Heat Recovery Off Screw Type Air Compressors

[JustAnotherChemE]

Hello,
I'm trying to spec out a way to recover heat off two 400hp screw type air compressors. These units have (or will shortly) their own four pass heat exchangers using an EG/water mix as cooling on the tube side. The oil from the compressor is in a closed loop.

Now the units developed by the air compressor company should be rejecting about 900,000-1MBTU/hr in the form of ~105 deg F EG/water flowing at about 70 GPH.

My scheme is to set up a secondary HX to recover this heat into a 55 deg F stream of city water which will then be mixed with the water headed to our [future] direct fire heaters.

One compressor is running pretty much 24/7 whereas the other might kick in 60% of the time when we enter peak production. We also have several 150hp compressors from the same company which might be worked into the system at a later date.

Now I'm fairly familar with shell and tube variety HXs, but I need to know if there are more efficient variaties for my application. One manufacturer suggested I consider a "Plate and Frame" type.


Suggestions?

 

[dcasto]

Efficency is a tricky word, you can build a heat exchanger from a garden hose and a roll of copper tubing and get 90% or buy one off the shelf and get 95% or custom fabricate and get 99.9999% of the heat out of the hot water. I guess what you really want is a cost per btu recovered. Shell and tube exchangers come in off the shelf sizes for your application out of catalogs such as Granger. You can also have customized unit build to get the exact size you need. The nice thing about plate and frame is that they are almost a cross beyween of the shelf and custom and thats why your vendor is recommending.

Get quotes for all of them and look at the cost per BTU recovered. Oh, don't for get the installation, here again, the plate and frame are very compact so installation should bve lower.

Q = U A delta T ln mean

 

[JustAnotherChemE]

Ah that sounds pretty good then. Footprint on this piece of equipment is important and this might work where the current defunct aftercoolers (~8-16 sqft) are now.

I think in this instace with the ability to use city water this system could recover most of the energy. It's all low grade heat afterall.

 

[pmover]

JustAnotherChemE,

while the information you've received is good, do be aware that plate heat exchangers are best suited for clean fluids. any minerals, salts, etc. can foul/restrict the SMALL passages in the plate type exchangers.

please note this thread: thread391-190222

meanwhile, good luck and it is good that the waste heat is being used!

good luck!
-pmover

 

[MintJulep]

Ok, I'm confused.

These units have (or will shortly) their own four pass heat exchangers using an EG/water mix as cooling on the tube side

I assume that "EG" means Exhaust Gas. Are the compressors dirven by and internal combustion engine? Or when you say EG do you mean the compressed air stream? What is the water that is mixed in? What is on the other side of this heat exchanger?

The oil from the compressor is in a closed loop.

Closed through what? Much of the cooling of a screw compressor is done through the lube oil. Are you planing to recover heat from the lube oil?

Is the compressor single or multi-stage? Where is the heat from the intercoolers and aftercoolers going?

Is there a refrigerated drier?

 

[JustAnotherChemE]

I dont mean to confuse. EG= Ethylene Glycol-water mix.

Each air compressor has it's own shell and tube heat exchanger that cools the oil from the compressor down from about 225 deg F to 180 deg F shell side. On the tube side these units are using what I believe to be a 50/50 EG water mix that comes in at 75 deg F and leaves at 105 deg F.

I want to introduce another heat exchanger that can capture the rejected heat from the 50/50 EG mix to preheat heater water instead of just cycling this stuff up to the rooftop cooling towers.

As for water cleanliness... well we have hardness issues here. Total water hardness ranges from 125-150 ppm seasonably. I think any exchanger used will need to be cleaned with sulfonic acid at least annually.... suggestions?

 

[dcasto]

You will have an extra exchanger in the EG loop, so you will need a few psi of pressure drop available from the pumping sytem and that should be available. Do you treat the water for hardness somewhere? If so use the treated water. I'm not a water chemist and you should ask one about you specific hardness and the minerals, but general chemistrys says if you warm up water it should hold more minerals in solution longer. Plate and frames tend to be easy to clean and you can see the degree of build up of salts on them from things like pressure drop, and temperature approach.

 

[JustAnotherChemE]

Well actually from what I've read the chemicals that contribute to water hardness show reverse solubility. i.e. they precipitate out as you increase temperature.

This EG water mix is prepared by a 3rd party contractor who also does the makeup water for our POS boiler system. It's possible he softens it. The city water is definitely hard, but the temperature rise may be low enough to avoid scale. (I hope!)

