Chiller Secondary Loop VFD Control

[gopack85]

I have a secondary loop VFD controlled with 10+ 2-way modulating valves and two 3 way valves. The VFD control is currently set to maintain a specific dp. In the winter time, the flow that this loop maintains is beyond what one chiller with a constant flow pump can provide (actual load in tons is ok for one chiller) so I'm forced to run two chillers and get a low delta T due a large amount recircing in the bypass line.

I am thinking about trying an alternate control method with the vfd secondary loop. Instead of maintaining a constant dp could I not use the actual % open of the valves? My control system would constantly be looking at the open % of each valve and attempt to drive the most open to 90% and stay there (by reducing VFD speed and total flow). If it gets above 90% then it would speed back up to maintain 90%. If another valve would open past 90% then that would become the new PV. Thoughts?

 

[DRWeig]

It sounds like you might have a good candidate for converstion to primary-only variable pumping.

http://www.eng-tips.com/viewthread.cfm?qid=327234

As to your question: If your secondary pump speed is controlling to a dp setpoint but the coil control valves are all mostly closed, why not just add a reset function to dp setpoint to achieve your goal. Lower the dp setpoint until you just barely satisfy the worst coil (as indicated by valve position), and raise it when the worst valve is fully open. Watch for another valve to come take the top spot, as you suggested. I have experience that says controlling a zone temperature or an AHU supply air temperature by modulating the pump speed (or air speed in fan systems) is NOT easy to stabilize, and the lag times are huge. I'd advise against it.

There are also many pitfalls and failure modes that do not exist in your current scheme. A stuck or dead valve actuator can affect one AHU or zone without interfering with the rest if controlling pump dp. If you're controlling pump speed to a valve position, that one failure can eliminate control of the system pressure completely. The other AHUs may be in trouble.




Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[DRWeig]

Sorry, I meant to paste this:

Primary Pumping

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[gopack85]

Wouldn't the lag carry over into the dp reset since its still based on valve %?

 

[gopack85]

I think I understand the method. What have you seen as a successfull reset method time? As in, how often do you look at the valve % and how much do you adjust the dp set point?

 

[urgross]

I don't think you need any special polling rate. You would need minimum flow rate for valve operation, probably about 6 psi. When valves are at closed, the dP would take over to maintain minimum pressure from off an in-line booster pump. Once valves start to open, valve position takes over. That's how I set a variable primary with distributed boosters.

 

[gopack85]

So you are actually controlling pump speed directly directly valve position once they start opening?

 

[ozmosis]

Some thoughts from a VFD manufacturer on the control of primary and secondary chilled water pumps with a VFD:
ftp://software.danfoss.com/Drives/RU/Drives/primenen_VLT_HVAC.pdf

Disclaimer: I do work for this company

 

[ChasBean1]

Keep maintaining the DP set point in your secondary loop. The two three-ways pass more water than the minimum flow required for each pump. Set circuit setters in each of the three way bypasses to maintain about a sixth of the secondary loop pump design gpm. If there are no circuit setters, close one of the two three-way bypass lines.

Stage chillers only on percent load (from the chiller panel) or tonnage. You can also stage to a second chiller based on inability to maintain the secondary loop temperature set point (e.g., 2°F above set point for X minutes), but you can’t stage down on temperature or differential temperature.

Don’t even THINK about staging to a second chiller based on a falsely high secondary load because flow is high through a couple of three way assemblies.

 

[DRWeig]

gopack,

ChasBean's comment about maintaining pump minimum flow is very important. You'll be limited in how much you can reset the dp. Implementing this strategy with air systems is much more effective, the range is greater. I've only done it on a hydronic loop once. I can't say the savings were worth the time, effort, and money.

To the questions you asked after my post above:

Yes, the time lag also applies to implementing dp reset. It is varying the pump speed directly in response to terminal conditions (valve position, temperature) that will neglect the possible needs of the other coils.

With today's building automation systems, polling the valve positions every scan or cycle does not add any burden. That makes it easy to determine the dead time you need to build in to accommodate the system time constant. Just data-log and measure the time between a step change and the beginning of a response. The minimum reset time must be at least 4X this value if you're doing PI control. Maximum is just a tuning issue. Expect a lot of unavoidable overshoot.

I have an IEEE paper somewhere on predictive-slope control, an alternative to PI / PID that avoids overshoot and handles loops with long time constants pretty well. I'll try to dig it up.

Oz - nice paper, as usual. Thanks.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[gopack85]

Thanks for the tips. My first thought was to try and change to a variable primary system but I'd have to show that the cost of retrofitting an existing system would be less than the overall savings. Right now I'm trying to do some "free" control modifications to optimize the current set up a little better.