Understanding How Our Grid Operates
Physics and Basic Operation
The US has three "grids." They are the Eastern Interconnect, Western Interconnect, and ERCOT. In technical terms, they are AC wide-area synchronous grids. Let's unpack that.
AC is alternating current. AC is what comes into our home and powers many of our appliances. In contrast to direct current (DC), AC "changes the direction and magnitude" of its voltage and current. How fast AC changes direction is its frequency. Synchronous grids share frequency but also the timing (phase). Wide-area grids can span whole continents.
US Interconnects; Source
Virtually all grids are AC. 100+ years ago, it was easy to make AC by spinning a generator, and analog transformers could change the voltage. Houses need low voltage, but low voltage requires large wires to transport electricity efficiently. The solution was to generate electricity in large power plants and transport it to users at high voltage, then step it down outside your house.
A wide-area grid like the Eastern Interconnect is similar to 20 water tanks connected with hoses. Each water tank is a vertically integrated monopoly utility like Duke Energy or an independent system operator (ISO) like SPP or PJM.
ISOs in the US; Source: Wikipedia
Each ISO or utility is responsible for balancing the water level in its tank as power plants add water and customers remove it. If there is a problem in SPP - say a cold snap spikes demand and trips power plants offline, the water level starts to decrease. The lower level creates a potential energy difference between SPPs tank and next door MISO. So water starts flowing from MISO's tank to SPP. MISO plants will try to add more water to maintain their level. More water will start flowing in from other tanks like PJM if MISO can't keep up. Hoses cost money, so there is only so much water that can flow between the tanks.
I'm simplifying concepts around voltage, frequency, and current, but the grid generally works this way.
There are many benefits to wide area synchronous grids. If one territory has problems, they automatically start getting help without any permission or human in the loop. Physics makes it happen instantly. The connections for this help are just wires.
The negative is that tail consequences go up. Water levels do not fall gracefully. If they fall below a line, the bottom drops out. Every other tank is attached to a black hole, pushing them towards the critical threshold. The reason is that out-of-specification frequency will damage equipment so that equipment shuts off to protect itself. Cascading failures can lead to wide outages from relatively benign beginnings. Operators and utilities across the grid must have tight cooperation to prevent these events.
Grids Favor Big, Traditional Power Plants
Economies of scale, maintenance requirements, and demand patterns lead to wide-area grids.
Several factors explain why power plants benefit from scale. First, they are complex construction projects. It takes years to build traditional power plants. Increasing the plant size tends to lower construction management costs as a share of the cost. Second, these plants are full of pipes and vessels. For cylinder-like shapes, the volume increases faster than the surface area. Bigger pipes, boilers, and turbines use less steel per watt and have similar maintenance requirements.
Thermal power plants (nuclear, geothermal, gas, coal, etc.) require weeks-long maintenance outages every year or two. Large grids spread output loss over many plants. France's electricity system would be untenable if they weren't in the European grid because their nuclear fleet can have issues that take many plants offline at once. Luckily they can lean on their neighbors to help supply electricity in years like 2022, when the nuclear fleet may only have a 60% capacity factor due to unplanned maintenance outages. Extended maintenance outages can cause issues when unexpected cold or hot weather happens. During winter storm Uri, Texas's ERCOT had almost 10 GW out of its 90 GW capacity offline for maintenance and repairs.
Electricity demand is highly variable by location and time. Distributing load over a wider region reduces the required power plant capacity by sharing resources.
Extensive grids minimize the investment in power plants.
The Grid and Emerging Technologies
The math for wind turbines, solar farms, and batteries differs from traditional power plants.
Wind sees even more benefit from wide-area grids than traditional power plants, even though an individual turbine is <1% of the output of a large power plant. The wind is likely to be blowing somewhere, but that somewhere isn't likely to be where the load is! Wind developers dream of mega transmission projects to take wind electricity from the Great Plains to population centers in the east.
Solar cells and battery cells are inherently low voltage electrical devices. They are flat and limited in size to minimize resistance. Even cylinder car batteries have flat electrodes rolled up and put into the can. The scaling effects are mild. Construction overhead is much smaller than traditional power plants because they are modular and easy to install. Solar insolation is much more consistent than wind speed, so the economics favor local generation. Solar farms require less maintenance than thermal power plants and rarely need to go offline. The grid is less beneficial for solar than other technologies.
The Difficulty of Change
Adding New Generators
Every power plant affects how electricity flows through the grid based on size and location. It is theoretically possible for a wind farm in South Dakota to impact transmission infrastructure near Chicago. The utility or ISO must study each new facility proposal to ensure there won't be issues and recommend grid upgrades to mitigate potential problems.
Vertically integrated utilities run studies and manage new facilities within their organization.
ISOs managing wholesale and transmission markets are different. They are medium-sized organizations (usually <1000 employees). There is limited engineering horsepower to model new power plants. That worked relatively well until the falling price of wind and solar swamped ISOs with new applications. Many like SPP or ERCOT have as many gigawatts in their interconnection queues waiting on studies as the entire ISO installed capacity.
The result has been an unfortunate spiral. Many proposed projects trigger infrastructure upgrades that cost more than the power plant. It hasn't been clear to developers which projects will incur these costs, so they spam the ISO with many applications. The excess applications mean it takes longer for the ISO to run studies. Attempts for ISOs to increase staff have struggled as developers poach engineers that can estimate what grid upgrade costs will be.
ISOs and FERC are proposing changes to address this. The first is to move from "First come, first serve" to "First ready, first serve." Projects that have land and customer agreements go first. But developers don't want to spend real money only to get hit by massive upgrade fees. They are pushing ISOs to generate visual data on where the grid has excess capacity and where it is constrained.
Even under the status quo, ISOs will eventually hire and train enough engineers to meet their own needs and fill the ranks of developers. The result will be power plants built where they best integrate with the grid and less spamming of requests. All participants welcome improvements that speed that process up.
Customers Increasing Usage
Lower voltage distribution systems also have problems. Any customer that wants to start using a lot of electricity must wait for a utility study. If we extend our "water tanks" analogy, a series of smaller tanks and hoses connect to customers like capillaries. If one customer starts charging fifty electric trucks, they can knock these smaller systems out of balance because the hoses from the big, high voltage tanks aren't big enough to send enough power. The utility has to determine what upgrades are needed to serve the load. It can take years to complete, even at more efficient utilities. Customers will often pay for part or all of the upgrade.
Who Pays for What?
The advantage of the grid - pooling resources, electricity freely flowing where it's needed - is also its major disadvantage. It becomes difficult to decide who should pay for what. The public nature of the investment means politics are heavily involved.
These issues were why most utilities were vertically integrated monopolies when the US electrified 100 years ago. It solved the "who pays for it?" problem. The utilities were either co-ops or investor-owned to take the massive CAPEX required off the government books.
In the 1990s, governments felt that the guaranteed rate of return utilities earned was too high and that new technology would allow wholesale markets to function successfully. Most US electricity now trades in wholesale markets. The Southeast and parts of the West are exceptions. Monopolies still handle distribution to customers. There were some hiccups along the way, but the process is generally considered a success.
These markets are undergoing massive transformations because of the change in the underlying economics of electricity generation. They aren't free like the No. 2 pencil market. There are thickets of rules coming from the Federal Energy Regulatory Commission (FERC), states, and individual ISOs. The rules are arcane and impenetrable in a way few topics I've come across are. Many small electricity producers using inverters instead of spinning generators don't fit well within the existing framework.
FERC, states, and ISOs have been incrementally grinding through problems like settlement periods, integrating demand response, and frequency management. The biggest issue is how to pay for upgrades to high voltage main lines. Most ISOs require the project that triggers upgrades to pay. It is like the straw that broke the camel's back where a 50-megawatt project might be responsible for a mainline improvement that could unlock gigawatts of transmission. The project gets canceled because it can't afford to pay, and the developers see it as unfair.
