Report prepared by.

Ezral Baharudin

Recursive Method in Multibody Systems

Multibody systems consist of collection of bodies that are constrained to move relative to one another by kinematic
connections between bodies. The system can have a tree or chain structure (open loop) or a structure with closed links
of bodies (closed loop). In the chain structure system, the path from an arbitrary body in the system to another arbitrary
body in the system is unique. To make it easy to understand, tree harvester truck can be assumed and simplified into
chain structure concept and in general it can be divided into several bodies as shown in Figure 1 below.

Figure 1. Tree harvester truck illustration into rigid bodies

The tree harvester truck mechanism is consists of 5 bodies, not include with the ground body. Bodies 1, 2, 3, 4 and
5 are connected to the previous bodies by revolute joints. Ground (B0) is designated as the base body. Body B1 is
attaching with body B0 by a revolute joint. Joints 4 consist of two revolute joints which are vertical and horizontal.
Therefore, this mechanical system has 6 degrees of freedom. Due to the movements is in horizontal and vertical; the
motion of the mechanism is in three dimensional.

Recursive Formalism

The field of multibody dynamics embraces the mathematical description of the dynamic behavior of multibody
systems. A recursive formulation is derived from the equation of motion of constrained mechanical system. The
kinematic properties such as position, virtual displacement, velocity and acceleration are developed based on the
relative coordinate between contiguous bodies which connected by joint [ ]. Hooker [ ] proposed this formalism to the
dynamic analysis of satellite and found the computational cost increase linearly with the number of bodies. This
algorithm has been used and extent by several researchers, and recently, this formalism has been generalized to improve
the implementation and efficiency [ ]. The dynamics equations in a recursive formulation are written in terms of the
system’s degree of freedom and lower dimensionality than those in an augmented formulation. To understand this
concept, chain topology of multibody system for tree harvester truck in Figure 2 could make better explanation.
 Figure 2. Chain topology of multibody system for tree harvester truck

From figure above, the bodies are connected with joint. Body B0 is called the base body and the number of body
start from the B1 until Bn. Each body number can be represented as Bn and the preceding body could be named as a Bn-1.
The arranged set of body number is called “body connection array” [ ]. The joint number is named before the body
number. As mentioned before, each joint can be consists of single or multiple axes such as revolute, prismatic,
cylindrical, translational and spherical joints.

Kinematics

Relative motion between neighboring bodies and constraints are the two main aspects in recursive kinematics to
generate the total system matrices and a solution for equation of motion for the multibody system. Figure 3 shows the
elementary system of two bodies interconnected by a joint.

Figure 3. Relationship of contiguous bodies

The orientation relationship of contiguous bodies as above is obtained by sequentially transforming from the body
reference frame on body B0 to the body reference frame on body Bn. The bodies system is consider as an open chain and
point 0 is the global reference frame. The relative kinematic of position for each body can be described with respect to
the global reference as
=
rn rn −1 + A n −1un (1)

whereas An-1 is an absolute rotation matrix from previous body, Bn-1 in three dimensional and un = [xn yn zn]T is relative
position vector of body Bn with respect to the previous body, Bn-1 describing rn relative to point O respectively. As an
example in this case, the absolute rotation matrix An-1 can be expressed as a multiplication of a series of n number of
bodies.
cos(θ n −1 ) − sin(θ n −1 ) 0 
A n −1* =  sin(θ n −1 ) cos(θ n −1 ) 0  (2)
 1

0 0
z − axis

The velocity of joint could be determined from the time derivative of equation (1).
 vn = rn
rn −1 + ω n −1un + A n −1u n
= (3)

r − u ω + H q
= n −1 n n −1 n n

where,
 du 
H n = A n −1  n 
 dt 
 du 
u n =  n  (θn )
 dt 
q = θ
n n

 0 −rnz rny 
  
=un  rnz 0 −rnx 
 −rny rnx 0 


Angular velocity ωn is the relative angular velocity of the body Bn with respect to the body Bn-1, and also function
of joint relative coordinate qn and joint relative velocity, qn as below.

