《纺织复合材料》课程参考文献（Composite Materials Handbook，Volume 3）Chapter 12 Lessons Learned

MIL-HDBK-17-3F Volume 3,Chapter 12-Lessons Learned CHAPTER 12 LESSONS LEARNED 12.1 INTRODUCTION The focus of much of what is in this handbook concentrates on establishing proper techniques for development and utilization of composite material property data.The motivation prompting specific choices is not always evident.This chapter provides a depository of knowledge gained from a number of involved contractors,agencies,and businesses for the purpose of disseminating lessons learned to po- tential users who might otherwise repeat past mistakes.Many of the contractors involved in developing the lessons learned are aerospace oriented.Thus,the lessons learned may have a decidedly aerospace viewpoint. The chapter starts with a discussion of some of the characteristics of composite materials that makes them different from metals.These characteristics are the primary cause for establishing the methods and techniques contained in the handbook. Specific lessons learned are defined in later sections.They contain the specific "rule of thumb"and the reason for its creation or the possible consequence if it is not followed.The lessons learned are or- ganized into six different categories for convenience. 12.2 UNIQUE ISSUES FOR COMPOSITES Composites are different from metals in several ways.These include their largely elastic response, their ability to be tailored in strength and stiffness,their damage tolerance characteristics,and their sensi- tivity to environmental factors.These differences force a different approach to analysis and design,proc- essing,fabrication and assembly,quality control,testing,and certification. 12.2.1 Elastic properties The elastic properties of a material are a measure of its stiffness.This property is necessary to de- termine the deformations that are produced by loads.In composites,the stiffness is dominated by the fibers;the role of the matrix is to prevent lateral deflections of the fibers and to provide a mechanism for shearing load from one fiber to another.Continuous fiber composites are transversely isotropic and in a two-dimensional stress state require four elastic properties to characterize the material: Modulus of elasticity parallel to the fiber,E Modulus of elasticity perpendicular to the fiber,E2 Shear modulus,Gi2 Major Poisson's ratio,v12 In general,material characterization may require additional properties not defined above.A thorough dis- cussion of this subject is given in Section 5.3.1.Only two elastic properties are required for isotropic ma- terials,the modulus of elasticity and Poisson's ratio. The stress-strain response of commonly used fiber-dominated orientations of composite materials is almost linear to failure although some glasses and ceramics have nonlinear or bilinear behavior.This is contrasted to metals that exhibit nonlinear response above the proportional limit and eventual plastic de- formation above the yield point.Many composites exhibit very little,if any,yielding in fiber dominated be- havior.Toughened materials and thermoplastics can show considerable yielding,particularly in matrix dominated directions.This factor requires composites to be given special consideration in structural de- tails where there are stress risers (holes,cutouts,notches,radii,tapers,etc.).These types of stress ris- ers in metal are not a major concern for static strength analysis(they do play a big role in durability and damage tolerance analysis,however).In composites they must be considered in static strength analysis. 12-1

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-1 CHAPTER 12 LESSONS LEARNED 12.1 INTRODUCTION The focus of much of what is in this handbook concentrates on establishing proper techniques for development and utilization of composite material property data. The motivation prompting specific choices is not always evident. This chapter provides a depository of knowledge gained from a number of involved contractors, agencies, and businesses for the purpose of disseminating lessons learned to po￾tential users who might otherwise repeat past mistakes. Many of the contractors involved in developing the lessons learned are aerospace oriented. Thus, the lessons learned may have a decidedly aerospace viewpoint. The chapter starts with a discussion of some of the characteristics of composite materials that makes them different from metals. These characteristics are the primary cause for establishing the methods and techniques contained in the handbook. Specific lessons learned are defined in later sections. They contain the specific "rule of thumb" and the reason for its creation or the possible consequence if it is not followed. The lessons learned are or￾ganized into six different categories for convenience. 12.2 UNIQUE ISSUES FOR COMPOSITES Composites are different from metals in several ways. These include their largely elastic response, their ability to be tailored in strength and stiffness, their damage tolerance characteristics, and their sensi￾tivity to environmental factors. These differences force a different approach to analysis and design, proc￾essing, fabrication and assembly, quality control, testing, and certification. 12.2.1 Elastic properties The elastic properties of a material are a measure of its stiffness. This property is necessary to de￾termine the deformations that are produced by loads. In composites, the stiffness is dominated by the fibers; the role of the matrix is to prevent lateral deflections of the fibers and to provide a mechanism for shearing load from one fiber to another. Continuous fiber composites are transversely isotropic and in a two-dimensional stress state require four elastic properties to characterize the material: Modulus of elasticity parallel to the fiber, E1 Modulus of elasticity perpendicular to the fiber, E2 Shear modulus, G12 Major Poisson's ratio, ν 12 In general, material characterization may require additional properties not defined above. A thorough dis￾cussion of this subject is given in Section 5.3.1. Only two elastic properties are required for isotropic ma￾terials, the modulus of elasticity and Poisson's ratio. The stress-strain response of commonly used fiber-dominated orientations of composite materials is almost linear to failure although some glasses and ceramics have nonlinear or bilinear behavior. This is contrasted to metals that exhibit nonlinear response above the proportional limit and eventual plastic de￾formation above the yield point. Many composites exhibit very little, if any, yielding in fiber dominated be￾havior. Toughened materials and thermoplastics can show considerable yielding, particularly in matrix dominated directions. This factor requires composites to be given special consideration in structural de￾tails where there are stress risers (holes, cutouts, notches, radii, tapers, etc.). These types of stress ris￾ers in metal are not a major concern for static strength analysis (they do play a big role in durability and damage tolerance analysis, however). In composites they must be considered in static strength analysis

