3.1 Thickness distribution
The laminate parts after cut and polish were showed in Figure 6 where parts in concave mold are shown in the lower surface and in convex mold in upper surface. The average thickness of S1 for different mold case was shown in Figure 7(a). It was divided into two main color, the blue one (30º angle part) and orange one (60º angle part). The dark color was used for concave molded part and light color for convex molded part. Obviously, the thickness of Ac area has significantly variation than other areas. As the mold angle raised, the thickness of Ac area in concave mold were greater than plate thickness and increased as mold angle raised. On the other hand, the thickness of Ac area in convex molds were lower than plate thickness and decreased as mold angle raised. This phenomenon verified the previous research, that the corners of the concave mold had a lower consolidation pressure, and the opposite for the convex mold.
Similar results were found in stacking thickness of S2 and S3 as showed in Fig. 7(b) and 7(c). The thickness of Ac area was larger than that of the plate in all concave molds, but smaller than that of the plate in all convex molds. Again, the thickness variations of 60° mold were larger than those in 30° mold for both concave and convex mold. It can be concluded that the thickness variation of the Ac areas increases with the angle of the mold. However, the plate thickness was about the same for molds of different angles, implying that the flat area had higher dimensional stability.
3.2 CoV analysis
Figure 8 showed the coefficient of variation (CoV) for thickness of laminates made in 30\(^\circ\) and 60\(^\circ\) concave or convex molds. The CoV value defined in Eq. (1) quantifies the thickness uniformity at the area A1 + Ac + A2, named as CoV_angle (blue bar), and the area Plate named as CoV_plate (orange bar). Obviously, laminate made in a convex mold has a lower CoV value than in concave mold for all sets, meaning that the part molded in the convex mold will have higher uniformity regardless of the stacking thickness. Moreover, the CoV value has a significant downward trend as the thickness of the laminate increases in both the concave and convex molds. On the other hand, the CoV value of plate were quite low, below 0.0065 in all cases. In all plate areas, the CoV values decreases with the increase of the laminate thickness. The CoV value in 60\(^\circ\) concave mold is higher than that in 30\(^\circ\) concave mold, which indicates that the thickness uniformity decreases with the increase of part angle for the case of the concave mold. However, there is no obvious trend for the part in the convex mold, which means that the mold angle of the convex mold had no direct effect on the thickness uniformity. Besides, the uniformity of the part manufactured by the convex mold are less affected by changes of the part angle.
In this study, the case of 60\(^\circ\)_concave_set_1 exhibited the highest variation in thickness with the CoV value of 0.0583. Compare to the previous study by Stella et al. [23] with the same stacking layers, the reported CoV value was 0.075 ~ 0.12 in their study. It could be concluded that the method of interleaved stacking and bagging arrangement in this study could effectively drain out the internal air of the laminate, and significantly improved the uniformity of the angle part.
3.3 CTD (corner thickness deviation)
The pressure was not evenly distributed in the corner area and also different in concave and convex molds. The pressure applied to corner area of part in the convex mold is larger than the plate area. On the contrary, the pressure applied to the corner area in the concave mold is smaller than the plate area. Corner thickness deviation (CTD) can clearly quantify the thickness change in the corner area. The CTD value is defined as:
CTD =\(\frac{{Corner}_{avg}-{total}_{avg}}{{total}_{avg}}\times 100\%\)
|
(2)
|
where \({Corner}_{avg}\) is the average thickness at the corner region and \({total}_{avg}\) is the average thickness of all regions.
The CTDs of angle parts with different thickness in different mold types and mold angles are shown in Fig. 9. In case of 30\(^\circ\) mold, a larger corner thickness change was noticed for thin parts (S1) for both concave and convex molds. In the figure, the CTD value is positive for the concave mold due to the thickening at the corner area, while it is negative for convex mold for the thinning at the corner area. The CTD value decreases with the laminate thickness in 30\(^\circ\) mold, which indicate the corner deformation reduced as laminate thickness increased. The same trend was observed in 60\(^\circ\) mold, as compared with the 30\(^\circ\) mold, the parts by 60\(^\circ\) concave and convex mold had higher CTD values, the similar result was observed in CoV analysis. Thus, it can be inferred that the relationship with CoV and CTD graph was proportional.
3.4 AD (angle deviation)
The angle deviation was affected by the residual stress caused by pressure and thermal stresses during the consolidation as well as the cutting force. The final angle of the angle part will be slightly deviated from the mold angle. The spring-back angle of the angle part was measured by a digital goniometer (Level Box, in precision of 0.1\(^\circ\)) after cutting. The detailed data of angle measurement is presented in Tables A1-A4. The percentage of angle deviation was defined as:
\({\theta }_{dev.}\) = \(\frac{\varDelta \theta }{\theta }\times 100\%\)
|
(3)
|
From the engineer’s intuition for metal materials, thin parts were more prone to angle changes after processing. The same result is expected for composite materials. Although the angle change in this experiment is very small, we can still find some significant trends. Figure 10 presented the angle deviation upon different stacking thickness in 30\(^\circ\)mold or 60\(^\circ\)mold. The set of S1 in concave mold has the largest angle deformation. As the thickness of the laminate increases, the angle change becomes lower in both concave and convex molds. The AD in the concave mold was lower than that in the convex mold, but both decreased monotonically with the increase of the thickness. Different from previous results, the increase of the mold angle did not aggravate the angle spring back. The angle deviation in 60\(^\circ\)mold was shown in Fig. 10, as compared with the case for 30\(^\circ\)mold, the angle deviation is slightly less. A similar result was found as metal materials, the lower angle caused greater angle deviation.
For both molds in 30\(^\circ\)and 60\(^\circ\), the angle spring back did not exceed 1.3%, which demonstrated good angle stability of the laminate in current study. The previous studies by Hörberg et al. [24], in the case of laminate thickness of 1 mm, reported the spring back angle being less than 2%, and the will decrease as the thickness of the laminate increases. Our results show unanimous conclusion that the selection of process parameters and materials has no direct influence on the angle change.
3.5 Effect of caul plate
For consolidation with caul plates, they were placed on top of the laminate at the corner areas of the part as shown in Fig. 3. We used two different caul plates on S2 for both concave and convex molds. As previous discussion, we quantified the quality of angle part by CoV, CTD, and AD values to investigate the effect of the caul plate. Figure 11 showed the CoV values of S2 with caul plate for consolidation. It can be observed that the Pb caul plate reduces the CoV value in most of cases. Therefore, the Pb caul plate successfully improves the pressure distribution on angle part. On the contrary, the Si-rubber caul plate does not have obvious effect on CoV value in all cases.
In order to demonstrate the influence of the caul plate/intensifier on the corner thickness more clearly, Fig. 12 presented the CTD of the parts using caul plates in concave and convex molds. The blue dot was the values for layup laminate without using caul plates. Notice that the CTD value is better as it close to zero. For the concave mold in upper part of Fig. 12, the lead caul plate of CTD had a slightly improvement of 3.2% in 60\(^\circ\) mold, but having a 19% improvement in 30\(^\circ\) mold. On the other hand, the Si-rubber caul plate is not helpful in suppressing corner thickness variations for both 30° and 60° concave molds.
Use the same method to evaluate the function of the caul plate in the convex mold, as shown in lower part of Fig. 12. It shows the similar results to concave mold. The lead caul plate could improve CTD values by 19.6% and 15.2% in 30\(^\circ\) and 60\(^\circ\) convex molds, respectively. Still, the Si-rubber caul plate cannot restrain the thickness variation in corner area in convex mold. For the effect of caul plate on corner thickness reported in previous researches [19, 21], it was reported that the caul plate of rubber did not contribute to the corner thickness change, which was consistent with our result. In conclusion, it can be inferred from the above experiments that the lead caul plate apparently has the effect of suppressing the corner thickness change in all cases. Especially in the convex mold, the lead caul plate can provide up to 19.6% improvement for corner thickness deviation.
For angle deviation (AD), both caul plates of Pb and Si-rubber presented excellent effect for enhancing angle stability. Again, the S2 laminate was chosen to demonstrate the effect of caul plate, on AD of the angle parts. Figure 13 shows AD values for all S2 laminates. In case of concave mold (upper part of Fig. 13), the caul plate of lead and Si-rubber showed lower AD values, having significantly improvement of 25% as compared with the original part (without caul plate). For the convex case (lower part of Fig. 13), greater improvement was also obtained for angle deviation. The caul plate of Si-rubber and lead successfully decreased the AD value from − 1.17% to -0.17% and − 0.78%, respectively, which showed more than 33% improvement at most.
3.6 Microstructure of angle parts
A digital microscope (Yuan Li Instrument Co., LTD) was used to take the micrographs of the cross section of the angle part. We proposed two scoring mechanisms as shown in Table 3 and Table 4 for the void size and amount to quantify the quality of various angle parts. The score of voids size was ranked on a scale of 1 to 5, with 1 corresponding to obvious void and the size was over 0.7 mm, and 5 corresponding to smaller void and the size was under 0.1 mm. Likewise, the score of void amounts was ranked on a scale of 1 to5, with 1 corresponding to significant porosity that is greater than 20 and 5 with a void less than 1. The score of void size and amount increased for parts with less and smaller voids. This scoring mechanism was evaluated for the part on the five different areas (plate1, A1, Ac, A2, and plate2) defined previously, and the final score was a sum of the scores for the five areas. Therefore, in this dual-scoring mechanism that quantities the void content of the angle part, the maximum score is 50.
