The issue

Back on October 1st, the FAA blindsided the Denver flying community by immediately banning pattern work at Centennial Airport (KAPA). Pattern work, meaning flying laps in roughly a rectangle near the airport with usually a touch-and-go landing on each lap, is critically important to flight training, as one of the hardest parts of learning to fly is learning to land well, and the best way to learn to land is to practice it a lot. KAPA is a busy airport for several reasons, but a major one is that a lot of flight training is done there. Now that’s been thrown into disarray.

The memo from the FAA, provided by Centennial Airport, describes the FAA’s reasoning, though not the lack of notice, nor of any investigation of alternative solutions. In short, the FAA claims there were a handful of instances, roughly 60 times between May 1, 2025 and May 1, 2026, where aircraft departing from the parallel runways at KAPA drifted within 600 ft laterally and 200 ft vertically of each other. They further classified two of these as “near-mid air collisions” (NMACs), which means within 500 ft laterally and 100 ft vertically. As a result of these “loss of separation”/”unmitigated convergence” incidents, the FAA banned simultaneous parallel departures at KAPA, which is a de facto ban on pattern work.

All of this is near and dear to me because I learned to fly at KAPA and still do the vast majority of my flying there. I have many friends who are flight instructors at KAPA, impacted for the reasons outlined above, and even non-flight-instructor pilots (like myself) still need to do a few laps in the pattern on occasion for various reasons.

The causes

The problem is thus: aircraft are getting close to each other during departures, potentially close enough to collide. The FAA specifically identifies drift due to wind, incorrect pilot turns, and delayed recognition of impending issues by ATC as causal factors.

With the parallel runway centerlines at KAPA being only 700 ft apart, it doesn’t take much directional deviation for two aircraft taking off simultaneously to get within 600 ft of each other, meeting the FAA’s threshold for a loss of separation.

I’ll add a few more potential causes to the mix:

• Confusion about the meaning of the “fly straight out” versus “fly runway heading” instructions from Tower during takeoff clearance. The former means to follow the extended centerline of the runway, i.e., to compensate for wind. The latter means to fly the heading of the runway, but don’t compensate for the wind. If one was meant but the other was done, well, that’s an obvious potential source of problems. (A corollary problem is that it can be difficult to fly perfectly on the extended centerline without a moving map, even when using visual landmarks.)
• Handoffs into the pattern being a little clunky sometimes. At KAPA, pattern work is usually done on the shorter parallel runway (17R/35L), but almost all initial takeoffs are done from the longer parallel runway (17L/35R). Moreover, the two runways often have different tower controllers. The transition from the upwind off the long parallel into closed traffic on the short parallel has on occasion seemed less coordinated than one would hope, presumably because of the added challenge of sequencing in new traffic or reduced situational awareness for each controller about the runway not under their control.
• The timing of touch-and-goes being difficult to consistently predict. Depending on where a plane touches down, how long it takes the pilot to clean up the aircraft configuration to take off again, and how quickly they get the throttle in, there can be considerable variation in the delay between being on final approach and being on climb-out again. Go-arounds add even more uncertainty. This of course makes takeoff clearances on the other runway more difficult to time correctly.

Independent research

I decided to pull the data and do my own analysis. While somewhat more coarse than the FAA’s radar-based data, it’s easy these days to pull ADS-B-sourced timestamped geolocation data from public archives. I used data from adsb.lol as aggregated by adsb.exposed in a ClickHouse database, which let me use SQL to narrowly select for close encounters in the sky. Specifically, I searched for all instances in the FAA’s aforementioned study window where two aircraft were within 600 ft laterally and 200 ft vertically, less than 7000 ft in altitude (specifically pressure altitude, because of ADS-B), greater than 50 kt ground speed (so as to remove false-positives with aircraft on the ground), and within 5 miles of KAPA. I then manually filtered out obvious formation flights of aircraft (e.g., when two aircraft intentionally take off in a close formation from the same runway). I found that the ADS-R data, which is (effectively) ADS-B retransmitted by the FAA, was less noisy than the directly received ADS-B data, so I further selected for that. What was left was 29 episodes.

