At present, there are two generally accepted theories for predicting the onset of yielding in ductile metals:
- Maximum Shear Stress Theory or Tresca Criterion
- von Mises Criterion or Distortion Energy Criterion
1. Maximum-Shear-Stress Theory or Tresca Criterion
The maximum-shear-stress theory, sometimes called the Tresca yield criterion, states that yielding will occur when the maximum shear stress reaches a critical value k. In terms of principal stresses \sigma_1, \sigma_2, \sigma_3, we can write:
\max\left\{\frac{|\sigma_1-\sigma_2|}{2}, \frac{|\sigma_1-\sigma_3|}{2}, \frac{|\sigma_2-\sigma_3|}{2}\right\}=k \tag{1}If \sigma_1\geq \sigma_2\geq \sigma_3, we previously showed that, the maximum shear stress is given by
\tau_{\max} = \frac{\sigma_1 - \sigma_3}{2} \tag{2}.For uniaxial tension \sigma_1 = \sigma_{yp}, \sigma_2 = \sigma_3 = 0, where \sigma_{yp} is the yield strength in simple tension. Therefore, the shearing yield stress for simple tension \tau_{yp} is equal to one-half of the tensile yield stress.
\tau_{yp} = \frac{\sigma_{yp}}{2}\tag{3}Substituting these values into the equation for the maximum shear stress results in
\tau_{\max} = \frac{\sigma_1 - \sigma_3}{2} = \tau_{yp} = \frac{\sigma_{yp}}{2} \tag{4}or
\sigma_1 - \sigma_3 = \sigma_{yp} \tag{5}This is sometimes written as
\sigma_1 - \sigma_3 = \sigma_1' - \sigma_3' = 2k \tag{6}where \sigma_1' and \sigma_3' are the deviators of the principal stresses and k is the yield stress for pure shear, i.e., the stress at which yielding occurs in torsion, where \sigma_1 = -\sigma_3.
The maximum-shear-stress theory is in good agreement with experimental results, being slightly on the safe side, and is widely used by designers for ductile metals.
The Tresca yielding condition is:
f=\sigma_1-\sigma_3-2k=0,\tag{7}However, in certain plasticity problems this simple form cannot be used as the yielding condition since it is not known which of the three principal stresses is the largest. A proper yield function must not depend on the arbitrary labeling "1, 2, 3".
Tresca’s actual statement is:
Thus, instead of assuming a particular ordering, we must write a form that treats all three differences symmetrically.
Symmetric Algebraic Form of Tresca Criterion
Tresca requires that one (not necessarily all) of the following be equal to 2k:
|\sigma_1 - \sigma_2| = 2k, \quad |\sigma_2 - \sigma_3| = 2k, \quad |\sigma_3 - \sigma_1| = 2k.\tag{8}A compact way to enforce the condition “at least one of these equals 2k” is to write
f = \big[ (\sigma_1 - \sigma_2)^2 - 4k^2 \big] \big[ (\sigma_2 - \sigma_3)^2 - 4k^2 \big] \big[ (\sigma_3 - \sigma_1)^2 - 4k^2 \big] = 0. \tag{9}This expression is:
- symmetric in the principal stresses,
- equal to zero if any pair difference satisfies |\sigma_i - \sigma_j| = 2k.
Thus, (9) is a proper analytic representation of Tresca’s condition.
Reuss’ Invariant Form
To write the yield criterion in terms of stress invariants, Reuss transformed the symmetric form (9) into an invariant expression involving the second and third invariants of the deviatoric stress tensor:
- J_2 = \dfrac12 \sigma'_{ij}\sigma'_{ij} (second invariant of the deviatoric stress),
- J_3 = \det(\sigma'_{ij}) (third invariant of the deviatoric stress).
Reuss showed that the product expression (9) is equivalent to the polynomial:
4J_3^2 - 27J_2^3 - 36k^2J_2^2 + 96k^4J_2 - 64k^6 = 0. \tag{10}This is a fully invariant form of the Tresca criterion, valid without assuming any ordering of the principal stresses. Obviously, such a complex relation will result in very cumbersome mathematics. It is for this reason that the yielding criterion that is discussed next is preferred in most theoretical work.
Plane Stress Problems
In plane stress problems where σ3 = 0, the Tresca criterion, which states that yielding begins when the maximum shear stress reaches σyp/2, simplifies to:
max {|σ1 – σ2|, |σ1|, |σ2|} = σyp
Since we do not know if σ1 or σ2 is the largest principal stress, this expression takes different forms in different quadrants of the σ1–σ2 plane.
Quadrants I and III (Same Signs)
When σ1 and σ2 have the same sign, the difference |σ1 – σ2| is always smaller than either |σ1| or |σ2| (or both). Therefore, in the first and third quadrants, the expression max{|σ1 – σ2|, |σ1|, |σ2|} is equal to either |σ1| or |σ2|.
