Net Shape Manufacturing and the Performance of Polymer Composites Under Dynamic Loads

Transcription

1 Net Shape Manufacturing and the Performance of Polymer Composites Under Dynamic Loads by D. Arola and M. Ramulu ABSTRACT--An experimental investigation was conducted to determine the influence of postmold trimming and resultant edge quality on the performance of fiber-reinforced plastics (FRP) under dynamic loads. Graphite/epoxy and graphitefoismaleimide laminates were machined using three state-of-the-art industrial techniques and subjected to threepoint bend impact to failure. The load load-line displacement records were used to obtain the load, bend deflection and energy absorbed to fracture. High-speed photography was also employed to identify the initiation and progression of failure and record the time dependent fracture process. From a comparison of edge quality and subsequent material performance for both polymeric composites, it was found that the impact response of FRPs is highly process dependent. In general, the load and energy absorbed to fracture decreases with increasing surface roughness. Reductions in the load and energy to the onset of fracture with degrading surface quality were as high as 20 percent. The dynamic response was also found to be dependent on the constituents, stacking sequence and impact velocity. KEY WORDS--Composite materials, impact strength, machining. Introduction Fiber-reinforced plastics (FRP) are used in various hightechnology industries due to the benefits provided by their exceptional strength and stiffness-to-weight ratios. Through the proper choice of constituents, primary processing and component layup, FRPs offer unlimited design flexibility. Although these materials have tremendous advantages, net shape trimming is a challenging task and often introduces edge-finishing defects to the component part. However, the effect of net shape trimming and consequent manufacturing defects on the performance of FRPs is not fully understood. The obstacles faced in net shape trimming of FRPs are a prevalent topic in advanced manufacturing research. Abrate and Walton 1'2 and Ramulu and Arola 3 recently reviewed the traditional and nontraditional methods used to obtain component parts. The most common techniques include abrasive diamond cutters (ADS), 4 polycrystalline diamond (PCD) cutting tools 5'6 and the abrasive waterjet (AWJ). 7,8 These are the "preferred" industrial methods based on their ability to provide high ma- D. Arola (SEM Member) is a Graduate Student, and M. Ramulu (SEM Member) is Professor, Department of Mechanical Engineering, University of Washington, Box , Seattle, WA Original manuscript submitted: May 9, Final manuscript received: September 24, terial removal rates, relatively low operating costs (through tool and/or auxiliary supplies per unit length) and a highquality trimmed edge. 9 But the importance of manufacturing and surface integrity to the performance of reinforced plastics in structural designs has not been adequately addressed. Preliminary investigations have shown that postmold processing is detrimental to the mechanical properties of FRPs, even under quasi-static loads, t~ However, no reports in the literature focus on the influence of surface integrity on the performance of FRPs under transient loads. Furthermore, no study has reported the effect of manufacturing defects on the impact strength of reinforced polymers. The purpose of this investigation was to determine the importance of net shape machining and subsequent edge quality to the performance of FRPs under dynamic loading. Three common industrial techniques were used to obtain impact specimens from graphite/epoxy (Gr/Ep) and graphite/bismaleimide (Gr/Bmi) laminates, The machined surfaces were evaluated qualitatively prior to the performance assessment in terms of the surface texture and so on. The influence of trimmed-edge quality on the mechanical behavior of FRPs under impact loading is reported, and its importance to composite structures is discussed. Experimental Materia/s Gr/Ep and Gr/Bmi laminates were used for this study. The Gr/Ep laminate consisted of resin and IM-6 fibers with a lamina-stacking sequence of [X,45 deg,(-45 deg,45 deg,90 deg,0 deg)2-45 deg,0 deg,-45 deg,-45 deg,90 deg,-45 deg,(0 deg,90 deg,45 deg,-45 deg)2,45 deg,x], where X represents a single cross ply weave oriented coincident with the specimen axis; each ply of the laminate is 200 ~tm in thickness, providing a total thickness of 5.2 mm. The Gr/Bmi laminate was composed of IM-7 fibers and BASF resin matrix with a lamina-stacking sequence of [0,45,90,-4516s, an average ply thickness of 125 ktm and a laminate thickness of slightly over 6 mm. The fiber volume fraction of each material is approximately 0.65, and the tape mechanical properties are listed in Table 1. Fabrication of Test Specimens Over 250 impact specimens were obtained from the two polymeric composites. A total of 100 specimens were machined from the Gr/Ep laminate, 20 each from ADS and PCD trimming and 60 specimens with the AWJ (20 each at three different cutting conditions). Thirty specimens were Experimental Mechanics 9 379

