PDA

View Full Version : Carters Widget Revealed (Just Facts - Nothing Else)







chillerman2006
23-08-2011, 02:22 AM
Originally Posted by mad fridgie
I have been teasing you guys with what maybe?? a potential way of increasing the net refrigeration effect of a refrigeration system. i would like to thank DesA and Chef for their input to date.
Basically, i have built a number of test riggs, with limited resources and knowledge. i have had a wide range of results from very good to the exact opposite. And very little or no repeatability.
before we get to technical in my mind the widget, is simple to make (low cost) for those who have the right machines (lathes ect), simple to install and commission. (But I could be wrong)
We do have to make a start and the best way is to ensure that we understand how the very basic system works.
I am looking for alternative opinions, so i will be asking the questions, in a method to build up how the widget works. (it is best not to guess what is coming next, but just to focus on the question being asked)
i would suggest that this will be a good method for those who are early in the trade to grasp the very fundementals of the refrigeration cycle (which in my opinion is now grossly overlooked in modern training)


Originally Posted by desA
Great thread. Promises to be a very interesting one. Keep it coming.





Originally Posted by mad fridgie
Starting with a blank page (do not think about previous threads, this will only mislead this part)

We have a normal simple refrigeration system (LT coldroom for example)

We are going to presume to keep it simple that there is no pressure drops, glide, heat gains or losses .

So we have high pressure liquid (from reciever) for ease at saturation but 100% liquid. It passes through the expansion valve into the evaporator. (no evap pressure drops)

So how do we determine the net refrigeration effect (how much cooling is taken from the coldroom)?




Originally Posted by desA
Net refrigeration effect:

Q'evap = m'r*(hout-hin) = m'r*(1-x)*hfg@Te,sat





Originally Posted by mad fridgie
So for those who may not be use to scientific or maths terms what does this mean! keeping it simple!





Originally Posted by desA
Key :
m'r = refrigerant mass flowrate [kg/s]
hout = enthalpy of refrigerant leaving evaporator [J/kg]
hin = enthalpy of refrigerant entering evaporator [J/kg]
x = vapour fraction of 2-phase fluid [kg/kg]
hfg = latent heat of vapourisation at evaporator saturation pressure [J/kg]

Strictly, to manage the exit superheat in the evaporator:

Q'evap = m'r*(h'out-hin) = m'r*(1-x)*hfg@Te,sat + m'r*Cpv*SH

where:
h'out = enthalpy of refrigerant leaving evaporator - superheated vapour state [J/kg]
CPv = exit vapour specific heat [J/kg.K]
SH = superheat [K]





Originally Posted by mad fridgie
OK, keeping it simple, if we had NO superheat at the exit of the evap, would the flash gas in the evap have any effect on the net cooling of the room. (energy only)





Originally Posted by desA
Yes.

Simply put, as flash gas is reduced, 'x' will reduce (ie. less vapour in incoming 2-phase fluid) - this will increase Q'evap (refrigeration effect). (A few simplifying assumptions in this, so far)





Originally Posted by mad fridgie
Are you sure?

I will re word. Liquid is at fixed pressure and temp (saturated). It enters the expansion valve, expands (reduction in pressure) So just before it enters the evaporator we a % of vapour by mass (weight) and a % of liquid by mass.
If there in no pressure drop in the evap and no superheat at the exit of the evap, 100% by mass is all vapour at saturation leaving the evap.
Does the original flash gas (that entering the evap) have any benefit to the net cooling.?

We are looking at energy only, not effect on heat transfer co-efficients





Originally Posted by Peter_1
Just noticed this one again... Interesting.

The flashgas is taken into account for calculation of the net refrigeration effect.
Will try to explain my vision: all the bubbles together present at the exit of the TXV have a certain mass at a certain enthalpy. During the pass through the evaporator, they will absorb heat, so the enthalpy will change and....

No, wrong Peter, twisting my brain again,..If it's already 100% gas and we add heat, then it becomes superheated. The bubbles are already at a complete gaseous phase, surrounded by liquid which still can absorb heat, So, the flashgas is already at the condition like we will find finally all the liquid on the right side of the saturated curve. Due to the fact that the flashgas is already 100% gas, it must be positioned on the right side of the curve.

What's Carters Widget, a little bit difficult for me to translate..what's widget anyway?





Originally Posted by Tayters
Right time to jump in with both feet.

The flash gas is caused where the energy contained in the liquid refrigerant entering the TEV is used to lower the temperature of itself to get to the boiling point in the evaporator.

i.e. Te is -30*C, Tc 35*C subcooling 5K means as ref. enters the TEV at 30*C it has to drop temperature by 60K to boil. Ideally the temperature of liquid feed to the TEV would be as low as possible as this shifts the expansion line on a PE chart over to the left into the higher % region of liquid/vapor mix. If it was 100% that means the liquid at the TEV was already at -30*C. If it was you wouldn't really need a refrigerant circuit as you somehow had the means to get refrigerant at the same temperature as Te so you'd use that means instead.

So in answer the flash gas has no net effect in cooling the room. It only cools the refrigerant to the evaporators boiling point.

How does that sound?

Cheers,
Andy.





Originally Posted by mad fridgie
mad fridgie = carter.
I have to agree with Tayters (well written)
Only when we introduce superheat in the evap does the flash gas have any effect on the net cooling (energy only)
A very simple statement " the amount of liquid which is boiled in the evaporator determines the net cooling effect"
More liquid boiled more refrigeration, less refrigerant boiled less refrigeration.
Can we are agree on this very basic statement.





Originally Posted by mad fridgie
The first few questions are not directly related to widget, just simple understand of how a refrigeration system works. We all need to agree on certain effects. They are not trick questions, just something we all always take for granted.
if I just stick the design out then we will get confused. It is coming!





Originally Posted by mad fridgie
At the end of a normal evap, we now have the combination of the original flash gas produced and the vapour produced during boiling (load) If this combination is superheated in the evap (what is used to control the TEV) then the total mass must absorb energy from the room, so extra net cooling is created by the flash gas proportion and vapourised proportion "superheat" (words are not my fortie)





Originally Posted by mad fridgie
Next question,
again keeping it simple.
We have the simplest of compressors suction and discharge. (fixed speed, fixed displacment)
We can maintain a constant outlet pressure
we have a suction pressure, and for ease at saturation ("O no that will f*** the comp)
This will give us a certain mass flow.
Again only looking at the compressor (not the system) How do we increase the mass flow through this compressor?
Not a trick question





Originally Posted by install monkey
increased suction pressure?





Originally Posted by mad fridgie
Yes! Simple but important later





Originally Posted by mad fridgie
So to clarify the above 2 questions 100% mass enters the compressor, 100% leaves the comp, 100% passes through the cond, 100% pass through the expansion device, 100% pass through the evap then back to the comp.





Originally Posted by mad fridgie
We are getting close.
Imagine if you can we have an intermediate pressure vessel (very much like an open flash economiser on a screw set)
High pressure liquid in, pressure drops, liquid at the bottom, vapour at the top.
For ease we will say that the pressure in mid way between the Saturated cond pressure and the Saturated Suction pressure.
Now Focusing on the liquid proportion "only" (the liquid which is leaveing the vessel) and we are talking about energy only, would the new liquid mass flow change the net refrigeration cooling





Originally Posted by mad fridgie
Well good input from both,
Infact if we just look at the "liquid part only", as determined in the first question, the net refrigeration remains the same, as the flash gas has no effect on the net cooling ( to keep it simple we are basing all observations when the refrigeratant is at saturation)





Originally Posted by mad fridgie
Now The vapour side of the vessel?
Is this vapour at a higher or lower pressure than that at the evap outlet. (and what would its properties be chef's question)





Originally Posted by mad fridgie
We are just about here.

Do you know what, a venturi, jet compressor, ejector or condell mover is?

basically a device that has a motive force (a high(er) pressure feed), a suction port (lowest pressure) and an outlet pressure (which is the sum of the motive and suction mass streams) medium pressure





Originally Posted by mad fridgie
The vapour from the vessel becomes the motive force (highest pressure)
The vapour exiting the evaporator enters the suction port (lowest pressure)
The combined flow leave the devices and enters the compressor.(medium pressure)

effectivly we have a 2 stage compression system.

The suction of the compressor would be higher than if it came just from the suction.

The compressor mass flow would increase (higher inlet pressure)

So the total system mass flow would increase, including increased liquid mass flow through the evaporator.

