Trace files are named based on the id of the process that needs to write to trace. But as process_id numbers are reusable oracle will append information to the existing trace file if one exists with the given process_id in its name. If such file does not exists Oracle will simply create a new one.