The center of mass (CM) energy in a collisional Penrose process - a collision taking place within the ergosphere of a Kerr black hole - can diverge under suitable extreme conditions (maximal Kerr, near horizon collision and suitable impact parameters). We present an analytic expression for the CM energy, refining expressions given in the literature. Even though the CM energy diverges, we show that the maximal energy attained by a particle that escapes the black hole's gravitational pull and reaches infinity is modest. We obtain an analytic expression for the energy of an escaping particle resulting from a collisional Penrose process, and apply it to derive the maximal energy and the maximal efficiency for several physical scenarios: pair annihilation, Compton scattering, and the elastic scattering of two massive particles. In all physically reasonable cases (in which the incident particles initially fall from infinity towards the black hole) the maximal energy (and the corresponding efficiency) are only one order of magnitude larger than the rest mass energy of the incident particles. The maximal efficiency found is ~ 13.92 and it is obtained for the scattering of an outgoing massless particle by a massive particle.

In a paper posted on the arXiv a few weeks ago Berti, Brito and Cardoso \cite{Berti+14} suggest that ultra-high-energy particles can emerge from collisions in a black hole's ergosphere. This can happen if the process involves a particle on an outgoing trajectory very close to the black hole. Clearly such a particle cannot emerge from the black hole. It is argued \cite{Berti+14} that this particle can arise in another collision. Thus the process involves two collisions: one in which an outgoing particle is produced extremely close to the horizon, and a second one in which energy is gained. The real efficiency of this process should take into account, therefore, the energy needed to produce the first particle. We show here that while this process is kinematically possible, it requires a deposition of energy that is divergently large compared with the energy of the escaping particle. Thus, in contradiction to claims of infinitely high efficiencies, the efficiency of the combined process is in fact extremely small, approaching zero for very high output energies. Even under more general conditions than those considered in \cite{Berti+14} the total energy gain never diverges, and is larger only by a factor of a few than the energy gain of the original collisional Penrose process that takes place between two infalling particles \cite{Piran+75,PiranShaham77,Bejger+12}