Carter Widget (simplified)





Originally Posted by mikeref
O.k. Stay of execution for now. Saturated vapour but not to the point of liquid to compressor. Has to be higher pressure than evap. outlet so it can be throttled down. Seperate line supplied to compressor?





Originally Posted by mad fridgie
No it goes through a device like a "venturi" for ease of desription and mixes together.

Ok another way of looking at.

You have 2 closed buckets, each has the same refrigerant in it, but at 2 completly different pressures/temperatures (saturation) High and Low. If there was a pipe between the 2 buckets and valve. If we opened the valve what would happen to the pressures inside the 2 buckets?





Originally Posted by chillerman2006
So (if I am understanding this correctl) the mass flow is increased to the widget by removing the vapour (same as push/pull recovery of a system)

The increase in mass increases the compressors mass flow rate

Which in turn increases system mass flow rate

????????





Originally Posted by mad fridgie
Yes basically, but it is a "non steady state" system to start (it does not happen instantly) until it reaches equalibrium, and is a circular calculation (no start or end), which does make hard to understand.





Originally Posted by mad fridgie
Try this very basic device.

A piece of 7/8 pipe 6' long, stick your oxy/act tip in on end (middlish if possible) leave this end still open to the air, turn on oxygen (really you should use nitrogen to be safe) but it is the tip that is important. (motive force)
What you should see is that air is pulled into the pipe at one end (suction) and quite alot flow coming out of the other end. ( alot more flow that just the oxygen flowing from the tip)





Originally Posted by mad fridgie
I am no specalist in venturi or the like ( i know nothing about them) only possible application. The question then comes down to what would the actual properties be of the outlet of the widget (The inlet to the compressor), When we move away from perfect design (which as we all know happens on every system) what then would be the effects on the system.
there are other possible beneifts as well. (or maybe)





Originally Posted by chillerman2006
Similar to how the oil is returned from the evaporator in a flooded phe system, high pressure vapour is driven down a pipe that is connected to the oil pipe and as it flows it pulls the oil into the vapour and the mixed vapour/oil then returns to the compressor/or oil reservoir

???





Originally Posted by mad fridgie
Basically "yes".

I am the first to know my limitions, and I am at this point now.
I have been talking to a number of true boffins (profs ect) all think is has the potential.





Originally Posted by mad fridgie
There must be a flaw in my methodology. Surely others must have looked at this simple system!





Originally Posted by mad fridgie
The idea is that the widget was to be quite simple (hence cheap) but yes could quite easily control with a sol valve and could also be made with proportinal control to keep optomised. (but we but the cart before the horse)
The very basic idea was the vessel was maintained at a constant pressure. (this makes sizeing the evap and expansion device easy) This gives us the oppotunity to allow the head pressure to fall, a lot lower that is presently classed as standard. ( so looking at the very big picture with lower head, we would get less flash gas, so the widget would not work as well, but we still maintain very high efficiency and system equalibrium) All the components could be produced as a single item, so installation would be very very similar to installing a suction accumulator with an internal liquid line heat exchanger. (Maybe with a removable nozzle orifice for different size systems, in a similar way we change the orifice size in an expansion valve)
I have but much thought and energy into this.





Originally Posted by mad fridgie
You need to read from the start (there is reason behind my madness, in the way i started the thread) Yes there is a flash tank, or I called a vessel.
It is at pressure in the middle between normal evap pressure and cond pressure.
It is positioned in the liquid line between the cond outlet anf the TEV (expansion device) inlet.
The flash gas in the vessel is something that would normally occur just after the expansion valve prior to evap inlet, and has in energy terms very little benefit to the refrigeration system. how ever we do have to pump it around the whole system.





Originally Posted by MikeHolm
I understand that it is not desireable to have flash prior to the evap so i can see removing it before the TEV so am I to understand that the amount that would be moved through the device is only that amount that would normally flash prematurely and not try to get more than usual?

again head on chopping block





Originally Posted by mad fridgie
Mike first, yes we only take part of what would normally be produced in standard system. And practically has very little use (my first question) but in a normal system still has to be 100% transported around the whole system "energy in"
CM
The vapour leaving the widget (including the ejector) is combined with the flow from the evap. The vapour can only flow if there is sufficient flash gas coming from the condensor liquid, If for example the liquid was very highly sub cooled prior to the widget, then no flash would be produced in the vessel and very little flow would occur through the ejector. (the ejector has avery small hole, need good pressure to flow)





Originally Posted by Magoo
Perhaps a compounded cycle with inter-cooler.
magoo





Originally Posted by Magoo
Hi Chillerman,
the low stage piston/s suck from a low load and discharge into a vessel [ intercooler ] that condenses partiallly, the resultant vapour becomes the suction of the high stage pistons, which discharges to final condenser, the partial liquid formed in intercooler is regengerated to supply low load topped from high stage liquid. [ jet valve area ]
A short cycle description, you have two levels of evaporation and two levels of condensing. easy with screws / piston ratios with recips.
magoo





Originally Posted by mad fridgie
Effectively the widget is that (without the inter cooler)

also known as two stage (in one compression chamber, out of this one into a second compression chamber.)





Originally Posted by mad fridgie
I configure my own heat exchange selection, slightly different to how most do it.
I presume you are talking about multipass HXs, The flow regime does not seem right for my application.





Originally Posted by mikeref
O.k. Mad. At the risk of a crash and burn probablity, there is another thought i have. Can there be a free energy effect,( minus mechanical losses) where one could use discharge pressure, or more likely liquid, while compressor is running to power a small turbine. My thought is.. the slave turbine in the suction line can increase suction pressure to compressor at the added benefit of reducing evaporator pressure, thus increasing evaporator efficiency? Seals will be a problem and so will driveshaft configuration. Increased pressure to compressor gives better efficiency on the low temp applications. So.. what do you say? Has this been debated before?.. Mike,( not the one sipping beverages on the beach :p).





Originally Posted by mad fridgie
There is no crash or burn on this thread, (not on my watch)
The discharge pressure has to be produced by power, (how ever some have looked at using ejectors to produce higher than motive pressures, but this is beyond my what I know)
using liquid soley as a driver (very little benefit) this watt a low wattage pump can pump L/S versurs head.
using the liquid then expanding through a turbine may work (energy purpose only), very similar to gary's attachment.
A lot of work has been done on efficiency and using ejectors, but I "believe" using the widget method is original and somewhat more practical (I not attepting to get 100% better performance 100% of the time)
good thoughts.
Even though you have a concept you still need to prove possible performance. I have my method, which may be right or wrong, but I am wanting others appraoch to see if it could be a goer.





Originally Posted by desA
If you split the flows as follows:
1. Vapour only (your 'flash gas') at m'fg;
2. Liquid only at m'l...
3. m'r = m'l + m'fg (total flow before split)

... then, if we know the pressure of each stream, we can locate the individual state points as follows:
1. Vapour - assume saturated vapour - located on right of bell;
2. Liquid - assume saturated liquid - located on left of bell.

If the fluids are mixed, then the mixture properties are constructed based on 'x', the vapour percentage in the mixture.

If the vapour is transported away, then only m'l goes towards the evaporator. The net refrigeration effect (assuming fully saturated liquid) is:

Q'evap = m'l*hfg

Where :
Q'evap = evaporator capacity [W]
m'l = liquid mass flowrate [kg/s]
hfg = latent heat of vapouraziation [J/kg]

So, the real amount of duty [W] managed by the evaporator will depend on how much liquid is actually going to it.





Originally Posted by Segei
I'm not an expert in venturi effect, but I have 2 concerns about this widget.
1. To create venturi, we need significant pressure difference between pressure in this vessel and suction pressure. This pressure difference will not be significant.
2. In verturi, main flow will suck secondary flow. As far as I know, main flow should be significantly greater than secondary flow. For example 10:1. Assume that flash gas is 20% of total gas flow. 80% of gas create by net refrigeration effect. It means that that 1 part of main flow will suck 4 parts of secondary flow. This will not happen.





Originally Posted by desA
A thought:
1. Replace TXV with small rotational device;
2. Liquid flashes across rotational device (RD1) to desired LP;
3. Rotational device (RD1) drives a rotational device (RD2) placed in suction line.
4. RD2 'stuffs' flow into compressor.





Originally Posted by mad fridgie
It is unlikely to be a venturi (as you know it) I have left this part open to what it actually is, Basically it is a Thermal Vapour Compresssor (TVR) or some form Condell effect mover. (these devices use less mass flow for the motive force as the driver).

