Lots of *potential* causes of ora-4031. I'd strongly suggest you have a look at MOS note 1088239.1.

But also, we have a diagnosis tool which will help as well. See MOS Note 1521925.1


You could do something like

SQL> select * from v$sgastat order by pool, bytes desc;

POOL         NAME                                          BYTES     CON_ID
------------ ---------------------------------------- ---------- ----------
java pool    free memory                                16777216          0
large pool   free memory                                33292288          0
large pool   PX msg pool                                  262144          0
shared pool  free memory                               115875992          0
shared pool  allocate segment latch re                  58802176          0
shared pool  KGLH0                                      39358864          0
shared pool  SQLA                                       27873360          0
shared pool  KQR X SO                                   13357016          0
shared pool  write state object                         13023264          0
shared pool  db_block_hash_buckets                      11538432          0
shared pool  event statistics per sess                  11512064          0
shared pool  KGLS                                       10598168          0
shared pool  KGLHD                                      10591064          0
shared pool  PLMCD                                       9017784          0
shared pool  row cache                                   8656032          0
...

to see what are the high consumers in the shared pool, but typically

a) the information is somewhat crytpic
b) changes rapidly

As a stopgap solution whilst you investigate the cause, you can look at increasing the size of the shared pool via init.ora parameter.