Chilled and hot water pumping

[ssn61]

Is there any specific requirements on pumping hot and chilled water in a HVAC system? I have seen water flow from outlet of the pump to the chiller and out to the building...hot water flow from outlet of pump to the building, return thru boiler and back to inlet of pump. Is there a merit to do the flow design this way or could we revers the flow as I have seen that done in the past as well?

 

[Noway2]

The pump will add heat to the water. In the chilled water case, you put the pump before the chiller so that the added heat gets removed. In the boiler case, the pump energy contributes to the heating.

 

[MTES93]

Another point of consideration is the air removal… where your air separator is.

Air comes out of solution at the lowest pressure and the warmest temperatures. On a chilled water system this would be on the chilled water return and on the suction side of the pump. Thus you are typically pumping into the chiller. Your expansion tank and air separator would be on the suction side of the pump on the chilled water return.

On boiler systems your warmest point would be on the outlet of the boiler. Thus you are pumping away from the boiler and out to the system. Here your expansion tank and air separator would be on the suction side of the pump on the heating water supply.

This is an overly simple view of systems that can be fairly complex depending on final arrangement of the system, building height and related. However, the general principal should be followed.

Always pump away from the expansion tank in either case.

 

[Artisi]

"Noway2 (Electrical)
10 Dec 12 9:33

The pump will add heat to the water. In the chilled water case, you put the pump before the chiller so that the added heat gets removed. In the boiler case, the pump energy contributes to the heating. "

Must be very inefficient pumps to be adding any heat into the pumped flow.





It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[Noway2]

@Artisi, this is a commonly accepted principle that has been put into practice at all of the chiller plants at the University I work and was a topic specifically discussed in a HVAC systems course I completed last fall. Keep in mind that a few degrees difference in the leaving chilled water temperature can have a major impact on the level of comfort cooling. The chiller will also be able to correctly compensate for the added heat in the return water much easier than in the leaving water.

 

[DubMac]

If the pump is adding a few degrees to the temperature of the water...then you got the wrong pump.

The heat added by a pump comes from its mechanical friction; the only way it can be added to the water is if the pump is operating far left on its performance curve, i.e. not moving enough water. It is not only adding heat, but wasting a bunch of energy also, as it is far left of its Best Efficiency Point.

 

[DRWeig]

The pump always adds a bit of heat to the water, as it is not 100 percent efficient. What magnitude? Y'all feel free to jump all over me if I do something wrong below. I'm tired, and I am aware that I shouldn't be attempting to think right now.

Here's a simple calc for a 100T system with 10°F CHW ΔT and 40 FT WC head at nominal full load flow:

100T x 3 GPM/T = 300 GPM

300 GPM @ 40 FT WC requires about 4 BHP when selected pretty close to the BEP of the attached pump curve from Deming / Crane.

Efficiency from curve at selection point is 78 percent.

4 BHP x 2550 BTU/HR/BHP x (1-0.78) = 2244 BTU/H added to water by pump paddlewheel work.

2244 BTU/H ÷ (300 GPM x 500) = 0.015°F ΔT across pump in the chilled water stream.

Now, even if the pump were only 20 percent efficient (unheard of), the answer comes out to 0.05°F ΔT.

Not much. This calc scales up very well.





Best to you,

Goober Dave

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

 

[Noway2]

This raises the point that it may also be a matter of scale. Your example, uses a 100T system, which is very small compared to the ones we use (a typical machine being about 2,500 tons). For example, right now we are operating one of our plants in economizer mode where we use the cooling tower and plate & frame exchangers. We are flowing 6000 GPM, with one pump drawing 63KW (this is about as low as our load will ever go). If the pump is 80% efficient, this is ~12.6Kw going into loss, including heat. In terms of tons of cooling, 1Kw ~= .28 tons (1kw = 3415 BTU/H and 1 Ton = 12,000 BTU /H). Consequently, this is about 3.5 tons of cooling loss in the pump out of the ~2340 tons that is being produced which is a fraction of a percent.

Looking at the SCADA system, I am also seeing about a .3 to .4 degree difference from the pump inlet and outlet. This is within the error of the RTDs, but it is pretty consistent amongst multiple meters.

 

[DRWeig]

Hi Noway2 -- I'm not saying your measurements are off, but I would be very suspicious of them. See calcs for the 6000 GPM / 12.6 kW case below.

