Questions on Plasticity

1. One dimensional return mapping algorithm Assuming a material with both (lin-
ear) kinematic and isotropic hardening, use the following algorithm for calculating the
material response to the following strain history

ε(t) = 0.01 sin(t) t ≤ 30 (1)

(1) Data Base {εpn , αn , qn }.

(2) Given Strain εn+1 = εn + ∆εn .
(3) Compute an “elastic trial state”
tr
σn+1 = E(εn+1 − εpn ) (2)
tr
qn+1 = qn (3)
tr
αn+1 = αn (4)

(3) Test this trial state for yield

fntr = |σn+1
tr tr
− qn+1 tr
| − (σY + Hαn+1 ) (5)

If fntr ≤ 0 then step is elastic and (·)n+1 = (·)tr

n where (·) is one of {σ, q, α}; else, the
step is plastic. Let

fntr
∆γ =
E+H +K
tr tr tr
σn+1 = σn+1 − ∆γsign(σn+1 − qn+1 ) (6)
p
εn+1 = εpn + ∆γsign(σn+1
tr tr
− qn+1 )
tr tr
qn+1 = qn + ∆γKsign(σn+1 − qn+1 ) (7)
αn+1 = αn + ∆γ

(a)Assume that E = 30 × 106 psi, σY = 60 × 103 psi, H = 40 × 103 psi, K = 60 × 103

psi and choose a time increment to provide sufficient resolution. Make plots of σ vs. t,
α vs. t, q vs. t, and σ vs. ε.

This algorithm is known as a return mapping algorithm. It is based on a backward Euler

integration of the flow rules, the requirement that fn+1 = 0 during plastic flow, and the
tr tr
observation that sign(σn+1 − qn+1 ) = sign(σn+1 − qn+1 ).

(b)Prove this last “assertion” by subtracting (7) from (6) and making use of the facts
that ∆γ > 0, E + K > 0, and for any real number x = |x|sign(x).
2. Plastic Dissipation Consider a closed stress cycle σ(t) over a time period [0, T ] – i.e.
one where σ(0) = σ(T ), which also implies that εe (0) = εe (T ). Show that the dissipation
in the cycle which is defined as
Z T
Dcycle := σ ε̇ dt (8)
0

can be expressed as Z T
Dcycle = σ ε̇p dt . (9)
0
This allows one to conclude that the dissipation rate in an elasto-plastic solid is given
by σ ε̇p .

3. One dimensional combined hardening Consider a one-dimensional plastic response

with kinematic and isotropic (linear) hardening. Find the stress rate response in terms
of the strain rate.

4. One dimensional kinematic and isotropic hardening Consider the following model:

f (σ, q, α) = |σ − q| − (σY + Hα) (10)

q̇ = K γ̇sign(σ − q) (11)
ε̇p = γ̇sign(σ − q) (12)
α̇ = γ̇ (13)

(a) Find γ̇ during plastic flow

(b) Find σ̇ for γ̇ = 0 and γ̇ 6= 0.
(c) Plot σ(t), α(t), q(t), and σ versus ε. Assume

E = 30 × 106 (14)
σY = 60 × 103 (15)
H = 40 × 103 (16)
K = 60 × 103 (17)
ε(t) = 0.01 sin(t) (18)
0 ≤ t ≤ 30 (19)

Use the following algorithm:

σ tr = E(εn+1 − εpn )

If f (σ tr , qn , αn ) ≤ 0, then the time step is elastic and σn+1 = σ tr , qn+1 = qn , αn+1 = αn ,

and εpn+1 = εpn . Else,

Page 2
 f (σ tr , qn , αn )
∆γ = (20)
E+H +K
qn+1 = qn + ∆γKsign(σ tr − qn ) (21)
εpn+1 = εpn + ∆γsign(σ tr − qn ) (22)
αn+1 = αn + ∆γ (23)
σn+1 = E(εn+1 − εpn+1 ) (24)

