Can I conclude from the above discussion that I cannot achieve my required output using SYNCSORT.
AFAIK, there's no extra cost for using a SYNCTOOL solution too. But before that you might want to answer the below question raised by Kolusu.
Why would you have a length of 255? You said you have a max of 10 occurrence per key combination and the total length of the fields you want to merge is 12 . so 10X12 = 120 bytes and then you have the key which is 12 bytes. So the total is 120 + 12 = 132 bytes. Why would you want to have 255 bytes as output when the data is only 132 bytes long?
The reson for using LRECL=255 was the input file .
The input file that I used to test the job was already having LRECL=133.
So, instead of creating a new dataset to test the job I used already exsisting dataset that met the same criteria.