The solutions have mainly been at the state level. Texas financed large transmission lines to bring wind from the West to its cities. Oklahoma passed laws to create complicated formulas to pay for upgrades. States with high wind penetration are usually further along. Some ISOs and FERC are considering a "cluster" approach where groups of applicants receive approval together and share upgrade costs. Each state and ISO will have to find its way.
What is the Smart Grid?
There is new hardware and software, but their usefulness varies.
The Tools of the Trade - Transmission
Cheaper Digital Valves
Some transmission paths can starve because electricity flows across the grid based on physics rather than human control. The line to a node with higher electricity prices might have more resistance to electricity flow than an alternative (the connecting hose is too small). Equipment can create resistance (or reactance for AC) on less desirable paths to force electricity to flow towards starved ones.
Monitoring Line Conditions
Utilities often have little data on real-time line conditions. Sensors that use LIDAR and electromagnetic sensors can determine line conditions. Wires get hotter when they move more electricity. That heating causes expansion and sagging in the line. Operators can use LIDAR to see the current load based on sag and push more electricity through under certain conditions, like high winds cooling the line.
Topology control means optimizing the settings of the grid to move more electricity. Many breakers or safety systems aren't dynamic because of difficulty understanding grid conditions and how changes propagate across the grid. More powerful computing can change settings on the fly to allow more capacity. Some estimates are that grid capacity could increase by 30%.
Wires, known as conductors, are usually aluminum and steel. New carbon core conductors can handle higher heat with less sag, increasing capacity by as much as 25% and lowering line losses. Many existing conductors are nearing their 70-year expected life. Replacements could increase grid capacity without new unpopular transmission lines and pay for their higher upfront cost through better efficiency. Other systems like substations may need upgrades to unlock extra throughput, making this a long-term play as equipment reaches the end of life.
Variable Frequency Transformers (VFTs)
VFTs link different AC wide-area grids without requiring the grids to be synchronized. There are several projects in service, but the high cost and the need for cooperation across multiple grids limit growth. The highest-profile project is known as Tres Amigas. The idea is to link the three main wide-area grids in the US near New Mexico, but the project is on life support after backers lost interest.
High-Voltage Direct Current (HVDC) Technology
Back-to-back HVDC stations can serve the same purpose as VFTs but are more expensive. Long-distance HVDC transmission lines have fewer electricity losses from resistance than AC lines. There have been many proposals for HVDC transmission to bring wind from the Great Plains to population centers further east. They've almost all failed after local resistance. The technology has seen more success in Europe, Brazil, and China.
Integrating Inverter-Based Resources
AC electricity frequency is set by how fast generators spin. Hydro, geothermal, nuclear, gas, oil, and coal power plants create work to spin a generator to produce electricity. Power plant generators turn at the same speed to remain in sync. These machines are relatively slow to adjust output, but the mass of the generators has inertia that helps balance the grid in milliseconds.
Wind turbines, batteries, and solar panels are inverter-based resources. They have no inertial response. But inverters are inherently digital, so they respond much faster than traditional turbines can ramp up and down. Many new battery installations have software and inverter designs to provide an inertia-like response.
The Tools of the Trade - Distribution
The main goal of the "smart grid" in the low voltage domain is managing customer demand. Information only flows one way in a classic analog grid. Operators can only balance the grid on the supply side. Maintaining resources to meet peak demand is expensive, and many power plants run <10% of the time. Adding information and incentives that allow the demand side to make adjustments can reduce the cost of supplying electricity by reducing the need for peak resources.
Digital electricity meters (smart meters) are the first upgrade utilities make. Meters reduce OPEX by automating meter reading and service starts/stops.
The next step is usually a connected digital thermostat. Space cooling is the driver of peak demand in most of the US, and the thermostat can relay information to the customer like the current electricity price in Time-Of-Use pricing schemes. Electric water heater controls have successful deployments, too. The main issue with these classic cases, especially space cooling, is that customers value AC and hot water, so an interruption in their usual schedule is unwelcome. Programs usually see fewer signups than hoped.
Most grid-tied rooftop solar installations in the US are "dumb." They run the meter backward when solar output exceeds household usage and dump electricity into the low voltage distribution network with zero control.
