Properties of Composite Trapezoidal Parts Manufactured with help of Filament Winding Technology using Taguchi Method

Properties of Composite Trapezoidal Parts Manufactured with help of Filament Winding Technology using Taguchi Method

PhD. Svetlana Risteska, PhD. Blagoja Samakoski, MSc. Maja Stefanovska

Institute for Advanced Composites and Robotics (IACR) Prilep, Macedonia

Abstract Aim of this work is to determine the influence of fiber orientation and winding tension on the structural and mechanical properties of trapezoidal glass fiber/epoxy resin beams, manufactured with help of filament winding technology. In the first part, with help of Taguchi method was achieved similar wall thickness on trapezoidal sides and corners. Beams with similar thickness where used for mechanical testing, described in the second part in this research. Values for stress, strain and module of elasticity were calculated using three-point bending test carried out on Shimadzu universal machine. Further, the paper investigates the content of voids inside the structure using optical and scanning electron microscope.

Keywordscomposite, filament winding, taguchi method, trapezoidal beam.

1. INTRODUCTION

   The importance of composite materials in todays society is well known. Due to the high strength-to-weight ratio, good stiffness properties, inherent corrosion resistance, low electromagnetic reflectance, thermal conductivity and weight, their use in different kinds of applications is accelerating rapidly [1]. Composite materials have become common engineering materials designed and manufactured for various applications and industries including automotive components, sporting goods, aerospace parts, consumer goods, marine, oil and infrastructure [2]. Decks for both pedestrian and vehicle bridges across waterways, railways and roadways are now a commercial reality in both North America and Europe, with some pedestrian bridges being built entirely from composites [3], [4]. The bridge deck sections, composed of hexagon and double-trapezoid profiles are bonded with a high-strength adhesive under controlled conditions. Composites can significantly reduce maintenance and replacement costs because of the material's excellent resistance to corrosion and fatigue. In essence, composite durability not only improves life-cycle costs but extends the life-cycle itself [5], [6]. During filament winding (FW) process voids are formed inside the composite structure due to a number of reasons, such as air entrapment during resin system mixture and moisture absorption during material storage or processing. These voids affect the material properties of the composite. Zhu et.al, [7] studied the microstructure of voids and investigated their influence on tensile strength and modulus of carbon/epoxy fabric laminates. They have reported, that the tensile strengths decreases with increased void content

   and cracks emanate in the composite when void content is 8.0% and 9.0% after tensile strength test. However, no cracks were found emanating from the voids in the composite parts with the void content of 0.4%.

   Because of the cost of failure tests of large structures, laboratory tests must be conducted on a much reduced scale. Thus, it is essential to have a correct method to extrapolate the results obtained from small laboratory specimens to much larger structural parts.

   In this paper, study of the influence of fiber orientation on the structural and mechanical properties of trapezoidal glass fiber/epoxy resin beams manufactured with help of filament winding technology is presented. With help of Taguchi method was achieved similar wall thickness on the sides and corners of manufactured trapezoidal beams, while mechanical properties studied include values of stress, strain and modulus of elasticity.

2. MATERIALS AND EQUIPMENT Trapezoidal beams used in this study are composed from

   long E-glass roving bers (P185 1200tex from Owens Corning) impregnated in epoxy resin system Araldite LY564/Aradur 917/Accelator 960 from Huntsman, based on diglycidylether of bisphenol A with an anhydride hardener. Contained volume fraction of glass fiber in composite beams is 55%. Wet winding process was carried out on laboratory filament winding machine MAW FB 6/1 with six aches, roller type resin bath and mechanical creel manufactured from Mikrosam AD. The glass fibers from the creel are passed through the resin bath, causing resin impregnation. The temperature of the resin bath was held constant, 37oC. The trapezoidal shape was fabricated by winding the roving fiberglass onto a rotating collapsible trapezoidal mandrel with pins. Mandrel length was 1200mm and 200mm high. Base at the bottom was 287mm width and base at the top was 172 mm width. It was coated with release agent (QZ13 from Huntsman) that allows easy removal of the part after processing. For smooth inside surface of produced beam and easy removal from the mandrel, firstly was used polypropylene film with 18g/m2 surface weight and 50mm width, applied in few layers and after that was winded Viledon® T1702 with 100mm width from Freudenberg, with 50% overlap. Each trapezoidal beam was manufactured with 11 winding layers with already specified winding design and

   10 bobbins. At the end, produced beam was cured with help of industrial heater at 100oC for 5 hours.

3. EXPERIMENTS AND RESULTS Improvement of beams characteristics was done in two

   parts. In the first part was improved the difference in wall and corners thickness (t) and in the second part was enhanced the bending strength of the beam.

