[ODE] Force calculations

[Gib Bogle]

Hello,
I'm modelling a system that is made up of a number (28) of rods 
connected together end-to-end.  The first and last rod are each hinged 
to a fixed axle lying on the z axis, separated by about 20 rod lengths.  
The other rods are hinged to each other (horizontal axes), except that 
the connections to the last rod at each end are ball-and-socket joints.  
The whole assemblage is therefore free to rotate about the z axis, and 
will assume a roughly paraboloid shape as it rotates.  Think of a 
rotating chain.

At each timestep the external forces applied to the rods are in the xy 
plane, i.e. there is a radial component and a tangential component.  
These forces, which tend to make the chain of linked rods rotate, is 
resisted by a torque applied to the top rod (one of those connected to 
the axle).  The constant torque can be applied either as a force applied 
to the c.g. of the rod in the tangential direction, or directly as a 
torque (it doesn't make any difference which way I do it).  When the  
steady resisting torque matches the external forces, a steady state is 
achieved with a constant rotational speed.

The problem is that the value of resisting torque that is needed to 
reach steady state is not equal to the sum of the applied moments.  The 
difference between these two values can be quite large, e.g. > 10%.  By 
"sum of the applied moments" I mean I compute the total of the moments 
contributions over one rotation and divide by the number of timesteps to 
obtain the average total moment of the external forces.  The rods are 5m 
long.  I'm using a time step of 0.01 sec, and all other solver 
parameters have their default values.  I'm not sure where the error 
arises, and whether or not I should expect ODE to model this system 
accurately.  Any suggestions or comments would be welcome.

Thanks
Gib