- First Quadrant: Both stresses are positive. Below the line σ2 = σ1 (where σ1 > σ2), the criterion gives σ1 = σyp. Above this bisection line, we must have σ2 = σyp.
- Third Quadrant: Both stresses are negative. By the same analogy, below the bisection line (where the magnitude of σ1 is larger), we must have σ1 = –σyp. Above the bisection, we must have σ2 = –σyp.
Quadrants II and IV (Opposite Signs)
When the stresses have opposite signs, the term |σ1 – σ2| represents the sum of the absolute values, which is larger than either individual magnitude. Therefore, the difference term controls the yielding.
Second Quadrant: σ1 is negative and σ2 is positive. Therefore:
max{|σ1 – σ2|, |σ1|, |σ2|} = |σ1 – σ2| = σ2 – σ1 = σyp
This describes a line connecting (–σyp, 0) to (0, σyp).Fourth Quadrant: σ1 is positive and σ2 is negative. Therefore:
max{|σ1 – σ2|, |σ1|, |σ2|} = |σ1 – σ2| = σ1 – σ2 = σyp
This describes a line connecting (σyp, 0) to (0, –σyp).

2. Von Mises, or Distortion-energy, Theory
In the previous section, we explained that the yield criterion for a material that is hydrostatic pressure insensitive, the yield surface takes the form
f(J_2,J_3)=C,where J_2 and J_3 are the second and the third invariants of the stress deviator, C is a material constant. The simplest form of the above equation is J_2=C, which is often written as
\boxed{J_2=k^2.}\tag{11}The development of this yield criterion is associated with the names of Von Mises, Hencky, Maxwell, and Huber, and now is often known as the von Mises criterion. Von Mises proposed this criterion in the invariant form given by the above equation primarily because it was mathematically simpler than the invariant form of the maximum-shear-stress theory given by Eq. (10). Subsequent experiments showed that Eq. (11) provides better over-all agreement with combined stress-yielding data than the maximum-shear-stress theory.
From the uniaxial tension test, \sigma_1=\sigma_{\rm yp}, and J_2=\frac{1}{3}\sigma_{ \rm yp}^2. Therefore, from this test, the constant k is determined as
k=\frac{\sigma_{\rm yp}}{\sqrt{3}}.\tag{12}From a pure shear test, J_2=\tau_{\rm yp}^2, and thus
k=\tau_{\rm yp}.Therefore, using the von Mises criterion, we imply that the tensile and shear yield strengths of a ductile material are related through \tau_{ \rm yp}=\sigma_{ \rm yp}/\sqrt{3}\approx 0.577 \sigma_{ \rm yp}.
The equation in the box can thus be written as
\boxed{J_2=\frac{\sigma_{ \rm yp}^2}{3}.}\tag{13}The above equation can also be written as
\sigma_{ \rm yp} = \frac{1}{\sqrt{2}} [(\sigma_1 - \sigma_2)^2 + (\sigma_2 - \sigma_3)^2 + (\sigma_3 - \sigma_1)^2]^{1/2} \tag{14}An equivalent way to express this criterion is by introducing the von Mises (equivalent) stress, \sigma_{\mathrm{v}} = \sqrt{3J_2},\tag{15} so that yielding occurs when \sigma_{\mathrm{v}} = \sigma_{ \rm yp}. Thus, the conditions J2 = k2 and σv = σyp are simply two different forms of the same yield criterion.
Physical Meaning
A number of attempts have been made to provide physical meaning to the von Mises yield criterion. One commonly accepted concept is that this yield criterion expresses the strain energy of distortion. On the basis of the distortion-energy concept, yielding will occur when the strain energy of distortion per unit volume exceeds the strain energy of distortion per unit volume for a specimen strained to the yield stress in uniaxial tension or compression. The derivation of Eq. (6) on the basis of distortion energy is given below. Another common physical interpretation of Eq. (6) is that it represents the critical value of the octahedral shear stress (discussed later).
The total elastic strain energy per unit volume (U_0=\frac{1}{2}\sigma_{ij}\epsilon_{ij}) can be divided into two components, the strain energy of distortion, U_d, and the strain energy of volume change, U_v.
We begin by decomposing the strain and stress tensors into their volumetric (mean) and deviatoric components:
\epsilon_{ij}=\epsilon_{m}\delta_{ij}+\epsilon_{ij}^\prime\sigma_{ij}=\sigma_{m}\delta_{ij}+\sigma_{ij}^\primewhere \delta_{ij} is the Kronecker tensor (\delta_{ij}=0 if i\neq j and \delta_{ij}=1 if i=j).
