Mahalanobis Distance

I am doing research concerning the degradation of a particular system... In this case, a cooling fan for a computer.

I am measuring three parameters: sound, rotations, and current.

I am testing the fan at different voltage...

For the training data, or the baseline... I am using 5 measurements from a new fan... and for comparison I am using an old fan and comparing it's MD value to determine if it is healthy or not.

The problem is... for 4 volts... I can not find a baseline measurement because the rotation parameter is constant for all 5 measurements which means that I can not calculate the MD values b/c I can not divide by zero.

I would consider this a parameter selection problem... but the problem is, the IPC standard regards the RPM, or rotation, as very important.

Does anyone have any idea of how to overcome this situation?