The high costs of continuous-fiber composites have driven the development of discontinuous-whisker and even particulate-reinforced composites. The potential performance of these discontinuous composites is much poorer, and the mechanism of strengthening may be only partially by fiber reinforcement with the dispersion strengthening being the major effect. Yet silicon-carbide-reinforced aluminum has a significantly higher modulus and strength, especially at higher temperatures. Although the normal compressive forming processes can shape the composite, the lower ductility and fracture toughness have limited its general application (Ref 36). It is competitive for applications such as snow tire studs, which should be wear and corrosion resistant, as well as low cost. Properties of Metal-Matrix Composites Metal-matrix composites offer significantly better mechanical properties than polymer-matrix composites for matrix- dominated properties, such as greater shear, compressive, and transverse tensile strengths. Reinforcement can increase the maximum-use temperature over that of the monolithic material. However, matrix behavior may not be as good as for the unreinforced matrix. For example, the transverse tensile strength of a metal-matrix composite, modeled by a matrix with poorly bonded fibers, is only about one-third of the strength of the unreinforced matrix. Furthermore, the microstructure of the matrix may not be as desirable, and together with the physical constraints placed on the fibers by the matrix, ductility of the matrix is generally much reduced. Metal-matrix composites also can have good electrical and thermal conductivity. The stress/strain behavior of metal-matrix composites is more complicated than resin-matrix composites because of work hardening and the change in the yield surface with different or multiple loading (Ref 37). (The same considerations make thermal expansion show hysteresis that changes with cycling.) If a residual stress-free metal-matrix composite is loaded parallel to the fiber axis, both the fiber and matrix will be elastically loaded initially (Fig. 13). Upon further loading, the matrix will finally be loaded to its yield stress and then plastically deform, but the fibers will still be loaded elastically. The effective modulus is decreased in this region. Then, substantial fiber failures occur and composites with high fiber volume fraction fail catastrophically. At low fiber volume fractions, fibers break up into critical lengths causing a substantial loss in modulus, and finally matrix failure occurs. These effects have been successfully included in describing the behavior of metal-matrix composites, but design of stable structures that are subjected to temperature cycles is more difficult. The details are beyond the scope of this article. Fig. 13 Stress/strain curve for nicalon silicon carbide fiber in aluminum (1100) matrix. The material has an initial modulus (E 1 ) of 87 GPa, which is representative of both fiber and matrix elastically deforming. The secondary modulus (E 2 ) of 70 GPa is indicative of fiber elastic deformation and matrix micro- yielding. For a relatively low- ductility matrix, failure often occurs at the end of the secondary modulus straight line, when the first fibers begin to fail. For the higher-ductility 1100 aluminum matrix, fiber fractures acc umulate during the curved part of the stress/strain plot until final failure occurs. Source: Ref 38 The plastic deformation that can occur in a metal-matrix composite is more apparent for off-axis and transverse loading to the fiber axis (Fig. 14). Loading at 30° produces a stress/strain curve that looks much the same as for the pure matrix except that the strain scale is reduced. The reason is that local matrix deformations around the fibers are much higher than the overall global strains, and failure finally occurs when local strains are similar to the failure strain in the unreinforced matrix. (Triaxial tensile strains are also present in the matrix, which also tends to reduce ductility.) Fig. 14 Effect of loading direction on uniaxial boron fiber/aluminum (7075). Results are for fibers oriented at 0, 30, and 90° with respect to the load and pure matrix in the T6 and O conditions. Source: Ref 39 Fatigue properties in the axial direction of the fibers is excellent, but fatigue for off-axis loading, which relies on the fatigue behavior of the matrix, can be worse than for the unreinforced matrix. The accumulation of failed fibers and matrix cracks in fatigue testing results in loss of modulus. This may continue until composite failure occurs. However, in some laminate constructions or composites with low fiber volume fraction, damage may develop to a stable condition (shakedown), whereupon further fiber or matrix damage occurs slowly. The drop in modulus may be as great as a factor of two during shakedown, which may require part removal because of reduced stiffness. References cited in this section 31. M. McLean, Directionally Sloughed Materials for High Temperature Service, The Metals Society, London, 1983 