(I was reminded of this 2018 post by the massive power outage in Spain and Portugal)
No, this post is not about sex…sorry. Nor is it primarily about electrical engineering, though it might at first give that impression.)
The often-interesting General Electric blog (now defunct, following the breakup of GE into three separate companies) had an article about drones, linked to a cloud-based AI platform, which are used to inspect power lines and detect incipient problems–for example, vegetation which is threatening to encroach on the lines and short them out, or a transformer with a tendency to overheat. The article mentions a 2003 event in which an encounter between an overgrown tree branch and a sagging power line resulted in a wide-area blackout that affected 50 million people.
The inspection drone sounds like a very useful and productivity-improving tool: obviously, inspecting thousands of miles of power lines is nontrivial job. But the deeper issue, IMO, is the fact that one problem in one place can propagate over such a wide area and affect such a vast number of people. Power system designers and the people who operate these systems are certainly aware of the need to minimize fault propagation: circuit breakers and fuses, network analysis tools, and the technologies of protective relaying were developed, by GE among others, precisely for reasons of fault localization. But experience shows that large-scale fault propagation still sometimes does take place.
This problem is not limited to electrical systems. The mention of the tree-branch-caused 2003 blackout reminded me of a passage from the historian Hendrik Willem Van Loon:
Unfortunately in the year 1914 the whole world was one large international workshop. A strike in the Argentine was apt to cause suffering in Berlin. A raise in the price of certain raw materials in London might spell disaster to tens of thousands of long-suffering Chinese coolies who had never even heard of the existence of the big city on the Thames. The invention of some obscure Privat-Dozent in a third-rate German university would often force dozens of Chilean banks to close their doors, while bad management on the part of an old commercial house in Gothenburg might deprive hundreds of little boys and girls in Australia of a chance to go to college.
This probably overstates the interconnectedness of the global economy as it existed in 1914, but would fit our present-day global economy very well. (The author was talking about the origins of WWI, which he blamed largely on economic interconnectedness…not correct, IMO, but the war was largely caused, or at least reached the scale that it did, because of another type of interconnectedness…in the shape of alliances.)
I think much of the reaction against globalization is due to a realization, often-subconscious, that coupling can have bad effects as well as good ones…especially, a feeling that events are beyond the control of those who are most affected by them and have instead passed under the control of those with no “skin in the game”, to use Nassim Taleb’s terminology.
The advantages of interconnectedness are real: in the case of power grids, heavy loads in one location can be picked up by generators located in areas where loads are light–and local generator failures can be backed up by the entire grid. In the case of food supply, crop failures in one country are far less-devastating than they once were. A recent article on coffee described the complicated worldwide supply chain that is involved in bringing you your morning cup, ending with these words:
Globalization has downsides, and these Indiana steelworkers might be a casualty. But free trade has been a great boon to humanity. The global poverty rate has fallen, lifespans have jumped. Consumers have been given access to a dizzying variety of goods. As the Stanford economics professor Russ Roberts says, “We already tried ‘Buy local.’ It’s called the Middle Ages.”
That closing quote, though, ignores lots of other distinctions between medieval food supply and that of the present day…better fertilizers, irrigation, crossbreeding of plants, the use of power equipment, and lots else. Looking beyond agriculture at the statements about the declines in poverty rates, the longer lifespans, the availability of more consumer goods, it would be ridiculous to credit all of this to trade while ignoring mass production, steam and electric power, railroads, antibiotics, etc. (Yet I have recently seen people, in the context of the tariff debates, doing precisely that)
And while the advantages of interconnectedness are real, so are the disadvantages. Easy travel between countries facilitates the spread of epidemics…easy immigration can facilitate the spread of toxic cultural values as well as beneficent ones. Globalization of the food supply means that an American political decision to turn corn into ethanol causes food prices to go up for low-income people in countries far away. Globalization of manufacturing means that Chinese wage rates have a big impact on wage rates in Michigan and Indiana. And, in general, interconnectedness tends to take control out of the hands of those most affected and place control in the hands of those to whom the matter in question is more abstract.
