- Email: [email protected]
UV-cured FRP joint thickness effect on coupled composite pipes
Composite Structures 80 (2007) 290–297 www.elsevier.com/locate/compstruct
UV-cured FRP joint thickness effect on coupled composite pipes Jerry Alan Peck
a,*
, Randy A. Jones b, Su-Seng Pang a, Guoqiang Li a, Brett H. Smith
c
a
c
Mechanical Engineering Department, Louisiana State University, Baton Rouge, LA 70803, United States b EDO Fiber Science, 506 N. Billy Mitchell Road, Salt Lake City, UT 84116, United States NASA Marshall Space Flight Center, Materials, Processes and Manufacturing Department/ED34, Huntsville, AL 35812, United States Available online 16 June 2006
Abstract In this study, 36 FRP composite pipes were joined with various layers of fiber-reinforced UV curing vinylester. Twelve of the composite pipes were joined with an eight-layer weld, 12 with five layers and 12 with three layers; each configuration used exactly one layer of E-glass woven fabric with the remaining layers being comprised of chopped strand mat of various widths. The joined pipes were cured with UV lamps at 80 milli-Watts per square centimeter. The mechanical properties of the cured pipe joints were evaluated by conducting internal pressure testing and simply supported four-point bending testing. The effect of joint thickness on the internal pressure rating, ultimate bending load, and stiffness was evaluated to determine the role of joint thickness in under-curing of UV curing vinylesters. There was a direct correlation observed between joint thickness and internal pressure rating. The five-layer and three-layer joints each performed better than the eight-layer joint in bursting. The mechanism for variation in the residual properties was found to be the degree of cure and was validated by finite element analysis that utilized a composite model that simulates variable cure. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: UV curing; Light intensity; FRP; Composite pipes; Joining
1. Introduction Fiber-reinforced polymer (FRP) composite materials boast qualities, which single them out for use in corrosive environments. It is difficult to estimate the precise fraction of plastics and composites used for corrosion control, because they are not used exclusively for their inherent resistance to corrosion, but also for their low weight, high strength-to-weight ratio, and other unique properties, which make them attractive alternatives to ferrites and alloys [1]. Industries that widely employ FRP composites include aerospace, offshore oil and gas, and chemical and petrochemical. Composite piping systems have been rigorously investigated for their potential development in these industries [2]. Joint introduction is inevitable in piping systems. Many factors dictate the choice of joining technique and the
*
Corresponding author. Tel.: +1 225 578 7009; fax: +1 225 578 5924. E-mail address: [email protected] (J.A. Peck).
0263-8223/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.compstruct.2006.05.009
material used to join including pipe material, the extent of elbows and other geometrical factors, and the everyday purpose of the piping system. The system response to component service life and the adaptability of the overall piping system must be optimized in order to realize more prolific industrial use, especially under the environmental conditions and with the working fluids that these systems will undoubtedly be exposed to. Increased study into joining techniques is crucial to effectively manage component repairs as well as carrying out upgrades and expansions of composite piping systems, thereby elevating the reliability of operation and repair of these systems to a level commiserate with industrial requirements. FRPs have been used extensively and successfully in joining composite pipes. The performance of the joint ultimately depends on the effectiveness of the adhesive used to bind the joint and join the pipes. The formation of an adhesive can be represented in two stages. First, the liquid adhesive spreads over the substrate and the joint, or weld, material. Secondly, the adhesive hardens and its ability to transfer service loads through the substrate and joint
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
material is quantified through direct use [3]. This hardening can be initiated either by chemical processes or polymerization, the latter being preferred in high-performance adhesion. Knowledge of the polymerization process is key when attempting to affect an increase in the reliability of a composite piping system that has joints because the mechanical properties of the joined piping system will depend on the extent of the adhesive cure along with the nature of the monomer. Ultraviolet (UV) curing resins are an attractive alternative to the adhesives most often used for joining composite pipes (ambient environmental curing epoxies and heat-activated curing prepregs). Shortened curing times result in a more employable system due to the inevitability of repair, regardless of material or function; and UV curing resins cure within a matter of minutes as opposed to several hours (prepregs), or over 24 h (epoxies). UV curing resins can also cure with sunlight alone and are therefore more practical for field use where additional effort would be necessary to protect the joint from the environment during a lengthier cure, or impractical levels of energy are input into the system to quicken the cure. However, despite their many benefits, UV curing resins are historically, as is true with many fast-curing resins, less structurally sound than their more protracted counterparts. There exist several factors for optimization of UV curing resins including the concentration of photoinitiators, intensity of UV light used for curing, type of monomer and the presence of oxygen in the curing environment [4]. Many studies have been conducted to evaluate the effect of light intensity on the curing of UV curing resins [3–11]. Most of these studies are dentistry based as UV curing resins offer the safety and speed of cure that is ideal for this discipline. Harris et al. [10] examined the dynamic modulus of elasticity of two composite materials, cured by three light intensities (180 mW/cm2, 350 mW/cm2 and 700 mW/ cm2). They concluded that the specimens that were cured at 180 mW/cm2 were weak and untestable; they further concluded that high curing light intensity might not achieve the most desirable results. Peinado et al. [3] investigated at the influence of the photoinitiator on the kinetics of polymerization for a specific acrylic system and photoinitiator and the applicability of fluorescence studies to quantify mechanical properties resulting from polymerization, concluding primary radical termination as the predominant mechanism during early curing; mechanical testing in this study was, however, limited to a simple lap joint test. Jo¨nsson and Hasselgren [4] examined ‘‘dark polymerization’’, so called for the degree of curing that takes place after initial irradiation. They concluded that a higher light irradiance leads to a higher degree of conversion and an increased polymerization rate. Scherzer and Decker [5] performed a spectroscopic study of the kinetics of photopolymerization induced by monochromatic UV light. Several practical applications are revealed in that study as powder coatings and printer inks. Other studies display similar scales, such as Soppera and Croutxe´-Barghorn [8] who
291