32. A.R. Champion, W.H. Kreuger, H.S. Hartman, and A.K. Dhringra, Proceedings of the 1978 International Conference on Composite Materials (ICCM2), TMS-AIME, 1978, p 883 33. T. Donomoto, N. Miura, K. Funitani, and N.Miyake, "Ceramic Fiber Reinforced Piston for High Performance Diesel Engine," Publication 83052, Society of Automotive Engineers, 1983 34. P.R. Smith and F.H. Froes, J. Met., Vol 36, 1984, p 19 35. P.A. Selmers, M.R. Jackson, R.L. Mehan, and J.R. Rairden, Production of Composite Structures by Low- Pressure Plasma Deposition, Ceram. Eng. Sci. Proc., Vol 6, 1985, p 896 36. W.A. Logsdon and P.K. Liaw, Eng. Fract. Mech., Vol 24, 1986, p 737 37. Y.A. Bahei-El- Din and G.J. Dvorak, Plastic Deformation Behavior of Fibrous Composite Materials, Proceedings of the 4th Japan U.S. Conference on Composite Materials, Technomic Publishing, 1989, p 118 38. J. Tanaka, H. Ishikawa, T. Hayase, K. Okamura, and T. Matsuzawa, Mechanical Properties of SiC Fiber Reinforced Al Composites, Progress in Science and Engineering of Composites, ICCM- IV, Japan Society of Composite Materials, Tokyo, 1982, p 1410 39. G.D. Swanson and J.R. Hancock, Off- Axis and Transverse Tensile Properties of Boron Reinforced Aluminum Alloys, Composite Materials: Testing and Design, STP 497, ASTM, 1971, p 472 Design with Composites R.J. Diefendorf, Clemson University Ceramic-Matrix Composites The demand for composites that can operate at higher temperature has driven the development of ceramic-matrix composites. Tremendous barriers stand in the path for successful application of these composites in which the properties of the fiber and matrix may be very similar, and indeed the chemical composition of both may be identical. The lack of cost-effective and creep-resistant fibers, stable crack-stopping interfacial coatings, and high strain-to-failure matrices all lead to poor-extended-life, oxidation-prone, high-temperature composites. Yet, the successful application of car-bon- fiber/carbon-matrix and silicon-carbide/silicon-carbide composites to rocket nozzles and reentry vehicles illustrates the potential of these materials if environmental stability can be achieved. While ceramic composites might be used at lower temperatures because of their corrosion resistance, they are mainly considered for high and extreme temperatures often as the last resort. The mechanical behavior of these composites, which have similar constituent moduli, is quite different from resin-matrix composites, especially because the strain to failure of the matrix is typically smaller than that for the fiber. The fibers are added not only to provide high-temperature creep resistance, but more importantly to provide damage tolerance and a more graceful failure at room temperature. When a ceramic-matrix composite with uniaxially oriented fibers is loaded in tension parallel to the fibers, the load is relatively evenly distributed between the fiber and matrix, the exact ratio being determined by the relative moduli and volume fractions. For relatively strong coupling between the fiber and the matrix, the composite fractures when the first crack occurs in the matrix (Fig. 15a). Often, there may already be surface cracks from specimen preparation. The composite strength may be lower than a monolithic specimen of pure matrix material, as it frequently is not possible to process the matrix to as high a quality with fibers present. The strength and toughness of commercially available composite products will generally be higher than for monolithic materials. Better processing eliminates many of the defects by filling in voids with more and better matrix material. Matrices with higher fracture toughness would help increase the strain at which first cracking occurs, as would lower modulus. However, the fracture toughness decreases as the refractoriness (bond strength) of the material increases. Low matrix fracture toughness is the likely result for the highest temperature applications. However, combining several techniques to improve fracture toughness, such as adding whiskers to the matrix and placing multiple crack-stopping interfaces within the matrix can increase the strain for first matrix failure. Fig. 15 Illustration of ceramic- matrix composite failure process. (a) The crack, which initiates in the matrix, propagates through both matrix and fiber s in the composite when strong interfacial bonding exists. (b) For intermediate or low interfacial bonding, the matrix crack runs through the matrix but around the fibers. Multiple cracks accumulate if the fibers bridging the cracks can sustain the load. ( c) After attaining multiple matrix cracks with an equilibrium spacing, fibers fail at flaws with increasing load, and not necessarily in the bridged regions of highest stress. (d) Fibers pull out of their matrix sockets with further extension. Decreasing the coupling between the fiber and matrix sufficiently allows the first crack in the matrix to deflect around the fibers such that the fibers will bridge the cracked matrix and sustain the load if the fiber strength and volume fraction are high enough (Fig. 15b). As