It's the grid, stupid!

“It’s the economy, stupid” was a phrase coined back in 1992 during a U.S. election campaign. It’s purpose was to keep people focused on big issues rather than being sidetracked by a sequence of lesser things. It’s kind of like when you are trying to deal with the health of a forest being cleared for cattle, but everybody is far too concerned with individual trees to notice; or perhaps about the latest Prince Harry gossip; or … [insert your favourite trivia here]. Adding the word “sStupid” was calculated to shock,; but it applies to all of us at times; too buried in our own obsessions to notice big things happening around us.

I wrote about this back in 2009. While Australia’s “green activist” movement was being born during the 1970s fighting to save 7 acres of bush (Kelly’s Bush) on Sydney’s wealthy North Shore, 50,000 acres of primary Rain Forest was being cleared for cattle by King Ranch along the Tulley River in north Queensland. I speculated that the Kelly’s Bush protesters may well have had BBQs and sandwiches with meat from those 50,000 acres. Why was this clearing in Queensland so relevant to actions in Sydney? Because of the central role of Jack Mundey and the Builder’s Labourers’ Green Bans in the Kelly’s Bush fight. Jack grew up in north Queensland, not far from the Tulley River.

People’s ability to be moved by a single example while impervious to a chilling statistic is a tool much used by journalists and other authors. It’s always good policy to start an article with a single example, even when your goal is broader and deeper. But a strategy morphs into a problem when you never get past those illustrative examples.

Poles and wires

So, I could begin with an example relevant to “the grid”. The $2.3 billion (900km) Energy Connect transmission line. It’s one example of what is required to integrate wind and solar plants into Australia’s grid.

And the chilling statistic? We need 10,000 kilometres of these new transmission lines.

What can you do with an example and a statistic?

Divide 10,000 by 900 and multiply by $2.3 billion. That’s $25 billion. That’Which is a pretty big investment to integrate cheap renewables, especially when you consider that transmission is probably less than a quarter of grid costs.

An Australian Competition and Consumer Commission (ACCC) report back in 2017 on the retail price of electricity found that 48% of the price was grid costs, often described as “poles and wires”, with just 22% of electricity’s prices due to the cost of producing electricity. So why are we so focused on the costs of producing electricity, rather than distributing it?

Because everybody understands solar panels. At the moment you can buy them and put them on your roof and save money. Who cares about the grid, that’s somebody else’s business. Our politicians love solar panels. Many were early adopters and enjoyed the big feed- in tariffs they legislated. Did anybody say conflict of interest? Probably not.

But a person’s decision to install solar panels has impacts outside their sphere of observation, in the same way that eating burgers has impacts on deforestation that are way out of sight and typically out of mind.

Could it be that everybody acting in their own best interests could totally bugger the grid and screw it for everybody?

It’s happened before.

We buy cars to optimise our personal travel, but they make cheap efficient mass transport impossible and we don’t notice that we have collectively broken our transport system until we are grid-locked in city traffic watching pedestrians overtake us; or until we measure the carbon emissions from building a vast network of concrete free-ways to try toand fix the mess.

What is the grid?

As simplifications go, calling our grid “poles and wires” isn’t too bad, but misses two really big components:; switches and transformers.

What’s wrong with ignoring these?

When you are dealing with high voltages, a switch isn’t just a switch. Most of us think of switches as like when somebody hot-wires a car (in a movie of course!). Take two wires and put them together or apart. That’s a switch. If you’ve ever had to start a car with jumper cables, you may have done it wrongly and seen a large arc of electricity.

With large voltages, things are rather different. Consider a 33,000 volt cable in a substation. Put two such cables within 35cm of each other and you’ll get an arc of electricity jumping between the two. To pull them apart, meaning to open a switch, or disconnect power, you need to separate wires by substantial distances. And if your switch is outdoors (most are), the gap an arc may jump is much bigger, depending on the weather.

Cables in outdoor settings need metres of separation to prevent arcing. How do you make a switch that separates wires by big distances quickly without huge arcs? This is part of the stuff that engineers spend years at University learning. Some go on to dream about making a perfect switch!

Switches wear out. What causes them to wear out? Each time the switch operates, it suffers damage; and they corrode. There are counters on substation switches to record the number of times they have operated; so that grid operators know when to replace them. More switching means more maintenance in the part of the system responsible for 48% of costs.

