Sketch of Rhea's interior
FYI, modelling the Rhea's interior is interesting because there is a controversy between geophysicists about the reading of the radio Doppler's data sent on Earth by the Cassini's flyby of the 2005. The radio Doppler data allows the misuration of the quadrupole of the gravitational field,
and in particular the value of its degree two (harmonics) C22 gravity coefficient, 
which rule the internal structure. 
Before such flyby, the C22 coefficient was constraint into a range of values through the Voyager I encounter and through the "bulk" data of the Cassini's first arrival in the Saturn's system, being assumed that Rhea were in hydrostatic equilibrium. From there, several models of Rhea's interior have been developed. The internal structure of Rhea show up to 10 different models of the Rhea's interior (BTW, in the paper there are also the formulas for calculating the C22 coefficient).
In short, from the paper above, for having Rhea a rocky core, supposed approximatively the same silicate mass fraction, the C22 coefficient would not be higher of 193 x 10^-6, while is found well up to 266.6 x 10^-6; value that establish a completely undifferentiated satellite with the ice-I <-> ice-II transition at about 400 km depth (the clear band in the screenshot: naively, the place in which an hypotetic liquid layer could sussist if Rhea were differentiated).
The controversy lie on the assumption or not of the hydrostatic equilibrium before to nest the radio Doppler data on the owed equations, which yields different values of C22. Nonetheless, the values under trial would seems definitely excluding a rocky core (and a liquid layer), being all the values higher of 223 x 10^-6. The original papers of the authors involved are behind paysites journals, but the table with the controversial C22 values for the year 2009 together with a brief account of the issue and an arrangement of the icy satellites to date is within Evolution of Icy Satellites
While I do not have found track of whatever happened with the last March's encounter, albeit without polar flybys the "controversy" is narrowest, the next Cassini's flyby of Rhea in January become important once more.
P.S.
Who is interested in knowing what are the "bulk data", the paper The gravity field of the saturnian system from satellite observations and spacecraft tracking data shows the list of the physical parameters sent to Earth by Cassini and how the geophysicists does accounts for it in their studies (together with the Saturn system's astrometrics updates). Among the others:
.
.
.
We also estimate the following parameters in the observation model:
1. Biases and drift rates in the on-board oscillator frequency for the one-way noncoherent Doppler.
2. Tracking station-dependent range biases. These are included in two forms, a global bias and a bias unique to each pass.
3. Solar corona parameters for the Voyager 1 1980 and Cassini 2004 and 2005 solar conjunctions. These are parameters in a model that corrects range data for the delay caused when the ranging signal passes through the solar corona.
4. Doppler biases during the solar conjunction periods. There is currently no model for the Doppler delay caused by the solar corona, so we use a bias parameter for each pass to correct for the delay.
5. Station-dependent ionosphere and troposphere media corrections for the VLBI data. Unlike Doppler and range, media calibrations are unavailable for the VLBI data.
6. Spacecraft camera pointing angles. The pointing for each picture is adjusted based on the background stars appearing in the picture.
7. Satellite-dependent phase angle biases in the optical data. These biases account for the error in determining the center of the image of a partially illuminated object.
8. Station-dependent timing biases for the ring occultations.
9. Radii of the occulting rings.
10. Positions of occulted stars.
11. Scale and orientation of the CCD detector in some of the Earth-based astrometry. A number of observers calibrated the scale and orientation of their detectors with a procedure that relies on positions of the satellites predicted from preexisting ephemerides, and we correct for errors in those ephemerides.
.
.
.
Very impressive, indeed.