the load is increased, more matrix cracks form and finally yield a specimen with a relatively uniformly spaced set of cracks in the matrix, all bridged by the fibers. The spacing of the cracks is determined by the fiber/matrix coupling: the poorer the coupling, the wider the spacing. Finally, there is insufficient length between cracks to transfer the load, which only the fibers are carrying in the bridging regions, to the matrix to cause further cracking. A large drop ( 50%) in modulus can occur with poor interfacial bonding when the matrix cracks. The design may be limited by this decrease in modulus or by resonant frequency changes. Further loading causes failure of fibers, and finally at one bridging section the remaining fibers can no longer support the load and the specimen fails. The probability for failure of the fibers is highest in the bridges, because the matrix carries no load. However, there is a distribution of strength-reducing flaws along the length of the fibers, and there will be fracture of fibers away from the bridged matrix crack, albeit with a decreasing frequency with distance from the crack (as the matrix picks up load) (Fig. 15c). The last step of the process is the pullout of the fractured fibers from their sockets in the matrix (Fig. 15d). Much energy can be absorbed, as the apparent strain can be quite high. Studies have shown that this pseudometallic, stress-strain behavior can be achieved in ceramic composites. Constituent Selection Ceramic-matrix composites are likely to be used at high temperature and are usually fabricated at high temperature. In addition to the usual considerations of physical properties such as specific modulus and strength, creep and stress rupture become the limiting mechanical properties for high-temperature applications (Ref 40). The most creep-resistant ceramic is graphite, followed by silicon carbide, titanium diboride, and silicon nitride. Yttrium-aluminum-garnet (YAG), mullite, and sapphire appear to be the most attractive oxides (Ref 41). Both fibers and matrices must be creep resistant. For lower- temperature applications, glass-matrix or glass/ceramic-matrix provide a combination of thermal expansion coefficient, fracture toughness, and low modulus that allow very good composites to be produced. The selection of fiber and matrix material combinations is very much limited by thermochemical and thermomechanical incompatibilities. For example, carbon fibers in any oxide matrix are thermodynamically unstable above 1500 °C. Similarly, a mismatch of thermal expansion coefficients much above 2 m/m · °C causes microcracking of the composite. The problem of selection of a ceramic composite system is compounded because a weak interface is necessary to control the fracture. A third interfacial phase between the fiber and matrix is often added, especially when the matrix and fiber have the same chemical composition, such as with a silicon-carbide-fiber/silicon-carbide-matrix composite. The interfacial materials generally have a layer structure, although highly porous interfacial phases have also been used. However, no environmentally stable layers have been found for the higher-temperature composites (>1200 °C). Model systems usually have used graphite. Boron nitride, which has some oxidation resistance to 1100 °C, has been found to be an effective crack-stopping interfacial material also. Oxide-layer compounds, such as synthetic micas, are useful for temperatures as high as 1100 °C. However, the interface between fiber and matrix remains a problem, and present interfacial coatings add substantially to the composite cost. Ceramic-matrix composites are not as well developed as polymer-matrix composites, with the exception of carbon- fiber/carbon-matrix composites. Carbon/carbon composites are used extensively for aircraft brakes, shuttle tiles, rocket motor nozzles, and reentry vehicles. For other ceramic-matrix composites, the designer must select one of the few systems that are available. Much stronger interaction with the materials supplier is required, because the processing and properties are very much producer dependent. Properties of Ceramic-Matrix Composites The elastic properties can be calculated using the equations described in the sections "Calculation of Lamina Properties" and "Symmetric In-Plane and Through-Thickness Laminates" in this article. The major difference with ceramic-matrix composites is that the moduli of the fiber and matrix are frequently similar, so that the elastic properties are more isotropic. The fracture behavior of ceramic-matrix composites differs from that of polymer- or metal-matrix composites in that the failure strain of the matrix is less than that for the reinforcement. The stress/strain curve for a uniaxially aligned fiber specimen, loaded parallel to the fibers is shown in Fig. 16 (Ref 42). Both matrix and fibers are being loaded initially, until the failure strain of the matrix is reached, and the matrix starts fracturing. There are three different limiting behaviors depending on the interfacial bonding: (1) high interfacial bonding