trade-offs with low velcocity (laminar flows)

[DavidButler]

Without going into the "whys"... I'm designing a chilled water A/C system that must operate over an unusually wide range of entering water temperature (40F-63F) and flow rates. The flows will be laminar the majority of time, with Reynolds numbers in the three fan coil circuits as low as 1000. Given the project design constraints, which I won't bore you with, I think it makes more sense to take a hit on heat exchange efficiency than the alternative.

Aside from a loss in coil heat transfer efficiency, I want to make sure I understand any downsides to operating at loop velocities as low as 0.5 fps in the air handler branches and less than 0.2 fps in the trunk. The loop is non-pressurized, and will include 5,000 gallons of highly purified (RO) water (essentially a large thermal battery), so sediment fall-out isn't an issue. Using smaller plumbing to increase velocity would cause head losses to spike during high flow conditions (e.g., when thermal battery is nearly depleted), and would do nothing to improve heat transfer efficiency in the fan coils.

I understand very well the trade-offs if I re-design to maintain turbulent flows at all times. But I'm not sure I understand the trade-offs of having such low flow rates, other than the obvious impact on coil performance. It's not something I've had to worry about before. Advice would be welcome.

 

[georgeverghese]

We've got heat input from the 3 fan coils into this chilled water; the thermal performance reduction from low flow velocity in these coils is acceptable from your estimates?? However, these velocities you've quoted are in the pipes and not within the fan coils itself, so how do you know there is a heat transfer debit in the coils?

The same may be said of the heat rejection side of this chilled water circuit - presume you have a HX of some some sort here also.

As long as total circulating flow in m3/hr remains the same as the design case for these HXs', the heat transfer duty wouldnt have changed. The same goes for dp in these HXs'.

Friction drop in the pipes would be much less as you say.

 

[LittleInch]

Downsides,

Mainly extra cost and space. If you don't have sediment then that isn't an issue.

Normal flow won't blow any air out unless you can achieve >3ft/sec every now and then, so good pipe runs with slopes and lots of air vents may be required to vent all the air out on a regular basis.

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

 

[DavidButler]

I know the # of circuits within each fan coil and the ID of the circuit tubing, so it's simple math to calculate velocities and Reynolds numbers for the coil circuits. In my original post, "coil circuit" refers to the tubing inside the fan coil, and branch circuit refers to the tubing that connects air handlers to main trunk and pump. As it turns out, the Reynolds numbers for the coil circuits are significantly lower than for the branch circuits, even though velocities are similar (due to small tubing size). However, all of the flows (circuits and branches) are laminar or transitional for most of the annual operating hours.

The only time fan coil HX efficiency and load loop head loss become factors (and it's a whopper) is when EWT approaches 60F, when thermal storage is nearly depleted. In that case, flows must be much higher (5:1 for one of the air handlers), and therefore the flows are turbulent all around. In fact, I'm relying on the excellent heat exchange efficiency of these fan coils to get the last BTU out of thermal storage before it cries uncle. If you haven't already figured it out, this is an off-grid project -- a new residence on 80 acres near Phoenix.

As for the heat rejection side, the chillers, which have internal HX and dedicated pumps, are hydraulically separated from the load loop by 5k of thermal storage. I designed the chiller-tank loop to ensure acceptable flow across the HX, based on the manufacturer's specs and chiller pump curve. Otherwise, a series booster pump would have been required.

Due to the unpressurized storage tank, there's going to be lots of entrained O[sup]2[/sup] in the water. No way around that. So all the components are designed for that environment. My biggest design headache was to make sure the pumps don't ever loose prime (out of necessity, they'll be mounted a couple of feet above the tank water level). The controller will have a purge cycle that speeds up the pump to dislodge any bubbles that accumulate in the fan coils. They'll be an air vent in the return trunk, situated at the highest point in the loop. I don't believe cavitation will be an issue since water will ever exceed 115F (in winter). However, I'm in uncharted territory, which is why I pose the question about low velocities. I don't want something I don't understand to bite me in the butt.

