Time Model System Stats DB/Inst: KTPRO/ktpro Snaps: 1552-1564
-> Total Time in Database calls 9967.6s (or 9967640228us)

Statistic Time (s) % of DB Time
----------------------------------- -------------------- ------------
DB CPU 6,858.6 68.8
DB time 9,967.6
PL/SQL compilation elapsed time 4.9 .0
PL/SQL execution elapsed time 9,218.9 92.5
background cpu time 15.0 .2
background elapsed time 126.6 1.3
connection management call elapsed 11.4 .1
failed parse elapsed time 0.2 .0
hard parse (bind mismatch) elapsed 0.0 .0
hard parse (sharing criteria) elaps 0.1 .0
hard parse elapsed time 3.8 .0
inbound PL/SQL rpc elapsed time 9,244.5 92.7
parse time elapsed 17.4 .2
sequence load elapsed time 0.3 .0
sql execute elapsed time 1,796.7 18.0
-------------------------------------------------------------


From the posted file 2 hour window. It looks like your app is transmitting most/ if not all of the PL/SQL to be executed into the server. Hence the high sql net activity.

What type of system and How many CPU's?

Can you modify the application to utilize named procedures/packages instead of anonymous procedures? Then use compiled PL/SQL and ping the packages.

BAD SQL:
SELECT ABS(:b1),ABS(:b2) FROM DUAL ----> Rewrite this, you don't need the DB to do this math work.
There seems to be a lot of that type thing. Each one by itself is not a lot, but it adds up. If you don't have to select the data, don't!
Instead:

var2 = abs(var1) --- If this is embedded in the app. Then no round trip to the DB is required... 100 x faster.

or if in PL/SQL
var2 := abs(:var1) ; (this is a function call and requires less overhead)