Professional Documents
Culture Documents
F.J. Schwan
Peliminary desian
Detailed desian
Member sizing - laminate design
Material specs
Manufacturing development
- I
Engineering
development
tests
- materals
- critical elements
- inspection techniques
Design - analyses
verification - prototypes
- testing
Fig. 31.1 The design process. - non-destructive evaluation
ing materials, the content of each box in the Table 31.1 Design requirements for an aircraft sta-
chart varies with many factors, including the bilizer
materials of construction. The intent of this dis- ~~~~~~~~
of an aircraft component are shown in Tables The data required to support preliminary
31.1 and 31.2 respectively. The component design investigations include stiffness, den-
selected for this example is a vertical stabilizer. sity, strengths, fatigue characteristics, data at
The design requirements define maximum tol- elevated and low temperature, forms in which
erable deflections, imposed loads, a weight the material is available and associated mater-
goal, production rate, cost goal, service life ial prices. In addition, some experience with
and operating temperature extremes. In real- the material in the fabrication process selected
ity, the set of preliminary design requirements is required to support decisions concerning
would be much larger including, perhaps, production rates and costs. Generally, reliable
maintenance and inspection characteristics, published data can be used for preliminary
descriptions of qualification tests, center of design. Only when no data is available is it
gravity characteristics, and numerous others. necessary to generate data in this phase of the
design.
712 Design of structure with composites
The sequence of decisions in this design materials incur increased engineering develop-
case would typically be to make a preliminary ment, testing, and increased raw material costs,
decision on a manufacturing technique, fol- data from previous applications is quite valu-
lowed by selection of one or more materials able in performing cost trades and making
based on cost and availability. Once this has preliminary design decisions.
been done, the component is 'sized' for stiff- Table 31.3 suggests the ever-increasing
ness, strength and service life in each scope of structural applications for reinforced
candidate material. Material properties and composites. The requirements identified in
strengths are used for these calculations. this table are as diverse and varied as the
Material selection may depend on weight components. The result of the wide range of
minimization or life cycle cost minimization requirements imposed upon composite mate-
or a combination of the two criteria. rials, compounded by the ingenuity of
The result of this process is a material selec- material developers, has led to an over-
tion and a preliminary design. This 'design' whelming array of constituent materials.
possesses enough detail to support confidence Many of these materials are very specific sys-
that (1)the material selection is justified, and tems with a narrow range of applications.
(2) that the design concept is feasible in this Some materials are suited to numerous uses.
material. The set of mechanical properties and This multiplicity of materials and features has
strengths was, by no means, a complete or final produced a plethora of design data. In order
set of design values, but rather contains those to make material comparisons and selections
characteristics needed to support preliminary on a rational basis, the designer needs to
design calculations. These values are generally establish a clear set of requirements for the
the 'best available' from material supplier data, material. These requirements are sometimes
literature publications, data from other projects captured in a document which governs mate-
and data from material specifications. rial procurement. The aerospace industry
While the dominant design goal of aerospace typically refers to these documents as mater-
designers is usually to reduce weight, the dri- ial specifications.
ving requirements vary widely from one The situation in which the structural
product to another. Table 31.3 provides a syn- designer usually finds himself at the initiation
opsis of the diversity of these requirements of a design includes a broad and usually
across defense and commercial industries. This incomplete set of requirements and a moun-
table indicates that requirements are numerous tain of literature on a vast array of materials. A
and, therefore, the derived design objectives are design checklist, such as that contained in
diverse and numerous. Payoffs are achieved in Table 31.4, may be of some help in getting
the form of weight savings or greater service started in this situation. This checklist offers a
life. Sometimes, as in the case of carbon-carbon systematic approach to determining what
rocket nozzles, the use of composite materials information is needed at each step and sug-
results in a new level of performance which is gests which design team members need to be
difficult to quantify due to a lack of economic involved at various points in the process.
alternatives. Table 31.3 shows a very large The structural design team will employ a
range in quantified payoffs. Each of these prod- process which resembles this one in order to
ucts reflect a history of successive applications develop a preliminary design which specifies
and successive generations of the same applica- configuration, materials, manufacturing
tion. With each successive development effort, process, and establishes a basis of confidence
the industry has become better at understand- to proceed with final design. This basis com-
ing and defining requirements, material prises having answers or a plan to develop
characteristics, and payoffs. Since composite answers to all key questions.