5. Nonlinear isotropic hardening Modify the plasticity algorithm from Problem 4 to

handle the case of zero kinematic hardening (K = 0) and non-linear isotropic harden-
ing.
Stress Strain:
σ = E(ε − εp )
Flow Rules:

ε̇p = γ̇sign(σ) (25)

α̇ = γ̇ (26)

Yield Condition:
f (σ, α) = |σ| − (σy + H[1 − exp(−α)])
Kuhn-Tucker Conditions:
γ̇ ≥ 0 f ≤0 γ̇f = 0
Consistency:
γ̇ f˙ = 0
Note that your equation for the consistency parameter ∆γ will be non-linear, and you
will need to use an iterative method to solve for it at each time step – Newton’s method
is a good choice. Assume E = 30 × 106 psi, σy = 60 × 103 psi, H = 40 × 103 psi, and
ε(t) = 0.01 sin(βt) for 0 ≤ t ≤ 30 sec, where β = 1 rad/sec. Make three plots: σ versus
t, α versus t, and σ versus ε.

6. Nonlinear isotropic hardening Modify the plasticity algorithm given in Problem 4

to work for zero kinematic hardening (K = 0) AND non-linear isotropic hardening.
Stress Strain:
σ = E(ε − εp )

Flow Rules:

ε̇p = γ̇sign(σ) (27)

α̇ = γ̇ (28)

Yield Condition:
f (σ, α) = |σ| − (σy + H[1 − exp(−100α)])

Page 3
 Kuhn-Tucker Conditions:
γ̇ ≥ 0 f ≤0 γ̇f = 0

Consistency:
γ̇ f˙ = 0

Step (1) Develop an algorithm by applying a Backward Euler integration rule to the
flow rules. Step (2) show the sign of the trial stress is the same as the sign of the actual
stress. Step (3) enforce consistency at the end of the time step, if yielding is currently
taking place; i.e. require
f (σn+1 , αn+1 ) = 0
This provides a non-linear equation for ∆γ; solve it by using a Newton iteration loop.
Assume E = 30 × 106 psi, σy = 60 × 103 psi, H = 40 × 103 psi, and ε(t) = 0.01 sin(βt) for
0 ≤ t ≤ 30 sec, where β = 1 rad/sec. Make plots of σ and α versus time and σ versus ε.
7. 1D Nonlinear power-law hardening Consider the following 1-D plasticity model
with power-law hardening
σ = E(ε − εp ) (29)
f (σ, α) = |σ| − (σY + Hαn ) (30)
∂f
ε̇p = γ̇ (31)
∂σ
α̇ = γ̇ . (32)
Show that during active plastic loading that
Esign(σ)ε̇
γ̇ = . (33)
E + Hnαn−1
Use this result to find and expression for σ̇ during active plastic loading in terms of ε̇, α
and the material parameters E, H and n.
8. Duvaut-Lions Model The one-dimensional Duvaut-Lions model for viscoplasticity is
obtained from our rate-independent plasticity model by replacing the Kuhn-Tucker con-
ditions and the plastic strain rate expression by the single (constitutive relation):
1
ε̇p = [σ − σ̄]
µE
where µ is a fluidity parameter and σ̄ represents the solution to the rate-independent
plasticity equations. By considering the relation
σ̇ = E[ε̇ − ε̇p ]
develop an efficient algorithm for the integration of the Duvaut-Lions model that uses the
solution to the rate-independent case as a known quantity. Check that your algorithm
gives the correct limits: ∆t/µ → 0 elastic response and ∆t/µ → ∞ inviscid plastic
response. [Hint: Exactly integrate exponential terms, approximate other terms.]

Page 4
 9. Tresca versus Hencky-von Mises The state of stress at a point is given as
 
70 0 0
N
σ →  0 σ22 0 
mm2
0 0 −35

What are the algebraically largest and smallest values of σ22 for elastic behavior according
to Tresca’s yield condition? According to von Mises’ yield condition? Assume the
uniaxial yield stress is σY = 170 N/mm2 and that both yield criteria are calibrated
using a uniaxial test.