ωn ωn −1 + Bn qn
= (4)

where Bn is the unit vector of the joint rotational axis whereas Bn = [0 0 1]T as shown in Fig. 5 in this specific case. The
velocity relationship between bodies could be determine by combining vn and ωn into a single form,
 rn   I − u n   rn −1   H n 
=     +   qn (5)
ω
 n  0 I ω
  n −1   Bn 
Equation (5) could be compacted into one equation which represents the total velocity as
Vn Cn Vn −1 + Sn qn
= (6)

By doing second time derivative of equation (6), the acceleration of each body could be determine as
W=
n Cn Wn −1 + S n q + Dn (7)

Where Dn = S n qn

Dynamics

The equation of motion for rigid body system could be derived after the kinematics of the elementary contiguous
bodies has been described as in [17]. In general, from the Newton second law of motion, the total force of body n is
M n W=n Qn + Rn (8)

where
m I 0 
Mn =  n 
 0 Jn 
 FT 
Qn =  n 
−ω J nω 
 Fn and Jn are representing vector applied force and inertia tensor of body Bn with respect to the mass center.
However, the idea of recursive method is to exclude the reaction force (Rn) from the equation. To eliminate the reaction
forces Rn from equation (8), both sides are multiplying by SnT.
S T M= W S TQ + S T R
n n n n n n n
(9)

From the ideality of constraint,

S nT Rn = 0 (10)

Consequently, the equation reduces to

S nT M n Wn = S nT Qn (11)

By substituting Wn from equation of total acceleration (7),

S nT M n (Cn Wn −1 + Sn qn + Dn ) =
SnT Qn (12)

 expression,
Then, reconstruct the above equation into q
−1
qn  SnT M n Sn  ( SnT Qn − S nT M n (Cn Wn −1 + Dn ))
= (13)

In compact form, Un and Fn can be represented as,

U n = SnT M n S n (14)

Fn =
( SnT Qn − S nT M n (Cn Wn −1 + Dn )) (15)

Knowing that the total inertia and force in the open loop system can be represented as
n

∑ (M W − Q ) + ... + (M
i =0
i i i n−2 Wn − 2 − Qn − 2 ) + (M*n −1Wn −1 − Qn*−1 ) =0 (16)

Whereas M*n-1 is the sum of mass matrices of bodies Bn-1 and Bn, and Q*n-1 is the total inertia force matrices at bodies
Bn-1 and Bn as below
M∗ = M + CT M C − CT M S U −1 S T M C
n-1 n −1 n n n n n n n n n
(17)
n
∗
Q
n −1 = Qn −1 − C (M n ( S n U S
T
n
−1
n
T
n ( Qn − M n Dn ) + Dn ) − Qn ) (18)

Continually applying above procedure from body Bn-1 till B0, the base equation of motion will be obtained as
M 0 W0 − Q0 =
0 (19)

If the constrained is not apply to the base body, B0

M 0 W0 = Q0 (20)

Recursive algorithm

The recursive method algorithm could be easily to understand using backward and forward movement approach in
order to acquire all needed components in equation of motion. Figure 4 shows the components which could be obtained
in backward and forward approach.
 Figure 4. Backward and forward movement approach

In general, the algorithm derived for generating the equation of motion for an open loop system (chain topology) is
as follows:
(1) Define the initial conditions for relative coordinate q (t ) and velocities q (t ) at t0.
(2) Compute all positions and orientation of all bodies in the system by using equation (1) and (2) for i=1 to nB
recursively.
(3) Compute matrices of Cn and Sn for each body recursively by using backward approach from body Bn to B1.
(4) Compute updated matrices of M*n-1 and Q*n-1 for each body in the system recursively using backward
approach from n till 1.
(5) Compute acceleration state, Wn recursively using forward approach from body B1 to Bn.
(6) Compute joint acceleration, q n of each body along the forward path from body 0 to body n.
From step (6), the relative coordinates will be determined by time integration. The position, rn and velocity, vn of
body, Bn can be computed from derivation of position, rn-1 and velocity, vn-1 of preceding adjacent body Bn-1.