied to the tooling for 1 min to extract as much air as possible from the mold cavity prior to infusion. The authors observed that the gauge on the vacuum pump during evacuation regularly reached a maximum of 0.8 bar (limit of the vacuum pump used) in 1 min or less for all the preforms. A clamping load that provides a molding pressure of 3.45 MPa (500 psi) is applied to the pair of molds to infuse resin into the dry perform and consolidate the part. Finally, pressure and temperature are maintained for 30 min to fully cure the part mold, the part is ejected using the ejector pins, and then the peel ply pieces protecting the vacuum hole and ejection pins are removed from the bottom of the part.

Characterization of the panels includes thickness, fiber volume fraction, and flexural properties. Details of the equipment and methodologies used are discussed below.

3.4.1. Thickness Measurements

The thickness of each panel was measured at its middle and all four corners using a Mitutoyo 547-520 thickness gauge (Mitutoyo, Kawasaki-shi, Japan).

3.4.2. Fiber Volume Fraction and Theoretical Permeability

Fiber volume fraction of a biocomposite is a critical material parameter for understanding the role of reinforcement and resin in overall mechanical properties. Unfortunately, using standard digestion, ignition, or carbonization per ASTM D2584 [33] and ASTM D3171, Test Method I [34] is impossible in our case, since the cellulosic fibers undergo complete decomposition along with the resin in all of these methods [35]. Hence, the V_f of a flat biocomposite laminate can be estimated using ASTM D3171, Test Method II [34] based on geometry and properties of the dry reinforcement plies and fiber alone, and by assuming that the void volume fraction V_v = 0. Specifically, V_f is found by dividing the fiber volume by the total volume to obtain the following:

$$V_f={\displaystyle \frac{\raisebox{1ex}{$n{\rho }_{ar}A$}\!\left/ \!\raisebox{-1ex}{${\rho }_f$}\right.}{hA}}={\displaystyle \frac{n{\rho }_{ar}}{h{\rho }_f}},$$

(7)

where n = number of plies in the laminate preform, ρ_ar = fabric mass/fabric area = areal density of the reinforcement fabric, and A = laminate area, ρ_f = density of the solid fiber, and h = laminate thickness. Fiber Volume fraction measurements were performed using a Mettler Toledo ML304 Analytical Balance with a 320 g capacity and a 0.1 mg readability, and length measurements were all performed using Mitutoyo 500-196-30CAL Digimatic Calipers with a 150 mm capacity and 0.005 mm resolution.

Principle permeability values as a function of V_f are predicted along the fiber direction ($K_{\parallel }$) and perpendicular to the fiber direction ($K_{\perp }$) using Gebart’s permeability model [36]:

$$K_{\left|\right|}={\displaystyle \frac{{8R}^2}{c}}{\displaystyle \frac{{\left({1-V}_f\right)}^3}{V_f^2}}$$

(8)

$$K_{\perp }=C_1{R}^2{\left(\sqrt{{\displaystyle \frac{V_{fmax}}{V_f}}}-1\right)}^{{\displaystyle \raisebox{1ex}{$5$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}$$

(9)

where R is the fiber radius. The parameters c, C₁ and V_fmax are dependent on the fiber arrangement (square or hexagonal) and are independent of the fiber radius or fiber volume fraction. The numerical model parameters and the equivalent Kozeny constant (corresponding to c) for flow along and perpendicular to the fibers are summarized in

Table 2. Numerical values of the parameters in Equations (8) and (9) [34].

Fiber Arrangement	c	C1	Vfmax
Square	57	${\displaystyle \frac{16}{9\pi \sqrt{2}}}$	${\displaystyle \frac{\pi }{4}}$
Hexagonal	53	${\displaystyle \frac{16}{9\pi \sqrt{6}}}$	${\displaystyle \frac{\pi }{2\sqrt{3}}}$

for two types of fiber arrangements.

3.4.3. Density and Flexural Property Testing Measurements

Flexural properties for all three composite material systems were determined experimentally. All specimens were prepared for evaluation in three-point bending using ASTM D790 [37], and the densities of the materials were determined using ASTM D2395 [38]. Five bending samples and three density samples were abrasive-waterjet cut from each panel for a total of 25 bending samples and 15 density samples for each reinforcement material. The densities of the test specimens and the dimensions of the bending samples were determined using procedures found in ASTM D2395 [38] and ASTM D790 [37], respectively. Biocomposite density can also be predicted using the rule of mixtures given as follows:

$$\rho =V_f{\rho }_f+\left(1-V_f\right){\rho }_m$$

(10)

where ρ_m = the matrix (resin) density.

All testing was conducted on an electromechanical load frame with the three-point bending fixture, shown in Figure 7, installed between the load cell and crosshead. Specifically, the ASTM D790 [37] three-point bending tests were conducted on a 25 kN capacity electro-mechanical fine screw load frame refurbished by Instru-Met Corporation (Union, NJ, USA). A 1 KN Series 11 load cell (SR# 1448) with 0.05 N resolution from Instru-Met Corporaton was installed on the load frame for all tests. The three-point test fixture used was a Model WTF-FL (SN#: CU-FL-40) manufactured by Wyoming Test Fixtures (Salt Lake City, UT, USA). A Model 3540-0500-ST displacement extensometer manufactured by Epsilon Technology Corporation (Jackson, WY, USA) with 12.7 mm of displacement capacity and 0.01 mm resolution was used to record mid-span deflection in all tests. The span between supports was calculated to be 61 mm, and the actual length of the specimens was 150 mm. A 1.1 kN load cell was used for all experiments, since this was determined to provide the best load resolution for the samples being tested. The crosshead extension rate was calculated using ASTM D790 [37] to be 1.5 mm/min based on the specimen thickness. Load and deflection using an extensometer were recorded for each specimen at a rate of 10 samples/s.

4. Results and Discussion

Five laminate panels were fabricated for each type of reinforcement. No resin ingress was observed in the vacuum hole or around the ejector pins when the semi-porous peel ply seals were used, and they were easily removed from the part after curing and ejection. Laminate panel examples are shown in Figure 8. The results and discussion for each physical or mechanical property measured are discussed in the following subsections.

4.1. Thickness Measurements

Average thickness and standard deviation for each material system are provided in

Table 3. Thickness measurements for biocomposite specimens.

Metric	BioMid	Hemp	Flax
Avg. Thickness, h (mm)	3.84	2.87	3.01
Standard Deviation	0.06	0.03	0.12

. The measured thicknesses varied from 2.87 mm (hemp) to 3.85 (BioMid), although these values depend on the number of plies used. Of particular interest is the variability as expressed by the standard deviation. Flax had the largest value at 0.12 mm, which is attributable to the course twill weave and difficulty of adjacent plies to nest together in a repeatable manner. The plain weave hemp had the smallest value at 0.03 mm.

4.2. Fiber Volume Fraction and Theoretical Permeability

Theoretical fiber volume fractions and theoretical permeability for each type of compression-molded laminate based on Equation (7) are provided in

Table 4. Fiber volume fractions and predicted permeabilities for each biocomposite system.

Variable (Units)	BioMid	Hemp	Flax	Notes
n (# plies)	18	4	5
ρar (kg/m3)	0.203	0.814	0.366	See Ref. [2]
ρf (kg/m3)	1500 [29]	1500 [39]	1400 [40]
h (m)	0.00384	0.00287	0.00301	From Table 2
Vf (%)	63.4	75.6	43.4
Technical Fiber Diameter Value/Range, 2R (μm)	11 [29]	106–142 [41]	35–150 [42]
K⊥ (m2)	$1.2\times {10}^{-13}$	$2.5\times {10}^{-12}$	$6.5\times {10}^{-11}$
K (m2)	$3.1\times {10}^{-12}$	$3.3\times {10}^{-12}$	$7.9\times {10}^{-12}$	From Table 1

. Flax has the lowest V_f at 43%, while hemp and BioMid are significantly higher at 76% and 63%, respectively. Although Gebart’s equation for resin flow perpendicular to the fiber direction is intended for unidirectional fiber layups, we can use it to predict permeability for woven textiles for comparison with the experimentally measured values in Table 1. Assuming a hexagonal fiber arrangement and the middle of the fiber range for hemp (124 μm) and flax (92.5 μm) for data obtained from the listed references, the predicted permeabilities are shown in Table 4. A comparison of the permeabilities provides mixed results—the BioMid value is over one order of magnitude lower, the flax is off by a factor of 0.75, and the flax value is an order of magnitude higher.

4.3. Density and Flexural Property Testing Measurements

The average (μ) and standard deviation (σ) for the dimensions, mass, and density of the 15 samples for each material system are summarized in

Table 5. Specimen dimensions and mass used to determine the density for the three material systems.

		Dim 1 (mm)	Dim 2 (mm)	Thickness (mm)	Mass (g)	Measured Density (kg/m3)	Calculated Density (kg/m3)
BioMid/Epoxy	μ	25.60	12.99	3.81	1.657	1309	1360
	σ	0.05	0.05	0.12	0.042	14	n/a
Hemp/Epoxy	μ	25.44	12.84	2.84	1.184	1276	1410
	σ	0.08	0.03	0.05	0.013	29	n/a
Flax/Epoxy	μ	25.48	12.84	2.94	1.230	1278	1285
	σ	0.10	0.02	0.15	0.060	18	n/a

. Using V_f and ρ_f from Table 4 and ρ_m = 1120 kg/m³ from Ref. [31] in Equation (10), the calculated density for each material system is provided in the last column of Table 5. When comparing the measured and calculated densities, there is reasonable agreement for BioMid/epoxy, poor agreement for hemp/epoxy, and excellent agreement for flax epoxy. The high discrepancy with hemp/epoxy is probably attributable to the inherent variation in reported hemp properties in the literature.

