Hunting down an air handler fan that is running too slowly and growing mildew (breaker panel power monitor)

I’ve been trying to reengineer the air conditioning in our house to match the new(ish) reduced cooling load after a hurricane low-E glass window retrofit by the previous owner (see ChatGPT is almost as bad at home maintenance as I am). Before I downsized the system, however, I decided that I had better make sure that the theoretical Manual J calculations of an 8.5-ton demand were correct. The goal was to see what percentage of the time the 12-ton current system (divided into three condensers/air-handlers) was running on hot days (e.g., when the NYT says South Florida is facing EXTREME DANGER).

I decided to install an inductive current monitor in the circuit breaker panel that could watch all three air handler breakers, specifically the Emporia Vue 2. This is supposed to be easy to install oneself and I have a Ph.D. in Electrical Engineering… so I decided to hire an electrician to do it properly. It took him less than one hour and he never shut off power to the panel, as the instructions suggest.

The software is reasonably good, but setup operations take longer to complete than you’d expect. Budget perhaps 30 minutes to get it all connected to WiFi and then to rename the ports. Here’s our Air Handler 3, a 3-ton system, on a day that was 125 degrees (NYT) or 91 degrees (Google/Apple). We can see that the variable-speed air handler (sadly, connected to a one-speed condenser) ramps up to about 500 watts and also that it is running most of the time (the calculated current demand for the upstairs was just 2.3 tons).

Here’s the a 5-ton air handler:

It’s drawing only 100 watts. Notice that I called it “AH2Try2” because I replaced the probe (myself!) and connected it to a different port because I assumed that the Emporia device was bad.

The installation guide for the Trane TEM6 air handler says that it should be drawing at least 500 watts:

I found that the unit was sweating on the outside and, opening it up, mildewing on the inside. The A/C contractor did the following:

  • replaced the blower (covered under warranty by Trane)
  • took the air handler apart and cleaned it
  • pumped out the refrigerant and cut the evaporator coil out and brought it down to the side yard and cleaned it thoroughly
  • cleaned out the air handler interior
  • replaced the plenum
  • replaced a failed UV sterilizer that had been in the old plenum with a REME HALO

With the new fan in place, power consumption went up to over 700 watts and the cabinet stopped sweating.

Given that air handlers are hard-wired, I don’t know of any other way to verify that they’re working properly. The regular A/C service guys don’t measure airflow carefully. And the power monitor is fun to have for investigating random appliance power consumption questions. Our 20-year-old last-legs KitchenAid refrigerator is consuming only 75 watts, for example.

What if you don’t want to spend $250-ish, including an electrician’s time? You can spend $thousands to replace your whole breaker panel and/or all of the breakers with “WiFi breakers”. Span will sell you a panel for $4,500 (plus the breakers?). Can you guess where this new company is located?

Eaton, which has been making panels for about 100 years, sells individual WiFi breakers that can report consumption and also be reset remotely. These seem to cost about $250 each, but if you already have an Eaton panel the installation could be cheap and simple.

Leviton makes a comprehensive system, but it will require replacing your panel(s). The panels themselves from Leviton seem to be cheap (less than $200). Once that’s done, an individual breaker can be as cheap as $54. Our electrician put this system in his own house and likes it.

Speaking of breakers, how long do they last in your experience? Our panels have spent 20 years in a hot/humid garage and the Cutler Hammer breakers inside don’t seem to be happy about it. Especially if a big one trips or is toggled it will tend to require replacement.

Full post, including comments

Biking without bike infrastructure: the Netherlands

In Danish happiness: bicycle infrastructure I described the Danish system of road/curb/bike path/curb/sidewalk. What if a significant percentage of a society used bicycles for transportation, but nobody bothered to build infrastructure? That’s the Netherlands!

I recently visited a friend in Delft, a university town south of Amsterdam. There are no generally no curbs at all in the downtown area. The road is informally divided into car/bike/pedestrian, but these divisions can change depending on what exactly is sticking out from a house, possibly forcing pedestrians into the bike area, or whether a truck is trying to use the road.