10. J2 Plasticity with combined hardening Consider the case of J2 plasticity. Compute
∂f
∂σij
for the case of linear kinematic plus linear isotropic hardening and determine γ̇.

11. Volume preservation of plastic flow Consider a von Mises yield condition f (σ) =
kσ 0 k − σY and associated flow. Show that the plastic strains are volume preserving –
i.e. show that tr[εp ] = 0.

12. Crack tip plastic zone For elastic bodies with cracks the stress fields near the crack
tips have singularities which indicates that yielding always takes place near the tip of a
crack for any applied loads. For the case of a crack loaded in tension, as shown below,
the stress field near the right-crack tip is given by
 
KI θ θ 3θ
σxx = √ cos 1 − sin sin (43)
2πr 2 2 2
 
KI θ θ 3θ
σyy = √ cos 1 + sin sin (44)
2πr 2 2 2
KI θ θ 3θ
σxy = √ sin cos cos (45)
2πr 2 2 2

where KI is a geometry dependent parameter called the stress intensity factor. The
coordinate origin is taken at the crack tip.
q
Assume a material governed by von Mises’ yield condition (f (σ) = kσ 0 k − 23 σY ).
Determine and plot the approximate shape of the plastic zone around the crack tip for
plane strain and generalized plane stress. You may assume that at the interface between
the elastic and plastic zones that the stress field given above is reasonably accurate. Also
you may assume that Poisson’s ratio ν = 0.3. Normalize all distances in your plot by
3KI2 /2πσY2 .

Page 5
 y

Crack Tips r
θ
x

Crack

13. Combined Kinematic and Isotropic Hardening Consider the following plasticity
model:

εij = εeij + εpij (46)

σij = Cijkl εekl (47)
r
0 2
f (σ, q, α) = kσ − qk − (σY + Hα) (48)
3
∂f
q̇ij = K γ̇ (49)
σij
∂f
ε̇pij = γ̇ (50)
σij
α̇ = γ̇ (51)

∂f
(a) Find σij

(b) Find γ̇ during plastic flow.

14. Non-associated flow Consider the following plasticity model, where q is a vector of
internal variables.

Page 6
 Stress-Strain:
σ = C : (ε − εp )
Flow Rules (not necessarily associative):

ε̇p = γ̇r(σ, q)
q̇ = −γ̇h(σ, q)

where r and h are given functions of stresses and internal variables σ and q. Yield
Function/Elastic domain:

f (σ, q)
Eσ = {(σ, q) | f (σ, q) ≤ 0}

Kuhn-Tucker Conditions:

γ̇ ≥ 0, f (σ, q) ≤ 0, γ̇f (σ, q) = 0

Consistency Condition:
γ̇ f˙(σ, q) = 0
1. Find an expression for γ̇ during plastic flow.
2. The elastoplastic modulus Cep is defined through the relation σ̇ = Cep : ε̇. Show
that 
 C if γ̇ = 0

Cep = C : r ⊗C : ∂∂f
 C − ∂f
 σ if γ̇ > 0
∂σ
: C : r + ∂f
∂q
·h

15. Plastic work based hardening Consider the following alternative characterization of
isotropic hardening. Let the yield function by given by

f (σ, W p ) = kσ 0 k − (σY + σ̂(W p ))

where W p is the total plastic work; i.e.

Z t
p dεp
W = σ(β) : (β) dβ
−∞ dβ
and σ̂(W p ) may in general be a non-linear function.
Assuming associated flow, ε̇p = γ̇∂f /∂σ, find an expression for γ̇ during plastic flow.
Notes:
1. It is OK to have the term dσ̂/dW p in your answer.
2. Recall Leibniz’s rule:
d g(t)
Z Z g(t)
dg df ∂h
h(t, s) ds = h(t, g(t)) − h(t, f (t)) + (t, s) ds
dt f (t) dt dt f (t) ∂t

Page 7
16. Drucker-Prager Consider the following yield function

f (σ) = kσ 0 : σ 0 k − (σY − k tr[σ]) , (61)

where k is a material constant. Assume an associative flow rule and find the plastic
strain rate.

17. Drucker-Prager Consider a Drucker-Prager yield surface:

f (σ) = kσ 0 k − (σY − Kp)

where σY and K are material constants and p = 31 σii is the presssure.

1. Assuming an associated flow rule ε̇pij = γ̇∂f /∂σij , show that

 0
σij

p K
ε̇ij = γ̇ + δij .
kσ 0 k 3

2. Assuming that the strain history εij (t) is known, find an expression for the consis-
tency parameter, γ̇, during plastic loading.
3. Assume a stress of the form
 
1 0 2
σ ∼ σY t  0 2 −1 
2 −1 3

and find the time t at which yield begins; assume that K = 41 .

4. Find the direction of the plastic strain rate at this moment.

18. Drucker-Prager Consider the Drucker-Prager yield condition

r
2
f (σ) = kσ 0 k − (σY − µp) ,
3
where σ 0 = [I − (1/3)1 ⊗ 1] : σ is the stress deviator, p = (1 : σ)/3 is the pressure, and
σY and µ are material parameters.

1. Find the flow direction assuming associative plastic flow ε̇p = γ∂f /∂σ; i.e. compute
∂f /∂σ.
2. Assume a value of µ = 1/3. If the stress is given by
 
1 0 3
σ ∼ σY · t ·  0 4 −2  ,
3 −2 3

find the time t at which yield begins.

Page 8
 3. Determine the direction of plastic flow at this moment.

19. Dissipation in Prandtl-Reuss model Consider Prandtl-Reuss plasticity where the

dissipation (rate) is given by D = σ : ε̇p .
1. Show that D = σ : ėp = σ 0 : ėp = σ 0 : ε̇p
2. By noting that Prandtl-Reuss
√ applies to isotropic materials, show that during plas-
tic flow γ̇ = σ 0 : ė/ 2kM . (kM is the calibration constant in von Mises condition.)

20. Dissipation rate for associated von Mises flow Show for perfect plasticity (i.e. no
hardening) that the rate of dissipation can be expressed as
r
2
D = γ̇ σY = ē˙ p σY
3
for the von Mises yield condition and associative flow.
21. Kinematic hardening Consider the following plasticity model:
εij = εeij + εpij (85)
σij = Cijkl εekl (86)
r
2
f (σ, q) = kσ 0 − qk − σY (87)
3
∂f
q̇ij = K γ̇ (88)
∂σij
∂f
ε̇pij = γ̇ (89)
∂σij
where Cijkl , K, and σY are known constants. Suppose for a known constant β and time
t ≥ 0, that  
2 0 0
σY 
σ 0 (0) → 0 −1 0 
3
0 0 −1
 
0 0 0
q(0) →  0 0 0 
0 0 0
and  
0 0 0
βt 
σ̇ 0 (t) → 0 −2 0 
3
0 0 2

(a) Show that σ 0 (t) remains on the yield surface ∀t > 0.

(b) Find an expression for q̇(t) in terms of known quantities and q; explicitly write out
the expression for q̇11 (t).

Page 9
22. Back stress Evolution Consider the model in Problem 13. At time t = 0 the state of
stress at a material point is
 
9000 0 0
σ→ 0 5000 0  psi (90)
0 0 3100
and for t ≥ 0 the state of stress is increasing at the constant rate
 
20 0 0
σ̇ →  0 45 0  psi/sec . (91)
0 0 40
The kinematic hardening constant is K√= 250 (psi) and the isotropic hardening constant
is H = 0. The yield stress σY = 40000 3 (psi). Assume, q = 0 at t = 0.
1. At what time t1 does the material at this point begin to yield?
2. At time t1 , what is q̇.
3. Set up three simultaneous nonlinear ordinary differnetial equations for the non-zero
components of q for time t ≥ t1 . Plot their solution for 12 (sec) beyond initial yield.
[Hint: (1) Use numerical integration. (2) Do not try and apply the return mapping
algorithm.]

