MIT - 16.20 Fall, 2002 Unit 9 Effects of the Environment Readings : Rivello 3.6, 3.7 T & G Ch. 13 (as background), sec. 154 specifically Paul A. Lagace, Ph.D. Professor of Aeronautics & Astronautics and Engineering Systems Paul A. Lagace ? 2001 MIT - 16.20 Fall, 2002 Thus far we have discussed mechanical loading and the stresses and strains caused by that. We noted, however, that the environment can have an effect on the behavior of materials and structures. Le t ’s first consider: Temperature and Its Effects 2 basic effects: ? expansion / contraction ? change of material properties Look, first, at the former: Concept of Thermal Stresses and Strains Materials and structures expand and contract as the temperature changes. Thus: ε T = α ( ? T) thermal temperature change strain coefficient of thermal expansion (C.T.E.) units: 1 degrees Paul A. Lagace ? 2001 Unit 9 - p . 2 MIT - 16.20 Fall, 2002 If these thermal expansions / contractions are resisted by some means, then “thermal stresses” can arise. Howeve r , “thermal stresses” is a misnomer, they are really … “stresses due to thermal effects ” -- stresses are always “ mechanical ” (we’ll see this via an example) --> Consider a 3-D generic material. Then we can write: T ? ε ij = α ij ? T ( 91 ) i, j = 1, 2, 3 (as before) α ij = 2nd order tensor The total strain of a material is the sum of the mechanical strain and the thermal strain . mechanical thermal M T ? ε ij = ε ij + ε ij ( 92 ) total Paul A. Lagace ? 2001 Unit 9 - p . 3 MIT - 16.20 Fall, 2002 ?( “ actual ” ) total strain ( ε ij ): that which you actually measure; the physical deformation of the part T ? thermal strain ( ε ij ): directly caused by temperature differences M ? mechanical strain ( ε ij ): that part of the strain which is directly related to the stress Relation of mechanical strain to stress is: M ε ij = S ijkl σ kl compliance Substituting this in the expression for total strain (equation 9-2) and using the expression for thermal strain (equation 9-1), we get: ε ij = S ijkl σ kl + α ij ? T ? S ijkl σ kl = ε ij ? α ij ? T We can multiply both sides by the inverse of the compliance … that is merely the elasticities : ? 1 S ijkl = E ijkl Paul A. Lagace ? 2001 Unit 9 - p . 4 MIT - 16.20 Fall, 2002 ? σ kl = E ijkl ε ij ? E ijkl α ij ? T This is the same equation as we had before except we have the thermal terms: E ijkl α ij ? T --> so how does a “thermal stress” arise? Consider this example : If you have a steel bar lying on a table and heat it, it will expand. Since it is unconstrained it expands freely and no stresses occur. That is, the thermal strain is equal to the total strain. Thus, the mechanical strain is zero and thus the “thermal stress” is zero. Figure 9.1 Free thermal expansion of a steel bar Paul A. Lagace ? 2001 Unit 9 - p . 5 MIT - 16.20 Fall, 2002 --> However, if the bar is constrained, say at both ends: Figure 9.2 Representation of constrained steel bar Then, as it is heated, the rod cannot lengthen. The thermal strain is the same as in the previous case but now the total strain is zero (i.e., no physical deformation). Starting with (in one direction): ε = ε M + ε T with: ε = 0 Thus, the mechanical strain is the negative of the thermal strain. Paul A. Lagace ? 2001 Unit 9 - p . 6 εεε MIT - 16.20 Fall, 2002 Stresses will arise due to the mechanical strain and these are the so-called “ thermal stresses ” . Due to equilibrium there must be a reaction at the boundaries. (must always have ∫ σ dA = Force for equilibrium) Think of this as a two-step process … Figure 9.3 Representation of stresses due to thermal expansion as two-step process expands due to ???? T, εεεε T Reaction force of boundaries related to mechanical strain, ε M ε M = - ε T ? ε total = 0 (no physical deformation) Paul A. Lagace ? 