My first AutoCascade -- building has begun

[Kevin Hotton]

Here are a couple pictures of the AutoCascade we are building. It uses all plate heat-exchangers. System consists of a water-cooled condenser, (no auxiliary condenser), two cascade condensers with their corresponding phase-separators and evaporator. The evaporator will cool circulating methanol from -30C to -60C. A heated water/propylene-glycol loop will reheat the methanol back to -30C. System also includes a suction accumulator and an oil-separator. An expansion tank may or may not be used. I calculate at 25C expected static pressure to be 169-psig. Cycle simulation predicts a duty of 1800W with the evaporator inlet at -73C and saturated vapor outlet at -43C. The compressor is a 12 m3/hr scroll (smallest Copeland 208VAC 3-phase model; 3.5-hp).

Kevin

[[XC] gomeler]

You have the best toys. Look forward to seeing how this comes together. What gases would necessitate 2 HXs? I'd think something simple like R600a/R170 would suffice?

[Kevin Hotton]

I'm going to use R600a and R1150. Also, there is not a hard rule that the number of phase-separators is equal to n-1 the number of refrigerants. The more phase-separating steps, the more pure the low-boiling-point refrigerant in the evaporator. My simulation model predicts about 95% pure ethylene in the evaporator. (See my post in Michael's cryoBUG thread for a graph of the evaporator temperatures).

Kevin

[mytekcontrols]

Wow

Looking good Kevin! And I have to agree with Chris, you guys do have the best toys.

I take it those blue things that look like oil seps are your phase separators. Was their originally intended manufactured purpose to be oil seps?

Edit: Yep when I looked at the picture again I see three of them. So unless I am mistaken they are oil seps, and one will truly be used as such, with the other two serving as phase separators.

[Kevin Hotton]

Yes they are Henry Technology helical oil separators used as phase-separators (and one for oil-seperation use). The units have drain ports that I can use to by-pass the internal float. I worried the density of my hydrocarbon refrigerant mix would not be dense enough to raise the internal float. The drain port solves this concern and should give steady outlet flow.

Kevin

[Kevin Hotton]

We ran the AutoCascade for the first time today. Didn't get to tune it at all and really were not at the design point but, it was encouraging. Attached is a screenshot of the LabView datalogging program taken about 30-minutes after start-up. We have some electrical noise due to the VFD's showing on our temperature readings -- I think. About 15-minutes later the evaporator temperature dropped below -70C. The main deviation from my design point is the compressor suction pressure -- it is about 20-psi too low. It is too soon to tell but, the captubes may be too long? I am expecting a cooling duty of 1.8kW but the screenshot shows only 0.6kW. We will run system again on Monday.

Kevin

[mytekcontrols]

...compressor suction pressure -- it is about 20-psi too low. It is too soon to tell but, the captubes may be too long? I am expecting a cooling duty of 1.8kW but the screenshot shows only 0.6kW.

For only running at 1/3 of full load capacity I would say that your running pressures of 17.3/212.4 are very reasonable. If you were expecting a suction pressure of 37 psig under these load conditions, that would seem a bit high to me, especially considering that this will come up as the heat load is increased.

As an example, here is the typical spec for a Polycold 660 series unit, note that the HC version is designed for a maximum heat load of 1500 Watts...

Polycold_660_specs.png

The suction pressure range is listed at 20-35 psig, but typically with a cooling water temperature of +22 C, the suction pressure realistically would be 25-28 psig.

Of course we are also talking a bit bigger on the compressor (6.5 HP) for a total CFM of 28. However this is a 6 cylinder reciprocating and you are using a scroll. But if your scroll is anything like an AC rotary compressor, then the efficiency should be considerably better than the reciprocating design. Factor this in with the additional stages being used on the 660 (3-stage autocascade w/final subcooler), and the argon for colder temps, and I'd say we should have a fairly close match between your system and the Polycold 660 running pressure wise. So I would think similar running pressures under similar loads.

But hey it looks to me like you are off to a terrific start. Congratulations

BTW I love the LabView screen

[sdumper]

Very impressive!!

[Kevin Hotton]

Hi Michael,

I can't take all the credit. My coworkers helped alot. Matt did the Labview control system and Joe did all the fabrication. I only did the cycle design. We did have the evaporator very warm before we started (maybe +60C) so that did not help. I am used to expansion valves that we can open up so cool down is quick (as with our large 2-stage cascade) so using captubes and an autocascade is all new. Seems cool-down is a slower process. Looking forward to Monday's system tuning -- I will keep you informed of our progress. Thanks so much for your input. Any suggestions you have are welcomed.