The next generation of technologies involves electric vehicles, batteries, and control software. They represent an opportunity to increase sources of flexible load and distributed generation:
Electric Vehicle Charging
Electric vehicles have much bigger batteries than what daily driving requires. Cars are only in use 5% of the time. There is significant latitude on when EVs charge. The grid sees massive benefits if cars charge off-peak and avoid charging during peak demand.
Big Hot Water Tanks
Daily hot water usage is relatively consistent, and hot water is cheap to store. Heating water makes up around 20% of a (non-EV) household's energy. Heating can be done when electricity is cheap and reduce demand during peak periods.
The Cut-Off List
Vehicle charging and water heaters require detailed information exchange between household appliances and the utility. That coordination is not free. Most utilities are switching customers to smart meters that allow remote disconnection of electricity. Utilities could enroll households willing to lose power for several hours at a time for a small payment or credit. Presumably, these customers would have generators or batteries to tide them over. A process like this would be simple for utilities to implement, still provide a demand response resource, and captures many benefits of variable electricity pricing without the utility having to change rate structures.
The Smart Grid is Not Always Cheaper
A combo of deregulation, low natural gas prices, and increasing solar and wind penetration led to a decline in wholesale electricity prices. The cost of distributing that electricity has increased roughly the same amount.
Spending on delivery is growing rapidly; Source: EIA
The general line given by the monopolies that control distribution is that increasing renewables and aging infrastructure require investments to keep the lights on and upgrade to the "smart grid." The cynical take is that these utilities are milking distribution after losing their monopoly on generation. Utility regulators approve because overall rates haven't been increasing.
Monopoly utilities regularly weight models they present to regulators to favor themselves. The "Beyond Coal" campaign proved this spectacularly. Utilities wanted to build coal plants, and their financial models showed coal was the cheapest path forward. A coalition of environmental groups, philanthropists, and even a famous oilman banded together to contest these models at regulatory hearings with better data. They won virtually every time. No one had ever shown up to these hearings with well-researched counter information.
Some technologies boost throughput at a much lower cost than others. To know whether utilities are juicing their distribution upgrades would take an in-depth investigation on a case-by-case basis.
Politics and Rate Structures Collide
Overview of Common Rate Structures
Utilities usually have two main types of rate structures.
Smaller customers pay a fixed price per kilowatt-hour and may also have a fixed monthly service charge. How much power you draw at any time doesn't impact the bill. There can also be tiers where more usage lowers the price, especially during seasons with lower demand.
Industrial-level customers have more complicated billing. They pay for the kilowatt-hours they use and for the peak power usage over the billing interval. The reasoning is that utilities have two main costs. They must pay the marginal cost to generate a kilowatt-hour and build enough power plants and transmission lines to deliver electricity during peak demand. A customer that randomly uses an enormous amount of electricity for ten minutes does not cost the utility much fuel, but they cost it dearly in infrastructure. Demand charges ensure utility revenue matches its cost. Demand charges are typically at least a third of the bill for customers that have 100% load factors. Demand charges dominate the monthly payments of customers running at lower load factors.
Utilities devised these structures when most small customers only used electricity for lighting and appliances. Large customers pay according to the resources used to prevent the utility from bleeding money. EVs, electric heating, and grid-tied solar installations can create perverse incentives when paired with rate structures lacking demand charges.
Net Energy Metering
Net Energy Metering (NEM) is the idea that a customer generating electricity can sell excess electricity back to the utility. An analog meter will turn backward, netting out energy use. The utility can program a digital meter to do the same. Typically the customer has a solar array that exports electricity to the grid during daylight, while the household relies on the grid for electricity in the evening. NEM customers have little incentive to use batteries.
The Feed-in-Tariff Interlude
A feed-in tariff is a subsidy scheme popular with renewable energy proponents that guarantees a price per kWh - typically above the market rate. These programs have had some hilarious outcomes where governments ended up with 10x as much solar capacity as they intended (Spain 2008 being the prime example). Solar prices have fallen so fast that once projects become economical at the fixed price, installations go bonkers. Governments become unhappy that their subsidy became more expensive than intended and ratchet down the price that projects earn.
Net metering has the same attributes as a feed-in tariff since customers earn at the fixed retail rate rather than a wholesale market rate. In places like California or Hawaii, NEM saw the same behavior as feed-in-tariffs, where governments and utilities were surprised by the volume of installations. They ended programs, made excess energy earn closer to wholesale rates, or increased fixed fees.
Low Voltage AC Grid Saturation
Things can become dicey for local distribution nodes if many homes and businesses have grid-tied solar arrays. Supply can exceed the demand and start causing the voltage to rise as power tries to flow backward through the substation. Hawaii had neighborhoods reaching this point before it canceled its net metering program. The utility had to deny new applications once a certain number of homes had solar to ensure grid stability.
Solving the problem while still allowing houses to use net metering is challenging. Utilities could spend money to build systems that manage power exports. But that requires adding monitoring equipment and managing each local node. They could also reconfigure substations to handle the reverse flow. Either way, the utility would be charging all ratepayers to benefit solar owners.
Hawaii's solution was to automate approval for new installations that don't export to the grid while capping those that export. Self-consumption radically simplifies the issues the utility has to manage. It also incentivizes the installation of batteries that can eventually help the grid be more reliable. Exporting electricity from distributed users adds incremental benefits over self-consumption but significantly increases cost and complexity.
The Perils of Seasonal Usage
There are many ways Texas's grid could have avoided disaster during winter storm Uri. Being synchronized to one of the other wide-area grids in the US is one way. Another is not to have ~50% of its households rely on electric heat.
Cold weather causes demand to spike while also hampering supply. ERCOT is not the only grid to have suffered significant supply outages during cold weather. But other grids like PJM in 2014 were bailed out by imports and lower shares of customers using electric heating.
Customers using electric heat don't pay the costs of their impact on the grid when they only pay a fixed price per kilowatt-hour. Electric resistance heaters and air source heat pumps see power usage spike dramatically during the coldest events. The overall kilowatt-hour usage only sees a slight increase on the monthly bill, but the peak power might be two or three times higher than the norm.
The Joys of Demand Charges
Demand charges can align what customers pay with what they use. Pricing problems with electric heating or net metering disappear once demand charges come into the picture. A customer that adds oversize solar panels to net out their energy use could pay a lot if they insist on using copious amounts of grid electricity during the evening. A customer using electric heating would pay for the infrastructure needed to heat their home during cold snaps.
I love proposals to expose all customers to wholesale rates, but this matters much less than ensuring customers pay for the infrastructure they use. An analysis of industrial power rates from my local utility shows that signing up for time-of-use and using zero electricity during peak hours only reduces the bill by 5%-10%. It seems like exposure to market-like rates should be more valuable, but grids have excess capacity 95% of the time. There aren't many times the utility benefits from customers using less electricity.
Demand charges also incentivize customers to purchase batteries or change their behavior to minimize their impact on the grid. They make sense where the grid has a lot of zero marginal cost generation sources and can align customers and utilities better than often unpopular variable electricity pricing schemes.
Consumer Exit is Powerful
Improving Batteries and Falling Prices Allow Consumer Exit
Off-grid electricity has been possible for a long time. But it came at a cost. Generators are expensive to fuel and require a lot of maintenance. Lead-acid batteries have short lifetimes and poor performance.
Cheaper lithium-ion batteries and solar panels improve the value proposition. If the power company raises rates, demands an easement to run service, takes years to complete studies to start or increase service, or isn't committed to reliability, saying "No Thanks" becomes realistic.
We can already see the future in some places. Roughly one-third of Hawaiian and Australian homes have rooftop solar. In California, a distribution company wanted electric trucks, but its grid connection wasn't adequate to charge them. They hired a company to build and finance a solar/battery/generator microgrid without even bothering to contact the utility, knowing they would take years to do a grid study and that the upgrade charge would probably be outrageous.
