This page summarizes/motivates how Kane’s dynamical equations of motion are derived and used. Basic understanding of vector mechanics and derivatives is assumed.
There are a few references I have used when preparing these notes, but the clearest was this thesis.
Review of Lagrange’s Equations:
The steps in Lagranges equations are:
Let ri be the location of the i-th particle in an inertial frame. Express this position as a function of a set of generalized coordinates:
ri=ri(q1,...q(3N−k))
where N is the number of particles in the system, 3N because each particle has a 3d coordinate, and −k because there might be k configuration constraints. So overall, we need 3N−k scalar quantities (degrees of freedom) to uniquely define the configuration of the system.
Determine the Lagrangian:
L=T−U
where T(q1,...,q3N−k,q˙1,...,q˙3N−k) is the kinetic energy of the system, and U(q1,...,q3N−k) is the potential energy of the system.
Then the equations of motion are:
dtd(∂q˙i∂L)−∂qi∂L=Qi,i={1,...,3N−k}
where
Qi=j=1∑NFj⋅∂qi∂rj
is the contribution of non-conservative/external forces.
Notice, we will have generated 3N−k 2nd order equations. If there are non-holonomic constraints, these are usually introduced as additional constraint equations, with lagrange multipliers. (I’ll explain this better once I understand it myself).
Derivation of Kane’s method (for system of particles)
Kanes method is basically a souped-up version of Newton’s 2nd law/D’Alemberts principle:
RPi−mPiaPi=0,i={1,...ν}
where
RPi is the net force on the i-th particle
mPi is the mass of the i-th particle
aPi is the acceleration of the i-th particle
ν is the number of particles.
Now suppose the position/configuration of the system can be uniquely described by a set of p scalar coefficients (and potentially time):
rPi=rPi(q1,...,qp,t)
then the velocity in some inertial frame is
vPi=dtdrPi=r=1∑p∂qr∂rPiq˙r+∂t∂rPi
Often, we can simplify this expression by choosing a convenient set of generalised speeds ui. In Lagranges formulation, the generalised speeds ur=q˙r. However, we can be more general, allowing functions of q,q˙:
ur=ur(q1,...,qp,q˙1,...,q˙p),r={1,...,p}
and we need p generalised speeds, so that this map is invertible (to be able to find q˙i in terms of u). This allows us to write
vPi=r=1∑pvrPiur+vtPi
where
vrPi=∂ur∂vPi,vtPi=∂t∂vPi
If there s non-holonomic constraints, we rearrange the variables such that last s generalised speeds can be represented as linear combinations of the first p−s generalised speeds. Then, we can also write
vPi=r=1∑p−sv~rPiur+v~tPi
where
v~rPi=∂ur∂vPi,v~tPi=∂t∂vPi
where we have expressed v only in terms of the first p−s generalised speeds, ur.
Now we return to Newton’s equation:
RPi−mPiaPi=0,i={1,...ν}
and notice that if we take the dot product of this equation with v~r, we get
v~rPi⋅RPi−mPiv~rPi⋅aPi=0,i={1,...ν}
and now we can sum across all particles, giving us an equation for each generalised speed, rather than for each particle in the system!
which gives p−s equations of motion, called Kane’s Dynamical Equations of Motion.
Kane’s Method for Bodies
The same general method is used, but we also need to introduce equations that describe the rotational dynamics of the system.
If a rigid body B belongs to a non-holonomic system S with p degrees of freedom in an inertial reference frame A, the set of all forces and torques on B can be summarised by a single resultant force RB acting through a point Q of B, and a torque T on B. Then the generalised active force is
F~rB=Aω~rB⋅T+v~rQ⋅RB,r={1,...,p}
and the inertial forces are:
FrB~∗=Aω~rB⋅T∗+v~rQ⋅R∗B,r={1,...,p}
where
R∗BT∗=−mBa∗=−α⋅I−ω×I⋅ω=−i=1∑βmiri×ai
where a∗ is the acceleration of the center of mass of B in A, and if there are β individual particles in B, the second version of T∗ can be used, or if the central inertia dyadic of B is known, the first version of T∗ can be used. If the principal central moments of inertia are known, and are aligned with a frame C, we can also write: