Data Centers: EMP Shielding, What it is and Why it Matters

Today I want to talk about a fascinating, frightening and complex topic, EMP. I’m sure many of you are familiar with the idea of an Electromagnetic Pulse, and have seen depictions of EMPs in popular culture and science fiction, however the reality of an EMP can be far scarier.

In my past life, I used to build high security facilities. One project I worked on was building an EMP shielded data center facility 225 feet below a mountain. Now, you might be thinking, if you are building something 225 feet below ground underneath a mountain of Limestone, wouldn’t that be enough to protect your equipment? The mountain would help shield you from the force of the blast, but it does almost nothing to protect you from EMP.

The reason for this is that EMP is like a radio signal, it is an induction current. So, if a radio signal can get through then so can the EMP. Unfortunately a whole mountain of rock does nothing to protect against this. This is also why it is so very hard to protect against EMP.

The type of EMP burst that I am going to spend the most time talking about is high altitude nuclear explosions. There are other things that cause some or all of the effects a high altitude burst. One can create a large burst with a Marx generator and a directional antenna. A Marx generator was invented by Otto Marx in 1924 and it is basically a cascading series of capacitors that all discharge simultaneously to produce a very high voltage pulse. When combined with a directional antenna, a device like this could be used to target a single building. There are non-nuclear cruise missiles that likely use this or similar technology to produce an EMP. 

There are naturally occurring events that also create large energy events. Solar flairs can produce some of the same effects as an EMP device, causing damage to long haul transmission and power delivery and generation systems. There is also the possibility of a late Gamma ray burst from a collapsing star creating a large EMP burst. Extraterrestrial events, while infrequent, do occur, and a large event like a burst of Gamma rays from the center of the galaxy would be very hard to anticipate.

HEMP
When a nuclear device is detonated in the upper atmosphere, it produces a series of effects that have a very large impact over a very large area. In fact, one single device could take down the electronics of the United States. It doesn’t even have to be a particularly high yield device, so long as it was detonated at the proper altitude (around 300 km). Specialized EMP nuclear weapons would be optimized to produce a large volume of Gamma rays to be mosts effective.

Much of what we know is speculative because we haven’t had any atmospheric nuclear detonations since the early 60’s which was a much simpler time electronically speaking. In addition, much of the research is highly classified, so there is a bit of estimation in all of this. We do have some data from some of these test that we can leverage, in particular the Starfish Prime detonation of a 1.4 megaton device 400km above Johnston island, which did some damage.

There are three phases to a high altitude nuclear EMP event. They are described as E1, E2 and E3. They all happen sequentially, but nearly simultaneously, and the combined effects are significant.

E1
When a nuclear explosion occurs, it releases a lot of high energy Gamma rays. These Gamma rays hit molecules in the atmosphere and scatter the electrons and often a proton in a process called pair production. This cascading effect releases a huge amount of electrons. These gamma rays have an energy of about 2MEV, some energy is lost when scattering electrons, so the electrons end up having an energy of about 1 MEV. These electrons travel perpendicular to the earth’s magnetic field at relativistic speed of up to 94% the speed of light. Because of this the mass of these high energy electrons increases to about 3 times their resting mass.

This creates a pulse at ground level that peaks at about 50,000 volts per meter or about 6.6 megawatts per square meter at the higher latitudes. Now if you think about that as the total electrical output of six large data centers in the footprint of a single rack, that is a lot of power. Now remember this is an induction current, so it needs something conductive to produce these high voltages. This would encompass most electronic devices. These power levels are achieved in nanoseconds which is much faster than any surge suppression or any TVSS device can react.

We call this the E1 portion of the EMP.

E2
The E2 portion of the EMP burst comes from the the gamma radiation produced by the neutrons from the nuclear device. The E2 pulse is much like a lightning burst, but unlike a lightning burst which is very localized, this is much more widespread. Because this burst is similar to a lightning strike, traditional surge suppression devices can mitigate it. This assumes that the surge suppression device was not destroyed by the E1 pulse, which is highly likely. That said, the E2 burst is of the least concern.

E3
The E3 pulse comes from the distortion of the earth’s magnetic field from the energy of the blast. Essentially, the earth’s magnetic field bulges out and then comes flying back, creating a very large electrical current. This discharge can also last for hundreds of seconds. Coupled with the fact that this follows the E1 burst which can destroy protection systems, it can be quite devastating. This typically impacts long haul cables and destroys components like line transformers. This is also the type of thing that happens during a very large solar coronal mass ejection which doesn’t produce E1 or E2 events. This is what takes out the electrical grid. To put some perspective on the damage, the lead time on these large transformers is often about a year, and that is assuming all of the transformers in the country don’t blow up at the same time. 

Impact
Instantaneously stopping all electrical devices across the countryman have devastating effects. Planes would fall out of the sky. People on life support would die.  Traffic lights would stop cars would shut down. SCADA systems that control so much of our infrastructure are likely very susceptible to the effects of an EMP burst and would fail. The fatalities could easily be in the millions within the first few minutes of the detonation. The bigger challenges come from the difficulties of restoring all of these systems. It would destroy all power production, water pumping, fuel production. Without vehicles, food and water transportation would stop. Most large cities only have a 30-60 day supply of food, so it could lead to mass starvation. Failure of sanitation systems and refrigeration systems would lead to disease outbreak. The point is it would be bad, potentially very bad.

So the question you might have is if the whole country is descending into chaos, why would I worry about my data center. The correct answer is you probably won’t be. One of the clients in this highly fortified data center I worked on was an insurance company, and their thinking was that at some point civilization would return to functioning, and they wanted the ability to help people settle their insurance claims.

Horizontal nuclear proliferation, where more small states have nuclear devices increases the likelihood of this scenario. A country that has only one or two nuclear devices would get far more out of their limited arsenal by launching an EMP attack than from bombing one or two cities.

Protecting your stuff
So how does one build an EMP shielded facility. As I already mentioned, an entire mountain can’t stop and EMP pulse. This is because rocks are transparent to these currents. What you need to do is enclose your equipment in a seamless contiguous conductive surface, and shunt all of that power to ground. People often know these as faraday cages, but these are very specific. As I mentioned previously, normal surge suppression devices don’t work in these scenarios. In typical surge suppression scenario, like a lightning strike, their is a build up of electrical potential that causes a breaker to trip. With E1 the entire surge is instantaneous. Additionally, it is a wide spectrum RF signal, so suppressing certain frequencies is not viable. 

Waveguide Image from https://www.kemtron.co.uk/

Copper is a really good conductor for this, but copper is very expensive, so galvanized steel is what is typically used. Doors have interleaved metal shielding and they seal like doors on a submarine. There must be no direct path for the electrons to travel into the facility. Copper ethernet cabling is converted to fiberoptic to enter the facility. To bring cooling in, Wave guides are used to allow airflow into the room while still blocking any direct electron path into the room.

Unfortunately there is only so much you can do to protect yourself. Even if you have all of your equipment sealed off, at some point you are going to be reliant on power or telecommunications equipment outside of your facility and outside of your control, but this is really a live to fight another day type of strategy.

What I know some of the experts in this field recommend is galvanized steel trash cans. When an attack is imminent, you can throw your electronics in there with some non conductive insulation, and they will likely be safe. Unfortunately, all my garbage cans around here are plastic these days.