2001 Unit 9 - p . 7 MIT - 16.20 Fall, 2002 Values of C.T.E. ’ s Note : α ij = α ji ? Anisotropic Materials 6 possibilities: α 11 , α 22 , α 33 , α 12 , α 13 , α 23 ? ? T can cause shear strains not true in “ engineering ” materials ? Orthotropic Materials 3 possibilities: α 11 , α 22 , α 33 ? ? T only causes extensional strains Notes : 1 . Generally we deal with planar structures and are interested only in α 11 and α 22 2. If we deal with the material in other than the principal material axes, we can “have” an α 12 Transformation obeys same law as strain (it ’ s a tensor). Paul A. Lagace ? 2001 Unit 9 - p . 8 MIT - 16.20 Fall, 2002 2-D form: α ? αβ = ll βγ α σγ ασ ? ? ? ? α 11 , α 22 (in - plane values) ? α 12 = 0 (in material axes) 3-D form: α ij = ll ? j l α kl ?ik So, in describing deformation in some axis system at an angle θ to the principal material axes ….. Figure 9.4 Representation of 2-D axis transformation ~ y 2 y 1 ~ θ y 2 θ + CCW y 1 Paul A. Lagace ? 2001 Unit 9 - p . 9 MIT - 16.20 Fall, 2002 ? α ? 11 = cos 2 θ α 11 + sin 2 θ α ? 22 ? α ? 22 = s in 2 θ α 11 + c os 2 θ α ? 22 ? ? α ? 12 = cos θ sin θ ( α 22 ? α 11 ) ? ? only exists if α 11 ≠ α 22 [isotropic ? no shear] ? Isotropic Materials 1 value: α is the same in all directions Typical Values for Materials: Units: x 10 -6 /°F μin/in/ °F strain/ °F ? μstrain/ °F Paul A. Lagace ? 2001 Unit 9 - p . 1 0 Material C.T.E. Uni Gr /Ep (perpendicular to fibers) Uni Gr /Ep (along fibers) Titanium Aluminum Steel 16 -0.2 5 12.5 6 MIT - 16.20 Fall, 2002 Notes : ? Graphite/epoxy has a negative C.T.E. in the fiber direction so it contracts when heated Implication : by laying up plies with various orientation can achieve a structure with a C.T.E. equal to zero in a desired direction ? C.T.E. of a structure depends on C.T.E. and elastic constants of the parts E = EE = 0 (perfectly compliant) 1 C.T.E. = 5 2 examples 1) total α of C.T.E. = 2 structure = 5 E = E 1 C.T.E. = 5 2) total α of C.T.E. = 2 structure E = ∞ (perfectly rigid) = 2 Paul A. Lagace ? 2001 Unit 9 - p . 1 1 MIT - 16.20 Fall, 2002 ? α = α (T) ? C.T.E. is a function of temperature (see MIL HDBK 5 for metals). Can be large difference. Implication : a zero C.T.E. structure may not truly be attainable since it may be C.T.E. at T 1 but not at T 2 ! --> Sources of temperature differential (heating) ? ambient environment (engine, polar environment, earth shadow, tropics, etc.) ? aerodynamic heating ? radiation (black-body) --> Constant ? T (with respect to spatial locations) In many cases, we are interested in a case where ? T (from some reference temperature) is constant through-the-thickness, etc. ? thin structures ? structures in ambient environment for long periods of time Relatively easy problem to solve. Use: ? equations of elasticity ? equilibrium ? stress-strain Paul A. Lagace ? 2001 Unit 9 - p . 1 2 MIT - 16.20 Fall, 2002 Example 1 - 2-material bar Total deformation is zero, but there will be nonzero total strain in the aluminum and steel --> stress is constant throughout --> match deformations Example 2 - Truss --> use equilibrium for forces in each member --> match displacements at each node Paul A. Lagace ? 2001 Unit 9 - p . 