Controller action with fail open (air to close) valve

I'm not sure exactly what you are asking, but that has never stopped me
before.

In a FTO system the safest situation is to have the valve open, possibly to
relieve pressure.  Fail closed attempts to contain whatever is in the system
or stop it from continuing onward.  While there is sometimes a relationship
between these and whether the valve control is reverse acting, they are
often two different analysis.  The failure analysis assumes no further
control is going to be possible and the system must be rendered safe.  The
control analysis assumes that control is ongoing and the system must be
maintained at some set point.

A simple example.  A pipe carries an inert to a reaction vessel.  If the
inert  flow is too high, the reaction will be quenched.  You desire to keep
it flowing at 10 lpm.  You use a negative acting valve to keep it from going
too high.  The failure mode is quite possibly fto, so that if there is a
power failure the reaction is quenched immediately.  The reactants may have
the same sort of negative action but have an ftc in order to stop reactant
flow to the vessel in the power failure.

The operator display will depend upon what sensors you have and the
intelligence of the system.  If the valve has position sensors, and they are
powered, they will read correctly.  If the valve position is dictated only
by air pressure or a voltage to the valve, with no feedback, then your
operator screen is pleasant fiction.

Michael Herman



In either of these, you can use reverse action if necessary.

I realize you didn't include Infi-90 in your list, but I will tell you
what we do at our plant, and it may be of some help.
First of all, after the valve action has been chosen, then you pick
your controller action.  A simple flow loop that has a Fail Closed
Valve may use a reverse acting controller.  Change the valve to Fail
Open and now the controller must be direct acting.  Of course, you
always pick the fail action of the valve first for what the process
needs, and then choose you controller action after.
Now about Fail Open valves...
There are a lot of people that will say this is a bad way of doing
things, but if the operators are aware of how things work, it seems to
do OK.
Most of our valves have valve positioners on them, but no feedback for
actual valve position comes back into the DCS.  We make the controller
output look like it is valve position (because 95% of the time it is)
on the graphic.  If a valve is Fail open with no characterization we
look at the manual/auto station output as valve position for the
operator.  Then we do a switch internally before the output to the
analog output card that makes 100% = 0% and 0% = 100%.  That way the
operator sees 0% as closed when there is really 100% drive signal going
to the valve to close it.  And he sees 100% as open even though there
is really 0% drive signal going to the valve to open it.

Mod 300 is old, so it's likely got a bunch of proprietary Function
Blocks in it and this can likely be done easily.  The new controllers
with all their open standards don't seem to have the flexibility that
the proprietary stuff has (so I've heard), and I've never done any
configuring with them so I can't help you in how it may be done in
code.  Hope this helps.
Curtis

Thanks for the information.

Infi-90 was not on my list of DCS's I am interested in. However, it is
probably pretty similar. Right now I am specifically interested in
Delta V and ABB's latest, 800XA. But I have heard that there is quite a
bit of Baily influence in the 800AX. I also understand that 800XA is
based on the IEC 61131-3 standard. I don't know if that covers this
issue.

Is Infi-90 based on open standards or proprietary functions?

Thanks.

Infi-90 was developed long before open standards were around.  It's all
proprietary.  The Delta V think is interesting - and makes good sense.
I would be the 800XA is similar.

Curtis

Anita,

DeltaV has two parameters for controlling the valves in this situation.
 The PID block is set up as direct or reverse acting according to the
process.  The AO block connected to the PID block has a check box under
either control options or I/O options (Sorry, I'm not in front of a
station right now) that is call increase to close.  Set this option for
a fail open valve.

With this option set, 0% out of the PID will close the valve and 100%
will open it.  The 0% is converted by the AO block to 20mA and the 100%
converted to 4mA if you are using 4-20mA outputs.  Of course, if the
increase to close checkbox is not checked, 0% gives 4 mA and 100% gives
20mA.

I prefer this approach.  A 0% out always closes the valve.  It prevents
operators from making mistakes.  In addition, a standard flow loop is
always reverse acting.

I think this came from the Provox days when the parameter was called
Air to Close.

Positioner feedback is handled much the same way.  The fail action of
the valve has nothing to do with it being opened, closed, or anywhere
in between, it has more to do with how it gets there and what happens
on power failures, loss of signal, etc.  A couple of the other
responses had very good input on where to us a FO vs a FC valve.

I can't help on the ABB products.  Hope this helps.

nathan

The ABB MOD 300 and Advant work the same way as the DeltaV (or
maybe that should be the other way around, given their relative ages).
Within a continuous control loop, the PID function is set for direct or
reverse acting to match the process while the analog output function
connected to the PID output is set for normal or inverted to match the
actuator. Normal scales 0-100% to the low-to-high range output counts
(i.e. 0-32000 for TRIO aka Genius I/O). Inverted scales 0-100% to the
high-to-low range output counts.

Mike

Anita, as you see, the answer to your question is "It Depends".

Mike, thanks for the good description of how Mod 300 handles fail open
valves.

I worked for Taylor Instruments (now a small part of ABB) when Mod 300
was developed and the issue of fail open (air to close) valves was
handled. A little background:

In the analog controller days Taylor (and I think other) controllers
would reflect the valve operation in the controller action. That is, a
flow loop with a fail open valve would be set to direct action.

