Having developed a degree of confidence with copper brazing, and reaching the potential limits of what can be learned, I have decided to move past my original work with the Ebullator. I document my progress with that device in Refrigeration Test Bench I, RTB II, and RTB III.
Ejector refrigeration devices were not something I had planned on messing with for some time, if ever. I spend a great deal of time at my work (truck driver in Portland) thinking about how I would like to proceed with my hobby. At times I run through hundreds of iterations of a particular device which I’m currently working on, and other times I will theorize on fanciful devices which may improve performance in a vapor compression system, but may never be explored further. A great tool in my mental kit, is the log(pressure) – enthalpy diagram. Other tools include factors like entropy, or more usefully: exergy.
While attempting to design a device which would preserve some of the lost exergy in the expansion of high pressure liquid refrigerant passing through a metering device into the low pressure portion of a refrigeration system, I considered the expanders used in some air refrigeration systems I’ve read about, which recover the potential energy of a a compressed gas, and return it to the compressor, generally as shaft work. The potential difficulties of such a device, especially concerning a very small application, are many, perhaps the worst being the small change in specific volume a refrigerant undergoes when it is lowered in pressure while remaining in the liquid state. A reciprocating expander is out of the question, and a turbine is quite simply beyond my technical abilities.
It was in quite a roundabout way, that I began to understand the way in which an ejector functions, or that it ever existed in the first place. I did not believe there was enough useful work to be recovered from the expansion process in a small tonnage refrigeration system to be of any use to perform the compression stage. Quite accidentally, and due to a degree of laziness on my part, I was attempting to simplify the construction of my next flooded evaporator design when I stumbled upon an alternative viewpoint. The liquid / gas separator column in my finned coil flooded evaporator, was going to have a number of copper lines brazed into it, and while designing one day in my truck, I realized that two of the lines may be simply combined as one while entering the separator column. These were the overflow from the evaporator, and the inlet from the expansion device. Having simplified my plumbing, it had occurred to me that these two refrigerant streams would likely be mixing at different velocities, with the expanding high pressure (lowered in pressure, but gaining velocity) refrigerant moving at much greater speed may impart some of that kinetic energy to the slower saturated refrigerant coming from the evaporator. At that moment I thought that if designed right, this thing could be made to pump, and therefore do some useful work! I wanted to perform this pumping as efficiently as possible, so I began to study the design of venturis on my Android. Further thought, and many iterations later, I believed that not only could I make it perform some pumping, but I could actually design it in such a way as to produce a small level of compression, before the main mechanical compressor!
Several more iterations, and I honestly thought this crazy thing might work. There were many questions remaining, but one that nagged at me, was whether or not this has been done before, because it just seemed like a really good idea. Within five minutes of internet searching when I got home from work, I found something like this:
Damn it – already been done. This was a moment of disappointment, and excitement. On one hand, I was a little bummed that it wasn’t purely my idea, but on the other hand, it’s pretty cool that I came up with something that apparently works. What’s more, is that this diagram (or one like it) was almost exactly what I envisioned in my head. Maybe I had seen it before, maybe not. What’s important is that it’s a really neat device.
What the ejector system does, is basically add two more components to a simple vapor compression machine: an ejector (or jet pump), and a liquid / gas separator.
Cool vapor, near saturation, is drawn into the compressor. Work is done on the vapor, reducing it in volume, and raising the pressure to that of the condenser. The now hot gas is discharged into the high side where it enters the condenser (or gas cooler), where heat is rejected, causing the vapor to condense. The pumping of the compressor forces the condensed liquid into the primary motive nozzle of the ejector, where the cross sectional area is reduced, causing the pressure to drop, and the velocity of the liquid to increase. Potential energy (higher pressure) is converted into velocity at lower pressure. This nozzle could be a fixed orifice, or a variable metering device. The high velocity liquid (with some flash gas, due to the temperature drop) is is discharged into the throat of a venturi device. The primary motive nozzle is generally mounted inside of a larger pipe that will lead up to the convergent nozzle of the venturi tube. Inside the annulus, vapor from the evaporator can be “drawn”, which will be mixed with the refrigerant discharging from the motive nozzle. As this evaporator vapor moves down the annulus, its velocity increases until it enters the convergent section where the cross sectional area decreases, and the velocity increases higher yet. Inside the throat of the venturi device, the two refrigerant streams mix at low pressure. The higher velocity stream from the motive nozzle gives up some kinetic energy to the vapor entering from the evaporator. Once these two streams are thoroughly mixed, they enter the divergent portion of the venturi. The cross sectional area widens, and the combined velocity of the stream slows, resulting in increased pressure.