   1. Influence of parameters on the thickness of the corners and walls of trapezoidal beam (Taguchi Method)

      Manufactured trapezoidal beams had different wall thicknesses on the sides and corners. To investigate the influence of winding tension and fibers orientation on wall thicknesses Taguchi method L4 was applied. Therefore, it was used winding tension in two levels, 120N (Level 1) and 180N (Level 2) and winding design in 3 levels, whereas winding speed was held constant, 30m/min. From 100% winded layers in composite beam in Level 1 46% belongs to winding angles 10o, 36% belongs to winding angles 90o and 18% belongs to winding angles 45o. In Level 2 were utilized 46% of winded layers with winding angles 20o, 18% from 90o and 36% from 45o. Beam winded with winding design in Level 3 had 18% from winding angles 10o, 36% from winding angles 90o and 46% from winding angles 45o. Table 1 shows three factors in three levels used in the experiment. If two levels were assigned to each of these factors and factorial experimental design was employed using each of these values, number of permutations would be 22 which reduce the number of experiments to four. The orthogonal array of L4 type was used and is represented in Table 2.

      TABLE I. LEVELS OF PROCESS PARAMETERS

      Symbol	Factor	Level
      		1	2	3
      A	Winding design	46%100/36% 900/18% 450	46% 200/18% 900/36% 450	18% 100 /36% 900/46% 450/
      B	Tension (N)	120	180	180
      C	Winding speed (m/min)	30	30	30

      TABLE II. TAGUCHI L4 ORTHOGONAL ARRAY

      No	Factors		t
      	A	B	mm
      1	1	1	4.675
      2	1	2	4.633
      3	2	1	4.988
      4	2	2	4.800
      5	3	2	3.435

      From given parameters in table 2 were produced four trapezoidal beams marked as No1, No2, No3 and No4. The wall thickness of the parts was measured on eight points shown on Fig. 1. Differences between the thickness (t) of the walls and corners of each beam are given in table 2. Thicknesses of sample walls were bigger than their corner thicknesses. For sample No1, t was 4.675mm. Similar result 4.633mm for t, were received at sample No2 where same design but bigger tension were used for winding. Difference in wall thickness of the trapezoidal part from 4.988mm was achieved at probe No3. In comparison to this sample, improved results were reached by probe No4 with t from 4.8mm. Here winding design was same as in probe No3, but with bigger tension, 180N. Samples were repeated. The end results according to Taguchi method L4 show that winding design has the biggest influence on the thickness difference between trapezoid walls and corners in almost 6%. Also, winding tension is showing an impact on the difference in wall and corners thickness in 0.8%. Similar issues were reported by Schultz [8] using epoxy pre-impregnated tow T300 12k and Fortal 50k carbon ber for winding geometries with circular and square cross section. Used mandrel had a square cross section that was about 62mm per side. Each corner had radius of 12.7mm.

      Fig. 1. Points of thickness measurement.

      Similar tests but not with FW works Hubert [9]. He had investigated the effect of fiber orientation, bagging conditions, material and tool type on compaction behavior for concave and convex shapes made of carbon-epoxy composites: AS4/3501-6 and AS4/8552. However, for both materials, convex shapes with [90] laminates are thinner at the corner. Almost identical findings are presented by Yang et. al, [10] for study of FRP tubes with square transversal cross section. Wall thickness of special shape which utilizing both circular and square geometry in transversal cross section was measured. In both cases was mentioned difference between sides and corners thickness.

      Very important characteristic for all produced parts is the presence of voids on the walls of each sample (Fig. 2 and Fig. 3). In comparison to them, on the corners there were no kinds of voids visible to the human eye. Void formation was observed in the work of Hubert [9]. More voids were presented at the corner compared to the flat section of the laminate, due to entrapped air during the lay-up, moisture in the material, volatiles released during the cure, difference between resin and resin viscosity.

      To investigate the effect of bigger winding angles presented in more per cents in a sample, was manufactured trapezoidal beam with the third design mentioned in table 1 and table 2 (sample No5-Level 3). Exchange of 10o winding angle with winding angle from 20o in filament winding technique is showing worse results. It is supposed that more voids are created when samples are produced with this angle. This is confirmed by the mechanical testing. Samples with 20o winding angle (No4) presents smaller bending properties even from samples with 45o (sample No5), which theoretically was not expected.

      1. Optical x150 (b) Optical x100

         (c) SEM x400 (d) SEM x3000

         Fig. 2. Optical (a, b) and SEM (c, d) results on voids presence in the sample.