examined free-radical photocurable hybrid sol–gels. In this study, spectroscopy was again used to study the photopolymerization upon irradiation. Five intensities were used: 290 mW/cm2, 180 mW/cm2, 100 mW/cm2, 70 mW/cm2 and 37 mW/cm2. Peck et al. [11] investigated the effect of light intensity on the ability of joined composite pipes to withstand internal pressure and bending loads. Three intensities were used: 80 mW/cm2, 35 mW/cm2 and 15 mW/cm2. This study showed that, while there existed a substantive increase in joint strength with increasing light intensity, the greatest internal pressure achieved was less than is practical for field application of the tested pipes; 5.5 MPa (800 psi), based on a preferred joint rating of 1.4 MPa (200 psi) and a prescribed factor of safety of four. Finite element analysis (FEA), performed in that study, suggested that a strength increase would result if the light intensity were held constant and the joint thickness were reduced, because the number of free radical chains would increase resulting in a higher degree of curing and a greater ability to transfer applied loads even though the reinforcement absorbing those loads were less. In the present study, the effects of three different joint thicknesses on the resulting mechanical performance of a FRP-joined composite piping system adhered with UV curing vinylester resin, were evaluated through direct experimentation. Eighteen composite pipe joints were prepared using E-glass fiber reinforced UV curing vinylester. The effects of thickness were evaluated using internal pressure testing, four-point bending testing and finite element analysis. 2. Raw material 2.1. FRP composite pipes A total of 39 composite pipes were provided for this study by EDO Specialty Plastics in Baton Rouge, Louisiana. The pipes were manufactured by filament winding E-glass fibers onto an epoxy vinylester matrix, marketed under the trade name FiberbondÒ. All pipes have inner diameters of 101.6 mm (4 in.) and wall thicknesses of 6.35 mm (0.25 in.). Thirty-six pipes were cut to a length of 304.8 mm (12 in.); the remaining three were cut to a length of 609.6 mm (24 in.) to be used as a stiffness control for bending tests. 2.2. FRP joint material Reinforcing joint material is comprised of various sequences of two materials, (1) chopped strand mat (CSM), a matrix of short, randomly oriented chopped Eglass fibers, and (2) woven roving, a matrix of bi-directional E-glass fibers. There were three variations of CSM layers (by width) designated as follows: ‘‘A’’ at 203.2 mm (8 in.) wide; ‘‘B’’ at 152.4 mm (6 in.) wide; and ‘‘C’’ at 95.25 mm (3.75 in.) wide. Each joint contained exactly
292
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
Table 1 Joint construction Eight-layer
Five-layer
Three-layer
Resin (mL)
500
312.5
187.5
Sequence
A/A/B/B/E/B/C/C
A/B/E/B/C
A/E/B
Graphic
one E-glass woven layer (designated ‘‘E’’), the width of which was 101.6 mm (4 in.). Three groups of joints were prepared by number of reinforcing layers – eight, five and three, the sequence for which is shown in Table 1 (sequences begin, left-to-right and bottom-to-top, from pipe surface). CSM is used in concert with the woven roving to form the structural cage of the joint. Typically, a synthetic veil is applied over the joint to be used as an external corrosion barrier; however, for the purposes of this study that material was not used as it serves little to no structural importance. 2.3. Joining resin UV curing vinylester resin, commercially available from Sunrez Corporation, was used to wet the joint material. The volume of resin used for each joint variation is listed in Table 1. 3. Sample preparation The procedure for weld preparation and application followed the established procedure set down by EDO Spe-
cialty Plastics, Baton Rouge, for use with FiberbondÒ piping systems [12]. The surfaces of each of the pipes to be joined were prepared by using an angle grinder to remove paint and to roughen the raw surface; Fig. 1(a). The pipes were then cleaned and bonded with a two-part epoxy adhesive, Scotch-Weld 1838 B/A Green; Fig. 1(b) and (c). The purpose of the bond is to fix the pipes for joining. After curing, the bonded pipes were wetted with resin along the outer circumference over the areas to be covered by the joint material. The butt-weld joints were prepared using the wet layup technique. UV curing vinylester was used to wet each weld according to the prescribed volume described in Table 1. Each layer was wetted and rolled onto the preceding layer, which was placed onto a glass substrate; Fig. 1(d). After the weld was fully wet, it was peeled from the substrate and placed over the bond line of the wetted pipe; Fig. 1(e). It was then excised with a roller in order to insure uniformity and to remove trapped air; Fig. 1(f). The 18 welded composite pipe samples were cured in three batches of six. Six 160-W UV fluorescent bulbs were installed onto three UV lamp high-output fixtures. The three fixtures were attached to a frame and positioned vertically at each point of an equilateral triangle; the side-
Fig. 1. (a) Surface preparation; (b) application of bonding adhesive; (c) bonding pipes for joining; (d) wetting of joint; (e) application of wetted joint onto bonded pipes and (f) excising of pipe joint.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
293
4. Experimental procedure Two test methods were employed to evaluate residual mechanical properties of the piping system, internal pressure (bursting) and four-point bending. 4.1. Internal pressure (bursting) testing Internal pressure testing was conducted at EDO Specialty Plastics, in Baton Rouge, per the ASTM-D1599 standard. Vented steel plugs were inserted into either end of the sample pipe. Upon initial pressurization, the plugs were sealed and the internal pressure was increased at a rate of 1.27 MPa/s (185 psi/s) until the peak value was reached; see Fig. 3(a). The pressure reading at joint failure was recorded and the leak location documented. 4.2. Four-point bending testing Simply supported four-point bending testing was conducted to determine the peak bending load and joint stiffness for the system. Testing was conducted with a MTS 810 machine; see Fig. 3(b). The span length for the test was 381 mm (15 in.). The loading rate was 1.27 mm/s (0.05 in./s). 5. Results and discussion Fig. 2. Curing procedure.
length of which was 0.36 m (14 in.); see Fig. 2. Reducing the known surface intensity, immediately in front of the bulbs, to a point source some distance behind that location allowed the estimation of the resulting irradiance using the inverse square law; that irradiance was estimated at 80 mW/cm2. The joined pipes were cured in stacks of three. The time of UV exposure was held constant at 60 min. In this way, the effective UV dose (the product of irradiance and time) was kept constant for all samples; the only variant was joint thickness.
5.1. Effects of irradiance on photopolymerization: observations All samples were, by all accounts, fully cured at the surface and not tacky to the touch. Gravity leaching was apparent in all samples; because the samples were cured vertically gravity leached some resin from the tops of each joint during curing; see Fig. 2. This is visually apparent due to the greenish tint of the cured resin and the white color of the composite layers that comprise the joint; a ring of white can be seen to varying degrees in all samples. The white portions remain, at least superficially, cured at the surface to the touch. Gravity leaching was not appreciably greater
Fig. 3. (a) Photograph of a joined pipe in internal pressure testing and (b) photograph of a joined pipe in four-point bending testing.