1 3 MIT - 16.20 Fall, 2002 Example 3 - bimetallic strip Each metal has different α ’s . What will happen? Bending! --> Concept of self-equilibrating stresses Must always be in equilibrium. General equation is: ∫ σ dA = F where: F = externally applied force If F = 0, can we still have stresses? Yes , but they must be “ self-equilibrating ” (satisfy equilibrium in and of themselves): ∫ σ dA = 0 This is the case of free expansion (thermal) of structures with varying properties or spatially-varying ? T (we’ll address this in a bit) Paul A. Lagace ? 2001 Unit 9 - p . 1 4 MIT - 16.20 Fall, 2002 If α 1 > α 2 and ? T > 0 T T ε ij > ε ij 1 2 (we’ll see more about this bending later) Bimetallic strip used as temperature sensors! --> ? T varies spatially (and possibly with time as well, we analyze at any constant in time) Must determine ? T by looking at heat flux into structure. Three basic methods: ? induction ? convection ? radiation Paul A. Lagace ? 2001 Unit 9 - p . 1 5 ? ? MIT - 16.20 Fall, 2002 Convection most important in aircraft: Aerodynamic Heating look at adiabatic wall temperature 2 ? T AW = ?? 1 + γ ? 1 r M ∞ ? T ∞ 2 where: γ = specific heat ratio (1.4 for air) r = " recovery factor" (0.8 - 0.9) M ∞ = Mach number T ∞ = ambient temperature ( ° K ) T AW is maximum temperature obtained on surface (for zero heat flux) Note : @40,000 ft. M = 2 ? T AW = 230 ° F M = 3 ? T AW = 600 ° F ? (much above M = 2, cannot use aluminum since properties are too degraded) worse in reentry Paul A. Lagace ? 2001 Unit 9 - p . 1 6 MIT - 16.20 Fall, 2002 Source of heat here is from air boundary layer: q = h (T AW - T s ) surface temperature of body heat flux heat transfer [watts/M 2 ] coefficient (convective constant) (h is determined from boundary layer theory) Radiation Always important but especially in space (or at high temperature in atmosphere). --> 2 considerations 1. Emissivity --> surface emits heat Paul A. Lagace ? 2001 Unit 9 - p . 1 7 MIT - 16.20 Fall, 2002 q = - ε σ T s 4 surface temperature heat flux Stefan- Boltzmann constant emissivity (a material property) 2. Absorptivity heat q = α I s λ angle factor flux intensity of source absorptivity (a material property) Figure 9.5 Representation of heat flux impinging on structure angle of structure like the sun I s (intensity of source) Paul A. Lagace ? 2001 Unit 9 - p . 1 8 ? ? ? ? MIT - 16.20 Fall, 2002 Heat conduction The general equation for heat conduction is: ? T q i T = ? k ij T ? x j where: T = temperature [ °K] T ? watts ? q i = heat flux per unit area in i direction ? m 2 ? T ? watts ? k ij = thermal conductivity ? mK ? ° (material properties) T The k ij are second order tensors Paul A. Lagace ? 2001 Unit 9 - p . 1 9 MIT - 16.20 Fall, 2002 consider: Figure 9.6 Representation of structure exposed to two environments Environment 1 Environment 2 look at a strip of width dz: Figure 9.7 Representation of heat flow through infinitesimal strip of material T q z T T q z + ? q z dz ? z Paul A. Lagace ? 2001 Unit 9 - p . 2 0 MIT - 16.20 Fall, 2002 Do a balance of energy: T q z T ? ? ? q z T + ? q z ? ? dz = ρ C ? T dz ? ? z ? ? t heat flux heat flux change in stored - = on side 1 on side 2 heat in strip where: ρ = density C = specific heat capacity t = time this becomes: T ? ? q z = ρ C ? T ? z ? t Recalling that: T q z = ? k T ? T z ? z Paul A. Lagace ? 2001 Unit 9 - p . 2 1 MIT - 16.20 Fall, 2002 we have: ? ? ? k z T ? T ? ? = ρ C ? T ? z ? ? z ? ? t If k z T and ρ c are constant with respect to z (for one material they are), then we get: k T ? 2 T ? T z = ρ C ? z 2 ? t Fourier ’ s equation We call: k T z = t hermal diffusivity ρ C More generally, for 3-D variation: ? ? ? k T ? T ?? + ? ? T ? T ? ? ? T ? T ? ? = C ρ ? T ? x ? x ? x ? ? y ?? k y ? y ?? + ? z ? ? k z ? z ? ? t thermal conductivities in x, y, and z directions Paul A. Lagace ? 2001 Unit 9 - p . 2 2 MIT - 16.20 Fall, 2002 Bottom line : use these equations to solve for temperature distribution in structure subject to B.C. ’s T (x, y, z) gives ? T (x, y, z) Note : These variations can be significant Example: Figure 9.8 Representation of plate in space 4 q in sun side black space side q in = α I s - εσ T s 4 q out = εσ T s Paul A. Lagace ? 2001 Unit 9 - p . 2 3 MIT - 16.20 Fall, 2002 could get other cases where T peaks in the center, etc. Result : ? Interna l stresses (generally) arise if T varies spatially . (unless it is a linear variation which is unlikely given the governing equations). Why ? consider an isotropic plate with ? T varying only in the y-direction Paul A. Lagace ? 2001 Unit 9 - p . 2 4 ??? ??? MIT - 16.20 Fall, 2002 Figure 9.9 Representation of isotropic plate with symmetric y-variation of ? T about x-axis (for the time being, limit ? T to be symmetric with respect to any of the axes) Paul A. Lagace ? 2001 Unit 9 - p . 2 5 MIT - 16.20 Fall, 2002 At first it would seem we get a deformation of a typical cross-section A-B as: x y This basic shape would not vary in x. Note, however, that this deformation in the x-direction (u) varies in y. ? u ? ≠ 0 ? y ? shear strain exists! But , ? T only causes extensional strains . Thus, this deformation cannot occur. (in some sense, we have “ plane sections must remain plane ” ) Thus, the deformation must be: Paul A. Lagace ? 2001 Unit 9 - p . 2 6 MIT - 16.20 Fall, 2002 locally In order to attain this deformation, stresses must arise. Consider two elements side by side Undeformed Deformed ???? T ↑↑↑↑ greater These two must deform the same longitudinally, so there must be stresses present to compress the top piece and elongate the bottom piece Paul A. Lagace ? 2001 Unit 9 - p . 2 7 MIT - 16.20 Fall, 2002 Thus: ε x = ε x (x) ε y = ε y (y) This physical argument shows we have thermal strains, mechanical strains and stresses. self-equilibrating a ∫ ? a σ y dx = 0 b ∫ ? b σ x dy = 0 Causes Paul A. Lagace ? 2001 Unit 9 - p . 2 8 MIT - 16.20 Fall, 2002 Solution Technique No different than any other elasticity problem. Use equations of elasticity subject to B. C. ’ s . ? exact solutions ? stress functions recall: ? 4 φ = ? E α ? 2 ( ? T ) ? ( 1 ? ν ) ? 2 V ? etc. (see Timoshenko ) --> Does this change for orthotropic materials? NO (stress-strain equations change) We’ve considered Thermal Strains and Stresses , now let ’s look at the other effect: Paul A. Lagace ? 2001 Unit 9 - p . 2 9 MIT - 16.20 Fall, 2002 Degradation of Material Properties (due to thermal effects) Here there are two major categories 1. “ Static ” Properties ? Modulus, yield stress, ultimate stress, etc. change with temperature (generally, T ↑ ? property ↓ ) ? Fracture behavior (fracture toughness) goes through a transition at “glass transition temperature ” ductile → brittle T g (see Rivello ) Paul A. Lagace ? 2001 Unit 9 - p . 3 0 MIT - 16.20 Fall, 2002 Figure 9.10 Representation of variation of ultimate stress with temperature Figure 9.11 Representation of change in stress-strain behavior with temperature ductile as T increases) (generally, behavior is more --> Thus, must use properties at appropriate temperature in analysis MIL-HDBK-5 has much data Paul A. Lagace ? 2001 Unit 9 - p . 3 1 MIT - 16.20 Fall, 2002 2. “ Time-Dependent ” Properties There is a phenomenon (time-dependent) in materials known as creep . This becomes especially important at elevated temperature. Figure 9.12 Representation of creep behavior Hang a load P and monitor strain with time resulting strain-time behavior This keeps aluminum from being used in supersonic aircraft in critical areas for aerodynamic heating. Paul A. Lagace ? 