MIL-HDBK-17-3F Volume 3,Chapter 12-Lessons Learned In general,if these stress risers are properly considered in design/analysis of laminated parts,fatigue loadings will not be critical. Another unique characteristic of composite material elastic response is its orthotropy.When metals are extended in one direction,they contract in the perpendicular direction in an amount equal to the Pois- son's ratio times the longitudinal strain.This is true regardless of which direction is extended.In compos- ites,an extension in the longitudinal(1 or x)direction produces a contraction in the transverse direction(2 or y)equal to the "major"Poisson's ratio,vxy,times the longitudinal extension.If this is reversed,an ex- tension in the transverse direction produces a much lower contraction in the longitudinal direction.In fiber dominated laminates,Poisson's ratio can vary from 0.5. The most unusual characteristic of composites is the response produced when the lay-up is unbal- anced and/or unsymmetric.Such a laminate exhibits anisotropic warping characteristics.In this condition an extension in one direction can produce an in-plane shear deformation.It can also cause an out-of-plane bending or torsional response.All these effects are sometimes observed in one laminate. This type of response is generally undesirable because of warping or built-in stresses that occur.Hence, most laminate configurations are balanced and symmetric. Classical lamination theory is used to combine the individual lamina properties to predict the linear elastic behavior of arbitrary laminates.Lamination theory requires the definition of lamina elastic proper- ties,their orientation within the laminate,and their stacking position.The process assumes plane sec- tions remain-plane and enforces equilibrium.Lamination theory will solve for the loads/stresses/strains for each lamina within the laminate at a given location for a given set of applied loads.This combined with appropriate failure theory will predict the strength of the laminate(empirically modified input ply prop- erties are often necessary). 12.2.2 Tailored properties and out-of-plane loads The properties of a composite laminate depend on the orientation of the individual plies.This pro- vides the engineer with the ability to tailor a laminate to fit a particular requirement.For high axial loads predominantly in one direction,the laminate should have a majority of its plies oriented parallel to that loading direction.If the laminate is loaded mostly in shear,there should be a high percent of t45 pairs. For loads in multi-directions,the laminate should be quasi-isotropic.An all 0 laminate represents the maximum strength and stiffness that can be attained in any given direction,but is impractical for most ap- plications since the transverse properties are so weak that machining and handling can cause damage. Fiber-dominated,balanced and symmetric,laminate designs that have a minimum of 10%of the plies in each of the0°，+45°，-45°，and90°directions are most commonly used. Tailoring also means an engineer is not able to cite a strength or stiffness value for a composite lami- nate until he knows the laminate's ply percentages in each direction.Carpet plots of various properties vs.the percent of plies in each direction are commonly used for balanced and symmetric laminates.An example for stiffness is shown in Figure 12.2.2.Similar plots for strength can also be developed. Out-of-plane loads can also be troublesome for composites.These loads cause interlaminar shear and tension in the laminate.Interlaminar shear stress can cause failure of the matrix or the fiber-matrix interphase region.Interlaminar shear and tensile stresses can delaminate or disbond a laminate.Such loading should be avoided if possible.Design situations that tend to create interlaminar shear loading include high out-of-plane loads(such as fuel pressure),buckling,abrupt changes in cross-section(such as stiffener terminations),ply drop-offs,and in some cases laminate ply orientations that cause unbal- anced or unsymmetric lay-ups.Interlaminar stresses will arise at any free edge.Interlaminar stresses will arise between plies of dissimilar orientation wherever there is a gradient in the components of in-plane stress. 12-2

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-2 In general, if these stress risers are properly considered in design/analysis of laminated parts, fatigue loadings will not be critical. Another unique characteristic of composite material elastic response is its orthotropy. When metals are extended in one direction, they contract in the perpendicular direction in an amount equal to the Pois￾son's ratio times the longitudinal strain. This is true regardless of which direction is extended. In compos￾ites, an extension in the longitudinal (1 or x) direction produces a contraction in the transverse direction (2 or y) equal to the "major" Poisson's ratio, ν xy , times the longitudinal extension. If this is reversed, an ex￾tension in the transverse direction produces a much lower contraction in the longitudinal direction. In fiber dominated laminates, Poisson's ratio can vary from 0.5. The most unusual characteristic of composites is the response produced when the lay-up is unbal￾anced and/or unsymmetric. Such a laminate exhibits anisotropic warping characteristics. In this condition an extension in one direction can produce an in-plane shear deformation. It can also cause an out-of-plane bending or torsional response. All these effects are sometimes observed in one laminate. This type of response is generally undesirable because of warping or built-in stresses that occur. Hence, most laminate configurations are balanced and symmetric. Classical lamination theory is used to combine the individual lamina properties to predict the linear elastic behavior of arbitrary laminates. Lamination theory requires the definition of lamina elastic proper￾ties, their orientation within the laminate, and their stacking position. The process assumes plane sec￾tions remain-plane and enforces equilibrium. Lamination theory will solve for the loads/stresses/strains for each lamina within the laminate at a given location for a given set of applied loads. This combined with appropriate failure theory will predict the strength of the laminate (empirically modified input ply prop￾erties are often necessary). 12.2.2 Tailored properties and out-of-plane loads The properties of a composite laminate depend on the orientation of the individual plies. This pro￾vides the engineer with the ability to tailor a laminate to fit a particular requirement. For high axial loads predominantly in one direction, the laminate should have a majority of its plies oriented parallel to that loading direction. If the laminate is loaded mostly in shear, there should be a high percent of ±45° pairs. For loads in multi-directions, the laminate should be quasi-isotropic. An all 0° laminate represents the maximum strength and stiffness that can be attained in any given direction, but is impractical for most ap￾plications since the transverse properties are so weak that machining and handling can cause damage. Fiber-dominated, balanced and symmetric, laminate designs that have a minimum of 10% of the plies in each of the 0°, +45°, -45°, and 90° directions are most commonly used. Tailoring also means an engineer is not able to cite a strength or stiffness value for a composite lami￾nate until he knows the laminate's ply percentages in each direction. Carpet plots of various properties vs. the percent of plies in each direction are commonly used for balanced and symmetric laminates. An example for stiffness is shown in Figure 12.2.2. Similar plots for strength can also be developed. Out-of-plane loads can also be troublesome for composites. These loads cause interlaminar shear and tension in the laminate. Interlaminar shear stress can cause failure of the matrix or the fiber-matrix interphase region. Interlaminar shear and tensile stresses can delaminate or disbond a laminate. Such loading should be avoided if possible. Design situations that tend to create interlaminar shear loading include high out-of-plane loads (such as fuel pressure), buckling, abrupt changes in cross-section (such as stiffener terminations), ply drop-offs, and in some cases laminate ply orientations that cause unbal￾anced or unsymmetric lay-ups. Interlaminar stresses will arise at any free edge. Interlaminar stresses will arise between plies of dissimilar orientation wherever there is a gradient in the components of in-plane stress

MIL-HDBK-17-3F Volume 3,Chapter 12-Lessons Learned 140 20 18 120 80%- 0's 16 100 14 60%- 0's 0 80 0 202 90'5 (2) 05 40% 十 40% 90's 60 (4) 90 60 20%-0's} 6 90's 40 80% 90' 0%-0's 20 2 0 0 10 20 30 4050 60 70 80 90 100 号45 PLIES ELASTIC MODULI FIGURE 12.2.2 Sample carpet plot. 12-3

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-3 FIGURE 12.2.2 Sample carpet plot

