Re: [foxboro] Feedforward control question
- From: "Pat Martens" <fox@xxxxxxxxxxxxxxx>
- To: <foxboro@xxxxxxxxxxxxx>
- Date: Thu, 19 Jul 2007 21:36:27 +0200
Greg,
We have some feedforward controls similar to your situation.
Reboiler temperature controllers can be a pain to tune properly (long lag
time) and why wait for the temperature controller to respond while you
already know that by increasing feed, you will have to increase reboiler
duty (unless maybe the feed temperature increases at the same time!)?
Another FF variable could be your steam temperature if this is not stable.
Most of our FF schemes (not a large amount; only done where significant
improvements were expected) were implemented to improve the 'basic-control'
layer to make sure we would get the most out of our multi-variable APC
control.
Before APC was introduced feed flows were more or less kept constant
according to product planning requirements. With the introduction of APC a
lot more variables are constantly being manipulated, including feed-flow,
column pressure, product flows etc ... etc.
Feed forward controls can improve the APC benefits and improve
controllability whenever APC is swithed of.
Some of them work really well, others were decommissioned because no real
improvements were noted (feed composition !?).
Our approach was to use 'conventional' algorithm blocks as apposed to the
FFTUNE/FBTUNE blocks. The main reason is that we like to know what we're
doing and how to adjust parameters in the right direction to 'fine-tune'
things.
I don't have this feeling when we would use FFTUNE/FBTUNE blocks despite
several attempts with simulations (in which you exactly know your process
gain/lag-time/dead-time).
If you really intend to use FFTUNE/FBTUNE I suggest you do some test on a
simulated process (and share your experiences).
We typically link the feed flow and/or feed temperature to a DTIME block in
series with a LLAG block and connect this directly (or through a CALC block)
to the temperature PIDA.BIAS (or 'PCT'; column pressure compensated
temperature controller); use BBIAS and KBIAS (or the CALC block) to scale
things and set the FF gain.
The FF value is limited within realistic constraints to prevent extreme
adjustments of the reboiler steam flow.
With multiplicative FF you need to be careful! If, for some reason, your FF
factor is 0.0, your controller output will be 0.0 and your temperature
feedback controller can do nothing to prevent this!
I personally feel using the BIAS input is more robust and easier to
understand.
Just my findings, do not claim to be an expert and would like to hear other
opinions!
Kind regards,
Patrick Martens
Process Automation Specialist
Total Raff. Ned. N.V.
-----Original Message-----
From: foxboro-bounce@xxxxxxxxxxxxx [mailto:foxboro-bounce@xxxxxxxxxxxxx] On
Behalf Of Gregory A Hurwitt
Sent: donderdag 19 juli 2007 19:41
To: foxboro@xxxxxxxxxxxxx
Subject: [foxboro] Feedforward control question
We have a new distillation tower where the steam to the reboiler will be
controlled by a sump temperature to steam flow cascade loop. The tower
will be run at various different rates; we would like to use the tower feed
flow rate as a feedforward control signal to the sump temperature
controller. (Control blocks are PIDA's running in a ZCP270.)
The PSS for the PIDA block states that "Multiplicative feedforward
compensation for a load flow is commonly used in a composition or
temperature control loop." Based on this and other statements in the PSS's
and Integrated Control Block Description documents for the PIDA/FFTUNE
blocks, I plan to:
1. Connect the feed flow signal to the MULTIN parameter of the temperature
controller. Define the range and units of the feed flow signal using the
HSCIN, LSCIN, and EIN parameters.
2. Build an FFTUNE block and connect it to the temperature controller by
setting the FFTUNE's PIDBLK parameter equal to <PIDAname>.BLKSTA.
3. Since the feedforward signal will be connected to the PIDA, I will not
connect it to one of the LOADx parameters of the FFTUNE.
Any comments/corrections on #1-3 above or on the general topic of
multiplicative vs. additive feedforward for this application?
Where do I go from here to "train" the FFTUNE? The documentation is short
and dense. Is there a more user-friendly "white paper" or other resource?
Should I also be considering the use of a FBTUNE block in conjunction with
(or prior to the implementation of) the feedforward signal and the FFTUNE?
If I can get reasonable tuning of the steam flow and sump temperature loops
under one (or more) typical load condition(s), can I get by without FBTUNE?
____________________________________________________________________________
__________
Greg Hurwitt
Engineering Associate
E-Mail: gregory.hurwitt@xxxxxxxx
BASF Corporation
Freeport, TX
BASF - The Chemical Company
_______________________________________________________________________
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
_______________________________________________________________________
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:
