Ozone dose control

Greetings everybody...
Well, I came to you with a little difficult problem controlling ozone dose for a process. The principal issue is to mantain certain ozone
concentration in a buffer tank even when water with the dissolved ozone (at x concentration measured in PPM parts per million) is consumed.
The process is relatively simple to describe:
1) The tank is filled until certain level. 2) When the level setpoint is reached, the sistem keeps monitoring it to mantain that level when the water begins to be drain out the tank. 3) At the same time that the level control is actuating, an ozone generator is turned on with a setpoint determined by the operator. The output of the PID block used to control the power of the generator is driven to it by a 4 - 20 mA signal.
The problem is the very slow response of the system to a variation of the CV, (about 3 minutes), so the use of a pure feedback control like PID that actuates based on the error between the SP an the PV seems to be unuseful in this case. All the time that the PV is lower than the SP the power of the ozone generator increases so when the PV is a little higher than the SP all the ozone dosificated when PV<SP increases the concentration and the process is never controlled.
I know maybe the constants of the PID loop are no adequate but I'm thinking that a feedforward control must help in this particulary case.
Any suggestions will be wellcome and very appreciated...!!!
P.D. I'm using a PLC to control all the system, is an Allen-Bradley's Compact Logix L32E with RSLogix 5000...
Thanks for all your help...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

You have a few choices to consider. One is to add some derivative action to the controller, so that it starts the generator producing more earlier in the problem. The second is to add a wee bit of integral control to bring the system back to setpoint.
If you know something about the demand, then it would be reasonable to add a bit of feedforward control to start the generator a little before a draw is taken.
Another consideration is to increase the size of your ozone reservoir, which makes the more advanced controls a little less critcal.
You might consider modelling the system to determine that it is indeed controllable to the extent you wish to control it. If the ozone generator takes a certain time to work, and has a certain ramp up time, you may not be able to make it recover much faster while staying within the concentration constraints of the system.
Michael
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear AleksF:

Say a little something about how the ozone is applied, and how the makeup flow is controlled.
The outflow does not present "anomalous demand", the inlet flow does. Feed forward on makeup flow would be good. Thorough mixing in the tank would be good. Ozone applied to the makeup flow would be better, with its own PID loop.
Poor boy's feed forward... If you had a constant makeup rate, you could have a "small step" that you add to the ozone generator power command signal, additive to the PID loop command. If the flow rate was a function of "level deficit" then you could scale that additive value proportionally.
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks Michael and David...
Your apportations and feedback is very important to me, and i want to thank you for your time and will to help me.
About the Michael's answer, indeed I've tried to increase a little the Derivative gain expecting that the loop anticipates the response of the system, but I note that the controller's output has a maximum only until the PV cross the SP with a positive slope. In other words, to make this works, I need to decrease the Proportional gain to guarantee a smooth response and increase the derivative gain to decrease the slope of the CV, but doing this I'm observing that the control action takes a lot of time to be observed in the output of the system.
Describing with a little detail the system I have:
1) A buffer tank with a total capacity of 4,000.00 litres. 2) The injection method is a 2" venturi using a recirculating bomb with a constant speed that takes water from the bottom of the tank, pass it through the venturi and feeding again the water to the tank by a lateral pipe. 3) I'm using an ozone analyzer to measure the concentration of the dissolved ozone into the tank. I'm using the signal of this analyzer as an analog input to my PLC using this information to control the system. 4) I have an additional bomb that drains ozonated water to the process, when the level runs down certain minimum level the system begins to restore the level by opening a valve on the top of the tank wich fills the tank with fresh water. 5) The ozonated water gets into a closed loop into the process, some part of this is rejected to drain and the rest enters again to my tank. 6) The valve that restores the level in my tank only act when the initial level is below the level setpoint and if the drained water flow is excesive and provoke a low level in my tank. Remember that I'm re-circulating water so in practice, the level in the tank is "constant".
About if the system is controllable, well, I knew about some other companies that can achieve control about +/- 10% of deviation of the proposed setpoint usign feedforward loops, one of these is Pacific Ozone with its DirectDrive technology.
I cannot pretend to achieve similar results, I really don't need it, as far as i know this strict control is only for human consumption products, when the setpoint of dissolved ozone rounds the 0.13 PPM where a difference of 10% compromise the quality of the product, in my case the setpoints rounds 2.0 PPM and my only goal is mantain the ozone in a "closer" neighboor near this SP.
I don't know if I can add images to my replies, maybe I can add some drawings or graphs that I've obtained during my tests.
Thanks for your quick answers, I'll be looking forward for your posts.
An apologize for the poor english, isn't in good shape yet...
Regards...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear AleksF:

...
"bomb" = "pump"

When I was designing these systems, I sampled the dissolved ozone at the outlet of this pump... this gave me a constant example of what was available to deliver to process and it was a a very constant pressure.

I made sure the recirculation pump for adding ozone, was tee'd off the connection that this pump draws from. One outlet, and monitoring the dissolved ozone at this point.

Is this an on-off type valve? Is this function controlled by the PLC?

When it goes to the process, does it go through a UV unit to destroy the ozone? If so, the return flow also has no ozone, and presents a demand.

The tank represents an unstirred "time delay" as water with essentially no ozone drifts slowly to the exit. You must add ozone directly to the influent streams, or you must stir the tank.

No need. I cannot see them in Google Groups anyway.

You are doing really well.
You will do much better if you: 1) stir the tank 2) add some ozone to the influent streams that have no ozone 3) stir the tank 4) reduce the volume of the tank. 5) stir the tank
By the way, did I mention "stir the tank"? ;>)
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Greetings David...
Thanks for the corrections, specially that of the "bomb", surely the CIA is now tracking my post due to my bad english...;)...
On the other hand, this is a Ozone CIP (Clean In Process) system, so no need to use UV lamps to destroy the ozone at the discharge of the pumps, and yes, the valve that get in fresh water into the tank is a valve controlled by my PLC based on the measures provided by a level sensor at the bottom of the tank.
The measuring point at the outlet of the pump seems to be a good idea if the process involves a filler but in this case, I drive the ozonated water through the pipes in the plant with the "deliery pump" and monitor the dissolved ozone at the outlet of the "return pump" (the pump that return water from process to my tank). When the concentration of dissolved ozone is greater than 0.1 PPM, then I close the drain valve and open the return valve which allow the water get into the tank again.
As you see, the water returns with a little amount of ozone (that can reaches 0.5 PPM in some cases) so tend toward increase the dissolved ozone measured in the tank.
Please let me know what you think about this...It's really helpful to me...
Thanks for your interest and your patience...
Regards...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear AleksF:

I would do similarly, if I had to try to communicate in your langauge. No shame.

... and no need for the "contact tank".

When you add water, add a couple more milliamps to the signal to the ozone generator, just a few seconds before the "slug" arrives. You can track the time between start of addition, and the associated drop in dissolved ozone. Maybe maintain the adder for the same duration the makeup valve was open.

Somone saved the price of another dissolved ozone monitor. Should have measured the dissolved ozone in (and control with it), and measured the return value separately. When it comes back with "enough" dissolved ozone, the piping is clean.

Yes. It is good that you do not return the flow with low ozone levels.

All the advice I gave you before still counts. Let me explain a little about your tank. These devices were intended to provide "contact time", providing water that had ozone applied for some period of time, to assure a kill with relativtely low levels of dissolved ozone (such that it would not adversely affect the plastic containers or make bromate). You wanted plug flow then, with no stirring. So the tank, and the piping as you describe it (piping volume divided by flow rate), adds a time delay that prevents tight control.
Stir the tank. Add a second monitor, and slave the ozone generator to that monitor.

Glad I can be of help. I hope you can implement some of these suggestions, especially "stir the tank". If you can do that, you will not need to do my silly programming suggestions, just "autotune" in the usual way.
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I have a feeling that the ozone generator itself is a little non linear, so that may complicate things a bit.
Do you have any flow rate information for water entering or leaving the tank?
Since you have a PLC, you can do a little more control than ordinary PID controls. You may be able to do a model based control.
I think I might set up one register for the "percent seconds" required of the ozone generator. A controller will decrement this by the seconds times the ozone generator percent. It will perhaps try to rebalance things as fast as possible, or within so many seconds.
A second set of controls will add to the percent seconds register to account for water added to the system. Basically, each water addition would require a certain number of percent seconds to be added. A third will check the actual concentration and make sure that the percent seconds is large enough to cover the needed amount, again using a concentration and volume model.
This type of model control may work a lot better than a standard PID control for you.
Michael
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Greetings everybody...
First I want to give a serious apologize because it's until now when I can post my results on this group. Unfortunately, I have to move to other plant where we are developing the same project and my access to the web became null.
Well, I finally found an answer to my problem, thanks to all the information and advices from each one of you, I've found a very good way to mantain my ozone concentration even when the system is perturbed by fresh water or ozonated water.
The process that I'm controlling indeed could be controlled by a PID, but the proportional gain of the loop must be very weak, this allow me to give enough time to the PV to react at every change of the CV, so basically, I give a low Kp so the system takes around 6 or 7 minutes reaching the desired SP wich normally rounds 3 or 4 PPM in a volume of 2000 litres of fresh water. The integral and proportional gains were 0.115 and 0.37, this integral gain helps to mantain the PV near the SP and derivative gives a smooth response of the system to changes in the PV.
Additionally to this I'm adding some extra percent to my CV (which in this case is a 4-20 mA power reference to the ozone generator), every time that fresh water are restored to the tank, this allow me to give extra power to the ozone generator and keeps the PV near the SP when the system is perturbed.
As a piece of advice to all that like me our core business isn't process control, in slow procesess like mine, don't expect that the PID loop can be able to stabilize the system too much quicker, as an example, my actual Kp is around 0.0004 and it takes around 15 to 20 minutes to obtain an smooth, flat and constant concentration in the tank.
I hope that this info helps you in future projects, sorry again for the long time disconnected of the group and thanks again to all people that dedicate some minutes to give a hand to a desesperate comrade.
I'll expecting your replies and success cases if there's one. I'll keep in touch...
Regards
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear AleksF:

