Hacker News new | past | comments | ask | show | jobs | submit login
A centrifugal spin on a 500-year-old air compressor design (newatlas.com)
225 points by joshwa 37 days ago | hide | past | favorite | 61 comments

Interesting. A few questions that come to mind:

1. What is the dew point of the compressed air? Many applications using compressed air require very dry air so as to avoid corrosion issues (e.g., air tools) as well as problems related to heating of water vapor (e.g., tires). For small applications, the dew point is reduced using simple traps, silica gel beads, or molecular sieves, but for even-drier air or larger applications, refrigeration-based dehumidifiers are needed. If the use of water in the compression stage leaves the dew point significantly higher than what you get from reciprocating or Roots-type compressors, that added complexity could be a deterrent to adoption.

2. What's the energy efficiency of the process? Does the claimed "20% lower cost of ownership" account for the difference in efficiency (whatever it is)?

3. What sort of pressures can be achieved with a tabletop or shop-sized compressor? Presumably, this is related to the speed at which the drum is spun, but what are the practical limits?

I mean, I use a decent amount of very dry air at 300 bar (yeah...), and I'd love to replace the noisy beast of multi-stage compressor with something quieter and more efficient.

Speaking as someone who uses compressors both for shops and breathing air - this is mostly non-interesting so far. For starters, 70db is actually pretty horrible.

Indoor Scroll Compressors are around 47db (IE 100-1000 times quieter than the 70db mentioned here). They are completely oil-free and produce class 1/class 0 air. That is, you can use them to breathe from They are not as energy efficient as rotary screws, but it's only off by about 20%.

In practice, the flow of air into a tank produces way more noise than the compressor itself with these.

Even rotary screws are around 60-65db, and there are both oil and water cooled versions. They are very energy efficient, and due to the design, can be put on VFD's and modulate air-demand incredibly well.

Oil carryover is typically <3ppm, and you can use oils meant for incidental food contact/etc or filter it down further.

Both the above have 100% duty cycles. They can be on continuously (and in fact, oil-flooded rotary screws get unhappy if they are used too little).

Without more, or at least, based on what this article says, it's really hard to see what part of the market this thing will inhabit. They also don't talk at all about how to modulate them, what their duty cycle is, etc. Nor the dewpoint, etc.

The trompe article goes into a bit more detail that might be of interest. You apparently modulate a trompe by changing the height of the intake and to some degree it’s diameter. I would assume that you modulate the Carnot device by changing the speed of the drum.

I am no expert here so I am making the assumption that perhaps the innovative part here has to do with using fewer moving parts than existent solutions, and potentially naturally producing cool dry compressed air without any oil mixed in. The idea of fewer parts/nothing but the air filter, belt, fan, maybe drum bearing to ever replace I’d quite appealing to someone who only has access to a pancake compressor that’s loud as all hell.

Yeah, this really does not sound like an energy efficient way to modulate down.

It's similar to some of the ways you see here: https://cascousa.com/compressed-air-101/types-of-compressors...

As for fewer moving parts and less maintenance, probably not gonna win on that. The state you want already exists. If you get a rotary screw, you are basically just going to replace oil and filters every so often until the air end dies. For a hobbyist, maybe never.

You might never run it for the 8000 hours the oil lasts or the filters last. If you ran it 3 hours a day, it would take 7 years :)

If you want simplicity and less moving parts, and the ability to fix them easily, you can get a rotary vane compressor.

They can't be modulated down as well using a VFD as you can a rotary screw (there is a minimal speed at which the vanes seal properly due to centrifugal force) , but they are ridiculously simple and have very few moving parts. They are also easily repairable.

Until it literally falls apart, you will just need to replace filters and oil every so often, and some carbon blades every so often (50k+ hours) or so of usage. The compressor itself will probably last 100k-200k hours, and so, again, for a hobbyist, may easily outlive you.

It's just a motor, connected to a slightly offset spinning circle with blades (so that as it rotates, the offset causes it to form progressively smaller chambers).

Looks like this: https://aircompressorworks.com/wp-content/uploads/2017/10/v-...

300 bar air.. Are you filling scuba tanks?

The wet air question is an interesting one. It seems like the compressed air would have more moisture in it than a piston-driven compressor, but at the same time having oil-contaminated air is problematic for moisture removal systems, especially dessicant-based ones. So you might need more thorough moisture removal, but with the dryer system having more life.

As far as efficiency goes, it's hard to tell because it's mostly a paper-company at this point and they don't have much in the way of product. Another press release mentions they can provide large volumes of air which usually translates to efficiency. The major source of work done in their system is to accelerate the water, so if you want more air, accelerate more water. It sounds pretty efficient to this armchair critic.