294
J.A. Peck et al. / Composite Structures 80 (2007) 290–297 Table 2 FEA element stresses resulting from internal pressure loading
Eight-layer Five-layer Three-layer
Fig. 4. Average internal pressure at leakage/bursting.
in one batch over any other and was therefore not considered a factor in whatever relative differences in residual mechanical properties amongst the three batches. 5.2. Internal pressure (bursting) testing: observations Fig. 4 shows the average bursting pressures for each of the three joint thickness groups (8, 5 and 3 layers). The data clearly shows that the pipes joined by five- and three-layer reinforcement out perform the eight-layer joints. The highest internal pressure attained was 9650 kPa (1400 psi), for a five-layer joint; the lowest was 2760 kPa (400 psi), for an eight-layer joint. The average internal pressure (bursting) values for five-layer and three-layer joints are equal to or greater than the current industry accepted value for a composite pipe joint for use with these pipes (5510 kPa (800 psi)). The resulting internal pressure rating relation with joint thickness may seem counterintuitive to the notion of reinforcing layers, but the importance of degree of cure is overtly defined by the observed data. The thickness of the eight-layer joint is not productive to sufficient curing under the conditions of this experiment. The penetration of UV radiation is not adequate enough to achieve the requisite amount of free radical chains to complete the cure. The degree of cure is demonstrably enhanced in the fivelayer joint and in the three-layer joint; the decrease in internal pressure rating for the three-layer joints is due to the reduction of reinforcing layers and not under-curing. 5.3. Internal pressure (bursting) testing: finite element analysis To validate the test results and assertions based on internal pressure observations, a finite element analysis was conducted. The COSMOS/M software package (version 2.7) was used to model the pipe-joint system. Three-dimensional, eight-node composite elements (SOLIDL) were used to model the pipe wall as well as the FRP joint. Eight-node solid elements (SOLID) were used to model the steel plugs employed at either end during internal pressure testing.
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
5.52 3.47 3.83
21.6 15.2 15.9
2.36 2.09 2.45
In order to simulate the curing profile observed in the FRP joint, the modulus of elasticity of the FRP layers was varied throughout. The modulus was explicitly defined as some percentage of the known, fully cured, modulus1 from the bottom of the joint and stepped down to the weaker top (by 16ths). Furthermore, the modulus was also decreased along the depth (at fifths) according to the general implications of absorption and refraction [3,4,6]. Eighty, distinct and orthotropic, modulus of elasticity sets thusly described each joint. The relative modulus percentages used to vary the ability of the joint to transfer loads, and to simulate gravity leaching, were 100% at the bottom of the joint and 80% at the top. Reducing the thicknesses of the elements varied the number of layers. Also the degree of cure was increased slightly along the depth of the joint as the number of layers was reduced, thus simulating greater penetration of the initial radiation and more complete cure. The percentages along the depth varied from 100–92% (for eight-layer joint), to 100–97% (for three-layer joint). The quantities used to distinguish the batches are estimated quantities based on the relative thicknesses, visible leaching and experimental observations. A 3.45 MPa (500 psi) internal pressure was applied to the model (pipe walls and steel plugs). Elements around the centerline of the joined pipes were weakened to simulate the bond. FEA results support the conclusions that failure occurs in large part because of the loss of load transfer capability resulting from a lower overall cure, which stems from the thicker joint. Table 2 lists the maximum resultant outputs, for each simulated thickness, for the peel stress, the hoop stress and the interfacial shear stress in the joint. The variation seen in Table 2, along with the stress distribution shown in Fig. 5(a)–(c) agrees well with experimental data and observation. The eight-layer model has both higher peak stresses and a wider distribution of high stresses, relative to the other two batches; see Fig. 5(a). From this figure, a clear reduction in peak hoop stress can be seen at the bond line in the five-layer simulation and the stresses in the joint are also less; Fig. 5(b). Fig. 5(c) shows the hoop stress distribution for the three-layer simulated joint. This model agrees well with the results of internal pressure testing, for which we see an increased pressure rating for the five-layer joint over the three-layer and eight-layer, respectively; see Fig. 4. Clear delineations between the ability to transfer load and the ability to withstand that transferred load 1
Fully cured modulus of elasticity: 10.8 GPa, 10.8 GPa and 36.3 GPa.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
295
Fig. 5. FEA results from internal pressure loading for (a) eight-layer; (b) five-layer and (c) three-layer pipe joints.
can be seen in these stress distributions; the modes of failure can be drawn from them as well, i.e. relative failure of adhesive and reinforcing layers. Stresses propagate from the weakened bond line and are distributed (as effectively as the simulated degree of cure allows for) throughout the reinforcing material. Fig. 5(a) shows a high peak stress at the bond line (internally) surrounded by a relatively wide distribution on either side. Externally, it can be seen that the stress at the bond line is much higher, tapering quickly on either side. This suggests that the load is not being effectively transferred to the outer layers. Fig. 5(b) and (c) shows a greater capacity for transferring the loads throughout the joint even though the joints are thinner and the relative reduction in modulus is only 3% and 5%. 5.4. Four-point bending testing: observations Table 3 lists the calculated average stiffnesses for each of the four sample groups (including the control pipes without
Table 3 Averaged stiffnesses and peak loads for four-point bending testing
Control pipes Eight-layer Five-layer Three-layer
Stiffness (N/mm)
Peak load (N)
4456 4478 4542 4390
43,421 27,263 38,066 35,491
joints). As can be seen by this table, the average stiffnesses for the four groups are remarkably similar; in fact the deviation of average stiffness between groups is <4%. The peak loads represent pipe failure in most cases. Most samples show pipe damage under bending loads and do not exhibit sufficient joint failure. When the joints did fail catastrophically they did so by delamination; see Fig. 6(a) and (b). In all other cases the four-point bending apparatus cut through the pipe material locally; see Fig. 6(c). Once this local failure occurred in the pipe substantially, the transferred loads were insufficient to cause failure in the joint material and the tests were halted. Stiffness calculations were performed using the linear portion of a force–displacement curve. This linear portion preceded the pipe failure shown in Fig. 6(c) and is a measure of the stiffness of the pipe system (with joint). While the deviation in stiffnesses between sample groups is too small to infer a definitive relation between joint thickness and joined-pipe stiffness, the stiffness trend of each batch is within scope of what would be expected, i.e. stiffer joints in fully cured samples possessing more reinforcing layers. Data suggests that little sacrifice in stiffness is necessary even with a five-layer reduction in reinforcement. While stresses will be more significant in the joint itself, the sufficient transfer of stress, such as is afforded by a more completely cured adhesive, mitigates significant stress changes in the pipe system. Strength, in this instance, is not greatly compromised by savings in time and material. In fact, the average peak loads are all stronger than an
Fig. 6. (a) Delamination in joint under four-point bending load; (b) crack initiation through delaminated joint section and (c) pipe failure resulting from point load of four-point apparatus.