The average (μ) and standard deviation (σ) for the bending sample dimensions are summarized in

Table 6. Three-point bending specimen dimensions for the three material systems being evaluated.

		Span (mm)	Width (mm)	Thickness (mm)
BioMid/Epoxy	μ	61.0	12.90	3.76
	σ	0.0	0.05	0.11
Hemp/Epoxy	μ	61.0	12.81	2.81
	σ	0.0	0.05	0.10
Flax/Epoxy	μ	61.0	12.799	2.887
	σ	0.0	0.049	0.133

. The order of bend testing was determined using a randomized complete block design scheme. Using the calculations recommended by ASTM D790 [37], the load and deflection data were converted into flexural stress and flexural strain values. The flexural stress vs. flexural strain curves for the twenty-five samples of each material system are found in Figure 8. The data from the flexural stress versus flexural strain curves shown in Figure 9 were used to calculate the maximum outer fiber stress and flexural modulus for each sample. These values were then divided by the sample density to determine the specific outer fiber stress at failure and the specific flexural modulus for each sample. These calculations are summarized in Figure 10 in the form of box–whisker plots. The p-values were computed to compare the specific moduli and specific outer fiber stress at failure for each combination of fiber types—BioMid–flax, BioMid–hemp, and flax–hemp. The statistic used was the difference in the means using the variances calculated by the data. The resulting p-values were all much lower than 0.05, leading to the conclusion that all the results were significantly different.

Several interesting observations can be made related to the material property comparison. Similarities in measured density from Table 5 are expected, since all three reinforcements are cellulosic. Furthermore, GreenPoxy 33 density is reported by the manufacturer as 1120 kg/m³, so the measured densities being in between the matrix and fiber densities are consistent with the ‘rule of mixtures,’ although the hemp/epoxy discrepancy between measured and calculated values is relatively high. Specific flexural modulus and outer fiber failure stress are similar for both BioMid and flax, whereas hemp values are essentially half as much. The authors believe that this inconsistency is attributable to differences in surface treatment. For both the BioMid and flax textiles woven by Absecon Mills and Lingrove, respectively, fiber surface preparation is intended for bonding with common composite thermoset resins such as epoxy. The hemp fiber from Hem Mills is intended for home décor applications, so no fiber surface preparation other than chemical degumming was performed.

5. Conclusions and Future Work

Wet Compression Molding, a relatively obscure advanced composite process involving through-thickness resin impregnation, is demonstrated for a transportation industry application of biocomposites (i.e., EV bus seats and backs) requiring short manufacturing cycle times and low waste. Previous manufacturing attempts using compression molding of a wet layup and VARTM were not deemed economically viable. The design of the WCM dry draping process and related tooling are demonstrated through a comparative investigation involving three different natural fibers (hemp, flax, BioMid) in woven textile form infused with a high-bio-content epoxy resin. Darcy’s law predicts that complete impregnation of the 3–4 mm thick, dry composite preforms with 3.45 MPa applied pressure is generally less than 2 s for all three material systems, using experimentally obtained permeability values. However, achieving a <10 min cycle time will require a snap-cure matrix, which was not available in a high-bio-content thermoset resin at the time of the investigation, but is available in a synthetic thermoset resin. Unlike traditional RTM methods that rely on in-plane impregnation, the process bottleneck for WCM is now the resin curing time. A relationship for estimating resin volume required for dry draping is provided and has been used for subsequent experimental runs. Critical and novel WCM tooling features validated by a series of successful preform impregnation and curing experiments include the following:

• O-ring seal incorporated into the top mold to maintain the high impregnation pressure and prevent resin leakage;
• Strategically located vacuum ports covered with a semi-permeable membrane (e.g., peel ply) in the bottom mold to remove trapped air and prevent resin ingress during processing;
• Ejection pins in the bottom mold, also covered with pieces of semi-permeable membrane, to quickly eject the part after curing.

Laminate panels made using all three material systems are characterized by measurements of thickness, fiber volume fraction, and flexural properties. High V_f values are estimated for hemp and BioMid, but the value for flax is quite low due to the poor nesting of adjacent twill weave plies. Thickness is shown to have relatively low variability for all composite panels (all 1σ < 4%). The specific flexural modulus and maximum stress of BioMid and flax are similar, but the values for hemp are about half as much. This inconsistency is attributable to the BioMid and flax textiles being processed for composite applications, whereas the hemp fibers are intended for home décor products.

Future research work planned for this project includes further investigation into the poor mechanical property performance of hemp shown in Figure 10, measuring the void volume fraction of the biocomposite parts, and the design of the final WCM molding and process for electric bus seats and backs.