I have to say that designing the building envelope and mechanicals for off-grid up-ends many of the design guidelines we take for granted.

 

[EdStainless]

What are you using for a biocide? Will some of the water be circulated through ion exchange to keep it clean? If the system is open and you don't continuously clean it you will have a tank of algae and mud in a few months.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[DavidButler]

@Ed, treatment details are not nailed down yet, but the idea is to start with clean tank & plumbing, and fill with industrial RO water. The tank will be closed to the environment with filtered vent, so evaporation and new contamination will be greatly reduced. One of my team members has designed and implemented two unpressurized thermal storage systems in Fairbanks so we're building on his experience. We do plan to treat the water with inhibitors (Rhomar or Fernox), the specifications of which are TBD. We probably won't install the tank until early next Spring. Please say more about ion exchange.

 

[georgeverghese]

Have yet to come across designs for heat exchangers with flow in the laminar region - the heat transfer coeff calcs for Nre <2000 or so are more complex and less exact than those for Nre >4000 or so.

 

[DavidButler]

@George, great point. I should have mentioned that an applications engineer with the fan coil manufacturer, MagicAire, confirmed that their New Magic 4 modeling tool internally calculates the Reynolds number and uses laminar heat transfer coefficients as appropriate. New Magic 4 is the best coil modeling tool I've come across, by far. I'm curious to know how it handles transitional flows.

 

[georgeverghese]

Guess there is always a "first time" for everything, but it may be reassuring to run a manual calc verification of this "MagicAire" simulator results for tubeside htc for 1 or 2 case runs in this laminar / transition flow region just to be sure it lives up to its brand name.

Still, it seems a poor use of the capital expense to be running these coils and HX in this flow region from a heat transfer perspective. If you plotted out a curve of the added expenses for OPEX and CAPEX over a reasonable payback period, you may find the curve indicating that we should move out to a higher circulation rate? Or alternatively, for a fixed chilling duty, that you could have done this with coils and HX with smaller heat transfer area ( with higher RO water recirc rate)?

 

[EdStainless]

I have worked on water systems that we used to cool high voltage electronics where the water was the insulator of last resort. we slip streamed about 10% of the flow through a ultra-filter and a mixed resin ion exchange bed to polish it. This was to take care of metals that were dissolving in the system, and other soluble trash. We protected the resin with the filter (today there are filters out there that would remove dissolved metals and salts).
If you keep the conductivity of the water low enough then you will not have galvanic corrosion issues. And it is hard for anything to grow with no nutrients. On a large system we actually injected hydrogen peroxide just where the flow went into the piping loop. It has poor persistence but does not add any detrimental chemicals to the system.
I have heard of system with UV lamps in the discharge line to kill in place of a biocide.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[DavidButler]

@George, to understand how I got to this design, a bit more explanation is necessary so bear with me (or ignore if you're not really interested)...

Battery capacity is expensive. Even more so than PV modules and inverters. Based on 20-year LCC, it costs less to add PV capacity and have two chillers operating only during prime daylight hours than to have one chiller supported by batteries. Although each chiller only draws 8 amps peak (240VAC-1P), that's by far the largest load in the house. Did I say being off-grid is expensive? I should also point out that in the Phoenix area, cooling ops are by far the long stick in the design, even given the significantly lower solar insolation in winter. Since the specified chiller is reverse-cycle, ample heating capacity is a no-cost byproduct of the cooling system. The only solar thermal in this project is a stand-alone solar water heater for DHW.

Needless to say, thermal storage is a bargain compared to electrical storage. The large thermal battery (5k gallons of chilled water) is designed to keep the house cool at night and during periods of low solar insolation (rainy season is mid-summer). The objective for sizing storage capacity (batteries & thermal storage) was to be able to sustain 3 days of negligible insolation in mid-summer. Once the last kWh and BTU are squeezed from storage, there's no choice but to fire up the generator.

