http://en.wikipedia.org/wiki/Coker_unit

to put it into laymans terms, a coker unit boils the tar out of the heavy resid (literally and figuratively).

The resid is heated and then flashed into the coker drum, anything that will volatize goes off the top and solids (coke) is left in the drum. You continue to flash the heated resid into the drum until the drum is full of solids. The full drum is then cooled with steam and then water. The bottom head is removed and then the water jets you noted above are used to cut the coke out of the drum.

You always see at least two drums in a coker unit, one being filled and the other being cooled/emptied.

The one aspect missing from the above presentation is the hydrogen content of the crude, products, and major fractions. Products contain only carbon and hydrogen, with the hydrogen content ranging between 14 and 15 wt%. The hydrogen content of crudes varies approximately linearly from 14% at 40 API to about 10wt% at 10 deg API (oil sands).

Vacuum residue (5 API for heavy crude) has about 8-9% hydrogen. Coke has 3-4% hydrogen. Thus, the raw products from coking of a vacuum residue (38 wt% coke yield from heavy crude with 24% "Conradson Carbon" are very "hydrogen deficient" with respect to refinery products (11% vs 14%).

If you assume the net yield from heavy crude is the same as light crude, you need to add a lot of hydrogen which is an input from the energy in natural gas. About 2 wt % H2 net, based on products will be required for a 22 wt% heavy crude converted into products. Less than 0.5 wt% will be required for light crudes, which can be obtained from the reforming process, which, incidently causes a reduction in volume vs the heavy naphtha feed.

Bottom line is that processing heavy crudes is no piece of cake-- more NG is required to achieve the same volumetric yield, which implies more hydrogen capacity, more hydrotreating capacity, more sulfur recovery capacity, and more utilities capacity.