PID question

The last project I did on just such a system consisted of a tank, heat exchanger recirc pump for heating and a pump to feed 95 Deg. C water to a cooker, the cooker was an in-line system that cooked rice (first in the world I believe it was even sold to the Chinese). The water was sprayed via 6 zones using a mark/space ratio depending on the speed & hence the feed ratio of the dry product, the screw feeder also agitated the product on a forward/back ratio. the original hot water supply struggled to maintain temperature as well as pressure so in effect caused variations in water takeup of the product. The valves were zoned so it was possible to have just some or even all valves open at the same time & as a consequence caused major pressure variations. We added a PID controller to try to compensate for the pressure increases/drops but no matter how we tried it we could not get the control we required. We then designed a new system with a larger capacity tank standby/duty pumps (made maintenance easier whithout shutdowns etc. We called in an expert for the design of the heat exchanger pack based on our requirement and although they were not involved in the pressure control of the feed water the company suggested from their vast experience to ditch the PID controller and use the reverse acting pressure reducing valve in the return line. This proved to be the best choice, no need to set up complicated PID settings, purely mechanical (well hydraulic) low maintenance and fast.
Sometimes the simplest is the best. looking at the pressure log showed almost a strait line but bearing in mind the logging frequency and the filter we added to the log.
 
I'm not doubting at all that a valve control would be faster and probably more accurate for an application such as spray heads. No way a VFD could react fast enough to maintain a tight pressure (short term) with the valves cycling quickly.

The advantage of a VFD for the heat exchanger type applications are:
- Energy Savings
- Water Hammer reduction
- Others? I'm sure there are.
 
This is what we found, however, there was no need to put a VFD on the heat exchanger recirculation pump as this was sized for the flow rate and heat uptake, the PID on this though had to be forced to a pre-set level during water addition to maintain the heating loop as this was injected directly into the heating loop and not the tank as this would drop the tank temperature so basically, the modulated steam valve was forced for a time during water addition to overcome the slower response of the PID heating loop, this maintained the overall storage tank temperature to the required setting. The feed pump to the injector loop was controlled by a VFD but was fixed at a level to maintain enough pressure when all injector valves were open so as not to overdrive the reverse acting pressure control valve.
 
Not looking to pick a fight, [...] I have seen very few pumps that are linear

Using a VFD and speed control to control pressure is very common. Unless you want to waste energy and use a pump chase or pressure control valve

Thank you.

A debate/discuission is not a fight, and anyway I repent in dust and ashes, and also revel in my participation in "the best way to get the right answer is to post a wrong answer" algorithm, even if I often seem to be on the latter half of that.

Regarding linear: of course pumps are non-linear, but what was meant by linear was "linear enough at that pressure" i.e. around the control range, which is enough to control*.
Caveat: control will get interesting when the number of open zones changes; still near-linear but the slope will change, so finding parameters suitable for all-zone-, one-zone-. and maybe no-zone-operation could be challenging.

Regarding VFD speed controlling pressure: Yes, I was speaking ignorantly and well outside my area of knowledge (a human's third greatest problem: thinking I am normative;-); a quick search on The Google yields plenty of examples, with discussions of pump curves and lots more.

"pump chase" - that is a new term to me; is that a recirc with pressure relief/reduction?

* "linear enough" is true of most processes, which is why 90% of engineering is multiplying by 1
 
I would be neat if DP across the pump mapped to torque, then the VFD could be set to use torque control.

That may be true on a PD pump, but centrifugals/fans no so much (if I increase DP by blocking flow to a cooling fan, which has a more or less constant-power motor, then the fan speeds up; since power ~ speed times torque, torque must be going down).

To answer the OP "Can someone explain me relation with DP and pump speed and how to set a pid loop parameters," the sum of the matter is that no one on this list knows enough about the process to provide specific answers; the best to hope for is someone saying "I have a working N-zone system and Kc is X; Ki is Y; Kd is Z," but those would have to be re-scaled to the actual inputs and outputs ([0:%]? [-16384:16384]? [-32768:32768]?), and it would still not be known if the system curves were similar enough.

Look at pump curves, run the system on manual to see how DP responds to pump speed in various configurations (zero to all zones open). If the results were posted here (spreadsheet or strip chart of speed and DP response vs. time for step changes in speed), then maybe someone here could estimate some conservative starting PID parameters.

Then google how to tune a PID: there are better methods available but they require a PO sent to a consultant. The main things are patience, keeping each adjustment small, and understanding how the PLC manufacturer's PID algorithm handles time. If those tuning guides, or proportions, are confusing, then keep a hand on the E-stop or write up that PO.
 
Paradigm shift

Assuming the motivation for VFD speed control of DP is energy savings, there may be bigger fish to fry thermodynamically.

Assumptions
  • Heating with combustion flue gas (i.e. fossil fuel e.g. gas)
  • Pumping hot water through some kind of radiator to heat air
  • Water flow rate (velocity, laminar vs. turbulent) is not critical because air heat-transfer coefficient is limiting the overall heat-transfer coefficient.

Given all those assumptions, the most important being about flue gas, a possibly more important aspect of energy savings is the temperature of the flue gas leaving the process (via stack, chimney, whatever). If the water is heated with electricity or summat else then the rest of this can be ignored.

The lower the final flue temperature, the more energy efficient the heating system.

To lower flue temperature, lower the temperature of the water return from the zones' radiators, to extract every available Btu out of the flue gas.

To lower the water return temperature, lower the flow rate: constant total heat load with lower flow more heat extracted per unit mass of water).

How low the return temperature can be is a function of the radiators' heat transfer characteristic (area, etc.) and the maximum heat load required (minimum outside temperature, target rooms' temperature, rooms-to-outside heat transfer characteristics).

The point is that the controllable parameter that might most strongly affects overall system energy efficiency is the water return temperature, not the differential pressure, although the design may be that the latter serves as an analogue for the former.

Of course, controlling return temperature will be very different as it will respond to changes in pump speed somewhat more slowly than the speed of sound. So while I might be on firm ground wrt thermo, it may not work in practice. Also, there could be second-order effects that cancel out any savings e.g. rooms will warm up more slowly because mean water temperature in radiators is lower, so times when all zones are closed and boiler/furnace/whatever is off will be less.
 
Thank you.

A debate/discuission is not a fight, and anyway I repent in dust and ashes, and also revel in my participation in "the best way to get the right answer is to post a wrong answer" algorithm, even if I often seem to be on the latter half of that.

Regarding linear: of course pumps are non-linear, but what was meant by linear was "linear enough at that pressure" i.e. around the control range, which is enough to control*.
Caveat: control will get interesting when the number of open zones changes; still near-linear but the slope will change, so finding parameters suitable for all-zone-, one-zone-. and maybe no-zone-operation could be challenging.

Regarding VFD speed controlling pressure: Yes, I was speaking ignorantly and well outside my area of knowledge (a human's third greatest problem: thinking I am normative;-); a quick search on The Google yields plenty of examples, with discussions of pump curves and lots more.

"pump chase" - that is a new term to me; is that a recirc with pressure relief/reduction?

* "linear enough" is true of most processes, which is why 90% of engineering is multiplying by 1

Pump Chase is a line that runs from the pump outlet to the pump inlet, usually in this line is a valve to control how much water is allowed to recirculate.

This is a "cheap" way to control pressure/flow.
 
To be honest, without knowing your exact application, flow from no valves open to all open & frequency i.e. are valves likely to be switching on/off on 3-5 sec intervals or longer etc. The reason we chose the system we did was it was a high pressure line 10 Bar and it made sense to dump the excess pressure continuously via the return line I.E. rather than just have a open return line (restricted) we fitted the reverse pressure reducing valve across the feed and return, this gave us total control so at our given flowrate and pressure of up to 20 bar (with control valve closed, we had flow and return. using the control valve it would automatically dump the extra 5 Bar to return, when the valves open the pressure control valve would adjust to maintain the 10 bar pressure feed. We monitored the flow with a pressure sensor in the line & given the sample rate of probably 0.5 seconds there was no noticeable change in pressure, agreed, it may have spiked for milliseconds but this never showed on the logging. The problem you have is also the flowrate difference from no valves open to all valves open may be quite high, ours however being spray bars the volume was quite small but these valves were switching at a variable rate of between 5 & 30 seconds depending on the required water take up required by the recipe & residence time, product flow.
 

Similar Topics

Been working with PLCs for a couple of decades, but almost 100% DirectLogic. Have a customer who wanted me to make a couple of simple changes to a...
Replies
3
Views
1,099
New to the EcoStruxure programming software (I'm an AutomationDirect guy) and I've been banging my head against the wall trying to figure out how...
Replies
2
Views
991
I haven't had to tune a PID loop in a very long time. It's actually a PI loop for a pulse width modulation s.v. What was the name of that tuning...
Replies
16
Views
4,072
So I have a PID loop on an 1756-L61 running V17 software just for background. Also the PID PV is a pressure transmitter and the CV is speed sent...
Replies
1
Views
863
Hi All I'm using Cohen-Coon method to calculate initial PID parameters. So, I have Model Gain that has been calculated from live data. Could...
Replies
55
Views
30,974
Back
Top Bottom