MIL-HDBK-17-3F Volume 3.Chapter 12-Lessons Learned 12.2.3 Damage tolerance Damage tolerance is the measure of the structure's ability to sustain a level of damage or presence of a defect and be able to perform its operating functions.The concern is with the damaged structure hav- ing adequate residual strength and stiffness to continue in service safely:1)until the damage can be de- tected by scheduled maintenance inspection and repaired,or 2)if the damage is undetected,for the re- mainder of the aircraft's life.Thus,safety is the primary goal of damage tolerance.Both static load and durability related damage tolerance must be interrogated experimentally because there are few,if any. accurate analytical methods. There are basically two types of damage that are categorized by their occurrence during the fabrica- tion and use of the part,i.e.,damage occurring during manufacturing or damage occurring in service.It is hoped that the occurrence of the majority of manufacturing associated damage,if beyond specification limits,will be detected by routine quality inspection.Nevertheless,some "rogue"defects or damage be- yond specification limits may go undetected.Consequently,their occurrence must be assumed in the design procedure and subsequent testing(static and fatigue)performed to verify the structural integrity. Service damage concerns are similar to those for manufacturing.Types of service damage include edge and surface gouges and cuts or foreign object collision and blunt object impact damage caused by dropped tools or contact with service equipment.A level of non-detectable damage should be established and verified by test that will not endanger the normal operation of the aircraft structure for two lifetimes.A certain level(maximum allowed)damage that can be found by inspection should be defined such that the vehicle can operate for a specified number of hours before repair or replacement at loads not exceeding design limit.This damage should also be tested(statically and in fatigue)to verify the structural integrity. Delaminations can also be critical defects.However,unless they are very large,historically more than 2 inches(50 mm)in diameter,the problem is mostly with thin laminates.Effects of manufacturing defects such as porosity and flawed fastener holes that are slightly in excess of the maximum allowable are usually less severe.They are generally accounted for by the use of design allowable properties that have been obtained by testing specimens with stress concentrations,e.g.,notches.Most commonly these are specimens with a centered hole.Open holes are typically used for compression specimens while either open or filled holes (holes with an installed fastener)are used for tension testing.(Open holes are more critical than filled holes for compression.Filled holes may be more critical in tension,es- pecially for laminates with ply orientations with a predominate number of plies in the load direction.)Con- sequently,the design allowables thus produced may be used to account for a nominal design stress con- centration caused by an installed or missing fastener,at least to a 0.25 inch(6.4 mm)diameter,as well as accounting for many other manufacturing defects.This is sometimes called the "rogue flaw"approach to laminate design,see Reference 12.2.3. 12.2.4 Durability Durability of a structure is its ability to maintain strength and stiffness throughout the service life of the structure.A structure must have adequate durability when subjected to the expected service loads and environment spectra to prevent excessive maintenance,repair,or modification costs over the service life. Thus,durability is primarily an economic consideration. Metallic structure can be very sensitive to durability issues;major factors limiting life are corrosion and fatigue.Metal fatigue is dictated by the number of load cycles required to start a crack(crack initia- tion)and the number of load cycles for the crack to grow to its critical length,reaching catastrophic failure (crack growth).Crack/damage growth rate is very dependent on the concentration of stress around the crack. In composites,it has been demonstrated that one of the most common damage growth mechanisms is intercracking(delamination).This makes composites most sensitive to compression-dominated fatigue loading.A second common fatigue failure mode is fastener hole wear caused by high bearing stresses. 12-4

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-4 12.2.3 Damage tolerance Damage tolerance is the measure of the structure's ability to sustain a level of damage or presence of a defect and be able to perform its operating functions. The concern is with the damaged structure hav￾ing adequate residual strength and stiffness to continue in service safely: 1) until the damage can be de￾tected by scheduled maintenance inspection and repaired, or 2) if the damage is undetected, for the re￾mainder of the aircraft's life. Thus, safety is the primary goal of damage tolerance. Both static load and durability related damage tolerance must be interrogated experimentally because there are few, if any, accurate analytical methods. There are basically two types of damage that are categorized by their occurrence during the fabrica￾tion and use of the part, i.e., damage occurring during manufacturing or damage occurring in service. It is hoped that the occurrence of the majority of manufacturing associated damage, if beyond specification limits, will be detected by routine quality inspection. Nevertheless, some "rogue" defects or damage be￾yond specification limits may go undetected. Consequently, their occurrence must be assumed in the design procedure and subsequent testing (static and fatigue) performed to verify the structural integrity. Service damage concerns are similar to those for manufacturing. Types of service damage include edge and surface gouges and cuts or foreign object collision and blunt object impact damage caused by dropped tools or contact with service equipment. A level of non-detectable damage should be established and verified by test that will not endanger the normal operation of the aircraft structure for two lifetimes. A certain level (maximum allowed) damage that can be found by inspection should be defined such that the vehicle can operate for a specified number of hours before repair or replacement at loads not exceeding design limit. This damage should also be tested (statically and in fatigue) to verify the structural integrity. Delaminations can also be critical defects. However, unless they are very large, historically more than 2 inches (50 mm) in diameter, the problem is mostly with thin laminates. Effects of manufacturing defects such as porosity and flawed fastener holes that are slightly in excess of the maximum allowable are usually less severe. They are generally accounted for by the use of design allowable properties that have been obtained by testing specimens with stress concentrations, e.g., notches. Most commonly these are specimens with a centered hole. Open holes are typically used for compression specimens while either open or filled holes (holes with an installed fastener) are used for tension testing. (Open holes are more critical than filled holes for compression. Filled holes may be more critical in tension, es￾pecially for laminates with ply orientations with a predominate number of plies in the load direction.) Con￾sequently, the design allowables thus produced may be used to account for a nominal design stress con￾centration caused by an installed or missing fastener, at least to a 0.25 inch (6.4 mm) diameter, as well as accounting for many other manufacturing defects. This is sometimes called the "rogue flaw" approach to laminate design, see Reference 12.2.3. 12.2.4 Durability Durability of a structure is its ability to maintain strength and stiffness throughout the service life of the structure. A structure must have adequate durability when subjected to the expected service loads and environment spectra to prevent excessive maintenance, repair, or modification costs over the service life. Thus, durability is primarily an economic consideration. Metallic structure can be very sensitive to durability issues; major factors limiting life are corrosion and fatigue. Metal fatigue is dictated by the number of load cycles required to start a crack (crack initia￾tion) and the number of load cycles for the crack to grow to its critical length, reaching catastrophic failure (crack growth). Crack/damage growth rate is very dependent on the concentration of stress around the crack. In composites, it has been demonstrated that one of the most common damage growth mechanisms is intercracking (delamination). This makes composites most sensitive to compression-dominated fatigue loading. A second common fatigue failure mode is fastener hole wear caused by high bearing stresses

MIL-HDBK-17-3F Volume 3.Chapter 12-Lessons Learned In this failure mode the hole gradually elongates.The most serious damage to composite parts is low velocity impact damage which can reduce static strength,fatigue strength,or residual strength after fa- tigue.Again,testing is a must! The strain level of composites in most actual vehicle applications to date has been held to relatively low values.Composites under in-plane loads have relatively flat stress-life(S-N)curves with high fatigue thresholds (endurance limits).These two factors combined have resulted in insensitivity to fatigue for most load cases.However,the greater variability found with composites requires an engineer to still characterize the composite's fatigue life to failure to correctly characterize its fatigue scatter. 12.2.5 Environmental sensitivity When a composite with a polymeric matrix is placed in a wet environment,the matrix will absorb moisture.The moisture absorption of most fibers used in practice is negligible;however,aramid fibers (e.g.,Kevlar)absorb significant amounts of moisture when exposed to high humidity.The absorption of moisture at the interface of glass/quartz fibers is a well-known degrading phenomena. When a composite has been exposed to moisture and sufficient time has elapsed,the moisture con- centration throughout the matrix will be uniform.A typical equilibrium moisture content for severe humidity exposure of common epoxy composites is 1.1 to 1.3 percent weight gain.The principal strength degrad- ing effect is related to a change in the glass transition temperature of the matrix material.As moisture is absorbed,the temperature at which the matrix changes from a glassy state to a viscous state decreases. Thus,the strength properties decrease with increasing moisture content.Current data indicate this proc- ess is reversible.When the moisture content is decreased,the glass transition temperature increases and the original strength properties return.With glass/quartz fibers there is additional degradation at the interface with the matrix.For aramid fibers there is additional degradation at the interface with the matrix and.also.in the fibers. The same considerations also apply for a temperature rise.The matrix,and therefore the lamina, loses strength and stiffness when the temperature rises.This effect is primarily important for the ma- trix-dominated properties.Temperature rise also worsens the fiber/matrix interface degradation for glass/quartz fibers and aramid fibers.The aramid fiber properties are also degraded by a rise in tempera- ture The approach for design purposes is to assume a worst case.If the material is assumed to be fully saturated and at the maximum temperature,material allowables can be derived for this extreme.This is a conservative approach,since typical service environments do not generate full saturation for most com- plex structures.Once the diffusivity of a composite material is known,the moisture content and through the thickness distribution can be accurately predicted by Fickian equations.This depends on an accurate characterization of the temperature-humidity service environment. Thermal expansion characteristics of common composites,like carbon/epoxy,are quite different from metals.In the (0 or 1)longitudinal direction,the thermal expansion coefficient of carbon/epoxy is almost zero.Transverse to the fiber(90 or 2 direction),the thermal expansion is the same magnitude as alumi- num.This property gives composites the ability to provide a dimensionally stable structure throughout a wide range of temperatures. Another feature of composites that is related to environment is resistance to corrosion.Polymer ma- trix composites (with the exception of some carbon/bismaleimides)are immune to salt water and most chemical substances as far as corrosion sensitivity.One precaution in this regard is galvanic corrosion. Carbon fiber is cathodic (noble);aluminum and steel are anodic (least noble).Thus carbon in contact with aluminum or steel promotes galvanic action which results in corrosion of the metal.Corrosion barri- ers(such as fiberglass and sealants)are placed at interfaces between composites and metals to prevent metal corrosion.Another precaution regards the use of paint strippers around most polymers.Chemical paint strippers are very powerful and attack the matrix of composites very destructively.Thus,chemical paint stripping is forbidden on composite structure. 12-5