I work with very big machines and plants too, university campus type stuff. I've never measured a difference across a pump that was larger than the error of the sensor/input combination except when the pump was running backward due to mis-wiring. Yours may indeed be an exception, but I can't explain how it would happen.

Now, I asked y'all above to shout out any silliness in my math. I'll be embarrassed as heck if there's a factor of 10 or 100 that I'm missing in here. Don't let me down. I can still do some math, but I'm no electricpete.

12.6 kW * 3413 BTU/KWH ÷ (6000 GPM x 8.33 LBM/G x 60 MIN/HR) = 0.014°F ΔT across pump in the chilled water stream.

Reversing the equation, the amount of power that would have to be lost in the water to create a 0.3°F ΔT across the pump would be:

0.3°F x 6000 GPM x 8.33 LBM/G x 60 MIN/HR ÷ 3413 BTU/HR/KW = 264 KW

It takes beaucoup heat to raise the temperature of 6000 GPM. 12 KW won't do the job.

OFF TOPIC TEMPERATURE SENSING RANT FOLLOWS! Read at your peril.

Note: the following pertains to HVAC type instrumentation. If you're in an industrial environment, your sensors might already be very accurate and your SCADA may be very precise. In our world, neither are really suitable for measuring tenths of a °F.

I've had to graduate into the extreme-accuracy RTDs and transmitters due to an absolutely silly building code in a country that I shall not name. The code requires that the chilled water temperature at the controller be ±0.05°C (±0.09°F). Oh my goodness. Even with a 1/10 DIN RTD (±0.03°C or ±0.054°F AT 0°C), a sniper-accurate transmitter, and 4-wire lead compensation, the ambient temperature variation error in both the sensor and the electronics knock the actual sensed accuracy up into unacceptable ranges for the building authority there. That is, unless you are his favorite local supplier. They actually take the whole rig -- sensor, transmitter, controller -- to the national standards lab to determine acceptability if you claim to be compliant and you do not have cigars each evening with the inspector. I have actually beaten that standard by quite a bit on the bench by careful calibration. But put it on a plane, send it across the an ocean, install it in the field and you'll find that it's out of the accuracy span by as much as 50 percent. That's verified by my own testing, too, not just the national lab there.

Describe your RTD / transmitter / SCADA type and wiring, Noway2, there may be a multiplication of error even if the RTD itself is accurate. Add the error sources above to the fact that the building automation system may be trying to resolve 0.1°F out of a 0-100°F 4-20 mA current loop with an A/D converter that might be only 8-bit. No way (no disrespect intended to your handle, Noway2 ).

It's tough to confirm that the spec I'm ranting about has been met unless the RTD is 4-wire, the transmitter is excellent ($$$), the controller has been calibrated to the nubs, and the ambient temperature of both the transmitter and the controller are kept fairly close to the temperature at which they were calibrated.

End of rant. Thank you for your patience. I hope I haven't offended any HVAC automation system suppliers.


Best to you,

Goober Dave

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

 

[ElSid1]

Pump will add an amount of heat to the system. Take a look at Air Side HVAC fan calculations (Pump=Fan just different medium). If the fan is after the cooling coil (suction side) then the fan heat is added to the calculation (even Trane Trace does this). Water and air have different thermal characteristics.

 

[DRWeig]

I've never measured a difference across a pump that was larger than the error of the sensor/input combination except when the pump was running backward due to mis-wiring.

Oops . I should qualify that statement by saying that I've only measured maybe three pump differential temperatures in my years. My sample is small. I don't have occasion to do that very often.

Best to you,

Goober Dave

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

 

[Noway2]

It is interesting that you should mention university type campus stuff, because that is exactly what this is.

If I recall correctly, the RTD (a Weed PT100) with a transmitter (Typically Rosemount - neither of which I could quote part numbers off hand) is about .25 to .5 degrees I am inclined to think it is largely error within the measurement too. It may be a 4 wire. These are then being read by an GE-MTL controller via a 4-20mA. The controller had an A/D bit accuracy that surprised me when I heard it (I think it was 20 bit or more but can't recall exact value). With respect to the A2D I also assumed that it would rapidly loose accuracy much beyond 12-16 bits due to the small signal nature. Calibration is an issue I have been dealing with. I am The historic practice was to use a dual block heater and try to calibrate to within .08 degrees, which is meaningless as the repeatability of the RTD is much worse than that. I've gotten them to go to a using an ice bath instead, but that is mostly on the building load side of things as that involves billing whereas the production is a cost side and kw/ton is the golden metric.