2 obtained from the Gr/Bmi laminate under each of the aforementioned conditions for a total of 150 specimens. Sample dimensions conform to the geometry specified for flexure tests in ASTM D790M (1994). 15 All of the ADS-machined specimens from both laminates were obtained with a #220 diamond grit wheel. Abrasive waterjet trimming was performed using three independent cutting conditions. The parametric conditions, denoted A, B and C, incorporated #50, #80 and #150 garnet abrasives, respectively, to provide different levels of machined surface quality according to an empirical model for AWJ machining of Gr/Ep. 80rthogonally trimmed specimens were precut with the AWJ and then edge trimmed using the appropriate orthogonal cutting tool inserts. The Gr/Ep laminate was trimmed using Compax 3000 PCD inserts with a 10-deg rake and 7-deg clearance angle, whereas C-4 carbide inserts with a 7-deg rake and 17-deg clearance angle were used for the Gr/Bmi laminate. Tool geometry of the PCD and carbide inserts was chosen according to results from a surface quality optimization study for edge-trimming FRP materials. 6 Experimental Procedure The surface texture and microstructural integrity of the impact specimens resulting from each of the three methods of machining were analyzed using contact profilometry and scanning electron microscopy (SEM). Profilometry was conducted with a SurfAnalyzer TM 4000 profilometer using a 5- ~tm diameter probe. Longitudinal profiles parallel to the cutting direction were recorded at 500-~tm intervals along the laminate thickness of each specimen. Transverse measurements were also obtained perpendicular to the cutting direction. A traverse length of 3.5 mm was used for all profiles, and the average surface roughness (Ra) and peak-to-valley height (Ry) were calculated according to ANSI B with a 0.8-mm cutoff length. The surface profiles were also examined using the probability density and cumulative height distribution scales. Microscopic analysis was conducted on a representative specimen from each finishing technique with a Jeol JSM-T330A SEM for a comparison of material removal features and subprofile constituent integrity. The impact tests were conducted using an instrumented, drop-weight, three-point flexure test apparatus as shown in Fig. l(a). A 16:1 span-to-depth ratio (L/d) was used for the simple support load configuration for both laminates as illustrated in Fig. l(b). Specimens were oriented with the machined edges coincident to the plane of the applied load. A tup diameter of 6.35 mm was used according to ASTM specifications, 15 and four pieces of masking tape were attached to the tup nose to reduce ringing of the accelerometer at the onset of impact. The impact response of the Gr/Bmi laminate was investigated at and 3.75-m/s drop-weight velocities, both of which were conducted with the minimum energy to failure established from precursory experimental analysis. The corresponding tup masses for the two velocities were 8.72 and 4.13 kg, which provided an impact energy of 22.1 and 29.0 N.m, respectively. The Gr/Ep laminate was tested at only the 2.25-m/s impact velocity with a tup mass of 4.13 kg (10.5 N-m energy). A piezoelectric load cell and two eddy current displacement transducers were used for the dynamic load and displacement measurements, respectively. A variable low-pass filter was selected to filter the load signatures recorded during the impact event with a cutoff fre- Fig. 1--Experimental equipment and test specimens: (a) instrumented drop weight impact apparatus, (b) specimen geometry and load configuration quency of 5 x 103 Hz. The impact load was recorded in both the filtered and nonfiltered states along with the displacement history using two Tektronix digital storage oscilloscopes. An Imacon high-speed camera was used to document the failure process with a framing module of 10,000 frames/s. Results Prior to analyzing the dynamic response of the Gr/Ep and Gr/Bmi laminates, the