The risk of injury has ballooned in the last few years due to the popularity of cargo bikes and electric bikes. Instead of getting hit by a 200 lb. person-bike combo going 8 mph you’ll get hit by a 400 lb. person-small person-groceries-bike combo going 15 mph. “Trouble in cyclists’ paradise: Amsterdam accused of favouring pedestrians” (Guardian 2021) describes the increasing conflict between walkers and bikers in Amsterdam.

There aren’t as many collisions as you’d imagine, but pedestrians are required to be constantly mindful. This works for the Dutch, but tourists are frequently wandering casually into near-collisions with cyclists. What the cyclists have gained is balanced by a loss of mental peace and capacity among pedestrians.

Here’s a narrow street designed for pedestrians in The Hague:

The bicycle is being used for transportation, not recreation, so it might be whipping by these pedestrians at 10-20 mph. Here are the two transportation modes interacting in Delft:

Maybe those white boxes are supposed to delineate between walking and biking? Or maybe there are two lanes for opposite directions? I didn’t figure it out.

Just a few of the bikes parked near the Amsterdam Zuid secondary station:

What if you choose “neither” but don’t have a car and/or don’t want to pay what my rich local friend said were insanely expensive parking fees? Take the tram!

My take-away: the Danes did it right.

Full post, including comments

Test deep submersible designs for 2-3X the number of dives that will carry humans?

Department of Fighting the Last War… let’s talk about ideas that could have prevented the Titan tragedy.

The potential for failure of a pressure vessel is something that aviation has been dealing with since at least the 1930s (Boeing 307). The cycles of pressurization and depressurization are known to cause metal fatigue and, sometimes, lead to catastrophic failure. Certification authorities, such as the FAA, require structural analysis to certify a cycles or hours limit and this may be extended once there is more experience with the airframe. Running the cycles up to 90,000 (many short hops) contributed to the failure of Aloha Airlines Flight 243, a Boeing 737.

The reasonable standard of safety for deep-sea exploration and tourism is lower than for commercial airline travel. However, what about a rule that you have to build at least two of each design and use one as a sanity check on the pressure vessel design and ability to tolerate cycling fatigue? Don’t send humans on a 50th dive in a machine unless its sister ship has done at least 100 or 150 robot-only dives to the same depth. (If additional protection from fatigue-related failures is desired, increase the number of sister ships and the multiplier for how many dives they must survive compared to the human-occupied dives.)

Note that this procedure wouldn’t guarantee safety. Portions of the de Havilland Comet were subjected to 16,000 simulated cycles and the finished design nonetheless suffered a catastrophic failure with passengers on board (though the airplane that failed had some construction method differences from the prototype that was tested). But it is better than finding out the cycle limit with humans on board. And every machine that goes to the ocean floor will have a cycle limit.

Full post, including comments

The economics of hydropower

We recently visited the Glen Canyon Dam, which destroyed Glen Canyon and replaced it with a reservoir to hold surplus water in a river that doesn’t have any surpluses (calculations were made in the early 20th century, a period of remarkable wetness when compared to the previous 800 years). From the Carl Hayden Visitor Center:

(Is the visitor center named after a senior engineer who made the dam possible? One or more of the 18 workers who died so that the dam could live? No. It is named for a U.S. Senator who funneled tax dollars into this project.)

The dam powers 400,000 households, which means that the trashing of what would have easily qualified as a National Park does not generate enough power for the houses that are occupied by a single year of immigration into the U.S. What did this cost?

$2.17 billion in pre-Biden (2015) money. The BLS says that this is roughly $2.82 billion in Bidies. If we can arrange for God to give us a new raging river ever 3-6 months suitable for damming, in other words, we can provide clean hydropower to the new American households formed by migrants at a capital cost of $7,000 each.

On the third hand, maybe it would cost us a lot more today to build a monster dam. At the Navajo Bridge, just downstream of the dam, the government notes that modern construction techniques are similar to what we used in 1929, but bureaucracy and regulation are dramatically more challenging:

How much inflation has there been in concrete-rich power-generating facilities? We can look at nuclear plant construction. This 2019 paper says that the cost has gone up about 10X, in constant dollars, compared to the 1970s (takes us twice as long as costs 5X as much per day). If we had help from God (new river ever 6 months), in other words, it would cost $70,000 per new household (see How much would an immigrant have to earn to defray the cost of added infrastructure?) to provision the power generation infrastructure.

(Comparison: progressive technocrats in California have spent $9.8 billion so far on their high-speed rail dream… without laying even one mile of track (CNBC).)

What would Glen Canyon look like if this massive silt-collecting dam hadn’t been built? Here’s the Horseshoe Bend, just downstream, photographed at 0.5X on the iPhone 14:

What does the dam look like?

A concrete salesman’s dream! Note that last bucket of concrete was poured in September 1963, the same year in which your beloved (I hope) blog host was born. The project was supposed to take 6.5 years, including all of the prep work and the bridge, but was finished after 6 years for substantially less than forecast (by Kiewit, whose Boston Harbor project was rendered lethal by government bureaucrats as described in Book review for Bostonians: Trapped Under the Sea).


Full post, including comments

How innovative is the OceanGate submersible that is in trouble?

Sad to think about the folks inside the OceanGate Titan right now. Usually the flip side of innovation is danger, as demonstrated by Deepwater Horizon (see Ten years since Deepwater Horizon set a depth record).

How innovative is the Titan? Here are the most critical specs:

Let’s compare to Alvin, built by a cereal company in 1964.

Alvin has the same payload as Titan, but weighs 38,000 lbs (65 percent more) and has dramatically less interior volume. A perhaps more significant difference is that Alvin has small windows. From a 2013 article:

When we set sail later this year, Alvin will have five windows: three up toward the front that are 7 inches across on the inside (17 inches across outside), and two smaller ones off to either side that are 5 inches in diameter on the inside (12 inches across outside). These are considerably larger than the windows we had before.

Titan reportedly has a 21-inch diameter viewport. Let’s hope that wasn’t the failure point. The other big difference is that Titan hull is “Carbon Fiber and Titanium” (above web site) while Alvin’s personnel sphere was entirely titanium (and could detach from the rest of the machine in an emergency, then rocket to the surface).

Here’s hoping that everyone comes out of this alive, though that seems unrealistic.


  • Remembering William Lewis Herndon, captain of the gold-laden SS Central America (the treasure hunters decided not to attempt sending humans down into 8,000′ of water, but to do it all with remotely operated equipment)
  • “DEEPSEA TITANIUM PRESSURE HULLS” (U-Boat Worx): In our deepest-diving submersibles, we use titanium alloys to achieve optimum size, weight and performance characteristics. Titanium has several distinct advantages – it is stronger than regular steel thereby enabling us to keep the weight of deep-diving models as low as possible. Other advantages include that it requires no maintenance, has an extended lifecycle, and has incomparable anti-corrosive properties.
Full post, including comments

Remembering William Lewis Herndon, captain of the gold-laden SS Central America

On this Memorial Day I’d like to celebrate the memory of William Lewis Herndon, author of Exploration of the Valley of the Amazon and captain of the SS Central America, a commercial ship with a U.S. Navy captain that sank off the Carolinas during a hurricane in 1857, resulting in a loss of 425 lives, mostly people returning from the California Gold Rush. Herndon could have escaped with his life, but chose to go down with the ship after ensuring that all women and children had been evacuated (including Lucy Dawson, the only black woman on board; we are informed today that Americans in 1857 were irredeemably racist, yet white men gave up their lives so that Ms. Dawson could keep hers).

Herndon is described in Ship of Gold in the Deep Blue Sea: The History and Discovery of the World’s Richest Shipwreck (Gary Kinder, 1998):

Married and the father of one daughter, Herndon was slight, and at forty-three balding; a red beard ran the fringe of his jaw from temple to temple. Though he looked like a professor or a banker more than a sea captain, he had been twenty-nine years at sea, in the Mexican War and the Second Seminole War, in the Atlantic and the Pacific, the Mediterranean and the Caribbean Sea. He knew sailing ships and steamers and had handled both in all weather. He was also an explorer, internationally known and greatly admired, who had seen things no other American and few white men had ever seen.

Herndon ordered Ashby and his first officer not to let a single man into the boats until all of the women and children were off. “While they were getting into the boats,” observed one man from the bailing lines, “there was the utmost coolness and self-control among the passengers; not a man attempted to get into the boats. Captain Herndon gave orders that none but the ladies and children should get into the boats, and he was obeyed to the letter.”

The ship took 30,000 lbs. of gold 8,000′ underwater, which is what led to the main story of the above book. This cargo was worth $8 million at the time and roughly 1 billion Bidies today (inflation of 125X or 12,500 percent).

If you can tolerate an old-style book in which race, gender ID, and sexual orientation are seldom mentioned, the story of engineering challenges being addressed one after another is fascinating. The hero of the book is Tommy Thompson, a self-motivated engineer who attends Ohio State, works for Key West treasure hunter Mel Fisher, and comes back to work for Batelle. While at Batelle, he comes up with the idea of salvaging wrecks in the deep ocean.

“A galleon drafted about fifteen feet,” Tommy told Bob, “so they generally hit reefs in about fifteen feet of water. It is not like men to leave gold lying in fifteen feet of water.” Most of the artifacts Fisher had found were at twelve feet, and the only reason Spanish salvage divers had not completely stripped the Atocha in 1622 is because a second, far bigger storm had hit the wreck site three weeks later.

During the three centuries following Columbus’s voyages to the New World, much of the gold and silver on earth had been transferred from the New World to the Old World, and 25 percent of it had been lost. But don’t search for it among the thousands of shallow-water shipwrecks in the Caribbean, said Tommy; the odds were too slim. Search for treasure where storms couldn’t buffet the remains, where ships were not piled on top of each other, where the bottom was hard and the currents slow, and where no government could stake a claim. Tommy told Bob he wanted to recover historic shipwrecks in the deep ocean.

A key enabler of the quest for the Central America‘s gold is Martin Klein, the inventor of practical side scan sonar, but this MIT graduate is not credited by the author. Once found, however, there is a question of how to conduct mining operations on the ocean floor with mid-1980s technology.

If you got your submersible safely into the water, your ship at the surface was rising and falling while your submersible was descending; each fall caused the cable to go slack, and each rise snapped the cable taut, like pulling a car with a chain. That load suddenly became ten times heavier than the submersible itself, and the cable often broke and you lost your submersible. That armored cable was filled with electromechanical wires that carried signals down to the sub and back again. If the snap loading didn’t break it, every time that cable passed over a pulley, the wires bent and straightened with the weight of the vehicle, and often ten times the weight of the vehicle, and the wires fatigued and parted. A replacement cable took three months to manufacture, and carrying a spare cable on board meant needing more space on a bigger ship, tended by a larger crew, for much more money. Attempting to land on the seafloor was risky and difficult for two reasons: First, the rocking of the ship would jerk the vehicle—one minute you’d be looking at the bottom, the next minute you’d see nothing, the next minute the camera would be in the mud. Second, hanging something heavy on the end of a cable twisted the cable; if you set that heavy weight on the seafloor and slackened the cable at the same time, the twisted cable tied itself in knots, like the cord on a telephone. When an armored cable with several thousand pounds on the end kinked up, and the bouncing of the ship topside jerked on those kinks, the cable again often broke, which meant you left your vehicle on the bottom and headed back to the beach for the rest of the season.

Everyone who had previously worked in this area was funded by militaries, which had essentially unlimited budgets to look for sunken submarines and similar valuable items. Tommy Thompson needed to do the job for $millions when others had failed with $1 billion budgets. There’s also an interesting legal challenge:

The Central America lay at the far reaches of the Economic Zone, almost two hundred miles offshore. No one had ever tried to recover an historic shipwreck so deep it lay beyond the three-mile boundary. Tommy could bring a piece of the Central America into the courtroom, but no one knew what would happen next.

