In order to display negative damping there should be a source of energy. The most common source of energy would be a steady flow. If the body is attached (an anchored ship, or a leg of of a oil-production platform...) instability of the wake (vortex shedding will drive an oscillation of the structure. In a linear theory if the body can be described as a mass-spring system one could say that the system displays negative damping if the vortex shedding frequency approaches the mechanical resonance frequency of the system. There are multitudes of such phenomena, which are described for the interaction of an air flow with an elastic structure as "Aero-elasticity". The oscillation of our vocal folds when producing voiced sounds is a typical example of "flutter" involving a negative damping.
Of course when the fluid is stagnant the energy can come from the motion of the floating object as described in the previous answer.
I have never worked your specific problem but I believe it is analogous to the physics involved in problems such as break squeal or whistling where a gross motion process (rotation or flow) interact with a small motion vibration process.
Processes of this type involve vibration taking up energy from the gross motion process. When computing complex modes, a stable system shows positive damping and an unstable system, i.e. system where the input excitation from the gross motion process exceeds the system internal damping, shows negative damping.
Therefore, negative damping can be interpreted as excitation being able to drive and reinforce a feedback loop.
See also http://en.wikipedia.org/wiki/Self-oscillation
The damping force depends on lots of factors, the motion you are investigating (Heave, Pitch, Roll), the hull shape, and existence of free surface and etc.
As I know, for the floating bodies, the damping force or moment are in the opposite side of the motion speed. For example, assume a ship entering water with the velocity of zdot, the heave force due to zdot is upward. However, in the maneuvering of bodies, the effects of other motion can cause negative damping. For instance, when a body have roll speed, it may have positive roll moment due to roll speed, but negative heave force due to roll speed. May be in you work, there are more than one motion, and as a result there are negative forces for damping.
In order to display negative damping there should be a source of energy. The most common source of energy would be a steady flow. If the body is attached (an anchored ship, or a leg of of a oil-production platform...) instability of the wake (vortex shedding will drive an oscillation of the structure. In a linear theory if the body can be described as a mass-spring system one could say that the system displays negative damping if the vortex shedding frequency approaches the mechanical resonance frequency of the system. There are multitudes of such phenomena, which are described for the interaction of an air flow with an elastic structure as "Aero-elasticity". The oscillation of our vocal folds when producing voiced sounds is a typical example of "flutter" involving a negative damping.
Of course when the fluid is stagnant the energy can come from the motion of the floating object as described in the previous answer.
I did forget to say that one of the conditions to find a transfer of energy from a continuous source of energy to a oscillating body is the presence of two degrees of freedom in the oscillating system. For aircraft wings this will in general be two mechanical degrees of freedom (coupling between two mechanical modes of vibration). Another example of an oscillation induced by coupling of a mechanical oscillator with a steady flow is the oscillation of the reed of a clarinet when you blow the instrument. In that case there is a coupling between a mechanical oscillator and an acoustic oscillator (pipe).
It is very normal to see the negative damping in hydrodynamic analysis, especially in seakeeping analysis. if the floating body is in the wave condition, the motions always involves in 6 DOF. The negative damping always occurs in the so-called coupled terms, for instance B53. The motion in heave and pitch directions are always coupled, which means, the heave motion will greatly influence the pitch motion. Therefore, in the numerical simulation you will find the heave induced pitch damping is negative in most of the cases. This is also true according to the model test results. not only the damp could be negative, the added mass of A35 (pitch induced heave added mass) could also be negative. So, this phenomenon is caused by the coupled behaviour between different motions.