The exiting stream of refrigerant is a saturated vapor / liquid mix, with a pressure higher than that of the evaporator, but lower than the high pressure liquid entering the motive nozzle. This medium pressure mix then enters a separation tank where the liquid portion can settle to the bottom, and the vapor be drawn back to the compressor. Due to the slightly higher pressure in the separator, liquid is forced through a metering device into the evaporator. This control probably has a larger orifice than the main refrigerant control, due to the relatively smaller pressure difference between the separator and the evaporator. Pressure drops across the throttle, liquid circulates through the evaporator, picks up heat, changes to a vapor phase, and due to the lower pressure in the ejector, it accelerates into the annulus of the ejector assembly, where starts the cycle again.
As a result of the added kinetic energy and increased velocity of the vapor stream sourced from the evaporator, the velocity drops and pressure rises, (kinetic energy to potential energy) produces a medium pressure, higher than the evaporator. What we have is compression.
In the basic vapor compression cycle, high pressure liquid passing through an orifice gains velocity as the pressure drops. Exiting into the evaporator (or separator in a flooded system), smooth laminar flow quickly becomes turbulent. Some of the kinetic energy is converted into internal energy, resulting in a portion of the liquid flashing to vapor. Overall, enthalpy remains constant, but entropy rises. In other words, work (or exergy) is lost.
What makes the ejector system different in some regards, is that some of this kinetic energy is used to accelerate evaporator vapor, and thus do useful work. As a result, suction pressure is higher than it would otherwise be at the desired evaporator temperature. Higher suction pressure means smaller compression ratios, lowered specific volume at intake, higher volumetric efficiency, increased mass flow given the power input, and an overall better coefficient of performance. Ejectors are sweet. 🙂
There are a great number of considerations to be made when designing an ejector, such as the mass flow rate through the evaporator, the velocity through the motive nozzle, the angles of the convergent and divergent cone, the length and diameter of the throat, not to mention the manufacturing process required to build an ejector. As of this date, I have built and tested a small ejector, and I’m happy to say it does in fact work. In the next article, I will discuss the process of constructing my ejector, as well as all of the physical iterations required to get there. Check out: Handmade Copper Ejectors, and Ejector Assembly Construction
-M.C. Pletcher
musings on entropy 
Pretty cool you figured that out by yourself!! Kudos!
I learned about this jet pump principle in relation to hydraulics – as an alternative to traditional mixing circuits (in case you have a high pressure / high temperature ‘source’, like a central heating system in an apartment building, and a low pressure heater, like the radiators in a flat).
My impression was it’s one of those clever ideas everybody appreciates from the technical perspective, but it is more difficult to implement due to missing standards, guidelines for sizing etc. though sizing is more important and more critical than with conventional hydraulics. But it seems it has become popular in Germany in the last 10 years.
I don’t know about its use in refrigerant cycles though – I would expect that it is hard to justify the additional costs for small systems. Re hydraulics of hot water circuits it is easier to argue as the jet pump replaces a classical pump with a lower efficiency.
Wow! You keep a large web presence: many sites!
Thank you for the kind words. I’m not sure if I figured much out, or if I just got lucky. I think that I was aware of mixing circuits for fluids like water in pressure washers, and paint in pneumatic sprayers, but the application of an ejector in refrigeration was foreign to me. I had to mention my thought process, because it represents so much of my waking hours, and independently inventing something like this (even on the theoretical level), was a real boost to my confidence.
There are many components which are rarely seen in small systems, for that very reason of economics, I’m sure. A manufacturer has competing priorities, starting with profit above all, staying competitive, environmental standards, product life expectancy, and efficiency. In my view, these machines we use should last a very long time, be easily serviced, and have components which are cheaply replaced. Developing a component which has no moving parts, and increases efficiency (even if only marginally), should be a no-brainer.
Making it work is another matter. ;P It is true that a refrigerant ejector (or jet pump) is not a simple project to undertake, although I haven’t had much trouble demonstrating its operation, I have yet to show it is effective at increasing efficiency. I’m doing a lot of fumbling around and guess work, but it is surely a great deal of fun!
I hope to implement some sort of ejector into the heat pump I’m building this fall. It is one thing to design it to work, it is another to make it work in varying conditions!
Whatever may come of it, the journey is fascinating and offers many lessons.
[…] work I have been doing with ejector and flooded evaporator is much too valuable (and so much fun), that I can’t let the dust […]