Information Technology Reference
In-Depth Information
You may remember that in the U.S. a couple of bridges collapsed over the last
couple of years. A lot of our infrastructure is old. It was built after the Second World
War. So we do not really know how well it holds up. Can we equip our infrastructure
with sensors that can actually figure out before something happens that things are not
what they should be? We all know that events do not happen at once: there are some
signals. The problem is, most of the time we do not take care of the signals because
we are unaware of them. Sensors allow us to actually understand that, understand the
signals and again make the decision that actually needs to be made.
Now, this brings me to another personal example. About three or four years ago, I
got in touch, together with some of my colleagues, with Shell because Shell had a
problem. The pumps in refineries occasionally failed and it is enough for one pump in
a refinery to fail and you stop the whole refinery. And we all know we do not have
too much refinery capacity around the globe at the moment. So they were looking at
ways to actually predict when a pump was going to give up so they could repair it
during a preventive maintenance cycle. Today what they do is, after the pump has
been used a given number of years or months, they actually replace it. That is not the
most efficient use because some of those pumps still go on for a couple years but you
want to know when they are going to give up.
In the hope of solving that, we looked at a little device that we have created: a 3D
accelerometer, which is extremely precise, to actually look at the vibrations. We took
the whole Shell team to our labs in Palo Alto where the device had been created to
actually go and discuss that project. One of the guys that was present was the chief
technology officer from the company and as soon as he heard what the device could
do, he stopped and he said no, we are not going to use this device to check pumps, we
are going to do something completely different.
What have we done? Well, the area he wanted to look at was seismic exploration.
Do you know how you find oil in the ground? It is actually pretty primitive in a way.
What you do is you put a bunch of sensors around the place, you then come in with a
big truck, like the one that is at the top there, you know. The truck for about 15
minutes says “boom, boom, boom, boom, boom, boom, boom,” and you listen to the
return of those vibrations and depending on those returns and where they are, you can
actually understand the structure of the underground. That gives you a vision of
whether there is a possibility of oil being there. The problem is that the devices that
they use are about as large as an old telephone. They're called optophones, if I
remember right. They cost $3000 apiece. So, the maximum they can put in is about
100,000 of them and, by the way, they need to cable them. So think about 100,000
devices over ten square kilometer with cables. Think about the time it takes to set it
up, the time it takes to take it down, the weight and so on. And because you can only
put 100,000, you have a certain visibility. You only take a signal every 10 meters or
something like that. What he wanted to do was use that sensor, embed that sensor in a
device, plug it into the ground and put one every meter. So instead of putting 100,000,
put a million of them. Now when you put a million, forget about cables, so you need
to go wireless. Well if you go wireless what do I do with energy, what do I do with
data transmission? So we have together been reinventing a completely new way of
doing seismic data capture.
Search WWH ::
Custom Search