Reading List for 06/21/2025
Construction Physics email stats, why appliances break so often, a special economic zone for Silicon valley, a river filtering pool, and more.
Welcome to the reading list, a weekly roundup of news and links related to buildings, infrastructure, and industrial technology. This week we look at some Construction Physics email stats, why appliances break so often, a special economic zone for Silicon valley, a river filtering pool, and more. Roughly 2/3rds of the newsletter is paywalled, so for full access become a paid subscriber.
A few housekeeping items:
I had planned to publish an essay about high housing costs in western states this week, but for various schedule reasons this got pushed till next week.
I was (briefly) on NPR’s Marketplace talking about shipbuilding this week.
Substack data
I vibe coded a little script to analyze the Construction Physics email list. The most surprising thing I discovered was how large the number of people who never open a single email are: over 25% of subscribers. What’s more, more than 10% of subscribers have never even received an email. I believe some of the latter is due to some email clients blocking tracking pixels, and some due to people signing up via the app and declining to receive emails, but not sure how much of either.
The other notable pattern is how bimodal open rates are. Most subscribers either open almost every email, or open almost none of the emails.
If you have a substack and are interested in analyzing your own email list, I’ve made the script available on github here. It’s easy to use if you have python installed: download your full email list (check “Export all columns”), then rename the file full_email.csv. Put the script in the same directory as the csv file, and run it. It will output a text file with the analysis. The output includes some simple analytics, as well as a count of subscribers from things like governments, media outlets, major corporations, and major universities. I’d be very interested in seeing the stats from anyone who decides to run it.
Iberian Peninsula blackout report
The preliminary official report on the causes of the Iberian Peninsula blackout in April of this year has been released (it can be found here). The report is 182 pages and (bizarrely) doesn’t have anything like an abstract or an executive summary. It’s also just difficult to read in general. Here’s a Claude 4 Sonnet-generated summary (full conversation available here): the key factor seems to be insufficient thermal power plants (nuclear, gas, coal) online to provide voltage control:
On April 28, 2025, at 12:33:30, the Spanish and Portuguese electrical systems experienced a complete voltage collapse that disconnected them from the European grid, affecting over 47 million people. The comprehensive investigation revealed that this unprecedented event resulted from a complex cascade of overvoltages—a "chain reaction" where high voltages caused generation facilities to disconnect for protection, which paradoxically caused voltages to rise even further, triggering more disconnections until the entire system collapsed. At the time of the incident, the system was operating under challenging conditions: demand was relatively low at 25,184 MW (compared to the historical peak of 44,876 MW), renewable generation comprised 82% of the mix with nuclear providing 10% and the remainder from gas, coal, and cogeneration, while 2,978 MW of pumped storage was consuming power to take advantage of low market prices.
The root cause was insufficient dynamic voltage control capability in the system. Only 11 thermal power plants were coupled with voltage control obligations—the lowest number recorded all year and fewer than the 12 originally planned due to an unplanned outage the previous evening in the critical southwestern zone. This shortage was compounded by evidence that several voltage-controlling generators failed to absorb the expected amounts of reactive power during high-voltage conditions, with one facility in the south zone behaving markedly differently from both other plants and its own historical performance. The system was further destabilized by two significant oscillations: first at 12:03 with an unusual 0.6 Hz frequency (attributed to anomalous behavior at a specific photovoltaic installation), followed by a more typical 0.2 Hz inter-area oscillation at 12:19 where the Iberian Peninsula oscillated against central Europe.
The final collapse sequence began at 12:32 when voltages started rising linearly across the transmission network, climbing from 413 kV to 428 kV in just 57 seconds at some substations. Three major generation loss events occurred in rapid succession: 355 MW lost at 12:32:57 in Granada, 730 MW lost at 12:33:16 in Badajoz, and 550 MW lost at 12:33:17 in Seville. Each disconnection reduced the system's ability to absorb reactive power and decreased line loading, both effects that further increased voltages and accelerated the cascade. Within approximately 12 seconds, the system had lost sufficient generation that it could no longer maintain synchronization with France, ultimately leading to complete voltage collapse throughout the peninsula.
The lack of voltage control, in turn, seems to be in part a regulatory issue:
Voltage control fundamentally depends on managing reactive power—the non-useful component of electrical energy that doesn't perform work but is essential for maintaining proper voltage levels. Synchronous generators can both produce and absorb large amounts of reactive power independently of their active power output. More importantly, they can do this dynamically in response to grid conditions.