2001 Unit 9 - p . 3 2 MIT - 16.20 Fall, 2002 “ Other ” Environmental Effects ? Temperature tends to be the dominating concern, but others may be important in both areas – atomic oxygen degrades properties – UV degrades properties – etc. ? Same effects may cause environmental strains like thermal strains: Example - moisture Materials can absorb moisture. Characterized by a “swelling coefficient ” = β ij Same “ operator ” as α ij (C. T. E.) except it operates on moisture concentration, c: s ε ij = β ij c “swelling ” moisture concentration strain swelling coefficient Paul A. Lagace ? 2001 Unit 9 - p . 3 3 MIT - 16.20 Fall, 2002 and then we have: ε ij = ε ij M + ε ij T + ε ij S total This can be generalized such that the strain due to an environmental effect is: environmental strain E ε ij = χ ij X environmental environmental operator scalar and the total strain is the sum of the mechanical strain(s) and the environmental strains A strain of this “type ” has become important in recent work. This deals with the field of Paul A. Lagace ? 2001 Unit 9 - p . 3 4 MIT - 16.20 Fall, 2002 Piezoelectricity A certain class of materials, known as piezoelectronics , have a coupling between electric field and strain such that: electric field causes deformation/strain strain results in electric field This can be looked at conceptually the same way as environmental strains except electric field is a vector (not a scalar). Thus, the basic relationship is: piezoelectric ε ij p = dE k ijk where: E k = electric field d ijk = piezoelectric constant units = [strain/field] a key difference here is that the “operator ” (d ijk ) is a third-order tensor (how transform? 3 direction cosines) Paul A. Lagace ? 2001 Unit 9 - p . 3 5 MIT - 16.20 Fall, 2002 And we add this strain to the others to get the total strain (consider the case with only mechanical and piezoelectric strain) ε ij = ε ij M + ε ij p Again, only the mechanical strain is related directly to the stress: ε ij = S ijmn σ mn + d ijk E k inverting gives: σ ij = E ijmn ε mn ? E ijmn d mnk E k (watch the switching of indices!) thus we have “ piezoelectric-induced ” stresses of: E ijmn d mnk E k if the piezoelectric expansion is physically resisted. Again, equilibrium ( ∫ σ = F ) must be satisfied. But , unlike environmental cases, the electric field is not just an external parameter from some uncoupled equation of state but there is a coupled equation: Paul A. Lagace ? 2001 Unit 9 - p . 3 6 MIT - 16.20 Fall, 2002 D i = e i k E k + d inm σ mn note switch in indices since this is transpose of dielectric constant from previous equation where: e ik = dielectric constant D i = electrical charge --> “ Normally ”, when piezoelectric materials are utilized, “ E-field control ” is assumed. That is, E k is the independent variable and the electrical charge is allowed to “floa t ” and take on whatever value results. But, when charge constraints are imposed the simultaneous set of equations: σ mn = E mnij ε ij ? E mnij d ijk E k D i = e i k E k + d inm σ mn must be solved. This is coupled with any other sources of strain (mechanical, etc.) Paul A. Lagace ? 2001 Unit 9 - p . 3 7 MIT - 16.20 Fall, 2002 Piezoelectrics useful for ? sensors ? control of structures (particularly dynamic effects) Note : electrical folk use a very different notation (e.g., S = strain) Now that we ’ve looked at the general “causes ” of stress and strain and how to manipulate, etc., consider general structures and stress and strain in that context. Paul A. Lagace ? 2001 Unit 9 - p . 3 8