A Thought Experiment

Thought Experiments are one of the more interesting but underappreciated tools Scientists use.  It has come into modern use through the Germans.  Their term "Gedankenexperiment" is literally a mashup of the German word for "thought" (gedanken) and the word "experiment".  The term and procedure first became popular among German scientists in the 1800's. Its international use became popular as a result of its frequent and very effective public use by Einstein.  But the concept actually dates back to the ancient Greeks who called the same process "deiknymi".

But okay.  None of us in the room are theoretical physicists.  So how it this relevant to us?  Like most scientific techniques anybody can use it.  And anybody can find it useful in surprising ways.  In fact, one of the principle attributes are the surprising things we can learn from a well constructed thought experiment.  And that's what I am going to do.  I am going to walk through the process of doing a thought experiment.  I hope to demonstrate that using your imagination, which is all a thought experiment really is, a disciplined use of our imaginations, can be surprisingly useful.

So what are we going to do?  We are going to build a large and complex piece of infrastructure.  But since we are only doing it in our minds it will be quick, cheap, easy, and generate no pollution.  And we can get a long way even though none of us really has the expertise to build the real thing.  Trust me!  It's going to be fun.  So what are we going to build?

Before doing that, let's take a digression and figure out why what we are going to build is useful.  We as a society have a problem.  Well, we have lots of problems but I am going to focus on just one.  We use a lot of electricity.  Most of it comes from "the grid", a complex and elaborate set of equipment that shuttles electricity from here to there.  In general the grid's job is to connect producers (power plants, etc.) to consumers (homes, businesses, etc.)  You can all relax.  I am not going to go into how all this works.  I am just going to note one thing.

The whole thing has to work instantly.  The producers have to produce exactly the right amount of electricity right now to meet the needs of consumers right now.  Handling this very difficult problem is extremely difficult.  But it has to be done.  Why?  Because batteries suck.  Fifty years ago they really sucked.  Now they only suck.  Anyone who has had their smartphone die because the battery has run out of juice knows what I am talking about.  Manufacturers are very aware of this.  If they could put a much better battery in, they would.  And its not a matter of cost.  A battery that is much better than the ones they currently use literally does not exist.

I am going to use "battery" as a generic term for anything that can store electricity.  In some cases the thing you use to store electr4icity is not literally a battery.  But for the purposes of this discussion I am going to call all electricity storage devices batteries even if they are actually something else.

So something that would help with the whole "instantly" problem would be to hook a giant battery up to the grid.  Then when you had extra capacity you could generate a little extra to charge up the battery and when you were short of capacity you could run the battery down to make up the difference.  That would make the job of the people who manage the grid much easier.  The problem is that batteries suck.

We know that the little batteries in our smart phones suck.  But they must be small and light.  So can we fix the problem with something that is big and heavy?  No!  Batteries suck.  Look at Teslas and other electric cars.  Why doesn't everybody buy an electric?  Well the obvious problem is that they are expensive.  Why?  Enough batteries to do a decent job cost a lot of money.  And they are heavy and take up a lot of space (not a problem in our "grid" situation but still . . .).  But Tesla has to do a lot of tricks to get their cars to go as far as they do.  And it takes forever to recharge them.  If you could "fill up" the battery in a car in the time it now takes to gas up then go 400 miles between fill ups (and the car was affordable) we'd all be driving electrics.  But we can't.  Why?  Batteries suck.

But we're still not talking industrial scale.  But the Tesla experience is illuminating.  Elon Musk, the Tesla guy, is trying to get into the electricity storage business using warehouses full of batteries.  But the batteries are really expensive and they can't store industrial scale amounts of energy.  Remind me again why it's a good idea to be able to hook a big battery to the grid.

Well, the cost of renewables has plunged.  Solar panel farms and wind farms can and do produce industrial scale quantities of electricity.  But they have a problem.  They are intermittent.  Wind farms can't produce electricity if the wind is calm or if the wind is blowing so hard the wind generators can't handle it.  Solar panel farms can't produce electricity at night or when it's dark.  And output is reduced by bad weather, the time of year, and other factors.

There is a clunky kind of solution.  Buy lots.  Then run only as many of them at a time as you need at that time.  That, for the most part, is what the electricity industry does now.  But this is inefficient.  You have to build two or three or possibly more times the capacity you really need.  This problem would go away if we had a good battery.  We could run everything all the time.  When we had more power than we needed we use the extra to charge the battery up.  When we are short we drain the battery to make up the difference.  If we have a good battery we need enough capacity to handle the average load plus a little more as an insurance policy.

So that's the problem.  We need a ginormous battery.  Now so far I have talked about "battery" batteries.  These are chemical reactions at heart.  That's why we refer to "lead acid batteries", traditional car batteries, or "carbon batteries ", old batteries for electronics, or "alkaline batteries", newer batteries for electronics, or "lithium batteries", modern batteries for electronics, cars, and (Musk would have you believe) industrial scale grid storage.  Is there another way?  Yes, of course there is.