         Reason for this is the percents of voids, which are showing an influence on the mechanical characteristics of the composites [11-13].

         Smaller t in the results are showing samples No2 and No5, where in more percents were used 45o winding angles. Due to the good interaction between the mandrel and winding angle, fewer voids were formed. Winding process had pressed the impregnated resin to the surface on the walls and corners.

   2. Mechanical Testing

      To see if the fibers orientation and winding tension are showing an influence on the mechanical properties of the beams, three point bending tests were carried out. The bigger basis of the trapezoidal composite part was cut into 6 or 7

      thus allowing six to seven reproducibility tests of each sample. Specimens where mounted on a two cylindrical supports with radii of 5mm. Central loading with radii of 5mm was located above the crack (Fig.5). Load-displacement curves were plotted for every sample and afterwards values for stress, strain and module of elasticity were calculated. The average results are shown in Fig. 6. The flexural stress, f in the outer surface of the test specimens occurred at the midpoint. These stresses were determined from the relationship:

      1. sides of samples (b) corners of samples

   Fig. 1. Optical pictures from the sides and corners of the samples.

   rectangular forms according to ISO 14125 with span-to-width

   3FL

   (1)

   ratio of 10 (Fig. 4). Micrometer was used to measure dimensions and thickness of the specimens. Values of the measurements are reported in table 3. The tests were carried out at room temperature on computer controlled universal testing machine from Shimadzu with maximal load 250N at a speed of 20 mm/min to minimize any dynamic effects. Load and displacement were recorded by an automatic data

   f 2bp

   Where, f is the flexural stress in MPa, F is the load in N, L is the support span in mm, b is the width of beam tested in mm, and h is the thickness of beam tested in mm.

   Flexural modulus of elasticity is:

   acquisition system for sample No1 and No3. To see the L3 F

   influence of winding angle 45o on the module of bending and flexural strength band samples from sample No5 were cut. A total of twenty bend specimens were involved in this testing,

   E f 4bp s

   (2)

   Where, Ef is the flexural modulus of elasticity in MPa, s is the difference between beam mid-point deflections in mm, F is the load difference in beam mid-point deflections in N.

   Flexural strain, f of the composite specimens was determinate using the fallowing equation:

   6sh

   (3)

   f L2

   In (3) f is the strain in the outer surface in mm/mm, s is the maximum deection of the center of the beam in mm, h is the thickness of beam tested in mm and L is support span length in mm. Received average results from three-point bending tests are shown in table 4.

   TABLE III. DIMENSIONS OF THREE-POINT BENDING SPECIMENS

   Sample Number	b (mm)	h (mm)	lo (mm)	L (mm)
   1-1	30.8	11.37	400	300
   1-2	30.62	12.1	400	300
   1-3	30.53	12.35	400	300
   1-4	30.68	12.32	400	300
   1-5	30.56	12.78	400	300
   1-6	30.54	12.65	400	300
   1-7	30.65	10.67	400	300
   3-1	30.82	12.17	400	300
   3-2	30.74	13.41	400	300
   3-3	30.7	12.77	400	300
   3-4	30.66	13.62/p>	400	300
   3-5	30.54	13.38	400	300
   3-6	30.61	12.08	400	300
   5-1	30.6	10.91	400	300
   5-2	30.66	11.6	400	300
   5-3	30.69	11.23	400	300
   5-4	30.52	11.76	400	300
   5-5	30.72	11.81	400	300
   5-6	30.63	11.93	400	300
   5-7	30.36	13.38	400	300

   Fig. 2. Geometry of three-point bending specimens.

   Fig. 3 Three-point loading arrangement according to ISO 14125.

   1. Average load-displacement curve for sample No1.

   2. Average load-displacement curve for sample No3.

   3. Average load-displacement curve for sample No5.

   Fig. 4. Average force-displacement curves for conducted test-samples.

   The results above shown that with change of angles and tension can be achieved very good results in t by sample No5. But bending results here are much smaller in comparison to samples No1 and No3 due to the smaller t. It can be seen that sample No1 have shown the best mechanical properties.

   Bumpus [14] had been investigating the mechanical properties of fiber reinforced composite structures manufactured with resin transfer molding technique using DOE. It has been reported values between 144MPa and 274MPa for maximal flexural stress. Further, flexural strain at maximal stress has been estimated between 1.65 and 2.24% and module of elasticity between 7.28 and 20.8GPa. The selection of our design (sample No1) shows better results.