The other press release mentions they are targeting 10hp and 20hp systems, and 100hp systems are feasible. I'd imagine 10hp is where industrial compressors start.

[edit] Found a press release from 2014 where they claim better energy efficiency than conventional methods. This really does sound like a breakthrough.

If you look at the linked article in TFA that discusses the trompe it claims that the air that comes out is completely dehumidified. I’m not sure how that process works but I guess big, if true.

The work done accelerating the water doesn't seem to be recovered anywhere...?

I would assume that to get any kind of respectable efficiency the high pressure water needs to go through an energy recovery turbine before being cooled and reused.

I didn't realize that air compression hardware has to deal with water removal but apparently it does. One article I found is suggesting that air at 7 bar can hold about 1/3 as much water. What confused me is that the a wing stall (low pressure above a wing) causes cloud formation, due to vapor condensing out of the air.

But 7 bar is ~7 times as much air per cubic centimeter, so that's 14% as much space containing 33% as much water. The partial pressure doesn't drop at a 1:1 ratio.

As a former owner of many water management devices from pet fountains to humidifiers to dehumidifiers, I'm more concerned about the fact that air is full of particulates, and you're going to wash the particulates through a fluid that stays in the pump permanently, so how quickly is the funk going to make that compressor unpleasant to be around?

Most industrial compressors (IE rotary screw, scroll, rotary vane, etc) have aftercoolers, because the temperature of compression is pretty high. The aftercooler causes most of the water to condense out.

So what you lose is efficiency, because it will take compressing more intake air to get the same yield of compressed air at a certain pressure/dewpoint.

(this is why you often see a lot of wet storage tanks for air - leaving it in the tank lets a lot of the intake air water condense at the bottom)

Aftercoolers remove a ton of of the water, and so you often get an 20 degree approach temperature.

Removing more than that requires some form of air dryer.

>you're going to wash the particulates through a fluid that stays in the pump permanently, so how quickly is the funk going to make that compressor unpleasant to be around?

Yeah, I was wondering about that too. But is the water going to be there permanently?

If the input air is quite dry, presumably (total guess) it would pick up some extra water. So you'd need to supply more. And probably not just tap water because of dissolved minerals and so forth - distilled seems like a better idea.

Conversely if the input air is humid, you might end up condensing water out of it, and need to bleed water from the system.

And then you're dissolving particulates in it either way...

All sounds like maintenance work, but maybe not a prohibitive amount of it.

Looks like there is a filter on the air intake, probably partially for that latter reason (larger particulates could be a problem mechanically also).

When you compress air, the moisture falls out, the amount that falls out depends on pressure and temperature. pretty much all normal compressors simply have a, I don't know the right english word, but.. a valve at the bottom and once in a while (like, once a year or less, depending how much you use it) you take the pressure mostly off the compressor and open this, then the water is blown out.

Fancy air compression systems (eg: not consumer) will have a drain valve that automatically opens when the pressure falls from turning system off for night/weekend/other maintenance so people don't forget to drain it.

Its also not just water that is collected at the bottom of the air tank, compressor oil also collects. Oil can also be deliberately added by either as a byproduct of compressor design or from separate equipment, into the airstream help lubricate and protect equipment from the moisture, reducing the need to remove the moisture through other means.

About the only thing practical thing that valve does is perhaps lengthen the time fir the tank to rust out. Any serious industrial installation winds up with a refrigerated water removal unit.

It also helps keep a slug of water from entering the air lines if employees (and management) puts off draining the accumulated water for too long. You can also get ones that are triggered by a float if the system is for 24/7 operation.

Any serious industrial installation will have multiple mitigation approaches.

Ours are on a timer to blow off a small amount of air at a few low spots in the system every five minutes, even though having a refrigeratant type demudifier.

One of the devices stopped working between services and the bottom reciever tank half filled with water.

Edit: they're call bleed valves

All I know is any compressed air tank you have needs water drained from it regularly. Air pipes count as a tank (just a small diameter) so you have to consider where the water will go.

I was wondering the same thing, re: dry (or not) air. I’ve spent a good bit of $ on a drying system for my setup.

Also: I’ve seen water vapor in the compressor ruin an otherwise nice paint job.

That and if ever accidentally you get water in your lines and you run piston based tools on them you'll end up with hydraulic lock and/or bits of your tools all over the place.

I'm certainly no expert, but as they pointed out this is basically a trompe, which...