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-5 In this failure mode the hole gradually elongates. The most serious damage to composite parts is low velocity impact damage which can reduce static strength, fatigue strength, or residual strength after fa￾tigue. Again, testing is a must! The strain level of composites in most actual vehicle applications to date has been held to relatively low values. Composites under in-plane loads have relatively flat stress-life (S-N) curves with high fatigue thresholds (endurance limits). These two factors combined have resulted in insensitivity to fatigue for most load cases. However, the greater variability found with composites requires an engineer to still characterize the composite's fatigue life to failure to correctly characterize its fatigue scatter. 12.2.5 Environmental sensitivity When a composite with a polymeric matrix is placed in a wet environment, the matrix will absorb moisture. The moisture absorption of most fibers used in practice is negligible; however, aramid fibers (e.g., Kevlar) absorb significant amounts of moisture when exposed to high humidity. The absorption of moisture at the interface of glass/quartz fibers is a well-known degrading phenomena. When a composite has been exposed to moisture and sufficient time has elapsed, the moisture con￾centration throughout the matrix will be uniform. A typical equilibrium moisture content for severe humidity exposure of common epoxy composites is 1.1 to 1.3 percent weight gain. The principal strength degrad￾ing effect is related to a change in the glass transition temperature of the matrix material. As moisture is absorbed, the temperature at which the matrix changes from a glassy state to a viscous state decreases. Thus, the strength properties decrease with increasing moisture content. Current data indicate this proc￾ess is reversible. When the moisture content is decreased, the glass transition temperature increases and the original strength properties return. With glass/quartz fibers there is additional degradation at the interface with the matrix. For aramid fibers there is additional degradation at the interface with the matrix and, also, in the fibers. The same considerations also apply for a temperature rise. The matrix, and therefore the lamina, loses strength and stiffness when the temperature rises. This effect is primarily important for the ma￾trix-dominated properties. Temperature rise also worsens the fiber/matrix interface degradation for glass/quartz fibers and aramid fibers. The aramid fiber properties are also degraded by a rise in tempera￾ture. The approach for design purposes is to assume a worst case. If the material is assumed to be fully saturated and at the maximum temperature, material allowables can be derived for this extreme. This is a conservative approach, since typical service environments do not generate full saturation for most com￾plex structures. Once the diffusivity of a composite material is known, the moisture content and through the thickness distribution can be accurately predicted by Fickian equations. This depends on an accurate characterization of the temperature-humidity service environment. Thermal expansion characteristics of common composites, like carbon/epoxy, are quite different from metals. In the (0 or 1) longitudinal direction, the thermal expansion coefficient of carbon/epoxy is almost zero. Transverse to the fiber (90 or 2 direction), the thermal expansion is the same magnitude as alumi￾num. This property gives composites the ability to provide a dimensionally stable structure throughout a wide range of temperatures. Another feature of composites that is related to environment is resistance to corrosion. Polymer ma￾trix composites (with the exception of some carbon/bismaleimides) are immune to salt water and most chemical substances as far as corrosion sensitivity. One precaution in this regard is galvanic corrosion. Carbon fiber is cathodic (noble); aluminum and steel are anodic (least noble). Thus carbon in contact with aluminum or steel promotes galvanic action which results in corrosion of the metal. Corrosion barri￾ers (such as fiberglass and sealants) are placed at interfaces between composites and metals to prevent metal corrosion. Another precaution regards the use of paint strippers around most polymers. Chemical paint strippers are very powerful and attack the matrix of composites very destructively. Thus, chemical paint stripping is forbidden on composite structure

MIL-HDBK-17-3F Volume 3.Chapter 12-Lessons Learned Other environmental effects worth noting include the effect of long term exposure to radiation.Ultra- violet rays from the sun can degrade epoxy resins.This is easily protected by a surface finish such as a coat of paint.Another factor is erosion or pitting caused by high speed impact with rain or dust particles. This is likely to occur on unprotected leading edges.There are surface finishes such as rain erosion coats and paints for preventing surface wear.Lightning strike is also a concern to composites.A direct strike can cause considerable damage to a laminate.Lightning strike protection in the form of conductive surfaces is applied in susceptible areas.In cases where substructure is also composite,the inside end of attachment bolts may need to be connected with each other and to ground by a conducting wire. 12.2.6 Joints 12.2.6.1 Mechanically-fastened joints Successful joint design relies on knowledge of potential failure modes.Failure modes depend on joint geometry and laminate lay-up for one given material.The type of fastener used can also influence the occurrence of a particular failure mode.Different materials will give different failure modes. Net-section tensile/compressive failures occur when the bolt diameter is a sufficiently large fraction of the strip width.For most successful designs,this fraction (D/W)is about one-quarter or more for near-isotropic lay-ups in carbon/epoxy systems that have a D/E of one-third or less. Shear-out and shear-out delamination failures occur because the bolt is too close to the edge of the laminate.Such a failure can be triggered when there is only a partial net-section tension or bearing fail- ure.D/t ratios should be 0.75 to 1.25. In some instances the bolt head may be pulled through the laminate after the bolt is bent and de- formed.This mode is frequently seen with countersunk fasteners and is highly dependent on the particu- lar fastener used. Bearing strength is a function of joint geometry,fastener and member stiffnesses.For a 0/45/90 family of laminates with 20-40%of 0 plies and 40-60%of t45plies,plus a minimum(10%)of 90 plies, the bearing strength is relatively constant.Fastener characteristics such as clamp-up force and head configuration have a significant effect.However,for a specific laminate family,a specific fastener,and equal thickness laminate joining members,the parameter with the greatest influence is D/t. Composite joints require smaller D/W and D/E ratios than do metals to get bearing failures. Composite joint strength characteristics differ from metals because the strength is influenced by the bypass load going around the joint.This occurs when two or more fasteners are arranged in a line to transfer the load through a joint.Since not all of the load is reacted by one fastener,some of the load by-passes it.The by-pass effects become prominent once the ratio of by-pass to fastener bearing load exceeds 20%. Titanium fasteners are the most common means of mechanical attachment in composites.This is because titanium is non-corrosive in the galvanic atmosphere created by the dissimilar materials.Tita- nium is closer to carbon on the cathodic scale. 12.2.6.2 Problems associated with adhesive bonding to peel-ply composite surfaces There are two schools of thought in regard to the adhesive bonding of fibrous composite laminates. One demands light but thorough mechanical abrasion,such as by low-pressure grit blasting,because the only such bonds never to fail prematurely were made to abraded surfaces on completely dry laminates. The other permits bonding directly to surfaces created by stripping off peel plies,with or without a drying requirement,using the justification that there is "adequate"initial strength,even though some of these joints have failed prematurely in service.It is also significant that no ultrasonic inspection technique has 12-6