surface roughness and cohesive integrity of the constituents resulting from each method of trimming were evaluated. From an unassisted visual assessment, the laminate trimmed with the ADS and AWJ exhibited a far superior surface integrity than that subjected to edge trimming with single-point cutting tools. Profilometry The surface roughness of the machined edges was determined from profiles taken parallel and transverse to the cutting direction. Standard roughness parameters from the Gr/Ep and Gr/Bmi laminates are listed in Table 2. The lowest average roughness (Ra) of both laminates was 0.2 lam and was received from the ADS specimens, as expected, due to the small diamond abrasive size of this cutter. Orthogonal trimming with the PCD and carbide tool inserts resulted in a 1.5-1am Ra, which was significantly greater than the ADStrimmed specimens but lower than that resulting from AWJ machining. The longitudinal Ra of the AWJ-machined Gr/Ep specimens ranged from 1.9 to 10.8 lam, and the roughness of the Gr/Bmi specimens ranged from 1.8 to 9.7 ~tm. Roughness measurements for the two abrasive material removal techniques were essentially the same in the transverse direction Vol. 37, No. 4, December 1997

3 TABLE 1--MECHANICAL PROPERTIES OF Gr/Ep AND Gr/Bmi Elastic Constants Ell E22 G12 Xt (GPa) (GPa) (GPa) v]2 (GPa) Gr/Ep Gr/Bmi ,4 TABLE 2--SURFACE ROUGHNESS OF THE MACHINED SPECIMENS Gr/Ep Surface Roughness Ra (~m) Ry (pm) Specimen long. tran. long. tran. ADS PCD AWJ A AWJ B AWJ C i Ultimate Strength Xc Yt Yc S (GPa) (MPa) (MPa) (MPa) Gr/Bmi Surface Roughness Ra (pro) R~, (itm) long. tran. long. tran (perpendicular to the cutting direction) and the longitudinal direction. The surface roughness was essentially independent of measurement direction or ply orientation. However, the surface roughness in the transverse direction of the PCDtrimmed material was significantly greater than it was in the longitudinal direction. The difference arises from trimmingrelated damage to the -45-deg plies which is a consequence of the material removal sensitivity to fiber orientation. 4'5 Extensive subsurface delamination often occurs in edge trimming of multidirectional laminates with single-point cutting tools in comparison to the abrasive cutting techniques due to large cutting forces. The -45-deg plies are subject to significant interlaminar tension and shear stresses by virtue of their orientation relative to the cutting direction. Large cutting forces are detrimental to the low matrix dependent strength components of reinforced polymers and often promote significant delamination. Representative statistical distributions of longitudinal profile height measurements from the Gr/Ep laminate are shown in Figs. 2(a)-(d). Two profiles from the PCD-trimmed surface are shown in Fig. 2(a), including one representative of the majority of the surface and another taken from the trimmed surface along a damaged -45-deg ply. The superior quality of the ADS specimens is evident from the comparison in probability density of profile height in Fig. 2(a) and the minor slope of the cumulative height distribution in Fig. 2(c). Furthermore, the increase in surface roughness with grit size of the AWJ-machined specimens is clearly apparent from the increase in profile height distribution about the mean line in Figs. 2(b) and 2(d). In contrast to the ADS- and AWJtrimmed surfaces, subsurface delamination resulting from the use of single-point cutting tools is apparent from the profile height distributions in Figs. 2(a) and 2(c). Edge-trimming process damage to the -45-deg plies of the Gr/Ep laminate extended in excess of 60/am below the plane of the trimmed surface. Microscopy From a comparison of the surface texture resulting from each process, the most extensive trimming damage of the impact specimens occurred to the -45-deg plies of the laminates trimmed with single-point cutting tools. The contrast in constituent relationship between edge trimming with singlepoint cutting tools and AWJ machining is illustrated in Fig. 3. g o l lo is :lo 21 N 35 n s io i~ 2o ~ N )~ Percent (%) Pedant (%},~) b) c) d) Fig. 2--Probability density and cumulative height distribution of the Gr/Ep trimmed edges: (a) probability density, ADS and PCD; (b) probability density, AWJ; (c) cumulative height distribution, ADS and PCD; (d) cumulative height distribution, AWJ Microscopic views of the -45-deg and 90-deg plies from the surface of a Gr/Bmi specimen trimmed with carbide-cutting tools are shown in Figs. 3(a) and 3(b), respectively. The micrographs of the AWJ-machined