 

[MintJulep]

Ok, the cycle makes sense now.

The city water headed to the heaters is for domestic use?

Ethylene-glycol is toxic. If you are going to run a toxic fluid opposite potable water in the same exchanger you might as well simply run the city water through what is currently the ethylene-glycol side and eliminate the second stage.

If you want to attempt to protect against contamination of the potable water then you need to change to something other than ethylene-glycol as the intermediate loop. This is probably a code requirement anyway.

Is the city water flow constant? Is the flow rate sufficient to absorb the heat?

 

[JustAnotherChemE]

Good points. That said, we don't have to use the EG/water mix. That's just was is used to cool ammonia compressors in other areas of the facility.

The coolant water is strictly for sanitation. I think I'd better investigate what water codes are required.

The city water flow can be adjusted. This water will be headed towards a digitally controlled mixing valve so the flow rate will only be dictated by the allowable pressure drop across the HX. If there really is a coding issue with the EG mix intermediate fluid then a separate pumping system will need to be introduced along side the HX.

I had really hoped to retain the option of at least being able to redirect the flow of coolant from the primary HXs to the cooling towers in the event of secondary HX failure. This might still be possible as the volume of liquid cycling through the secondary system should be minor compared to the ammonia, limiting any dilution upon mixing.

How often does tube failure/leakage occur in operating heat exchangers? As I get further into this project more and more obstacles pop up, damn.

 

[zdas04]

Contamination concerns can be reduced with pressure. The EG loop can be at any pressure you want (you are pumping it after all), so if your potable water loop is at 70-90 psig like most city water systems and if you maintaing your EG system at 30-40 psig then a leak would go from the potable water to the glycol and overflow your EG make-up tank pretty quickly. While you need to check the codes, this kind of system is used pretty widly (usually with a system shutdown on the EG tank).

David

 

[unclesyd]

If you are going to tie in a EG/Water system to Potable Water supply a break tank will be needed on the Potable Water side.
I would also use a PG/Water system in lieu of The EG/Water system. PG is less toxic than EG.

zdas04
You cannot depend on a pressure differential to prevent mixing of the two streams. The biggest problem is that where the connection/leak occurs it is possible to have aspiration between the two systems.

 

[MintJulep]

Typically the lube oil in a screw runs 40 - 60 psi above the discharge air pressure, which would put it in the neighborhood of 200 psig.

 

[zdas04]

That's true, but the water is not in contact with the oil. A tube failure in the oil/glycol HX would overflow EG makeup tank prior to becoming a risk to the potable water.

David

 

[crjones]

On an oil flooded screw compressor they are usually a pressure feed system based on the operating pressure. Some manufacturers use an additional pump for bearing oil. so it is generally not higher than discharge pressure. so it is critical that extra pressure drop not be added. Or it would have to be made up with a pump.

 

[JustAnotherChemE]

Well the oiler is going through a separate shell and tube installed by the compressor company. I believe these compressors have an allowable 21 psig drop anyhow.

As for the makeup water flowing through these HXs I'll just spec out an appropriate sized pump. I was hoping to be able to avoid a pump on the secondary HXs using city water pressure to flow the city water shell side.

 

[crjones]

On a 110/125 psig plant air compressor the oil side pressure drop is generally designed in the 10 psid. The coolant side varies a bit more but generally it's 15 psid on the oil cooler and 3 psid on the aftercooler. Most compressors have a water regulator where you lose 1 to 3 psid. Make sure that you don't over cool the oil, there is a thermostatic valve in the oil system that is set for 140 degF that will shut off oil flow through the cooler if it's too cold. The regulator on the water will choke off the water flow.

 

[dcasto]

PG propylene glycol,is an approved cooling medium near foods.

 

[JustAnotherChemE]

Actually I've been looking into doping the closed loop water with EDTA.

I also found a shell and tube made by ITT Fluid Technology that was in McMaster Carr. its 7'Dx29"L, 4 pass with 29 3/8" tubes, made of 316 stainless.

None of the plate and frames listed meet our capacity though.

 

[JustAnotherChemE]

How embarassing. The entire sizing of this system was futile, I had a small unit conversion error on my mass balance (multiplied 7.48 instead of dividing) and the real reject heat is only ~19,000 btu/hr.

Lesson Learned: Double check those conversions before optimizing/sizing!

I small plate and frame might work.. but the payback would be like 10 years. :~/