My first AutoCascade -- building has begun

[mytekcontrols]

I am not sure what you mean by "loader"?

Kevin

My guess would be that he wants to see what you are using to present a heat load to the system. But that's just a guess.


Do you think you might try the single-stage AutoC since it looks better under simulation?

[n00b 0f l337]

The 1 separator simulation seems good. The numbers are interesting, especially if we assume they'd be close to the real thing. The discharge temperature however, seems a bit too high being in the 90C. What is the electric motor heat in the suction line for? Is that the compressors motor being accounted for?

[n00b 0f l337]

What version of Aspen HYSYS are you using?

I got Version 3.2 and am playing around with it now but bit busy with school so will have to wait for next week I think.

[mytekcontrols]

The 1 separator simulation seems good. The numbers are interesting, especially if we assume they'd be close to the real thing. The discharge temperature however, seems a bit too high being in the 90C. What is the electric motor heat in the suction line for? Is that the compressors motor being accounted for?

Adam
--
90C is quite normal for a Scroll Compressor, and well within it's comfort zone. In my business, AutoC's with Scrolls quite commonly run with 90-100C discharge temperature.

[n00b 0f l337]

Ah fair enough. I've got one sitting but have yet to get to use one. Lookin forward to it

[Kevin Hotton]

Let me see if I can answer a few of the questions presented.

1) the "loader" (if it is what Michael thinks) consists of the refrigerant cooled evaporator. The load is a pumped methanol flow. To replace the heat being taken out, we have a vertical tank (2-1/8 inch OD copper 45-inch long with a 1500W resistive heater wrapped around and insulated. We can calculate heat load by measuring methanol flow rate with a paddle-wheel flow meter and using the temperature change in the methanol.

2) Yes I plan to reconfigure our AutoCascade to the simpler single phase-separator with auxilary condenser design.

3) The Hysys version I am using is 7.2 released in 2010 (I think).

4) Yes I add heat for the electric motor (before the compressor when it is a scroll and after the compressor if it is a rotary compressor).

5) The discharge temperature is maybe a bit high because the scroll model I have is not the most efficienct (only about 65%). It is the smallest 3-phase model and I suspect the scroll height is very short so relative tip losses are higher. I also input estimates of its volumetric efficiency (which change with pressure ratio).

I have refined the single phase-separator design and think n-butane and ethylene is a good two component blend. I only used the iso-pentane, propene, ethylene combination because those are the one I have on hand. I am ordering the n-butane and will be going with that. Of interest, I also tried iso-butane, ethylene but it is not a good fit (need the still lower vapor pressure of n-butane -- concidentially that is what the blend of iso-pentane and propene was approximating (the vapor pressure of n-butane).

Kevin

[ultralo1]

Buffer tank with pressure controlled inlet valve is pretty much standard in reducing and maintaining discharge pressures. It will also help with pull down and temp recovery should the final evap come under varying loads.

It is pretty much standard operation to over build the refrigeration system and control your final product temp with electric PID controlled heaters. You can maintain 0.1-0.5 temp variances with this style of control. Its alot easier to add a little energy to a system than take it away.

Best fabricating practices are a must for these. Simple turbulence created by burrs left from cutting the tubing can effect your flows enough to effect your final temp.

I cant add much more than that to these discussions. AutoCs are fascinating but they can make you pull your hair out trying to tweak them into working the way you think they should. They all have personalities, no two are alike.

[Kevin Hotton]

I can relate to your comment, "they all have personalities". I do know that without the Aspen Hysys software to model the cycle, that designing and tuning would have been very difficult. All and all I have been happy with our results. The main disapointment has been the sub-par compressor performance (the unit I have is not efficient). My computer simulation using fairly conservative compressor performance levels, predicts significantly less electrical power consumption (and discharge temperature). Today we swapped out this scroll compressor for a slightly larger model (14 m3/hr vs 12 m3/hr). I expect its performance to be better...we will start testing tomorrow.

Kevin

[Kevin Hotton]

Today we ran with the larger (14 m3/hr displacement) scroll compressor. We removed the VFD (varible frequency drive) and ran with simple "ON" relay / contactor. Amps were 12-13 (208VAC 3-phase). This compressor's spec's claim better motor and compressor efficiencies, but power-factor is low (only 0.70 so only really 9.1 net Amps = 13 * 0.70 and 208*9.1*sqrt3 = 3278W consumed).

