3.6.4 · D4Spacecraft Structures & Systems Engineering

Exercises — Hooke's law in 3D — generalized stress-strain (tensor)

2,383 words11 min readBack to topic

Before we compute anything, let us fix every word and symbol so none is used before it is earned.

Throughout, when a material is named, use these values (typical aluminium alloy for spacecraft):


L1 — Recognition

1.1 In the symbol , which face of the cube is loaded and in what direction does the force point? Is equal to , and why?

1.2 A material has and all shear strains zero. Write down the volumetric strain and say in words what it means.

1.3 Using the compliance matrix reproduced above, which single entry tells you the Poisson coupling between a stress in and the strain it causes in ? State its value in terms of .

Recall Solutions — L1

1.1 First index : the face whose outward normal points along the -axis (the -face). Second index : the force on that face points along the -axis (). So is a shear on the -face pushing in . Yes, : if they differed the tiny cube would feel a net twisting moment and spin up infinitely fast — moment balance forbids it, so the stress tensor is symmetric.

1.2 . It is the fractional change in volume: a cube of this material swells by in volume.

1.3 The off-diagonal entry in the top-left block: . So picks up — the Poisson pull-in.


L2 — Application

2.1 A bar is loaded uniaxially with MPa, everything else zero ( GPa, ). Find , , .

2.2 A block is under equal triaxial tension MPa, no shear. Find and the volumetric strain .

2.3 Pure shear MPa acts alone (recall , and the engineering shear strain is ). Find . Confirm no normal strains appear.

Recall Solutions — L2

2.1 Only is nonzero, so . WHAT we did: dropped the two zero cross-stresses. WHY: uniaxial means the stresses vanish. The bar stretches along and thins each transverse way — the Poisson effect.

2.2 With all three normal stresses equal to MPa, .

=\frac{100\times10^6\,(1-0.66)}{70\times10^9}=4.857\times10^{-4}.$$ $$\varepsilon_{kk}=3\varepsilon_{xx}=1.457\times10^{-3}.$$ Note it is positive (tension swells volume) and small because $(1-2\nu)$ is small. **2.3** Shear is decoupled, so use $\gamma_{xy}=\tau_{xy}/G$ with $G=\dfrac{E}{2(1+\nu)}=\dfrac{70\times10^9}{2(1.33)}=26.316\times10^9$ Pa: $$\gamma_{xy}=\frac{60\times10^6}{26.316\times10^9}=2.28\times10^{-3}.$$ Normal strains come only from normal stresses in the compliance matrix; all normal stresses are zero here, so $\varepsilon_{xx}=\varepsilon_{yy}=\varepsilon_{zz}=0$. Pure distortion, no volume change.

L3 — Analysis

3.1 (Plane stress) A thin flat panel has (nothing pushes through its thickness). Given MPa, MPa, , find . Explain why it is nonzero even though .

3.2 (Plane strain) A long tunnel-like body is fully constrained along so . With MPa, MPa given, find the induced that the constraint must supply.

3.3 For the plane-strain body above, using the you found, compute .

The figure below contrasts the two cases. Left (plane stress): a thin panel is pulled by in-plane forces (black arrows) while its two flat faces are free; the red arrows show it thinning through the thickness — but . Right (plane strain): the same in-plane pull, but the top and bottom faces are clamped (red hatching) so the thickness cannot change — , and the clamps must supply a hidden . The accent colour marks, in each panel, the thing that is not zero.

Figure — Hooke's law in 3D — generalized stress-strain (tensor)
Recall Solutions — L3

3.1 Plane stress: , so MPa.

=\frac{1}{70\times10^9}[0-0.33(280\times10^6)]=-1.32\times10^{-3}.$$ *WHY nonzero:* no *force* pushes through the thickness, but the in-plane pulls squeeze the panel thinner through the Poisson effect. Zero stress $\ne$ zero strain. **3.2** Plane strain forces $\varepsilon_{zz}=0$. Set the $z$-strain equation to zero: $$0=\frac1E[\sigma_{zz}-\nu(\sigma_{xx}+\sigma_{yy})]\ \Rightarrow\ \sigma_{zz}=\nu(\sigma_{xx}+\sigma_{yy})=0.33(280\times10^6)=92.4\ \text{MPa}.$$ *WHY:* to stop the natural Poisson thinning, the walls must *push back* along $z$ with exactly this stress. The constraint manufactures a real stress out of nothing but geometry. **3.3** Now $\sigma_{kk}=\sigma_{xx}+\sigma_{yy}+\sigma_{zz}=200+80+92.4=372.4$ MPa. The Poisson pull-in on $\varepsilon_{xx}$ comes from $\sigma_{yy}+\sigma_{zz}=80+92.4=172.4$ MPa: $$\varepsilon_{xx}=\frac1E[\sigma_{xx}-\nu(\sigma_{yy}+\sigma_{zz})] =\frac{[200-0.33(172.4)]\times10^6}{70\times10^9}=2.04\times10^{-3}.$$ Compare: the induced $\sigma_{zz}$ adds extra Poisson pull-in versus a free panel, but $\sigma_{xx}$ still dominates, so $\varepsilon_{xx}$ stays positive (the body stretches along $x$).