296
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
environmental curing epoxy adhered FRP joint from a previous study (22.0 kN) [13]. 5.5. Four-point bending testing: finite element analysis FEA loading conditions for four-point bending mimicked the MTS apparatus shown in Fig. 3(b). This was accomplished by applying elemental pressure forces at the location of point-applied loads at the 381 mm (15 in.) span length per the standards followed during testing. Results from FEA using the same model generated for internal pressure testing simulation, but under four-point bending loading conditions, agree well with observation and conclusions drawn from recorded data. Tables 4 and 5 show peak stresses in the FRP layers and substrate, respectively. As predicted by experimental performance, the simulation shows increased peak stresses with decreased joint thickness, yet little change in peak stress is observed in the pipe wall. The peak stresses in the joint, as given in Table 4, stem from decreased thickness in the reinforcing material. Fig. 7(a)–(c) shows the peel stress distributions of the three simulated joints under four-point loading. The deformed shape has been generated to illus-
Table 4 FEA elemental stresses in FRP joint resulting from four-point bending loading
Eight-layer Five-layer Three-layer
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
32.5 51.9 85.7
54.6 55.9 76.8
11.7 11.1 15.2
6. Conclusions Based on the results of this study, it can be said that there exists a correlation between the load carrying capacity of FRP joints wetted with the described UV curing resin and the thickness of that joint. Through experimentation and finite element analysis, it was found that reducing the number of reinforcing layers increased the internal pressure rating of joined pipes. The mechanism for the increase in load carrying capacity is the degree of cure of the adhesive used to bind the reinforcing layers. By achieving a more complete cure, the capacity for the adhesive to transfer load is increased and the reinforcing layers can be optimally utilized. With the results of this study, the volume of reinforcing material, which is the volume currently in use for joining the described sample pipes, can be reduced as can the cost of resin and the time required to cure the joint. While reducing the number of reinforcing layers in the joint reduces the bending strength of the joint, the increased capacity for load transfer minimizes this reduction. Similar stresses were found in the pipe wall, under bending loads, regardless of the number of layers. Acknowledgements
Table 5 FEA elemental stresses in bonded pipes resulting from four-point bending loading
Eight-layer Five-layer Three-layer
trate the mechanism for increased peak stress. As can be seen in the simulations, the point-applied loads act on the joint in a more destructive way as the layers decrease. However, this increase in deformation does not translate into prodigious stress increases in the pipe itself because the interfacial shear stress increases at a much more muted rate due to the increased degree of cure.
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
Top
Bottom
Top
Bottom
Top
Bottom
14.5 14.9 15.6
26.3 26.8 27.0
9.52 9.61 9.66
19.9 19.7 19.6
4.49 4.60 4.62
4.63 4.73 4.75
This investigation was partially sponsored by the Louisiana Board of Regents (BoR) ITRS program, under contract number LEQSF(2001-04)-RD-B-03, and the LaSPACE program, under the Louisiana NASA EPSCoR Dart subprogram contract number NASA/LEQSF(200304)-DART-09. The authors wish to acknowledge EDO Specialty Plastics and Dr. Nikhil Gupta for their contributions to this work. The contents of this paper were presented at the 2004 SAMPE (Society for the Advancement of Material and Process Engineering) annual conference in Long Beach, California.
Fig. 7. FEA results from four-point bending loading for (a) eight-layer; (b) five-layer and (c) three-layer pipe joints.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
References [1] Koch GH, Brongers PH, Thompson NG, Virmani TP, Payer JH. Corrosion cost and preventive strategies in the United States. Office of Infrastructure Research and Development. Federal Highway Administration, Report No. FHWA-RD-01-XXX, September 2001. [2] Pang SS, Li G, Jerro HD, Peck JA, Stubblefield MA. Fast joining of composite pipes using UV curing FRP composites. Polym Compos 2004;25(3):298–306. [3] Peinado C, Salvador EF, Alonso A, Corrales T, Baselga J, Catalina F. Ultraviolet curing of acrylic systems: real-time Fourier transform infrared, mechanical, and fluorescence studies. J Polym Sci: Part A: Polym Chem 2002;40:4236. [4] Jo¨nsson S, Hasselgren C. Secrets of the dark. Fusion UV Systems, Inc., 910 Clopper Road, Gaithersburg, MD 20878. Available from: http://www.fusionuv.com. [5] Scherzer T, Decker U. Real-time FTIR-ATR spectroscopy to study the kinetics of ultrafast photopolymerization reactions induced by monochromatic UV light. Vibr Spectrosc 1999;19:385. [6] Kindernay J, Blazˇkova´ A, Ruda´ J, Jancˇovicˇova V, Jakubı´kova´ Z. Effect of UV light source intensity and spectral distribution on the photopolymerization reactions of a multifunctional acrylated monomer. J Photochem Photobiol A: Chem 2002;151:229.
297
[7] Wu Y, Demachi Y, Tsutsumi O, Kanazawa A, Shino T, Ikeda T. Photoinduced alignment of polymer liquid crystals containing azobenzene moieties in the side chain. 1. Effect of light intensity on alignment behavior. Macromolecules 1998;31:349. [8] Soppera O, Croutxe´-Barghorn C. Real-time fourier transform infrared study of the free-radical ultraviolet-induced polymerization of a hybrid sol–gel. II. The effect of physicochemical parameters on the photopolymerization kinetics. J Polym Sci: Part A: Polym Chem 2003;43:831. [9] Russell GT, Gilbert RG, Napper DH. Chain-length-dependent termination rate processes in free-radical polymerizations. 1. Theory. Macromolecules 1992;25:2459. [10] Harris JS, Jacobsen PH, O’Doherty DM. The effect of curing light intensity and test temperature on the dynamic mechanical properties of two polymer composites. J Oral Rehab 1999;26(8):635. [11] Peck JA, Li G, Pang SS, Stubblefield MA. Light intensity effect on UV cured FRP coupled composite pipe joints. Compos Struct 2004;64(3–4):539–49. [12] Wells K. Welding procedures for FIBERBONDÒ fiberglass piping systems. EDO Specialty Plastics, Baton Rouge, Louisiana. Available from: http://www.fiberbond.com, October 2002. [13] Li G, Pourmohamadian N, Cygan A, Peck JA, Helms JE, Pang SS. Fast repair of laminated beams using UV curing composites. Compos Struct 2003;60(1):73.