Design process 713
c
a,
Y
a
8
.3
4
3
714 Design of structure with coinposites
Figure 31.2 shows the makeup of a typical preliminary design. The primary assumption
design team. For designs with composite mate- made in this section is that the materials of
rials, the team may be larger than design teams construction are laminated composites with
using other engineering materials. The focus of continuous fiber reinforcement. These materi-
the team is the individual with design responsi- als are typical of structural applications.
bility. Essential members are representatives of
manufacturing, tooling, and materials procure-
31.3.1 MATERIALS
ment. Others might include materials
suppliers, and specialists in testing and analy- The subject of material design data, and the
sis. In some product development activities, the closely related subjects of material characteri-
team includes subcontract managers or mater- zation and specifications, assume greater
ial procurement specialists who interact both importance for composites than for conven-
with other team members and with other com- tional materials. There are several primary
panies who also need to be considered as team reasons for this: lack of a single, comprehen-
members. More important than the list itself, is sive design database or source; greater
the early and continued involvement of all variability in measured values from one mate-
members. Interaction among these team mem- rial to another and from one lot of the same
bers produces the design. As shown in Fig. 31.1, material to another; differences between pre-
preliminary design begins with requirements dicted and measured component performance
definition and produces a design complete to introduced by variables associated with
the point of materials identification, fabrication 'workmanship' in the manufacture of the part.
approaches, size and shape, and preliminary The most basic reason, however, is the simple
lay-ups and joint designs. The development of fact that the engineering development process
this information requires the constant input encompasses design of both the material and
and review of all key team members. the structure. For these reasons, the designer
One feature of the design team which is of composite structure needs to focus more
unusual is the inclusion of the material sup- attention, and usually more resources, on the
plier in the design team. There are several subject of materials, than his counterpart
reasons for doing this. Sometimes, existing using only conventional materials.
materials do not 'fit the bill'. In these instances, Material data is required in order to make
further material development or modification key design decisions, first in the preliminary
may become part of the design process. The design phase and later in detailed design. The
material specification can be looked at as a type and extent of the data depends on the
blueprint for the material supplier to use in application, but, typically, data on the stiff-
design of the desired material. It is important nesses, strengths, and densities of candidate
to recognize the material supplier as a member materials is needed.
of the design team so that his inputs and Figure 31.3 shows relative ranges of specific
development efforts can be integrated into the tensile modulus for composites with various
design. Otherwise, some beneficial options fiber reinforcements. Specific modulus is a
may be excluded, both in terms of existing term which refers to the ratio of composite
materials and near term material development modulus to composite density. Specificmodu-
efforts which may be of value to the project. lus is a measure of stiffness per kilogram of
material, and is commonly used as an aid in
material comparisons. Figure 31.4 shows the
31.3 PRELIMINARY DESIGN
ranges of material costs for these same families
The discussions which follow concern the por- of materials. Both of these figures were con-
tion of the design process characterized as structed considering the entire range of
716 Design of structure with composites
300.00 T
cn 250.00 --
-aa
200.00 --
$g
-‘8 3 150.00 --
8
=a
En
u
z
s2. 100.00 --
u
8
2 50.00 --
0.00 I
I
I
I
I ,
I
I
EGLASS ARAMID T300 M I-M
FIEF! FlBER GRAPHITE GRAPHKE GRAPHITE
FER
Material
Fig. 31.3 Specific tensile modulus of various composite materials.
properties available from each specific mater- materials is usually made on the basis of per-
ial. These ranges are the result of including formance. Figure 31.4 considers only material
different forms of the material and the entire cost, which in some applications, is the major
range of laminate values, including quasi- portion of product cost. In other applications,
isotropic values on one end and unidirectional material cost is completely overwhelmed by
values on the other. manufacturing costs or other life-cycle costs.
Figure 31.4 suggests some general conclu- Material cost is typically about 30% of compo-
sions. First, there are three more or less distinct nent cost in aircraft parts, less than 10% in
regions of composites. The first group is glass- satellite components, and generally higher than
reinforced composites. The second is 30% in most commercial applications. It is
intermediate modulus graphite and aramid therefore critical that product cost analysis
fiber-reinforced composites. The third group include all important cost contributors.
includes high and ultra high modulus graphite Generally, ’you get what you pay for’ or,
fiber composites. Each of these types is suitable more precisely, ’you must pay for what you
for a distinct range of applications. A compari- need.’ However, the range of properties and
son of where the common metals fall on this costs within a family of materials is an indica-
figure provides a pretty good explanation of tion of the latitude which a structural designer
why aramid and graphite have replaced metals has when employing reinforced composites.
in applications where the market can ’pay for The upper end of each range of specific
performance’. Glass has replaced metals in modulus in Fig. 31.3 is established by values
those applications where some performance for a unidirectional laminate. While this is one
can be sacrificed for lower cost, increased envi- equitable basis for comparing reinforced mate-
ronmental resistance or longer life. rials to each other, it is not a good basis to use
The second conclusion suggested by Fig 31.4 for design calculations because so many
is that there is extensive overlap in the price of requirements drive the design towards multi-
aramid and intermediate modulus graphite directional reinforcement. Foremost among
fiber composites. Selection from this group of these is the need for properties in more than
Preliminary design 717
1000 -
900 --
800 --
P '0°--
600-- W HMSYMBOUZES HIGH MODULUS FIBERS
INCLUDING GY70, M60J, PlOO & P120
500.-
IMSYMBOUZESINERMEDIATEMODULUS
4
EI- 400-
FIBERS INCLUDING T40, T50, T650 AND
M40J
9 300.-
200 --
100 --
3
I-
Fig. 31.4 Costs of composite materials.