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-6 Other environmental effects worth noting include the effect of long term exposure to radiation. Ultra￾violet rays from the sun can degrade epoxy resins. This is easily protected by a surface finish such as a coat of paint. Another factor is erosion or pitting caused by high speed impact with rain or dust particles. This is likely to occur on unprotected leading edges. There are surface finishes such as rain erosion coats and paints for preventing surface wear. Lightning strike is also a concern to composites. A direct strike can cause considerable damage to a laminate. Lightning strike protection in the form of conductive surfaces is applied in susceptible areas. In cases where substructure is also composite, the inside end of attachment bolts may need to be connected with each other and to ground by a conducting wire. 12.2.6 Joints 12.2.6.1 Mechanically-fastened joints Successful joint design relies on knowledge of potential failure modes. Failure modes depend on joint geometry and laminate lay-up for one given material. The type of fastener used can also influence the occurrence of a particular failure mode. Different materials will give different failure modes. Net-section tensile/compressive failures occur when the bolt diameter is a sufficiently large fraction of the strip width. For most successful designs, this fraction (D/W) is about one-quarter or more for near-isotropic lay-ups in carbon/epoxy systems that have a D/E of one-third or less. Shear-out and shear-out delamination failures occur because the bolt is too close to the edge of the laminate. Such a failure can be triggered when there is only a partial net-section tension or bearing fail￾ure. D/t ratios should be 0.75 to 1.25. In some instances the bolt head may be pulled through the laminate after the bolt is bent and de￾formed. This mode is frequently seen with countersunk fasteners and is highly dependent on the particu￾lar fastener used. Bearing strength is a function of joint geometry, fastener and member stiffnesses. For a 0/±45/90 family of laminates with 20-40% of 0° plies and 40-60% of ±45° plies, plus a minimum (10%) of 90° plies, the bearing strength is relatively constant. Fastener characteristics such as clamp-up force and head configuration have a significant effect. However, for a specific laminate family, a specific fastener, and equal thickness laminate joining members, the parameter with the greatest influence is D/t. Composite joints require smaller D/W and D/E ratios than do metals to get bearing failures. Composite joint strength characteristics differ from metals because the strength is influenced by the bypass load going around the joint. This occurs when two or more fasteners are arranged in a line to transfer the load through a joint. Since not all of the load is reacted by one fastener, some of the load by-passes it. The by-pass effects become prominent once the ratio of by-pass to fastener bearing load exceeds 20%. Titanium fasteners are the most common means of mechanical attachment in composites. This is because titanium is non-corrosive in the galvanic atmosphere created by the dissimilar materials. Tita￾nium is closer to carbon on the cathodic scale. 12.2.6.2 Problems associated with adhesive bonding to peel-ply composite surfaces There are two schools of thought in regard to the adhesive bonding of fibrous composite laminates. One demands light but thorough mechanical abrasion, such as by low-pressure grit blasting, because the only such bonds never to fail prematurely were made to abraded surfaces on completely dry laminates. The other permits bonding directly to surfaces created by stripping off peel plies, with or without a drying requirement, using the justification that there is “adequate” initial strength, even though some of these joints have failed prematurely in service. It is also significant that no ultrasonic inspection technique has

MIL-HDBK-17-3F Volume 3,Chapter 12-Lessons Learned been able to distinguish between bonded joints which will fail in service and those which will not.In addi- tion,most traveler specimens do not represent the same cure conditions as experienced by adjacent large parts and,therefore,mechanical testing also often fails to identify defective bonding.One must de- pend on process control of techniques which can be relied upon 100 percent of the time and on thorough validation of the processes before committing them to production. Consider a surface,created by stripping off a peel ply,which is then bonded as part of an adhesive joint.The resulting adhesive bond "sticks"well enough to pass all inspections;however,may fail prema- turely at the interface between the laminate and the adhesive.All premature bond failures,other than those caused by incomplete cure,occur at the interface between the adhesive and the resin in the lami- nate.Structurally sound bonds either fail outside the joint area,cohesively within the layer of adhesive,or interlaminarly in the resin matrix between the surface fibers and the adhesive layer.These premature failures can occur either when uncured adhesive is bonded to precured laminates,or when uncured pre- preg is cured against cured adhesive films used to stabilize honeycomb cores and the like. There are several ways in which peel plies can create surfaces on which reliable durable bonds are not possible. The peel ply can be coated with a release agent,which transfers to the cured laminate when the peel ply is stripped off. The surface of the peel ply's fibers must be sufficiently inert that the ply can be removed without damaging the laminate.The grooves left in the laminate (or glue layer)by stripping off the peel ply may retain enough inert surface that the resin which is subsequently cured onto it may simply fail to adhere.Adhesion requires more than cleanliness:surface tension is also critical.In the absence of cohesion at the interface,a bonded joint relies only on mechanical interlocking,which is far weaker in peel than it is in shear. The peel-ply surface in the laminate consists of innumerable short grooves separated by sharp edges where the resin between the filaments in the peel ply fractured as the peel ply was stripped off.Moisture on (or in)the adhesive or the laminate can be trapped in these grooves.If this moisture cannot escape during the curing of the adhesive (or of a co-cured face sheet),the trapped moisture will result in a slick bond when examined microscopically after failure. It should be noted that the latter two mechanisms function without any contamination. One aircraft company's process specification has,for decades,required that any peel-ply surface to be bonded must first be thoroughly abraded to remove all traces of the texture of the peel ply.In the ab- sence of the ridges between the grooves,it is presumed that moisture could escape,as it turned to steam during cure,unless the part was too large and too poorly ventilated.Using these requirements,this air- craft company has had no disbonds in those secondarily bonded composite structures which were grit blasted before bonding.The same cannot be claimed for bonds made to unabraded (or only scuff- sanded)peel-ply surfaces.In two instances,on different aircraft types,disbonds were traced to transfer of release agents from silicone-coated peel plies,the use of which is now banned throughout all docu- ments,not only the approved materials lists. On another aircraft type,interfacial failures on peel-ply surfaces appear to be the result of prebond moisture,the exact origin of which has yet to be established.An accident with one test panel during pro- cess qualification by a supplier revealed the consequences of condensate on adhesive film (the roll of adhesive had not been properly sealed when returned to the freezer after the previous use).There was absolutely no adhesion between the resin and the adhesive,even though the lap-shear numbers seemed to be acceptable.Microscopic examination of the surfaces clearly showed perfect imprints of the peel ply texture on both surfaces,with the surface in all grooves as smooth as glass and all of the resin on one surface and all of the adhesive on the other.However,with thicker-than-normal(0.123 inch(3.2 mm)) adherends of the same unidirectional carbon/epoxy,bonds made with the same nonreleased peel ply and the same kind of adhesive achieved cohesive failure of the bond at the same strength level attained by 12-7