causes catastrophic failure, (2) intermediate bonding causes a change in slope, (3) poor bonding causes a load drop with a subsequent slope proportional to just the modulus and volume fraction of the fibers. For low or intermediate interfacial bonding and a volume fraction and fiber strength sufficiently high to support the applied load after matrix fracture, composite fracture does not occur, and multiple matrix cracks will accumulate. In the limit, as bond strength goes to zero, the ceramic matrix behaves as if it has holes. These "holes" act as stress raisers. The final crack spacing is determined by the strength of the matrix and inversely on the maximum shear stress that can be sustained at the fiber/matrix interface. The relationships for multiple matrix cracking, and the final matrix crack spacing are (Ref 43): CMu < Fu V F (Eq 12) L = V M Mu R/V F (Eq 13) where CMu is the composite stress when first matrix cracking occurs; Mu and Fu are the matrix and fiber ultimate strengths, respectively; R is the radius of the fiber; is the shear stress at the fiber/matrix interface; and L is the spacing between matrix cracks. Fig. 16 Stress/strain curve for a typical uniaxial ceramic- matrix composite loaded parallel to the fibers. The solid line (A) shows the behavior for strong interfacial bonding and catastrophic failure with the first matrix crack. The dotted line (B) indicates intermediate bonding behavior such that the fibers bridge the matrix cracks, but with short fiber/matrix debonding near the cracks. The dashed line (C) illustrates the stress/strain curve f or very weakly coupled fiber and matrix. In the limit, the matrix contribution to the modulus is completely lost after matrix fracture. In ceramic-matrix composite systems that retain high frictional or chemical bonding interaction between the fiber and the matrix after matrix fracture, a simple change in slope occurs in the stress/strain curve as the matrix begins fracturing (Fig. 16). In other systems, there is loss of the matrix stiffness if coupling between fiber and matrix is lost when matrix fracture occurs. The stress/strain curve under displacement control may show a load drop much like yielding, or become horizontal, until the load increases again with strain, but this time only with the modulus contribution from the fibers. (The actual curve may differ because of residual stresses between the fiber and matrix.) The microcracking of the matrix not only results in loss of modulus, but also allows internal oxidation to occur. Parts could be designed to stress or strain levels below which matrix cracking occurs. While it might be expected that the matrix would crack at similar strains as for the unreinforced ceramic matrix (0.05 to 0.10%), the strain for matrix cracking was shown to be enhanced (Ref 43): Mu = [24 O M /E C D F (1 - V F ] 1/3 + E F V F T/E C (Eq 14) where Mu is the reinforced matrix strain at failure, O is the interfacial shear strength, M is the matrix fracture energy, D F is the fiber diameter, is the difference in thermal expansion coefficient between fiber and matrix, and T is the difference between the stress-free temperature and the use temperature. The stress-free temperature is often assumed to be the processing temperature. Although the fractional exponent on the first term minimizes the effect of the parameters, the wide range that some of the parameters can have produces significant changes in the strain. High matrix fracture energy, high fiber modulus and volume fraction, small fiber diameter, and low matrix modulus all increase the matrix failure strain. Increasing the interfacial shear strength also raises the matrix failure strain, but the value must not exceed that which causes brittle failure. One approach, which has doubled the matrix microcracking strain, is reinforcing the matrix with about 15% of fine whiskers. The second term in the equation is the residual stress that arises from the mismatch in the coefficients of thermal expansion between the fiber and matrix. An axial compressive residual stress would be present in the matrix at room temperature if the fiber thermal expansion coefficient is larger than that for the matrix, because the composite is generally processed at high temperature. Unfortunately, the thermal expansion coefficient of the fiber is likely to be smaller than the matrix, placing the matrix in residual axial tension. The average matrix strain-to-failure in a composite can frequently be increased to values of 0.4% or more, which combined with the 1% or better strain to failure of the fiber, can produce a stress/strain curve mimicking a ductile metal, albeit with very limited strain capability. A problem is that the first few matrix failures are observed at strains only slightly higher than the unreinforced matrix. Therefore, a prudent assumption is to assume that matrix microcracks will always be present that may allow internal oxidation and embrittlement. The ultimate tensile strength of uniaxially aligned composite, loaded parallel to the fibers, is given by the bundle strength and volume fraction of