This problem has been around a long time and smart people have been trying to fix it the whole time it has been around.  You can transform back and forth between electricity and other forms of energy.  So people have suggested using big heavy flywheels.  You use a motor to spin them up (storage) and you hook them up to a generator to run them down (drain) by hooking them to a generator.  And it turns out that compared to even a lithium battery flywheels work pretty well.  They are relatively cheap, we know how to make them, and they store a lot more power than a similar amount of lithium batteries.  But people haven't figured how to do flywheels at industrial scale.

Another idea people have had is to seal up a big cave.  Then you pump air in to raise the pressure (charge).  Later you discharge the compressed air through a turbine (first cousin to t jet engine) that is hooked up to a generator (drain).  No one has actually tried to do this at industrial scale.  There are lots of other ideas.  But, like flywheels and caves full of compressed air, people for the most part haven't figured out how to make them work.

So is there anything that has been tried and works at industrial scale?  Yes.  It goes by the generic term "pumped storage".  I live in a part of the country that has lots of hydroelectric dams.  You dam up a river.  Then you periodically drain the water through a penstock (a fourth cousin to a turbine) and hook that up to a generator.  It works great if you have a nice river to dam up.  And this has been done lots of times and works well.

There is a variation you can do.  What if you have a lake high above a river?  If you drain the lake into the river you can do the dam thing and make electricity in exactly the same way.  But eventually the lake goes dry.  But what if you use surplus power to pump water up to the lake when you have more capacity than you need?  Then you can keep the lake from running dry.  That's the idea behind pumped storage.  There's an example of this not too far from me called Banks Lake.  When there is extra capacity water is pumped from a nearby river up to Banks Lake.  When capacity is short they drain the lake through the same kind of setup that is used for a dam and electricity comes out the end.

So problem solved, right?  Unfortunately, no.  You need just the right setup for an installation like Banks Lake to work.  And there are only a few places where just the right setup exists.  As a result only a few pumped storage facilities have been built anywhere.  A Wikipedia article on the subject states that the total pumped storage capacity of the European Union is only 5% of total generating capacity.  And 97% of US "grid-scale energy storage" is pumped storage.  So at this time there is really no alternative to pumped storage when it comes to grid-scale, what I have been calling industrial scale, energy storage.

Enough already.  We are finally ready to start work on our thought experiment.  The Banks Lake pumped storage project is part natural and part artificial.  The river and the lake were provided by nature.  The artificial part, the pipes, pumps, generators, etc., had to be added before it would all work.  As a thought experiment, let's build a completely artificial pumped storage facility.  To do so we need to make some decisions.  but first let's talk about the givens.  We need a "high" reservoir that we pump water up to and a "low" reservoir that the water can drain down to.  These will be big water tanks.  We as a society know how to build big tanks so we'll just take it as a given that these tanks can and will be built.  Then we need the between machinery.  It will be the same sort of equipment that is used in the Banks Lake facility.  So we will also take it as a given that this machinery can and will be built.

So what do we have to decide?  We have to decide what the capacity will be.  I am going to arbitrarily decide that the plant will have a capacity of one megawatt-hour.  That means it can put out a million watts of power for a hour.  So how much is that?  My recent electricity bill says I used a little less than 1,200 kilowatt hours over a two month period.  That's a rate of consumption of roughly a kilowatt-hour per hour.  So out plant would be capable of powering about 1,000 homes like mine for an hour.  That seems like a lot.  But in 2010 the US had over 20 gigawatts of pumped storage capacity.  So our plant's capacity would be 20,000 times less.  Put that way. it seems like not very much.

But what I have in mind for my thought experiment is to come up with something that could be turned out in large numbers assembly line style.  It turns out there are about 50,000 wind turbines in the US and the average capacity is about 1 megawatt per turbine.  So our plant would be a close capacity match to one wind turbine.  Is that a good choice?  I don't know.  But it is a starting point.  And the nice thing about thought experiments is that you can easily tweak them.

So what else do we need to decide?    We need to decide on the height difference between the two tanks.  The height difference between the two reservoirs at Banks Lake is 280'.  I'm going to go with 100 meters or about 330 feet.  It is a nice round number.  Is it the right number?  I don't know.  But as it is close to the Banks Lake number it follows that the kind of machinery necessary to do the pumping, draining, generating, etc. is readily available.

If we know this then we can size the tanks, pipes, pumps, etc.  We also need some water.  But this is a closed system.  We move the water around.  But once the system has been loaded up all we need to do is replace small losses.  So we can't site our installation out in the middle of nowhere completely away from any water at all.  But once we have done the initial fill we only need access to a little water.  So lots of places can work.  And we don't need drinking quality water.  We are just going to pump it around.  We don't want the water to be so nasty that it rots the machinery.  But with the right kinds of paint and that sort of thing the water can be pretty nasty and still work just fine.

And we are building the whole thing from scratch.  We are going to put the high tank on a tower so we don't need dramatic landscape.  If we have dramatic landscape we can take advantage of it to reduce costs.  But even flat landscape should do.  The idea is to have a basic design that with little or no modification can be put pretty much everywhere.

A key item is how much it is going to cost.  And I don't know the answer.  But someone like a civil engineer who has experience with large construction projects should be able to quickly and inexpensively come up with a rough number, a "back of the envelope estimate".  And for our thought experiment that's all we need.

We are not going to actually build it.  We are just trying to answer two basic questions.  The first and most important one is "can it be built at all"?  The second question is "how much would it cost"?  And this second question is actually two questions rolled into one.  The first is "what is the construction cost"?  And that is a question I really can't answer.  The second question is "what is the operating cost"?  Based on operations like Banks Lake the operating cost, exclusive of the energy costs is "very low".  It should require very little effort to operate and the maintenance costs should be low too.

But this energy cost is important.  To answer it we need to know the operating efficiency.   The science of thermodynamics says that nothing ever operates with 100% efficiency.  There are always losses.  And that is true of pumped storage facilities.  Most of them seem to operate with an efficiency in the 70% to 80% range.  That is if you spend 100 kilowatts pumping water up you will get 70 to 80 kilowatts back when you run it down through the turbines.  So between 20% and 30% of the energy you put in will be lost.  But the idea is that excess wind farm capacity or solar farm capacity otherwise goes to waste.  If we use this capacity to charge our pumped storage facility  we will be ahead on costs in the end.

Given all the "I don't know"s we have racked up as we have worked through our thought experiment it would seem at this point that the whole thing was a waste of time.  But surprisingly it is not.

Our thought experiment has shown that there is a proven method for creating as much grid-scale energy storage capacity as we want.  That's good to know.  It has never been clear that enough chemical battery based energy storage could be built to make a difference.  The same is true of the flywheel, compressed air, and other approaches I have seen.  Knowing that a problem has a solution is valuable information.

And a civil engineer could quickly come up with a "back of the envelope" quality estimate for what such a facility would cost.  This number, whatever it turned out to be, also turns out to be useful information.  Let's say the facility would cost ten million dollars.  What that does is give us a benchmark against which to judge other potential solutions.  How much would a similar sized chemical battery facility cost to build?  If the answer is "a lot more" then we should forget about chemical based battery solutions.  The same thing applies to other approaches.  If it is obvious that they would cost a lot more they are not worth looking into further.

Now I just made the ten million dollar number up.  What if the number was actually a hundred million dollars or a billion dollars?  It is still easy to use whatever number eventually turns out to be the right one as a benchmark against which to measure other alternatives.  Certainly the lower the cost of our "back of the envelope', "thought experiment" design is, the worse it makes possible alternatives look.

And what if the number looks expensive but not wildly expensive.  Then it might just be a good idea to actually build one.  The cost of wind turbines has dropped dramatically as more and more are built.  The same is true of solar panel farms and many other things.  If it turns out that out that the rough estimate of the cost of our first facility is high but not completely out of the question high it may turn out that the hundredth or the thousandth one might be quite inexpensive.  So another thing this cost experiment does is give us a starting point for deciding whether the "artificial pumped storage" idea deserves a serious look.

And that's how it often goes with thought experiments.  You can figure out a lot without having to invest a lot of time, effort, and money.  And you often find out surprising things.  And you can easily imagine doing something that would be either dangerous or flat out impossible.  After all, it's all made up anyhow.  Scientists often ask questions like "if I was inside a worm hole what would it be like"?  Scientists who actually asked that question decided the answer was "I wouldn't know because I would be killed instantly".

So what that particular thought experiment told us was "don't bother even trying to figure out how to put people through worm holes because if you succeeded it would kill them".  In our far less dramatic example we can safely conclude that "there are better approaches than warehouses full of chemical batteries or flywheels or tunnels full of compressed air" for solving the grid-scale energy storage problem.  That's something that is important for the officials in government, industry, and the investment community to know when they are making the decision on whether to fund a project or not.

And thought experiments don't have to be technical or esoteric.  They can be things like a "what would it be like if I want to Mazatlán on vacation?" thought experiment.  This can be compared to a "what would it be like if I want to Paris on vacation?" thought experiment.  Or it could be applied to picking a car or deciding on the route you are going to take to work today or any number of other things. 

And the nice thing about a thought experiment is you are not confined to the practical or even the possible.  You just come up with a scenario and try to answer "what would happen" or "what would it be like" type questions.  Often a lot can be learned by getting an approximate idea of how things stack up.  And the specifics of the thought experiment can be tweaked instantly.  It's not like you are already in Mazatlán or Paris or wherever your thought experiment takes you.  In a thought experiment if you change your mind all you are out is a little time and effort.  And that's their beauty.