The report explains that thermal plants provide "dynamic voltage control by setpoint"—they continuously monitor grid voltage and automatically adjust their reactive power output to maintain voltage within target ranges. As the report notes: "the higher the overvoltage, the greater the reactive power that must be absorbed by the generation to help contain the voltage." This is like having an automatic voltage regulator that responds in real-time to changing conditions.
While renewable technologies have been technologically capable of voltage control for years, they are currently regulated differently. Under Spanish regulations (Royal Decree 413/2014), renewables must operate at a fixed power factor (typically 0.98), meaning their reactive power output is directly proportional to their active power generation. When solar output drops due to clouds or market signals, their reactive power absorption drops proportionally, contributing to voltage increases—exactly what happened during the April 28 event.
The report specifically identifies this as a critical issue: "a reduction in generation from fixed power factor technologies puts a double upward pressure on voltage (lower reactive power absorption, and higher reactive power production in less loaded grids)." This is why the system became vulnerable when renewable generation decreased while thermal generation remained insufficient to compensate.
Federal land auction
A huge fraction of land in the western half of the US (ranging from 30 to 80% depending on the state) is owned by the federal government. Much of this land is protected (National Parks, Forests, and Wildlife Refuges), but much of it is also used for things like grazing cattle and drilling for oil. Every so often a proposal comes along to sell some federal land: now one is being considered as a way to raise money as part of the “Big Beautiful Bill” making its way through congress. From the New York Times:
Senate Republicans are resurrecting a plan to sell millions of acres of federal lands as part of President Trump’s giant tax and spending bill, setting up a fight within the party.
The proposal would require the Bureau of Land Management and the U.S. Forest Service to identify and sell between 2.2 million and 3.3 million acres of public lands across 11 Western states to build housing…
The new plan to sell public lands was included in draft legislation issued on Wednesday by the Senate Energy and Natural Resources Committee that is part of Mr. Trump’s “big, beautiful bill.” The draft envisions raising as much as $10 billion by selling land for housing in Alaska, Arizona, California, Colorado, Idaho, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming over the next five years.
And the Wall Street Journal gives a bit more information here:
The areas up for sale would exclude land with dedicated uses such as national parks and monuments, and land previously leased for mining, energy production and other activities. The plan directs federal agencies to prioritize selling land that is near existing roads or suitable for home-building. Agencies would consult governors before auctioning land in their states, and state and local governments would have the right to make an offer before private bidders.
A map of the lands planned to be eligible for sale is available at Wilderness.org
The idea that this will be used for “housing” seems somewhat dubious to me: most of this is BLM or Forest Service land that isn’t anywhere near places where there’s actual demand for housing. But some of it does seem to be near developed areas: scrolling around on the map reveals a few locations where there’s large BLM or FS tracts across the road from residential developments. Perhaps there’s enough parcels like that for the proposal to make sense (it would be worth someone doing an analysis of this). 3 million acres sounds like a lot, but its less than 0.5% of the 600 million-plus acres of land the federal government owns, so the proportion of develop-able land does not need to be particularly high.
It’s also worth noting that selling off federal land isn’t exactly new, though the extent of sales seems to be different this time than it has been in recent history. Between 1990 and 2018 the federal government sold around 31 million acres of federal land. But most of this was in Alaska, as well as Department of Defense divestments following the end of the Cold War.
Appliances
The Wirecutter has a good article on drilling down into why today’s appliances seem to have much lower lifespans than they did historically (as someone who’s bought several increasingly shitty versions of the Ninja blender, I was very interested in the conclusion here). They conclude it’s a mix of survivorship bias (most old appliances didn’t have all that great lifespans), government regulations (and consumer desires) requiring constant product changes, and competition driving down the quality of parts and repairability.
Regulations have improved appliances by many measures: They’ve removed leaky pilot lights in gas stoves, replaced ozone-depleting freon in refrigerators, prevented cooktop fires and burns through safety sensors and better insulation, and reduced water and power used by washers and dishwashers exponentially, saving consumers money.
But on top of that, they have generally added layers of complexity to appliances that had previously stayed the same for decades, said Zilka, the repair-industry veteran.
The main culprit is the set of efficiency standards for water and energy use for all cooking, refrigeration, and cleaning appliances. Those standards have been managed by the Department of Energy since 1987 and frequently updated; the rules are evaluated at least once every six years after each amendment, according to AHAM, and potentially strengthened. Most of the significant changes arrived in the early 2000s, and they have been updated several times since.
Before those updates, appliances were road-tested for decades, said Zilka. Brand representatives would even ride along with his technicians as they worked on appliances “and monitor the failures, and what goes on inside,” he said.
Now, the models change a lot more often…