the fibers bridging the matrix cracks: Cu Fu V F (Eq 15) where Cu and Fu are the ultimate tensile strengths of the composite and fiber bundles, respectively. The bundle strength of the fibers cannot be measured on a gage length that approximates the short gage length in the cracked matrix region of the composite, but might be approximated from resin- bonded strand tests. Because fiber volume fractions often are about 50% in a uniaxially aligned ceramic composite, the axial ultimate tensile strength will be about one-half of the fiber bundle strength, if the interfacial coupling has been properly adjusted to prevent brittle fracture. If a fabric is used, somewhat less than 25% of the tensile strength of a bundle can be obtained because of the over/under construction in a fabric and the effective fiber volume fraction in the load direction. The toughness of a ceramic-matrix composite can be caused by at least five different mechanisms, which can act independently or in a combined manner. However, the work to fracture, the area under the stress/strain curve, is generally dominated by the "pull out" of broken fibers from their matrix sockets. Although it is generally desirable to have little scatter (high Weibull modulus) in the strengths of fibers in order to attain the highest composite, ultimate tensile strength, no pullout occurs when there is no variation in fiber strength. All fibers would fail simultaneously in the highest-loaded regions at the matrix cracks. Pullout requires that fibers fail at flaws away from the matrix cracks, and then the broken fibers drawn out from the matrix sockets. The work of pullout (W P ) for a multiple matrix cracked composite (in ft · lbf) is (Ref 44): W P = 0.083 [(m - 1) (m + 2)/m 2 (m + 1) 2/(m + 1) ] V F (0.5D F ) 3m/(m + 1) [ / (m - 1)/(m + 1) ] (Eq 16) where m is the Weibull modulus, O is the Weibull strength scale parameter, D F is the fiber diameter, and is the interfacial shear strength. For m = 8, typical for commercially produced fibers, the equation reduces to: W P = 0.008 V F [ / 0.8 ] (Eq 17) The optimal Weibull modulus for maximizing work of fracture is about 4. (There will probably not be any development of fibers optimized for maximum work of pullout in the near future due to the limited market.) A high volume fraction of large-diameter fibers with high strength, but with a low interfacial strength produces a maximum work of pullout. A comparison with the matrix microcrack strain equation shows that high fiber volume fraction and strength are beneficial to both. However, the effects for fiber diameter and interfacial shear strength are opposite. A ceramic composite cannot be optimized for both the work of pullout and the matrix microcrack strain simultaneously. The strength of a uniaxial ceramic-matrix composite that is loaded off-axis is low. This result is a consequence of the poor interfacial shear strength between the fiber and matrix that is required to control fracture. Multiple-ply laminates with off- axis oriented plies must be used to sustain off-axis loads. The materials selected depend very much on the application. For inert atmosphere or very short time applications (tens of minutes), unprotected carbon fibers in a carbon matrix provide the best mechanical properties, especially at very high temperature. Creep is low until 2200 °C. Coated carbon/carbon composites with internal additives can have lifetimes in oxidizing environments up to 1700 °C for as long as 100 h, but not very reproducibly. Silicon carbide or nitride provide much better oxidation resistance, but would probably be creep limited at temperatures above 1500 °C for the carbide and somewhat lower for the nitride, even anticipating future improvements. At present, oxide systems appear limited to 1200 °C and up to 1300 °C in the future. The exact values all depend on stresses, time, and temperature, and the development of suitable interfacial materials. References cited in this section 40. A. Kelly, Design of a Possible Microstructure for High Temperature Service, Ceram. Trans., Vol 57, 1995, p 117 41. W.B. Hillig, A Methodology for Estimating the Mechanical Properties of Oxides at High Temperatures, J. Am. Ceram. Soc., Vol 76 (No. 1), 1993, p 129 42. A.G. Evans and F.W. Zok, The Physics and Mechanics of Fibre Reinforced Brittle Matrix Composites, J. Mater. Sci., Vol 29, 1994, p 3857 43. J. Aveston, G.A. Cooper, and A. Kelly, Single and Multiple Fracture, Properties of Fiber Composites, IPC Science and Technology, 1971, p 15 44. M. Sutco, Weibull Statistics Applied to Fiber Failure in Ceramic Composites and Work of Fracture, Acta Metall., Vol 37 (No. 2), 1989, p 651 Design with Composites R.J. Diefendorf, Clemson University Large Composite Structures: Joints, Connections, Cutting, and Repair A major advantage of composites is that large integrated structures can often be designed with a major reduction in parts count and assembly time. The number of mechanical fasteners is often a small fraction of the number used in metal structures. While it is sometimes possible to design a large composite in which