AWHP superheat & sub-cooling

[Gary]

This is the essential statement of importance. The system is prone to hunting. It is, by nature, an unstable process.

Does it or does it not improve capacity?

That's what is important, because if it doesn't improve capacity then stability is a moot point.

[Gary]

The simulator is actually quite good, if managed carefully. It happens to be the best process simulator on the market, with a price tag that would scare the faint-hearted.

If, during simulation, process instabilities are around, then to converge the solution is always a tough, tough call. I've been simulating for most of my trained life & know when the process non-linearities begin to bite...

If you look at the VIC article I posted, you will see what they say about this process instability - it is real & should be carefully managed, in my view. I would go so far as to install a demister pad, or filter-drier somewhere between VIC & compressor to prevent droplet fling under unstable conditions...

Or limit the charge?

[desA]

Does it or does it not improve capacity?

That's what is important, because if it doesn't improve capacity then stability is a moot point.

Where I see it improving capacity is on the superheat side of the high pressure side of the system.

With raised Tcomp,suct will come a pretty high gas discharge temperature. The Alfa-Laval example gives 130.4'C for example. This is going to require some careful management, in order to not over-stress the compressor.

The additional superheat can then be taken care of with a dedicated desuperheater, so that the condenser receives a superheat it can manage.

The scroll compressor in the lab rig, has a line on the envelope at 25'C suction gas return temperature. They must be considering this to be a safe upper limit - I'd imagine.

So using VIC with a scroll compressor, could possibly bring some issues, which an alternative compressor could manage better.

[desA]

DesA - This simple equation might explain why you get Te drift

Te2=Ta + (Te-Ta)*[m2(h12-h42)*u]/[m(h1-h4)*u2]

For a starting state at Tw=45C and ending at Tw2=65C and where all suffix 2 are the second state (which is the hot water at temperature state).

Assuming a start point of Ta=30C and Te=10C
The 1st iteration gives Te2 as 14C and the 2nd iteration gives it as 16c whilst convergence is finally 15.4C.

If U and U2 are included it reduces Te2 by a few percent but also increases the SH and so reduces the m2 and about balances out. However the inclusion of fluid backup into the condensor and subcooling changes also alter the final value but that gets too complex to write in a simple equation and really needs simulation.

Chef

Feedback on the Te,sat drift simulation:

I built an equation (to back-check Chef's findings) which uses the evap heat-transfer coeff, including superheat etc, then subjected it to various sensitivity trials.

The upshot seems to be as follows:
1. Major player is x (vapour fraction) - seems to concur with your simulation, in principle;
2. Adjusting air mass flow (fan speed) offers a solution;
3. Increasing sub-cooling - hence decreasing x, also offers a solution.

There is still a 'closing correlation' to go into it - eg. relationship between SH & TXV m'g (mass delivery) - for now, it was per observation.
I also need to model a correlation between RH% & effect on (UA) - where U=heat-trfr coeff & A=reference area.

This drift is an issue, & the parametric variations hold promise, for fine-tuning. I'm still convinced, however, that a CPR-style guarantee for the maximum suction pressure into the compressor needs to be guaranteed.

So, thus far, the combination of CPR (SV), fan speed tuning, & sub-cooling control, seem to offer the promise of protecting the compressor under severe local temperature conditions.

[Gary]

Where I see it improving capacity is on the superheat side of the high pressure side of the system.

With raised Tcomp,suct will come a pretty high gas discharge temperature. The Alfa-Laval example gives 130.4'C for example. This is going to require some careful management, in order to not over-stress the compressor.

The additional superheat can then be taken care of with a dedicated desuperheater, so that the condenser receives a superheat it can manage.

The scroll compressor in the lab rig, has a line on the envelope at 25'C suction gas return temperature. They must be considering this to be a safe upper limit - I'd imagine.

So using VIC with a scroll compressor, could possibly bring some issues, which an alternative compressor could manage better.

The Alfa-laval example has the bulb on the inlet, not the outlet. That changes everything.

With the bulb on the outlet:

There is no raised Tcomp,suct. There is no additional superheat. There is no overstressing the compressor.

The Alfa-laval example is irrelevant, all of the above concerns do not apply... and the capacity improvement will be on the low side, not the high side.

In one post you are concerned about flooding the compressor. In the next post you are concerned about starving the compressor. Which is it? These are total opposites. You can't have it both ways. And Alfa-laval can't have it both ways.

[desA]

As I read the Alfa-Laval position, their recommendation to best manage the instability (overfeed/underfeed) from the VIC on the vapour side, is to re-position the TXV SH sensor to after the VIC - as we have been working with.

This signal should operate in a reverse sense to what a normal bulb would do & actually shut back on high SH, rather than increase flow - hence their suggestion for an electronic controller.

With surging, unstable flows, you could very well have both flooding & starving within short durations of time, if the system dynamics are not carefully managed.

Personally, I like simple TXV bulbs - they are slow acting & don't introduce unnecessary system disturbances of their own.

Is there a way to have a 'reverse acting' TXV bulb i.e. one that shuts back flow slightly on rising SH - within a control band?

[Gary]

Try an AEV and see if your simulator likes that better.

[desA]

Try an AEV and see if your simulator likes that better.

Actually, the simulator, as it's currently set up for the VIC has a simple valve type - so there is no additional feedback dynamics, yet, to further complicate the picture. So, basically, it's like a fixed orifice, controlling mass flow to be a fixed value at constant reduced pressure.

The system instability comes from liquid line, to vapour time-lags & huge mismatch between latent heat of vapourisation & gas specific heat capacity.
---------------

Let's talk through the AEV concept a little more. What mass-flow target would you set it to, in the first place & how would you go about setting it up on a new system?

Could we emulate an AEV, using a TXV, with the bulb placed in something, to basically de-activate the bulb-temperature effect? I have a few spare TXV's on hand. Is there a way to trick the TXV bulb to go into a 'reverse mode'?

I can size an Alfa-Laval unit suitable for the VIC & have it brought up, to be put into the test rig - no problem. We can just confirm the duty requirements. What to do about the excessive Tcomp,discharge?

[Gary]

In the Alfa-laval example, they are showing 5K superheat at the evap outlet and 30K superheat at the VIC outlet. If the bulb were mounted at the VIC outlet there would be 5K superheat at that point. There would be no excessive Tcomp,disch. The numbers in the example are all wrong. The temps would be different throughout the system.

And the equalizer line needs to be moved along with the bulb.

[Gary]

The system instability comes from liquid line, to vapour time-lags & huge mismatch between latent heat of vapourisation & gas specific heat capacity.

Okay... try this simulation: Move the TXV from the VIC liquid out to the VIC liquid in. We will need to use an externally equalized TXV with the equalizer line after the bulb.

[desA]

Ok, fair-enough. Point taken.

What is a typical maximum SH range that can be taken up on a TXV?

[Gary]

Ok, fair-enough. Point taken.

What is a typical maximum SH range that can be taken up on a TXV?

I'm not sure if this answers your question, but in general a TXV can control down to about 1/3 of it's rated load before it starts hunting.

[Gary]

Let's talk through the AEV concept a little more. What mass-flow target would you set it to, in the first place & how would you go about setting it up on a new system?

The AEV holds the evap pressure, and therefore the mass flow, constant. The superheat decreases as load decreases and increases as load increases.

The charge must be limited to avoid excessively low superheat at low loads (below setpoint).

The fan speed needs to be regulated in order to avoid excessively high superheat at high loads (above setpoint).

[Gary]

Okay... try this simulation: Move the TXV from the VIC liquid out to the VIC liquid in. We will need to use an externally equalized TXV with the equalizer line after the bulb.

Actually I like this configuration better as it not only recovers the heat from the liquid but also the energy from the pressure drop and bypasses both around the coil.

And your simulator may like the matchup of flashing on one side and superheating on the other side.

[Gary]

I can size an Alfa-Laval unit suitable for the VIC & have it brought up, to be put into the test rig - no problem. We can just confirm the duty requirements.

As I recall, something like 25% of the mass flow flashes off as it enters the coil in order to cool the remainder down to saturation temp. This is the heat that I hope to recover and transfer with the VIC.

[desA]

I'm not sure if this answers your question, but in general a TXV can control down to about 1/3 of it's rated load before it starts hunting.

Thanks Gary. This is a very useful number to know.

[desA]

I have the VIC now running stably, but, there is a twist in the tail...

I'll work it up a little more & see where the sweet-spots lie.

[desA]

As I recall, something like 25% of the mass flow flashes off as it enters the coil in order to cool the remainder down to saturation temp. This is the heat that I hope to recover and transfer with the VIC.