There was always a way to indicate to the operator the position of the
valve rather than the level of the signal. Early pneumatic controllers
had stick on OPEN and CLOSE labels; later electronic controllers had
slide switches that would invert the operation of the output meter and
manual open and close buttons.

In the late 70's the early computer control mini-computers were
connected to analog output boxes and would send up or down pulses to
analog boxes that would generate the 4 - 20 ma signals. The analog
boxes had reverse action functions to decrease the current when an up
pulse was received. So from the stand point of the computer -- the PID
algorithm, the operator displays, and other functions -- a flow loop
was always reverse acting. That made the software configuration much
simpler. To configure the PID action you looked only at the process. In
a complex loop (override, etc.) the signal was in terms of percent
valve position.

In the early 80's we carried the same philosophy into Mod 300. A
function block handles the analog output and is configured with such
things as the channel number and whether or not the signal was
inverted. The output of the PID function block represented the valve
position. That PID output is displayed to the operator. The color of
valves on graphics, etc. is based on the output of the PID without
regard to the valve action.

Personally, I think this is much simpler, particularly in the more
complex loops. 100% is always open (even if the analog output converts
it to 4 ma.)

I have heard that most, but not all, DCSs use the same scheme as the
Mod 300.

There has been some confusion among customers and engineering firms
that worked with other types of controls that used the older analog
form. (I have received more than a few calls at home at night from
sites going through startups). I have a drawing illustrating the two
methods of handling fail open valves that I used when I was at ABB. It
is on my web site at <a
href="http://www.jashaw.com/pid/action">www.jashaw.com/pid/action</a> .

I do not know how the new 800XA handles the issue. I think that there
is more Bailey (recently acquired by ABB) than Mod 300 influence. I
will soon find out about how 800 does it.

This is an important issue (nothing can mess up a control loop more
than having the wrong action - and, with complex control loops,
correcting it can be time consuming during a startup!). Unfortunately,
it is very difficult to ask a DCS vendor and get an explanation. I
don't think IEC 61131 addresses the issue.

Anita, as you see, the answer to your question is "It Depends".

Mike, thanks for the good description of how Mod 300 handles fail open
valves.

I worked for Taylor Instruments (now a small part of ABB) when Mod 300
was developed and the issue of fail open (air to close) valves was
handled. A little background:

In the analog controller days Taylor (and I think other) controllers
would reflect the valve operation in the controller action. That is, a
flow loop with a fail open valve would be set to direct action.

There was always a way to indicate to the operator the position of the
valve rather than the level of the signal. Early pneumatic controllers
had stick on OPEN and CLOSE labels; later electronic controllers had
slide switches that would invert the operation of the output meter and
manual open and close buttons.

In the late 70's the early computer control mini-computers were
connected to analog output boxes and would send up or down pulses to
analog boxes that would generate the 4 - 20 ma signals. The analog
boxes had reverse action functions to decrease the current when an up
pulse was received. So from the stand point of the computer -- the PID
algorithm, the operator displays, and other functions -- a flow loop
was always reverse acting. That made the software configuration much
simpler. To configure the PID action you looked only at the process. In

a complex loop (override, etc.) the signal was in terms of percent
valve position.

In the early 80's we carried the same philosophy into Mod 300. A
function block handles the analog output and is configured with such
things as the channel number and whether or not the signal was
inverted. The output of the PID function block represented the valve
position. That PID output is displayed to the operator. The color of
valves on graphics, etc. is based on the output of the PID without
regard to the valve action.

Personally, I think this is much simpler, particularly in the more
complex loops. 100% is always open (even if the analog output converts
it to 4 ma.)

I have heard that most, but not all, DCSs use the same scheme as the
Mod 300.

There has been some confusion among customers and engineering firms
that worked with other types of controls that used the older analog
form. (I have received more than a few calls at home at night from
sites going through startups). I have a drawing illustrating the two
methods of handling fail open valves that I used when I was at ABB. It
is on my web site at www.jashaw.com/pid/action.

I do not know how the new 800XA handles the issue. I think that there
is more Bailey (recently acquired by ABB) than Mod 300 influence. I
will soon find out about how 800 does it.

This is an important issue (nothing can mess up a control loop more
than having the wrong action - and, with complex control loops,
correcting it can be time consuming during a startup!). Unfortunately,
it is very difficult to ask a DCS vendor and get an explanation. I
don't think IEC 61131 addresses the issue.

Thanks to all for your help. I would like to hear about other DCSs, but
the ABB 800XA is the one I need to know about most (Delta V was the
other; Nathan described it well).

It would be helpful if this was standardized. Does IEC-61131 or any
other standard address it?

I certainly prefer the method Nathan described for Delta V and Mike and
John described for Mod 300. That is, configure the PID block based on
the process and configure the analog output to invert the signal if the
valve is fail open.

Anita.

The ABB 800XA handles air to close valves through the use of a function
block to invert the output before the analog output. So the PID
function is configured based on the process, without regard to the
valve action. Then, if the valve is air to close (fail open) a function
is added to invert the signal. Anything connected to the PID block will
see a PID output based on the desired valve position, not the signal.

I don't know about other DCSs not discussed in this thread, but I
think any DCS could be configured like the ABB 800 by setting the PID
based on the process and using a function block to invert the signal.

I do not think that IEC 61131 or IEC 61499 address the issue.