140
120
100
h
0
u) 80
3
#
5
t
s 60
40
20
one direction. Figure 31.5 shows the relation- forced composite materials. This table shows
ship of tensile modulus in x and y directions mechanical properties and strengths for four
and inplane shear modulus (xy) for a varying fiber-reinforced materials. These material
lay-up angle, 13.The same type of relationship properties and strengths are the basic data
exists for strengths and thermal expansion from which laminate values are calculated. As
coefficients. A design which maximizes modu- explained in Chapter 30, lamination theory
lus in one direction at the same time simply uses layer values to determine engi-
minimizes modulus in the direction transverse neering constants and to estimate strengths for
to this and minimizes inplane shear modulus any desired laminate. Materials testing is
as well. Strength in the transverse direction therefore conducted primarily at the layer
and inplane shear strength will be low in mag- level. Laminate test coupons are employed to
nitude also. These low properties and confirm predictions and 'workmanship'. The
strengths in the 'secondary directions' will exceptions to this statement are strengths
severely limit the ability of the laminate to which are laminate-specific, such as bearing
resist load and deformation in these direc- strength and interlaminar shear strength.
tions. Highly directional laminates are However, these are values which are impor-
therefore not appropriate for structures which tant to only specific regions of the component,
act as plates or shells, and for beams with sig- and generally do not enter into material selec-
nificant secondary loads. Highly directional tion decisions.
laminates also present severe design con- For preliminary design, laminate values are
straints in joint regions where loads are typically predicted from available ply-level
multi-directional in nature. values. Design calculations and trades are con-
Multi-directional reinforcement offers a ducted with these predicted laminate values.
more robust design because strengths and Later, in detailed design, predictions are con-
properties are dominated by fiber properties firmed with selective tests of design laminates.
in all inplane directions. Furthermore, lami-
nates with multi-directional reinforcement
Carpet plots
offer more desirable cure characteristics and
usually avoid high levels of residual stress, The infinite number of combinations of ply
internal damage or warpage in thin parts, thickness, ply angle and material reinforce-
induced by cooldown from cure temperatures. ment is both a curse and a blessing in
Successful instances of unidirectionally rein- preliminary design. Once the design require-
forced parts are nearly non-existent. ments have been identified and quantified, the
Table 31.5 provides a typical list of materials question of how to meet these requirements
data required for structural design in rein- with any given material presents itself. An
extremely useful tool in addressing this ques- forcement in each of three specified directions.
tion is the simple carpet plot. This graphical While 0, 45 and 90" plies are most commonly
representation provides a way to present the used, the three angles can be any selected val-
entire range of properties, or strengths, avail- ues. Another common combination is
able from a particular family of laminates. 0/30/60. Typically, values are calculated using
Figure 31.6 shows a carpet plot of tensile mod- a difference of 10%between successive points,
ulus for laminates comprised of layers of which requires about 1000 data points per set
T300-epoxy with orientations of 0,45 and 90". of curves. For this reason, creation of carpet
The plot contains all possible proportions of plots is done with the aid of computers.
the three directions, and is read as the legend Development of carpet plots of strength
indicates. This type of plot treats laminates as values is less straightforward. The primary
homogeneous combinations of various per- complication is the definition of failure of the
centages of layers. Carpet plots do not laminate. An extensive discussion of laminate
recognize lay-up sequence. Generally, this pre- failure modes is contained in Chapter 30.
sents no limitation, but for laminates Calculation of laminate strength involves con-
composed of very few plies, or laminates sideration of progressive failure of each layer
employing significantly unbalanced stacking and construction of strength envelopes for
sequences, the assumption can lead to incor- combined states of loading. Construction of a
rect strengths. carpet plot requires an approximation of ply
Carpet plots of material properties are con- strength which is adequate for preliminary
structed directly from a table of values for trade studies and initial member sizing. The
many specific laminates. Values of moduli, recommended approach is to define laminate
Poisson ratios, and expansion coefficients are strength as the product of allowable fiber
calculated from lamination theory for a series strain and the laminate modulus in the appro-
of laminates with varying percentages of rein- priate direction. This calculation assumes that
140
120
-0- 0% 0 LAYERS
++ 10% 0 LAYERS
-C-20% 0 LAYERS
4 3 0 % OLAYERS
+40% OLAYERS
43- 50% 0 LAYERS
+60% OLAYERS
4 7 0 % 0 LAYERS
+80% OLAYERS
+90% 0 LAYERS
+100% 0 LAYERS
0 1 I
0 10 20 30 40 50 60 70 80 90 100
PERCENTAGE4M)EGREE LAYERS
the laminate contains fibers oriented in at least 427 MPa and 15.5 GPa, respectively. If this set
two directions, and further ignores the Poisson of properties were not suitable, the three carpet
effect in the calculation of strength. While this plots would be used to iterate to an optimum
approximation tends to overpredict strength, set of values and an accompanying laminate.
it has proven to be suitable for preliminary The use of carpet plots can greatly facilitate
laminate selection in strength critical struc- the laminate selection process in the early
ture. Using this definition of strength, carpet stages of design. It should be noted that these
plots can be directly constructed from allow- plots are simply visual representations of the
able tensile and compressive strain values, results of lamination theory calculations, using
and predicted modulus values. ply level input values. Carpet plots of all
Carpet plots provide a quick method for inplane material properties: elastic moduli,
selecting a candidate laminate for the prelimi- Poisson ratios, thermal expansion coefficients
nary design process. For example, let us and moisture expansion coefficients can be
assume that design calculations indicate that a developed from the appropriate equations.
modulus of about 90 GPa is required. The car-
pet plot of Fig. 31.6 offers a large number of
laminates which meet this single requirement. Design values
One is a laminate comprised of about 55% Appropriate values for properties and
0" layers, 35% 45" layers and 10%90" layers. If strengths are required in both of the primary
tensile strength and inplane shear stiffness design phases - preliminary and detailed. For
were important to the design, it would be nec- preliminary design, values are typically
essary to consider the values associated with acquired from existing sources. These can
the selected laminate. Related carpet plots pro- include data published or provided by mater-
vide ranges of tensile strength (Fig. 31.7) and ial suppliers in the form of brochures, design
shear modulus (Fig. 31.8). These are about handbooks, data sheets, data published in
'0° T
-D- 0% 0 LAYERS
2- 600
A
X- 10% 0 LAYERS
+20% OLAYERS
I
&w 500 -A- 30% 0 LAYERS
+40% 0 LAYERS
a +50% 0 LAYERS
b
w
400
+60% 0 LAYERS
d -A-70% 0 LAYERS
v)
300 +EO% 0 LAYERS
t-
+90% 0 LAYERS
200 +loo% 0 LAYERS
100
0 1
0 10 20 30 40 50 60 70 80 90 100
PERCEMAGE W E G R E E LAYERS
40 T
t
I
0 0 0 0 0 0 0 0 0
7 01 r n d In (D b 00 : g
F
PERCENTAGE&DEGREE LAYERS
professional society proceedings or publica- Test methods also tend to vary from one lab-
tions, data from textbooks such as this one, or oratory to another. In general, it is best to be
data developed on previous projects within wary of specimens other than the standard
the organization. An abbreviated list of ones. Standard coupons and test methods are
sources of preliminary design data is pre- defined by the American Society for Testing
sented at the end of this chapter. and Materials (ASTM), and other organiza-
As might be expected, this available data tions such as Suppliers of Advanced
comes in various forms, so comparison of data Composite Materials Association (SACMA).
from different sources may require some judg- The primary value of employing standard
ment and mathematical manipulation. Fiber coupons and tests is the confidence established
volume fraction usually varies from one with time and existing databases. While each
source to another. The data is sometimes nor- standard coupon may possess some shortcom-
malized to a 'standard' value, typically about ing, it does offer consistency and a basis of
60%. This normalization process is rarely comparison with existing data.
explained in the literature, but most often con- Fiber volume and coupon configuration are
sists of simply multiplying measured values only two variants present among published
by the ratio of 0.60 to whatever fiber volume data. Others include the resin system, material
was measured. The values most frequently form, and method of fabricating panels from
normalized include tensile and compressive which specimens are taken. These variants
modulus, and tensile and compressive underscore the reasons why existing data can
strength. The validity of normalizing compres- be useful for preliminary design, but almost
sive strength is debatable because always needs to be augmented or replaced by
compression strength of reinforced composites data specific to the design application for final
is dependent, to some extent, on the matrix. design. Two different approaches are
The validity of normalizing values other than employed in the development of design val-
the four mentioned above is dubious. ues. The first of these is to buy existing
722 Design of structure with composites
commercial grade composite material, charac- displacement and rotation at both ends, and
terize it, and design to those values. The loaded by a uniform load along the entire
second is to include the material supplier in the length, I, as shown in Fig. 31.9. Design require-
process by jointly developing a specification to ments are summarized below:
which the delivered material must comply.
While the second approach may increase mate- w = 2.0 nVmm
rial costs, it guarantees the suitability of the A
material and usually results in the use of S h
higher or more specific design values.
I = 1500 mm
A material specification typically defines 4
acceptable values of all critical characteristicsof
the material to be purchased. The specification
may be a simple one page document speclfylng Fig. 31.9 Example: trade study of beam with a uni-
reinforcement, resin system, and limiting val- form load.
ues of fiber volume, or a multiple page
document including minimum values of mod-
1. Maximum beam weight is not to exceed
ulus, strengths, cure characteristics, and other
5.0 kg or 3.33 g/mm excluding end fittings.
items. Material procurement cost will tend to
2. Maximum beam deflection (at midspan)
follow the number and stringency of the criteria
must be less than 2.5 mm under the load
specified, since the material supplier will need
shown in Fig. 31.9.
more quality assurance activity before delivery
3. Maximum stresses in the beam must not
and may need to perform additional material
exceed allowables, including a factor of
development.The more demanding a specifica-
safety of 2.0.
tion becomes, the greater the need to identify
4. The cross section of the beam must be
and work with one (or more) material suppliers
closed and rectangular as shown.
in its development. One reason for this is sim-
ply to ensure the development of a specification Four materials will be considered for this
which can be met by at least one supplier. A application: aluminum (7075 alloy), E-
good specification needs to recognize the nor- glass/ epoxy composite, T300 graphite / epoxy
mal variation in material characteristics from composite, and M60J graphite/epoxy compos-
one manufactured lot of material to another. ite. Formulas for deflection, maximum
Data illustrating this lot-to-lot variation will be compressive stress, and beam weight, in
shown and discussed in a later section of this pounds per inch, are presented below:
chapter.
Deflection: d = (w14/384EI)
where €I = (Ebth2)/2
31.3.2 DESIGN EXAMPLE
Max. stress: f, = M c / I
Key aspects of the preliminary design process f, = ( M / W
have been described in this chapter. An exam- = (wL2/4hbt)
ple, consisting of the design of a composite
Weight: W = v(2h+2b)t < 3.33 g/mm
beam, is included here to illustrate the
methodology typically employed in prelimi- One additional constraint must be introduced to
nary design, and to provide a demonstration the design. Buckling of one side of the rectangu-
of a trade study performed for the purpose of lar section is governed by elastic moduli and by
material selection. the width-to-thickness ratio of that side. The
The structural element selected is a beam of crippling strength of a side can be sigruficantly
constant cross section, fixed against lower than the laminate compressive strength or
Detailed design 723
inplane shear strength. In order to prevent com- However, the cost of a beam fabricated in T300-
pressive or shear buckling of the sides and epoxy may still exceed the cost of an aluminum
flanges, a rule of thumb is to limit the side beam, despite the fact that less T300 material is
dimension, h, to no more than twenty times the required. The next calculation might be to esti-
thickness, and to limit the flange dimension, b, mate the fabricated costs, including materials,
to no more than fifteen times the thickness. of the T300 and aluminum designs, to ascertain
Using these relations, h and b are eliminated the cost differential. A final comparison can be
from the equations and replaced by multiples of made between T300 and M60J fiber reinforced
t. The three equations are then solved by itera- beams. As presented in Table 31.6, the addi-
tion for each of the four materials using tional weight savings associated with an M60J
properties and strengths shown in Table 31.6. beam is relatively small. The cost differential is
The summary table presents a comparison of going to be substantial because M60J prepreg is
the calculated design thicknesses and corre- over $200 per kilogram while T300 prepreg is
sponding beam weights for each of the four about $20 per kilogram. Only a substantial
candidate materials. The deflection requirement payoff in increased component life or overall
proved to be the dominant requirement for all system performance would justify the selection
four materials, i.e. the thickness required to sat- of M60J over T300.
isfy the deflection requirement was greater, in In a real design situation, the trade would
each case, than that required to limit imposed probably be more complex. Perhaps, the load
stresses to acceptable values. The estimated pre- would be a cyclic load or an intermittent load,
liminary design weight of each design is less leading to consideration of fatigue allowables.
than 5 kg, as required. Or, perhaps the natural frequency of the beam
Several conclusions can be drawn from this is another important design consideration in
comparison. First, the aluminum and E-glass the structural system. The trade methodology
designs are comparable on a weight basis due remains the same, but additional characteris-
to the fact that the lower density of fiberglass tics would need to be considered.
offsets its lower stiffness. End fittings have
been ignored, along with numerous other
31.4 DETAILED DESIGN
details, in this preliminary study, but it is
worth noting that the additional weight intro- Many of the activities which constitute
duced by metallic end fittings would probably detailed design are extensions or iterations of
cause the E-glass detailed design to be heavier activities already performed in preliminary
than the aluminum design. The T3OO/epoxy design. However, a few topics require special
design offers a significant weight savings over attention. This section will focus on three areas
aluminum and E-glass, deriving primarily specifically. These include a discussion of the
from its higher modulus-to-density ratio. development and use of material design data,
the approach to joining and the development measurement of this property, and the
of joining details, and the construction of expected variation in this characteristic.
drawings. Differences in design values and measured
strengths are larger. However, preliminary
design values are all exceeded by average mea-
31.4.1 USE OF MATERIAL DESIGN DATA
sured values, confirming the fact that strength
The first of these three areas is an extension of critical regions are conservatively designed.
the discussion of material data introduced in There are some differences in values mea-
the section on preliminary design. The point sured from one lot of material to another. The
was made in that section that material data kind and amount of variation is described in
relies heavily on ply-level values. A typical set more detail in Figs. 31.10 and 31.11. Specified
of mechanical material data is shown in Table 'acceptance values' are shown on each figure
31.7. The material of interest is T50 graphite for reference. In this case, these acceptance val-
fiber in an epoxy resin system. The material ues are lower than those used for preliminary
was ordered in significant quantity to an aero- design. The figures also show results of tests
space specification and was manufactured in conducted by the material supplier and by the
ten lots. Table 31.7 is subdivided into material receiving company. Differences in these results
properties and strength values. The values provide some insight into the influence of test
shown as 'measured' are average for all speci- methods, facilities, and personnel. The first
mens from all ten lots of material. The table figure compares tensile modulus and the sec-
also includes preliminary design values for the ond tensile strength. Figure 31.10 indicates
sake of comparison. that, while values are different from lot to lot,
This comparison shows excellent agreement all lots exceed the specification value for ten-
between measured and preliminary design sile modulus and are therefore acceptable. The
material properties. The largest differences are measured ply-level data confirms the design
in coefficient of thermal expansion (CTE) val- values of tensile modulus. The measured val-
ues, and these are on the order of 20-25%. This ues for tensile strength exhibit larger
is excellent agreement considering the difficul- variations, but are also higher than the speci-
ties and uncertainties associated with the fied value. Once again, all measured values
Table 31.7 Mechanical design values for T50/epoxy tape: preliminary and measured layer properties and
strengths
300 T
iii 250
n
9
;
v)
-
200
0
I
W LMSC DATA
d 150 HEXCEL DATA
6
I- +LMSCSPEClFlCAl-ION VAI-UE
0
1 2 3 4 5 6 7 8 9 10
LOT NUMBER
Fig. 31.10 Lot-to-lot variation in tensile modulus - T50/F384 unidirectional tape (manufactured by
Hexcel).
1400 1
g 1200
n
3
I
6 1000
E
F 800
HEXCEL DATA
d mLMSC DATA
u)
z 600
+
W +LMSC SPEClFlCPiTlON VAI.LIE
0
1 2 3 4 5 6 7 8 9 10
LOT NUMBER
Fig. 31.11 Lot-to-lot variation in tensile strength - T50/F384 unidirectional tape (manufacturedby Hexcel).
726 Design of structure with composites
exceed specification levels and all lots are database for establishing preliminary design
acceptable. values for the next application.
This data is typical of aerospace applica- Table 31.8 contains a further comparison of
tions. The types and amounts of data will, of predicted and measured laminate values for
course, depend on the specific application. T50-epoxy tape, developed in the same appli-
The major difference between this data and cation. The values in this table correspond to
preliminary design data is that the data sup- specific design laminates. Three sets of values
porting final design will be specific to the are shown for each of the design laminates -
material form, resin system, and fiber volume the first developed from preliminary design
selected. Final design data is an extension of layer values, the second from measured layer
data available for preliminary design. It will values, and the third directly from tests per-
generally focus on ply level values since this is formed on laminate coupons. Using measured
the most cost-effective way to obtain data for layer properties produces a better correspon-
all of the laminates which will be employed in dence between predictions and measurements,
final design of the part or assembly. These val- but not an exact overlay. The primary differ-
ues now become an important part of the ence is workmanship, i.e. changes introduced
Table 31.8 Comparison of measured and predicted material design values for T50/F584 graphite epoxy
tape laminates
by handling and fabrication that simply are allowables, and so provides maximum lever-
not accounted for by theory. These differences age in terms of applying testing resources to
are larger for strengths than for properties, development of design confidence.
and account for the knockdown factors nor- A less extensive series of laminate tests
mally used in developing material design may be required in addition to the unidirec-
strengths. tional coupon testing. The purpose of these
One final point to be made concerning these tests would be to provide some selective con-
comparisons is that differences exist, but are firmation of key design properties and
not typically an inhibitor to designing with strengths for comparison with values pre-
composites. These variations must be recog- dicted from measured layer values. In
nized in the design process, and can be addition, some very specific strength charac-
handled with statistically-derived knockdown teristics, such as bearing strength, are specific
factors, and conservative but realistic design to laminates, and cannot be reliably predicted
approaches. Key design values can also be from layer values. If this allowable is critical
controlled through development of intelligent to the design, tests of specific design lami-
material specifications. In most designs, the nates may be necessary.
usual variations are easy to accommodate. The
important point is that they need to be quanti-
31.4.2 JOINING
fied and included in the design process.