Thanks Gary.

That's an interesting number. The coil flashing effect is noticeable in that the TXV exit temperature is a few degrees higher than Te,sat in the evaporator.

Where does most of this flashing occur? Distributor? Feeder lines?

[Gary]

Thanks Gary.

That's an interesting number. The coil flashing effect is noticeable in that the TXV exit temperature is a few degrees higher than Te,sat in the evaporator.

Here's a number you will like even better: I expect the capacity of the coil to increase by about 1/3... but perhaps that is overly optimistic... lol

[Gary]

I have the VIC now running stably, but, there is a twist in the tail...

I'll work it up a little more & see where the sweet-spots lie.

Is that with the TXV re-located?

You may need to increase orifice size. Does that help?

[desA]

This is in the simulator. Now please tell me if this makes sense.

What I found stabilises the system, under certain states is as follows:
1. Flash flow - say at 0.6MPa(a) - in liquid line, prior to entry into the VIC. The fluid is now 2-phase & holds temperature across the VIC.
2. The 2-phase fluid now moves onto the TXV/orifice/what you will, prior to entry into the evap.

I'll snapshot the flow diagram & let you make sense of it.

Under this scenario, I can get manageable stable spots in the simulation. The reason for all this malarky has to do with the difference in refrigerant latent heat & its gas specific heat. The flow tries to jump into all sorts of quasi-stable states, that probably would not exist for long, in a real machine - where damping occurs. In numerics, there is no real damping to speak of, unless I build it in...

[desA]

Circuit diagram. (Ignore the inter-cooler, 2nd compressor & desuperheater - they are not active in this simulation)



VIC evaporation duty, under liquid line - 1st flash

[Gary]

This is in the simulator. Now please tell me if this makes sense.

What I found stabilises the system, under certain states is as follows:
1. Flash flow - say at 0.6MPa(a) - in liquid line, prior to entry into the VIC. The fluid is now 2-phase & holds temperature across the VIC.
2. The 2-phase fluid now moves onto the TXV/orifice/what you will, prior to entry into the evap.

Try moving the TXV/orifice to the entrance of the VIC. This should provide 2-phase flow in the VIC, without flashing in the liquid line. In your diagram, #10 (TXV) is eliminated and #3 (Valve) becomes TXV.

[desA]

^ It's a simulator, so think of line (8) as having zero length...

Ok, so on the real VIC, we hard-mount the valve directly onto the unit - or provide a small impact plate to protect the HE.

Next question - from the VIC, the feed is 2-phase to the evap TXV... Is this doable, or should a different valve/orifice be used here, instead? I'm thinking from a control perspective.

[Gary]

^ It's a simulator, so think of line (8) as having zero length...

Ok, so on the real VIC, we hard-mount the valve directly onto the unit - or provide a small impact plate to protect the HE.

Next question - from the VIC, the feed is 2-phase to the evap TXV... Is this doable, or should a different valve/orifice be used here, instead? I'm thinking from a control perspective.

I'm thinking move the TXV to the other side of the VIC.

Liquid line>TXV>VIC>evap>VIC>bulb>compressor

[desA]

Ok, let me give that a whirl...

[desA]

I'm thinking move the TXV to the other side of the VIC.

Liquid line>TXV>VIC>evap>VIC>bulb>compressor

To get a temp differential across the VIC, an orifice, or 2nd valve will be required as follows:

Liquid line>TXV>VIC>valve>evap>VIC>bulb>compressor

The TXV will also have to open at higher line pressure. I'll simulate some more & then try to make sense of the findings. I'm looking for stability as much a possible.

[Gary]

Now you've got me doing it. I'm trying to make the simulator work instead of trying to make the VIC work.

Let's go back to the original configuration, i.e. liquid line>VIC>TXV>evap>VIC>bulb>equalizer>suction line.

It doesn't need 2-phase flow. It doesn't need that extra valve before the VIC. It will work. Let's build it and then deal with the issues, if there are any.

Tell Alfa-laval to move the bulb AND the equalizer line to the VIC outlet and come up with some real numbers. Changing the equalizer line location along with the bulb location will minimize or eliminate the hunting, TXV's do not act in reverse... and the VIC is NOT going to cause high superheat.

I don't know what's wrong with your simulator. Is it too late to get your money back?

