Odd dispute
If your reason for placing a datacenter in Arizona is anything but "cheap land and solar power," you might have a flawed analysis. It's no surprise that a desert city dislikes your plan to convert water into brine.
Microsoft has agreed not to use evaporative cooling in future datacenters at its PHX 10-11 campus in Goodyear, near Phoenix, Arizona, a move the Windows giant touted as a sustainability improvement because it will reduce water consumption in the drought-prone region. However, the City of Goodyear's recounting of the events …
Slight bit more complicated mate...the city almost certainly got dollar signs ($$$ not that they will be worth much) in their eyes and allowed some range of compromise. Only now it's not looking so good, so all sides are pulling back. Record high Colorado river basin snowpack...at least both sides are looking beyond that?
Well I might be mistaken but they do happen to have these Datacenters located in an area which sees a lot of sun, most of the year around. Now I'm sure there's a technology that exists that could easily be located on top of said datacenters (or even on the ground around them), that absorbs sunlight directly on some panels and converts it to .... what? No you're just going to burn some extra gas and coal. Ok yeah fine go with that....
In fairness, a quick search indicates MS have done a deal with a local solar generator to build out capacity for them to run mostly on solar (presumably not at night!).
What's weird though is the datacentres themselves have no panels on the roof. This isn't because there's other plant in the way - the cooling plant is all at ground level around the DC. Sure, I guess it doesn't make any difference to the bean counters whether you put panels on the roof or buy it in from down the road. But land in Phoenix isn't free (even if it's cheap - relatively speaking). And from a building-cooling perspective, actively sun-shading your roof with half a megawatt of solar panels is even better than painting it white. It does mean the roof needs the structural capacity for the weight of panels, but when you're building from scratch that's something you can reasonably accommodate.
absorbs sunlight directly on some panels and converts it to ....
Thing is, you see, let's think about it. (yes, I did see the icon...)
What is the footprint of a typical datacentre rack? Let's be very generous and imagine it's about the same as a typical roof-mounted solar panel (that's about 1m x 1.75m by the way, considerably larger than most racks I know).
Let's be even more generous and imagine that there is only one floor in this datacentre and that this floor has an area of 20x the area of a rack for each rack installed - that's to accommodate the plant, the staff facilities, store rooms, corridors etc. We'll assume that the whole of the roof is available for solar panel installation.
Now let's assume the panels can be ideally angled to the hot Arizona sun and that 1m of panel requires 1m of roof to avoid shading and indeed, how about motorising them so that their angle can be varied through the year so that they are always square-on to the sun?
A typical panel of the size I've mentioned might have a peak output of 350W so again, let's be generous and imagine 400W and ignore inefficiencies caused by heat and all the other stuff which means that kWp is rarely reached in practice.
According to Wikipedia, Phoenix receives about 3,800h of "bright sunshine" a year, which is only a little less than if we assume 12h of sunshine a day, so let's go with 12 hours of generation per day and that all of this is at peak output.
In a typical day then, the 20 panels available for each rack could potentially produce a maximum of a generously-calculated 20 × 0.4kW × 12h = 96kWh of electricity, which is quite a lot.
How much electricity does a rack actually consume? I don't build racks (not this sort anyway), but I've had a look around and it seems as if a low-density rack (I'm being generous again) might be rated for 10kW and require about half that again for cooling, so the power consumption of a rack over a day is 15kW × 24h = 360kWh.
Not even close.
Ok, so I'm being a bit devil's advocate, and I have to point out that if someone did cover the roof of each datacentre with panels and generate ¼ of the electricity consumed onsite (it'll actually be less, but the point is the same), that's quite a lot of electricity that doesn't have to be sourced from coal, oil and gas so it's definitely beneficial if you can afford the initial costs...
M.
I'm a little confused how evaporation is somehow turning fresh water into brine? Where is the salt coming from?
The fresh water. It's not distilled, and whilst it's not as concentrated as sea water, it still contains minerals and salts (that's where limescale comes from).
If you evaporate off millions of gallons of fresh water (which is what these DCs are doing daily during peak periods), you'll have a briny sludge of minerals and salts left over.
Exactly. And in fact water sources in the US Southwest tend to be quite high in mineral content. The water we get from our well in northern New Mexico has around 200 ppm of calcium dissolved in it (and a pH of 9, both of which are outside EPA recommendations). Which is what you'd expect; dig down half a meter and you come to a karst ("caliche" is the local term) layer that's about 50-100cm deep, a white streak through the ground that's calcium (and some magnesium) leached out of upstream soils and percolated down into the ground from natural flows and three centuries of irrigation.
Our situation is better for cooling than Phoenix's, because at our altitude things cool off very quickly once the sun sets. Our house doesn't have A/C or evaporative cooling and we don't need it; nighttime passive cooling and lots of thermal mass keep the daytime temperatures comfortable. I wouldn't want to run a DC here though.
"However, according to the city, Microsoft almost immediately abandoned this plan — apparently due to trouble with the governing regulatory bodies. Microsoft later sought an amendment from the city allowing it to dump its untreated wastewater into the city's sanitary sewer system instead."
Maybe a little creative accounting can get around the regulations. Instead of Microsoft pre-treating the water, they should have the city pre-treat the water, at a city-owned treatment plant that just happens to be located on the Microsoft campus. Then the city can deal with the regulatory bodies, of which they have plenty of experience.
Maybe a little creative accounting can get around the regulations. Instead of Microsoft pre-treating the water, they should have the city pre-treat the water, at a city-owned treatment plant that just happens to be located on the Microsoft campus. Then the city can deal with the regulatory bodies, of which they have plenty of experience
Yeh, but water and riparian rights have a long and ignoble tradition, especially as demands for water continue to grow. So a whole slew of local, city, state and federal regulations to navigate. Plus fun stuff like agreements between states for who gets how much water.
But common sense would seem to suggest... ermm.. Collocation. Anyone who's ever played a city builder like Sim CIty knows this. Pre-treat water so it's good enough for Microsoft, treat the dischage, throw in a few SMRs to keep it all powered. Then it's just the cost sharing to figure out.
Easier to move data than water?
Put the DC next to a hydro-dam (say, Washington state?). Find a data network (say, the innernet?) to carry what few bytes Arizona needs?
Land costs CAN'T be the issue. You can stack data very high, like those container-farms in Europe (OVHCloud's SBG2 Strasbourg fire), built on a sliver between the Rhine river and an industrial railroad yard.
Good job on missing the point people. It's your lord and master Microsoft. It has a right to your water and power. You need to learn to deal with some salty tap water. You owe them for making jobs. If you want more water treatment plants and power, pay for it out of your own pockets, or they walk away.
"For example, direct liquid cooling or immersion cooling, both of which are significantly more efficient compared to air-cooled systems." - won't help, as all the heat needs to be vented into atmosphere somehow. Data centres in the southern states either use adiabatic coolers, or pretty much double the energy consumption - either way it's not sustainable.
The river, not the state.
El Reg readers probably haven't been following the ongoing situation with regard to this river. The problem is that we've been overdrawing on it for decades so we're going to have to re-divvy up its water among the states. California has quite a big share of the water, its mostly for agriculture in the Imperial Valley. but even if we gave this up entirely (unlikely) there wouldn't be enough for the other states to carry on as they have been doing.
Phoenix is a city that shouldn't have been allowed to develop. Its too hot and too dry to support a sizeable population. It grew because land is cheap -- a large flat expanse of desert is easy to build on provided you don't think about the long term. Semiconductor plants are water hogs and there's policy to build a couple more, that water's got to come from somewhere, and the only place they're going to get it is from squeezing existing users.
Evaporative coolers -- known colloquially as 'swamp coolers' -- are common on older cheap tract homes because they're really cheap to both install and run compared to full A/C. A/C itself is a bit problematical since the units won't be very efficient in 40+C temperatures (expect more than average unit failures as well). It all really adds up to Nature's Way of trying to tell people that perhaps this isn't the best place to locate major new industries (and people). But we're good at ignoring warning signs.
I(ncidentally, for those Brits who think that this climate's no big deal I'd recommend a few days in Phoenix in summer. Its beyond description -- its when nightime lows don't go much below the mid 90sF that you realize that the only way to cope is to head to the north of the state.)