configuring two slave loops from 1 master loop

RE: configuring two slave loops from 1 master loop

Pierre_Dion — Mon, 17 Jun 2013 12:05:30 GMT

Thanks SMAQLL. I thought about that way of doing it too. My concern is, although the two control valves are identical "at the moment", what if one is replaced by a different one, or a different controller is installed, which needs a slightly different tuning?

I'm thinking about building two complete loops, but enabling each loop with a calc block, depending which reboiler is in use.

RE: configuring two slave loops from 1 master loop

SMAQLL — Sat, 15 Jun 2013 03:20:26 GMT

I think Pierre mentioned that there's only one flow meter. Considering this and the fact that both reboilers are identical I would recommend the configuration based on one master control loop and two identical flow slave loops (this will eliminate any tuning concern as both slave would need the same tuning constans, asumming both control valves are identical). I think the simplest way to retrieve the BKCAL's from the two flow loops is using a calc block with one input used as selecting value (I usually wire this calck block input from an input parameter that has a named set that allows operators to select a mode from the corresponding operating picture, for example: Flow Loop A and Flow Loop B modes. The input parameter of course is created only for this purpose). This way you can select the feedback from the only control valve that is being used.

Let me know if you still have trouble building this control strategy.

RE: configuring two slave loops from 1 master loop

Lou Heavner — Fri, 14 Jun 2013 20:31:14 GMT

I would use a flow loop for each steam valve, assuming there is a steam flowmeter. And I would use a selector block. The bkcal of each downstream loop would come into the selector block and the bkcal coming out of the selector (going to the temperature loop) would be based on the selected loop. But even if there is no steam flow PID, the AO driving the valve will also have a bkcal that can be used to update the selector block.

I assume if this is a case of in service vs out-of-service for cleaning, there are going to be isolation valves and decon procedures and it would probably be a batchlike operation. If the isolation valves are automatically controlled by the DeltaV, you could use the selector state as an input to logic or a calc block that also directs the isolation valves.

RE: configuring two slave loops from 1 master loop

Pierre_Dion — Fri, 14 Jun 2013 19:20:27 GMT

Thanks Lou.

The reboilers are identical, and they are switched when the one in use becomes too dirty. So they are operated one at a time.

My problem is the feedback from the FCV. I can output the same value for both, only the one in auto will follow the setpoint but I have to read the feedback from the right one. I thought about a calc block but wasn't sure it would work.

I will try it th is way.

RE: configuring two slave loops from 1 master loop

Lou Heavner — Fri, 14 Jun 2013 15:10:37 GMT

My first inclination would be to use logic to switch between one reboiler and the other rather than a splitter. But this is something I haven’t really seen, so I have to ask… Are the reboilers more or less identical, or are they different and selected based on what is being separated? If the reboilers are not essentially identical substitutions, then you are probably going to have a different process gain and your temperature loop tuning will need to be changed depending on which reboiler is in service.

Is the process bottoms diverted from one to the other depending on which reboiler is in service? Are you using logic to manage flow on the process side? Most of the examples of dual reboilers I have observed used both reboilers simultaneously. In one case, the column was built with two identical reboilers and we used a simple splitter which sent equal outputs to both reboiler steam flow loops. In other cases, the reboilers are heated with streams of differing value, perhaps a utility like steam or hot oil in one and waste heat from another part of the process in the other. Generally in that case, it was desirable to minimize the use of the utility, which simply supplemented the recovered process heat.

If you use a splitter or a selector, then the bkcal from the downstream flow loops both come back to the block and it generates the required bkcal for the temperature loop PID block. Alternatively, you can intercept the bkcals in a calc block and output the one you want back to the PID block. I’ve done both ways in the past. If you decide to use Predict Pro, there will be some issues with switching the MPC PRO block between local and MPC as regards the mode of the downstream loop, so it needs to be planned out, but it isn’t that difficult. I’ve done that before, as well.

Regards,

Lou Heavner