This leads to a discussion of the role of test- Joining of composites normally presents some
ing in the development of design allowables. It challenges and tends to influence the configu-
was previously pointed out that normally little ration of the part. The two recognized
or no testing is performed in support of pre- methods for joining composite structure to
liminary design. In fact, the only testing which other composite or to metallic parts are adhe-
is essential is that performed to determine crit- sive bonding and mechanical fastening. The
ical values which are just not available. advantages and disadvantages of each method
In the process of detailed design, however, will be discussed, followed by some guide-
the picture changes. Once a material is lines for selection between the two. The
selected and a design acquires some maturity, discussion is not an exhaustive one by any
the values critical to establishing sufficient means, but is intended to highlight key con-
confidence in the final design can be identified siderations which determine joint designs and
and addressed through test. The emphasis is provide some initial direction to the designer.
usually placed upon layer level coupons for Numerous factors need to be considered in
two reasons. The first is that the information the selection process. Figure 31.12 offers a list
obtained from unidirectional, standard of the most common requirements. It should
coupons is the most direct information avail- be kept in mind that some of these 'require-
able about the material. Failure modes are ments' may become design variables in the
predictable and repeatable, and results can be course of the process. Geometry of the mem-
directly converted to allowables. An addi- bers being joined, for instance, could be
tional benefit deriving from the use of altered locally to facilitate joint design.
unidirectional coupons is that direct compar- Reliability potentially includes an array of
isons to other materials and different lots of requirements, one of which is the implications
the same material can easily be made. of joint failure on system performance. Is it
The second reason for allocating the major- catastrophic or benign? Is it easily detectable
ity of testing resources to ply-level testing is prior to failure or insidious?
that this data can be used for prediction of a Determination of this list of requirements
wide range of laminate properties and will lead directly to the formulation of key
728 Design of structure with composites
How important
i
,
is weight?
How critical
is cost?
Reliability
t Service life
Implications of
fai Iure
and so the joint geometry is established prior other sections, and so this discussion is aimed
to finalization of the adhesive joint design. at providing a brief summary of the process.
Factors including how well the mating parts The process begins with the determination
will 'fit up' to each other after allowances for of a configuration for the joint. Single lap joints
all specified dimensional tolerances, and the are normally adequate for thin laminates (up
quality of the adherend surfaces need to be to about 5 mm (0.2 in) in thickness). Fastener
recognized and defined. bending and initial bearing failure are primary
The first step in the process is to determine concerns. Double lap joints are better for cyclic
a dimensional configuration which minimizes (fatigue) loads, and generally stronger.
tensile and peel stresses in the bonded joint. The next step is to select the type of fastener
Once this is accomplished, the next task is to required. Fastener selection usually raises
select an adhesive system which best satisfies issues requiring decisions concerning laminate
static strength, fatigue life, and environmental reinforcement, hole sizes and locations,
requirements. This selection is sometimes the drilling, fastener installation, and inspection.
result of a simple trade study among several Table 31.9 identifies issues and proven design
candidates. The third step is the development approaches to each issue. The table reveals
of formal or informal process specifications for that the complexity of designing bolted joints
the joint. This specification will include details derives from two primary sources: (1) com-
for surface preparation, curing the joint, and posite laminates cannot redistribute high local
maintaining pressure during cure if necessary. loads by yielding and plasticity; (2) compos-
The joint strength is typically verified analyti- ites are more easily damaged by drilling and
cally or by structural test or both. fastener installation than metals.
Similarly, a design process can be outlined Design of local reinforcement of the lami-
for mechanical joints. The primary design con- nate to resist local stresses is an important step
siderations for bolted joints include joint in the design of bolted joints. If reinforcement
strength, fastener type, local reinforcement, is required, a proven approach is to increase
joint configuration, holes, and preload. All of laminate thickness by the addition of plies
these subjects are discussed extensively in placed at 345 and 90" to the primary load
Local reinforcement
As in metal structures, local reinforcement is
,
w
(1 .O)
Scale: 118
generally required where any hole or cutout is See view B view
placed in a structural part. Analogous to metal 0.10 (reference). .- .K-
, _ f ~ . 2(reference)
~
_f I
parts, reinforcement can be bonded or fas- I
drawing are unique. There are countless varia- table defining the material: fiber reinforce-
tions on the format of a drawing, so the ment, resin system, material form, resin
discussion will be more or less limited to the content, material source or sources, and sur-
content, with a specific format shown only for face treatments. The second is to invoke a
the purpose of providing an example. document, such as a specification, which
A drawing detailing a composite part needs defines the material required. If the second
to specify geometry, materials, sources of approach is used, it may still be a good idea to
those materials, ply sequence for each unique provide a brief identification of the material
laminate, details of transition regions, process- on the face of the drawing to facilitate inter-
ing definition, post fabrication instructions for pretation by analysts, tool designers,
drilling, sawing, etc. and any inspection manufacturing and inspection personnel.
requirements. Other details, such as individ- Table 31.10 shows an example of material
ual ply patterns, may be necessary for a description employing both methods.
specific part, but this list is suggested as one of Specifying a fabric requires the designer to be
general applicability. familiar with all of the variable characteristics
Dimensions are defined in the same manner of the material form - weave style, surface
as for metal parts with the exception of thick- treatment, and end count. This is necessary
nesses. Thickness is really defined by laminate because fabrics are available in numerous
sequence meaning that the part will be as thick weave styles, with various surface treatments,
as the cured laminate dictates. A thickness and can be woven with any of several differ-
dimension is normally shown as a reference ent size yarns.