Your are not really sucking, as this applies a stastic flow, as we know we have flow through the suction.

The first question has to be does it have the potential to work, and how to we prove what the results would be in a perfect world.
Secondly if the question is yes then what piece of equipment is required to achieve the results.

I have formulated my own calculations, and have built a test rigg(s) to test. Thats does not mean my method is correct, nor the parts chosen for the test rigg are even close to what is required.





Originally Posted by mad fridgie
In this industry they do not use the terms primary and secondary, for that reason, "motive and suction" flow

How do we prove if it works? Then How can we make work?





Originally Posted by mad fridgie
It is some what more complicated in the theory, how ever I do not want show how may calculations have been achieved. I do not need some to check that i can add my apples calculation, when in fact what i should calculatting is bananas, in other words my method of determining efficiency/performance may well be flawed, and I do not want to influence, others thought patterns on what the result maybe.
The practical objective is to reduce the compressor displacment, increase COP.





Originally Posted by mad fridgie
The pic is just a snapshot in time at a steady condition!
Practically and what is required to prove performance is a the system moving to reach a steady state.
The compressor inlet pressure will only increase when the process is underway, so to start with it will run similar to a normal system, as we start to recompressor, the compressor mass flow will increase, which inturn makes more flash gas in the vessel, which gives greater levels of recompression and so on until equalbrium is reached.





Originally Posted by mad fridgie
The drawing show the liquid/vapor seperation in the vessel.
100% in liquid (35C sat)
Reduce pressure (0C sat)
Liquid leaving 69% (0Csat)
Vapour leaving 31% (0C sat)
This would remain a constant as long as the incoming liquid remained constant.
How ever the actual total mass flow SHOULD increase over time (as the process fires up).
PS figures care of old faithful Coolpack!





Originally Posted by mad fridgie
Long story short!
First rigg, just modified an existing rigg nothing to special, to see if there was potential, had possible results.
(also some flawed testing) because part of the design is to consider what happens if there is a failure with the widget (cause and effect)
Built purpose rigg, with 100% isolation and by-pass of each and every component. So it can be run as 100% conventional or 100% with widget and all steps im-between
can control load (flow and temp), pressure, Superheat, sub cooling, and condensing flow rates.
Basically we are measuring for results cooling capacity, power draw and heat rejected, whilst measuring the parts of the refrigeration system to ensure that we keep with component specifications.
The TVR could be wrong (venturi type device) more than one size and the wrong type, the vessel could be wrong, asumed peak performance may be wrong.





Originally Posted by mikeref
Mad, so using your control rig, if it is only a scale model of the real working size, and identical rig with device in place, would there not be changes such as weight of refrigerant and volume of flow that has to be taken into account? Maybe gravity and ambients have effects. ( Assuming these are scale models as i can't see you moving coldrooms to UK on Air N.Z.)





Originally Posted by mad fridgie
Mike,
The unit is a full size model, could be made bigger, could be made smaller, i would expect the complete widget to be the size of an accumulator (the size that would suit a certain size plant)
The test rigg is basically a water cooled, water chiller. (it is some what easier to control and monitor water on both sides) the actual application is not important, it is only a refrigeration circuit.
Copeland ZB 45 on R134a "Brand new"
The bigger the compression ratio, then it would seem the bigger the increase in performance.
for example (using my clacs which could be completely wrong!!) an Ac improvement would only 7-9% increase max net cooling, verses a LT system upto 55% increase in net cooling. (personally I was looking at it for hot water heat pumps)





Originally Posted by mad fridgie
We have proven that the flash gas (just after TEV) in a normal system has little benefit to the net cooling effect.
By having an indermidiate pressure, we retain the mass of liquid required to meet the refrigeration load, plus we have vapour at pressure higher than the suction, this vapour has the potential to do some work (chosing the words wisely I hope, as not to influence your thought patterns) How would you then determine the work available and how would this be applied, remembering that at the end all streams will pass through the compressor.





Originally Posted by mad fridgie
I normally use the term expansion device, just used TEV for short, but if was a TEV, then i would presume at the evap. So your next aargument is that we will have some superheat, therefore the flash gas would indeed slightly increase the net cooling effect.
For simplistic purposes I am just using saturation points (subcooling, superheat will of course effect what happens within the system, as it always does)
On the other hand, if the widget has a whole (this how i invisage it) is mounted next to the evap, then maybe the bulb could be mounted after the widget to ensure a fully wetted coil.





Originally Posted by mad fridgie
very witty,

Looking forward to that one coming.

If the thing works, and we can prove it, there are plenty of other considerations that to be needed to be accounted for. (many I already know, likely some that i do not) But unless I can get independent proof of concept and some level of possible commercial practiabilty (can be bought, sold and installed for the right price, I believe this side is possible) then other considerations are not worth looking at.

I have spent the last year working on this (not full time) as much for fun as commercial gain. I am presently stumped! hit the wall! in the dark! can not see the wood for trees. Hence bring out into the public eye.





Originally Posted by mad fridgie
Well I hope you can use thread to help with your project,

hers some more to help you along the way.

Vortex tube to further sub cool the liquid, and to use of joul thompson to reduce compressor SH, whilst increasing mass flow, of course will need high side intercoler.

a simple turbo charger, would do the trick

You could use a an ejector and go internal sonic an elevate the exit pressure above that of the motive force, recondense and re-introduce into the liquid line.

there are good few other options but just best suited to very stable steady systems.

hope this help for a starter.





Originally Posted by Peter_1
Just reading now on page 3..what you describe MF is the working principle of a ramjet engine.





Originally Posted by Peter_1
Such an interesting thread during these busy days. MF, your schematic at the end explained a little bit more about the intermediate pressure. Have to read back now. Will continue reading later.
I also had a similar widget about increasing evaporator capacity with 0K SH and making the needed SH elsewhere with an additional coil which made also a higher SH then needed to use this to heat water. I think I once made a sketch I've send to DesA . But just like you, there must be something wrong in my thinking process.





Originally Posted by simon@parker
have just had the penny drop big style yr basically squirting higher pressure vapour at the compressor causing a sucking vacume effect to increase the velocity of the gas leaving the evap so causing less work done by comp bloody genius :) and the part yr gonna fit is like you say very basic wow think u just re invented the wheel





Originally Posted by mad fridgie
Well done, now get your think cap on.
Rememeber that the high pressure vapour, would be classed as intermidiate pressure vapour, so not to confuse it with the normal discharge pressure.
The intermediate vapour is that would just normally be wasted in standard refrigeration system. So the smarts is the combination of more than one principle and the simplistic nature of the device if we can resolve both the theorectical, then the practical issus.
On very large systems it could be more complex to achieve optiumum across the whole working range. (modulating everything)





Originally Posted by desA
As I read & re-read through this very interesting thread, I keep wondering about a few niggles going on in my engineering 'gut':

1. The ejector/motive device will be governed my a momentum interchange between the two fluids.
2. Momentum = mass*velocity = density*volume flowrate*velocity
3. Could the pressure drop in the motive device cause back pressure into the separation tank?
4. Would the liquid flash into the separation tank?
5. A niggly feeling of system feedback driven backwards from the motive device into the liquid circuit.
6. Expectation of some level of system 'bouncing'.

Niggles - can't quite put my finger on them. Will try to set up some simulations to play with gas/liquid dynamics.

Originally Posted by mad fridgie
3. only if the nozzle is to small and chokes (other wise no, unless there is some for of reverse shockwave), this would balance itself out, as the pressure would increase reduce the flash gas and increase thre liquid flow through the evap expansion device (this also comes down how we control the pressure in the vessel, up stream or down stream)
4. the vessel is at reduced pressure than the liquid line (for design purposes) if the liquid is highly sub cooled or a pressure close to that of the vessel the amount of flash gas will reduce, "this would normall indicate the system is running very efficiently anyway, and has been considered in the practical application"
5. Could be a possibilty as at the start the evap would be very starved, give low compressor mass flow, so lo flash gas. (can be resolved easily with a simple widget by-pass)
6. Likely to be abit of instability, can live with this (fairly normal for most refrigeration system, example pressure controlled cycling condensing fans)

Originally Posted by mad fridgie
Now we are thinking.
Firstly we have to consider that end results is time based, and built up over time.
You have covered this by stating that the system in its self could be a negative feedback system. (when we first start, mass flow is very low, does it continue to fall or will it rise, and then where is equalibrium reached.)
The main question (ignore the system as a whole for a while) is what would be the pressure/temperature of the two combine vapour streams? ( evap outlet and the intermediate vessel vapour outlet) (the vapour that would enter the normal compressor is the combination), or i should say best possible result.
At this stage I would have to presume (wrightly or wrongly) that the streams properties are at saturation.