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-7 been able to distinguish between bonded joints which will fail in service and those which will not. In addi￾tion, most traveler specimens do not represent the same cure conditions as experienced by adjacent large parts and, therefore, mechanical testing also often fails to identify defective bonding. One must de￾pend on process control of techniques which can be relied upon 100 percent of the time and on thorough validation of the processes before committing them to production. Consider a surface, created by stripping off a peel ply, which is then bonded as part of an adhesive joint. The resulting adhesive bond “sticks” well enough to pass all inspections; however, may fail prema￾turely at the interface between the laminate and the adhesive. All premature bond failures, other than those caused by incomplete cure, occur at the interface between the adhesive and the resin in the lami￾nate. Structurally sound bonds either fail outside the joint area, cohesively within the layer of adhesive, or interlaminarly in the resin matrix between the surface fibers and the adhesive layer. These premature failures can occur either when uncured adhesive is bonded to precured laminates, or when uncured pre￾preg is cured against cured adhesive films used to stabilize honeycomb cores and the like. There are several ways in which peel plies can create surfaces on which reliable durable bonds are not possible. • The peel ply can be coated with a release agent, which transfers to the cured laminate when the peel ply is stripped off. • The surface of the peel ply’s fibers must be sufficiently inert that the ply can be removed without damaging the laminate. The grooves left in the laminate (or glue layer) by stripping off the peel ply may retain enough inert surface that the resin which is subsequently cured onto it may simply fail to adhere. Adhesion requires more than cleanliness; surface tension is also critical. In the absence of cohesion at the interface, a bonded joint relies only on mechanical interlocking, which is far weaker in peel than it is in shear. • The peel-ply surface in the laminate consists of innumerable short grooves separated by sharp edges where the resin between the filaments in the peel ply fractured as the peel ply was stripped off. Moisture on (or in) the adhesive or the laminate can be trapped in these grooves. If this moisture cannot escape during the curing of the adhesive (or of a co-cured face sheet), the trapped moisture will result in a slick bond when examined microscopically after failure. It should be noted that the latter two mechanisms function without any contamination. One aircraft company’s process specification has, for decades, required that any peel-ply surface to be bonded must first be thoroughly abraded to remove all traces of the texture of the peel ply. In the ab￾sence of the ridges between the grooves, it is presumed that moisture could escape, as it turned to steam during cure, unless the part was too large and too poorly ventilated. Using these requirements, this air￾craft company has had no disbonds in those secondarily bonded composite structures which were grit blasted before bonding. The same cannot be claimed for bonds made to unabraded (or only scuff￾sanded) peel-ply surfaces. In two instances, on different aircraft types, disbonds were traced to transfer of release agents from silicone-coated peel plies, the use of which is now banned throughout all docu￾ments, not only the approved materials lists. On another aircraft type, interfacial failures on peel-ply surfaces appear to be the result of prebond moisture, the exact origin of which has yet to be established. An accident with one test panel during pro￾cess qualification by a supplier revealed the consequences of condensate on adhesive film (the roll of adhesive had not been properly sealed when returned to the freezer after the previous use). There was absolutely no adhesion between the resin and the adhesive, even though the lap-shear numbers seemed to be acceptable. Microscopic examination of the surfaces clearly showed perfect imprints of the peel ply texture on both surfaces, with the surface in all grooves as smooth as glass and all of the resin on one surface and all of the adhesive on the other. However, with thicker-than-normal (0.123 inch (3.2 mm)) adherends of the same unidirectional carbon/epoxy, bonds made with the same nonreleased peel ply and the same kind of adhesive achieved cohesive failure of the bond at the same strength level attained by

MIL-HDBK-17-3F Volume 3.Chapter 12-Lessons Learned metal-to-metal bonding (6,000 psi(40 MPa)or so).With normal thickness composite adherends,only half this strength was reached,because the resin between the surface fibers and adhesive layer then failed in peel,leaving resin clearly covering both surfaces.This problem can be minimized by maintaining very tight time limits between making parts and bonding them together,with a requirement to thoroughly dry everything before bonding if the time constraints are exceeded.Careful scheduling can avoid this added drying step.The same high-strength cohesive bond failures had previously been achieved by an- other supplier of composite structures using grit-blasted surfaces and 0.080 inch(2.0 mm)thick unidirec- tional laminates. In considering adhesive bond strength,it is vital to note that the specimen testing validates the proc- ess,NOT the part.There is no requirement for the specimen to look like the actual part.Indeed.in a properly designed bonded joint,the bond will not fail first.Consequently,the use of specimens which are "similar"to the part and which are evaluated in terms of the "adequacy"of the load carried in relation to the stresses in the part,is not sufficient to ensure the integrity of the bonded composite structure.This issue is complicated because,only with unidirectional tape laminates is it possible to develop sufficient load to fail a high-strength adhesive bond cohesively.Therefore,only such specimens can provide any assurance that the part they are intended to substantiate has been bonded properly.However,in real parts made from woven-fabric laminates,failures within bundles of fibers at 90 to the applied load will trigger interlaminate failures before such bond strengths can be attained. In all cases,the one condition which can be detected visually on test specimens and failed parts alike which is a guaranteed indicator of a defective bond is an interfacial failure with all of the resin on one side and all of the adhesive on the other,with a clear imprint of the peel ply texture on both surfaces. 12.2.7 Design The design of composite structure is complicated by the fact that every ply must be defined.Draw- ings or design packages must describe the ply orientation,its position within the stack,and its boundaries. This is straightforward for a simple,constant thickness laminate.For complex parts with tapered thick- nesses and ply build-ups around joints and cutouts,this can become extremely complex.The need to maintain relative balance and symmetry throughout the structure increases the difficulty. Composites can not be designed without concurrence.Design details depend on tooling and proc- essing as does assembly and inspection.Parts and processes are so interdependent it could be disas- trous to attempt sequential design and manufacturing phasing. Another factor approached differently in composite design is the accommodation of thickness toler- ances at interfaces.If a composite part must fit into a space between two other parts or between a sub- structure and an outer mold line,the thickness requires special tolerances.The composite part thickness is controlled by the number of plies and the per-ply-thickness.Each ply has a range of possible thick- nesses.When these are layed up to form the laminate they may not match the space available for as- sembly within other constraints.This discrepancy can be handled by using shims or by adding "sacrifi- cial"plies to the laminate(for subsequent machining to a closer tolerance than is possible with nominal per-ply-thickness variations).The use of shims has design implications regarding load eccentricities. Another approach is to use closed die molding at the fit-up edges to mold to exact thickness needed. The anisotropy of special laminates,while more complicated,enables a designer to tailor a structure for desired deflection characteristics.This has been applied to some extent for aeroelastic tailoring of wing skins. Composites are most efficient when used in large,relatively uninterrupted structures.The cost is also related to the number of detail parts and the number of fasteners required.These two factors drive de- signs towards integration of features into large cocured structures.The nature of composites enables this possibility.Well designed,high quality tooling will reduce manufacturing and inspection cost and rejection rate and result in high quality parts. 12-8