Based upon your numbers, I am starting to wonder why the pump before the chiller due to heating is so commonly pushed (and yes, this was definitely taught in the technical development program I recently attended). Interestingly, in response to this discussion, I took a quick look at our main plants. The two newer ones have the pumps before the chillers and are variable primary, the newest of which even runs the chillers on VFDs for variable capacity. The two older ones have pumps both before and after, but use dual headers and are primary secondary (or really primary-secondary-tertiary by the time you get to the building). Then our absolutely newest (a medium sized standalone facility at about 4000Tons) is variable primary with the pumps after the chillers.

 

[DRWeig]

Cool, Noway2. Your measuring stuff is definitely a couple classes above what I usually see in central plants. Your facilities team must actually care about doing things right instead of at least possible cost. That extra bit of accuracy can be employed to better control the plant and pay for the extra cost quickly in a plant that size.

I think that folks may generally be remembering that AHU fans add a much more significant temperature rise, maybe a degree F or more with typical HVAC static pressures. It's just not that big with water, but we think about it anyway.

The two latest designs I've seen have been variable-flow, all-primary, with variable speed chillers. Some amazing chiller kW/ton numbers, getting down to the 0.25 range at the 25 percent load mark. All of those had the pumps before the chillers.

Of course, throw in variable speed cooling tower fans and variable speed condenser water pumping, and my controls job takes MUCH more thought and work than before. If only owners would pay more...



Best to you,

Goober Dave

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

 

[ElSid1]

Yes, fans are different due to the properties of air vs water. Non the less, an amount of heat is added, by the pump...usually insignificant
Set up sounds like primary secondary for the original and primary variable for the newer system. Just a guess

 

[Noway2]

@DRWeig, the impact on the controls and programmers has been immense. In truth, over the last few years, the changes have required a complete shift in thinking and paradigm of thinking. It used to be, prior to my joining, that the mission was to produce chilled water and as long as we output enough tonnage to meet the demand and keep the differential pressure high enough that was all that mattered. The tightening of the economy has changed that.

Today, our newest plant(s) also use variable speed on the fans, condenser pumping, etc. In fact, we have been working with the energy engagement division of facilities to identify the buildings that would benefit by having VFDs and getting them in place there. Of course this is in addition to identifying the buildings that have poor return temperatures and high gallons per tonnage and trying to get these things fixed.

In some of the older plants, such as the ones with constant speed chillers, we've been adopting control strategies, such as finding better operation points, which are typically about 85% load, and adjusting the flow through them to achieve higher efficiencies. We've got such a wide swing in our wet-bulb temperatures here that we will go from about ~3,000-4,000 tons in the winter (largely process cooling, data centers, and research facilities), to about 40,000 tons in the summer. The spring and fall months have been opportunities for us to take advantage of the variability. One thing that we've been doing is finding ways to match production to the load, rather than simple step loading, at least during the spring, fall, and winter. This gave our programmers, especially those who were around for the old thinking, a lot of heartburn, but it works beautifully.

Other things we've done include using reclaim water for cooling towers. Use cooling towers with low approach, bigger 18 blade fans, and increasing the number of cells in use per chiller rather than stick to the conventional 1:1 design. Overall, we currently operate at about .65Kw/ton average for the year and have the goal of getting this down to .6 kw/ton. Of course this includes things like using the P&F in the winter. Going forward we are looking at getting some steam turbine drive machines for summer base load (we have a CHP plant on campus that needs a summer steam load) and replacing the old absorbers. Other things we do include the use of thermal storage (about 60,000 ton hours worth) and design to reduce pump lift (e.g. don't put cooling towers on the roof).

Yes, our philosophy has been to do the job right to achieve overall best costs rather than just what is cheapest to purchase (surprising for a govt institution).

 

[DubMac]

ENjoy the depth of the conversation and rigorous calculations, but.....

At the end of the day, if heat added by the pump has any measurable effect on a chilled water system, then you have the wrong pump.

 

[DRWeig]

Nicely said, Noway2. Truth enough, DubMac.

An enjoyable thread.

Best to you,

Goober Dave

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