While 29 is less than 60, I have no reason to think this isn’t a representative sample. The discrepancy in count can be explained by the FAA also including TCAS alerts (as those aren’t necessarily going to show up in ADS-B data), timestamp coarseness (the data had roughly 10-second intervals), aircraft without ADS-B reporting (notably, at least around KAPA, most military aircraft and most law enforcement aircraft), or other items that could have been on radar tapes but not ADS-B data. Also, the accuracy of altitude and position data itself is subject to tolerances.

(Disclosure: about half of the episodes involved aircraft associated with at least one of the two flying clubs I’m a member of at KAPA. I have personally flown several of the listed planes, though not during the listed episodes. Also, I personally know at least some of the pilots who were at the controls in some of the episodes.)

Several of the 29 episodes were spot-checked against graphical flight path plots on adsbexchange.com; all spot-checked flights did appear to involve aircraft losing separation due to drift or similar during departure. None of the spot-checked flights involved a loss of separation on approach.

Of the 29 episodes I identified, 3 fit the NMAC (near-miss) criteria (500/200 ft), with the closest being one with 437 ft of lateral separation and zero vertical separation. This is slightly different than the FAA’s findings, which indicated 2 NMACs; the caveats mentioned earlier about data apply equally here.

Unfortunately, the ATC audio from 2025 and early 2026 is no longer available from free public sources, so I can’t determine whether or not a tower controller advised any of the involved aircraft about their situations nor what, if any, specific instructions were given.

So what does a loss of separation look like in terms of flight paths? Here’s an example of the September 3, 2025 incident involving N739LV and N1330N:

In this case, N739LV drifted a touch to the west. Winds on the ground reported five minutes earlier were from 110 degrees at 4 kt, i.e., light winds from the east.

So was this specific situation dangerous? Well, maybe or maybe not. N1330N was already in the pattern in left traffic on 35L, and N739LV was turning to join the pattern on the same runway. If N739LV had turned about 10 seconds earlier, the situation could have potentially been not great. On the other hand, we don’t have the tower audio, so it’s probable that Tower was keeping an eye on both of them and kept N739LV flying straight(-ish) out until they were well clear of any conflict for the turn. Instructions along those lines, where planes are told to fly straight out and must not turn until specifically cleared to do so by tower, are common.

The FAA made an unfair comparison

The FAA listed the airports they looked at: Centennial (KAPA), Charlotte (KCLT), Dallas-Fort Worth (KDFW), Fort Worth (KFTW), George Bush-Houston (KIAH), Chicago O’Hare (KORD), DeKalb-Peachtree (KPDK), Seattle-Tacoma (KSEA), and San Francisco (KSFO). All of those except for KPDK and KAPA are busy class B airports, which effectively means that the vast majority of operations are part 121 commercial jets, not exactly a fair comparison to the part 91 general aviation operations that dominate at KAPA. Even KPDK isn’t really a peer: it has fewer than half the number of annual operations compared to KAPA (169,000 vs 305,000 in 2025).

Comparing GA-heavy KAPA with a bunch of large commercial airports borders on ridiculous.

Suspicion around the campfire is that the FAA bowed to pressure from the numerous anti-airport people in the vicinity of KAPA to cut down traffic by eliminating pattern work. I have no evidence that’s the case; I’m just sharing the rumor.

A better comparison

A far more fair comparison would have been Deer Valley (KDVT) north of Phoenix. Not only does KDVT have about 60% more traffic than KAPA (490,000 operations vs 305,000 in 2025), it also has parallel runways that are spaced exactly as far as at KAPA, 700 ft. (Note that some sources incorrectly list KAPA as having 600 ft separation, but this is easily refuted by measuring with a GIS tool like Google Earth.) Importantly, KDVT is heavily used for flight training, like KAPA, and, also like KAPA, doesn’t have any scheduled passenger jet traffic.

I pulled the data for near-misses in the vicinity of KDVT just as I had for KAPA, in the same time period, and wouldn’t you know it: there were FAR FAR more issues at KDVT, on average one every couple of days. I counted 154 such events, including 24 meeting the NMAC criteria.

One particularly spicy encounter on April 13, 2026 saw N6296W and N714SC come within 279 ft of each other, 217 ft laterally and 175 ft vertically. Or, since that technically doesn’t count as an NMAC, consider the May 18, 2025 close encounter between N169AG and N222TD with 360 ft of lateral and 25 ft of vertical separation. Both of those were considerably closer than anything that happened at KAPA during that time period.