>Compressed air from a trompe is at the temperature of the water, and its partial pressure of water vapor is that of the dewpoint of the water's temperature. If the water is cool, the compressed air can be made very dry by passing it through pipes that are warmer than the water. Often, ordinary outside air can warm the pipes enough to produce dry, cool compressed air.


So use a better heat exchanger to cool the water, and then heat up the air until it's dry enough, I guess?

A heat pump could be used to cool the water to below-ambient and inject that heat directly into the outgoing compressed air. I wonder if a system like that would be more efficient than installing a separate refrigeration-based system downstream?

> this is basically a trompe

Is it though? They don't really explain how it works or show any diagrams, and they said it was "inspired by" the trompe. So I'm a bit confused on that point.

It really looks more like a mashup of a centrifugal compressor and a diffusion pump. One where water instead of oil is used as the working fluid.

As I see it, the trompe has the problem of needing a lot of distance for gravity to operate it. This basically uses a centrifuge to increase the "gravity" and thus reduce the size.

Ah yup. I was misremembering what a tromp was--I was thinking one of those cyclic pumps that uses water hammer (but to compress air rather than pump water).

If there was liquid with vapor benign in your use case, you could substitute that. Also, since you're doing compression, you would have less water vapor than with something like a water aspirator (vacuum ejector, or Venturi effect vacuum pump.)

If the water is saturated with a salt, it could actually dehumidify the air, https://www.youtube.com/watch?v=R_g4nT4a28U

Right, but if the point of moisture reduction is corrosion resistance, salt is probably not the best option.

Similarly, using mercury as the rotating fluid might give you higher compression ratios, but then you'd have to deal with mercury vapor in the compressed air...

I suspect that Carnot is aiming to compete with liquid ring compressors https://en.wikipedia.org/wiki/Liquid-ring_pump, which have much of the same advantages and disadvantages (near isothermal compression, water saturation). I'm guessing the article didn't mention this because every home compressor user wants a less noisy one and they get better propagation of the article around the internet that way. I considered building a centrifugal trompe a while ago for a project of mine, mainly because its really hard to find sub-horsepower liquid ring compressors, but frankly my experience with DIY centrifuges has been explosively scary.

You can run a pump like this with kerosene instead of water, in order to reduce the amount of saturated vapor that comes out with the compressed air, but that can also cause certain flammability concerns. I once had one of my homemade vacuum pumps blow kerosene all over me when a gas bubble entrained a lot of fluid. This is one of those reasons I don't smoke - it can kill you.

"isothermal" is the reason this system is more efficient right?

But is it the theoretically most efficient way to compress air? Is there a (theoretical) way to compress the air adiabatic'ly (letting it heat up as you squeeze it), and do better overall?

Yes ... and kinda yes.

So, if you just want to reach a certain pressure regardless of temperature, you will put in less energy with adiabatic compression, because all the energy you put into the gas stays there. However, in many gas compression applications, the gas is not going to stay hot, but will cool to something closer to the environmental temperature before being used. For these cases, then the hot gas is going to lose some of that energy to the environment. If you compress it isothermally instead of adiabatically, then it still loses heat energy to the environment, but it loses it as it is being compressed rather than after the compression is done. This requires less work.

So, in short, if you want hot compressed gas (like a diesel engine) go adiabatic, but if you want room-temperature compressed gas (like compressed air which sits in a container), go isothermal.

Could you do it with mineral or tool oil?

In theory yes, but kerosene has a lower viscosity while still having a low vapor pressure. The low viscosity can be important when the liquid is moving a lot, to prevent losses and heating.

Grady over at Practical Engineering did nice a video[1] on the trompe, which they claim this design is inspired by. As usual with some nice history and a working prototype.

[1]: https://www.youtube.com/watch?v=uvf0lD5xzH0

Am I the only one twitching and reaching for Close Tab when the video labeled atmospheric air intake as "o2"?

It was probably a graphic designer who was going for the chemical formula aesthetic and wanted to be consistent with that and thought air = oxygen. N2 would've been acceptable but best would've been to just spell air and water outright.


The legacy design is called a trompe, here is a video of how it worked from Ragged Chutes Air plant near Cobalt, Ontario Canada. https://www.youtube.com/watch?v=UtYLVLkWyGc

From the diagram, this seems like it wouldn't be terribly hard to DIY: you can take an off the shelf heat exchanger and a fan, then all you'd need is the centrifuge part and tubing for the pressurized gas.

Would there be any special considerations in DIYing this that I'm overlooking? The only part that I can see being tricky is the centrifuge mechanism and sealing the bottom of that to a compressed air tube... not quite sure how to build that (presumably it'd have to withstand both high RPM and high pressure.)

