Re: [foxboro] CALC vs. IND question
- From: Corey R Clingo <corey.clingo@xxxxxxxx>
- To: foxboro@xxxxxxxxxxxxx
- Date: Wed, 26 Mar 2008 16:09:41 -0500
I can't give you a direct correlation between CALCs and INDs, but I can
share some personal experience.
We had an application of about 40 lines of HLBL doing some arcane
calculations, maybe 8-10 equations total (I rue the day I ever told a
process engineer we could do this :). I'd estimate it would have taken
8-10 CALC blocks to do. But like you, I wanted to be able to decipher it
after I was done, so I took my first dip into the IND river.
Well, the first pirahna bite I got was that it buried the CP30 I was
testing it on. It was a testing system, fortunately, but my one IND
block, running at 2 sec, dropped the IDLETM about 30% (I'll defer the
discussion of IDLETM as a real bellwether of CP load for now). I pulled
out the books, and found the BPCSTM (number of HLBL statements per IND
block scan) parameter that defaulted to 100. In essence, it was running
my entire IND every scan. I had no WAITs in it, and all my external
connections were to RIxxxx inputs, not coded in the HLBL. I lowered that
number to 10, and the loading disappeared. Of course my IND now took
10-15 secs to run, but that wasn't a problem in my application. I could
have used WAITs as well, but the BPCSTM got the job done.
The 8-10 CALC blocks would have happily executed in 2 secs, with no
detectable loading increase, so I'd say that's at least a 100:1 ratio. But
my brain IDLETM would have dropped to 0 trying to figure them out a year
from now.
Corey Clingo
BASF Corporation
Mike_Adams@xxxxxxxxxxxxx
Sent by: foxboro-bounce@xxxxxxxxxxxxx
03/26/2008 04:44 PM
Please respond to
foxboro@xxxxxxxxxxxxx
To
foxboro@xxxxxxxxxxxxx
cc
Subject
[foxboro] CALC vs. IND question
Hello list,
I have some existing compounds in 4 I/A nodes that each have about 20 CALC
blocks running at a 2 sec period with pretty good phasing. These blocks
are running somewhat arcane mathematical calculations with 15 to 25 steps
each but only 2 or 3 real outputs are generated per CALC block.
The CALC steps are not commented at all. We are running into problems
where it is becoming harder to keep up with what is going on in these
CALCs
and it is very hard to edit them efficiently. I am interested in
replacing
these CALCs with IND blocks that would do the same work but be much easier
to troubleshoot, document & edit.
I know that INDs load the system more than CALCs but I'm not sure by what
degree. All 4 control processors involved are CP30s. Two of these CPs
are
lightly loaded but the other 2 are heavily loaded.
If I configured an IND to run at the same period as a CALC and do the same
calculations, what would a reasonable guesstimate of the increase in load
be? I just want to get a rough idea on a 1:1 comparison for starters.
I realize I can probably do the following to ease the load increase
incurred by going to INDs:
- lengthen the period to 5 sec & phase them well
- put 2 or 3 CALC blocks worth of steps into 1 IND
- place strategic WAIT statements in the IND code
Thanks in advance,
Mike Adams
Mitsubishi Polyester Film, LLC
Greer, SC USA
864-879-5231
_______________________________________________________________________
This mailing list is neither sponsored nor endorsed by Invensys Process
Systems (formerly The Foxboro Company). Use the info you obtain here at
your own risks. Read http://www.thecassandraproject.org/disclaimer.html
foxboro mailing list: //www.freelists.org/list/foxboro
to subscribe: mailto:foxboro-request@xxxxxxxxxxxxx?subject=join
to unsubscribe: mailto:foxboro-request@xxxxxxxxxxxxx?subject=leave
Other related posts:
