calculating loop pressure drop with variable pumps in series

[DavidButler]

Attached is a diagram showing how thermal energy storage (TES) can be installed in parallel between chiller and load. The 2nd page is an excerpt from an ASHRAE Journal article by Kent Peterson on which this design is based. When only the chiller is operating, the zone valve on the fan coil is closed and the chiller will charge the tank. When only the fan coil is operating, the chiller zone valve is closed and the fan coil will discharge the tank. When both are operating, the tank will either charge or discharge depending on which flow rate is greater.

In the diagram, I show the fan coil having the larger flow rate, so the tank will discharge the difference between the two flow rates. This is an elegant solution as it avoids the need for a dedicated pump for TES with associated valves and controls.

When only one pump is operating, the loop pressure drop includes the Tee-branch into and out of the tank, but what I don't understand is how to estimate the PD at the Tee that each pump "sees" when both are operating. Hopefully someone here can assist. Thx
.

 

[LittleInch]

You need to just look at each loop flow as it occurs. To "charge" the storage tank the flow will be 5 in reverse flow to that shown. When the tes discharges with no flow through the chiller, the flow is 8 in the direction shown on youvre sketched. Third option is what you've shown. Pressure at any tee is the same in both branches. So long as your pumps are pd type then flow can be assured regardless of pressure loss, but with centrifugal pumps it will be very difficult

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[DavidButler]

LittleInch wrote: "as your pumps are pd type then flow can be assured regardless of pressure loss"

But I need to know what the losses are at various conditions in order to select my pumps. Depending on what's happening at the Tee, the worst case may be when both pumps are operating. The system is actually a bit more complicated (multiple air handlers and two chillers) but the pressures at all other points in the loop are straightforward.

 

[Ziggypump]

System curve, don't forget the system curve.

 

[DavidButler]

Ziggy wrote "don't forget the system curve"

Well, that's why I asked the question... I need to know the PD's at every point in the loop to develop the system curve, no?

Since my original post, I found a very helpful resource for this exact scenario: Caleffi Idtronics #17 Thermal Storage in Hydronics Systems (July 2015). Rather than attempting to characterize the PD at the junction of the chiller loop and load loop, the article recommends locating that junction close to the TES and making the junction large so there's virtually no head loss between tank and junction (ref: page 9).

 

[Ziggypump]

Pressure losses along the loop won't determine the system curve. Only the pump can do that. You can calculate what is needed, but you cannot dictate how the pump will work, unless you are creating the pump too.

 

[LittleInch]

Your diagram shows only one mode. You needed to draw all the modes, work out the flows in each section then add the losses together at those flows. The key is that the head at any junction has to be the same so some iteration is often required.

If your pumps are not positive displacement then you'll probably never going to work this out.

I didn't follow your latest info, needs a drawing.

Your charging off the TES will need reverse flow. It is not easy to control flows without control valves.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[DavidButler]

See concept drawing. This is for an off-grid home near Phoenix. To take maximum advantage of TES capacity, the load side flow rate will vary from less than 1 gpm with single zone call for smallest zone and water is in low 40's, up to about 13 gpm when water temperature climbs into low 60's during a multi-day chiller curtailment event (e.g., hot weather but with little or no sun), which is the reason for having large TES. We're controlling the proportional zone valves with custom control that sets the target H2O delta-T for each AHU based on EWT.

@LittleInch, why do you suggest positive displacement pumps? We haven't selected the load side pump(s) yet, but we will likely need 2 in parallel to increase flow range in order to get closer to the wide design range. The chillers comes with OEM Wilo Yonos Para RS 25/7.5 (European, bare-bones pump with PWM input).

I removed the zone valves on chillers (initially included to prevent TES bypass when chiller pumps is off), but I believe the chiller HX will create enough resistance (~8 ft of head) to keep leakage through that circuit to a minimum.

 

[LittleInch]

Ok, More and detailed info makes more sense than the somewhat simplified sketch provided earlier.

On this basis I would be tempted to make the chiller pumps PD so that they push through a fixed volume of water regardless of where that water is going. The load side pump should be centrifugal and probably needs to be two or three pumps in parallel to cope with your extreme turndown rate (13:1) or a set of small/med/ large pumps, say 1 gpm, 4 gpm and 8 gpm and bring them on and off in various combinations as required.

Then put your manifold next to the TES and your chillers some distance away so that the preferential path is via the TES when the flow through the primary load loop is higher than the load through the chillers.

Your chiller PD pumps need only consider the loop from manifold via the pump, through the chillers and back to the manifold. Your main pump needs only look at the loop from manifold, through all your valves and AHUs back to the manifold. You should make the pressure loss from manifold to manifold via the TES as low as you can (bigger pipes) so it can be ignored in the main pump pressure loss

Then your system should automatically balance. If you have more flow from chillers than you need in the house, this all goes via the TES, If vice versa, the main pump should then pick the slack from the TES. so the advice earlier is probably correct - make your manifold quite big and close to the TES

I would still install on/off valves on the chillers - simple solenoid valves open when chiller running, closed when not - not much power required and stops flow going through the chiller pumps when not in use.

does that help ( I forget what the original query was now....)

It means your TES will need to be designed to handle the full 13 gpm if both chillers are off.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Artisi]

LittleInch, Hope there is a consultancy fee payable at the end of all this

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.)

 

[LittleInch]

You can only hope.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[DavidButler]

I would be tempted to make the chiller pumps PD

These chillers are little 2-ton monoblocks with an integrated variable speed pump (the WiLo I referenced in my previous comment) and a Mitsubishi inverter compressor. The chiller's control board uses PWM to maintain the user-selected LWT and Hx delta-T. The chiller's internal control board includes an auxiliary pump control output in case the source loop PD exceeds the internal pump's range. We won't have that problem.

I would still install on/off valves on the chillers

The chiller maintains a trickle flow between cycles to ensure valid EWT readings. At present, we don't plan to (or want to) modify the chiller internals (e.g., use different pump or control strategy). The only mod we plan is to move the internal EWT sensor over to the TES to prevent short-cycling the compressor on false EWT readings (due to potential heat gain in the 20-foot buried trunk on hot days).

The load side pump... probably needs to be two or three pumps in parallel to cope with your extreme turndown rate (13:1)

We'll almost certainly end up with 2 load-side pumps. My expertise is mostly air-side so I asked the guy who's building my control system to specify the pumps. He was adamant about using equals. I'm not too concerned if we can't get all the way down to 1 gpm for the smallest zone. That's the PV Room (batteries & inverters for the solar system). The only downside of over-supplying the zone is that the air handler will short cycle. It's a little ceiling-hung 300 cfm unit.

Your chiller PD pumps need only consider the loop from manifold via the pump, through the chillers and back to the manifold.

My original question was how to evaluate the PD of the Tee branch where the chiller loop joins the manifold. What I was missing (and figured out when I read the Idronics TES issue) is that I can ignore the impact of the other loop on that Tee by making the Tee large and locating it at the TES.

In case you're interested, I'm attaching an elevation drawing of the tank juxtaposed to the house. This was just to get the piping elevations correct. The chillers are behind the courtyard wall and the source trunks will parallel those shown. I still need to add side-view sketch of the 4-way valve.

I appreciate your input!

 

[LittleInch]

Ok, I think you now have it clear, the further info would have been good to know from the start....

As an aside, I assume / hope you're going to insulate the tank walls / floor / roof otherwise your water won't stay cold for very long, but maybe that's just not shown for clarity.

I've just been insulating some hot tanks and the impact of insulation is significant.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[DavidButler]

The tank lid will have 4" XPS and high reflectance/low emissivity coating on top. Upper 4 feet of tank body (corrugated culvert) will have three layers of flexible EPS wrap (~R-18), middle 4 feet will have two layers and the bottom portion one layer, with 1" XPS under the slab. I'm not sure yet how the owner plans to cover/protect the above-grade portion. We looked at having an EPS shell made to fit with similar R-values but cost was 3x. We also looked at fully burying the tank but that would have put pumps above the water level - not good in an open system.