UV-cured FRP joint thickness effect on coupled composite pipes Jerry Alan Peck
a,*
, Randy A. Jones b, Su-Seng Pang a, Guoqiang Li a, Brett H. Smith
c
a
c
Mechanical Engineering Department, Louisiana State University, Baton Rouge, LA 70803, United States b EDO Fiber Science, 506 N. Billy Mitchell Road, Salt Lake City, UT 84116, United States NASA Marshall Space Flight Center, Materials, Processes and Manufacturing Department/ED34, Huntsville, AL 35812, United States Available online 16 June 2006
Abstract In this study, 36 FRP composite pipes were joined with various layers of fiber-reinforced UV curing vinylester. Twelve of the composite pipes were joined with an eight-layer weld, 12 with five layers and 12 with three layers; each configuration used exactly one layer of E-glass woven fabric with the remaining layers being comprised of chopped strand mat of various widths. The joined pipes were cured with UV lamps at 80 milli-Watts per square centimeter. The mechanical properties of the cured pipe joints were evaluated by conducting internal pressure testing and simply supported four-point bending testing. The effect of joint thickness on the internal pressure rating, ultimate bending load, and stiffness was evaluated to determine the role of joint thickness in under-curing of UV curing vinylesters. There was a direct correlation observed between joint thickness and internal pressure rating. The five-layer and three-layer joints each performed better than the eight-layer joint in bursting. The mechanism for variation in the residual properties was found to be the degree of cure and was validated by finite element analysis that utilized a composite model that simulates variable cure. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: UV curing; Light intensity; FRP; Composite pipes; Joining
1. Introduction Fiber-reinforced polymer (FRP) composite materials boast qualities, which single them out for use in corrosive environments. It is difficult to estimate the precise fraction of plastics and composites used for corrosion control, because they are not used exclusively for their inherent resistance to corrosion, but also for their low weight, high strength-to-weight ratio, and other unique properties, which make them attractive alternatives to ferrites and alloys [1]. Industries that widely employ FRP composites include aerospace, offshore oil and gas, and chemical and petrochemical. Composite piping systems have been rigorously investigated for their potential development in these industries [2]. Joint introduction is inevitable in piping systems. Many factors dictate the choice of joining technique and the
*
Corresponding author. Tel.: +1 225 578 7009; fax: +1 225 578 5924. E-mail address: [email protected] (J.A. Peck).
0263-8223/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.compstruct.2006.05.009
material used to join including pipe material, the extent of elbows and other geometrical factors, and the everyday purpose of the piping system. The system response to component service life and the adaptability of the overall piping system must be optimized in order to realize more prolific industrial use, especially under the environmental conditions and with the working fluids that these systems will undoubtedly be exposed to. Increased study into joining techniques is crucial to effectively manage component repairs as well as carrying out upgrades and expansions of composite piping systems, thereby elevating the reliability of operation and repair of these systems to a level commiserate with industrial requirements. FRPs have been used extensively and successfully in joining composite pipes. The performance of the joint ultimately depends on the effectiveness of the adhesive used to bind the joint and join the pipes. The formation of an adhesive can be represented in two stages. First, the liquid adhesive spreads over the substrate and the joint, or weld, material. Secondly, the adhesive hardens and its ability to transfer service loads through the substrate and joint
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
material is quantified through direct use [3]. This hardening can be initiated either by chemical processes or polymerization, the latter being preferred in high-performance adhesion. Knowledge of the polymerization process is key when attempting to affect an increase in the reliability of a composite piping system that has joints because the mechanical properties of the joined piping system will depend on the extent of the adhesive cure along with the nature of the monomer. Ultraviolet (UV) curing resins are an attractive alternative to the adhesives most often used for joining composite pipes (ambient environmental curing epoxies and heat-activated curing prepregs). Shortened curing times result in a more employable system due to the inevitability of repair, regardless of material or function; and UV curing resins cure within a matter of minutes as opposed to several hours (prepregs), or over 24 h (epoxies). UV curing resins can also cure with sunlight alone and are therefore more practical for field use where additional effort would be necessary to protect the joint from the environment during a lengthier cure, or impractical levels of energy are input into the system to quicken the cure. However, despite their many benefits, UV curing resins are historically, as is true with many fast-curing resins, less structurally sound than their more protracted counterparts. There exist several factors for optimization of UV curing resins including the concentration of photoinitiators, intensity of UV light used for curing, type of monomer and the presence of oxygen in the curing environment [4]. Many studies have been conducted to evaluate the effect of light intensity on the curing of UV curing resins [3–11]. Most of these studies are dentistry based as UV curing resins offer the safety and speed of cure that is ideal for this discipline. Harris et al. [10] examined the dynamic modulus of elasticity of two composite materials, cured by three light intensities (180 mW/cm2, 350 mW/cm2 and 700 mW/ cm2). They concluded that the specimens that were cured at 180 mW/cm2 were weak and untestable; they further concluded that high curing light intensity might not achieve the most desirable results. Peinado et al. [3] investigated at the influence of the photoinitiator on the kinetics of polymerization for a specific acrylic system and photoinitiator and the applicability of fluorescence studies to quantify mechanical properties resulting from polymerization, concluding primary radical termination as the predominant mechanism during early curing; mechanical testing in this study was, however, limited to a simple lap joint test. Jo¨nsson and Hasselgren [4] examined ‘‘dark polymerization’’, so called for the degree of curing that takes place after initial irradiation. They concluded that a higher light irradiance leads to a higher degree of conversion and an increased polymerization rate. Scherzer and Decker [5] performed a spectroscopic study of the kinetics of photopolymerization induced by monochromatic UV light. Several practical applications are revealed in that study as powder coatings and printer inks. Other studies display similar scales, such as Soppera and Croutxe´-Barghorn [8] who
291