As to sizing, you are sizing a DX/liquid subcooler coil to chill the liquid flowing through the liquid line from 60C to 20C. On the DX side the Te,sat=15C.

[mad fridgie]

Now you've got me doing it. I'm trying to make the simulator work instead of trying to make the VIC work.

Let's go back to the original configuration, i.e. liquid line>VIC>TXV>evap>VIC>bulb>equalizer>suction line.
correct 100% use of evap for phase change
It doesn't need 2-phase flow. It doesn't need that extra valve before the VIC. It will work. Let's build it and then deal with the issues, if there are any.
Correct
Tell Alfa-laval to move the bulb AND the equalizer line to the VIC outlet and come up with some real numbers. Changing the equalizer line location along with the bulb location will minimize or eliminate the hunting... and it is NOT going to cause high superheat.
The vapor entering the VIC from the evap may be wet, the TXV will control the superheat, the superheat will be lower in this configuration,
I don't know what's wrong with your simulator. Is it too late to get your money back?
To me the simulator is only steady state with moving averages, as we all know the system is always in Flux.

As to sizing, you are sizing a DX/liquid subcooler coil to chill the liquid flowing through the liquid line from 60C to 20C. On the DX side the Te,sat=15C.

Correct, the DX side will be by mass % vapor (as most work has already been completed by the evap

[desA]

I'll do the re-routing - it's been done umpteen times in the last weeks.

I'll predict again, ahead of time, that as long as the evaporator remains an evaporator, with full evaporation service & the VIC acts as a sub-cooler/super-heater, then the system will remain smooth. The instant that the main evaporator is called on to have a wet exit, & the VIC to take on a partial evaporation load, is where all hell breaks loose.

The reason for this effect lies in the huge difference numerically between the latent heat of evaporation of the refrigerant while evaporating & its specific heat, as a gas.

This swing over function for the VIC from a super-heater, to partial evaporator seems to cause system instabilities. The next stable point, under this scenario is a liquid temp of something like -60'C - to get the correct heat balance across the VIC.

To my mind, it looks like there is a mismatch of heat & an attempt to defeat the laws of thermodynamics - the simulator will not allow this to occur.

I will lay my hat on the table that, when we've built this animal, that it will hunt & that we'll have to damp it down somehow.

I don't know what's wrong with your simulator. Is it too late to get your money back?

I'll bet you a new heat-pump that we are breaking thermodynamic rules somewhere & that the simulator is trying to help us not to...

[Gary]

I'll do the re-routing - it's been done umpteen times in the last weeks.

I'll predict again, ahead of time, that as long as the evaporator remains an evaporator, with full evaporation service & the VIC acts as a sub-cooler/super-heater, then the system will remain smooth. The instant that the main evaporator is called on to have a wet exit, & the VIC to take on a partial evaporation load, is where all hell breaks loose.

Unless there is not enough refrigerant in the system to allow hell to break loose... limited charge.

[desA]

Unless there is not enough refrigerant in the system to allow hell to break loose... limited charge.

This is the wise way to go. Keep the filter-drier vertical & set up for tight charge control.

How effective would a suction-line filter-drier be at catching small liquid fling-overs? I wouldn't want to go as far as a suction accumulator as this would just defeat the whole charge-conservation principle.

Another question of importance:
When I spec the VIC, do I spec it as a sub-cooler/super-heater (sensible heat transfer), or partial evaporator (evap on LP side)? I'm thinking that the likes of Alfa-Laval & SWEP would like to know this.

[Gary]

As the charge approaches its limit there would be 2-phase liquid entering the VIC.

[desA]

I'm working on a table that shows the relationships between latent heat & specific heats for the liquid line & LP side. The ratio has units of Kelvin i.e. a temp difference.

The size of these ratios is what causes huge, huge numerical instability in the simulator & will cause some level of boiling instability in the VIC - this is a fact of physics.

[desA]

As the charge approaches its limit there would be 2-phase liquid entering the VIC.

Ok, I'll spec the VIC as a partial evaporator, & get guidance from the plate folks on the useful operating range - turn-down ratios etc.

[mad fridgie]

One option that may help stability is a "Henry Superheat Sensor" I have never used one, Maybe gary has expericience with these (US made I believe)

[Gary]

Carefully limiting the orifice size can help, too.

I think the important point here is that we can work out any issues we run into.