Originally Posted by MikeHolm
MF, at a given point, the amount of flash will partially be determined by the surface area and insulation of the "vessel" be it a pipe or a tank and the exterior temp around the vessel and the boundary (surface) area between liquid/gas. If the the vessel is long and thin like a low loss header there will be more surface area but less liquid/gas boundary area.

Will this aid in producing a higher quality gas? I assume the above factors will help to determine the quantity of usable flash gas to be injected. What it seems to me is we are looking at a steam boiler (sort of)


Originally Posted by mad fridgie
You are correct this have to be considered in the vessel design.
First we have to seperate the liquid and vapour , so first we have slow down the exit velocity leaving expansion device (squirt effect), (could be baffle or gause, or can use it to aid seperation centrifugal force) then we have to ensure the internal velocity is low enough the gravity will actually allow the liquid to drop out, and suspose the height also becomes important to give enough time for it to happen.

the actual boundry layer between the mass of liquid and vapour is always being broken due to the falling droplets. as vapour and liquid at saturation no energy exchange occurs, but you may get a reverse action to the drops failing "splash back"

Now in heat ingress to the vessel, if in the example the vessell was not insulated, the we would likely be superheating the vapour prior to the exit, and adding a slight amount of energy in the liquid causing slightly more flash gas to drive the TVR.





Originally Posted by desA
So far, the simulations are showing up a few interesting leads to pursue (the ejector pull/momentum dynamics haven't yet been modeled):

1. The AEV seems to be required to generate a flash, otherwise the vessel stays as liquid only, with no vapour feed stream to the ejector.
1A. Practically some flash could be caused by small pipe changes/blockages as well.

2. If the liquid line is split (remains as liquid) - one part to TXV/evap, then the other part flashing on its way to the ejector - could create a sizable vapour speed/momentum to the ejector.





Originally Posted by mad fridgie
1 yes AEV or similar is always required.

2 Unless you change the conditions of the liquid entering the TEV at the evap all you a have done is reduce the mass flow to the evap, reducing net cooling, then on the other side, will the expanding liquid, pulling in the evap vapour, lift the suction pressure to the compressor enough to increase the compressor mass first to equal the original short fall and secondary to increase the mass flow to make it a increase the net cooling over a standard so that the system becomes commercially viable. (it is somehat similar to know what will be the outlet pressure and temperature of the TVR/ejector orcompressor inlet

Originally Posted by Gary
There is no doubt in my mind that reduced flashing in the coil increases coil efficiency. We can see this where there is mechanical subcooling. The part I am having trouble wrapping my head around is the contribution of the ejector.

What is missing in the drawing is the pressure and temperature at the compressor inlet.


Originally Posted by mad fridgie
That is the unknown? Not something you can find in your local refrige wholesale book. And the ejector is unlikely to be something you just buy of the shelf.





Originally Posted by Gary
The numbers in your drawing are theoretical then?... not measured values from your test rig?





Originally Posted by mad fridgie
Yes the numbers given are theorectical, I picked those because they are common in the industry.
I am unable to get repeatable results, (I have what would seem very good, similar and very poor results in comparission to standard) The reason for opening up the idea is to attempt to find a best theorectical benchmark.
I have choosen a system with a relatively big compression ratio as it would seem that this where most improvement is likely.
So the question is?
What would be the compressor inlet pressure/temperature be? When the vapour from the evap is mixed (joined) with the vapour from the intermediate seperator vessel, using some form 100% efficient device !





Originally Posted by Gary
I'm more of a measure it kinda guy than a calculate it kinda guy. The more interesting question, from my point of view, is why are you unable to get repeatable results? What are the variables? What changed?





Originally Posted by mad fridgie
That is not an unresonable question.
Direct measurment and understanding the results, only mean something when, when you compare to knowledge you already have. (you know how fridge works no doubt, so any changes in system can be easily evaluated)

Now it comes to the rigg, this has been built to operated as a standard sytem, or full widget system and all parts inbetween. (all parts can be isolated and by-passed).

All flows/temp can be set (water evap and cond), pressures can be maintained, so can superheat and subcooliing.

I can not measure refrigerant mass directly (i do not have the quality of equipment required)

But temps, pressures, water flows are well looked after.

Issues, because we are dealing with vapours and liquid at saturation it very difficult to know, if the streams are 100% seperated (and this can make a hugh difference).

and the TVR, ejector(s) I have zero data on, when asked your country men, "blank", so best GUESS was made.
This guess was baseed upon what if there was a failure of the widget, would the system still operate in some form of fashion and what was the likely hood of damage to the other component within the system.

Others issue can include changes in heat transfer properties of the heat exchangers.

All i was after with rigg is to show some regular improvent (not oerfect but enough) so the the technology could be sold to those who could develope further and bring to market. And here we are today!





Originally Posted by Gary
I'm thinking R404a might fractionate in the separator. Not sure what effect this would have, but it might be a significant unknown... possibly a de-stabilizing influence.

I wonder if R507 might be the better choice.





Originally Posted by mad fridgie
The actual rigg was R134a, but SCT of upto 80C would seem frightning to most on here. R404a, was just a refrigerant of choice because it is common.
Glide could be seem as an issue, i do not think there would be sufficent time to end up into true seperation, but of course one of the considerations, that would be needed to take into acccount further down the track.
It is the lack of repeatabilty, which is giving a head ache, repeatable failure would be better, at least we would have a constant to work with.
Hence the attempt to look deep into the theorectical side.
I am in some ways similar to your self, I have idea, do a bit of theory, then just build it and modifiy to achieve best results. belt and braces just is not working for this.





Originally Posted by Gary
When you say the results were not repeatable, what results are you talking about?





Originally Posted by mad fridgie
The widget is there for the puropose of increasing nett cooling effect, and increased efficiency

So actual cooling duty
Power draw
heat off rejection.

To explain in normal mode, all othe parts isolated and bypassed. I can turn on the system set my load and the usal parts, get a result. turn of the machine for a while, turn back on, and the results are the same (within a couple of %) This what i would expect.
I introduce the full widget, same procedure, on and off and on, differing results (by as much as 60% diff in cooling loads)
And of course all the standard sort of readings are also different as you would expect with these changes.
This type of inconsistancy is through the full range of operation, swinging in boths directions, at varying amounts of difference.

I can say hand on heart that I have on numourous occassions seen performance dramatically increased above standard, with the same power draw, plus the complete opposite.

I am not here to prove scientific annomilies, It has to work or it does not, either way I happy to find out the truth.

Just because I can not get it to work, does not mean that somebody eles can!

As one of your greater leaders

" I choose to do what I do, not because it is easy, but because it is hard" or something like that??





Originally Posted by Gary
Do the results vary during the run?... or just from one run to the next run?





Originally Posted by mad fridgie
Apart for start-up (which i actually exclude from overall results) as equalisation takes place (as with any normal system) and deos not really seem to be much longer, the system is quite stable. I can get about 8-10% variance in load, but these changes are in conjuction with the EEV changes as it attempts to find its level in the PID controll. I think there is also some historical control within the controller. (over time very stable) 3% shift on EEV, and flat line of evap SH (0.1C resolution) I





Originally Posted by Gary
Have you tried increasing the setting on the constant pressure valve? Presumably, this would increase the ejector dP, which would increase the motive flow and thus the suction flow as well. There may be some minimum dP needed for the ejector to function properly.





Originally Posted by Gary
I'm wondering too how oil return might affect the ejector operation.





Originally Posted by mad fridgie
Now your thinking!

Yes increasing the pressure in the vessel, would seem to indicate more motive force (greater pressure difference), but on the other side you will have reduced mass (less flash gas produced).

I have played with increasing pressures, decreasing pressures, changing orifice size in ejector. (present on are US made air movers).

I am sure there is an issue with ejector, but I do not have enough experience or knowledge to evaluate the ejector performance correctly. The ones chosen, and one I made maybe completly the wrong style and/or size.

If I could just of got about 30% of what I predicted (repeatable) I would have been well happy, and could of got investor or passed onto to those better well funded.





Originally Posted by mikeref
Mad, since this device has proven efficiency increase, than results on further tests can be duplicated. What has changed? Would it be the increase load on evaporator that causes increased flow through the first stage temperature reduction and maybe reducing saturated vapour available? Or vapour that you are harvesting is being redirected at times and continuing on to evap. TX valve. Superheat readings would pickup that problem. Would pressure stray too low and TX fail to function, after all, conventional expansion valves do not expect cold, low pressure liquid going through them now do they? Your thoughts Gary?