Here’s the filtered dataset for KDVT for you to draw your own conclusions. (If you find any issues, such as formation flights that I missed, please let me know!)

My point is that it’s very odd that the FAA is coming down hard on KAPA when the problem is considerably worse elsewhere.

Solutions

So, even though I think KAPA is being picked on unfairly, I will readily admit that it would be far better to have fewer or no losses of separation between departing aircraft at KAPA. Here are some suggestions from my perspective as a pilot; I’m not an air traffic controller, so take all of these with a grain of salt.

• Instruct all touch-and-go traffic to fly their upwind (takeoff) leg angled out about 15 degrees relative to the other runway. This should provide additional lateral separation to compensate for potential drift, regardless of whether the touch-and-go or the non-local departing traffic was actually at fault. It also would provide margin for things like unexpected go-arounds on the parallel runway, where timing-based approaches would be less effective.
• Instruct all non-local departing traffic to fly their upwind angled out about 15 degrees the other way. This could be done with or without the previous suggestion.
• Add more buffer time between touch-and-go traffic and non-local traffic.
• Be more willing to call go-arounds for touch-and-go traffic on short final if it looks like the timing will be tight.
• Better coordination between tower controllers when different people are controlling each of the two parallel runways.

Just some ideas. All I hope is for pattern work to resume soon!

 

Sean and I flew a Cessna 172 from Denver to Oshkosh, Nebraska — and then Oshkosh, Wisconsin — in July for EAA AirVenture. It was nuts.  It was fun!

It wasn’t supposed to be done in a 172 at all, but rather a Diamond DA-40. However, while pre-flighting the plane a few days before the trip, I discovered a delamination in the prop, grounding the plane for repairs. Fortunately, Sean had the foresight to reserve a 172 with a different flying club. After a last-minute checkout for me with that club so I could also legitimately be at the controls, we set off into the sky. We would be slower than expected, but better a slow bird than one threatening to shed part of its propeller.

The defining flight condition for the eastbound trip was smoke. Northern Minnesota and a giant swath of Canada was on fire, which pushed the flight visibility down to during parts of the trek down to nearly instrument conditions. In fact, Oshkosh itself had been IFR most of the day before we arrived, went IFR again a few hours after we arrived, and was IFR most of the day after we arrived. Since IFR arrivals and departures are severely rate-limited during AirVenture, pretty much all of the 10,000 or so planes that fly in must do so VFR. That meant a mad rush to get in during the brief periods the field wasn’t IFR. We lucked out; nobody was ahead of or behind us for miles during our transit of the Fisk Arrival, and by the time things were getting spicy, we were safely on the ground.

No pilot can resist stopping what they’re doing to watch an airplane land. When multiple craft are coming in per minute, and when stress is high, ATC instructions are sometimes contradictory, compliance is iffy, and winds aren’t exactly favorable, the entertainment factor is high. In some ways, it’s like watching racing as a racer: nobody wants to see any machines get damaged or people get hurt, and you can relate to the person at the controls in the cockpit, but it’s more exciting when things go just a little wrong. We saw plenty of sketchy approaches, bounced landings, and at least one prop strike, first from under our wing parked in the North 40 at the airport, and then from the roof of a nearby hotel.

In contrast to the chaos of the main airfield, the seaplane base for AirVenture was just as chill as advertised. Perched on the shore of Lake Winnebago at the end of a short trail walk through the woods, we found an opening filled with people sitting around campfires and lounging in the shade on numerous Adirondack chairs. Seaplanes were leisurely bobbing in the water next to docks. I might admit to snoozing at least a tiny bit while watching seaplanes come and go; it was great.

The next few days went by quickly. Air shows were watched. Aviation exhibits were visited. And quirky locals were encountered at neighborhood bars and restaurants, including Sean trying to convince one Oshkoshian that the world really is round. It was soon time for our return trip to Colorado, loaded up with as much Spotted Cow as we could take within weight and CG limits.

We departed Oshkosh IFR, which we justified partially because there were in fact some scattered low clouds on our route, and partially because we’d been lucky enough to actually get one of the rare IFR departure reservations. As it turned out, the smoke was bad enough upon our arrival for a stop in Minneapolis that we ended up shooting an instrument approach to the Flying Cloud Airport due to necessity rather than discretion.