[Gary]

One option that may help stability is a "Henry Superheat Sensor" I have never used one, Maybe gary has expericience with these (US made I believe)

Sorry... I am familiar with Henry, but haven't come across this.

[Gary]

I wouldn't want to go as far as a suction accumulator as this would just defeat the whole charge-conservation principle.

I don't understand this. Can you elaborate?

[Gary]

Unless there is not enough refrigerant in the system to allow hell to break loose... limited charge.

Can your simulator handle this? Can it reduce the load until the TXV starts hunting and then reduce the charge until it stops hunting?

[Gary]

Hmmm... You have a 4 row coil. What if the first 3 rows were the evap and the 4th row were the superheater, with the VIC in between?

liquid line>VIC>TXV>evap123>VIC>evap4>bulb>equalizer>suction line

[desA]

Originally Posted by desA
I wouldn't want to go as far as a suction accumulator as this would just defeat the whole charge-conservation principle.

Gary:
I don't understand this. Can you elaborate?

As I understand the accumulator principle, a certain amount of liquid will reside in it during operation - at very best, it would only have vapour in it, during operation.

This adds both cost (the accumulator) & additional refrigerant mass.

[desA]

Originally Posted by Gary
Unless there is not enough refrigerant in the system to allow hell to break loose... limited charge.

Gary:
Can your simulator handle this? Can it reduce the load until the TXV starts hunting and then reduce the charge until it stops hunting?

It could, if I built in all the bells & whistles - physical vessel & pipe volumes etc. At this point, the simulator is set up as a simple bare-bones thermodynamic calculator - something along the lines of Coolpack - but with more options to add in, when required. It's a build-as-you-go set-up.

That's why I use the lab machine... a lot less simulation headaches

[Gary]

As I understand the accumulator principle, a certain amount of liquid will reside in it during operation - at very best, it would only have vapour in it, during operation.

This adds both cost (the accumulator) & additional refrigerant mass.

Rotary compressors all have small accumulators mounted on the side of the shell (and for good reason). I'm betting these are very inexpensive.

[desA]

Hmmm... You have a 4 row coil. What if the first 3 rows were the evap and the 4th row were the superheater, with the VIC in between?

liquid line>VIC>TXV>evap123>VIC>evap4>bulb>equalizer>suction line

I'll send through a schematic of the model - let me know if the connectivity makes sense. I'll run it up in the morning.

Ignore valve V1 for now - is has zero pressure drop & is just inserted in case we need to play with inlet pressures later.

[desA]

Originally Posted by desA
As I understand the accumulator principle, a certain amount of liquid will reside in it during operation - at very best, it would only have vapour in it, during operation.

This adds both cost (the accumulator) & additional refrigerant mass.

Gary:
Rotary compressors all have small accumulators mounted on the side of the shell (and for good reason). I'm betting these are very inexpensive.

I looked into that one & my understanding was that for many of these rotary compressors, the compressor sits in the bottom of the casing. The accumulator sits above that & feeds down into the compressor suction. I think Sanyo, or even one of the Korean brands did this.

There are most probably other reasons, of course.

[desA]

There we go. Evap>VIC>s/h

I'll run up some tests tomorrow to see what comes out...

[Gary]

I looked into that one & my understanding was that for many of these rotary compressors, the compressor sits in the bottom of the casing. The accumulator sits above that & feeds down into the compressor suction. I think Sanyo, or even one of the Korean brands did this.

There are most probably other reasons, of course.

As I understand it, on a rotary compressor the suction gas feeds directly to the compressor section and thus has near-zero tolerance for liquid, hence the accumulator.

The discharge from the compressor section dumps into the shell. IOW, the entire compressor is on the high side of the system. The shell runs very hot.

[Gary]

There we go. Evap>VIC>s/h

I'll run up some tests tomorrow to see what comes out...

I'm thinking the air should go through the evap first and then the superheater... but I suppose we could try it both ways.

Whichever gives us the higher air dT is the better way.

[desA]

^ The super-heater coil row is generally placed in the path of hottest air stream - ie. at air inlet. The air leaving super-heater is then used to evaporate the lower temp liquid.

Another thing to notice is that, in general, the cold air outlet temp is rarely colder than the superheated refrigerant outlet temp. In other words, very little 'temperature cross' occurs.

My commercial evaps do actually have around 1-1.5'C temp cross, but the lab machine can't get there - inefficient coil design.