L4 — Synthesis

4.1 (Pressure-vessel wall) A thin cylindrical propellant tank has hoop stress , axial stress , radial (plane stress). With MPa, m, mm, GPa, , find , , then the axial strain and hoop strain . See Thin-walled pressure vessels.

4.2 For the same tank, compute the von Mises equivalent stress (plane-stress form) and compare to a yield strength of MPa — does the wall yield? See Von Mises yield criterion.

4.3 Derive the bulk modulus purely from the isotropic law by loading with pure hydrostatic pressure , then plug GPa, for a number.

Recall Solutions — L4

4.1 Stresses first:

\sigma_z=\tfrac12\sigma_\theta=125\ \text{MPa}.$$ Axial strain (radial stress $=0$, so only $\sigma_\theta$ is the Poisson partner): $$\varepsilon_z=\frac1E[\sigma_z-\nu\sigma_\theta] =\frac{[125-0.33(250)]\times10^6}{70\times10^9}=6.07\times10^{-4}.$$ Hoop strain (its Poisson partner is $\sigma_z$): $$\varepsilon_\theta=\frac1E[\sigma_\theta-\nu\sigma_z] =\frac{[250-0.33(125)]\times10^6}{70\times10^9}=2.98\times10^{-3}.$$ *WHY the coupling:* in the thin wall the hoop and axial directions are perpendicular, so each acts as the *other's* Poisson partner. The big hoop stress tries to thin the wall axially (reducing $\varepsilon_z$ via the $-\nu\sigma_\theta$ term), while the smaller axial stress only mildly reduces the hoop stretch (the $-\nu\sigma_z$ term). That is why $\varepsilon_\theta$ ends up much larger than $\varepsilon_z$ — and why sizing bolted flanges must account for this coupled thinning. **4.2** $\sigma_v=\sqrt{250^2-250\cdot125+125^2}=\sqrt{62500-31250+15625}=\sqrt{46875}=216.5$ MPa. Since $216.5 < 270$ MPa, the wall does **not** yield — but the margin is only about $20\%$, tight for flight hardware. **4.3** With $\sigma_{xx}=\sigma_{yy}=\sigma_{zz}=-p$: each normal strain is $\varepsilon_{xx}=\frac1E[-p-\nu(-2p)]=\frac{-p(1-2\nu)}{E}$. Then $\varepsilon_{kk}=3\varepsilon_{xx}=\frac{-3p(1-2\nu)}{E}$. Bulk modulus is $K\equiv -p/\varepsilon_{kk}$: $$K=\frac{E}{3(1-2\nu)}=\frac{70\times10^9}{3(1-0.66)}=\frac{70\times10^9}{1.02}=68.6\ \text{GPa}.$$

L5 — Mastery

5.1 Starting from the strain-from-stress law , derive the shear relation and thereby prove from first principles.

5.2 (Energy) The elastic strain-energy density is (see Elastic strain energy density). For the uniaxial bar of 2.1 ( MPa, others zero), show and evaluate it in J/m³.

5.3 (Incompressibility limit) Show algebraically that as , the bulk modulus , and explain physically why is forbidden. Then compute how much changes if a rubber-like material has vs (take MPa).

Recall Solutions — L5

5.1 For the off-diagonal component , the Kronecker delta , so the whole trace term drops: Engineering shear is . By definition , so matching coefficients: WHAT/WHY: the delta term is the only thing linking shear to the volumetric part; killing it () isolates pure shear, and the factor-of-2 comes straight from the convention.

5.2 With only nonzero, the sum has just one surviving term (all other ):

=\tfrac12\,\frac{(140\times10^6)^2}{70\times10^9}=1.4\times10^{5}\ \text{J/m}^3.$$ (Cross terms like $\sigma_{yy}\varepsilon_{yy}$ vanish because $\sigma_{yy}=0$, even though $\varepsilon_{yy}\ne0$ — energy needs *both* factors.) **5.3** $K=\dfrac{E}{3(1-2\nu)}$. As $\nu\to\tfrac12$, the denominator $(1-2\nu)\to0^+$, so $K\to+\infty$: it takes infinite pressure to change the volume — the material is **incompressible**. If $\nu>\tfrac12$ then $(1-2\nu)<0$ and $K<0$, meaning squeezing would make it *expand*: a spontaneous instability, physically impossible for a passive solid. Hence $\nu\le\tfrac12$. Numbers with $E=10$ MPa: $$K(0.499)=\frac{10^7}{3(1-0.998)}=\frac{10^7}{0.006}=1.667\times10^{9}\ \text{Pa},$$ $$K(0.4999)=\frac{10^7}{3(1-0.9998)}=\frac{10^7}{0.0006}=1.667\times10^{10}\ \text{Pa}.$$ A tenfold approach of $\nu$ toward $\tfrac12$ makes $K$ ten times larger — the blow-up is steep.

Recall One-line self-check before you leave

Which convention pitfall bit you most — face-vs-force indices, -vs- shear, or plane-stress-vs-plane-strain? ::: All three are "you fixed the wrong one of a symmetric pair" — read the physics first, then choose which member is zero.