Along the way, a resolutely uncertain pilot flying VFR repeatedly asked Minneapolis Center if some MOA in Wisconsin was active. (This despite the question being somewhat moot: VFR flights don’t have any obligation to avoid active MOAs.) The Center controller calmly explained that no, the MOA was not active, and that in fact they deliberately keep that MOA cold during the week of Airventure. The pilot acknowledged hearing Center’s response, but apparently unsatisfied with the controller’s answer, the pilot then repeated basically the same question. The controller gave basically the same response. Then the pilot did it again, at which point the controller said, “I really don’t know how else to tell you that the MOA is cold.” The pilot then diverted around the MOA anyway. A few minutes later, another pilot in a different plane piped up, “Center, is that MOA hot”? No response came from Center, so after a beat, the pilot spoke again: “That was a joke.” Center then immediately acknowledged with a laugh.

From Flying Cloud, we continued across South Dakota so as to avoid convective weather in Nebraska.

While we were in some benign clouds west of Pierre, the battery-charge ammeter started acting oddly, sometimes showing an implausibly high charge current, sometimes showing a normal current, and sometimes rapidly slewing between those states. The plane was flying fine, and everything electrical seemed to be correct, so after some extensive troubleshooting, we decided it must be a bad sensor. Still, warry of the possibility there was actually a real problem, we kept rapt focus on the airplane’s systems all the way to Rapid City.

At KRAP, we were happy to find that they had an A&P (mechanic) on staff and able to immediately look at our plane. It took some effort for him to reproduce the problem, and even longer to fix it — foiled by a first attempted replacement part, an “ACU”, turning out to be bad new out of the box — but we were grateful that he was willing to set everything else aside to help us. By the time the plane was repaired, the weather had turned in Denver, and we would have been arriving well after nightfall, so we decided to do a local test flight around Mt. Rushmore, spend the night in Rapid, and proceed onward in the morning.

Morning dawned rainy, misty, and with low ceilings, so we departed IFR and entered the clouds at about 600 ft AGL. A couple thousand feet higher, we popped out above the clouds, and from there it was an uneventful trip back to Denver.

A couple weeks ago, the aviation community woke up to the latest editions of the VFR Sectional charts from the FAA in their EFBs and had a surprise on their hands: it was suddenly much, much more difficult to read the charts. The colors had changed!

Where the difference between things like Class B airspace boundaries and Victor airways had once been unambiguously clear shades of blue, they were now all kind of similar muddled variations on green.

Surely it was a mistake. After all, the FAA has a specification for the VFR sectional and TAC charts. The colors in the specification have been stable since at least 2022, and the blues in that specification are indeed Pantone 307U with various halftones; no mention of greens.

And indeed, in short order, the FAA acknowledged that the colors were changed a bit in the digital versions and even took the extraordinary step of reissuing the October 2, 2025 series of sectional charts with slightly tweaked colors (official notice here).

Except… that didn’t put things back to the way they were. Still green, just slightly darker green. No blues, at least not outside of water features. Here’s a back-to-back-to-back comparison of the old charts, the new charts, and the “revised” charts.

Reading the FAA’s announcement closer, it turned out that their goal was to make the digital versions of the charts match the print versions more closely. To a rounding error, nobody in aviation in the USA actually uses printed charts anymore, me included. It took a bit of effort for me to track down print versions of both the new October 2, 2025 edition of the Denver sectional as well as an earlier version from April 2023 so that I could compare them to the digital versions everyone really uses.

Wouldn’t you know it: both the new and the old printed sectionals matched the “green” of the new digital sectionals! We’d all been so focused on the digital versions for so long that we’d all forgotten what the printed versions even looked like.

In fact, the only real differences between the old and new printed charts are that the terrain-height browns are a bit darker on the newer ones and that they switched the printing from spot color to process color.  I’m not sure why the browns changed, since the spec didn’t; it doesn’t seem to be a gamut issue. The change from spot to process color seems to have happened at some point between April 2023 and October 2025, but the result is pretty similar in print.

 

The color issue for the blues seems to be solely about the digital versions.

As best I can tell, the digital versions of the charts prior to October 2, 2025 used pure hues according to some sort of hierarchy, with no mixing of hues. For example, the dark blue of a Class B airspace boundary had high priority, so it was shown on the chart as just dark blue. Similarly, the Victor airways were moderately high priority, so they were depicted as just light blue with no mixing in of the underlying yellows or browns. Stated differently, higher-priority colors would effectively mask out lower-priority colors.