One of the more unusual challenges was how to bring up gold coins without scratching them, which would reduce their value to collectors. The team of salvors came up with the idea of a silicone injection process that would embed gold objects in a block of the soft substance before it was all brought to the surface.

If you love engineering, I recommend Ship of Gold in the Deep Blue Sea. Even if you don’t love engineering, I hope that you’ll join me in remember Captain Herndon and his decades of service in what was a hazardous job back then (wooden ships combined with no GPS and no weather forecasts).

(Since 1998, Mr. Thompson’s career has developed some warts. I don’t want to spoil the book for you, but let’s just say that, as in family court, a big pot of gold can lead to accusations of unfairness.)


Full post, including comments

Lift the Wankel/battery powertrain from a Mazda to use in an airplane?

Piston-powered airplanes subject pilot and passengers to unpleasant vibration. Battery-electric airplanes have minimal range.

There have been some successful applications of Wankel rotary engines in aircraft. The homebuilt folks have had some success with Mazda RX-7/RX-8 engines. Here’s an article from EAA’s Sport Aviation in 2002:

While incredibly reliable in automotive use, car engines haven’t done well running at high power settings all day every day in airplanes. The aviation-specific rotary engines thus far, such as Diamond’s AE50R, are low power engines designed for self-launching gliders and UAVs.

What if the smooth rotary engine were used to generate electricity buffered through a battery pack? Then it wouldn’t matter if the engine failed more often than 1930s-style Continental and Lycoming piston engines. An engine failure would mean using a 20-minute battery reserve to land. Is there a mass-market low-cost battery+Wankel combination available? Yes! From “The Hybrid Wankel Rotary-Powered Mazda MX-30 R-EV Is Finally Here. Here’s How It Works” (Autopian, January 2023):

For starters, the engine doesn’t drive the wheels. It only serves as a generator connected to a motor/generator unit to send power to the battery pack. The battery pack then provides juice to an electric motor which powers the wheels. This means that despite burning gasoline, the MX-30 R-EV should theoretically have the seamless power delivery of an EV, and it should be able to keep the Wankel engine at its “sweet spot” for efficiency for a significant portion of its on-time.

As for deeper details on that rotary engine, there’s the presence of direct injection, something never attempted before on a production rotary engine. The side housings are aluminum and coated with plasma for low weight and friction management respectively, all while being just 80 mm wide. For the sake of longevity, the apex seals are 25 percent wider than the ones on an RX-8’s RENESIS engine, clocking in at 2.5 mm. The result is 73.7 horsepower from just 830 cc of displacement. Curiously, although rotary engines love to rev, Mazda claims that peak power hits at just 4,700 RPM. That might sound weird for a high-revving Wankel, but it should translate to very low noise.

At 214 pounds, it looks like this engine is fairly heavy for its horsepower (a little heavier than an 80 hp aluminum piston engine), but given the high efficiency of electric drive maybe this would still work out well for a 2-seater.

Readers: Where’s the flaw in this path toward aircraft powered by a mass-market powertrain?


Full post, including comments

How did Hurricane Fiona, a Category 1 storm, knock out Puerto Rico’s power?

A few media-following friends in the Northeast have been checking in, concerned that Hurricane Fiona, which knocked out power in Puerto Rico, is also trashing our neighborhood. They are reassured to learn that Puerto Rico is 1,000 miles from Palm Beach County, but it has made me wonder… given that (1) Fiona is only a Category 1 storm, (2) Puerto Rico can expect something similar every year or two (history), and (3) the power grid in Puerto Rico was recently rebuilt to the latest standards (after the 2017 Category 5 Hurricane Irma), why were the reported 85 mph winds enough to take the system out?

Is it simply impossible to make above-ground lines robust enough to handle 85 mph winds? Is the problem that trees will inevitably come down and break the lines even if the lines wouldn’t have been blown down? (But a newly engineered grid should be able to handle quite a few individual tree impacts because the power would be routed around the cut line.)

