Don't Go Being Politically Insane You Climate Change Skeptics

[Ivanhoe]

1 hour ago, rmgill said:

Oh another consideration for data center operating temperatures is staff. Putting staff working long hours in a 110 degree data center is NOT conducive to good working environments. Those human factors are serious considerations.

Use interns for inside the server room, and use all plenum-rated to minimize the toxic smoke. Problem solved.

[rmgill]

1 hour ago, sunday said:

Ok, ok.

So both ultimately dump the heat to air, then?

Yes. Generally. Most of the time data centers seem to not be in urban areas, at least new big builds. There are smaller ones in old telecom facilities (telegraph) that originate in the days of railroad, those are key spaces for switching points, pops and some light colo. They're expenseive. The BIG data centers they're in rural areas. They could sell heat to their neighbors, but those are ALSO data centers. The cost to pipe hot water or steam to neighbors is minimal. THEN you also get into the factor of being a utility that's supplying a service and if your data center is lightly loaded, how do you provide continuity of service. 

Here's a concentration of them in Ashburn, va. 
https://www.google.com/maps/place/Ashburn,+VA/@39.0094811,-77.4657617,1455m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu

And more

https://www.google.com/maps/place/Ashburn,+VA/@38.9995232,-77.4800779,1881m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu


And more

https://www.google.com/maps/place/Ashburn,+VA/@39.018033,-77.4620185,830m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu


And more

https://www.google.com/maps/place/Ashburn,+VA/@39.0222043,-77.451839,830m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu

And more

https://www.google.com/maps/place/Ashburn,+VA/@39.0244263,-77.4351421,492m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu


and more

https://www.google.com/maps/place/Ashburn,+VA/@39.0021263,-77.4389889,630m/data=!3m1!1e3!4m6!3m5!1s0x89b615f166fcc957:0x9da316eb11e3d5b!8m2!3d39.0437567!4d-77.4874416!16zL20vMDVwNDZr?entry=ttu



Every time I'm in that area, there are more data centers. Running the pipe to the nearby customers would be interesting....But you'd be dumping the heat in the air most of the year. 


 

[rmgill]

14 minutes ago, Ivanhoe said:

Use interns for inside the server room, and use all plenum-rated to minimize the toxic smoke. Problem solved.

Then you will have outages and problems because your interns need to know what hte'yre doing. I've cleaned up after interns who didn't know what they were doing. 

There's a reddit for this sort of mess. 
 

https://www.dotcom-monitor.com/blog/server-cable-hell-worst-wiring-jobs-ever/

And after looking at those, I need a drink. 

[sunday]

1 minute ago, rmgill said:

your data center is lightly loaded, how do you provide continuity of service. 

Major drawback here.

If the data center is continuously mining crypto, however...

[rmgill]

But if you're running AI and it's surgy. Or you're doing financial markets so you're busy from 9am to 5pm. At CNN's site got busy as the markets opened and people got to work it would surge upwards ads folks ended their work day and drop off to 3 separate plateaus as 5pm hit eastern, standard and central, then drop off to low levels after 5pm pacific time. 

But yeah, IF you're providing steam or hot water to clients near by, you've now become a local utility and you've made your business more complex. Now you have to invest in pipe to feed clients AND you have to support them at all manner of hours. Do you get a steam generator to maintain a certain constant temperature through winter despite load needs on your data center? 
 

Edited June 24, 2024 by rmgill

[sunday]

2 minutes ago, rmgill said:

But if you're running AI and it's surgy. Or you're doing financial markets so you're busy from 9am to 5pm. At CNN's site got busy as the markets opened and people got to work it would surge upwards ads folks ended their work day and drop off to 3 separate plateaus as 5pm hit eastern, standard and central, then drop off to low levels after 5pm pacific time. 

Yes, yes, you are right on those scenarios.

The original article announced the use of the rejected heat of a crypto mining operation for district heating, and mining should be a nearly constant load.

Edited June 24, 2024 by sunday

[bojan]

If the peak load is 9 to 5, only way it is useful is providing hot air to other office buildings nearby during colder period of year.

There are a lot of weird things where waste heat can be useful, but that is all niche use.

Weirdest cooling/heating system I have seen was on the river dredging ship, it had multiple compressors needed for work, running basically all the time with plenty of "exhaust" air. So it had some vortex tubes built in, and it provided heat to cabin (small one) during a winter and cool air during summer. Ingenious solution, not using any moving part of separate heating elements , but totally niche application based on surplus of "waste" compressed air.

 

Edited June 24, 2024 by bojan

[rmgill]

Niche case in a boutique market. I like the efficiency. But when I started trying to think about doing that with the three data centers under my purview (we are tenants in 2) it got complicated. Plus all the nearby customers would be transient. For us to get them to put in water based heating for winter, not worth the squeeze. 

Now, if I had a data center in a mixed use area or mixed use building, that could work. For CNN, we might have been able to use rejected heat from chillers to sell to the Omni Hotel, the distances would have been shorter, but it would have only been useful part of the year. 

The cost of the hardware compared to the amortized cost of the normal domestic hot water....not so ideal. And in many office towers you'll find the local domestic hot water is heated at point of use with a small water heater. 

[rmgill]

4 minutes ago, bojan said:

Weirdest cooling/heating system I have seen was on the river dredging ship, it had multiple compressors needed for work, running basically all the time with plenty of "exhaust" air. So it had some vortex tubes built in, and it provided heat to cabin (small one) during a winter and cool air during summer. Ingenious solution, not using any moving part of separate heating elements , but totally niche application based on surplus of "waste" compressed air.

 

Not any different really than your own car's heater being used to heat a core for keeping the passenger compartment warm. Where it gets interesting is when someone has an electric car and they don't understand why their range is so limited in winter when they run the heat. 

[sunday]

Thanks, guys.

[jmsaari]

On 6/24/2024 at 11:44 PM, rmgill said:

But yeah, IF you're providing steam or hot water to clients near by, you've now become a local utility and you've made your business more complex. Now you have to invest in pipe to feed clients AND you have to support them at all manner of hours. Do you get a steam generator to maintain a certain constant temperature through winter despite load needs on your data center? 

The data centers wouldn't be entering the utilities market, they'd just sell the heat to the local energy company, who would take care of pipe maintenance. Even if it were 8-12 hours a day, if there's a lot of heat in those hours, it would prob make sense to locate the data center so there's a market for the heat. When the data center doesn't have heat to sell, the energy company just uses something else to produce the heat with. Hot-water tanks are also an easy way to balance the supply if the peak supply hours exceed the demand, and sometimes vice versa.

[jmsaari]

On 6/24/2024 at 10:18 AM, Soren Ras said:

Using waste heat from power plants (whose primary output is electricity), data centres or other industrial manufacturing for district heating is quite well established and hardly unusual in the Nordics. The only novel bit about this Finnish exercise I can see is that it is heat from a Bitcoin farm as opposed to a regular data center.  Water pipes are insulated pretty much by default in any case, but you do need to build (or rebuild) the data center specifically for this to work and as I understand it, you need a certain critical mass in terms of DC size before it makes sense (and the bigger the better). 

Soren

It really depends on the local energy company's policy... the idea of "prosumers" who are able to buy and/or sell heat through the network depending on situation is picking up with some, idea is to be able to tap into some pretty small heat sources to reduce fuel costs.. but if any length of new pipeline is needed to connect to network, the obviously it needs to be big. For a data center that's able to supply most of the heat demand of a small town, it's worth digging a bit of pipeline...

[jmsaari]

On 6/22/2024 at 2:53 PM, sunday said:

regardless of CO2 driving the temperature increase, or the inverse, and that is something that is far from being scientifically settled.

Well, several known mechanisms working both ways exist and are quite settled, but T isn't the only thing that's driving CO2 and CO2 not the only thing driving T. (and thus you can set up a signal processing approach to react to what you want, and get published if not in a decent journal, at least with a publisher that in last few years has been nosediving into predatory category...)

The big effect driving CO2 up from temperature increases is oceans degassing as their temperature increases. But this time we know that as temperatures have gone up, the CO2 content in seawater has gone up with it, while CO2 content in atmosphere has increased only by about half as much as what'd be expected if all CO2 from fossil fuel combustion would remain in the atmosphere. 

[sunday]

34 minutes ago, jmsaari said:

But this time we know that as temperatures have gone up, the CO2 content in seawater has gone up with i

Is there a long enough record of CO2 content in seawater?

Also thanks for not even bothering to show that you tried to understand the statistics on the linked article.

Edited June 26, 2024 by sunday

[rmgill]

1 hour ago, jmsaari said:

The data centers wouldn't be entering the utilities market, they'd just sell the heat to the local energy company, who would take care of pipe maintenance. Even if it were 8-12 hours a day, if there's a lot of heat in those hours, it would prob make sense to locate the data center so there's a market for the heat. When the data center doesn't have heat to sell, the energy company just uses something else to produce the heat with. Hot-water tanks are also an easy way to balance the supply if the peak supply hours exceed the demand, and sometimes vice versa.

You locate data centers based on space, proximity to power, water and connectivity. Doing so based on access to someone to dump your waste heat is a tertiary function. 
 

 

[jmsaari]

32 minutes ago, rmgill said:

You locate data centers based on space, proximity to power, water and connectivity. Doing so based on access to someone to dump your waste heat is a tertiary function. 

I find it quite hard to believe that with the power consumptions of data centers, it wouldn't be a factor in siting choices if you can re-sell most of your energy consumption to someone as heat, at a fairly high fraction of the purchased el.purchase price.

And it's not like you'd be terribly limited in siting options by looking for heat customers in Nordic countries...

[jmsaari]

1 hour ago, sunday said:

Is there a long enough record of CO2 content in seawater?

Also thanks for not even bothering to show that you tried to understand the statistics on the linked article.

About 30 years of measured record IIRC, showing a clear trend that goes up with atmospheric CO2, not down as you'd expect with temp-driven CO2 increase, and obviously way beyond that its proxy data. 

And yes, i only skimmed through it; when a paper has to ignore a lot of what *is* actually known, from its base assumptions to its conclusionms and gets published by MDPI in 2023(*)m it's not something i'm willing to spend a lot of time going through... 

(*) they were always low-tier but kinda ok for publishing things that aren't exactly groundbreaking science but have enough value and novelty that you dont want to leave unpublished. At least reasonably pretended to follow decent peer review standards until the last couple of years. And yes i have to admit i did publish there, in hindsight after the point i really shouldnt have after the last couple of reviews i did for a few of their journals. It took a district heating related paper where the "expert" reviewer wanted me to describe what kind of anti-freeze we use for district heating water in Finland since surely the water must freeze in our climate otherwise, until i finally decided i'm done with them...

[rmgill]

I don’t think you can resell sell most of your energy usage as heat. 
 

There would have to be some substantial efficiency losses there. 
 

I am kinda curious about this wondering if we could do that with some of our sites.  
 

The one I am sitting in right now can run on one 2 megawatt gen sets during a power outage. For chillers typically run one 600 ton chiller about 600 hours a month. But at about 80% capacity. That’s near to 1.6 mw of heat based on my back of the envelope calcs

I don’t see how that efficiency losses would work out to nearly that much out at the far end. 

 

The siting to proximity to good power, ie multiple grid/substations, diverse fiber paths and good water supplies is a critical set of criteria. 
 

 

Edited June 26, 2024 by rmgill

[jmsaari]

43 minutes ago, rmgill said:

I don’t think you can resell sell most of your energy usage as heat. 
 

There would have to be some substantial efficiency losses there. 
 

I am kinda curious about this wondering if we could do that with some of our sites.  
 

The one I am sitting in right now can run on one 2 megawatt gen sets during a power outage. For chillers typically run one 600 ton chiller about 600 hours a month. But at about 80% capacity. That’s near to 1.6 mw of heat based on my back of the envelope calcs

I don’t see how that efficiency losses would work out to nearly that much out at the far end.

I'm no expert on data centers, but seems there shouldnt be very much losses along the way when you're converting electricity to heat? Well the electricity that gets used gets anyway all converted to heat first, and if you're cooling the space with a heat pump, the condenser will dump almost all that plus the compressor's power consumption at a higher temperature into the district heating water, shouldn't be any huge losses of heat anywhere along the way there..

There would be some losses between the end user and site of course that would affect how much the energy company would be willing to pay for the heat, but that's not much either... <10 percent in a big network, or <20% even in a very small one. 

Might get hard to make it work without a pre-existing network though, which i guess would be the case in most of north america...

Edited June 26, 2024 by jmsaari

[rmgill]

Well the buildings themselves are not 100 % insulated. There are losses there. 
 

Also, are computers 100% efficient at converting energy to heat? 

[sunday]

3 hours ago, jmsaari said:

About 30 years of measured record IIRC, showing a clear trend that goes up with atmospheric CO2, not down as you'd expect with temp-driven CO2 increase, and obviously way beyond that its proxy data. 

And yes, i only skimmed through it; when a paper has to ignore a lot of what *is* actually known, from its base assumptions to its conclusionms and gets published by MDPI in 2023(*)m it's not something i'm willing to spend a lot of time going through...

There is correlation, and there is causation.

The paper shows causation, not correlation, IMHO.

You are free to dismiss that, of course.

[rmgill]

Hmmm. As I think about this more and chat with some co-workers I am wondering if this is something more feasible. 
 

Make note, that my view is shifting in this conversation of a few posts to the other idea being potentially more workable than not. 
 

The question is the potential cost recovery when combined with the offset cost of the added infrastructure. 

[Ivanhoe]

4 hours ago, rmgill said:

Also, are computers 100% efficient at converting energy to heat? 

No, 90% goes to heat and 10% goes to software bugs.

[lucklucky]

Quote

 

Due to its complex role within the radiation budget of the Earth, cloud cover plays an important role in climate variability at local, regional or continental scale. Consequently, there is a need for better understanding the causes and feedbacks of changes in cloud cover. 

 

“Since clouds in reality have three-dimensional (3D) structures, the simulation of radiative transfer (RT) in clouds should ideally consider the transport of radiation in both vertical and horizontal directions (referred to as ‘3D RT’).”

However, “operational bispectral cloud retrievals are almost exclusively based on the one-dimensional (1D) RT theory that considers only the vertical and ignores the net horizontal transport of radiation.”

Consequently, “the radiative properties of clouds under 3D RT are substantially different from those under 1D RT.”

 

The sorry state of climate simulation.

https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/joc.6841

[sunday]

Quote

Summary

Henrik argues cosmic rays and sunspots drive global warming more than anything else. He has found that cosmic rays increase cloud formation, which cools—and warms—the climate.

Henrik has also discovered a strong link between cosmic ray flux, cloud cover and global temperatures.

In other words, more cosmic rays mean more clouds and more climate change.

His experiments show that cosmic rays boost cloud condensation nuclei formation.

In short, humans do not cause climate change.

 

https://jermwarfare.com/conversations/henrik-svensmark

 

Edited June 28, 2024 by sunday