...
Thanks for the update. Your life would have been much simpler with a dissolved ozone monitor on the loop that adds ozone, and another on the return from the loop to be sanitized.
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear David,
Indeed you're right again, maybe I haven't been so specific describing my system but actually it has a dissolved ozone monitor in the preparation (and recuperation) tank, that allow me to prepare my CIP cleaning agent (ozoned water), then I send it using a pump to the process, at the return of this black box that can be a tank or a line, i have another dissolved ozone monitor that gives me the feedback of the concentration at the return point. If the concentration is lower than a desired setpoint at the return, then the water is drained. On the contrary, if the water reaches this setpoint and holds the concentration over it, then the water is returned to my tank (recirculation).
Thanks for your reply, personally, I'm want to thank all the time that you take helping me with my problem.
Regards.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear AleksF:
...

This is much better. But it is different than you described here: http://groups.google.com/group/sci.engr.control/msg/322c0ca5096eb56f ... but this does not matter now.

You have been very appreciative. Do not feel that you have not been. I only posted "over your shoulder" to posterity, so that those that come across this thread on the internet, get a (more) full recap. See, you are famous now... http://www.docendi.org/re-t78990.html http://newsgroups.derkeiler.com/Archive/Sci/sci.engr.control/2008-03/msg00001.html http://www.talkaboutscience.com/group/sci.engr.control/messages/47336.html ... in addition to the dozen google.groups copies in various langauges

To you as well.
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 03 Mar 2008 20:48:47 -0800, AleksF wrote:

Per dlzc, feedforward on the inflow rate seems like a good idea. This will let you yank the CV around with a great deal more authority than you apparently can with the PID. Assuming a more or less linear system, in fact, you can do just about anything you want to with adding in commands to the CV without affecting stability.
I suspect, however, that no matter what you do, you can get an inflow that will overwhelm the ability of your generator to keep the concentration up. Depending on how important it is to keep the concentration steady, how rapidly the inflow must change, and the dynamics of the ozone generator, you may need to control both the inflow and the ozone generator (at the risk of running the buffer tank dry) to keep the concentration within limits. Some simulation or on-paper calculations may be a good idea, to make sure you're getting it right before you make expensive changes.
Some sort of premixing of the inflow, so that it hits the tank contents at roughly the correct ozone concentration, would be a Good Thing* -- there's not much you can do with the ozone concentration when you slosh a bunch of fresh water into it, so making sure to slosh pretreated water into it will help a bunch.
* I don't do process control, so I have absolutely No Clue about how one may do the premixing, or if it is even practical.
--
Tim Wescott
Control systems and communications consulting
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dear Tim Wescott:
...

It is and your advice is good.
Premixing is OK, but he has not told us everything about his loop yet, I think. With this volume tank, he will still get a slug of "low ozone", since it is plug flow downwards, and the ozone generator is supplying makeup only for "natural ozone decay" until the slug arrives.
I did one of these loops for a bottled water manufacturer, and there are a couple of ways to improve it. Easiest (in terms of premixing) is to add the makeup flow to the inlet of the ozone application pump.
Better still is to eliminate the tank, apply the ozone directly to the flow stream, and degas the bubbles out before the process gets the fluid ("GDT process", whom I worked for until we made a profit :-O ).
David A. Smith
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.