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-8 metal-to-metal bonding (6,000 psi (40 MPa) or so). With normal thickness composite adherends, only half this strength was reached, because the resin between the surface fibers and adhesive layer then failed in peel, leaving resin clearly covering both surfaces. This problem can be minimized by maintaining very tight time limits between making parts and bonding them together, with a requirement to thoroughly dry everything before bonding if the time constraints are exceeded. Careful scheduling can avoid this added drying step. The same high-strength cohesive bond failures had previously been achieved by an￾other supplier of composite structures using grit-blasted surfaces and 0.080 inch (2.0 mm) thick unidirec￾tional laminates. In considering adhesive bond strength, it is vital to note that the specimen testing validates the proc￾ess, NOT the part. There is no requirement for the specimen to look like the actual part. Indeed, in a properly designed bonded joint, the bond will not fail first. Consequently, the use of specimens which are “similar” to the part and which are evaluated in terms of the “adequacy” of the load carried in relation to the stresses in the part, is not sufficient to ensure the integrity of the bonded composite structure. This issue is complicated because, only with unidirectional tape laminates is it possible to develop sufficient load to fail a high-strength adhesive bond cohesively. Therefore, only such specimens can provide any assurance that the part they are intended to substantiate has been bonded properly. However, in real parts made from woven-fabric laminates, failures within bundles of fibers at 90° to the applied load will trigger interlaminate failures before such bond strengths can be attained. In all cases, the one condition which can be detected visually on test specimens and failed parts alike which is a guaranteed indicator of a defective bond is an interfacial failure with all of the resin on one side and all of the adhesive on the other, with a clear imprint of the peel ply texture on both surfaces. 12.2.7 Design The design of composite structure is complicated by the fact that every ply must be defined. Draw￾ings or design packages must describe the ply orientation, its position within the stack, and its boundaries. This is straightforward for a simple, constant thickness laminate. For complex parts with tapered thick￾nesses and ply build-ups around joints and cutouts, this can become extremely complex. The need to maintain relative balance and symmetry throughout the structure increases the difficulty. Composites can not be designed without concurrence. Design details depend on tooling and proc￾essing as does assembly and inspection. Parts and processes are so interdependent it could be disas￾trous to attempt sequential design and manufacturing phasing. Another factor approached differently in composite design is the accommodation of thickness toler￾ances at interfaces. If a composite part must fit into a space between two other parts or between a sub￾structure and an outer mold line, the thickness requires special tolerances. The composite part thickness is controlled by the number of plies and the per-ply-thickness. Each ply has a range of possible thick￾nesses. When these are layed up to form the laminate they may not match the space available for as￾sembly within other constraints. This discrepancy can be handled by using shims or by adding "sacrifi￾cial" plies to the laminate (for subsequent machining to a closer tolerance than is possible with nominal per-ply-thickness variations). The use of shims has design implications regarding load eccentricities. Another approach is to use closed die molding at the fit-up edges to mold to exact thickness needed. The anisotropy of special laminates, while more complicated, enables a designer to tailor a structure for desired deflection characteristics. This has been applied to some extent for aeroelastic tailoring of wing skins. Composites are most efficient when used in large, relatively uninterrupted structures. The cost is also related to the number of detail parts and the number of fasteners required. These two factors drive de￾signs towards integration of features into large cocured structures. The nature of composites enables this possibility. Well designed, high quality tooling will reduce manufacturing and inspection cost and rejection rate and result in high quality parts

MIL-HDBK-17-3F Volume 3.Chapter 12-Lessons Learned 12.2.8 Handling and storage Epoxy resins are the most common form of matrix material used in composites.Epoxies are perish- able.They must be stored below freezing temperature and even then have limited shelf life.Once the material is brought out of storage there is limited time it can be used to make parts(30 days is common). For very complex parts with many plies,the material's permissible out-time can be a controlling factor.If the material is not completely used,it may be returned to storage.An out-time record should be kept.In addition,freezer storage of these materials is usually limited by the vendor to 6 to 12 months.Overage material will produce laminates with a high level of porosity. The perishability of the material also requires that it be shipped refrigerated from the supplier.Upon arrival at the contractor's facility,there must be provisions to prevent it being left on-dock for long periods of time Tack is another composite material characteristic that is unique.Tack is"stickiness"of the prepreg.It is both an aid and a hindrance.Tack is helpful to maintain location of a ply once it is placed in position.It also makes it difficult to adjust the location once the ply has been placed. 12.2.9 Processing and fabrication Composite parts are fabricated by successive placement of plies one after the other.Parts are built-up rather than machined down.Many metal fabrication steps require successive removal of material starting from large ingots,plates,or forgings.Prepreg "tape"material typically comes in rolls of relatively thin strips(0.005-0.015 inches or 0.13-0.38 mm).These strips are a variety of widths:3",6",and 36". Prepreg "fabric"is usually thicker than tape(0.007-0.020 inches or 0.18-0.51 mm)and usually comes in 36-inch(0.9 m)wide rolls. Fabrication of a detail part requires the material to be taken out of the freezer in a sealed bag and allowed to come to room temperature prior to any operations.Placement of the prepreg on the tool(if not automated)requires care.The plies must be aligned properly to the desired angle and stacked in the prescribed sequence.Prepreg plies come with a backing material to keep them from sticking together on the rolls.This backing material must be removed to prevent contamination of the laminate.Care must be exercised when handling the material to prevent splinters from piercing the hands. Part lay-up(particularly when done by hand)can lead to air entrapment between plies.This creates difficulty when the part is cured because the air may not escape,causing porosity.Thus.thick parts are normally pre-compacted using a vacuum periodically during the lay-up. Some prepreg materials contain an excess of resin.This excess is expected to be"bled"away during cure.Bleeder plies are placed under the vacuum bag to soak up the excess resin.However,most cur- rent prepreg materials are "net resin"so no bleeding is required. Composite processing requires careful attention to tool design.The tools must sustain high pres- sures under elevated temperature conditions.The composite material has different expansion character- istics than most tooling materials,thus thermal stresses are created in the part and in the tool.Tool sur- faces are treated with a release agent to facilitate removal of the part after cure.Tools must also be pres- sure tight because autoclave processing requires application of a vacuum on the laminate as well as posi- tive autoclave pressure.Lastly,tool design must account for the rate of manufacture and the number of parts to be processed. Prepreg material is not fully cured.Curing requires application of heat and pressure that is usually performed in the autoclave.Autoclaves typically apply 85 psi(590 kPa)pressure up to 350F(180C). They can go beyond these values if required for other materials(such as polyimides),but they must be qualified for higher extremes.Autoclave size may limit the size of a part to be designed and manufac- tured.Very large autoclaves are available,but they are expensive and costly to run.Common problems 12-9

