keep a small mining farm cool with only air conditioners.

[mikeywith]

No, the efficiency actually changes:



TC is Capacity PI is power consumption.
It's a pain in the ass-protected pdf file, but I'm pretty sure even your model has the same kind of manual.
Again, look at the pipe length, this rating above is for 5m linear, and once you count elevation it's a mess.

Couldn't find a similar manual on the AC, the distributor of said brand guaranteed that the BTUs labeled on the AC will be achieved as long as the outside temp is below 52c.

Anyway, looking at the specs you posted, you can clearly see that the AC  has the highest TC when the room is hot and the outside temp is cold, so the hotter the room the more the cooling capacity, and, the greater the difference between in and out temp also the greater the cooling is, which makes sense, and that's what I picture the farm to look like, always hot inside - warm to cool outside.

It will bring the room temperature slowly up till it reaches the exhaust temperature at which point they will probably shut down themselves.
Let's look at it from the thermodynamics point of view as heat is just energy:

At one point you have one extra unit of produced heat in a unit of time, if this process goes continuously in a closed environment, after 1000 units of time it's 1000 extra units of heat. In a million units of time one million units of heat!
This will not stop at any point if we don't assume the equipment will melt itself because you're still feeding more heat in that room every second than you extract, 10C, 40C,
100C, 300C it's like blowing into a balloon, the heat has nowhere to go.
Of course, this is in a perfectly sealed environment, so there is heat escaping the room through the floor and the walls

There is nothing that can perfectly store energy in its heat form, the better the insulation the harder for heat to escape, but still, heat will always escape the room as long as the outside temp is is cooler, in the above example I gave, if 1 ac is turned on heating mode in a closed room, there is no way it would heat the room to melting point, in fact, if storing heat was that easy, then in the winter we would just use a small 1 ton BTU to heat a whole building, just wait for temps to keep rising since units of heat will keep accumulating, but we know this isn't the case, which is why we need to size cooling and heating and account for all the losses.

but if we talk about 30KW extra there is no way you will reach an equilibrium before you melt everything!
Especially since you also want to soundproof this room which will add to insulation and make less heat able to escape through the walls.

That's true, the 1 AC example is pretty "misleading", the more energy inside the room the hotter the room gets, there is limited energy/heat that can escape the room per unit of time, all depending on the insulation, sun, wind, outside temp, I'd imagine if you run a miner in a closed sealed room where outside temps at -40 the room would still be in a negative temp, while if you run it in the same room when temps outside are 40c, the pace/amount of heat leaving the room will be a lot lower, the potential difference between inside and outside is key.

With that said, I will by no means gamble with anything less than a 1:1 ratio at least for the initial feasibility study, in fact, it would be wise to assume that even 1:1 ratio isn't good enough and there will be days when we would have to lower the miner's frequency as Phill mentioned, there will be a lot of trial and error.

Thanks for your input Powell, do you recall the total power consumption of your farm vs BTUs?

For the humidity part, the place is very dry, there should be no issues with humidity at all.

As for the noise coming from the compressor /outside unit, it should be manageable, I checked a unit running and it was at 65 decibels at pretty much close to distance, noise levels will increase by 3 decibels every time the noise source is doubled, so 10 of them will output 75 decibels, that's a lot but still a lot less noise than a single miner doing that fan test, I doubt we will face issues with the noise coming from the ACs, I am more concerned about those Whatminers, since a single unit does 84 db when fans are at 100%, 50 of them will do close to 101 dB, I have exactly no way of knowing how much the insulation is going to reduce, but we kind of need the noise to be below 75 dB, could get away with a bit higher, but we will see.

@phill

There is a close to 10 degrees difference between the env temps of 192.168.1.232 and 192.168.1.233, there is something wrong there, can you try to swap them and see?

[1714230398]

1714230398

[1714230398]

1714230398

[philipma1957]

@mikey I know why its hot 🥵 environment.

I have shifted units it is due to two really  stubborn people that don’t want to vent the area just a bit more.
and downclock just a bit more gear.

I am still running 15 L3s that I don’t want to run. burning 🥵 12kwatts to make about 15 usd a day.

I would have sold them a long time ago. But I am outvoted so they will run to the LTC ½ ing.

[KiaKia]

It depends on how big the room is and some solution works better base on location, I think working air cooling fan into the farm is better in some.cases, I am using my own farm to analyst this question though, I don't want to run a air conditioner unless there are now newest air conditioner that takes less power consumption, the best I can find here are 1.5HP Air Conditioners and man that's a lot for me.

You said your electricity Bill is very cheap? Then you can run anything without worries of high bills on electricity then, AC draws a lot of power and that makes me want to forever stay away from them.

[spectre71]

Completely sealed is close minded.

My datacenter (not  mining) consumed about 11Mw give or take 24/7/365.

Developed an outside air system that blended cooler outside air into the filtered intakes of the AC units and opened dampers for escape. If it was a bit humid outside the AC would take care of it. Considerable savings especially in the winter months.

Side note: People mentioning mixing cold air into the hotside before the AC is just plain dumb, don't do that.

[philipma1957]

Completely sealed is close minded.

My datacenter (not  mining) consumed about 11Mw give or take 24/7/365.

Developed an outside air system that blended cooler outside air into the filtered intakes of the AC units and opened dampers for escape. If it was a bit humid outside the AC would take care of it. Considerable savings especially in the winter months.

Side note: People mentioning mixing cold air into the hotside before the AC is just plain dumb, don't do that.

an 11 mega watt data center is quite different to a 175kwatt sealed room setup.


At Mikeywith how did this room end up doing for you.

My 150kwatt under vented partially ac'd room has been holding up for  the summer. We have had to power off some gear (20kwatts) from 1pm to 8pm but we are doing pretty good for the last 6 weeks.

Maybe 90% of top hashrate. I am looking forward to Sept and Oct we will be able to boost the hashrate on the gear to around 105%

[DaveF]

Side note, since Phil bumped this the other day one of the DCs we have server equipment / routers in sent out an email overnight that they have lost backup cooling in one of the suites.
They have n+2 cooling, so sometime in the last few hours they lost 2 units. Not sure what happened, the last email was they have repair coming but no ETA.

It was more of an advisory since as of now there is no effect on us. This has happened once before in a different location. But the point remains the same. As you get more and more units, the odds go up that sooner or later one of them will have an issue. Every once in a great while, 2 will have issues. So if you really have no wiggle room be prepared for this.

This is a multi story DC built with all kinds of other backup plans it's not a big deal, and because suite we are in is about 20% empty it means even less, but if it was at 100% packed to the walls it would be different.

-Dave

[philipma1957]

Side note, since Phil bumped this the other day one of the DCs we have server equipment / routers in sent out an email overnight that they have lost backup cooling in one of the suites.
They have n+2 cooling, so sometime in the last few hours they lost 2 units. Not sure what happened, the last email was they have repair coming but no ETA.

It was more of an advisory since as of now there is no effect on us. This has happened once before in a different location. But the point remains the same. As you get more and more units, the odds go up that sooner or later one of them will have an issue. Every once in a great while, 2 will have issues. So if you really have no wiggle room be prepared for this.

This is a multi story DC built with all kinds of other backup plans it's not a big deal, and because suite we are in is about 20% empty it means even less, but if it was at 100% packed to the walls it would be different.

-Dave

Yep the more gear in a space the more likely something breaks.

Our setup is 2x 3 ton ac's lots of fans in and 2 fans out.

[stompix]

an 11 mega watt data center is quite different to a 175kwatt sealed room setup.

Talking about two orders of magnitude, the difference between firing a pellet gun and 155mm gun in a room or nearly the sam thing as saying that my computer with its GPU it's having no problem running in a room without AC and I'm not bothered by it.

175kw is much by some standards but at the same time is outside the rounding error for others.

Back on topic, I too wonder how Mikey is doing with the room!
Might be difficult but as some say, If it can be done someone will be doing it sooner or later.

It was more of an advisory since as of now there is no effect on us. This has happened once before in a different location. But the point remains the same. As you get more and more units, the odds go up that sooner or later one of them will have an issue. Every once in a great while, 2 will have issues. So if you really have no wiggle room be prepared for this.

A tiny advantage Mikey has, since it's a closed room it means it can set a standard temperature and stop everything at the moment of a single fluctuation, he knows that if one fails there is no way the setup will keep working as intended so the moment one goes down he can shut down  the same amount of consumption. Far easier to implement and less of an issue compared to a server farm for which a 20% cut can be as bad as an 80% in some cases. The local logistic hub of my company doesn't even have that luxury, it's either black or white, and everything works or everything goes down till is repaired, we stop and go all offline data entries that need to be checked before being added when everything is up again.

[philipma1957]

Also he has a lot of whatsminers.

their software can allow for shutdown or underclock really fast.

So I do have an interest in what he is going to do.

Off topic I found a new spot for mining.

I will be checking it out this week.

we may get 1200 amps at 240 watts.

[citb0in]

out of curiosity: how many people are you in the team and what was your invest for that farm? sounds nice, but also big and expensive :p

[LoyceV]

Room size is 10m*6m*3m / 32ft*19ft*9.8ft.

Each 60k btu AC pulls about 6kw worth of power, basically 1/3 of the miner's power.

If noise is a problem for the outside, how about heat? You'll be emitting 175 + 60 = 235 kW. That's the equivalent of 20 of those things on full power:

(image source)

Anyone walking by may be in for a surprise.

Based on that (assuming it's correct) if the outside temp in the above example was 15c, then I would need less cooling, simply put, I don't need the mining room to be at 15c, I am okay with 30c, which means probably using half the 10,236 cooling BTU will be enough to keep the room at 30c while the outside temp is 15c.

That would mean the other half of the heat has to flow through the walls. I'm not used to using BTUs, so I'd just keep everything in W. You produce 175kW, and have to remove 175kW. The efficiency (COP) of the airco will go up if the outside temperature is lower. Usually the COP is around 3-4, so when it's not too hot outside, you'll need 175/4kW for your airco. If it's hot outside, you'll need 175/3kW. That keeps the math much easier than bringing in BTUs.

Note that I have no practical experience whatsoever with running miners.

[philipma1957]

out of curiosity: how many people are you in the team and what was your invest for that farm? sounds nice, but also big and expensive :p

my current farm earns 9-10k a month

my share is 20%

last month I cleared 1800 usd.

[Flexystar]

If it is sealed then are we forgetting about the input and output math about the energy here?
If it is sealed and the heat output that is getting released into the divided environment is more than the Air Conditioners capacity, then obviously it is going to be a big issue. For one reference purpose, it is evident that there will be war between the miners and AC units in either heating up or down the room environment.

However, I am also adding one more point about "how sealed the room is?" Is it a perfectly sealed room or it does have few cracks due to door hinges, windows etc? In a small room the pressure wont be much thus outer air can easily parse through these crack openings.

If so then it changes everything about that room. In summer hot air can easily take away the coldness with it. In cold it might have some positive effects. I am sure these details will help a lot when it comes to mining operation that will be run for virtually permeant.

[citb0in]

out of curiosity: how many people are you in the team and what was your invest for that farm? sounds nice, but also big and expensive :p

my current farm earns 9-10k a month

my share is 20%

last month I cleared 1800 usd.

Thanks for your kind and quick reply. Sounds nice, congrats man and keep up the good profits

[mikeywith]

If noise is a problem for the outside, how about heat? You'll be emitting 175 + 60 = 235 kW.

On average there is 30kw worth of heat reaching from the sun for every 1 square meter on earth, the effect of that 235kw on the outside tempters is so negligible unless someone stands right in front of the external AC unit, they won't even notice that there is much heat being generated from the building.

Even with the normal setup of using only airflow, you would need to stand right in front of the exhaust fans to feel the heat, standing near the other sides or even leaning against the wall won't be enough to notice the heat, probably a little warmer to set next right next to the building during winter, but nothing much.

Furthermore, the building is isolated, the only problem is the noise, miners can be very noisy, I took a measurement of my farm running 200 S9s + 100 of a mixture between M30 and M20s, when the miners boot at start up the overall noise on the exhaust fan of the building reaches up to 110 db, that's as loud as a subway train, I went to the farm once in the winter around 11 pm it was very quiet in the area, I was able to hear the fans of a distance over 1000m (I made sure it the noise was coming from the farm).

Honestly, I am now more concerned about the noise than the cooling for this project, cooling will be solved by just adding more ACs, if 10 of them don't cut it, just power off or underclock some miners till we can add more ACs, the noise part is not something you can predict, it has to be a trial and error process until we reach desired noise levels.

That would mean the other half of the heat has to flow through the walls.

Yes pretty much, the room will lose heat as long as the outside temp is colder, the colder it is outside the more heat the room loses, of course, insulation and the type of wall plays a very big role in this.

And ya, 1:3-4 ratio is the correct math "on paper", but we will find out in time.


@stompix and philipma1957

We paused the operation for a while, turns out we have a bigger problem than cooling which is the noise, with the door closed the noise was still high, especially during the fan check phase, the place has the capacity to grow to 1MW so we want to do it the right way accounting for 150 gears, we tested with 14* M30s for a few mins before the room got hot (no ACs yet) and the noise was somewhat bad, the noise level will increase 3db with every doubling of the miners, so it will only get worse.

We are now considering two plans:

1- Build a second wall to reduce the noise (this will make cooling more difficult but probably cheaper than plan B)
2- Get rid of the stock fans and use fans with bigger bladers, less RPM, and the same airflow (will be expensive and need some extra work)


We are not rushing this for a few reasons, one being that it's pretty hot and will be until mid-September, and two is the fact that this is a huge investment and we must test everything before taking a risk, I ordered an inline fan that does 500cfm which should be enough for whatsminer, its advertised noise level is 60db (way below the 85db of whatsminer), will need to print a 3D shroud to fit it to the miner and see how it performs, meanwhile we are getting quotes for the second wall.

This will take a while to finish but hopefully, we will get there before winter.

If so then it changes everything about that room. In summer hot air can easily take away the coldness with it. In cold it might have some positive effects. I am sure these details will help a lot when it comes to mining operation that will be run for virtually permeant.

No windows, just a door that isn't perfectly sealed, your point is valid, we want to seal the cold side as much as possible while trying to get away with not so much sealing on the hot side to allow air to escape, doesn't matter summer or winter the exhaust side of the room will always be hotter than the outside temp so we could use every little crack or just bad insulation, but then again, every little crack will allow the noise to escape, it is extremely difficult to balance the two aspects here, you want

1- low noise.
2- less ACs.
3- less spending.

if option B mentioned above works, the noise level could be low enough that might even allow us to make a few holes at the exhaust side so that more heat can dissipate, but then having to buy an inline fan for every miner, printing the shrouds, buying different expensive shelves that are big enough for the miner + fan could be way too expensive, it might be better to build a second wall, make things worse as far as heat is concerned, and then just add more ACs than initially planned.

It's like trying to win in competitive eating without having to eat too much  , not an easy task, I knew it wasn't going to be easy, I just didn't think it would be this difficult, so ya, going to take a lot of time and money for this to happen.

[philipma1957]

If noise is a problem for the outside, how about heat? You'll be emitting 175 + 60 = 235 kW.

On average there is 30kw worth of heat reaching from the sun for every 1 square meter on earth, the effect of that 235kw on the outside tempters is so negligible unless someone stands right in front of the external AC unit, they won't even notice that there is much heat being generated from the building.

Even with the normal setup of using only airflow, you would need to stand right in front of the exhaust fans to feel the heat, standing near the other sides or even leaning against the wall won't be enough to notice the heat, probably a little warmer to set next right next to the building during winter, but nothing much.

Furthermore, the building is isolated, the only problem is the noise, miners can be very noisy, I took a measurement of my farm running 200 S9s + 100 of a mixture between M30 and M20s, when the miners boot at start up the overall noise on the exhaust fan of the building reaches up to 110 db, that's as loud as a subway train, I went to the farm once in the winter around 11 pm it was very quiet in the area, I was able to hear the fans of a distance over 1000m (I made sure it the noise was coming from the farm).

Honestly, I am now more concerned about the noise than the cooling for this project, cooling will be solved by just adding more ACs, if 10 of them don't cut it, just power off or underclock some miners till we can add more ACs, the noise part is not something you can predict, it has to be a trial and error process until we reach desired noise levels.

That would mean the other half of the heat has to flow through the walls.

Yes pretty much, the room will lose heat as long as the outside temp is colder, the colder it is outside the more heat the room loses, of course, insulation and the type of wall plays a very big role in this.

And ya, 1:3-4 ratio is the correct math "on paper", but we will find out in time.


@stompix and philipma1957

We paused the operation for a while, turns out we have a bigger problem than cooling which is the noise, with the door closed the noise was still high, especially during the fan check phase, the place has the capacity to grow to 1MW so we want to do it the right way accounting for 150 gears, we tested with 14* M30s for a few mins before the room got hot (no ACs yet) and the noise was somewhat bad, the noise level will increase 3db with every doubling of the miners, so it will only get worse.

We are now considering two plans:

1- Build a second wall to reduce the noise (this will make cooling more difficult but probably cheaper than plan B)
2- Get rid of the stock fans and use fans with bigger bladers, less RPM, and the same airflow (will be expensive and need some extra work)


We are not rushing this for a few reasons, one being that it's pretty hot and will be until mid-September, and two is the fact that this is a huge investment and we must test everything before taking a risk, I ordered an inline fan that does 500cfm which should be enough for whatsminer, its advertised noise level is 60db (way below the 85db of whatsminer), will need to print a 3D shroud to fit it to the miner and see how it performs, meanwhile we are getting quotes for the second wall.

This will take a while to finish but hopefully, we will get there before winter.

If so then it changes everything about that room. In summer hot air can easily take away the coldness with it. In cold it might have some positive effects. I am sure these details will help a lot when it comes to mining operation that will be run for virtually permeant.

No windows, just a door that isn't perfectly sealed, your point is valid, we want to seal the cold side as much as possible while trying to get away with not so much sealing on the hot side to allow air to escape, doesn't matter summer or winter the exhaust side of the room will always be hotter than the outside temp so we could use every little crack or just bad insulation, but then again, every little crack will allow the noise to escape, it is extremely difficult to balance the two aspects here, you want

1- low noise.
2- less ACs.
3- less spending.

if option B mentioned above works, the noise level could be low enough that might even allow us to make a few holes at the exhaust side so that more heat can dissipate, but then having to buy an inline fan for every miner, printing the shrouds, buying different expensive shelves that are big enough for the miner + fan could be way too expensive, it might be better to build a second wall, make things worse as far as heat is concerned, and then just add more ACs than initially planned.

It's like trying to win in competitive eating without having to eat too much  , not an easy task, I knew it wasn't going to be easy, I just didn't think it would be this difficult, so ya, going to take a lot of time and money for this to happen.

Soundproofing is hard as you are in a wet humid climate.

you can't use mdf it is not water resistant.

can you get sheets of pond fabric?


https://www.amazon.com/gp/product/B0964RTYPT/ref=ox_sc_act_title_1?

it is pretty good at stopping sound.



__________________
!___!____!___!__!___


layer of rubber
ribs of 1 inch pvc
layer of rubber

or

layer of rubber
2 by 4 pressure treated wood
Layer of rubber

that 3 part wall will stop a lot of sound

I am not sure if you get pvc pipe

or if rot resistant wood is easy.

these strips would work
https://www.lowes.com/pd/PVC-EMB-1X2-1-1-2-8/5013353085

use these screws
https://www.lowes.com/pd/Tuftex-10-x-1-in-Tan-Galvanized-Self-Drilling-Roofing-Screws-100-Count/3018299


attach top sheet  and bottom sheet  to same pvc strip in the middle

it is water proof and very sound resistant.

[LoyceV]

If noise is a problem for the outside, how about heat? You'll be emitting 175 + 60 = 235 kW.

On average there is 30kw worth of heat reaching from the sun for every 1 square meter on earth

That's incorrect:

The average annual solar radiation arriving at the top of the Earth's atmosphere is about 1361 W/m².

On the surface of the earth, it's less. Depending on your location and time of day, most of the time it will be much less. So 235 kW is really a lot.

Even with the normal setup of using only airflow, you would need to stand right in front of the exhaust fans to feel the heat, standing near the other sides or even leaning against the wall won't be enough to notice the heat, probably a little warmer to set next right next to the building during winter, but nothing much.

Compare the 20 patio heaters

Furthermore, the building is isolated

That solves all worries I was thinking of a small building in a crowded city.

the building reaches up to 110 db, that's as loud as a subway train

That's hearing damage level.

We are now considering two plans:

1- Build a second wall to reduce the noise (this will make cooling more difficult but probably cheaper than plan B)

Does the noise come through the walls of the building, or does it reach the outside through the airco system? Or both?

[coindigitalcoin]

heat might not be as bad if you keep them in an open area, instead of in a tight closed space.

[mikeywith]

you can't use mdf it is not water resistant.
can you get sheets of pond fabric?

We could use MDF inside, we have to build a second wall, it doesn't mean it needs to be outside the current wall, although technically, it would be easier to build outside.

Those pond fabrics seem like good sound insulators, they are high-density which is the most important element in soundproofing, will check to see if I can find them around, hopefully the price is reasonable.

That's incorrect:

The average annual solar radiation arriving at the top of the Earth's atmosphere is about 1361 W/m².

On the surface of the earth, it's less. Depending on your location and time of day, most of the time it will be much less. So 235 kW is really a lot.

Everyday, the sun beams to earth about 10,560 BTUs of energy per square foot.

source: https://www.extension.purdue.edu/extmedia/ae/ae-99.html#:~:text=Everyday%2C%20the%20sun%20beams%20to,available%20at%20the%20earth's%20surface.

1 BTU = 0.293 watts
so 10,560 BTU = 3,094 W per square foot
1 square meter = 10.76 squer foot
3,094 * 10.76 = 33,291w
1kw = 1000w
33,291w = 33.291kw per square meter.

ya, a lot of that is lost along the way, but really, 0.25mw is not a lot as far as ambient is concerned, have you ever walked next to a hotel or a large building and felt that it was "strangely warm" in the street? ya standing on the exhaust side of the HVAC or near the external ACs units will be too hot, but just by walking around, it would be hard to tell if the building is consuming 1mw or none just by the "feel of it".

0.25mw is a lot of "heat" in a closed place if the size is not too large, but as far as the outside it's too tiny to be noticeable.

Compare the 20 patio heaters

what happens when you put 20 patio heaters inside the house and walk outside, are you still going to feel the heat? probably nothing much unless there is enough airflow forcing that heat to your direction, I have personal experience walking in and out of mining farms as large as 2MW, it's only warm at the exhaust side because heat is being pushed outside into that direction, walking around the other 3 walls you feel nothing.
 

That solves all worries I was thinking of a small building in a crowded city.

I wouldn't be even attempting that in a crowded city, the nearest building is roughly 100m away.

Does the noise come through the walls of the building, or does it reach the outside through the airco system? Or both?

We don't have the ACs in place yet, the noise is coming out the wall and more so from under the door, but sound travels in all directions so it goes through everything at different levels, there is at least a 20db drop between door open and close, but generally, it's not close to as quiet as we need it to be.

[philipma1957]

you can't use mdf it is not water resistant.
can you get sheets of pond fabric?

We could use MDF inside, we have to build a second wall, it doesn't mean it needs to be outside the current wall, although technically, it would be easier to build outside.

Those pond fabrics seem like good sound insulators, they are high-density which is the most important element in soundproofing, will check to see if I can find them around, hopefully the price is reasonable.

That's incorrect:

The average annual solar radiation arriving at the top of the Earth's atmosphere is about 1361 W/m².

On the surface of the earth, it's less. Depending on your location and time of day, most of the time it will be much less. So 235 kW is really a lot.

Everyday, the sun beams to earth about 10,560 BTUs of energy per square foot.

source: https://www.extension.purdue.edu/extmedia/ae/ae-99.html#:~:text=Everyday%2C%20the%20sun%20beams%20to,available%20at%20the%20earth's%20surface.

1 BTU = 0.293 watts
so 10,560 BTU = 3,094 W per square foot
1 square meter = 10.76 squer foot
3,094 * 10.76 = 33,291w
1kw = 1000w
33,291w = 33.291kw per square meter.

ya, a lot of that is lost along the way, but really, 0.25mw is not a lot as far as ambient is concerned, have you ever walked next to a hotel or a large building and felt that it was "strangely warm" in the street? ya standing on the exhaust side of the HVAC or near the external ACs units will be too hot, but just by walking around, it would be hard to tell if the building is consuming 1mw or none just by the "feel of it".

0.25mw is a lot of "heat" in a closed place if the size is not too large, but as far as the outside it's too tiny to be noticeable.

Compare the 20 patio heaters

what happens when you put 20 patio heaters inside the house and walk outside, are you still going to feel the heat? probably nothing much unless there is enough airflow forcing that heat to your direction, I have personal experience walking in and out of mining farms as large as 2MW, it's only warm at the exhaust side because heat is being pushed outside into that direction, walking around the other 3 walls you feel nothing.
  

That solves all worries I was thinking of a small building in a crowded city.

I wouldn't be even attempting that in a crowded city, the nearest building is roughly 100m away.

Does the noise come through the walls of the building, or does it reach the outside through the airco system? Or both?

We don't have the ACs in place yet, the noise is coming out the wall and more so from under the door, but sound travels in all directions so it goes through everything at different levels, there is at least a 20db drop between door open and close, but generally, it's not close to as quiet as we need it to be.

built a lot of speaker cases and sound proofed a few recording studios.

If you can do mdf inside make a sandwich.

mdf/pond fabric/mdf/pond fabric/mdf

you can use 1/2 inch mdf sheets . the wall would be about  2 inches thick and stop a lot of the noise as most of the noise is higher than 400hz

https://www.youtube.com/watch?v=rXu61z8yiMc

stopping that is easier that stopping this 40 hz

https://www.youtube.com/watch?v=rXu61z8yiMc

note they (sonic electronix ) have a lot of frequency you can test.


yeah I found the paper on mdf/rubber/mdf/rubber


http://www.woodresearch.sk/wr/201904/16.pdf


notice the improvement they got with multi layer

mdf/rubber and glue

I did 5 layers

3 of 12.5mm mdf
2 of pond fabric
 I glued, clamped and screwed them together.

I was hoping to find the paper.

I was building enclosures 1994-2004  these guys expanded and tested a lot of the work I did with a few others in the Brooklyn navy yard with our speaker building group.


 this paper clearly shows that the sandwich method of

mdf-glue-Rubber-glue-mdf  works really well


'CONCLUSIONS
By using rubber-based composite materials, it is possible to create a sandwich structure with superior acoustic performance while retaining the same thickness of the single-layer density fibreboard for sound insulation performance. Using MDF face sheets with a rubber core increases the sound insulation at the resonant frequency and causes the coincidence frequency to move to higher frequencies, effectively inhibiting the anastomosis effect and making the coincidence valley shallower. Moreover, core materials with high damping performance corresponded to multi-layered composites with improved acoustic performance. As the thickness of the rubber layer increased, so too did the loss modulus of the composite. The weighted sound insulation of single-layer MDF was 28.0 dB, which increased to 37.4 dB for the wood damping composite material, increased by 25.1%. As the thickness of the rubber increased, the storage modulus and loss factor of the composite both increased accordingly. The greater the damping loss factor, the greater the energy loss, making the composite material more resistant to sound waves caused by stronger vibrations. Compared to the rubber samples, the multi-layered composites showed higher storage modulus and loss modulus values. Since the loss modulus is a measure of the energy dissipation, a higher loss modulus of the multi-layered composite denotes that more acoustic energy would be dissipated during sound wave propagation in the material. Therefore, the sound insulation ability of the composites can be enhanced by controlling the rubber thickness, keeping in mind that a good balance between the sound insulation ability and the mechanical properties is required. The density of rubber is not significant for the sound insulation performance of composite materials.'

source is above