   TABLE IV. AVERAGE RESULTS FROM THREE-POINT BENINDG TESTS

   No	f average	average	Ef average
   	MPa	MPa	GPa
   1	495.26	9.96	24.574
   3	237.97	5.11	14.113
   5	333.11	6.39	16.554

   Composites made of glass fiber/epoxy resin and carbon fiber/epoxy resin made by vacuum bagging technique were investigated in the work of Rathnakar et. al, [15] using three- point flexural test for evaluation of strength and stiffness. There have been presented load vs. displacement graphs, which had shown linear behavior until failure, similar with the graphs in Fig. 6. Results with improve performance of a laminate gives Brooks [16]. In [16] three-point bending tests on Hexcel glass fiber reinforced polymer (GFRP) samples of rectangular cross section with 16:1 support span-to-depth ratio have been done. Maximal stress reported had value between 475.67MPa and 528.66MPa. GFRP laminates exhibit progressive failure consisting of fiber failure, debonding and delamination. In this paper isnt discussed or

   calculated the content of voids. However, voids with bigger dimensions can be observed on some places (optical and SEM image) who will cause reduction of laminates mechanical characteristics.

   Mathematical model of the influence of the parameters on the content of voids in trapezoidal beams will be present in future.

4. CONCLUSION

The experimental results confirmed that the best winding design for manufacturing trapezoidal beams by FW technology have shown sample No1 with large flexure stress. These results have confirmed the theoretical calculations. If bigger winding angles are used, less bending strain will be initiated and vice versa. Using these angles smaller t will be reached with less voids. The purpose of this paper is to choose proper combination of winding angles for production of laminate that will have high flexure stress. Since the fact that main cause for occurrence of voids in the laminate of the final product is not a novel issue, the idea of this paper is to initiate the use of appropriate design for production of trapezoidal beam for bridges with FW technology. Validity of this study lies in high efficiency of fabrication of trapezoid by FW technology, better quality of the composite trapezoid final products and competitive advantage for the producers and beneficiaries of these products.

REFERENCES

1. Lukáová M, Analysis of technical and structural properties of composite materials, Transfer inovácií 25/2013.

2. Sanjay K. Mazumdar, Composites Manufacturing Materials, Product, and Process Engineering, CRC Press LLC, USA 2002.

3. Gorrochategui I, Manteca C et. al., Composite material pedestrian bridge for the Port of Bilbao, International Conference on Structural Nano Composites (NANOSTRUC 2012), 2012.

Fiber SPAN Decking for Pedestrian Bridges, Composite advantage, Spring 2013 Newsletter, 2013.

4. Hastak, M, Mirmiran A et. al., A framework for life-cycle cost assessment of composites in construction, Journal of reinforced plastic and composites, Vol. 22, No. 15/2003.

5. Li M, Ranade R et. al., On improving the infrastructure service life using ECC to mitigate rebar corrosion, edited by K. van Breugel, G. Ye, and Y. Yuan, 2010, pp. 773-782.

6. Zhu H, Wu B et al., Influence of voids on the tensile performance of carbon/epoxy fabric laminates, Journal of Material Science and Technology, 2011, 27(1), 69-73.

7. Schultz M, Energy absorption capacity of graphite-epoxy composite tubes, Master thesis, 1998.

8. Hubert P, Aspects of flow and compaction of laminated composite shapes during cure, PhD Thesis, Department of Metals and Materials Engineering, The University of British Columbia, 1996.

9. Yang Y, Hamada H, A study on design of fiber-reinforced plastic (FRP) tubes as energy absorption element in vehicles, Energy Technology and Management, Prof. Tauseef Aized (Ed.), ISBN: 978- 953-307-742-0, InTech, 2011.

10. Saidpour H, Tehranchi H, Determination of void distribution and fiber volume fraction in filament winding composites, Research open access repository, University of East London, 2001.

11. Kukliski H, Influence of voids and layers number on mechanical properties of hand lay-up bended laminates, Journal of Polish CIMAC, ISSN 1231-3998, Vol. 6, No 3, page 69-74, 2011.

12. Liu L, Zhang B et. al., Effects of cure pressure of Induced voids on the mechanical strength of carbon/epoxy laminates, Journal of materials science and technology, Vol. 21, No 1, 2005.

13. Bumpus SRJ, Experimental setup and testing of fiber reinforce composite structures, Master Thesis, Department of Mechanical Engineering, University of Victoria, 2005.

14. Rathnakar G, Shivanan HK, Experimental evaluation of strength and stiffness of fiber reinforced composites under flexural load,

    International Journal of Engineering and Innovative Technology (IJEIT), Volume 2, Issue 7, January 2013

15. Brooks E, Bend testing of composite materials, poster presentation, Materials Engineering, Cal Poly San Luis Obispo, Junior Year, Spring Quarter.