Gr/Bmi in Figs. 3(c) and 3(d) emphasize the high degree of interstitial integrity along the fiber/matrix interface and the lack of constituent disruption resulting from the abrasive-based material removal. Although only two ply orientations are shown here, the features are consistent with those of the remaining ply orientations and are discussed more extensively in Refs. 7 and 8. In contrast to the characteristics of the two abrasive removal techniques, the plies trimmed with single-point cutting tools in Figs. 3(a) and 3(b) exhibit a poor interface constituent integrity, which is evident by the fractured fibers and fiber pullout in these micrographs. Furthermore, matrix reconsolidation visible on the surface of the 90-deg ply implies that surface texture quantified using profilometry may misrepresent the actual surface integrity. Identical plies of an AWJ B sample are shown in Figs. 3(c) and 3(d). The microscopic, machined surface features of the Gr/Ep laminate are consistent with those of the Gr/Bmi laminate and have been reported previously. 14 Micrographs of the ADS specimens Experimental Mechanics 9 381

4 I~ t" 14 Fracture PCD 8o a) b) time (ms) a) b) ; I0 time (ms) 3ooo~ Fracture., ~/ AWJB~eO ~,lsook r / " I ~" c) d) Fig. 3--Micrographs of the edge-trimmed Gr/Bmi laminate (1000x magnification): (a) -45-deg ply, PCD; (b) 90-deg ply, PCD; (c) -45-deg ply, AWJ B; (d) 90-deg ply, AWJ B are not shown, but the features are essentially identical to those resulting from AWJ trimming in Fig. 3. Furthermore, characteristics of the AWJ-machined surfaces in Fig. 3 are also representative of the specimens received from the use of conditions AWJ A and C with #50 and #150 abrasive mesh sizes, respectively. Impact Loading Impact tests were conducted to failure with the FRP specimens under drop-weight three-point bend loading. A typical load load-line displacement profile from an ADS-trimmed Gr/Bmi specimen at a 2.25-m/s impact velocity is shown in Fig. 4(a). Similarly, the load load-line displacement profiles from a representative PCD and AWJ B specimen are shown in Figs. 4(b) and 4(c). Noteworthy features of the impact event include the peak load and peak fracture load as noted in Fig. 4(a), as well as the laminate bend deflection and energy absorbed to the peak fracture load. The fracture energy is the integral of the tup load with respect to displacement of the peak fracture load which demarcates the onset of irreversible damage to the laminate. The peak load is essentially the result of the energy required of the tup to accelerate the laminate from rest to a velocity equivalent to the tup velocity. An average of the aforementioned properties defining the impact response of the five specimen groups for the Gr/Ep laminate are listed in Table 3. The equivalent properties for the Gr/Bmi laminate subjected to 2.25-rn/s and 3.75-m/s drop-weight velocities are listed in Tables 4 and 5, respectively. Only minimal variations in the peak impact load with the method of trimming were noted, as expected. However, the peak fracture load, lateral deflection and the energy dissipated to the onset of fracture were clearly manufacturing process dependent. The ADS specimens with the highest surface quality had the largest peak fracture load for each laminate '~176)... 1,o o... r-;, Io time (ms) Fig. 4~Typical load load-line displacement of the Gr/Bmi laminate at 2.25 m/s impact velocity (solid line = load, dashed line = displacement): (a) ADS, (b) PCD, (c) AWJ B material and under both impact velocities. Contrary to expectations from the surface roughness measurements, the onset of irreversible damage of the PCD-trimmed specimens occurred at a significantly lower fracture load and energy than that of the ADS and AWJ C specimens. The energy levels to the onset of fracture under a 2.25-rn/s impact velocity for the PCD-tfimmed Gr/Ep and Gr/Bmi laminates were I 1 percent and 18 percent lower, respectively, than they were for the corresponding ADS-trimmed laminate. But the reduction in energy to fracture of the PCD-trimmed Gr/Bmi laminate at a 3.75-m/s impact velocity exceeded 35 percent. The most interesting feature of the PCD-trimmed specimen performance was the extensive reduction in impact load to the onset of fracture. The impact load of the PCD-trimmed Gr/Bmi and Gr/Ep was 20 percent less than that of the ADS (control) specimens under both drop velocities. Also, the performance of the AWJ-machined laminates clearly decreased with increasing surface roughness. Specimens that were machined with the smallest grit size in group AWJ C (#150, 180-pm effective diameter) had only marginal reductions in the failure parameters characterizing the impact response. However, the AWJ A specimens with the highest surface roughness had reductions in peak fracture load from 10 percent to 13 percent and reduction in energy to fracture from 11 percent to 35 percent. The time to the onset of fracture corresponded with the extent of bend deflection for each method of processing. For the Gr/Bmi laminate subjected to drop-weight impact at 2.25 m/s, the average time to fracture of the ADS and PCD specimens was 4.7 and 4.0 ms, respectively. The corresponding time to fracture of the AWJ specimens was 4.1, 4.4 and 4.6 ms for groups A, B and C, respectively. Similar behavior in the time to fracture was also noted for the other two impact conditions. Discussion An evaluation of the machined surface and impact test results from this study has shown that indeed the impact re Vol. 37, No. 4, December 1997

5 TABLE 3--IMPACT RESPONSE OF Gr/Ep AT LOW-VELOCITY IMPACT Peak Load Fracture Load Fracture Displacement Energy at Failure Method of Trimming (N) (N) (mm) (Nmm) ADS PCD AWJ A AWJ B AWJ C TABLE 4--IMPACT RESPONSE OF Gr/Bmi AT LOW-VELOCITY IMPACT Peak Load i Fracture Load Fracture Displacement Energy at Failure Method of Trimming (N) (N) (mm) (N.mm) ADS 3062:1: PC D AWJ A AWJ B AWJ C TABLE 5--IMPACT RESPONSE OF GdBmi AT HIGH-VELOCITY IMPACT Peak Load Fracture Load i Fracture Displacement Energy at Failure Method of Trimming (N) (N) (mm) (N.mm) ADS PCD AWJ A AWJ B AWJ C sponse of reinforced polymeric materials is dependent on the trimmed-edge quality and presence of manufacturing defects. The laminates trimmed with the ADS exhibited the highest machined quality and material performance in terms of the load, deflection and energy to the onset of fracture. A plot of the distribution in peak fracture load and total impact energy with average surface roughness for the Gr/Ep and Gr/Bmi at 2.25 m/s and 3.75 m/s impact velocities are shown in Figs. 5(a)-(c), respectively. In general, both the peak fracture load and impact energy decreased with an increase in the average surface roughness (Ra). An exception to this trend was evident in the fracture energy of the Gr/Bmi at high-velocity impact. The decrease in peak fracture load and absorbed energy with surface texture of the AWJ-machined laminates appears linear, but an extrapolation to the surface roughness of the ADS-trimmed specimens does not correspond to the impact response resulting from this method of trimming. Furthermore, the PCD-trimmed laminates' response did not comply with the trends between surface roughness and dynamic strength resulting from AWJ machining. Hence, using average surface roughness alone to infer the surface integrity of a FRP subjected to net shape trimming can be misleading. An alternate evaluation of the impact response for the laminates is shown in Figs. 6(a)-(c), in which the impact energy is plotted in terms of the instantaneous bend deflection. Although the onset of fracture of the PCD-trimmed laminates occurred at a much lower impact load and energy, they continued to absorb tup energy equivalent to that of the laminates trimmed with the other processing techniques. In fact, the PCD-trimmed Gr/Ep absorbed more energy per unit displacement after the onset of fracture than the ADS- or AWJmachined laminates as shown in Fig. 6(a). This conflicts with the response of the Gr/Bmi laminate as shown in Figs. 6(b) and 6(c), in which the ADS specimens absorbed more energy per unit bend deflection with respect to the laminate trimmed with the other two techniques. Figure 6 suggests i :.... 'I, I0 o 2 4 ~ a I0 Average Rous ~m) Average Rouglme~s t~tm) i) b) c) Fig. 5--Influence of surface roughness on load and energy to fracture: (a) Gr/Ep, 2.25 m/s impact; (b) Gr/Bmi, 2.25 m/s impact; (c) Gr/Bmi, 3.75 m/s impact that although surface texture and/or manufacturing defects accelerate the onset of failure, they do not necessarily affect the progression of the failure process. However, this behavior is obviously stacking sequence dependent as noted from comparing the postinitiation energy to fracture of the PCDtrimmed Gr/Ep and Gr/Bmi laminates. A survey of the dynamic images recorded with the highspeed camera during fracture did not indicate significant differences in failure initiation according to the method of trimming. The onset of fracture under impact loading occurred via bending-induced tension and comprised the first few outer plies depending on the laminate. Failure of the Gr/Bmi originated in the first 90-deg ply in the stacking sequence, and failure of the Gr/Ep was initiated at the boundary of the 45- deg/-45-deg plies. Figure 7(a) shows a typical dynamic Experimental Mechanics 9 383