The mean (volumetric) strain and stress are defined as
\epsilon_m=\frac{\epsilon_{kk}}{3}=\frac{\epsilon_{xx}+\epsilon_{yy}+\epsilon_{zz}}{3},and
\sigma_m=\frac{\sigma_{kk}}{3}=\frac{\sigma_{xx}+\sigma_{yy}+\sigma_{zz}}{3},with summation taken over repeated indices.
Because the traces of deviatoric tensors are zero, we have
\sigma_{kk}^\prime=0,\quad \epsilon_{kk}^\prime =0. \quad (\text{summation over }k)since the trace of a deviatoric tensor is zero.
Recall that
\sigma_{ij}^\prime=2G\epsilon_{ij}^\primewhere G=\frac{E}{2(1+\nu)} is the shear modulus and
\epsilon_{kk}=\frac{1}{E}(1-2\nu)(\sigma_{xx}+\sigma_{yy}+\sigma_{zz}) \Rightarrow \epsilon_{m}=\frac{\sigma_{m}}{3K},where K=\frac{E}{3(1-2\nu)} is the bulk modulus.
Therefore,
\sigma_{ij}=3K\epsilon_{m}\delta_{ij}+2G\epsilon_{ij}^\primeand
\begin{aligned} U_0&=\frac{1}{2}\sigma_{ij}\epsilon_{ij}\\ &=\frac{1}{2}(3K\epsilon_{m}\delta_{ij}+2G\epsilon_{ij}^\prime)(\epsilon_m\delta_{ij}+\epsilon_{ij}^\prime)\\ &=\frac{3K}{2} \epsilon_m^2\delta_{ii}+\frac{3K}{2}\epsilon_m \epsilon_{ii}^\prime+G\epsilon_m\epsilon_{ii}^\prime+G\epsilon_{ij}^\prime\epsilon_{ij}^\prime\\ &=\frac{9K}{2}\epsilon_m^2+0+0+G\epsilon_{ij}^\prime\epsilon_{ij}^\prime\\ &=\frac{\sigma_{m}^2}{2K}+\frac{1}{4G}\sigma_{ij}^\prime\sigma_{ij}^\prime \end{aligned}\tag{16}Let's define
U_v=\frac{\sigma_m^2}{2K},\quad U_d=\frac{1}{4G}\sigma_{ij}^\prime\sigma_{ij}^\prime=\frac{J_2}{2G}.\tag{17}So U_0=U_v+U_d, and distortion energy is associated purely with shape change, while U_v is associated with volume change.
The maximum distortion energy states that the yielding begins when U_d reaches a critical value U_d^y equal to the distortion energy at yielding in a uniaxial test:
U_d=\frac{J_2}{2G}=U_d^y \Rightarrow J_2=2GU_d^y=k^2.\tag{18}Therefore:
The von Mises criterion J2 = k2 is exactly equivalent to the “distortion energy reaches a critical value” criterion for isotropic linear elasticity.
Plane Stress Problems (von Mises)
Similar to the Tresca criterion, the von Mises criterion simplifies significantly for cases of plane stress, where one principal stress is zero (let \sigma_3 = 0).
We start with the general von Mises criterion expressed in terms of principal stresses (Eq. 14):
\sigma_{\rm yp} = \frac{1}{\sqrt{2}} [(\sigma_1 - \sigma_2)^2 + (\sigma_2 - \sigma_3)^2 + (\sigma_3 - \sigma_1)^2]^{1/2}Squaring both sides provides a slightly easier form to work with:
2\sigma_{\rm yp}^2 = (\sigma_1 - \sigma_2)^2 + (\sigma_2 - \sigma_3)^2 + (\sigma_3 - \sigma_1)^2Substituting the plane stress condition \sigma_3 = 0 into this equation yields:
\begin{aligned} 2\sigma_{\rm yp}^2 &= (\sigma_1 - \sigma_2)^2 + (\sigma_2 - 0)^2 + (0 - \sigma_1)^2 \\ 2\sigma_{\rm yp}^2 &= (\sigma_1^2 - 2\sigma_1\sigma_2 + \sigma_2^2) + \sigma_2^2 + \sigma_1^2 \\ 2\sigma_{\rm yp}^2 &= 2\sigma_1^2 - 2\sigma_1\sigma_2 + 2\sigma_2^2 \end{aligned}Dividing the entire equation by 2 results in the governing equation for von Mises yielding under plane stress:
\boxed{\sigma_1^2 - \sigma_1\sigma_2 + \sigma_2^2 = \sigma_{\rm yp}^2} \tag{19}This is the equation of an ellipse in the \sigma_1–\sigma_2 plane, with its major axis oriented at a 45° angle to the \sigma_1 and \sigma_2 axes.