Thermal storage capacity of a tank is largely a function of how much EWT swing can be tolerated. Minimum EWT is limited (by chiller design) to about 42F. Maximum acceptable EWT is largely constrained by the air handler and indoor design temp, which is allowed to slip to 80F when operating on storage for an extended period. To this end, I specified much larger hydronic air handlers than necessary to meet the design loads (e.g, a nominal 5 ton AHU for the Main Zone, against a design load for that zone of just over 2 tons) in order to raise the highest acceptable EWT to about 63F. By oversizing the air handlers (blower + fan coil), I'm able to keep pump and blower energy under control, while meeting the design load with 63F water. I refer to that as constrained ops. More than a couple of degrees higher runs into a wall. As it turns out, the ECM blower motors and water pump collectively represent the largest single electrical load that has to be supported by batteries, so head and air-side static losses were the determining factors in arriving at the constrained operating point. If I were to use smaller fan coils, the maximum EWT would necessarily have been lower, and tank quickly gets a lot larger. At 5k gallons, it's already far larger than the owner originally envisioned and tank prices jump disproportionately past that.

Ideally this home should have had radiant floors to eliminate those blower loads as well as leverage the slab mass. Unfortunately, that wasn't an option by the time I got involved. Even radiant panels would have been better than forced air, but the owner wouldn't sign on to that (and I'm normally the guy arguing for forced air over radiant, but that's another conversation).

OK, now the flip side. When EWT is in the low 40's (by far the largest portion of operating hours), it's necessary to dramatically slow down the blower and H[sub]2[/sub]O flow rate to avoid severe short-cycling of the air handlers (what I refer to as "motel room syndrome"). For the main zone, the gpm turn-down ratio needs to be about 4.5 to 1, which is about as far as I could push it. During constrained ops, the flows are into the turbulent range . During nominal ops when water is cold, the NRe's are down around 1000, so flows are laminar.

But here's the thing that's not apparent unless you consider the entire system... making up for lost heat transfer efficiency during nominal ops has zero impact on the overall design, since the pump is at the minimum operating point. I can't tell from the modeling s/w how much of a hit laminar flow imposes, but let's say it's 30%. That just means I need 0.7 gpm instead of 0.5 gpm. The challenge was finding the sweet spot (EWT) at the high end. It was a balancing act between storage capacity, AHU selection, pump & blower pressure drops during constrained ops (and associated energy penalties), battery capacity, and concessions made by the client on indoor temperature during constrained ops.

As I mentioned in an earlier comment, much of how we think about mechanical design gets turned on its head when you go off-grid.

 

[DavidButler]

@Ed, thanks for the tips on water processing. One of my team members has implemented large non-pressurized thermal storage on a couple of projects in Fairbanks. He used trucked water (by necessity) but it wasn't particularly clean. His loop water is now turning gray after one heating season. My goal is to clean the tank and loop as much as possible and use RO water with minimum treatment. The water will have very low conductivity (but TDS will not be zero, you don't want that). Once we gain experience with what turns up in the water, we can add targeted filtration as necessary. I saved your comment for future reference.

 

[hydrae]

One thought

Since you have the pumps on drives, could you cycle the pumps instead of low flowing them during the low demand period, thus using the thermal storage inside the fan coil unit?
eg instead of running the pump at 10% all the time, run at 20% half the time. The flow is turbulent during operation. Keep the fans running to avoid the hotel room problem

Hydrae

 

[DavidButler]

@hydrae, by 'thermal storage inside the fan coil unit', I assume you mean the mass of the fan coil and coil circuit volume? In any case, the dynamic viscosity of 42 degree water is a lot higher than at 63F, a big reason why Reynolds numbers are so low during nominal ops. When water is cold, the pump would have to push 85% to 130% of required flows into each zone to raise coil circuit to 4000 NRe (% varies depending on zone).

I'm not inclined to redesign the system just to maintain laminar flows unless there's an insurmountable rationale for doing so. The compromises imposed on the design would be pretty serious. It's just that neither of the other two engineers on my team have dealt with such low flow rates before so there's understandably some push-back. Fortunately, none of the comments posted (thus far) suggest there's cause for concern.

 

[georgeverghese]

It was after my last post that I realised that this may be perhaps what one would do on a off grid A/C system. Maybe one way to reduce power consumption further is to use a pressurised accumulator - fill this up during times when the pump runs at higher hydraulic eff and then let it deplete into a lower pressure vessel or tank via the fancoils ( with Nre = 1000 as you have now) during low demand (i.e with the recirc RO water pump off).

Another suggestion would be to pack in more fins and increase extended surface area on these fan coils, provided it doesnt add much on the air blower dp.

 

[DavidButler]

With the current design, the pump will gradually ramp up during an extended period of reserve ops as storage gradually "loses its cool". The controller will make this happen based on a sliding delta-T. Sorta like a delta-T reset curve based on EWT. So the pump only operates at relatively high hydraulic efficiency when weather events lead to several days of reserve ops, and only near the end.

If I'm understanding your suggestion, we would want to charge the accummulator during nominal ops to some cut-out threshold, then let the accumulator feed the air handlers at the required low flow rates. The cut-in point for the pump would be when the calling zone(s) can no longer meet their respective load demands.

Interesting idea, but it looks like a solution in search of a problem. We would still need the 5k gallon thermal storage tank for reserve ops, so what does extra complexity and cost this buy me? Or said differently, what problem does this solve that I may not be aware of? Keep in mind that pump efficiency when the sun is shining and water is cold is totally irrelevant.

I haven't said much about the chiller side, but one thing I had to get my mind around is the need to operate the chillers at their least efficient operating point... when return water is 42F, the EER is much lower than when the return water is 65F. But again, during nominal ops when sun is plentiful, efficiency is unimportant. Reserve capacity (and recovery thereof) totally drives the design.

As for the fan coils, they're already at the maximum (rows, circuits and fin spacing) offered by this manufacturer. Good point.

 

[georgeverghese]

On the accumulator suggestion, you wouldnt need to run the recirc pump at this low flow when the accumulator is on rundown mode? The 5000gallon storage would then be part of this accumulator. That would also have the added benefit of increased reliability since you wouldnt need the recirc pump and a spare to keep the RO water circulating all the time.

 

[hydrae]

What is the 5k storage tank? is is something fancy or just a 8 ft diameter 12 ft tall (or 12 ft dia x 8 ft tall) plastic water tank set in the basement? yes tank prices jump at the 5k point but that does not mean you can not use 2 tanks. The shipping of the larger tanks also become a major cost issue because you get into oversize loads, but having 2 tanks will add to your thermal storage, and also allow operational flexibility. If foot print is the concern on storage volume then other sizes may be more effective.

Hydrae

 

[DavidButler]

@hydrae... owner is investigating tanks, but there's no basement and no space to spare, so the tank will be buried outside, near the house. I recommended a vertical tank to minimize surface area and avoid the potential for poor lateral mixing with a low-boy or horizontal cylinder. The Fairbanks engineer on my team built a large storage tank out of wood and pond liner.

Two tanks would be twice as expensive, so not very incremental. But I haven't heard a compelling reason to add storage just to reduce operating temperature range. For example, if I were to limit the range to 11 degrees -- 52F min/63F max, the flow rates would still be laminar at 52F with these air handlers, which I oversized to satisfy load when water is 63F. Likewise, if I limit range to 42F min/53F max, I could have used smaller air handlers (too late for that), but flows would still be laminar at the lower end of the range, which as I said, is the majority of the time.

As engineers, these discussions are interesting but I'm still unsure why I should consider changing the design. What problem would raising the flow rates solve?