I know it's well after the point where this is relevant, but I came across this when looking up the snow system post once again and now understand this topic a lot better, so I figured I'd add some more pertinent information.

Eyewall cycles are part of the natural evolution of a tropical cyclone. Many hurricane modeling studies have been performed using what is called the Sawyer-Eliassen non-linear balance model. In layman's terms, they took the typical atmospheric balances (in wind and mass) and developed a set of equations that captures the atmospheric motions and processes related to changes in the wind and mass (and heating) fields. Originally developed for mid-latitude storms, it has been successfully applied to hurricanes as well. This is relevant because the application of this model to hurricanes has shown that these cycles are a natural part of a hurricane's evolution and have several implications on the wind profile and future evolution of the storm.

These eyewall replacement cycles take place as a secondary wind maximum develops in the outer edges of the inner core of the hurricane. Enhanced convection is often found in association with these wind maxima; between the eyewall, or primary wind maximum, and the secondary maximum, you usually see a moat, or area of weakened rainfall rates. This results from the convergence at the surface needed to create and maintain the secondary (and primary) wind maxima. Rising motion that results near these maxima result in sinking motion outward -- both inside the storm and at larger radii -- thus resulting in weaker rainfall rates at ~50-200km radii in many instances (and perhaps the dry air surrounding the storm on water vapor imagery).

The natural tendency of this wind maximum is to contract with time. As it strengthens, however, the aforementioned sinking motion away from it begins to affect the eyewall itself. Eventually, the original eyewall begins to erode and a new eyewall is set up at a larger radius with the secondary wind maximum due to strong sinking motion. As a result, the storm has weakened (at least temporarily) and has grown in overall expanse (see Frances from this year for a good case....a small storm to begin with becoming a monster). Another eyewall cycle may or may not take place from here on out.

What is still unknown is the exact predictability of when one is going to occur as well as how fast they will occur and how far they will reach towards the center before resulting in the collapse of the original eyewall. Atmospheric conditions modulate whether or not another cycle will take place; if conditions become no longer favorable for TC development (e.g. shear, sea temperatures, etc.), another eyewall cycle may not take place and the storm may be left in a weakened state (or with a flat wind profile not entirely unlike an extratropical storm). This was likely the case with both Isabel in '03 and Frances in '04 -- upwelling from Fabian (for Isabel) and Frances itself (for Frances) resulted in unfavorable conditions for another eyewall cycle to take place, resulting in a weakened storm.

If we had full, real-time data about the evolution of the storm, we might be able to predict these secondary wind maxima as well as the evolution of the resulting concentric eyewall cycles (whether two or even three -- in rare instances -- eyewalls) to a better degree. However, due to their large sensitivity to the actual atmospheric conditions, we're likely still a ways off of actually being able to model them in a near-real-time basis. (Heck, maybe I'll be able to do it someday....it is sort of an extension of my current work on wind field expansion during extratropical transition.)

Hope this provides some insight, several months late...