It is possible to greatly reduce the effect of pile-up by extracting data from an annulus about the source centre. The optimum inner radius to use depends on the strength of the pile-up but a typical first guess would be 20 arcseconds. The task calculates the encircled energy correction for annuli using the accurate, energy-dependent PSF described in section 7.5. The task epatplot can be used to investigate pile-up issues.
While a basic pile-up correction implementation is available, this feature is not currently being maintained.
arfgen provides a feature for correcting for the loss in detection efficiency due to over-lapping events occuring within the same frame. It does not, however, correct for spectral distortion as a consequence of pile-up. This is caused when two events of the same pattern overlap exactly; in this situation, the onboard electronics will recognise the pair as a single event with a charge equal to the sum of the charges due to the individual events. The approach of  was used to correct for flux loss. He provides formulae for flux loss for each of the four patterns accepted by the on-board detector electronics, given the pattern morphology and contributions of each pattern to the incident photon spectrum. In order to correct for pile-up, the task requires a rough estimate for the incident photon spectrum. This should be provided by the user in the form of a DAL dataset. The pile-up correction is itself dependent on the incident photon spectrum, which of course the ARF, RMF and observed counts spectrum are used to determine. An iterative analysis procedure is suggested, using the rmfgen pile-up correction feature only. See the rmfgen description for details.