examined free-radical photocurable hybrid sol–gels. In this study, spectroscopy was again used to study the photopolymerization upon irradiation. Five intensities were used: 290 mW/cm2, 180 mW/cm2, 100 mW/cm2, 70 mW/cm2 and 37 mW/cm2. Peck et al. [11] investigated the effect of light intensity on the ability of joined composite pipes to withstand internal pressure and bending loads. Three intensities were used: 80 mW/cm2, 35 mW/cm2 and 15 mW/cm2. This study showed that, while there existed a substantive increase in joint strength with increasing light intensity, the greatest internal pressure achieved was less than is practical for field application of the tested pipes; 5.5 MPa (800 psi), based on a preferred joint rating of 1.4 MPa (200 psi) and a prescribed factor of safety of four. Finite element analysis (FEA), performed in that study, suggested that a strength increase would result if the light intensity were held constant and the joint thickness were reduced, because the number of free radical chains would increase resulting in a higher degree of curing and a greater ability to transfer applied loads even though the reinforcement absorbing those loads were less. In the present study, the effects of three different joint thicknesses on the resulting mechanical performance of a FRP-joined composite piping system adhered with UV curing vinylester resin, were evaluated through direct experimentation. Eighteen composite pipe joints were prepared using E-glass fiber reinforced UV curing vinylester. The effects of thickness were evaluated using internal pressure testing, four-point bending testing and finite element analysis. 2. Raw material 2.1. FRP composite pipes A total of 39 composite pipes were provided for this study by EDO Specialty Plastics in Baton Rouge, Louisiana. The pipes were manufactured by filament winding E-glass fibers onto an epoxy vinylester matrix, marketed under the trade name FiberbondÒ. All pipes have inner diameters of 101.6 mm (4 in.) and wall thicknesses of 6.35 mm (0.25 in.). Thirty-six pipes were cut to a length of 304.8 mm (12 in.); the remaining three were cut to a length of 609.6 mm (24 in.) to be used as a stiffness control for bending tests. 2.2. FRP joint material Reinforcing joint material is comprised of various sequences of two materials, (1) chopped strand mat (CSM), a matrix of short, randomly oriented chopped Eglass fibers, and (2) woven roving, a matrix of bi-directional E-glass fibers. There were three variations of CSM layers (by width) designated as follows: ‘‘A’’ at 203.2 mm (8 in.) wide; ‘‘B’’ at 152.4 mm (6 in.) wide; and ‘‘C’’ at 95.25 mm (3.75 in.) wide. Each joint contained exactly
292
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
Table 1 Joint construction Eight-layer
Five-layer
Three-layer
Resin (mL)
500
312.5
187.5
Sequence
A/A/B/B/E/B/C/C
A/B/E/B/C
A/E/B
Graphic
one E-glass woven layer (designated ‘‘E’’), the width of which was 101.6 mm (4 in.). Three groups of joints were prepared by number of reinforcing layers – eight, five and three, the sequence for which is shown in Table 1 (sequences begin, left-to-right and bottom-to-top, from pipe surface). CSM is used in concert with the woven roving to form the structural cage of the joint. Typically, a synthetic veil is applied over the joint to be used as an external corrosion barrier; however, for the purposes of this study that material was not used as it serves little to no structural importance. 2.3. Joining resin UV curing vinylester resin, commercially available from Sunrez Corporation, was used to wet the joint material. The volume of resin used for each joint variation is listed in Table 1. 3. Sample preparation The procedure for weld preparation and application followed the established procedure set down by EDO Spe-
cialty Plastics, Baton Rouge, for use with FiberbondÒ piping systems [12]. The surfaces of each of the pipes to be joined were prepared by using an angle grinder to remove paint and to roughen the raw surface; Fig. 1(a). The pipes were then cleaned and bonded with a two-part epoxy adhesive, Scotch-Weld 1838 B/A Green; Fig. 1(b) and (c). The purpose of the bond is to fix the pipes for joining. After curing, the bonded pipes were wetted with resin along the outer circumference over the areas to be covered by the joint material. The butt-weld joints were prepared using the wet layup technique. UV curing vinylester was used to wet each weld according to the prescribed volume described in Table 1. Each layer was wetted and rolled onto the preceding layer, which was placed onto a glass substrate; Fig. 1(d). After the weld was fully wet, it was peeled from the substrate and placed over the bond line of the wetted pipe; Fig. 1(e). It was then excised with a roller in order to insure uniformity and to remove trapped air; Fig. 1(f). The 18 welded composite pipe samples were cured in three batches of six. Six 160-W UV fluorescent bulbs were installed onto three UV lamp high-output fixtures. The three fixtures were attached to a frame and positioned vertically at each point of an equilateral triangle; the side-
Fig. 1. (a) Surface preparation; (b) application of bonding adhesive; (c) bonding pipes for joining; (d) wetting of joint; (e) application of wetted joint onto bonded pipes and (f) excising of pipe joint.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
293
4. Experimental procedure Two test methods were employed to evaluate residual mechanical properties of the piping system, internal pressure (bursting) and four-point bending. 4.1. Internal pressure (bursting) testing Internal pressure testing was conducted at EDO Specialty Plastics, in Baton Rouge, per the ASTM-D1599 standard. Vented steel plugs were inserted into either end of the sample pipe. Upon initial pressurization, the plugs were sealed and the internal pressure was increased at a rate of 1.27 MPa/s (185 psi/s) until the peak value was reached; see Fig. 3(a). The pressure reading at joint failure was recorded and the leak location documented. 4.2. Four-point bending testing Simply supported four-point bending testing was conducted to determine the peak bending load and joint stiffness for the system. Testing was conducted with a MTS 810 machine; see Fig. 3(b). The span length for the test was 381 mm (15 in.). The loading rate was 1.27 mm/s (0.05 in./s). 5. Results and discussion Fig. 2. Curing procedure.
length of which was 0.36 m (14 in.); see Fig. 2. Reducing the known surface intensity, immediately in front of the bulbs, to a point source some distance behind that location allowed the estimation of the resulting irradiance using the inverse square law; that irradiance was estimated at 80 mW/cm2. The joined pipes were cured in stacks of three. The time of UV exposure was held constant at 60 min. In this way, the effective UV dose (the product of irradiance and time) was kept constant for all samples; the only variant was joint thickness.
5.1. Effects of irradiance on photopolymerization: observations All samples were, by all accounts, fully cured at the surface and not tacky to the touch. Gravity leaching was apparent in all samples; because the samples were cured vertically gravity leached some resin from the tops of each joint during curing; see Fig. 2. This is visually apparent due to the greenish tint of the cured resin and the white color of the composite layers that comprise the joint; a ring of white can be seen to varying degrees in all samples. The white portions remain, at least superficially, cured at the surface to the touch. Gravity leaching was not appreciably greater
Fig. 3. (a) Photograph of a joined pipe in internal pressure testing and (b) photograph of a joined pipe in four-point bending testing.
