The problem of hadronic cluster production in quark-hadron phase trans
ition in heavy-ion collisions is studied by cellular automata. Previou
s result on the scaling behavior is extended to include variations in
the drift speed. It is also shown that coalescence is more important t
han growth in generating scaling. A new set of rules is adopted to fre
e the clusters from being rigid. It is found that the scaling exponent
is independent of not only the shapes of the clusters, but also the p
robability of the breakup of the clusters. The universality of the sca
ling behavior is now extended to a wide range of physical properties c
haracterizing the geometry and dynamics of the phase transition proces
s.