I tweaked the cycle slightly (used 55% iso-butane 45% ethylene (by mass)) and shortened the captubes (all 0.064"ID; 1st 6-ft, 2nd 7-ft, 3rd 5.5-ft). As expected duty improved. We could use the full 1440W heater (with the added methanol pump heat and heat loss to the room) and keep the 2.2 Liter-per-minute Methanol flow chilled to below -60C. Calculated duty based on Hysys model was 1600W (so I think we are close). Based on above power consumption, COP = 1600 / 3278 = 0.488 I need to get this efficiency higher for this Autocascade cycle to be competitive to our larger 2-stage cascade (its COP = 0.729)

The only major factor limiting further increases in duty and improvements in cycle efficiency, are the two cascade-condenser plate HX's (both 20-plate). Their calculated overall heat-transfer coefficients are about 180 W/m2-C for CC-1, and 120 W/m2-C for CC-2. These low values force greater delta temperatures across the HX's (over 20C) and this hurts efficiency. I will be trying a larger CC-2 HX (the 40-plate model I first used as an evaporator) to see if I can further improve performance. I plan on using a "comb" distribution tube to deliver the captube exit flow from the second phase-separator across the full depth of the HX. These cycle changes are modeled to improve duty by about 200W with no increase in compressor power (I will need to lengthen the second captube) and change blend to 56% i-butane, 44% ethylene.

Kevin

[ultralo1]

Kevin, It sounds like you are making good progress. I am having a little trouble following your descriptions though.

These low values force greater delta temperatures across the HX's (over 20C) and this hurts efficiency.

This line in particluar is giving me trouble. As a technician I usually get to see the following temps on a system.
1: 1st stage evap in and out
2: 2nd stage evap in and out
And so on for as many evaps as you have.

If I am guessing correctly, you are seeing a Delta T of 20c across the evap in and out on each HX. If so, it sounds like you do not have enough refrigerant flowing in the evaps of the HX. I usually see alot less than 20C difference between the in and out. What do your pressures look like at temp?

[Kevin Hotton]

Hi Ultralo1,

What I meant by the delta temperature of 20C, is referring to the two flows in a counter-flowing heat-exchanger. One stream is the warmer one and it is being cooled by the colder stream. The better the heat-exchanger design, the difference between the steam temperatures will be less. For example, if warm stream is cooled from 50C to 20C with a cold stream warming from 10C to 40C, than the delta temperature I was talking about is 10C. The closer the approach in temperature between the two streams, the better the thermodynamic efficiency of the refrigeration cycle.

Kevin

[Patrickclouds]

maybe a pipe in pipe coax hx will result in better temps? the refrigerant stream passes the hx very fast.

i've usually seen mixed gas units running a coax hx. yours was the first with a plate hx. i think mytek is the man with a lot experience to answer the question about the right hx.

[ultralo1]

Thanks for the clarification Kevin. If you measure the evap Inlet and the evap out temp you will see if the evap has enough refrigerant or if its starving. Usually you will only see a Delta T of 5c-8c between those two if the evap is not starved. It would be easier to tweek the charge than to change HX.

I dont think that you would see much Glide since you are only using 2 refrigerants.

[Kevin Hotton]

We have been continuing our tuning of the autocascade with the goal of higher duty and efficiency. Today I used a blend of n-butane, propene, and ethylene (in 40% 20% 40% mass fractions). Our system is reacting completely differently as if the captubes are too short and too much refrigerant is being recycled by-passing the final evaporator and hurting duty. I lengthened the 1st captube to 8-ft and the 2nd to 10-ft, but even with this change my simulation suggests the too little refrigerant is making it to the evaporator. I can no longer understand how earlier we seemed to be getting over 1400W duty and now even 1200W is a challenge. I do know that we still have some leaks -- so that is altering the charge during a run. I am going to remove the suction accumulator because I believe it is pooling some butane and altering the mix in unpredictable ways during a run. Michael if you see this I have a question: " did PolyCold use suction accumulators in any of their units?

The plate HX's are proving very difficult to predict their performance. The run today indicates the large 40-Plate 2nd cascade condenser HX was only reaching an overall heat-transfer-coefficient of 51 W/m2-C (very low value). Earlier testing had indicated the 20-plate model when used as the second cascade condenser was delivering 115 W/m2-C. I am getting discouraged because my simulations indicate that a cooling duty as high as 1900W might be possible. I would be very happy if we reached a steady-state duty of +1500W.

Kevin

[n00b 0f l337]

The question then, is your simulation data incorrect, or are the plate heat exchangers failing to meet expectations.
I'd assume a cycle with manual tube in tube HX's in next, as that seems to be the standard.