Originally Posted by mad fridgie
I would not say the device is proven, as it has not been 100% repeatable, even though some test have and others have not.
When testing you do your best to keep all variables constant. (water flow through the evap at a fixed inlet temperature)
You are correct about TEVs they do have limitations (for large application swings), for this reason I chose a carel EEV (sized to suit all possible pressure drops and duties)
I have sight glasses in the liquid lines, it would seem there is no bubbles when running constant. (sight glass is an aid)





Originally Posted by chillerman2006
MF

I know your trying to keep the system simple but

what about liquid line heat exchanger after your vessel, that the evap outlet runs through

would this give you superheated vapour to the widget & make a better mix going back to the compressor and also subcool the liquid to evap, further increasing efficiency ???


Originally Posted by mad fridgie
Well that is good question, but for the purpose of calculations I have assumed that all is at saturation, because I thought this maybe an easy starting point, I may be wrong!

Adding excess superheat to the compressor reduces the possible mass flow through the compressor.

Superheat will be needed at the compressor inlet, so depending upon what happens when the stream joins determines where the superheat is to be added.

On the rigg superheat was at the evaporator (as i am trying to replicate a standard system)

I could very easily give the impression of proof (i have even fooled myself), but thats not my game, i only want to give genuine proof.





Originally Posted by chillerman2006
MF

I was thinking the other way, good superheat from heat exchanger to balance out too much % of saturation from the vapour coming from the vessel

I see the problem now, It's knowing exactly what condition the vapour is at first - all chicken & egg or vice versa

& wind up would have been more polite ;)





Originally Posted by mad fridgie
if i was to make a guess (and i am the only one who can on the rigg) and if the concept of the process is correct.
Then likely issues are;
1; on start up some form of backwash is being produced in the ejector, (not flowing where it should within the chamber itself)
2; liquid carry over in the motive feed into the ejector (vapour from the vessel), this also reduces the mass flow to the evap.
3. Heat transfer coefficient is reduced in heat exchangers (reducing entry flash gas to evap, reduces turbulence, reduction in heat transfer)





Originally Posted by Gary
I would suggest that placing the EEV sensor at the ejector exit might better regulate the flow proportions... then adjust the AEV for optimum performance.


Originally Posted by mad fridgie
Valid point.
A couple of considerations.
As you touched upon on the other thread, as we expand the gas in the ejector a cooling effect will occur, on the other side the suction gas should be being compressed so "heating up" (all working well of course) so it would be difficult to know what to set at (apart from compressor protection), but not a silly idea.
second, by changing this we have moved away from the standard design, so results could look to be improved/or derated, only due to more normal factors, for example the evap could become fully wetted increase evap duty.
I can make it look as if it is doing more than anormal system, but it is actually small changes in the system itself that is improving performance, not always the widget. This is an issue I have always had to consider during testing. Many energy saving devices promote massive savings, when infact they are almost a gimick, improvements they portray tend to be on very poorly designed systems in the first place. I do not want this to be one of them, if this is a failure then so be, but let it be a failure on merit, not a success on Bull Shyte!

Originally Posted by install monkey
i've reviewed the concept the design spec ,the fundmentals also the logic and expected outputs and cost savings and i have a query over the widget.-what colour is it???

mad fridgie
23-08-2011, 02:46 AM
Thanks CM, you are a computer whizz. Can you re post the drawings, on this thread please.

chillerman2006
23-08-2011, 03:11 AM
'Carters Widget Drawing's'

Gary
23-08-2011, 07:01 AM
There is no doubt in my mind that reduced flashing in the coil increases coil efficiency. We can see this where there is mechanical subcooling. The part I am having trouble wrapping my head around is the contribution of the ejector.

What is missing in the drawing is the pressure and temperature at the compressor inlet.

mad fridgie
23-08-2011, 07:26 AM
The part I am having trouble wrapping my head around is the contribution of the ejector.

What is missing in the drawing is the pressure and temperature at the compressor inlet.

That is the unknown? Not something you can find in your local refrige wholesale book. And the ejector is unlikely to be something you just buy of the shelf.

Gary
23-08-2011, 03:49 PM
The numbers in your drawing are theoretical then?... not measured values from your test rig?

chillerman2006
23-08-2011, 08:34 PM
SEE ABOVE - (moved up)

chillerman2006
23-08-2011, 10:33 PM
SEE ABOVE - (Moved up)

mad fridgie
23-08-2011, 11:41 PM
The numbers in your drawing are theoretical then?... not measured values from your test rig?

Yes the numbers given are theorectical, I picked those because they are common in the industry.
I am unable to get repeatable results, (I have what would seem very good, similar and very poor results in comparission to standard) The reason for opening up the idea is to attempt to find a best theorectical benchmark.
I have choosen a system with a relatively big compression ratio as it would seem that this where most improvement is likely.
So the question is?
What would be the compressor inlet pressure/temperature be? When the vapour from the evap is mixed (joined) with the vapour from the intermediate seperator vessel, using some form 100% efficient device !

Gary
24-08-2011, 12:17 AM
I'm more of a measure it kinda guy than a calculate it kinda guy. The more interesting question, from my point of view, is why are you unable to get repeatable results? What are the variables? What changed?

mad fridgie
24-08-2011, 12:41 AM
I'm more of a measure it kinda guy than a calculate it kinda guy. The more interesting question, from my point of view, is why are you unable to get repeatable results? What are the variables? What changed?

That is not an unresonable question.
Direct measurment and understanding the results, only mean something when, when you compare to knowledge you already have. (you know how fridge works no doubt, so any changes in system can be easily evaluated)

Now it comes to the rigg, this has been built to operated as a standard sytem, or full widget system and all parts inbetween. (all parts can be isolated and by-passed).

All flows/temp can be set (water evap and cond), pressures can be maintained, so can superheat and subcooliing.

I can not measure refrigerant mass directly (i do not have the quality of equipment required)

But temps, pressures, water flows are well looked after.

Issues, because we are dealing with vapours and liquid at saturation it very difficult to know, if the streams are 100% seperated (and this can make a hugh difference).

and the TVR, ejector(s) I have zero data on, when asked your country men, "blank", so best GUESS was made.
This guess was baseed upon what if there was a failure of the widget, would the system still operate in some form of fashion and what was the likely hood of damage to the other component within the system.

Others issue can include changes in heat transfer properties of the heat exchangers.

All i was after with rigg is to show some regular improvent (not oerfect but enough) so the the technology could be sold to those who could develope further and bring to market. And here we are today!

mikeref
24-08-2011, 12:43 AM
Gary, going to be some changes to your weather real soon. Battern down the hatches time.:off topic:

chillerman2006
24-08-2011, 01:36 AM
Moved Up - (See Above)

Gary
24-08-2011, 02:17 AM
I'm thinking R404a might fractionate in the separator. Not sure what effect this would have, but it might be a significant unknown... possibly a de-stabilizing influence.

I wonder if R507 might be the better choice.

mad fridgie
24-08-2011, 03:28 AM
The actual rigg was R134a, but SCT of upto 80C would seem frightning to most on here. R404a, was just a refrigerant of choice because it is common.
Glide could be seem as an issue, i do not think there would be sufficent time to end up into true seperation, but of course one of the considerations, that would be needed to take into acccount further down the track.
It is the lack of repeatabilty, which is giving a head ache, repeatable failure would be better, at least we would have a constant to work with.
Hence the attempt to look deep into the theorectical side.
I am in some ways similar to your self, I have idea, do a bit of theory, then just build it and modifiy to achieve best results. belt and braces just is not working for this.

Gary
24-08-2011, 04:03 AM
When you say the results were not repeatable, what results are you talking about?

mad fridgie
24-08-2011, 04:26 AM
The widget is there for the puropose of increasing nett cooling effect, and increased efficiency

So actual cooling duty
Power draw
heat off rejection.

To explain in normal mode, all othe parts isolated and bypassed. I can turn on the system set my load and the usal parts, get a result. turn of the machine for a while, turn back on, and the results are the same (within a couple of %) This what i would expect.
I introduce the full widget, same procedure, on and off and on, differing results (by as much as 60% diff in cooling loads)
And of course all the standard sort of readings are also different as you would expect with these changes.
This type of inconsistancy is through the full range of operation, swinging in boths directions, at varying amounts of difference.

I can say hand on heart that I have on numourous occassions seen performance dramatically increased above standard, with the same power draw, plus the complete opposite.

I am not here to prove scientific annomilies, It has to work or it does not, either way I happy to find out the truth.

Just because I can not get it to work, does not mean that somebody eles can!

As one of your greater leaders

" I choose to do what I do, not because it is easy, but because it is hard" or something like that??

Gary
24-08-2011, 04:54 AM
Do the results vary during the run?... or just from one run to the next run?

mad fridgie
24-08-2011, 05:12 AM
Apart for start-up (which i actually exclude from overall results) as equalisation takes place (as with any normal system) and deos not really seem to be much longer, the system is quite stable. I can get about 8-10% variance in load, but these changes are in conjuction with the EEV changes as it attempts to find its level in the PID controll. I think there is also some historical control within the controller. (over time very stable) 3% shift on EEV, and flat line of evap SH (0.1C resolution) I

Gary
24-08-2011, 06:08 AM
Have you tried increasing the setting on the constant pressure valve? Presumably, this would increase the ejector dP, which would increase the motive flow and thus the suction flow as well. There may be some minimum dP needed for the ejector to function properly.

Gary
24-08-2011, 06:15 AM
I'm wondering too how oil return might affect the ejector operation.

mad fridgie
24-08-2011, 06:52 AM
Now your thinking!

Yes increasing the pressure in the vessel, would seem to indicate more motive force (greater pressure difference), but on the other side you will have reduced mass (less flash gas produced).

I have played with increasing pressures, decreasing pressures, changing orifice size in ejector. (present on are US made air movers).

I am sure there is an issue with ejector, but I do not have enough experience or knowledge to evaluate the ejector performance correctly. The ones chosen, and one I made maybe completly the wrong style and/or size.

If I could just of got about 30% of what I predicted (repeatable) I would have been well happy, and could of got investor or passed onto to those better well funded.

mikeref
24-08-2011, 07:36 AM
Mad, since this device has proven efficiency increase, than results on further tests can be duplicated. What has changed? Would it be the increase load on evaporator that causes increased flow through the first stage temperature reduction and maybe reducing saturated vapour available? Or vapour that you are harvesting is being redirected at times and continuing on to evap. TX valve. Superheat readings would pickup that problem. Would pressure stray too low and TX fail to function, after all, conventional expansion valves do not expect cold, low pressure liquid going through them now do they? Your thoughts Gary?

mad fridgie
24-08-2011, 07:57 AM
I would not say the device is proven, as it has not been 100% repeatable, even though some test have and others have not.
When testing you do your best to keep all variables constant. (water flow through the evap at a fixed inlet temperature)
You are correct about TEVs they do have limitations (for large application swings), for this reason I chose a carel EEV (sized to suit all possible pressure drops and duties)
I have sight glasses in the liquid lines, it would seem there is no bubbles when running constant. (sight glass is an aid)

chillerman2006
24-08-2011, 07:57 AM
That is not an unresonable question.
Direct measurment and understanding the results, only mean something when, when you compare to knowledge you already have. (you know how fridge works no doubt, so any changes in system can be easily evaluated)

Now it comes to the rigg, this has been built to operated as a standard sytem, or full widget system and all parts inbetween. (all parts can be isolated and by-passed).

All flows/temp can be set (water evap and cond), pressures can be maintained, so can superheat and subcooliing.

I can not measure refrigerant mass directly (i do not have the quality of equipment required)

But temps, pressures, water flows are well looked after.

Issues, because we are dealing with vapours and liquid at saturation it very difficult to know, if the streams are 100% seperated (and this can make a hugh difference).

and the TVR, ejector(s) I have zero data on, when asked your country men, "blank", so best GUESS was made.
This guess was baseed upon what if there was a failure of the widget, would the system still operate in some form of fashion and what was the likely hood of damage to the other component within the system.

Others issue can include changes in heat transfer properties of the heat exchangers.

All i was after with rigg is to show some regular improvent (not oerfect but enough) so the the technology could be sold to those who could develope further and bring to market. And here we are today!

MF

I know your trying to keep the system simple but

what about liquid line heat exchanger after your vessel, that the evap outlet runs through

would this give you superheated vapour to the widget & make a better mix going back to the compressor and also subcool the liquid to evap, further increasing efficiency ???

mad fridgie
24-08-2011, 08:10 AM
MF

I know your trying to keep the system simple but

what about liquid line heat exchanger after your vessel, that the evap outlet runs through

would this give you superheated vapour to the widget & make a better mix going back to the compressor and also subcool the liquid to evap, further increasing efficiency ???

Well that is good question, but for the purpose of calculations I have assumed that all is at saturation, because I thought this maybe an easy starting point, I may be wrong!

Adding excess superheat to the compressor reduces the possible mass flow through the compressor.

Superheat will be needed at the compressor inlet, so depending upon what happens when the stream joins determines where the superheat is to be added.

On the rigg superheat was at the evaporator (as i am trying to replicate a standard system)

I could very easily give the impression of proof (i have even fooled myself), but thats not my game, i only want to give genuine proof.

chillerman2006
24-08-2011, 08:22 AM
Well that is good question, but for the purpose of calculations I have assumed that all is at saturation, because I thought this maybe an easy starting point, I may be wrong!

Adding excess superheat to the compressor reduces the possible mass flow through the compressor.

Superheat will be needed at the compressor inlet, so depending upon what happens when the stream joins determines where the superheat is to be added.

On the rigg superheat was at the evaporator (as i am trying to replicate a standard system)

I could very easily give the impression of proof (i have even fooled myself), but thats not my game, i only want to give genuine proof.

MF

I was thinking the other way, good superheat from heat exchanger to balance out too much % of saturation from the vapour coming from the vessel

I see the problem now, It's knowing exactly what condition the vapour is at first - all chicken & egg or vice versa

& wind up would have been more polite ;)

mad fridgie
24-08-2011, 08:34 AM
if i was to make a guess (and i am the only one who can on the rigg) and if the concept of the process is correct.
Then likely issues are;
1; on start up some form of backwash is being produced in the ejector, (not flowing where it should within the chamber itself)
2; liquid carry over in the motive feed into the ejector (vapour from the vessel), this also reduces the mass flow to the evap.
3. Heat transfer coefficient is reduced in heat exchangers (reducing entry flash gas to evap, reduces turbulence, reduction in heat transfer)

Gary
24-08-2011, 02:45 PM
I would suggest that placing the EEV sensor at the ejector exit might better regulate the flow proportions... then adjust the AEV for optimum performance.

mad fridgie
24-08-2011, 11:05 PM
I would suggest that placing the EEV sensor at the ejector exit might better regulate the flow proportions... then adjust the AEV for optimum performance.

Valid point.
A couple of considerations.
As you touched upon on the other thread, as we expand the gas in the ejector a cooling effect will occur, on the other side the suction gas should be being compressed so "heating up" (all working well of course) so it would be difficult to know what to set at (apart from compressor protection), but not a silly idea.
second, by changing this we have moved away from the standard design, so results could look to be improved/or derated, only due to more normal factors, for example the evap could become fully wetted increase evap duty.
I can make it look as if it is doing more than anormal system, but it is actually small changes in the system itself that is improving performance, not always the widget. This is an issue I have always had to consider during testing. Many energy saving devices promote massive savings, when infact they are almost a gimick, improvements they portray tend to be on very poorly designed systems in the first place. I do not want this to be one of them, if this ia a failure then so be, but let it be a failure on merit, not a success on Bull Shyte!

al
24-08-2011, 11:40 PM
MF et al

Great thread, can't say i understand much of it but hey!

Could the fwidget (mfwidget silent m) be used to ultimately allow the evap run flooded, could it be the expansion device?

Reason for this is purely commercial, if aimed at smaller condensing units (1hp up), cost is going to be a major issue, i wonder could we use outsourcing between the members to progress this?

alec

mad fridgie
24-08-2011, 11:56 PM
Hi alec, during the concept process, the commercial side has always been a consideration. so has to be priced to meet the market. I believe that it would be well priced.

If sucesseful, then you would save a little on the compressor as it would be smaller) the vessel could be used as the liquid reciever (saving a bit)

There is the possibilty the evap could run fully wetted (not quite flooded) however gary showed a system which was very similar to what you are describing. When the ejector, is sort of used a pump/mover (inbetween a pump recirc and thermosyphon system)

al
25-08-2011, 12:16 AM
That would definitely be a way to market, any way for us to assist with your testing, see where the bugs lie?