For the October 2, 2025 digital editions, this approach seems to have changed. The hues depicted in the digital versions of the charts came to be mixed in the same way that they already had been on the printed charts: blues overlying yellows mixed to became greens, for example; no masking, no hierarchy, no priorities.

I think the FAA misunderstood the criticism of the new October 2, 2025 digital chart releases. People don’t want the digital versions to match the printed version more accurately. Instead, people want the digital versions to go back to the old way of masking, such as using pure blues instead of mixed greens for Class B airspaces (among other places).

As it stands, the new-style digital charts are really difficult to read. It’s bad in Denver, but it’s a disaster in places like Philadelphia where everything just melds together into one indecipherable mess. The only practical solution is to turn on things like digital airspace overlays in the EFBs.

Maybe we’ll all get used to it; after all, the printed charts have been this way for a long time. Or maybe somebody will realize that the old-style digital depictions really were superior and all of the color depictions will move back to that.

I’m a fan of 24-hour endurance races in motorsports. Since 2020, I’ve been contesting the true-24 race run by the 24 Hours of Lemons, first as a part of another team, and since 2021 with our own car. Here’s our 1999 Miata endurance car showing off its fancy lights:

One thing I’ve noticed is that it can be hard to find lists of all of the true-24-hour (or longer) races. To my knowledge, there are currently 10 such races for cars and at least 5 for motorcycles, though I know that my list of motorcycle races is definitely incomplete. Here are the races I’m aware of:

Cars:

24 Hours of Daytona – January – GT and prototype cars – Daytona International Speedway (road course), Florida, USA – IMSA

24H Dubai – January – GT cars – Dubai Autodrome, Dubai, UAE – 24H Series

Qlispe Classic 24 – May – Production cars – Qlispe Raceway Park, Washington, USA – Independent

24 Hours of Le Mans – June – GT and prototype cars – Circuit de la Sarthe, Le Mans, France – ACO

24 Hours of Spa – June – GT cars – Spa-Francorchamps, Stavelot, Belgium – GT World Challenge

24h Nurburgring – June – Production and GT cars – Nurburgring Nordschleife, Rhineland-Palatinate, Germany – ADAC

The 25:01 (25 hours plus 1 minute) – June – Production cars – Thunderhill Raceway, Willows, California, USA, – 24 Hours of Lemons

25h Fun Cup – July – Spec Beetle-silhouette cars – Spa-Francorchamps, Stavelot, Belgium – Fun Cup

ChampCar 24 Hour Classic – August – Production cars – Virginia International Raceway, Virginia, USA – ChampCar

24 Hours of Lemons True-24 – September – Production cars – High Plains Raceway, Colorado, USA – 24 Hours of Lemons

24H Barcelona – September – GT cars – Circuit de Bercelona-Catalunya, Barcelona, Spain – 24H Series

NASA 25 Hours (formerly the 25 Hours of Thunderhill) – November – Production cars – Ozarks International, Missouri, USA – NASA

 

Motorcycles (incomplete):

24 Heures Motos – April – Superbikes and similar – Bugatti Circuit, Le Mans, France – ACO

Vespa The Resistance – May – Vespa and Lambretta scooters – Zuera International Circuit, Zaragoza, Spain – Pinasco

Perry Mountain 24 Hour Challenge – June – Dirt bikes – Plantersville, Alabama – Perry Mountain Motorcycle Club

Bol d’or – September – Various classes – Circuit Paul Richard, Le Castellet, France – FIM EWC

 

Now, all of this isn’t to take away from the other great endurance races of the world, such as the Baja 1000 or the Dakar. Those simply aren’t 24-hour races in the sense meant here. Also, I’m vaguely aware of at least several karting 24-hour endurance races, but I’ll admit that I haven’t looked into them in depth yet.

Know of a true-24-hour (or more) motorsport race (even karting!) that I didn’t list here? Email me! I’ll update this list as new information comes to my attention.

Last update: July 4, 2025 (added 25h Fun Cup)

Sean and I were an hour east of Avoca, Iowa near Des Moines when it became apparent that we would not make it to Cedar Rapids on our extant tank of fuel. I was driving; it was Sean’s truck, and our F1000  race car was in our 24ft trailer behind us.