loads are predominantly carried by the fibers, usually there will be regions that rely only on the matrix for load transfer from one major volume to another. Several parts may be cocured together so that the matrix resin acts as the adhesive. This area behaves much as a bonded joint. Mechanical fasteners have been added to increase the reliability, but surprisingly performance has sometimes been decreased because of the stress concentrations and delamination damage at the free edge of holes. Extensive use of mechanical fasteners is costly with carbon-fiber/resin composites, because the fasteners are generally made of titanium to minimize galvanic corrosion; use of coated fasteners would avoid corrosion and lower cost. Stitching has also been applied successfully using a tough fiber such as an aramid as the stitching yarn, as a method of improving interlaminar properties and minimizing edge delamination. Stitching pierces the laminate with a minimum of laminate fiber damage and displacement. Optimization of stitch spacing and yarn denier and tension has been performed for a number of different laminates and structures. While composite materials are especially amenable for making large integrated structures, it should always be realized that damage is likely to occur during the lifetime of the structure. The structure must be designed for easy and efficient repair. Consideration of where cuts should be for removal of damaged material, and flanges for attachment or bonding of repaired sections, and so forth, is required. Composite materials also offer the possibility for in situ structural-integrity monitoring by incorporating sensing fibers within the composite. A number of studies have shown that optical and strain-sensitive fibers can be used to measure loads and structural integrity. However, these techniques have yet to achieve wide application. Variance and Scaling Considerations Composites traditionally have used brittle fibers, because of their otherwise superior properties. For simplicity, consider a composite with the fibers all aligned parallel to the load direction. While the most probable state of the composite would be to have the worst flaws in each fiber randomly distributed along the fiber length, there is a small chance that all the worst flaws are located at the same distance along each fiber. (If there are periodic processing-induced defects, the odds become worse that the flaws correlate.) The question is how does the variability, as measured by the Weibull parameter, of composites compare to that of engineering metals? For well-developed composite systems, made in simple, flat laminates, the Weibull parameter for small specimens can be as high (20 to 24) as for aluminum sheet in the rolling direction, and even better than aluminum in the transverse direction. However, strengths of longer laminates have often been lower than predicted by simple Weibull scaling (Ref 45). There are two problems. Firstly, a large composite laminate may not be made in the same way as the smaller test laminates and more and worse flaws are introduced. Secondly, the small test specimens do not sample the worst flaws very well, which actually cause failure of the long laminate. By contrast, Weibull scaling of strength, when the width is increased, underpredicts the strength of the composite. (The strength does not drop off as fast as predicted.) The designer must be aware of the effect of scale on strength. Additional discussion of Weibull statistics is provided in the article "Design with Brittle Materials" in this Volume. Load, Heat, and Electrical Current Transfer The large difference in properties that often exist between fiber and matrix make transfer of load, heat, and electricity to a composite part very inefficient. The problem is easily illustrated in resin-matrix composites, because of the low modulus, and low thermal and electrical conductivity of most polymers. Most of the load, heat, or electrical current is usually transferred to a composite through the resin to the fibers from the sides and not directly to the fibers from the end (Fig. 17). In the case of mechanical loads, the load is transferred through shear stresses at the surface, and a tensile stress builds up in the composite from the end. A gradient in tensile stress also is present through the thickness, but disappears with sufficient length as the load is diffused throughout the composite. The distance along the composite that is required to reach approximately a constant stress is much longer in a composite than in a metal. While the stress can be considered constant in an isotropic metal within approximately two thicknesses for a sheet, the value for composites can vary from 6 to 40, the exact value depending on the composite architecture and materials. Hence, a designer must be more careful in transferring load to a composite to prevent overloading outer fibers. Load-transfer distances are much greater than in metals, and joints are less efficient. Fig. 17 Transfer of load into a uniaxial composite. The transfer of load through shear produces high stresses in the outer ply fibers at the grips. The distance along the composite that it takes for the load to diffuse throughout the cross section is much greater than in a metal. The stress profiles through the thickness are illustrated at several positions along the length of the composites Fiber-reinforced composites have also been considered for heat-transfer applications, because of the development of very high thermal conductivity carbon fibers (five times the thermal conductivity of copper at room temperature). The thermal conductivity is at its maximum value somewhat below room temperature and decreases to low values near 0 K. Above the maxima, the thermal conductivity decreases approximately inversely with temperature. Hence, the very high thermal conductivity can only be achieved in near-room-temperature applications. These high thermal conductivity carbon fibers have been used in resin, metal (particularly aluminum), and carbon matrices. Flash laser thermal diffusivity experiments have shown that high thermal conductivity can be achieved in resin-matrix composites, when the laser beam directly impinges on the fiber ends in a uniaxial composite. However, the effective thermal conductivity in many real designs is much lower, because the heat must be transferred through the resin matrix, and the heat flow never becomes uniform through the cross section of the part. In the corresponding case for electrical conduction, uniform current flow is even more difficult to achieve. Adding an electrically or thermally conducting powder to the matrix resin can help alleviate the problem. Metal-matrix composites allow better transfer of heat to the high thermal conductivity fibers, provided good fiber/matrix interfacial thermal contact can be maintained. Although aluminum can galvanically corrode with carbon fibers, the low density and high thermal conductivity makes this combination attractive especially for space and aeronautical applications. Protective coatings are available for carbon fibers to minimize corrosion. Joint Design Only one-half of the potential weight savings is often achieved when using composites because of the necessity of joints. The inherently inefficient load transfer into most composites, especially polymer-matrix composites, means that joint design is often a driving factor in composite design. Frequently, sections are thick and load paths complicated, which requires careful, and three-dimensional, analysis. Design of a joint is often more difficult than that of the rest of the composite part. A primary decision is whether to use bonded or mechanically fastened joints. Studies have shown that bonded joints can be more efficient than mechanically fastened joints. However, the questionable reliability and longevity of these bonded joints, especially if off-axis peeling stresses are present, have often resulted in the use of mechanical fasteners. Mechanical fasteners must be used if easy disassembly or access is required. Mechanically Fastened Joints. The behavior of mechanically fastened joints is similar to but more complicated than the behavior of similar joints in metal. Not only is the strength dependent on geometry as with a metal, but also on the fiber orientations. Numerous composite joint configurations have been tested and evaluated. Only the major conclusions are presented. The important factors affecting joint strength are joint type, fastener, geometry, and failure mode. Several types of mechanical joints have been used with composite materials. The choice of joint type depends on the application. Single- or double-lap designs with rivet (pinning) or bolting are frequently used. Riveted joints often provide adequate strength in carbon- or glass-fiber- reinforced resin-matrix composites, but bolts offer the greatest strength. Higher-bearing-strength bolted joints result if the clamping pressure is increased to an optimum by using a higher bolt tightening torque. Additionally, washers, and a carefully reamed hole that closely fits the bolt, produce higher-strength joints. Mechanically fastened joints in composites display the same failure modes as observed in metals: net-section tension, edge-section shear, end-section bearing, and combined modes (Fig. 18). Geometrical factors as well as fiber orientations determine the failure mode. The geometrical factors of width (pitch), end distance, and diameter influence the behavior of composite joints, just as they do in metals (Ref 47, 48) (Fig. 19). End distance is defined as the distance from the end of the joint to the center of the closest hole, and width is the distance from the sides of the joint to the center of the nearest hole. End distance and width must be above a certain minimum if full bearing strength is to be obtained. Below these minima, tensile failure occurs if the width is too small, and shear-out or cleavage if the end distance is too small (Fig. 18). While all materials follow the same trend, the full strength occurs at differing w/d and e/d ratios for different materials and stacking sequences. However, values of w/d and e/d of 5 appear adequate to achieve a bearing failure in most resin- matrix composites. Adding fabric plies for reinforcement around holes is a very efficient way to reduce w/d and e/d values to below 5, although the thickness of the composite will be increased in the local region.The hole diameter compared to the laminate thickness, d/t, has a negligible effect if the hole is optimally clamped, but should be below 1 if not. [...]