294
J.A. Peck et al. / Composite Structures 80 (2007) 290–297 Table 2 FEA element stresses resulting from internal pressure loading
Eight-layer Five-layer Three-layer
Fig. 4. Average internal pressure at leakage/bursting.
in one batch over any other and was therefore not considered a factor in whatever relative differences in residual mechanical properties amongst the three batches. 5.2. Internal pressure (bursting) testing: observations Fig. 4 shows the average bursting pressures for each of the three joint thickness groups (8, 5 and 3 layers). The data clearly shows that the pipes joined by five- and three-layer reinforcement out perform the eight-layer joints. The highest internal pressure attained was 9650 kPa (1400 psi), for a five-layer joint; the lowest was 2760 kPa (400 psi), for an eight-layer joint. The average internal pressure (bursting) values for five-layer and three-layer joints are equal to or greater than the current industry accepted value for a composite pipe joint for use with these pipes (5510 kPa (800 psi)). The resulting internal pressure rating relation with joint thickness may seem counterintuitive to the notion of reinforcing layers, but the importance of degree of cure is overtly defined by the observed data. The thickness of the eight-layer joint is not productive to sufficient curing under the conditions of this experiment. The penetration of UV radiation is not adequate enough to achieve the requisite amount of free radical chains to complete the cure. The degree of cure is demonstrably enhanced in the fivelayer joint and in the three-layer joint; the decrease in internal pressure rating for the three-layer joints is due to the reduction of reinforcing layers and not under-curing. 5.3. Internal pressure (bursting) testing: finite element analysis To validate the test results and assertions based on internal pressure observations, a finite element analysis was conducted. The COSMOS/M software package (version 2.7) was used to model the pipe-joint system. Three-dimensional, eight-node composite elements (SOLIDL) were used to model the pipe wall as well as the FRP joint. Eight-node solid elements (SOLID) were used to model the steel plugs employed at either end during internal pressure testing.
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
5.52 3.47 3.83
21.6 15.2 15.9
2.36 2.09 2.45
In order to simulate the curing profile observed in the FRP joint, the modulus of elasticity of the FRP layers was varied throughout. The modulus was explicitly defined as some percentage of the known, fully cured, modulus1 from the bottom of the joint and stepped down to the weaker top (by 16ths). Furthermore, the modulus was also decreased along the depth (at fifths) according to the general implications of absorption and refraction [3,4,6]. Eighty, distinct and orthotropic, modulus of elasticity sets thusly described each joint. The relative modulus percentages used to vary the ability of the joint to transfer loads, and to simulate gravity leaching, were 100% at the bottom of the joint and 80% at the top. Reducing the thicknesses of the elements varied the number of layers. Also the degree of cure was increased slightly along the depth of the joint as the number of layers was reduced, thus simulating greater penetration of the initial radiation and more complete cure. The percentages along the depth varied from 100–92% (for eight-layer joint), to 100–97% (for three-layer joint). The quantities used to distinguish the batches are estimated quantities based on the relative thicknesses, visible leaching and experimental observations. A 3.45 MPa (500 psi) internal pressure was applied to the model (pipe walls and steel plugs). Elements around the centerline of the joined pipes were weakened to simulate the bond. FEA results support the conclusions that failure occurs in large part because of the loss of load transfer capability resulting from a lower overall cure, which stems from the thicker joint. Table 2 lists the maximum resultant outputs, for each simulated thickness, for the peel stress, the hoop stress and the interfacial shear stress in the joint. The variation seen in Table 2, along with the stress distribution shown in Fig. 5(a)–(c) agrees well with experimental data and observation. The eight-layer model has both higher peak stresses and a wider distribution of high stresses, relative to the other two batches; see Fig. 5(a). From this figure, a clear reduction in peak hoop stress can be seen at the bond line in the five-layer simulation and the stresses in the joint are also less; Fig. 5(b). Fig. 5(c) shows the hoop stress distribution for the three-layer simulated joint. This model agrees well with the results of internal pressure testing, for which we see an increased pressure rating for the five-layer joint over the three-layer and eight-layer, respectively; see Fig. 4. Clear delineations between the ability to transfer load and the ability to withstand that transferred load 1
Fully cured modulus of elasticity: 10.8 GPa, 10.8 GPa and 36.3 GPa.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
295
Fig. 5. FEA results from internal pressure loading for (a) eight-layer; (b) five-layer and (c) three-layer pipe joints.
can be seen in these stress distributions; the modes of failure can be drawn from them as well, i.e. relative failure of adhesive and reinforcing layers. Stresses propagate from the weakened bond line and are distributed (as effectively as the simulated degree of cure allows for) throughout the reinforcing material. Fig. 5(a) shows a high peak stress at the bond line (internally) surrounded by a relatively wide distribution on either side. Externally, it can be seen that the stress at the bond line is much higher, tapering quickly on either side. This suggests that the load is not being effectively transferred to the outer layers. Fig. 5(b) and (c) shows a greater capacity for transferring the loads throughout the joint even though the joints are thinner and the relative reduction in modulus is only 3% and 5%. 5.4. Four-point bending testing: observations Table 3 lists the calculated average stiffnesses for each of the four sample groups (including the control pipes without
Table 3 Averaged stiffnesses and peak loads for four-point bending testing
Control pipes Eight-layer Five-layer Three-layer
Stiffness (N/mm)
Peak load (N)
4456 4478 4542 4390
43,421 27,263 38,066 35,491
joints). As can be seen by this table, the average stiffnesses for the four groups are remarkably similar; in fact the deviation of average stiffness between groups is <4%. The peak loads represent pipe failure in most cases. Most samples show pipe damage under bending loads and do not exhibit sufficient joint failure. When the joints did fail catastrophically they did so by delamination; see Fig. 6(a) and (b). In all other cases the four-point bending apparatus cut through the pipe material locally; see Fig. 6(c). Once this local failure occurred in the pipe substantially, the transferred loads were insufficient to cause failure in the joint material and the tests were halted. Stiffness calculations were performed using the linear portion of a force–displacement curve. This linear portion preceded the pipe failure shown in Fig. 6(c) and is a measure of the stiffness of the pipe system (with joint). While the deviation in stiffnesses between sample groups is too small to infer a definitive relation between joint thickness and joined-pipe stiffness, the stiffness trend of each batch is within scope of what would be expected, i.e. stiffer joints in fully cured samples possessing more reinforcing layers. Data suggests that little sacrifice in stiffness is necessary even with a five-layer reduction in reinforcement. While stresses will be more significant in the joint itself, the sufficient transfer of stress, such as is afforded by a more completely cured adhesive, mitigates significant stress changes in the pipe system. Strength, in this instance, is not greatly compromised by savings in time and material. In fact, the average peak loads are all stronger than an
Fig. 6. (a) Delamination in joint under four-point bending load; (b) crack initiation through delaminated joint section and (c) pipe failure resulting from point load of four-point apparatus.