Factors Driving Cost Decreases
Further cost declines in solar and batteries allow more customers to defect (fully or partially) from the grid.
Residential solar in the US is very expensive compared to places like Australia. Most of the cost difference comes from direct and indirect permitting costs plus higher marketing and sales expenses. Municipal governments make it hard to do installations effectively. Poorly designed incentives combine with long installation times to encourage byzantine financial structures that take a lot of effort to sell to homeowners.
Efforts like SolarAPP for automated permitting and more sales using simple loans or cash promise to decrease costs. Suppliers are trying to incorporate more components into a single package. The equipment will come in one piece instead of separate batteries, charge controllers, and inverters. The US should eventually match countries like Australia, where residential installations are ~30% more than utility-scale.
It seems likely that most houses will have solar panels within the next two decades. Whether they stay connected to the grid is less clear.
A House Can Be a Smart Grid
An off-grid home might need to double or triple the rated panel capacity than the norm to ensure reliability on cloudy days. The cost goes up but so does electricity production.
The extra energy can power electric cars and heat water. Cars get charged on sunny days, and hot water tanks can store water to reduce heating on cloudy days, improving the economics of the system. If batteries are expensive, most homeowners will only buy enough to get through the night and overbuild solar capacity. If costs fall dramatically, then batteries can carry more of the load.
Another effect is that the last 1% of reliability gets increasingly expensive. At some point, it is better to trickle charge the batteries a few times a year with a gasoline generator than add more solar panels. Utilities could still be in the game for these services. Serving an additional household has a low cost. The utility might not be able to compete on price for bulk electricity, but they could alter rate structures to help both parties. Utilities could charge a very high rate for customers with low per month kWh usage and implement demand charges instead of a fixed monthly line item on solar customers. "Off-Grid" users will trickle charge their batteries when their software projects a deficit. They will be willing to pay higher rates since meeting their tail reliability is expensive, and demand charges ensure they pay for the utility to maintain the infrastructure.
Electric heating may play some role but is a surprisingly complex topic. A drafty house with an air-source heat pump would be a disaster for a solar + storage home system. A well-insulated home with a ground-source heat pump might barely see a blip in its usage. An efficient house where the owners prefer colder nighttime temperatures might only need to run the heater during the day. Off-grid electric heating depends on a lot of factors.
Software to manage household battery charging, vehicle charging, and a hot water tank is straightforward. The hardware is falling rapidly in cost. Coordination problems mostly go away. The idea of leaving the grid dumb while making houses "smart" is underrated.
The Grid Does Not Easily Facilitate Abundance
Costs are Fixed
Energy abundance means power generation is cheap, has excess capacity, and has a wide spatial distribution. These attributes are the opposite of the problems the grid solves. The grid itself has a low variable cost and high fixed cost. Even if the wholesale price is $0, there is still a substantial cost to deliver it to customers.
Grid operators could reduce their fixed costs through automation and CAPEX-light debottlenecking. But the challenges are substantial because consensus on upgrades, price, and incentives are hard to reach.
Technologies that bypass the grid have inherent advantages.
Technologies that Avoid the Grid
Solar + Batteries
The breakeven for a solar + storage off-grid system currently ranges between $0.20-$0.40/kWh, depending on the details. The solar and battery industries are immature compared to the global scale of the energy industry. There are still many possible levers to pull to decrease costs.
Solar panels will become cheaper, lighter, and more efficient. The upfront cost will decrease, the surfaces panels can mount on will grow, and more buildings with limited space will be able to generate enough electricity on their own.
Transportation applications dominate the demand for batteries and can afford to pay more than storage applications. Prices can fall once supply catches up to transportation demand or alternative chemistries that are cheap but not useful for vehicles emerge. The best candidates for storage chemistries are lithium ferro-phosphate (LFP) batteries and sodium-ion batteries. LFP batteries use earth-abundant materials. Lithium is the bottleneck. They are useful for transportation, so supply will have to grow faster than demand. Sodium-ion batteries don't have the volumetric energy density to compete in most transportation applications, and there are no rare materials in formulations like CATL is planning. The last decade has shown how flexible users can be in switching between chemistries, so there could always be surprises. I expect home storage prices to fall 50%-80% over the next decade as the industry scales.