Can the ejector be machined, i wonder would a machine shop make a range of sizes for you, as you say above it could be that you are using one designed for air, would one specific to 404 be smaller(hole wise?)

Alec

Gary
25-08-2011, 12:43 AM
Valid point.
A couple of considerations.
As you touched upon on the other thread, as we expand the gas in the ejector a cooling effect will occur, on the other side the suction gas should be being compressed so "heating up" (all working well of course) so it would be difficult to know what to set at (apart from compressor protection), but not a silly idea.
second, by changing this we have moved away from the standard design, so results could look to be improved/or derated, only due to more normal factors, for example the evap could become fully wetted increase evap duty.
I can make it look as if it is doing more than anormal system, but it is actually small changes in the system itself that is improving performance, not always the widget. This is an issue I have always had to consider during testing. Many energy saving devices promote massive savings, when infact they are almost a gimick, improvements they portray tend to be on very poorly designed systems in the first place. I do not want this to be one of them, if this ia a failure then so be, but let it be a failure on merit, not a success on Bull Shyte!

If nothing else, it may help identify problems... and if it can help determine the optimum AEV setting, that would be a big plus. As to superheat setting, I would set for compressor protection.

Wet evap? Let the EEV tell us if that's what is needed... or not.

mad fridgie
25-08-2011, 01:06 AM
That would definitely be a way to market, any way for us to assist with your testing, see where the bugs lie?

Can the ejector be machined, i wonder would a machine shop make a range of sizes for you, as you say above it could be that you are using one designed for air, would one specific to 404 be smaller(hole wise?)

Alec
The ejector/TVR?condell mover, would be very easy to make in a machine shop. Once made a CNC machine would pump out very quickly.
I would like to say I could give you design for one of these thing, but unfortunately not. very likely an airo engineering could do in his sleep. (I just made best guess,)
It then becomes a design for perfect conditions and then we bring in reality (all the other conditions) and see what happens, make some comprimises to ensure average good performance. Finding that balance wil lalso be an art.
Once the RD work is done then I do not see any great issues, it would just be like selectiong an expansion valve.
But of course the principle has to be proven without doubt (which I am pretty sure the concept is right).

Gary
25-08-2011, 03:15 PM
I haven't been able to find anything on Condell air mover. I'm wondering if you might be referring to a Coanda air mover?

http://www.laesieworks.com/ifo/how/pict/air-amplifier.gif

mad fridgie
25-08-2011, 10:13 PM
I haven't been able to find anything on Condell air mover. I'm wondering if you might be referring to a Coanda air mover?

http://www.laesieworks.com/ifo/how/pict/air-amplifier.gif

Thanks Gary, you are correct, I apolgise to all, not meant to mislead.

cheers

mad

Gary
26-08-2011, 04:28 AM
In looking at the diagram, it occurs to me that there could be substantial heat transfer (by conduction) between the two flow streams, within the ejector. The motive flow stream being near or at saturation, there could be condensation before the slit inlet followed by rapid flashing at the slit outlet. This could cause turbulence, which is the enemy of coanda effect. A solution might be to increase superheat in one or both flow streams.

I would heat the top of the separator. It would seem preferable to increase the superheat of the motive flow and then balance this against lower evap superheat, resulting in optimum compressor inlet superheat.

mad fridgie
26-08-2011, 05:52 AM
Intersting is it not.
gary you have sort of hit an issue on its head, when dealing with gas/liquids at saturation, it is very difficult to determine if for example that the vapour is just that, and not a vapour with a small trace of liquid in it.
On the two purchased units, with motive force is introduced on the circumference, yes larger surface area, and I do not think your suggestion should be discounted. (even though during some of the test , heat was added to the motive line to ensure liquid was not present) If we used a liquid line to motive heat exchanger then we slightly reduce the flash gas being produced. (the energy remains neutral in the system), if we use an external force, then we are adding energy to the system, without benefiting the system directly.
The one I built was more like an industrial TVR, single nozzle in the centre of the body, very little heat exchange would be possible. (these tend to be used for higher compression, but less suction flow) The hand build one I just guessed. (did seem OK until i tried to make it better) These things really do need science behind them, not just for perfection (ideal refrig conditions) but to understand what happens when we move away from ideal, which we all know is the absolute reality with small to meadium systems.
Thanks gary for your thoughts, always valued.

Gary
26-08-2011, 04:27 PM
The fact that it sometimes works tells us that the flow geometry is in the ballpark, so it comes down to the condition. proportions and overall volume of the two flow streams. What is needed is a step by step means of determining what the optimum conditions are and adjusting the flow stream variables.

It seems to me that the intent of the widget is to induce a pressure difference between the evap outlet and the ejector outlet, so perhaps this dP should be viewed as the measure of success, and the variables adjusted to whatever values gives us the best dP.

I would start with both flow streams highly superheated... then adjust the AEV to whatever yields best dP... then the EEV... tweaking each in turn to determine optimum values.

Adding outside heat may not be directly beneficial to the system, but it may be beneficial in troubleshooting the system... and resolving the repeatability issues.

Gary
26-08-2011, 05:24 PM
Gary, going to be some changes to your weather real soon. Battern down the hatches time.:off topic:

Sometimes it pays to be on the Gulf coast rather than the Atlantic coast. Just another day in paradise... so far.

Tesla
27-08-2011, 03:44 AM
Hi MF
Please excuse my late entry - I have not long returned from overseas travel.
Gary is definitely on track. Perhaps an adjustable ejector is required for the dynamics of the system for prototyping. This would allow for faster system stable state and varying loads. It would enable the sweet spot to be found. I know this will complicate the system and introduce a dual control loop. Understanding dynamic fluid flow and what is happening with all three dimensions (mechanically) and liquid/vapor phases. Viktor Schauberger's work in Living Energies by Callum Coats covers fluid flow dynamics with the effects of geometry and temperature in simple terms. Roy dossat also covers how fluid flows and shows calculations.
The concept as I understand it is achieving energy savings by increasing the potential energy transfer through the evap by precooling the refrigerant which reduces flash gas at start of evap. So thinking about it, the fluid in the evap is moving with a certain type of flow, we know the turbulence is reduced at the start, liquid would be swirling around the out side of the tube with vapor forming in the center of the tube more towards the end. The return bends in evap would reintroduce a more turbulent flow.
We could use a calculation of what might be happening at the ejector where the two gasses meet. We have 31% mass flow through the ejector with the remainder from the evap. This leaves the velocity as a variable and the quality of the refrigerant mixes. What are your thoughts on this - have I gone off an a tangent?

mad fridgie
27-08-2011, 05:41 AM
Hi MF
Please excuse my late entry - I have not long returned from overseas travel.
Gary is definitely on track. Perhaps an adjustable ejector is required for the dynamics of the system for prototyping. This would allow for faster system stable state and varying loads. It would enable the sweet spot to be found. I know this will complicate the system and introduce a dual control loop. Understanding dynamic fluid flow and what is happening with all three dimensions (mechanically) and liquid/vapor phases. Viktor Schauberger's work in Living Energies by Callum Coats covers fluid flow dynamics with the effects of geometry and temperature in simple terms. Roy dossat also covers how fluid flows and shows calculations.
The concept as I understand it is achieving energy savings by increasing the potential energy transfer through the evap by precooling the refrigerant which reduces flash gas at start of evap. So thinking about it, the fluid in the evap is moving with a certain type of flow, we know the turbulence is reduced at the start, liquid would be swirling around the out side of the tube with vapor forming in the center of the tube more towards the end. The return bends in evap would reintroduce a more turbulent flow.
We could use a calculation of what might be happening at the ejector where the two gasses meet. We have 31% mass flow through the ejector with the remainder from the evap. This leaves the velocity as a variable and the quality of the refrigerant mixes. What are your thoughts on this - have I gone off an a tangent?