From state-sponsored NPR in 2021:

It’s been four years since Hurricane Maria devastated Puerto Rico’s electric power grid. Yet even after billions of dollars were allocated by the federal government to repair it, the island’s energy infrastructure is still in terrible shape. Blackouts continued this summer as the two entities responsible for operating the grid pointed fingers at each other over who is to blame. One of those two entities is Luma, a private company that was awarded a contract last year to distribute electricity around the island. The other is the Puerto Rico Electric Power Authority, known as PREPA, which used to be in charge of the whole system and now continues to operate the power plants.

The restoration process is very bureaucratic because you have Luma going through FEMA’s process, going through the Puerto Rico Energy Bureau’s process. And you also have Luma going through federal process and going through Puerto Rican process. And you know what? There’s not a single work already done with reconstruction funds. They’re still planning and designing. So this will take a lot of years before we see something better.

An IEEE article from 2018 doesn’t explain any of the engineering or technical details:

This past December, I traveled to Puerto Rico to report on this massive undertaking. I found contradictions everywhere I went. I saw utility workers fanned out across the island, yet progress remained excruciatingly slow. I met rank-and-file PREPA employees working flat out to restore power, yet each day brought a new report of fumbles at the utility’s top levels. And I heard many smart and exciting ideas for how to build a modern, resilient grid in Puerto Rico, even as the urgent need to restore power meant resurrecting the vulnerable existing system.

KSUA to TJSJ (skyvector):

Are we going to see “I stand with Puerto Rico” Facebook profile images? Or will people stick with this one:

Full post, including comments

Audible’s tale of an engineering hero: The Man Who Knew the Way to the Moon

It is rare for engineering to be the subject of literature and entertainment and even rarer for an engineer to be the subject. Audible’s The Man Who Knew the Way to the Moon is a welcome outlier. Although I was once a proud Fortran programmer at NASA’s Goddard Space Flight Center (the Pioneer Venus project), I hadn’t realized that the original idea for the moon landing was to fly a huge vehicle and enough fuel for the return trip straight to the lunar surface. Audible’s work is about John Houbolt, who fought the conventional thinking and endured all of the bureaucratic infighting to promote the idea of a small vehicle that would land on the moon, thus requiring only a tiny fraction of the fuel to get back to Earth. After escaping the moon’s gravity, the small vehicle would rendezvous with a bigger spacecraft in lunar orbit (“lunar orbit rendezvous” or “LOR”) and then the astronauts could all go home.

Trigger warning: the book implies that members of the 2SLGBTQQIA+ community, engineers of color, and engineers who identify as “women” played no role in getting astronauts to the moon.


Full post, including comments

Is 5G a total fraud?

Mobile phone service back in Maskachusetts was generally terrible, whether the iPhone 12 Pro Max indicated “LTE” or “5G” up at the top right. I attributed this to hills generating multipath and the righteous demanding that cell towers be built in someone else’s town.

We’re living in Florida, though, where a municipal landfill is the only hill, and the government encourages any kind of useful infrastructure. I think that all of the preconditions for awesome mobile data service have been fulfilled:

  • I’m fully vaccinated and so is our golden retriever, Mindy the Crippler
  • The Verizon bill is on autopay
  • the iPhone usually shows 3 or 4 bars of 5G
  • there are no tall buildings or hills around

Yet the service simply doesn’t work. It can take minutes to send a single photo via iMessage, for example. Looking up stuff on Google can be impossible. Navigating via Google Maps results in an “offline” display, even when the phone shows 3 bars of 5G.

Could it be that there is a working LTE service in most locations, but the phone sees 5G and latches onto it even when the 5G radios are simply broken? I’ve experimented with telling the phone to use LTE only, but that didn’t seem to help. Sometimes the Verizon network yields impressive numbers on a Speedtest, comparable to high quality home broadband circa 2010, but for any given request it is unpredictable whether it will take a fraction of a second or minutes.

Is this issue unique to my iPhone 12 and it will be #ProblemSolved when I upgrade to the glorious world of iPhone 13? Or are other folks having similar issues (3 or 4 bars of coverage yet it is tough to download an ordinary web page)?

Waiting for a page to load on 5G:


Full post, including comments