Switching gaps can be reduced if you have something other than air in the space between the wires. Gas-insulated switchgear is common in Japan because less land puts a premium on small switchgear. It’s also common in offshore wind farms because of the harsh environment.

Have a look at the picture below. It looks like plumbing, because it is. Except that instead of water, it contains gas. The gas is SF6, a (non-toxic) gas that which is tens of thousands of times more potent as a greenhouse gas than carbon dioxide. This is something your really don’t want to leak! If you think this looks more expensive than your local substation (assuming you’ve ever noticed it), then you’d be right.

Gas- Insulated Switchgear

The switches in the image above are a small portion of those inside this yellow structure in the North Sea:

Dolwin Kappa power hub

The yellow building connects to the coast via a 90 km pair of 320kV transmission lines, which will carry the output of the 900 MW Dolwin6 wind farm. I’m trying to give you a feel for the scale of modern “poles and wires”; massive doesn’t begin to capture it.

The above yellow building is Dolwin Kappa, you can find it on the lower right in this image; it’s one of many such structures.

Wind power plans for North Sea

Aside: It’s amazing how many environmentalists rage against deep sea mining and yet are quite comfortable about the level of destruction offshore wind turbines and their associated mass of cables require. It’s amazing how many environmentalists reckon we can deal with climate change by “returning” to some simpler less material existence. It’s easy to view an onshore or offshore wind turbine shimmering and idly turning in the distance and forget the massive industrial complex behind it. The (currently) largest wind turbine, the Chinese CSSC H260-18MW is just 40m shy of topping the Eiffel tower and the tips of its blades are typically travelling at over 300 kph.

Switches are used in the grid for the same reason they are used everywhere else. To stop or start electricity flows. Transformers are needed to change the voltage up or down. The power coming out of a power station has whatever voltage it has, usually quite low. It is increased to tens of thousands of volts before being sent out, because high voltage reduces energy loss in transmission lines. Another transformer drops it down as it gets closer to its destination.

Power might pass few a few transformers before it hits it’s final delivery voltage of 240 Volts at your wall socket (in Australia).

Electrical generators work by spinning a magnet and the resulting electricity has voltage levels that are naturally oscillating; up and down, alternating between positive and negative. This is called alternating current (AC). Transmission cables have thus always carried AC. It’s long been known that if you convert this electricity to a form where the the voltage is constant, then the energy lost in transmission can be reduced.

The problem with DC transmission has always been cost; it still is. There are three parts to the problem. Convert to DC, transmit along cables, convert back to AC. If the cables are long enough, then you can save enough energy to pay for the conversions. The distance at which DC transmission is now cost-effective is about 500 kms. The cost effective distance in the ocean is about 60 kms for some really complicated reasons (see Appendix).

Wind and solar power are intrinsically linked to geography and require long transmission lines. But while DC transmission can make such transmission feasible, it will never be cheap. As we saw above in reference to the new transmission connection between SA and NSW. This connection is 900 kms, but it is still AC. That’s because it doesn’t just connect two points, but several intermediate points.

So while DC transmission (typically high voltage or ultrahigh voltage DC) enables long distance transmission with low losses, it isn’t cheap and will still cost you a bundle of money to use the cheapest form of energy generation. And not just money. Wildlife pay the heaviest price, with millions of birds killed annually.

Ok, so having transmitted your electricity to somewhere near your customers, the final stage is called distribution. Distribution is complex. This kind of low voltage electricity can be transmitted in underground cables, but the cost is considerably higher, so poles and wires are more common. You can’t put the high voltage lines underground for the same kinds of complicated reasons I alluded to with regard to undersea cables from wind farms.

The transformers at the end of the chain are constantly adjusting the final output voltage, depending on the demand. Back when life was simple, this would just depend on when people decided to cook, or perhaps turn on something for heating or cooling. But stick a lot of photovoltaic (PV) panels onto roofs and the final transformer will need to change its output more often and over a wider range.

Think about it and look at the following graph of the demand over a few days in South Australia; the Australian state with one of the highest renewable penetration rates in the world.

South Australia, Friday 13 2023

If graphs aren’t your thing, then take a deep breath and hasten slowly.

Left to right across the page are just over three days' worth of data. The highest yellow peak is for Friday, the 13th of January.

The up and down waves are the rises and falls in electricity demand, measured in mega-watts (MW); the peak demand was at about 4pm on each day; Friday and Saturday were particularly hot but the temperatures aren’t marked.

The area of each colour gives you the total energy from each source. Think about a rectangle. It has a height and a width. Multiplying them gives you the area. On this graph we have a height and a width, but the height is measured in megawatts, and the width in hours (which aren’t shown). Energy is power multiplied by time; height x width, so it’s the area. One kilowatt-hour of energy is what you get from operating a one thousand watts power source for one hour.

South Australia has 1.7 million people, so it’s a smallish, but non-trivial, grid. The colours tell you how much energy is coming from which source. Yellow is solar. Green is wind power. The grey shaded bars are night time, starting at 10pm. The data comes from OpenNem; they decided on 10pm, it’s actually dark by 9pm at the moment in South Australia.

Compare Friday 13th of January with the day before.

On the 13th, it didn’t drop below 36 degrees between 4pm and 7pm. But there wasn’t much wind until almost 10pm, so the demand at 7.30 when the solar had dropped had to be supplied by gas. You can see the gas colours underneath the green. There are four types of gas generator.

Why so many?

Three of them, open-cycle gas turbine (OCGT), Steam and Reciprocating, are cheap, simple, and not very efficient. combined-cycle gas turbine (CCGT) is the most modern highly efficient technology. Trace its brown ribbon back to the previous days. CCGT is clearly the gas technology of choice. The others get used when the grid is desperate and will use any bloody thing which works; regardless of cost, efficiency or pollution. You can see the desperation level in the little bright red blip … we even had to burn diesel to keep the grid up; not to mention our “big” (110 MW) battery.

Regardless of this desperate effort, South Australia would have crashed and burned without importing electricity from Victoria.

Eventually, it got much windier and the gas was turned (mostly) off.

The Australian Energy Market Operator (AEMO) Renewable Integration Study: Stage 1 (RIS) has a special appendix dealing with these kinds of surges in demands, called, appropriately, ramps. When the South Australian battery was installed in 2016 it was famously, the biggest in the world. The graph shows how utterly trivial battery storage is in relation to firming up our grid. Globally, Li-ion battery capacity is expected to hit about 2.5 TWh per annum by 2030, and most of that will go to electric vehicles (EV). The EU gas storage capacity is some 1,119 TWh and they are worried that it isn’t enough! There may be an alternate universe where batteries can firm up a non-trivial grid, but it isn’t this one.

Think about it and you’ll see that the ramps will get steeper as renewable penetration increases. Steeper than anything which any grid has ever faced in the past.

Any ramp is accompanied by a frenzy of switching. A grid is naturally partitioned into high and low-voltage regions. In the low-voltage region, there used to be no generators. Meaning the disconnection was simple –; flip a switch. Think about a house. Disconnect it from its supply and the wires are safe to touch. But if you have solar PV on your roof, you need to disconnect that also. A house is on a street and you used to be able to disconnect a street at the point where its power arrived. But with generators on the street. That is no longer so simple. Fixing this requires changes to switches, transformers, and fault isolation mechanisms.

There’s a nice little table in the AEMO RIS which lists grid problems. One of them is “tap settings”. The setting on a transformer is called a “tap”. Transformers in many places are getting too hot, because they are being used more than they were designed for. Ditto the switches and the rest of the grid.

The issue isn’t “Can it be done?” –, it can. The issue is why rebuild the grid when we don’t need to? We have more than enough jobs to do to hit net zero, so why take on a job we never needed to? Of course, if building grids is how you make money, then it’s a great move. It’s also terrific fun for engineers who love solving tricky problems.

Is that all?

So far, all we’ve seen is that the entire grid has to be rebuilt, and rebuilt with more metal and heavier duty switches when we add large numbers of renewables. This is pretty much what you see when you look at AEMO’s RIS and the host of problems at every level.

Let’s think about a much simpler problem. Not operating a grid for 12 months with 100% renewables, but running it for just one second.

AEMO recently released a report on this problem, its – Engineering Roadmap to 100% Renewables. This is actually about this much simpler one-second problem; a precondition for the much harder problem. You’d think the easier problem would be easy, but according to AEMO:

“Operating a gigawatt-scale power system at 100% instantaneous renewable generation is a feat unparalleled worldwide.”

Nobody has done it.

It’s not that South Australia doesn’t have enough renewables to supply demand when it’s sunny and windy. We do. But the risk of cascading failures due to inverter behaviour and sudden losses of power is too great. So AEMO always maintains a minimum supply of reliable generators. Any excess from renewables is removed.