The evap issue, is to be considered but later, (but in short, the reduction in flash gas can detrate the evap or it can also increase the evap duty. but this is a story on its own)
Try to to think that the system build up to its performance.
Keeping it simple, the prime objective is to increase the inlet pressure to the compressor (higher than the evap inlet)
This in its self would increase the comp mass flow.
It has been presumed that I have choosen the correct ejector (one was adjustable), but there are many different tpye styles and designs. So even though garys suggestions are extermly valid (as one would expect), I believe that i fundementally chosen the wrong type, but the only ones I could find, that would seem to suit my purpose. But with help forms other (thanks to you all) the supply of others may be easier, or to the point that a design for manufacture maybe another option.
Like most things is lack of riskable resources that has stopped the R&D process.
(if I was looking at many megawatt systems then there are standard units avaliable, then would best suit the process, normally used in steam systems)

Gary
27-08-2011, 02:53 PM
The evap issue, is to be considered but later, (but in short, the reduction in flash gas can detrate the evap or it can also increase the evap duty. but this is a story on its own)
Try to to think that the system build up to its performance.
Keeping it simple, the prime objective is to increase the inlet pressure to the compressor (higher than the evap inlet)


Targeting the compressor inlet pressure versus evap inlet pressure adds all of the evap variables to confuse the issue. Targeting compressor inlet pressure versus evap outlet pressure narrows the focus down to the ejector performance, making ejector troubleshooting/adjustment easier.

What factors would cause the system to build up to it's performance?

Gary
27-08-2011, 05:46 PM
In reading through various papers, it is my (possibly flawed) understanding that high motive/low suction is more stable, but less productive... while low motive/high suction is more productive, but less stable. Seems there needs to be some form of flow proportion control for optimum operation.

I'm thinking this might be achieved by setting the AEV to a fixed optimum value and varying the evap flow for fluctuating conditions. This would require the EEV to sense and control evap flow, thus flow proportions, according to ejector output.

Gary
27-08-2011, 06:37 PM
Hmmm... in addition to everything else, it seems there is a maximum discharge pressure, which when exceeded causes the system to become unstable, resulting in backflow through the ejector.

mad fridgie
27-08-2011, 10:57 PM
The issues with the ejector are many fold, as you have noted.
It would seem that there are possible designs, that may not have an optium compared to others, but are more consistent over a range? which would be required to a refrigeration system. If the widget could be built is has to be easy to install and commssion. I am practical and was looking at achieving a % of peak performance, the 80-20 rule.
It seems that you have a feel for the system, in your opinion, without worring about the details "is the fundemental concept correct"

thanks for your input

Mad

Gary
28-08-2011, 05:21 AM
The issues with the ejector are many fold, as you have noted.
It would seem that there are possible designs, that may not have an optium compared to others, but are more consistent over a range? which would be required to a refrigeration system. If the widget could be built is has to be easy to install and commssion. I am practical and was looking at achieving a % of peak performance, the 80-20 rule.
It seems that you have a feel for the system, in your opinion, without worring about the details "is the fundemental concept correct"

thanks for your input

Mad

Induced flow is a proven concept. Your particular setup is not yet proven. You say you have had positive results and I believe you... but those results need to be made stable and repeatable.

desA
28-08-2011, 11:31 AM
I suspect that the system may be flipping between multiple steady, or quasi-steady operational points.

If an additional 'controlled' motive gas were to be injected into the ejector e.g. via hot-gas bypass, until system begins to stabilise in the required flow directions - then this HGB is slowly backed off until system stabilises as desired.

A few thoughts for you to ponder MF.

mad fridgie
28-08-2011, 12:03 PM
I suspect that the system may be flipping between multiple steady, or quasi-steady operational points.

If an additional 'controlled' motive gas were to be injected into the ejector e.g. via hot-gas bypass, until system begins to stabilise in the required flow directions - then this HGB is slowly backed off until system stabilises as desired.

A few thoughts for you to ponder MF.


The flipping of the system is i think a dead cert at the moment.

I do have a number of by-passes, but not the HGB.

This would be stealing some mass flow from condensor outlet, but should not be disregarded as a method of achieving some of of stabilty.

I am sure that the concept has merit, but need to ensure that we have a theoretical positive steady result, then look what a happens when we move away from design.

When this proved beyound a doubt as theorectical possibilty, then a descion has to be made about employing a specailist(s) for the design of individual components, which a can be applied to a test rig.

I believe that the present rigg has a possiblity of proving concept, but I think that the proof point is a very critical point (which could take for ever to find) "I am feeling lucky punk" NO

Gary
28-08-2011, 09:17 PM
I'm wondering if there is sufficient liquid pressure to feed the evap coil?

mad fridgie
29-08-2011, 08:05 AM
I'm wondering if there is sufficient liquid pressure to feed the evap coil?
This had to be considered, when designing the rigg, I chose a large EEV!
I am able to control the superheat right down to 1.5 bar difference, then the evap does become starved.
How ever heat transfer co-efficients may be effected. (how ever I did attempt to simulate this during standard tests)
Thanks again gary for your thoughts

chillerman2006
29-08-2011, 08:42 AM
In looking at the diagram, it occurs to me that there could be substantial heat transfer (by conduction) between the two flow streams, within the ejector. The motive flow stream being near or at saturation, there could be condensation before the slit inlet followed by rapid flashing at the slit outlet. This could cause turbulence, which is the enemy of coanda effect. A solution might be to increase superheat in one or both flow streams.

I would heat the top of the separator. It would seem preferable to increase the superheat of the motive flow and then balance this against lower evap superheat, resulting in optimum compressor inlet superheat.


Intersting is it not.
gary you have sort of hit an issue on its head, when dealing with gas/liquids at saturation, it is very difficult to determine if for example that the vapour is just that, and not a vapour with a small trace of liquid in it.
On the two purchased units, with motive force is introduced on the circumference, yes larger surface area, and I do not think your suggestion should be discounted. (even though during some of the test , heat was added to the motive line to ensure liquid was not present) If we used a liquid line to motive heat exchanger then we slightly reduce the flash gas being produced. (the energy remains neutral in the system), if we use an external force, then we are adding energy to the system, without benefiting the system directly.
The one I built was more like an industrial TVR, single nozzle in the centre of the body, very little heat exchange would be possible. (these tend to be used for higher compression, but less suction flow) The hand build one I just guessed. (did seem OK until i tried to make it better) These things really do need science behind them, not just for perfection (ideal refrig conditions) but to understand what happens when we move away from ideal, which we all know is the absolute reality with small to meadium systems.
Thanks gary for your thoughts, always valued.

combining your thoughts here gents, is it possible to run the discharge pipe & seperator outlet through a small heat exchanger, ensuring the motive force is just vapour & maybe then allowing for a lower superheat evap outlet ???

Gary
30-08-2011, 04:39 AM
combining your thoughts here gents, is it possible to run the discharge pipe & seperator outlet through a small heat exchanger, ensuring the motive force is just vapour & maybe then allowing for a lower superheat evap outlet ???

Hmmm... you have pumped heat from the low side to the high side. Now you want to transfer that heat back to the low side... to be pumped again. Doesn't sound very efficient.

chillerman2006
30-08-2011, 01:14 PM
Hmmm... you have pumped heat from the low side to the high side. Now you want to transfer that heat back to the low side... to be pumped again. Doesn't sound very efficient.

Hi Gary

liquid injection agreed, but the minimal discharge superheat required prior to condensor to ensure no liquid to fwidget, surely makes sense, the problem I have is I can not get my head around the quasistatic process required to fathom equilibrium, best left to the real brains of the industry, as not many will, unless you can do us the honour of explaining the process in detail with all your calcs,

R's CM

Gary
30-08-2011, 02:38 PM
Calcs? You are confusing me with someone who does math. :confused:

chillerman2006
30-08-2011, 03:30 PM
Ok, mate we shall have to hope Chef, DesA or Peter_1 does us the honour of explaining it one day & you can add that to the other 99.9999% of the industry you already have tucked away, thanks anyway mate

Tesla
31-08-2011, 08:56 AM
Hi mf
Just thinking outside the square - is it possible to calculate roughly when it is not working? with graph gradients or measurements verses time. Say on start up it could be that the sweet spot is not found till after x seconds and then check for additional times. There may be some common parameter values at start up which equal definite failure of experiment. You could be looking at a resonant response for success.

mad fridgie
31-08-2011, 10:09 AM
All tests were completed on a 20 min cycle, with only one change per made at any single time.
So about 12 tests a day.
Plus short term tests when a major change is made. (some times you can see when things are just wrong)
As with any refrig system, time is taken to reach equilibrium, even though this section was ignored for the performance results, it was all ways monitored.

Good thoughts

Tesla

CM; your point also has merit, but as Gary says we are introducing extra energy, what you should of considered is using the liquid line (how ever this has its own issues "less flash gas"

Individual parts can be made to work with satisfaction, but refrigeration is not a group of parts, it is a system!

chillerman2006
31-08-2011, 08:43 PM
Post removed - as again have miss-understood...not a designer...just mr nuts & bolts !
:)