23. Back stress evolution Consider the model in Problem 13. At time t = 0 the state of
stress at a material point is
 
38000 0 0
σ→  0 0 0  psi (92)
0 0 0
and for t ≥ 0 the state of stress is increasing at the constant rate
 
0 0 0
σ̇ →  0 1000 0  psi/sec . (93)
0 0 −1000
The kinematic hardening constant is K = 250 (psi) and the isotropic hardening constant
is H = 0. The yield stress σY = 40000 (psi). Assume, q = 0 at t = 0.
1. At what time t1 does the material at this point begin to yield?
2. At time t1 , what is q̇?
3. Compute the tragectory of the elastic domain in the π-plane. [Hint: (1) The center
of the elastic domain is governed by a system of non-linear ODES. (2) Do not try to
use the return mapping algorithm. The statement of this problem is stress driven so
it is more easily approached from a direct manipulation of the governing equations.
(3) Integrate the non-linear ODEs using the built-in time integrator in Matlab or
Mathematica.]

Page 10
24. Yield surface evolution with kinematic hardening An associative plasticity model
with linear kinematic hardening and a von Mises yield condition. At time t = 0 assume
the state of stress at a material point is
 
ζσY 0 0
σ→ 0 0 0  (94)
0 0 0
and for t ≥ 0 the state of stress is increasing at the constant rate
 
0 0 0
σ̇ →  0 c 0  , (95)
0 0 −c
where ζ = 0.95 and c > 0 is a given constant. Further assume, q = 0 at t = 0.
Qualitatively, but with reasonable precision, describe the time evolution of the center
of the elastic zone – i.e. the trajectory of q as a function of time in the π-plane (the
octahedral plane). Make sure to identify all important moments in time.
25. Upper Bound Theorem for Plastic Collapse Consider a body Ω that is subjected
to surface tractions ct̄i on ∂Ωt , where t̄i represents a load pattern and c is a scaling factor
for the load. Assume zero body forces. A classical question is to estimate the value of c
associated with gross plastic deformation in the body – the situation typically associated
with collapse.
In this problem you will prove the upper bound theorem for plastic collapse. I will
provide the steps and you need to provide the reasoning for each step. To start, first
note that the Cauchy-Schwarz inequality tells us that |A : B| ≤ kAk · kBk for any two
second order tensors.
We begin by assuming a (virtual) displacement field δw such that δw is zero on ∂Ωu .
Then it follow by the Cauchy-Schwarz inequality that
σ 0 : ε(δw) ≤ |σ 0 : ε(δw)| ≤ kσ 0 k · kε(δw)k , (96)
where the notation ε(v) ≡ (1/2)[∇v + ∇v T ] defines the symmetric gradient – i.e. the
strain field associated with a generic displacement field v.
Since δw is virtual we can arbitrarily assume that its symmetric gradient is traceless
(i.e. deviatoric), resulting in
σ : ε(δw) ≤ kσ 0 k · kε(δw)k . (97)
(a) Show how (97) follows from (96).
Assume now that the material obeys they von Mises yield condition. This implies
r
2
σ : ε(δw) ≤ σY kε(δw)k . (98)
3

Page 11
 (b) Show how (98) follows from (97).
We can now integrate (98) over the body to give
Z Z r
2
σ : ε(δw) dV ≤ σY kε(δw)k dV (99)
Ω Ω 3
It can then be shown that
Z Z r
2
c t̄ · δw dA ≤ σY kε(δw)k dV . (100)
∂Ωt Ω 3
(c) Show how (100) follows from (99). [Hint: weak equilibrium]

Equation (100), implies by simple division that the load factor can be no greater than a
specific value:
R q2
Ω
σ kε(δw)k dV
3 Y
c≤ R
∂Ωt
t̄ · δw dA
The utility of this result is very wide ranging. Pick any virtual displacement field and
the right hand side provides and upper bound to the load factor. By judiciously picking
virtual motions that are near the true collapse motion of system, one can obtain quite
good estimates for the collapse loads.
26. Plastic Torsion In the Prandtl stress formulation of torsion one has that

σ13 = Ψ,2 and σ23 = −Ψ,1 .