We were towing to Road America, a trip we’ve done many times, and we really wanted to make our next stop at a particular Kwik Star truck stop on US-151. (They had genuinely good food.) The distance to that Kwik Star: 102 miles. The estimated range per the truck’s computer: 100 miles.

We needed a draft to make it.

The key to a good draft is that you initially need to go just a little faster than your target speed so that you can come up upon a draftee that’s meeting that target on the nose. I considered a few prospect semis, but they were all going too slow. A few more miles ticked by. Then we came upon a blue tractor with an unassuming off-white trailer. Haz-mat, per his placards, “1906” and “1760” – cleaning solutions.

I hopped into his dirty air at a respectful yet efficient distance and instantly saw our mileage rise. He maintained speed with what seemed like cruise control; I modulated to hold station with my right foot.

As with many budding relationships, things soon got a little rocky. Truckers generally don’t like being drafted, especially by non-semis. When the two of our vehicles came up behind a slow-moving truck a few minutes in, my draftee merged into the left lane not far ahead of a passenger car, preventing me from following. He then slowed way down, giving me plenty of room to go in front of him and onward. I chose to stay behind the truck in the right lane, betting that my target wouldn’t like going so slowly for very long. Indeed, he sped back up to his previous pace, and I maneuvered back to my place behind him.

Things might have stayed somewhat strained, but for two truths: first, we came upon more traffic a few miles later, and second, my draftee was courteously aggressive about staying in the right lane except for passing. Of course, the problem with being courteously aggressive about staying in the right lane is that it’s easy to be screwed by traffic in the left lane when someone ahead is driving slowly in the right.

An opportunity presented itself when my draftee changed to the left lane for a pass, I followed, and he returned to the right after completing the pass. I, however, stayed in the left lane, as I could see he would very soon come up behind slow-moving traffic in the right lane. I could also see several cars behind me waiting to pounce. So, to the minor detriment of the cars behind me, I matched his speed but in the left lane, staying slightly back from his rear door. “I’m going to play some 2D chess,” I told Sean. I’d meant “3D chess”.

Sure enough, a minute or two later, he came up on the traffic in the right lane, and I slowed slightly more so as to make a big, obvious gap for him to come back left. Not wanting to be too pushy, I deliberately chose not to flash my lights in signal. He hesitated a moment, then politely merged back left in front of me, and promptly gave me the “hazard light blink”.

“Oh good, we’re friends again!” I told Sean.

And so it went. I’d draft him, then when we came up on traffic, I’d go left and block for him so he could move over without slowing down. He’d go over, give me a couple hazard blinks, and then we’d make the pass together.

This continued along I-80 for over an hour, then onto I-380 towards Cedar Rapids.

I almost dropped the ball on I-380 with a nearly missed block. When I realized the window was closing, I might have been slightly aggressive about moving over to, let’s be honest, cut off a car that was steaming up the outside. “Sorry!” I thought to the car now behind me — hey, I had to protect my friend.

When we got within a few miles of the US-151 turnoff from I-380, I slowed well-down, stayed right, and my semi friend went ahead in the left lane. He gave one more set of hazard blinks in goodbye, and traffic soon blocked my view of him.

I thought that was then end of it, but no! Like the awkwardness of discovering you’ve parked in the same direction as an acquaintance after already completing your goodbyes, it turned out that he too was following US-151, and we came up behind him again on the ramp. I wasn’t sure what to do; I didn’t want to be weird about it.

When we both made yet another turn to continue along US-151, I figured, “Eh, why not?” and we fell back into old patterns.

Thanks to the draft, our mileage in Sean’s truck had improved from 6.2 mpg to 7.5 mpg, and our target Kwik Star came into view with an estimated 35 miles of range left.

Sean and I had shared the previous four turns with our big-rig friend, but not our fifth: at the light for the Kwik Star, he stopped in the rightmost straight-through lane, and we slid into the right-turn-only lane.

As we came up alongside his rig, I slowed more than was needed and looked to my left at his cab with a smile and a wave. Inside, I could see his head turned back at me, a wave on his hand as well. He gave a couple quick toots of his horn in salute, and as his light turned green as we turned right.

Our highway friend lumbered off into the distance; we will never see him again.