Betavoltaic batteries and radioisotope thermoelectric generators are examples of nuclear "batteries." They could power our homes and appliances. NASA has also made traditional reactors that output roughly one kilowatt. These technologies allow home generation without a surface footprint. Their costs are far from competitive but intermediate markets await in aerospace and remote power.
Solar and batteries may always be cheaper for powering most homes. But nuclear batteries can leap ahead by powering individual appliances. Reliability, simplicity, and convenience all improve. Devices without plugs should be the ultimate goal.
Small Wind Turbines
Large wind turbines typically produce electricity at a lower cost. There is the possibility that technology crossover from the automotive industry adds some wrinkles. Car companies have been relentlessly making electric motors (which generate electricity if spun backward) smaller and cheaper. A small wind turbine only has to be cheaper than retail electricity rates. These turbines can reduce the number of solar panels and batteries needed to go off-grid since wind electricity production has a low correlation with solar output. The wind doesn't blow everywhere, and not everyone wants a 25' tall wind tower in their backyard, meaning deployments won't be ubiquitous.
There is a lot of diversity in industrial power users. Some, like aluminum plants, devour electricity at the same rate 24/7/365. It isn't an exaggeration to say utilities built the grid around these facilities. On the other side is light manufacturing running one shift or users with weird usage patterns like our previous California distribution company example.
Aluminum and chemical plants make the most sense to continue to serve from the grid. Constant usage leads to improved fixed cost absorption. Reliability is extremely valuable since any interruption can damage the plant.
It is the opposite for facilities with irregular power usage. Demand charges punish them for not fully utilizing the grid infrastructure they require. Their scale means that micro-grid companies are willing to install generation equipment onsite with no upfront cost. A factory that runs one shift during the day might only need battery storage for brief periods of high power usage. Defection can be partial where some facilities install batteries to reduce demand charges or micro-grids to increase electricity consumption. As costs decrease, more and more electricity generation will move onsite. The size of these installations means permitting and customer acquisition costs are a smaller share of the cost, decreasing the cost difference from utility-scale solar farms.
An important reason for this change is that these new technologies have low maintenance and labor requirements. Having them onsite is not a burden like a diesel generator or a gas turbine is. Small modular nuclear reactors or geothermal with solid-state heat engines could compete in applications where solar + storage is weaker if they can reduce costs.
Choose Your Adventure
Utilities, regulators, and politicians can choose the path. On one extreme is the harmonious grid and on the other is anarcho-energy.
In the harmonious grid scenario, stakeholders work together to forge a path to change utility incentives, integrate information from and align incentives with all users, and foster better markets. Delivery costs can decrease as operators use inexpensive technology to meet increased usage and improve fixed cost absorption. Most generation would be from the lowest cost resources as operators use demand from EVs, batteries, water heaters, and industry to maximize transmission assets and avoid using expensive sources like coal and natural gas.
The byzantine structure means things may not wind down gracefully at the other extreme. The number of stakeholders and regulations means the grid is brittle to change. As customers defect, governments and utilities will fight over raising rates and other cost-saving measures. Rising rates and possibly falling reliability will push more and more customers to defect. Politicians will try to soak the utilities' best customers to stem residential rate increases. Probably the best outcome would be a railroad-style reform where utilities gain pricing power while drastically shrinking their network, lenders take some haircuts on utility debt, and the government handles the customers that are the equivalent of passenger rail. AMTRAK for electricity, anyone? Hopefully cash subsidies prevail. Most customers will see better service and prices. Cross-subsidized ratepayers will be a political liability and likely win subsidies.
We should hope that the grid dies as our technology improves. The squabbling stakeholders, who all see the grid as a chokepoint to use for their benefit, will determine how long it lasts. Users will have more choice and power to exit every year.
 France has many reactors of the same design. If a problem crops up at one, they all likely have it. The latest issue is unexpected corrosion leading to safety concerns.