Kevin

[n00b 0f l337]

There are ways to make the cooldown quicker though.
Bypass valves and such for the upper stages.

[mytekcontrols]

There are ways to make the cooldown quicker though.
Bypass valves and such for the upper stages.

Yes AutoC's using cap tubes are a bit slow on the cool-down, but as Adam suggested, you can accelerate things by providing extra flow with a bypass. The simplest approach on your system would be to have a solenoid valve in series with a high flow cap tube, and then parallel this with your normal lower flow final cap tube going to the evap. This could be controlled either through a timer (solenoid on for first 15-20 minutes) or you could do it based on temperature. And obviously both functions could be programmed into yor LabView VI.

Another method of bypass would be to put the solenoid valve cap tube apparatus between the liquid outlet of phase separator #1 and the evaporator inlet. And perhaps yet another one from phase separator #2 to the evaporator inlet. Start up sequence would begin with both bypass valves on, then within a few minutes the first bypass is shut down, and then several more minutes later the #2 bypass is also shut down. This would most likely give you maximum turbo boost so to speak. Back in the 80's a fellow engineer and I tried this on a Polycold unit. It was amazing, in about 1/3 of the time the evaporator was down to cryogenic temperatures. In fact it worked so well that we got a patent on it (which has long since expired).

[Kevin Hotton]

Monday test results:

Over the weekend it turned out we had a slight leak. Of course this alters the refrigerant mixture so I had to pump it all out. We fixed the leak and recharged fresh. The problem we kept getting was high compressor discharge (overload tip). I believe how we avoided this on Friday was I had started with a light charge and we added charge twice while the system was running. Today I just started with the full charge. It seems to be too much load for the compressor to transition through. Towards the end of the day we took out about half the charge and it did respond with reduced load but temperatures were not getting much below -40C and suction pressure is always low < 15 psig. I have decided to shorten the final captube from 10-ft to 7-ft.

My original reason for a 10-ft length was liquid sub-cooling. The second cascade condenser heat-exchanger is somewhat oversized and the large delta temperature predicted between flows led me to expect significant sub-cooling of the liquid. When I calculated captube length using the method I posted, over 3-ft of captube length was needed just to drop pressure enough to allow the sub-cooled liquid to begin to boil. I just think I need to get the compressor suction pressure higher by allowing high-side flow an easier path to low-side. Michael as you suggested, we tried increasing evaporator load, but suction pressure did not go up – just discharge pressure. I hope the shorten captube helps. If not, I do have a bellows valve that I can install in series with the captube to add pressure drop.

One other change made had to do with a liquid injection feature of the scroll compressor we are using. If the discharge temperature gets high (as we were encountering due to low suction pressure / high pressure ratio) a valve opens and liquid refrigerant is injected directly into the scroll to cool it. In our system I made the mistake of taking this liquid from just in front of the final captube feeding the evaporator. So high discharge temperatures caused by the warm evaporator was made worse by the constant loss of liquid refrigerant being by-passed back to the scroll. I have now moved this liquid refrigerant injection line to the liquid line at the first phase-separator. This should allow the evaporator to chill down more easily.

Another problem we are battling is condenser discharge temperature. Because this system is meant to be a sub-scale version of a large unit (likely 300kW cooling duty !), I have to match cooling tower water temperatures of about 25C. However, tap water temperature is around 10C and the condenser is running too cold. This results in an incorrect vapor/liquid ratio in the first phase-separator.

Finally, it may be necessary to implement the solenoid by-pass circuits you mention. I am just trying to keep things simple (if possible) for now.

Kevin, Matt and Joe

[mytekcontrols]

A good way to deal with high pressure problems on start, is to implement an expansion/buffer tank similar to what I did on my AC2 project.

So in other words, you would want a fairly good size tank (8-10" diameter, 18-20" tall) and use something like a Sporlan B6 valve connected between it and the vapor out of your 2nd phase separator. Set the valve to open for 10 seconds during compressor start and buffer the high pressure gases into the tank. Then have a cap tube coming off the bottom of the tank going to the compressor suction (72" x 0.042 ID), this will let the system draw the refrigerant back in at a controlled rate.

You could also use a high pressure switch to control the buffer solenoid valve, or a combination of both methods.