296
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
environmental curing epoxy adhered FRP joint from a previous study (22.0 kN) [13]. 5.5. Four-point bending testing: finite element analysis FEA loading conditions for four-point bending mimicked the MTS apparatus shown in Fig. 3(b). This was accomplished by applying elemental pressure forces at the location of point-applied loads at the 381 mm (15 in.) span length per the standards followed during testing. Results from FEA using the same model generated for internal pressure testing simulation, but under four-point bending loading conditions, agree well with observation and conclusions drawn from recorded data. Tables 4 and 5 show peak stresses in the FRP layers and substrate, respectively. As predicted by experimental performance, the simulation shows increased peak stresses with decreased joint thickness, yet little change in peak stress is observed in the pipe wall. The peak stresses in the joint, as given in Table 4, stem from decreased thickness in the reinforcing material. Fig. 7(a)–(c) shows the peel stress distributions of the three simulated joints under four-point loading. The deformed shape has been generated to illus-
Table 4 FEA elemental stresses in FRP joint resulting from four-point bending loading
Eight-layer Five-layer Three-layer
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
32.5 51.9 85.7
54.6 55.9 76.8
11.7 11.1 15.2
6. Conclusions Based on the results of this study, it can be said that there exists a correlation between the load carrying capacity of FRP joints wetted with the described UV curing resin and the thickness of that joint. Through experimentation and finite element analysis, it was found that reducing the number of reinforcing layers increased the internal pressure rating of joined pipes. The mechanism for the increase in load carrying capacity is the degree of cure of the adhesive used to bind the reinforcing layers. By achieving a more complete cure, the capacity for the adhesive to transfer load is increased and the reinforcing layers can be optimally utilized. With the results of this study, the volume of reinforcing material, which is the volume currently in use for joining the described sample pipes, can be reduced as can the cost of resin and the time required to cure the joint. While reducing the number of reinforcing layers in the joint reduces the bending strength of the joint, the increased capacity for load transfer minimizes this reduction. Similar stresses were found in the pipe wall, under bending loads, regardless of the number of layers. Acknowledgements
Table 5 FEA elemental stresses in bonded pipes resulting from four-point bending loading
Eight-layer Five-layer Three-layer
trate the mechanism for increased peak stress. As can be seen in the simulations, the point-applied loads act on the joint in a more destructive way as the layers decrease. However, this increase in deformation does not translate into prodigious stress increases in the pipe itself because the interfacial shear stress increases at a much more muted rate due to the increased degree of cure.
Peel stress (MPa)
Hoop stress (MPa)
Interfacial shear stress (MPa)
Top
Bottom
Top
Bottom
Top
Bottom
14.5 14.9 15.6
26.3 26.8 27.0
9.52 9.61 9.66
19.9 19.7 19.6
4.49 4.60 4.62
4.63 4.73 4.75
This investigation was partially sponsored by the Louisiana Board of Regents (BoR) ITRS program, under contract number LEQSF(2001-04)-RD-B-03, and the LaSPACE program, under the Louisiana NASA EPSCoR Dart subprogram contract number NASA/LEQSF(200304)-DART-09. The authors wish to acknowledge EDO Specialty Plastics and Dr. Nikhil Gupta for their contributions to this work. The contents of this paper were presented at the 2004 SAMPE (Society for the Advancement of Material and Process Engineering) annual conference in Long Beach, California.
Fig. 7. FEA results from four-point bending loading for (a) eight-layer; (b) five-layer and (c) three-layer pipe joints.
J.A. Peck et al. / Composite Structures 80 (2007) 290–297
References [1] Koch GH, Brongers PH, Thompson NG, Virmani TP, Payer JH. Corrosion cost and preventive strategies in the United States. Office of Infrastructure Research and Development. Federal Highway Administration, Report No. FHWA-RD-01-XXX, September 2001. [2] Pang SS, Li G, Jerro HD, Peck JA, Stubblefield MA. Fast joining of composite pipes using UV curing FRP composites. Polym Compos 2004;25(3):298–306. [3] Peinado C, Salvador EF, Alonso A, Corrales T, Baselga J, Catalina F. Ultraviolet curing of acrylic systems: real-time Fourier transform infrared, mechanical, and fluorescence studies. J Polym Sci: Part A: Polym Chem 2002;40:4236. [4] Jo¨nsson S, Hasselgren C. Secrets of the dark. Fusion UV Systems, Inc., 910 Clopper Road, Gaithersburg, MD 20878. Available from: http://www.fusionuv.com. [5] Scherzer T, Decker U. Real-time FTIR-ATR spectroscopy to study the kinetics of ultrafast photopolymerization reactions induced by monochromatic UV light. Vibr Spectrosc 1999;19:385. [6] Kindernay J, Blazˇkova´ A, Ruda´ J, Jancˇovicˇova V, Jakubı´kova´ Z. Effect of UV light source intensity and spectral distribution on the photopolymerization reactions of a multifunctional acrylated monomer. J Photochem Photobiol A: Chem 2002;151:229.
297
[7] Wu Y, Demachi Y, Tsutsumi O, Kanazawa A, Shino T, Ikeda T. Photoinduced alignment of polymer liquid crystals containing azobenzene moieties in the side chain. 1. Effect of light intensity on alignment behavior. Macromolecules 1998;31:349. [8] Soppera O, Croutxe´-Barghorn C. Real-time fourier transform infrared study of the free-radical ultraviolet-induced polymerization of a hybrid sol–gel. II. The effect of physicochemical parameters on the photopolymerization kinetics. J Polym Sci: Part A: Polym Chem 2003;43:831. [9] Russell GT, Gilbert RG, Napper DH. Chain-length-dependent termination rate processes in free-radical polymerizations. 1. Theory. Macromolecules 1992;25:2459. [10] Harris JS, Jacobsen PH, O’Doherty DM. The effect of curing light intensity and test temperature on the dynamic mechanical properties of two polymer composites. J Oral Rehab 1999;26(8):635. [11] Peck JA, Li G, Pang SS, Stubblefield MA. Light intensity effect on UV cured FRP coupled composite pipe joints. Compos Struct 2004;64(3–4):539–49. [12] Wells K. Welding procedures for FIBERBONDÒ fiberglass piping systems. EDO Specialty Plastics, Baton Rouge, Louisiana. Available from: http://www.fiberbond.com, October 2002. [13] Li G, Pourmohamadian N, Cygan A, Peck JA, Helms JE, Pang SS. Fast repair of laminated beams using UV curing composites. Compos Struct 2003;60(1):73.