6 o~ a ~ C ; v I a) 1 I I I 6 8 I0 12 deflection (ram) b) I I I I 8 I deflection (ram) E' E z v toooo I 500O 4 e) I I I I $ I deflection (ram) Fig. 6---Impact energy over the bend deflection: (a) Gr/Ep, 2.25 m/s impact; (b) Gr/Bmi, 2.25 m/s impact; (c) Gr/Bmi, 3.75 m/s impact record of failure initiation and progression for a Gr/Ep specimen machined with the AWJ. An equivalent record for the onset of failure of the Gr/Bmi laminate trimmed with singlepoint cutting tools is shown in Fig. 7(b). Contrary to the dynamic records, an inspection of the postfractured specimens provided some justification to the suspect performance of the PCD-trimmed Gr/Ep. In contrast to the confined fracture of the ADS- and AWJ-machined laminate, extensive interlaminar delamination occurred in the PCD-trimmed specimens, emanating from beneath the tup and extending from 15 mm to an excess of 25 mm in length along the -45-deg plies. The "fracture zone" of the Gr/Bmi laminate remained localized near the tup regardless of the method used for machining and did not extend along the damaged plies of the PCD-trimmed laminate. Hence, the decrease in fiber/matrix cohesion of the Gr/Ep resulting from machining with single-point cutring tools coupled with the free edge stresses resulted in an increase in the energy absorbed under the dynamic bending load. However, the Gr/Bmi specimens did not exhibit the same characteristic dynamic response. The difference in performance between the Gr/Bmi and Gr/Ep trimmed with single-point cutting tools highlights the importance of considering the lamina-stacking sequence and free edge stresses with the process-dependent surface integrity. By dissipating the tup energy through delamination along the ply direction as surface energy, the dynamic crack growth propagating perpendicular to the laminate thickness was temporarily suppressed. Therefore, the poor interfacial integrity and surface flaws imposed by PCD trimming increased the ability of the Gr/Ep laminate to absorb postfracture initiation tup energy. No one has yet documented this phenomenon in an assessment of polymeric composites subjected to dynamic loads. The reduction in performance of the AWJ-machined laminates with surface roughness cannot be explained with the analogy used for the PCD-trimmed specimens. From the microscopic analysis and micrographs in Figs. 3(c) and 3(d), there was no evident changes in interfacial integrity with the three different AWJ machining conditions. This implies that the reduction in impact strength with surface roughness was attributed solely to the geometric stress concentrations resulting from machining and their influence on the magnitude of the stresses near the free edge. The reduction in load, deflection and energy to failure with increasing surface roughness occurred despite differences in the constituents and lamina-stacking sequence, unlike the response from the PCD-trimmed laminates. Therefore, the influence of postmold manufacturing on the performance of FRPs under dynamic loads is dependent on the surface roughness, constituent integrity, location of significant manufacturing defects and lamina-stacking sequence. Our future work will concentrate on the development of analytical tools to incorporate manufacturing effects and machined surface integrity in the design of FRP structures. Conclusions A graphite/epoxy (Gr/Ep) and a graphite/bismaleimide (Gr/Bmi) laminate were machined using an abrasive diamond saw (ADS), polycrystalline diamond (PCD) orthogonal cutting tools and the abrasive waterjet (AWJ). The surface integrity of the trimmed edges received by each method of machining was assessed using profilometry and electron microscopy. Following the evaluation, trimmed laminates were subjected to three-point bend impact to failure. From a comparison of the machined edge quality, and its effect on the material performance under impact loading, the following conclusions were made: 1. Both of the laminates trimmed with the ADS exhibited an average surface roughness (Ra) of 0.2 Hm, which VoL 37, No. 4, December 1997