Edit: I thought I'd add a picture to better illustrate what I'm talking about...
Buffer_Tank.png

When evacuating or charging your system you will want to tie into the expansion tank's access valve (use a 4 valve charging manifold). Be sure to use at least a 3/8" OD tubing from the #2 phase separator's vapor out -to- the buffer solenoid valve -to- the expansion tank, in order to minimize any delay due to pressure drop.

[Kevin Hotton]

Thank you Michael for the diagram and information. I expect to be implementing what you have posted. Today we ran our autocascade for several hours. It was a struggle to get the captubes to all flow liquid. I am wondering if having such large phase-separators is a negative (they are big -- approximately 2.3 Litre interior volume; 4-inch ID 12-inch tall). We have been adding more and more charge to try and get the captubes to flow liquid. Believe it or not, we finally reached a charge of nearly 60-oz wt. (over 1.7 kg) before the first two captubes flowed with very few bubbles in the sight glass. I must have a large liquid hold-up in both phase-separators (I don't think there is any other place for such a large volume of refrigerant to reside. The mixture ratio is 60% iso-butane, 40% ethylene by weight.

Until the captubes were flowing liquid, the suction pressure was very low (under 10 psig) and the compressor discharge was about 230 psig, so a big pressure ratio and corresponding high discharge temperatures (over 110C). Near the end of the day (with the big charge) the suction pressure finally climbed above 25 psig (my design point is 43 psig). I am not yet getting good cooling duty. I need to refine the method being used to maintain a steady methanol flow rate at an appropriate temperature. The current set-up is to crude and it fluctuates between no load (flow off) and too much load (flow on but with too much heating so methanol circulation warms and return too hot -- causing huge cooling load).

The attached screenshot shows system conditions maybe 2-minutes after shutting off methanol flow. The temperature traces with the obvious one dropping from about 0C to -46C is the evaporator outlet temperature. This temperature is important because it provides cooling at the phase-separators and consequently condenses the ethylene so that it can flow throught the final captube and provide cooling duty. I still am not getting full liquid flow in this final captube (sight glass shows considerable vapor content). The captube sizes being used are 10-ft lengths 0.064-inch ID for both phase-separator captubes and 7-ft of 0.052-inch ID for the final evaporator captube.

I must say that I have both optimistic and pessimistic feelings. I am getting cold temperatures (actually without any insulation besides some on the HX's) but as yet cooling duty is low. Many of the temperatures are within a few degress of the design simulation. However, it is frustrating to not know what "knob-to-turn" to effect system performance in the direction I would like it to go.

Any additional help from Micheal or others would be appreciated.

Thanks,
Kevin

[Kevin Hotton]

Final note: The liquid flow out of both phase-separators exits at the bottom of the vessels and then flows through a liquid filter drier and then "upward" through the sight-glass and then a few inches of 1/4-inch OD tubing before finally entering the captubes. I wonder if this is causing much of the problem with getting full liquid flow in the captubes? Should we have downward gravity assisted flow into the captubes? Is that vitally important? I wonder...Kevin

[mytekcontrols]

Final note: The liquid flow out of both phase-separators exits at the bottom of the vessels and then flows through a liquid filter drier and then "upward" through the sight-glass and then a few inches of 1/4-inch OD tubing before finally entering the captubes. I wonder if this is causing much of the problem with getting full liquid flow in the captubes? Should we have downward gravity assisted flow into the captubes? Is that vitally important? I wonder...Kevin

Definitely a Big YES to maintaining downward flow to all cap tubes. This is especially important in the colder stages where liquid quality begins to suffer.

I am also worried about the use of a liquid line filter dryer instead of a simple bullet style strainer prior to the cap tubes. It has been my experience that the material used within these dryers does not work well when getting very cold, and can begin to restrict flow as a result. There is also no benefit to use this in place of a normal strainer, which after all was there intended application.

Even after these problems are rectified, don't be surprised to see bubbles in the liquid feeding your coldest stage, especially when under higher loads. Achieving 100% liquid under all conditions is doubtful, and even more so as the number of stages is increased. In my experience working with and designing Polycold chillers, we often times had to increase the final cap tube flow rates higher than the ones in the preceding stages to make up for this poorer quality condensate.

I think if you address the problems I listed you will begin to see better more consistent results. Not to say there isn't a problem with what you have selected to be your phase separators, although this will have to be determined (I suggest cutting one open and see how much liquid it takes before it can exit).

[Kevin Hotton]

Today went much better than yesterday’s. I pumped out the charge into a large recovery cylinder (about 55 Liter volume). We opened the recovery cylinder and let pressure equalize before starting the compressor. This helped start-up and had another unintended consequence. After getting going I could tell that about 1.5 lbs less refrigerant had been drawn in. The majority of it must have been iso-butane because the base of the recovery cylinder was cold and the suction pressure was high enough to stop boil-off. I believe we had over-charged (and predominately iso-butane trying to get all bubbles out of the first two sight-glasses). This lesser charge with a higher fraction ethylene was better. Still it was difficult to get the system fully cooled down without overloading the compressor. We finally did however, and managed to get pretty good performance.

I am designing for a target duty of 1800W (if everything works perfectly). The load is circulating methanol which needs to be cooled from -30C to -60C. Of course at start-up the methanol is at room temperature and when you start flowing it, it creates a huge cooling load with quickly overload the system (so we have to cycle the flow on and off). But once the methanol gets cold, heat needs to be added to warm it back to the required -30C inlet condition. We have been using a water/glycol pumped circulation. This would work in principle, but not so good in practice. To avoid overloading the system, the warm water/glycol need to be cycled on/off and the flow kept low initially. Unfortunately, before we could get the system stable at the design-point we repeatedly froze up the heat-exchanger and had to stop. The included screenshot was shortly before this happened.

In the screenshot Matt added boxes below the measured temperatures (yellow box is actual temp. / grey box is dT above-or-below design-point temperature. You can see that overall the temperatures are too cold in the range of 3-to-15 degrees C, and that my suction pressure is still low at 23.3-psig. The methanol flow meter has been steady, but the water flow rate meter is erratic (electrical interference?) and so the condenser cooling duty of inaccurate. I do think the methanol cooling duty is accurate (1400W in this screenshot). So I am happy that I do seem to be getting significant cooling duty now.

To allow better start-up we are adding the by-pass circuit from the second phase-separator vapor outlet to the compressor inlet. Also I will be going to an electrically heated methanol system that will solve the freezing problem and give finer temperature control. All and all, we are pretty happy with the performance we are now seeing. I do need more sustained and steady data to make final adjustments to charge and captube lengths.

As for the filter/driers, our coldest is only -40C so maybe they will be ok? Also, the liquid out of the oil-separators-being-used-as-phase-separators does not depend on the internal float. There is an oil drain port we are using, so any liquid in the vessel should be able to drain out.

Kevin, Matt, and Joe

[mytekcontrols]

As for the filter/driers, our coldest is only -40C so maybe they will be ok?

Kevin, Matt, and Joe

They might be, but I've seen problems using them at even -30C (this was with a 1/4" Sporlan Liquid Line Dryer). So since you don't need them for the purpose that you are using them for, you might want to substitute conventional cap tube strainers when you get a chance.

The thing you mentioned about the isobutane being trapped in the tank, was the reason why I showed the use of a bottom port in my buffer/expansion tank diagram.

But hey, it looks like you are getting closer to your goal, congratulations

[Kevin Hotton]

Hi Michael,

Thanks again for your very helpful advice. I don't know if Joe will be in the mood, but I will try and get the filter/driers removed. Once we are able to run stable under-load for an extended period of time I will likely ask your help to refine captube lengths and refrigerant blend to best converge on our design-point (of course if you our willing -- know you are a busy man).

Kevin

[Kevin Hotton]

Hi Michael (and other’s with AutoCascade experience),

We ran our autocascade Friday and acheived stable operation with cold methanol flow. Attached is a screenshot. I have a design-point cycle modeled in Hysys and have altered it to get close agreement with the screenshot performance. These are the key observations:

1) The system charge is slightly high in iso-butane. I wanted 60% iso-butane, 40% ethylene by weight but seem to have 63% and 37%.
2) Vapor/liquid split at the high-stage phase-separator is somewhat below design at quality 0.62 (design is 0.70 and this low value would come up with a higher ethylene blend). However, low-stage phase-separator quality is much too low at 0.37 (should be 0.50) Both of these offsets are partly attributable to too little ethylene
3) Suction pressure is too low. I want a suction pressure of 43-psig and I am getting about 20-psig (I suspect the pressure transducer at -83C location is deviating from true pressure reading because it indicates a lower evaporator pressure than compressor suction pressure).

My initial design had calculated captubes of: 12-ft 0.064-inch ID for both low and high stage phase-separators and 10-ft of 0.052-inch ID captube for the evaporator. I have since shorted the first two to 10-ft and the last to 7-ft. I am considering shortening the second phase-separator (low-stage) captube to 9-ft and the final evaporator captube to 6-ft. However, I am not very clear on the behavior of an autocascade and if these changes will result in the intended outcome (higher second phase-separator quality, higher suction pressure and greater cooling duty)?

Low second phase-separator quality (I think) is caused by too cold a temperature in the second phase-separator. This I think would be caused by too much liquid being let out of the first phase-separator (so maybe I should not have shortened this one from 12-ft to 10-ft ?).

I am getting to the right (design-point) temperature at the inlet of the final evaporator captube but the outlet pressure is too low. This seem to suggest shortening this captube. Is this correct? It is very easy to get confused thinking about an autocascade system.

I guess if I shorten the second and final captubes by 1-ft each this will in effect “lengthen” the first captube (as to its relative flow rate) and hopefully get me the performance results I am seeking.

Any comments recommendations are welcomed.

Kevin, Matt, and Joe

PS: I have also attached the “design-point” and “as tested” Hysys models.
PSS: I just realized that I can compare the "design-point" and "as tested" Hysys models to compare the captube flowrates. Doing so, it is apparent that I need to shorten both the second and even more so the third captube.


This is Screenshot


Design-Point Hysys Model I am trying to reach


Altered Hysys Design-Point Model to match system test results. The yellow outline on the compressor indicates liquid present. Good thing we have a suction accumulator. This also shows that refrigeration duty is not being used where intended.

[Kevin Hotton]

Hi All,

We ran our AutoCascade today and reached a steady-state condition holding a 1200W load. I believe we have the charge about right (maybe a bit over-charged) as well as the captube lengths. The major problem seems to be the evaporator heat-exchanger. Even though it is a 40-plate (1.28m2 area) model, its performance is terrible. I will need to source a different heat-exchanger. We ran into the same problem with our large cascade system. I have read that even flow distribution is critical to good evaporator performance. GEA brand plate HX's even make a specialized model with an integrated flow distribution manifold for this purpose. They are more expensive so I tried the DudaDiesel model. I can tell that the suction into the compressor is wet -- so unevaporated refrigerant was not being used to cool the methanol (and compressor discharge temperature is lower than expected at 72C.

Joe eliminated the smaller filter/driers and reconfigured the sight-glasses so they lie flat for good flow. This did help. He added a larger filter on the compressor discharge just after the oil separator. After fully charged, the first two sight-glasses were full (very little and infrequent bubbles) and the final captube was about 1/3 full but it drains directly downward into the bullet strainer and captube.

Matt is in the first photo with his hand near the iced-up methanol lines.

I don't know the exact charge (we alternately added iso-butane and ethylene) but I know it has to be close to 60% iso-B / 40% ethylene by weight with a total charge of about 3-lbs. It does appear that a fade-out (buffer tank) will be needed as start-up with a static pressure over 150-psig is not possible.

Kevin, Matt and Joe

[n00b 0f l337]

I believe your issue with evaporator HX comes down to the orientation of your HX. Your refrigerant is pooling in the bottom, and crossing the bottom set of plates, while not using the capacity that mixed gas/liquid would have if it used all the plates.
Have you considered just using a standard coil and tank configuration? Or a coil in coil evaporator designed for this size load? These methods leave one pathway for the refrigerant to carry through, and if done in a downward fashion will allow the liquid pass through that can occur versus trapping liquid.

[Kevin Hotton]

Thanks for the input nOOb,

Maybe I should try evaporating downward. The final evaporator is mounted vertical (you can see in the photo) and the methanol flows downward with the refrigerant entering at the bottom and flowing upward. Based on the simulation model, the total liquid fraction entering the evaporator is only 0.38 Liter per minute (about 1 oz every five seconds). That could easily pool in the bottom of the heat-exchanger.

Kevin

[n00b 0f l337]

Ohhh! Didn't see it, thought it was in the rack with the horizontal ones.
Yes I would suggestion downward flow and upward flow the methanol. You are trapping isobutane in there most likely which will reduce in less efficiency as the machine runs longer and longer.

Maybe this has been addressed, but looking at your diagram, do you not have a HX before your first phase separator?

[mytekcontrols]

Ohhh! Didn't see it, thought it was in the rack with the horizontal ones.
Yes I would suggestion downward flow and upward flow the methanol. You are trapping isobutane in there most likely which will reduce in less efficiency as the machine runs longer and longer.

Maybe this has been addressed, but looking at your diagram, do you not have a HX before your first phase separator?

Hi Adam... Yes they do have a a HX before each sep, but funny thing is I must have completely blanked on the orientation of all the plates (good that you weren't falling asleep on the job). For best operation all those plates really should be mounted vertically.