In South Australia, grid operators disconnect PV panels when the level of PV is too high. Any inverter installed in South Australia since 2020 must allow remote disconnection. This has already happened a few times, most recently back in November 2022.

Power systems are not plug and play

Many people are familiar with the plug and play concept from computing.

For a plug and play device, you just plug it in and it works. This is possible for small devices, but not so easy with big ones. You don’t need to be an engineer to buy a bunch of computers and some Cat 6 cable and cobble it all together into a network.

Electrical grids are very different.

What’s the problem?

Transients.

The word denotes things that come and go, but in electrical power systems they are a little like Facebook, they move quickly and break stuff.

When you plug a device in, a surge of current floods into the device, creating a surge which your device has to be designed to handle. That's a transient. Plugging a solar or wind farm into the grid and you get a similar surge, but larger. How large? Before answering that question let’s think about lightning. When lightning strikes a power line, it sends a surge down the line and the line and all attached equipment has to be designed to handle it. That's also a transient, but it's massive. How does a computer handle a lightning strike? It doesn't. It will break. Grids get them on a regular basis and survive.

Computer software is used to run the calculations to estimate lightning surges and their impacts. It is used every time you design the interface between a solar or wind farm and a grid. You don’t just plug a standard solar farm plug into a standard grid socket. That doesn’t work; each connection is unique and the impacts of lightning and switching have to be calculated specifically for each one.

The software to do this is based on some very complicated maths which models some even more complex physics. How complex? The software used by AEMO is called PSCAD/EMTDC. PSCAD stands for Power System Computer Aided Design and EMTDC stands for ElectroMagnetic Transients including Direct Current. EMTDC is what’s called the backend. This is jargon for something which is like a crew of really smart slaves you can call on to do the intellectual grunt work when you have a problem. It’s like being able to write a set of equations for some problem on a piece of paper and hand it to somebody else to solve – the team in the back room. They do the work and hand you back the solution. EMTDC is like that. PSCAD allows you to define the problem and EMTDC solves it. The first code for EMTDC was written in 1975. It’s written in fortran!

Fortran was released as a computer language in 1957.

You won’t find Fortran in anybody’s list of top 20 programming languages, but you’d be surprised how much Fortran is still kicking around in the backends of really important software.

If a piece of computer code got the right answer to a mathematical problem in 1975, then it will get the right answer today. This isn’t Instagram, fashion or politics, this is maths.

Writing this kind of software is really hard, so there is plenty of it still in use. The latest Fortran standard was released in 2018 and there are various well- maintained compilers which support it, including versions from both Intel, GNU and NVidia. High-end gaming machines or bitcoin miners often run on NVidia chipsets and NVidia has a Fortran compiler optimised for highly parallel algorithms to run on chips which implement its popular CUDA architecture. Even in the world of computers, where it often appears that change is rapid, complex algorithms are tough to develop and working code can live for decades.

If you want to connect a solar farm to the grid, for example, you would use PSCAD/EMTDC to ensure that your design meets requirements. The software would calculate the surges through the cables during connection and operation. Of course, you need to know the length and type of the cables together with the electrical characteristics of the devices. The backend knows the physics and can solve the equations for the transients; the surges due to a connection to the grid or a lightning strike. This may sound incredibly complex, but my simplified description is just that, very simplified. The appendix has a pointer to just how much deeper down the rabbit hole you can go if you are after the truth.

How many web browsers can you name? Chrome, Edge, Opera, Safari perhaps. A web browser is an extraordinarily complex piece of software, but there are dozens of them. Again there is the distinction between the front end that you see and the backend that does the harder stuff; there are far fewer backends (just as with EMTDC).

How many pieces of power system software are there like PSCAD/EMTDC?

Not many. There is a list here, and it’s much shorter than the browser list, and they are not all full-featured products capable of handling large state or national grids.

You may have heard about the Australian Stock Exchange writing off $255 million in its failed attempt to rewrite a piece of stock managing software. This was just software that kept track of things. The things themselves didn’t interact with each other in complex physical ways, so in one sense, this kind of software is pretty simple compared to power engineering software. But there is a long history of software projects failing, despite being conceptually simple. Consider, for example, the US Air Traffic system. All it has to do is keep track of planes. How hard can that be?