... eliminating variability and randomness from the process, and on making nonvalue-added operations such as orienting and handling as simple and easy to perform as possible Many product design departments now use design for assembly techniques to improve the ease with which products are assembled Design for assembly, which is discussed in detail in the article "Design for Manufacture and Assembly," which... 1996 6 M Andreasen, S Kahler, and T Lund, Design for Assembly, IFS Publishing, 1983 Introduction to Manufacturing and Design Henry W Stoll, Northwestern University Interaction between Design and Manufacturing The design of a product and its method of manufacture are intimately connected This interdependence cuts across all aspects of the manufacturing system and exists at the part, assembly, and system... on the interaction between design and specific manufacturing processes Still others, such as failure mode and effects analysis and value engineering, can be used at all levels and in all contexts to improve both product design and the method of manufacture Related coverage is provided in the Sections "The Design Process" and "Criteria and Concepts in Design" in this Volume Design for life-cycle manufacturing... within constraints imposed by other design requirements This is done by first reducing the number of parts and then ensuring that the remaining parts are easy to assemble Design for "X" (DFX) Methods A variety of "design for" methods and approaches are being developed and used to improve the designs of parts and products with respect to specific manufacturing processes and activities Examples of methods... separate but interrelated design issues of assembly and manufacturing processes Therefore, there are two fundamental aspects to producing efficient designs: DFA to help simplify the product and quantify assembly costs and the early implementation of DFM to quantify parts cost and allow trade-off decisions to be made for design proposals and material and process selection Design for Assembly The DFA... be produced quickly and consistently However, large production volumes are often required to offset tooling cost Design and manufacturing practice for mass-conserving processes generally focus on reducing tooling cost and development time and on setting process parameters to produce high-quality parts with low cycle times Detail design of the part and material selection are of particular importance... activities include design for service, design for testing, design for disassembly, and so forth Design for casting, design for plastic injection molding, and design for machining are examples of DFX methods directed toward improving the design of parts that are to be manufactured using a specific manufacturing process Many companies have specialized manufacturing facilities that can only process parts or assemblies... requirements, and standardizing procedures for batch-type production Design engineers use GT to reduce design time and effort as well as part and tooling proliferation With increasing emphasis on flexible and integrated manufacturing, GT is also an effective first step in structuring and building an integrated database Standardized process planning, accurate cost estimation, efficient purchasing, and assessment... processes, and detail design of the individual parts and components that make up the finished assembly Therefore, each part must be designed so that it not only meets functional requirements, but also can be manufactured and assembled economically and with relative ease in the production environment of the company Effective part design requires early consideration of the characteristics, capabilities, and. .. between design and manufacture has resulted in a variety of design and manufacturing practices This section provides a brief overview of various practices that are of general importance to design and manufacturing Some of these practices, such as standardization and group technology, span several organizations within a company and have their greatest effect at the systems level Others, such as design . Composite Materials, American Society for Metals, 1965 14. H.T. Hahn and J.G. Williams, Compression Failure Mechanisms in Unidirectional Composites, Composite Materials: Testing and Design. Manufacturing and Design Henry W. Stoll, Northwestern University Introduction THIS ARTICLE introduces and describes general concepts and practices related to manufacturing and design. It is. relationship among design, material selection, and manufacturing • Summarize modern design for manufacture practices being widely used in industry today The main focus is on how design and manufacturing