Table 3
Scoring mechanism of void size
Magnitude
|
Description
|
Score
|
1.
|
over 0.7mm
|
1
|
2.
|
0.7mm ~ 0.5mm
|
2
|
3.
|
0.5mm ~ 0.3mm
|
3
|
4.
|
0.3mm ~ 0.1mm
|
4
|
5.
|
under 0.1 mm
|
5
|
Table 4
Scoring mechanism of void amount
Amount
|
Description
|
Score
|
1.
|
over 20
|
1
|
2.
|
19 ~ 12
|
2
|
3.
|
11 ~ 6
|
3
|
4.
|
5 ~ 1
|
4
|
5.
|
under 1
|
5
|
Figure 14 reported the scores of angle parts by 30 ° and 60 ° molds, individually. For both 30 ° and 60 ° cases, the scores of parts by convex molds were higher than that by concave molds in all cases, which showed the convex mold would result in parts with lower porosity than concave mold. In addition, it can be observed that the scores increased as thickness increased for both concave and convex molds. Overall, the convex molds presented higher scores due to the greater consolidation pressure during processing, resulting in lower a void content at the corner region. Figure 15 showed the cross sections of angle parts with a high score or a low score by case of 60°_convex_S1 and 60°_concave_S1, respectively. In addition, the caul plate had no significantly effect on reducing void content of angle parts in both 30° and 60° molds.
Void contents could also be estimated by the ImageJ scientific image analysis program. The flange and corner regions were analyzed separately. In all cases, void contents were calculated as the ratio of the total void area over the laminate area. The images of cross section were converted from color RGB to 8-bit binary using an intensity threshold together with some manual adjustments. Figure 16 showed the cross sections of concave and convex parts by ImageJ where blue areas presented the voids, and the yellow line was used to highlight the boundary of the cross section. In this section, we focused on S1 to investigate the void contents under different mold angles. Figure 17 showed the void content of angle parts by concave or convex molds with different angles. It was observed that the part by the convex mold had lower void content in two different angles. The overall void content is within 2.5%.
3.7 The predicting model of corner thickness deviation
From the results of the experiments, it can be seen that the thickness of the part was not uniform in the corner area. In the previous research, the corner thickness deviation predicting models were mainly focused on L-shaped, U-shaped or right-angle parts [28–30]. To further understanding the corner thickness deviation in other angle molds, we proposed a new experimental predicting model of CTD prediction for different mold angles.
Through the experimental data, it is found that the mold radius (\({R}_{m}\)), plate thickness (\({t}_{p}\)) and mold angle (\(\theta\)) were highly correlated with CTD. A semi-empirical model was proposed to satisfy three conditions derived from experimental results. First the corner thickness deviation (CTD) decreased as the radius of mold angle increased, and decreased with plate thickness (\({t}_{p}\)) increased. Second, the corner thickness deviation CTD increased with the mold angle. The semi-empirical model was written as:
\(\text{C}\text{T}\text{D}=\frac{\text{(K+A}\times \stackrel{-}{\text{R}}\text{+B)}}{{\text{R}}_{\text{m}}/{\text{t}}_{{\text{p}}_{0}}}\times {\alpha }\times\) C (\({\theta }\))
|
(16)
|
where, \(K=\frac{(1.5\beta -1)}{{t}_{p}/{t}_{{p}_{0}}}\), \(\beta =\frac{{t}_{i}}{{t}_{p}}=\) \(\frac{{t}_{debulked}}{{t}_{cured}}=1.2\), \(\alpha\)= mold type factor, \({\text{t}}_{{\text{p}}_{0}}=1\), C (\(\theta\)) = mold angle coefficient. After simplified, the CTD predicting model can be written as below:
\(\text{C}\text{T}\text{D}\)= ( \(\frac{0.8+A\times {R}_{m}}{{t}_{p}\times {R}_{m}}+\) \(\frac{\text{B}}{{R}_{m}}\) ) \(\times \alpha\)C(\(\theta\))
|
(17)
|
where the coefficients A and B depend on the corner angle as given in Table 5. The parameters of A and B were derived by fitting the equation to the experimental data.
Table 5
Coefficient of simplified predicting formula
|
\(\alpha\)
|
A
|
B
|
C\(\left(\theta \right)\)
|
Concave, \(\theta\) = 30\(^\circ\)
|
1
|
1.7
|
1.1
|
5
|
Concave, \(\theta\) = 60\(^\circ\)
|
1
|
1.7
|
1.1
|
6.5
|
Convex, \(\theta\) = 30\(^\circ\)
|
-1
|
0.4
|
2.7
|
5
|
Convex, \(\theta\) = 60\(^\circ\)
|
-1
|
0.4
|
2.7
|
6.5
|
Figure 18(a) and (b) showed the comparison result of predicting model with experimental data in concave mold. For case of concave mold, the mold type factor (\(\alpha\)) was equal 1 and the CTD value was positive. The predicting model of concave mold showed reasonable CTD prediction, and the error percentage was under 17.6 in both 30\(^\circ\) and 60\(^\circ\) molds. In addition, for the convex mold, the mold type factor (\(\alpha\)) was equal to -1 and the CTD value was negative. The predicting results for convex mold were shown in Fig. 18 (c) and (d). The CTD predicting error percentage was below 16.1% in both 30\(^\circ\) and 60\(^\circ\) molds.