Unlike stock bids, planes move and signals get lost. So planes are obviously harder, but how hard? In 2014, one part of the US air traffic system crashed and hundreds of flights were impacted. And that was only about $2 billion worth of the full system, which is much bigger, and has been under construction for years and is expected to be finished by about 2030.

How much harder is the redesign of our grid to handle millions of renewable components? The AEMO Engineering Roadmap to 100% Renewables (ERM100) report describes it as:

“Progressing this uplift, while simultaneously operating a real, gigawatt-scale power system is akin to “rebuilding a plane while flying it.”

“Progressing this uplift” is just bureaucratic jargon for “Making these changes”. The description is quite inaccurate. It’s far more accurate to describe the changes as being like turning a plane into a fleet of helicopters …; while flying it.

So what exactly is it that makes renewables hard? Has the physics changed? No. Obviously, given the example I used above, PSCAD/EMTDC can handle solar farms. So what can’t it handle?

AEMO Operations Technology Roadmap

The AEMO Operations Technology Roadmap lists a number of tasks that need to be done to upgrade a grid for renewables, as well as some of the roadblocks to getting them done. I won’t (and can’t!) explain all of these, but the list is formidable:

1. Lack of dynamic models for distributed energy resources (DER), distributed photovoltaics (DPV), battery energy storage systems (BESS) There are models for calculating transients when you connect a big thing like a solar or wind farm, but connecting clusters of hundreds of thousands of things? No.

2. Lack of EFCS modelling (emergency frequency control system) The current systems may work wrongly with increased penetration of rooftop solar.

3. Lack of participant dynamic response and secondary control systems. The current grid is centrally controlled and directed. Tools to control a grid with high levels of generation on rooftops are missing or primitive.

4. Lack of important operational data.

5. Multiple models are managed by different teams.

6. Different functions use different model variations.

As the report proceeds, it lays out specific software tools which are necessary but either totally missing or deficient:

1. Tools to predict stability to provide operational management tools … note that this is two complex tasks, not one.

2. Tools to predict/detect the behaviour of IBR actions

3. Weather tools not aligned with weather-dependent generation. You need to know the weather at a solar farm, not at some weather station located somewhere else for historical reasons.

4. Limited ability to identify oscillations and other emerging forms of instability. This is deep magic, and my understanding is far too limited to try and explain it.

5. Missing pathways for telemetry and control signals to and from assets and AEMO

6. No frequency security tool in place.

7. Volatility can be driven by unpredictable market prices and responses.

8. Emerging challenge … “Decline in model quality, while system operation becomes increasingly complex”.

9. Increasingly computationally-intensive studies. Limited high-performance computing capability.

10. Develop tools for small signal stability assessment in high IBR systems

11. Develop online monitoring and dynamic security assessment tools for converter-driven instabilities.

12. Update existing tools VAr Dispatch Scheduler (VDS) and Voltage Security Assessment (VSAT).

All up, there are calls for 23 new software tools and 10 enhanced tools.

Does that sound harder than keeping track of share transactions? Just a tad.

All over the world people are coming up with lists like this. If you are a software developer, you’ll be seeing big dollar signs.

How many of these projects will prove to be simple and how many will end up floundering? And which projects will be in which category? The one thing I am quite sure of, is that the ASX failure will have plenty of company from grid developments if we continue in the direction we are travelling.

On the other hand, a grid with a significant nuclear component and a small rooftop solar component is just a normal well understood traditional grid. Except that it’s more environmentally benign and wildlife-friendly than anything with plenty of wind, solar and batteries.

Appendix: Cables and electricity

It’s convenient to think of electricity flowing through wires, and it’s a very useful model. But it’s a model that gives you wrong answers in many situations, and the “wrongness” rises with voltage!

Physicists use much better models based on electromagnetic (EM) fields that surround wires and interact with whatever is in the vicinity. These are described by some simple-looking equations which are profound in their implications: the Maxwell-Heaviside equations. In a sense, the wire guides the fields rather than transmitting anything directly. High voltage transmission cables are a long way off the ground because of what happens in these EM fields, not the wires. These fields provide massive challenges for all underground and undersea cables.

If you want a glimpse of the challenging world of EM engineering, with some terrific pictures, animations and physical examples, but no real mathematics, then check out Derek Muller’s great youtube clip: How electricity actually works. Perhaps I can update Shakespeare a little: there are more things in these equations, Horatio, than are dreamt of in all the imaginations of the world’s playwrights.