Chattering is the limitation of sliding mode control. Chattering can limit the performance of mechanical system. It can exicet  un-modelled high frequency dynamics. In presence of high frequency control (because of chattering) , actuator saturation or failure may appear.

thanks for the answer. If the actuator is a flywheel, how does the chattering influence its structure? What I really like to know is the specific physical respond to chattering that might happen. what happens to the wheels? failure in what sense?

Samira,

Repetitive driving of the flywheel (is this really a momentum wheel?) shouldn't cause any undue ill-effect - they tend to fail at their bearings and loading and unloading of the wheel by a few % of rotation rate shouldn't cause any deterioration in the bearings.

Thanks a lot for the answers. Can we say, with a certain level of certainty, that  the chattering shows its influence in attitude pointing accuracy? I believe the worst case scenario in s/c application could be instability in the attitude and ultimately loss of control.

Samira,

I think that it is reasonable to link overall pointing accuracy in some way to the chatter amplitude and period. One can map the perturbing torque in each excursion from the steady state, to an excursion in pointing angle by integrating the torque and accounting for the inertia of the s/c.

Samira,

Chattering occurs in the vicinity of s=0 and it will effect the precision of attitude control, and yes it do effect the actuators. 

Hello Samira,

The work of A. Levant in 'Higher order sliding modes as a natural phenomenon in control theory' may be relevant to your question. It says that the dynamics of the actuator may cause a discontinuous control input signal to appear continuous, filtering it so that the chattering does not transfer to the system output. Therefore it is possible that there be no physical sign on the system that the control input to the actuator is of classic sliding mode control noted for chattering. Perhaps a similar filtering might occur from the actuator to the variable under control also.

As I think hinted by James Garry, whether or not rapid damage occurs on the actuator because of the character of the control signal depends on its construction. Inertia, a slow response, may insulate the actuator and system under control from chattering.

For a simple first order system $\dot{x}=u$ with sliding mode controller $u=-sgn(x)$, and finite frequency switching, the chattering would transfer to the output: $x$ will oscillate around zero. For a higher order system with several integrators, chattering becomes less likely to transfer to the output. Thus it more or less becomes just the problem of the actuator.