dimension and provided for tooling design. It Each unique laminate in the part requires
should be recognized that specifymg a thick- definition sufficient to allow a fabricator to lay-
ness as a fixed dimension demands up the part. Figure 31.15 presents a typical
compliance to that dimension. Unless manu- example of the usage of a ply sequence table on
facturing ‘tool try’ units have been fabricated the face of a drawing. The figure identifies the
to determine cured laminate thickness with surface that ply number 1 is placed upon and
the particular manufacturing process speci- then details the ply stacking sequence to be fol-
fied, there is no guarantee that the thickness lowed. It is desirable, but not always practical,
can be achieved with the number of plies and that the first ply shown in the table be the first
resin content specified. ply physically laid up on the tool. The ply table
Dimensional tolerances are generally deter- contains a column identifying the material of
mined by tooling design and the amount of each ply since sometimes a single laminate
material shrinkage. Experience with the contains plies of different materials.
process and material is the only reliable guide Areas of transition are handled as shown
in setting tolerances. in Fig. 31.16. The two adjacent regions of
Material definition can be accomplished in ‘constant’ thickness are defined by appropri-
two ways. The first is to create a note or a ate tables as shown. The tables, however,
anything else. These operations are normally The advantage of documenting each NDE
specified on the drawing by means of notes. activity is traceability. If a problem arises in
Notes need to identify items such as the type development or production, NDE records can
of drill bit and cutting speed. usually identify potential sources and trends.
The disadvantage is cost. Each company per-
forms its own cost/payoff determination, and
31.4.4 DESIGN VERIFICATION PROCESS
develops its own verification processes from
A few words about design verification are that determination. One point for the designer
appropriate to a section dealing with compo- to keep in mind is the necessity and value of
nent design. Design verification encompasses this verification for composite parts. Variability
more than the nondestructive evaluation in raw materials, processes, and workmanship
(NDE) procedures specified on the compo- can and will affect the final part. The designer
nent drawing. Design verification is integral needs to be aware of all potential sources of
to the entire design and manufacturing variation, so that he knows where to look when
process. Figure 31.17 identifies verification final inspection reveals defects in his part.
activities in each phase of the component Part inspection methods are usually identi-
development process. Each of these activities fied on the drawing. These requirements
ranges from a formal, documented step in actually form the accept/reject basis for the
some industries, to undocumented, highly finished part, and can range from visual only
integrated steps in other industries. The key to extensive ultrasound methods. Table 31.11
point is that verification is the part of the presents a summary of which defects are
development process which ensures that the detectable with each of several established
developed component fulfills all design NDE techniques. The most commonly used
requirements. methods are X-ray and ultrasonic, particularly
Development tests
Engineering Material
development Procurement Material acceptance tests
Tool tries
In-process inspection
F)
fabrication
Trim
Bonding
Drilling holes
Sawing
Witness panels
Dimensional verification
Non-destructive inspection
acceptance
Proof test / acoustic
emissions
C-scan. These methods require proper equip- SOURCES OF MATERIALS DATA FOR
ment, cumbersome setup in some instances, PRELIMINARY DESIGN
and time. X-ray and ultrasonic methods can
detect a wide range of defects, and are sensi-
tive in ranges where other methods are not. Military Specifications
However, these methods have definite limits. Mil-R-9300B Resin, Epoxy, Low Pressure
In fact, no single method can detect all types Laminating
of defects or variations. Moreover, it is impor- Mil-C-47257C Compound, Epoxy, Filament
tant to understand, in advance, what a defect Winding
looks like. For this reason, it is common prac- Mil-R-24719 Resins, Vinyl Ester, Low
tice to perform calibration with intentional Pressure Laminating
defects built into test panels. Mil-Y-1140 Yam, Cord, Sleeving, Cloth
Table 31.11 is an oversimplification of an
and Tape, Glass
extensive and complicated subject. It is Mil-R-60346 Roving Glass, Fibrous (for
included here as a handy reference for the
Filament Winding
designer and a starting point for determining Applications)
which techniques are appropriate for a given Mil-Y-83371 Yarns, Graphite, High
part. Inspection requirements will, of course,
Modulus, Continuous
be related to the function of the part, and will
Filament
be focused on the critical aspects of the design. Mil-Y-87125A Yam, Graphite, 1000/3000
In some companies, it is common practice to
Filaments
limit part inspection to simple dimensional Mil-F-87121A Fabric, Graphite Fiber
and visual checks, and to rely on a ‘proof’ test
Mil-T-29586/ 1 Thermosetting Polymer
to verify the structural adequacy of the com- (Epoxy) Matrix, 350°F (177°C)
ponent. Since inspection requirements tend to
Cure, Intermediate Modulus
be unique to each component, it is usually nec- Carbon Fiber Reinforced
essary to make inspection requirements an Prepreg Tape
integral part of the design drawing.
Sources of materials data for prelimina y design 735