7 machined surface texture on the dynamic strength of the Gr/Bmi laminate increased with impact velocity. a). The impact energy absorbed by the PCD-trimmed Gr/Ep specimens beyond the onset of fracture exceeded that of the laminates machined using the ADS and AWL The source of this phenomena stems from the dynamic crack growth perpendicular to the plane of the applied load along the damaged -45-deg plies of the laminate. The absorbed impact energy of the PCDtrimmed Gr/Bmi was equivalent to that resulting from the laminates trimmed using the other two techniques. The difference in response between the two laminates is due to the difference in ply-stacking sequence of the laminates and subsequent free edge stresses. Hence, the influence of net shape machining on the performance of FRPs under dynamic loads is dependent on the surface roughness, constituent integrity, location of significant manufacturing defects and lamina-stacking sequence. Acknowledgments The authors wish to thank the Boeing Company for its generous donation of materials and Albert S. Kobayashi for his gracious support and encouragement throughout this study. References b) Fig. 7--Typical dynamic records of failure initiation and progression (2x): (a) Gr/Ep, (b) Gr/Bmi. 3. was the lowest of the three techniques used. Trimming with single-point cutting tools provided an Ra of 1.4 ~tm, which was superior to that received from AWJ machining. The highest peak load, bend deflection, and energy to the onset of fracture of the two FRP laminates resulted from specimens obtained with the ADS. The laminates trimmed with PCD single-point cutting tools had the poorest mechanical performance under dynamic loading in terms of both the load and energy to fracture. Reductions in the load to the onset of fracture were in excess of 20 percent lower than the corresponding values from laminates processed using the ADS. The corresponding decrease in the energy to fracture ranged from 10 percent to 35 percent. The peak load, deflection and energy to the onset of fracture clearly decreased with increasing surface roughness. The maximum reductions in fracture load and energy to failure of the AWJ-machined laminates were 13 percent and 35 percent, both of which resulted from specimens with an average surface roughness of approximately 10/am. In addition, the influence of 1. Abrate, S. and Walton, D.A., "Machining of Composite Materials. Part h Traditional Methods," Composites Manufact., 3 (2), (1992). 2. Abrate, S. and Walton, D.A., "Machining of Composite Materials. Part II: Non-traditional Methods," Composites Manufact., 3 (2), (1992). 3. Ramulu, M. andarola, D., "Traditional andnon-traditionai Machining of Fiber Reinforced Plastic Composites," Proc. 39th SAMPE Syrup., 39 (1), (1994). 4. Colligan, K. and Ramulu, M., "Edge Trimming of Graphite/Epoxy with Diamond Abrasive Cutters," Syrup. Machining Adv. Composites, ASME Bound Volume, PED-Vol (1993). 5. Wang, D.H., Ramulu, M., and Arola, D., "Orthogonal Cutting Mechanisms of Graphite/Epoxy Composite. Part h Unidirectional Laminate," Int. J. Mach. Tool Manufact., 35 (12), (1995). 6. Wang, D.H., Ramulu, M., and Arola, D., "Orthogonal Cutting Mechanisms of Graphite~Epoxy Composite. Part ll: Multi-directional Laminate," Int. J. Mach. Tool Manufact., 35 (12), (1995). 7. Ramulu, M, and Arola, D., "Waterjet and Abrasive Waterjet Cutting of Unidirectional Graphite~Epoxy Composite," Composites, 24 (4), (1993). 8. Ramulu, M. and Arola, D., "The Influence of Abrasive Waterjet Cutting Conditions on the Surface Quality of Graphite~Epoxy Laminates," Int. J. Mach. Tool Manufact., 34 (3), (1994). 9. Colligan, K., "Edge Machining Effects on the Compressive Strength of Graphite~Epoxy Composite," MS thesis, University of Washington (1993). 10. Garret, R.A., "'Effect of Manufacturing Defects and Service lnduced Damage on the Strength of Aircraft Composite Structures," Composite Materials: Testing and Design, ASTM STP 893, 5-33 (1986). 11. Howarth, S.G. and Strong, A.B., "Edge Effects with Waterjet and Laser Beam Cutting of Advanced Composite Materials," Proc. 35th Int. SAMPE Syrup., (1990). 12. Tagliaferri, V., Caprino, G., and DiterlizzL A., "Effect of Drilling Parameters on the Finish and Mechanical Properties of GFRP Composites," Int. J. Mach. Tool Manufact., 30 (1), (1990). 13. Muller de Almeida, S. E and Candido, G.M., "Effect of the Free Edge Finishing on the Tensile Strength of Carbon/Epoxy Laminates," Composite Struct., 25 (1), (1993). 14. Arola, D. and Ramulu, M., "Machining Induced Surface Texture Effects on the Flexural Properties of a Graphite~Epoxy Laminate,'" Composites, 25 (8), (1994). 15. ASTM D790M, "Standard Test Methods for Flexural Properties of Unreinforced and Reinforced Plastics and Electrical Insulating Materials," American Society for Testing and Materials, Philadelphia, PA (1994). Experimental Mechanics 9 385