1. Show that if the material yields according to the von Mises condition, then
r
1
k∇Ψk − σY ≤ 0 .
3
p
[The 1/3 is NOT a typo.]
2. At the limitp load, every point on the cross-section will have yielded and thus
k∇Ψk = 1/3σY everywhere on the cross-section. This implies that the surface
Ψ(x1 , x2 ) has the property that its gradient has constant norm everywhere.
(a) For a torsion rod with circular cross-section, this implies that Ψ is inpthe shape
of a right-circular cone, where the height of the apex of the cone is 1/3σY R,
where R is the radius of the rod. Using this information compute the ultimate
torque (limit load) using the relation
Z Z
T = x1 σ23 − x2 σ13 dA = −x1 Ψ,1 − x2 Ψ,2 dA
A A

OR the relation Z
T =2 Ψ dA.
A

Page 12
 (b) Compute the ultimate torque (limit load) for rod with square cross-section a×a
using one of the two previous relations. Note that Ψ will be in the shape of
flat facets that must be pieced together to form a continuous surface, p
where on
each facet the norm of the gradient of Ψ has the same constant value 1/3σY .
The figure below depicts this situation. [Hint: You can perform the integral
over only a part of the cross-section and then multiply by an appropriate factor
to account for the entire cross-section.]

A common way of thinking about this surface is to imagine what shape a “heap”
of sand would take if you poured a bunch of sand on top of a board cut out
in the shape of the cross-section. This is known as the sand-pile or sand-heap
analogy.
3. (Extra) Compute the ultimate torque for a rectangular bar with cross-sectional
dimensions a × b, where b > a. The function Ψ looks like:

Page 13
27. Material testing The plane strain compression test is one of the better methods for
obtaining a compressive stress-strain curve for a metallic material. A schematic figure
is show below.

Page 14
 F

b w

x2

x1
F x3

The indenting dies have a width w and a breadth b. The specimen has a width w and a
thickness t. Recommended ratios for (w/b) and (b/t) for such a test are (w/b) > 6 and
2 < (b/t) < 4. As a load is applied the metal between the indenters is prevented from
moving in the e3 direction by the constraint from the unstressed material adjacent to
the indented region. This gives a state of (approximate) plane strain. Neglecting the
effects of elasticity and friction, show that during plastic deformation in this mode:
(a) the flow strength Y of the material is related to the mean pressure p = F/bw under
the dies by √
3
Y = p
2
and
(b) that the equivalent plastic strain
 
p 2 ti
ē = √ ln
3 tf
q
where ē = 23 ε̇p : ε̇p , and ti and tf are the initial and final thicknesses (under the dies)
˙ p

of the strip specimen. [Assume J2 plasticity for parts (a) and (b).]

Page 15
28. Plastically loaded thick-walled cylinder Consider a thick walled cylinder subject
to an internal pressure pi and an external pressure po ; internal radius ri and external
∂
radius ro ; plane strain conditions ∂z (·) = 0 and uz = 0; made of an elastic perfectly
plastic Tresca material with Tresca constant kT .

(1) Determine the expressions for all the stresses when the cylinder is entirely elastic.
(2) Find the condition on the applied pressures for initiation of yield.
(3) Determine the radial and hoop stresses when the elastic-plastic interface is located
at r = R.
(4) Determine the relation for computing the elastic-plastic interface radius when the
applied pressures are known.

[Note that the compatibility equation here is the same in the sphere problem; i.e. εrr =
(rεθθ ),r .

ro
For part 5 and 6, assume that po = 0 and ri
= 2.

(5) Plot σrr /kT , σθθ /kT , and σzz /kT vs. r/ri when pi =1/2kT .
(6) Plot σrr /kT and σθθ /kT vs. r/ri when R/ri = 3/2.

Page 16