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-9 12.2.8 Handling and storage Epoxy resins are the most common form of matrix material used in composites. Epoxies are perish￾able. They must be stored below freezing temperature and even then have limited shelf life. Once the material is brought out of storage there is limited time it can be used to make parts (30 days is common). For very complex parts with many plies, the material's permissible out-time can be a controlling factor. If the material is not completely used, it may be returned to storage. An out-time record should be kept. In addition, freezer storage of these materials is usually limited by the vendor to 6 to 12 months. Overage material will produce laminates with a high level of porosity. The perishability of the material also requires that it be shipped refrigerated from the supplier. Upon arrival at the contractor's facility, there must be provisions to prevent it being left on-dock for long periods of time. Tack is another composite material characteristic that is unique. Tack is "stickiness" of the prepreg. It is both an aid and a hindrance. Tack is helpful to maintain location of a ply once it is placed in position. It also makes it difficult to adjust the location once the ply has been placed. 12.2.9 Processing and fabrication Composite parts are fabricated by successive placement of plies one after the other. Parts are built-up rather than machined down. Many metal fabrication steps require successive removal of material starting from large ingots, plates, or forgings. Prepreg "tape" material typically comes in rolls of relatively thin strips (0.005-0.015 inches or 0.13 - 0.38 mm). These strips are a variety of widths: 3", 6", and 36". Prepreg "fabric" is usually thicker than tape (0.007-0.020 inches or 0.18 - 0.51 mm) and usually comes in 36-inch (0.9 m) wide rolls. Fabrication of a detail part requires the material to be taken out of the freezer in a sealed bag and allowed to come to room temperature prior to any operations. Placement of the prepreg on the tool (if not automated) requires care. The plies must be aligned properly to the desired angle and stacked in the prescribed sequence. Prepreg plies come with a backing material to keep them from sticking together on the rolls. This backing material must be removed to prevent contamination of the laminate. Care must be exercised when handling the material to prevent splinters from piercing the hands. Part lay-up (particularly when done by hand) can lead to air entrapment between plies. This creates difficulty when the part is cured because the air may not escape, causing porosity. Thus, thick parts are normally pre-compacted using a vacuum periodically during the lay-up. Some prepreg materials contain an excess of resin. This excess is expected to be "bled" away during cure. Bleeder plies are placed under the vacuum bag to soak up the excess resin. However, most cur￾rent prepreg materials are "net resin" so no bleeding is required. Composite processing requires careful attention to tool design. The tools must sustain high pres￾sures under elevated temperature conditions. The composite material has different expansion character￾istics than most tooling materials, thus thermal stresses are created in the part and in the tool. Tool sur￾faces are treated with a release agent to facilitate removal of the part after cure. Tools must also be pres￾sure tight because autoclave processing requires application of a vacuum on the laminate as well as posi￾tive autoclave pressure. Lastly, tool design must account for the rate of manufacture and the number of parts to be processed. Prepreg material is not fully cured. Curing requires application of heat and pressure that is usually performed in the autoclave. Autoclaves typically apply 85 psi (590 kPa) pressure up to 350°F (180°C). They can go beyond these values if required for other materials (such as polyimides), but they must be qualified for higher extremes. Autoclave size may limit the size of a part to be designed and manufac￾tured. Very large autoclaves are available, but they are expensive and costly to run. Common problems

MIL-HDBK-17-3F Volume 3,Chapter 12-Lessons Learned that occur in autoclave operations include blown vacuum bags,improper heat-up rates,and loss of pres- sure. Once the part is cured it may still require drilling,trimming and machining.Drilling of composites re- quires very sharp bits,careful feed and speed,and support of the back face to prevent splintering.Wa- ter-jet cutters are very useful for trimming.Machining produces a fine dust that requires protection for the operator's safety. 12.2.9.1 Quality control The quality control function for composite materials starts at a much earlier phase than for metals. There is much coordination and interaction occurring between the material supplier and the user before the material is ever shipped.These controls are defined by the material and process specifications and in some cases design allowables requirements.The supplier is often required to perform chemical and me- chanical tests on the material prior to shipment.These involve the individual material constituents,the prepreg,and cured laminates. Material processing and handling must be monitored throughout the various manufacturing phases. Receiving inspections are performed on the prepreg and cured laminates when the material first comes in.From this time on the material is tracked to account for its shelf life and out-time. Quality control activities include verification of the ply lay-up angle,its position in the stack,the num- ber of plies,and the proper trim.During lay-up it is necessary to ensure all potential contaminates and foreign materials are not allowed to invade the material. The curing process is monitored to ensure proper conformance to time-temperature-pressure profiles. These records are maintained for complete traceability of the parts. After the part is cured,there are a number of methods to verify its adequacy.One of the most com- mon is Through-Transmission-Ultrasonics (TTU).Parts with high porosity or delaminations can not transmit sound as well as unflawed parts.Thus ultrasound transmission is attenuated in a flawed part. Other techniques used to verify part quality include traveler specimens,specimens cut from excess mate- rial on the part,tracer yarns within the laminate,and in some cases proof loading.Visual inspections, thickness measurements,and tap testing also serve to interrogate composite parts. One of the most crucial aspects of quality control is information on the effect of defects.It is not enough to discover a flaw or suspected non-conformity.There must also be sufficient information to evaluate the impact of that rejection.The quality control function in its entirety includes the dispositioning of exposed non-conformances.Dispositioning includes acceptance as-is,repair or rework,and scrap- page.If proper dispositioning is not possible because of a lack of knowledge about the effect of defects, an inordinate expense will be incurred scrapping or reworking affected parts. 12-10

MIL-HDBK-17-3F Volume 3, Chapter 12 - Lessons Learned 12-10 that occur in autoclave operations include blown vacuum bags, improper heat-up rates, and loss of pres￾sure. Once the part is cured it may still require drilling, trimming and machining. Drilling of composites re￾quires very sharp bits, careful feed and speed, and support of the back face to prevent splintering. Wa￾ter-jet cutters are very useful for trimming. Machining produces a fine dust that requires protection for the operator's safety. 12.2.9.1 Quality control The quality control function for composite materials starts at a much earlier phase than for metals. There is much coordination and interaction occurring between the material supplier and the user before the material is ever shipped. These controls are defined by the material and process specifications and in some cases design allowables requirements. The supplier is often required to perform chemical and me￾chanical tests on the material prior to shipment. These involve the individual material constituents, the prepreg, and cured laminates. Material processing and handling must be monitored throughout the various manufacturing phases. Receiving inspections are performed on the prepreg and cured laminates when the material first comes in. From this time on the material is tracked to account for its shelf life and out-time. Quality control activities include verification of the ply lay-up angle, its position in the stack, the num￾ber of plies, and the proper trim. During lay-up it is necessary to ensure all potential contaminates and foreign materials are not allowed to invade the material. The curing process is monitored to ensure proper conformance to time-temperature-pressure profiles. These records are maintained for complete traceability of the parts. After the part is cured, there are a number of methods to verify its adequacy. One of the most com￾mon is Through-Transmission-Ultrasonics (TTU). Parts with high porosity or delaminations can not transmit sound as well as unflawed parts. Thus ultrasound transmission is attenuated in a flawed part. Other techniques used to verify part quality include traveler specimens, specimens cut from excess mate￾rial on the part, tracer yarns within the laminate, and in some cases proof loading. Visual inspections, thickness measurements, and tap testing also serve to interrogate composite parts. One of the most crucial aspects of quality control is information on the effect of defects. It is not enough to discover a flaw or suspected non-conformity. There must also be sufficient information to evaluate the impact of that rejection. The quality control function in its entirety includes the dispositioning of exposed non-conformances. Dispositioning includes acceptance as-is, repair or rework, and scrap￾page. If proper dispositioning is not possible because of a lack of knowledge about the effect of defects, an inordinate expense will be incurred scrapping or reworking affected parts