Hacker News Reader

Comment by eli | Hacker News Reader

I attended a lecture about this a few years back. NOAA has a satellite system to warn us about large solar flares... but it's old and severely underfunded. Which is somewhat surprising, because that 20 minute heads up is the difference between GPS satellites and electric grid components being safely taken offline vs permanently damaged.

View on HN

Yeah, when I started getting involved in web operations tools, and realizing all the levels of redundancy that the big players like Google and Netflix have, it really got me thinking about our hydro and communications infrasture.

For instance, Netflix has it's chaos monkey, than randomly kills servers and availability zones and generally wreaks havoc. They set it loose, and learn to keep things operating. I really thinkg it would be a responsible societal effort to introduce chaos monkeys into our archaic infrastructure outside digital.

And as for that underfunded satellite, it's sitting in the sweet spot between the gravitation tug of the earth and sun called the L1 langrangian point, where it takes no effort to keep it exactly in place. But we really need more drills to learned whether the hydro/comm infrastructure on one hemisphere of the earth can actually shut everything down in 20 minutes, day or night.

The 100-year old underground electric wires running under the street where I work have caught on fire three times in the past year. Something tells me our archaic infrastructure has quite a backlog of existing chaos monkeys to deal with already.

The spirit of the idea is good, but it would not work well for hardware. With software, if something breaks, you restart the servers. With hardware, if something breaks, you have to buy a new one. Throwing chaos monkeys to see which transformers blow up would end up quite expensive. :)

Haha, was thinking the same thing. This would make a great xkcd.

Then at least simulations on how it affects electric grids coyld help quite a bit. Are this simulations taking place?

I interned at Areva T&D in 2006 (now Alstom Grid). While I was not working on that specific project, I remember they had a very powerful EMS (Energy Management System), which let them simulate incidents (e.g. "what happens if these power lines are destroyed").

Doing a quick search, this seems to be called a "dispatcher training simulator": http://en.wikipedia.org/wiki/Dispatcher_training_simulator

http://en.wikipedia.org/wiki/Energy_management_system

http://www.alstom.com/Global/Grid/Resources/Documents/Automa...

The relevant department within NOAA is the space weather prediction center, http://www.swpc.noaa.gov (beta site: http://origin-www.swpc.noaa.gov).

I toured their facilities in Boulder, CO a few years ago (I develop a satellite data product, http://link.springer.com/article/10.1007%2Fs11207-014-0529-3, that is used in forecasting). The forecasting is done partly with numerical models, partly with observational data, and partly with old school intuition and experience. It partly involves a small group of grizzled experts sitting in front of a bunch of monitors displaying images and time series.

They are trying to automate it as much as possible and bring in new data sources, especially near real time data sources, from NASA and NOAA satellites. It's too much to go into in the scope of a comment, but suffice it to say that, with as many photons as the Sun puts out, you can gather a lot of information about the entire 3D solar atmosphere and magnetic environment. One day, these measurements would feed a forward model of the same type used in terrestrial weather prediction.

There are a lot of modern facilities providing solar flare predictions, many are old and many are new. Some are space based and a lot are ground based (radio telescopes and coronagraphs) Government agencies amalgamate data from all the sources to come up with the best predictions. We always have more than 12 hours to prepare after spotting something from surface on the Sun, the accuracy of predictions improves as time goes by, the 20 minute mark is generally the time it takes for a fast CME to go from L1 Lagrange point to Earth. L1 is where multiple aging solar observatories collect insitu data, but we usually know before then if a strong event is Earth directed.

What could we do to satellites in 20 minutes that would protect them from a coronal mass ejection? It doesn't seem to me that simply powering them down would protect them.

We can't save satellites but when those begin to fail, we have 20 minutes to power down things before earth itself is hit.

You can save satellites - you can order them to rotate instrumentation away from the flare vector, and you can shutdown sensitive electronics or high-gain radios and the like (most satellites have many levels of shutdown for exactly this reason).

under adverse space weather conditions, launch personnel may delay a launch, and satellite operators may postpone certain operations (e.g., thruster firings). For the spacecraft industry, however, the primary approach to mitigating the effects of space weather is to design satellites to operate under extreme environmental conditions to the maximum extent possible within cost and resource constraints.

http://www.nap.edu/openbook.php?record_id=12507

Reply on news.ycombinator.com