The key to establishing and running resilient systems is the ability to benefit from interconnectedness while limiting the propagation of failures throughout the system. This issue does not appear to be well-understood by our political class.
I’ve read several stories about police being called to deal with extremely bad…even dangerous…behavior of schoolkids. The pattern seems to be that teachers and administrators are prohibited or at least strongly discouraged from doing anything about disruptive and hostile behavior…suspensions, expulsions, and the like…until it becomes so bad as to be a matter for the cops. Reminds me of those idiots who used to respond to a blown fuse by putting a penny in the fusebox, thereby keeping their toaster toasting, but possibly burning their houses down.
In our society today, we have by analogy too often not only put pennies in the fuseboxes, but have also bypassed the high-voltage circuit breakers that protect the transmission and distribution lines.
Finishing up the original post, I saw that GE had another piece on power grid protection, this one dealing with the problems of stability and the minimization of failure propagation. It gives some idea of the complexity of the issue…and this system, unlike economic and social systems, is largely deterministic.
Discuss.
Interconnectedness clearly has benefits. It also has costs. The issue is that the benefits from interconnectedness are certain, measurable, and here & now — the costs are uncertain
(might China refuse to sell goods to Walmart?), difficult to measure (because they have not happened yet), and somewhere in the future (if they indeed occur).
Human beings tend to discount the future. And that tendency is even more extreme among our ruling Political Class. So things roll along … until they don’t.
I have some notes I took on some related insights, for myself to work from later.
This was from https://accordingtohoyt.com/2025/01/09/the-illusion-of-knowledge/ and some of BGE’s comments there.
“The math around cascades is very interesting (ang least to me, I’m odd) and very applicable to (e.g.,) stock market crashes. What I have taken away is that you can predict whether there will be crashes (depending on the load and interconnectedness of the system) but not when or how big. Power grid failures are the paradigm case, but it’s very applicable.”
“Preference cascades are like any other cascade. essentially, it comes down to a system being tightly linked and overloaded. When it is, the failure of a single component can cause cascading failure, for stock markets, the load factor is financial leverage and the linkage is rehypothecation. Other cascades would have different factors, but the load and linkage drive it. That’s why X being free again is so important.”
I’ve been working a bit on preference falsification preference cascades, and a related behavioral cascade that is a little different, but I have not pulled up my notes for the last.
I actually already had the wiki page for ‘power-flow analysis’ up a day or two before the black out. You can linearize the system of equations by treating the AC currents as DC currents. But, obviously that costs you the phase/frequency information that seems to have been relevant for this black out.
The network theory of AC circuits is based a bit on ‘transmission line’ theory, which looks at two conductors and a time varying signal at some frequency. One of the key elements is there is an electrical length, scaled a bit by wavelength, which basically is the function that for one point in time shows you the points along that line that are max amplitude, adn also which points are zero amplitude.
One of the key results is the imaginary element of the signal and also of the source and the load are important. Frequency controls a lot of how much power gets reflected where.
So this is a bit tricky, and very important when you are both trying to understand how to manage an operating AC power grid, and also to understand precisely what happened in which order after a major failure.
The power-flow analysis wiki page wanted a citation for Newton-Raphson being the most popular solver for this application, which is a little silly. Newton methods are very easy, and are gonna be a pretty common first attempt.
Well, also, this is kinda a really sucky problem, and complex optimization breaks some of the usual alternatives. Some usual methods are on reals, and can tell they have gone too far when the difference switches signs. For complex numbers, norms are more valid, and Newton-Raphson is one of the methods which can use those.
Anyway, newton methods are a little related to newton fractals. IIRC, the page for newton fractals has a graph that relates to the optimization over sine x. That basically shows why newton methods over complex variables, for AC power network analysis, or for another problem, really really suck.
I’m not sure how that relates to certain more elaborate formula of complex variables, but it certainly shows one way that the infinite roots can show up, and why one needs to be able to freaking tell which root of those might be the correct ones.
Anyway, I hate this stuff a bit right now. But I would hate more being bored to the point that I would hit myself in the head with a hammer, so whatevs on hating the topic and hating my own incompetence.
Anyway, I don’t understand AC network analysis for the power grid case, so I can believe that a Spanish politician could understand less than I do, and could have added some critical extra bits of screw up, and Bob is your uncle. The known contributing factors include going to 80% renewable about two weeks prior to the incident.
The telegraph actually had a pretty good layman summary of how that influences the complex mathematics of the network, and how that could have contributed to the incident. (The physical inertia and electro-mechanical coupling of the French turbines are a time dependent effect, and so are basically also the same thing as the frequency/phase stuff that is contained in the imaginary parts of the network variables.)
We are now in the early stages of some sort of trust bubble collapsing involving some fo the trust that was networked through academic circles.
A lot of the selection of pure academics who have not previously been intensely studying these problems are puzzled by the fact that a lot of the public does not care about their woes, and would perhaps liek to see them hurt a bit more.
Me, I know that academics are way overdue for a reckoning, and I fear that fighting back now is mainly certain to make that reckoning worse for us. We best serve the public trust by fixing our problems early and fast, and where we have refused to do so, we are being corrected now by the populists. If we try to fight the populists, we will only make the trust correction to the endemic fraud into a bit more costly of an experience.
Compared to the ultimate costs of continuing as if, say, there hasn’t been a lot of uncontrolled fraud, stopping academic work entirely for some years is quite affordable. Much of the purported opportunity cost of not doing current projects is down to interested parties taking the book value of that proposal at face value. Having the academics with a monopoly on assigning official valuation for ‘educated’ academic work, and for ‘uneducated’ work is a conflict of interest, that is no longer really an avoidable problem.
A Carrington event — the damage caused by solar electromagnetic storms — was named after the guy who puzzled them out. In similar fashion, we should name the failure of industrial networks due to (entirely predictable) reliance on green magical thinking after the chief advocate of that magical thinking. We should call them Thunberg events.
Reminds me of the thesis of a book from the 80s called “Normal Accidents.”
The author got a bit carried away by his premise and really didn’t have a tight argument but yes, too much interconnectedness can be a worrisome thing, prone to collapse.
And utility companies can NOT afford to offer perfect reliability of electric power since few would pay the exponentially rising price of perfection.
You’ll get what you and the other customers are willing to pay for.
Hendrik Willem Van Loon….a loon indeed. I’m not a historian, but one would think the BIG message regarding interconnectedness that comes from 1914 is not economic but political. Don’t let some “damn fool thing in the Balkans” set the entire continent of Europe ablaze! It was the tightly interconnected system of treaties that lead to WWI. Once the first domino fell, everyone lined up and marched off to war – wrongly believing that it would all be over quickly.
The risks of interconnectedness were starkly revealed by covid, but have been laying in plain site all along. In many cases systems didn’t have enough redundancy built-in to cope with a breakdown in a single component. Well there is no free lunch — you either pay up front by bearing the inefficiencies of multiple suppliers, multiple manufacturers, strategic stockpiles, sourcing from multiple countries, etc. – OR you pay the piper in the case of a Black Swan event.
In my view the smart people pay up front (think of it as insurance) but politicians generally don’t want to do this (Case in point – LA Mayor Karen Bass and the Pacific Palisades fire) but are willing to gamble that nothing bad will happen on their watch.
“But free trade has been a great boon to humanity. The global poverty rate has fallen, lifespans have jumped.”
well, yea. it occurs to me that (to go clintonian for a bit) it depends on what your definition of “free trade” is. I suspect that a convenient definition is one in which there are **NO** barriers. Whereas what we’ve had since WWII, and what has become the default understanding of the term) has been more like trade w/o too many tariffs by the US and piracy-free trade routes facilitated by the US Navy.
Too much government money in energy. Let the market decide. For what it’s worth, solar works in limited and focused situations. I have a solar powered camera in a birdhouse connected to the internet and its great. A solar powered well in the middle of nowhere would be great. I put solar on my house and it works-but with the subsidies and the ability to sell back to the grid it pays for itself in five years. Longish payback
For an on demand, information economy, we need big big increases in the amount of electricity we generate. Only one way to do it: nuclear power (which is cheap). People say nuclear is expensive but it’s only expensive because of the regulations