[Kevin Hotton]

I think the simulation is accurate (at least to within 10%). So the mystery is mainly captube size/lengths. I do believe that the best HX design for autocascade is NOT plate HX's (they have a large flow cross-section and short flow length). I feel that tube-in-tube is better (which have small flow cross-section and long flow length). However to transfer 2550W and 2000W (simulation values for 1st and 2nd cascade condensers) would need difficult to fabricate tube-in-tube HX's. For example, I modeled Michael's "morph'd AC autocascade and the largest heat load at any HX was under 600W and this needed a 10-ft length of twin 3/16 tubes within a 1/2 tube. So for 2550 Watts, suggests I would need FOUR of these in parallel (not easy to fabricate).

Kevin

[n00b 0f l337]

Scale up.
Three 1/8 or 3/16 in 5/8, possibly two of those in parallel.
Or just purchase premade tube-in-tube with the ratings you need.

[mytekcontrols]

Here's what a 3600 watt at -110C capable 3-stage AutoC HX stack looks like, utilizing tube-in-tube construction...

Back
1100Stack_back.jpg

Front
1100Stack_front.jpg

Left
1100Stack_left.jpg

Right
1100Stack_right.jpg

Top (small coils are final sub-coolers)
1100Stack_top.jpg

To keep things reasonable, it is made up from two identical HX Stacks in Parallel. This way the tubing size doesn't get near impossible to bend in a compact fashion. A company called T&B did the actual machine fabrication of the HX's, Spinco made up the phase separators, and the rest was assembled by hand. This particular HX Stack design was based on earlier work by myself and a fellow Engineer Scott Forrest, while working at Polycold in the late 80's.

Here's my AC-2 HX Stack. Look familiar

[n00b 0f l337]

Well damn.
Trying to trace it all, the phase separators are interesting, thats for sure. Separate subcoolers?

[mytekcontrols]

Well damn.
Trying to trace it all, the phase separators are interesting, thats for sure. Separate subcoolers?

Remember these are parallel HX Stacks. So that includes the sub-coolers.

[n00b 0f l337]

Heh mytek, wasn't sure if they merged at the end, or merged pre-subcooler, etc. Just asking

[mytekcontrols]

Heh mytek, wasn't sure if they merged at the end, or merged pre-subcooler, etc. Just asking

No reason it couldn't be a single sub-cooler, but in this case everything was kept separate just in case we wanted to use just one of those stacks, or 3 of those stacks (very high load system). All the blending is done with the outlets of the phase separators being parallel connected prior to feeding the cap tubes. So in other words, the liquid outlets of all the #1 phase separators are teed together, then sent to a common bullet strainer, feeding all of the #1 cap tubes simultaneously.

[n00b 0f l337]

Yeah I saw that. Seems like one heck of a solid design.

[Kevin Hotton]

Hi XS's

We ran our autocascade the last two days with different charges. Yesterday was a 40% n-butane 20% propene 40% ethylene blend and today it was 55% iso-butane 45% (both mass fractions). The total charge has been increased from around 2.5-lbs to 3.5-lbs and this has helped. I believe the very large helical oil-separators I am using for phase-separators is requiring a large refrigerant inventory (if I did this over I would use smaller phase-separators). I also believe removing the suction accumulator was a positive step. Both days testing would hold a load of about 1700W (both a 1440W and 216W heater on full power plus all pump heat additions and heat loss to room). Compressor amps was just under 13.

I went back and reviewed data along with my captube calculations and convinced myself that I need to make all the captubes longer. So I did. I am now using 8-ft of 0.064" ID for the 1st captube, 11-ft of 0.064" for the 2nd, and 8-ft of 0.052" for the 3rd (at evaporator). Below is a screenshot near the end of our run. The 1766W duty shown is not constant (varies) but based on the total electric heat input, we must be holding in excess of 1600W.

The remaining problem is all about the HX's we are using. Notice that we have increased the methanol flow to over 3.5 LPM to heip heat-transfer. The evaporator is cold -73C but the exit methanol is only -54C (poor temperature approach). The key take away is that with autocascades, the design and sizing of all the HX's is critically important. With good HX's performance can be great, but with mediocre HX's performance suffers considerably. I truly believe that with optimal HX's I could have reach the peak duty near 1900W.

Kevin, Matt, and Joe

[Buckeye]

mytek should be banned for posting cooling stack pron like that

Man those sure look nice