Obviously a trompe is pretty easy to DIY if you have a steady stream of water and some serious elevation change, but the vast majority of places don't have that.

Building a DIY centrifuge is a great way to become a Darwin award candidate.

Eh, build it at small enough scale and encase it in metal and you'll be fine.

Very interesting machine. The target market is for food and other processes that require food grade ingredients or just very clean air. If they can get cleaner air with less power and maintanance then its a go. And the initial model is slated to be 10 to 20 hp ,so the size of a big deepfreez,hundreds of pounds. 100% duty cycle,air and water filtration,clean outs,drains,etc,all has to be assumed to fit into an industrial environment.

Air compressors need not be noisy.

Your fridge at home likely has a cylinder air compressor in it, pumping gas up to 30 bar (400 psi) and it only makes a quiet humming.

The trick is that air compressor is attached to a big vibration damping weight, mounted on big springs, which are mounted into an oil filled heavy steel box, which is itself mounted onto rubber feet.

> The trick is that air compressor is attached to a big vibration damping weight, mounted on big springs, which are mounted into an oil filled heavy steel box, which is itself mounted onto rubber feet

So easy... If it was that easy, it would have already been done, but...

1) compressor are maybe 100lb of solid cast steel

2) owners need to fix mechanical wear, not simply "replace the whole thing" - head gasket fails, gauges fails, hose leak (think thermal expansion / contraction), oil need to be changed, etc.

3) refrigerator system are sealed, air compressor need to pull air from the outside (ie. it can be DIRTY), which means lot of the noise actually come from the intake

Isn't the difference that a refrigerator is a closed system whereas many industrial/workshop air compressors are open and must maintain relatively high pressure due to a "controlled leak"? I am certainly no expert but they don't seem the same to me?

Cut a pipe on the back of a fridge and it'll run open circuit just fine. Doesn't make it appreciably louder. (beware, some fridge gasses are illegal to vent to the atmosphere, and others are pretty explosive)

> Cut a pipe on the back of a fridge and it'll run open circuit just fine

Try to run a plasma cutter, or a 1" impact wrench out of a fridge compressor...

Those things also don't need 400psi.

A modified fridge compressor could have 4x the capacity at 100psi and be suitable for a small workshop.

Do you understand the concept of air flow at a specific pressure ? 15CFM at 90psi is no small feat.

Lowering working pressure does not increase CFM, pump have a fixed displacement at their most efficient RPM, it's not gonna change...

The CFM rating of a fridge is nowhere close to the CFM requirement of industrial processes.

Scroll compressors are also known for being relatively quiet.

I think they need to do far better than 70db to make this viable compared to existing technology.

My home shop has a California air tools compressor in it that I chose specifically for its <70db volume. You can easily have a conversation when it’s running, and compared to my dust collector or other tools, it’s nearly silent.

I wonder if a pump like this could be made smaller using a perfluorocarbon liquid like FC-70 (Perfluorotripentylamine)—nearly double the density of water, higher boiling point, non-flammable, and would not impart water vapor to the compressed air.

I wouldn't necessarily be able to say whether it'd be smaller, but it'd probably work. It works due to the air getting entrained in the liquid as it increases in pressure. So there shouldn't be a requirement that the air dissolve into the fluid or anything like that.

A very common failure mode for machines that use water is freezing. No information for the eventuality of the water becoming ice. Maybe they just won't sell these in freezing climates.


some previous discussion: https://news.ycombinator.com/item?id=24760625

So by this same reasoning, adding a little water spray to the input of my air compressor, and a water seperator on the output, would reduce energy consumption since the water would absorb heat during the compression stroke making the compression closer to isothermal rather than adiabatic?

Do any commercial products do this?

You would have to demineralize the water first or you are going to get deposits building up on the warm parts of the compressor (most of it) where the water evaporates on contact.

It would seem to be better to have a closed loop water cooling system around the compressor head, but this is a very rare feature. So many home compressors treat cooling like "LOL 5% duty cycle for you", it's pretty annoying. Most of the time the cooling fans aren't even running unless the compressor is squeezing air, so the whole thing just soaks in its own heat once the tank is full.

I'm more thinking cooling should theoretically ~halve the energy required to compress air...

For some industrial applications, electricity used to compress air is substantial, so there might be big cost savings to be had.

Even if you are pumping up bicycle tyres by hand there are benefits to cooling - by keeping the air cool, the same human muscle power can pump up twice as many tyres.

Isnt this how water ring vacuum pumps work, used for vacuum generation, instead of air compression?

Can this model be used in a closed loop, with a different fluid for refrigeration?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact