Results 1 to 13 of 13
12-02-2012, 11:22 PM #1
A few questions on evaporator size, liquid capacity, liquid slugging prevention, etc.
I've got a couple questions on system design and evaporator sizing for a liquid chiller project for engineering school.
Is it true that in designing a refrigeration system, especially when the evaporator stays at the process temperature when the compressor is not running, the evaporator should have enough volume to hold the entire refrigerant charge in liquid form, since that is where it will condense to when the system equalizes? The system will be a R-134a cap tube system and I plan on using a small 10 plate heat exchanger as the evaporator, which has approximately 5 oz liquid refrigerant capacity by my calculations. What happens if the required charge ends up being more than that 5 oz capacity?
Also, how does this all affect possible liquid slugging on compressor start up? I will be using hermetic reciprocating refrigerator compressors. Doesn't the suction return go straight into the compressor shell space, which means liquid refrigerant is highly unlikely to feed directly into the pump intake, making this a relative non issue so long as some reasonable liquid trap U bend is made in the suction tubing?
Another consideration I am up against is the choice of liquid line driers. I know the copper bullet driers are more or less useless for anything other than very clean, dry systems, but they also hold less liquid volume. I'd rather specify a sporlan catch-all, but would that also increase the required refrigerant charge?
Any insight is much appreciated!
12-02-2012, 11:57 PM #2
Firstly, I am truly jealous, of the skill you have in CAD drawing. It looks just great.
Looking at your pic, why are feeding the liquid into the top of the PHE as this increases the chance of liquid slugging (on start up) and a wet suction when running. (oil dilution problems)
12-03-2012, 12:47 AM #3
12-03-2012, 01:07 AM #4
12-03-2012, 02:06 AM #5
It's probably going to see around -12 to -15 C evap temps most of the time, but I haven't yet actually done the calculations to relate coolant entering/leaving temps with evap temps. It gets complicated because the coolant flow rates will change, the heat exchangers must be piped in parallel, and the ice rink control system will demand a certain supply temperature to go to the ice rink floor. Initial estimations were that the HX were going to be oversize, but 10 plates are the smallest available. Correlating with some experimental data found in a scientific journal article on the use of plate HX as evaporators, the estimated overall heat transfer coefficient (U value) is around 3500 w/m^2 K for the smaller compressors and 3900 W/m^2 K for the larger one, with a heat transfer area of 0.12 m^2. This worked out to yield a log mean deltaT between coolant and refrigerant of 0.5-1 deg C across the range of expected evaporating temps/compressor performance for the small compressors, and 1-1.75 deg C for the large one. The U values are a little low since the cross sectional flow area is relatively large and drops the refrigerant mass flux a bit, but that seemed to bring the evaporating heat transfer coefficient into the 5000 W/m^2 K range versus the 7000 W/m^2 K range, but that difference doesn't affect the U value that much, as the coolant flow rate range still maintains turbulent flow on the liquid side, which is estimated to yield around 15000 W/m^2 K according to the experimental results. To make a long story short, according to the numbers, the coolant flow restriction is no problem, the approach temps will stay nice and low, as intended, but the lower refrigerant flow velocity could possibly cause oil return issues. I am a bit constrained in the available dimensions of small size plate heat exchangers that also fit within the project budget, but a longer, narrower HX would probably be more ideal. The alternative is a custom tube/liquid heat exchanger similar to the condenser design, but then all I have is regular smooth bore tubing, which would potentially limit evaporation heat transfer coefficient performance and also be more difficult to design to maintain a high coolant side convection coefficient with a varying flow rate.
12-03-2012, 02:33 AM #6
Going back a step.
What are you trying to do and what are you trying to prove.
Are you just trying to make a sheet of ice, (mini ice rink).
12-03-2012, 02:58 AM #7
The project as a whole is developing a zoned ice rink floor to be able to adjust coolant flow rates through each zone in proportion with the average heat load experienced by that zone, as heat load on an ice rink surface is rarely uniform, causing ice surface temperature variations given an industry standard static single circuit cooling floor, which causes inconsistencies in performance for more advanced skaters as well as causing problems in non-uniform ice thickness maintenance, not to mention wasted energy cooling some areas too much in the effort to compensate for the worst case surface conditions. The main temperature control will use a special neural network type controller and will read zone heat loading states using surface temperature via non contact IR sensors and coolant liquid return temp sensors. Since the whole network is being supplied from a single chiller plant, the control system will utilize a genetic algorithm to optimize the selection of zone flow control valve positions to maximize required coolant supply temperature based on the most heated zone in order to maximize chiller plant efficiency. The coolant pump will be speed controlled using a proportional closed loop control reading discharge pressure in order to keep the pressure drops across the zones constant and not induce more complex flow dynamics induced control oscillations. The main controller will dynamically vary the supply temperature setpoint, which will be forwarded to the chiller plant control, which will control the stage sequencing of the 3 compressors using a PI or PID control structure in order to meet the demand.
12-03-2012, 03:15 AM #8
All good, for it to work you first will have to undertake CFD of energy infiltration to establish both loop configuration and IR positioning and some method of calibration in changes in reflective index of the ice during use.
But this side I will leave to you.
Down to the refrigeration, you may be better at looking at using as single circuit (and compressor), using VRV (variable refrigerant volume), either by a VSD or a digital control (a scroll for example). This would reduce many of your other issues. And gut feeling is that it would be no more expansive than your present design. The refrigerant mass flow can then be controlled directly by your process variables.
12-03-2012, 03:58 AM #9
Yeah, our project faculty adviser decided to have us make the zones into equal rectangular divisions to focus on proof of concept of the control method, as the simulation and modeling of actual ice rink heat loading and zoning design would be almost an entire project in itself.
I considered a variable capacity compressor, and that would be the best method for maximizing efficiency, but our biggest constraint besides time is cost of new cutting edge equipment versus what we have available. Since I already had two compressors that were free scrap, that made the current setup already more favorable. Single phase 120V variable speed drives are hard to come by and are pretty expensive. We were looking into a simple custom designed VFD, but that got pretty complicated unless we wanted to stay with a modified sine wave inverter architecture, which may have problems starting the compressor, and would still get pretty complicated since we'd still be trying to digitally control both duty cycle and frequency. I had a variable frequency PWM sine wave signal generator working pretty well on an arduino microcontroller board, but the timing signal synchronization to toggle the two sides of the MOSFET bridge started looking like a nightmare and all I have are mechanical engineers (I would consider myself quite a bit above average in the circuits and programmable logic control department as far as ME's go, but certainly no EE, at least not a good one). We decided to abandon it and move on with more straightforward obstacles as we were (are!) running out of time and the primary focus of the project is the neural network temp control algorithm. I'm the one with the EPA card, so I get to be the one to figure out the source of cold glycol to make it work.
12-03-2012, 04:42 AM #10
Large scale, variable refrigerant flow is not an issue, and is done every day.
You really are not going to be able to simulate the refrigeration with what you have.
For this i would still simplify the refrig system and use a thermal buffer. Have a controlled loop of which you draw from (its own dedicated pump and 3 port valve), and if necessary have a simple heater as false load.
Fixed flow to evaps.
Monitor the 3 port valve position or the 4-20mA drive, you could use this to show simulate savings made on an industrial refrig plant.
You will not be sure how well the old comps are working, bite bullet buy a single compressor, one evap and one cond.
For example run your buffer between -23C on and -26C off, or if you want a little bit more stabilty add a PCM into the buffer say -22.
You should be able to control the loop quite with high degree of accuracy
12-03-2012, 02:04 PM #11
12-03-2012, 02:34 PM #12
You have mis-understood,
You have a fixed flow through evap and the control loop (3 port valve) care of the thermal buffer (no requirement for fancy refrigerant flow control),
the buffer is sized for 6 mins of load (to ensure number of comp starts per hour)
You draw a predetermined temp from the control loop, to your rink control. At this point you can glycol flow in which ever way you require.
A single compressor, a single PHE and Cond, will be cheaper than 3. Be careful not to be penny wise, pound foolish.
Remember that you do require some level of superheat, so you do need to increase your TD a bit, or introduce external of the evap.
12-03-2012, 09:23 PM #13
A simple way to deal with it, not considering valve response profiles and all that is to add restriction to the bypass loop to approximately equal that of the rink floor, so the head pressure on the pump is constant regardless of valve position...except that the restriction presented by the dynamic rink floor is always changing. The rink flow demand will not swing wildly, though, as there can only be so much difference in heat load between one zone and another in the real world. There will always be one zone control valve running wide open and the rest proportioned down in relation to the heat load imbalance, so flow rate will always be maintained fairly high. It has to to keep the glycol delta T small enough so the ice temperature does not vary too much across the zone.
The second option would be to speed control the pump using flow rate so as to maintain a constant flow rate, and then just deal with the coupled hydraulic dynamics within the rink floor, since the pressure drops across the zones will depend on each other's state of flow control and the state of the 3 way bypass valve...I have no idea in hell where to start with that one as far as whether the control system will be able to deal with that effectively.
The third option, which I tried to explain last post but didn't get it right, would be to have a closed loop head pressure based control on 3 way valve and flow based speed control on the pump. That way the 3 way valve will change position based on how much flow is demanded by each of the zones in order to maintain a constant pressure feed on the zone coolant distribution header, eliminating or drastically reducing coupled flow dynamics, and have flow based speed control on the pump to keep the flow rate through the evaporator constant with the changing flow restrictions presented to it. That would be possible and might be worth considering. Flow sensors are nightmarishly expensive for a student project, though, unless we concoct some sort of venturi and pressure sensor, but I do have another idea that could use one of those cheap turbine flow meters used by the gaming computer overclockers for their water cooling system monitoring.
One big compressor will probably be cheaper than 3 smaller ones if they are all bought new. The thing is, two I already have that were free, and the third I got off ebay for a really good deal. One advantage to going with one large compressor is the ability to dump cap tube in favor of a TXV. Pricing out 1 hp low temp R-134a compressors is a sticker shock experience compared to what I have so far, unless I get really lucky on ebay or something. They're not nearly as common as domestic fridge size compressors, and I haven't seen one for less than ~$600. At this capacity, it might make sense to go with R-404a, as compressors are a little more common (read "there's more likely to be deals on ebay") and possibly more efficient, but then there's the issue of acquiring gas, as I'm not too excited to go out and pay up for a whole cylinder of 404 for one project, but I guess I could sell off the rest if I really wanted to (or wait until it's phasing out and worth its weight in gold...then sell it). I'm also not too excited about dealing with a blend with the temperature glide and fractionating if there's a leak and all that fun stuff. Plus, if one large compressor is the better way to go, why is it that every ice rink chiller plant I've seen runs multiple smaller compressors? But yes, given enough glycol in the loop, a 1200 W compressor could be made to run a lower limit duty cycle of ~25% without short cycling. It would also be much easier to analyze and also much easier to prove the energy savings difference between dynamic flow control and fixed control. The chiller control system would be simpler, too...I already have a single stage PLC I built over the summer complete with a status display and current sensing. Just needs some I and/or D control added to it...oh, and I hate cap tubing... I might have changed my mind... Only problem is will I be able to start a ~50 LRA compressor on a standard 15-20A 120V circuit? We have to exhibit the project at the science museum in the spring, and will be constrained to common power receptacles.
Not sure if I misunderstand, but the delta T numbers quoted above are the delta T between the evaporating temp and the coolant log mean temp across the HX, where lower would mean a more efficient HX, as the evap temps can run closer to the desired process temp. The glycol delta T from inlet to outlet can vary anywhere from 1 to 6 deg C depending on the pump operating conditions and the number of stages running.