Runway Width vs. Runway Excursions

It would be labor intensive, but an average number of operations per airport would get the percentages more accurate?

Do you think pilots see “big runway” and just relax a bit too much?
 
455 Bravo Uniform said:
It would be labor intensive, but an average number of operations per airport would get the percentages more accurate?

Do you think pilots see “big runway” and just relax a bit too much?
Click to expand...
How'd you like my 40' runway? 10c is 36' and the flight school there is slammed. I did most of my landings for ppl at those 2 widths.

Conversely large runways absolutely mess dad up.

But yeah, total airport ops and average number of landings per excursion of would be interesting
 
Ok, I can now see what you were trying to explain. Thanks for sharing that.

Could it be that runway widths are not reported at wider airports that have accidents without excursions? That would make the denominator smaller for the big airports. And maybe the widths are reported more often for the smaller airport non-excursions to somehow indicate that they looked at it and was not a factor?
 
I love your statistical analysis.

I have had more than one learner where I needed to take control to keep them from touching down off the edge of the 150 foot wide runway.

I was often surprised when we changed to the 75 foot wide cross wind runway that they still landed very near the edge.

With most it was a sighting (focus) problem or not coming straight back with the cyclic (stick) in the landing flare.

This was usually early on and always correctable.
 
Most videos I’ve seen here… Pilot can’t se to land in centerline.

For that matter, they can’t even take off of centerline.
 
Vance Breese said:
I was often surprised when we changed to the 75 foot wide cross wind runway that they still landed very near the edge.
Click to expand...
Was flying right seat in a T-18 with the older owner. On final, he was well off the centerline...

"Hey, Cecil, we're kinda to the left."

"I line up on the edge of the runway because it has better contrast." Turns out he had cataracts.....

Ron Wanttaja
 
My wife always does much better on narrow runways. I suspect it is because she focuses more. Big, wide runways seem to intimidate her. But runway width has never been a deciding factor as to where we land, as long as it is sufficient for a Bonanza.
 
One factor that might skew the data is the smaller the runway the less likely it will be reported.

*edit* I see this has already been brought up. oops.
 
Wider and generally longer runways attract operations in A/C that have higher Vso's?
 
No solid conclusions can be drawn from these data. Needs to be normalized for usage within each width range.

Even if you normalize by the amount of runways in each range, it changes the ratio. There are a lot more 71-85’ wide runways than any other width.

Nonetheless, I’d hypothesize that use, especially by flight training, would increase the ratio in that range.
 
Kritchlow said:
Most videos I’ve seen here… Pilot can’t se to land in centerline.

For that matter, they can’t even take off of centerline.
Click to expand...

Or roll the nose wheel anywhere near the painted taxi line.

IMO, it all relates to back to the centerline. Whether landing, rolling or climbing out, if the aircraft is lined up with the centerline the width of the runway irrelevant.

Any significant excursion off the centerline indicates the pilot was not capable of successfully controlling the plane during whatever factor initiated the loss of directional control.
 
Student pilot raises hand. "I Know, I know the answer!" "We need narrower runways!"
I had one of those for a student once. I don't know what ever happened to him....or how he passed the checkride. But he did.
 
On a wide runway, you can get quite crooked before your get close enough to the edge to notice, by which time it may be too late to correct.
 
This is a totally bogus assumption. Aircraft runoff small runways at uncontrolled airports with smaller runways that are unreported. In some cases, even when there should have been an NTSB notification.

Another issue is weather. No one is doing low instrument approaches in a pouring rain with terrible braking reports on to a 2500 x 25 runway.
 
Clip4 said:
Another issue is weather. No one is doing low instrument approaches in a pouring rain with terrible braking reports on to a 2500 x 25 runway.
Click to expand...
Which shouldn’t be an issue. If it’s too crappy to land, they shouldn’t be landing.
 
Wide runways are more likely to be reported because they are more likely to be at towered airports, airports with active airport management, and in general, more people around to witness.

Narrower runways are at smaller airports, with less traffic, or more likely to be private airports as well. A runway excursion gets gathered up and quietly put away in a hangar, 49 CFR 830 be damned.

Also, narrower runways often have comparable surfaces off the runway that result in no or lesser damage, so they don't get reported. An excursion at a larger airport might involve hitting taxiway lights, signs, or dropping off pavement onto a lower surface causing a prop strike...or maybe even heading into a drainage ditch.
 
When I was instructing, despite having 150 foot wide runways, I always taught and expected centerline between the mains. Then the width of the runway is irrelevant.

Our local DPE expressed amazement at how all my PP candidates could maintain the centerline.
 
Last edited:
AvNavCom said:
Landed on an incredibly wide runway the other day. It was only 150' long though, so that was challenging.
Click to expand...
All part of the fun-and-games of working with database information input by humans. CEN23LA175, for example, lists the runway as 100 feet long and 6,000 feet wide.

1778773389214.png
Ron Wanttaja
 
midwestpa24 said:
When I was instructing, despite having 150 foot wide runways, I always taught and expected centerline between the mains. Then the width of the runway is irrelevant.

Our local DPE expressed amazement at how all my PP candidates could maintain the centerline.
Click to expand...
I used to have copilots who got ****ed off because I wouldn’t let them land on a 50-ft-wide taxiway with a dogleg when they couldn’t keep the ailerons on opposite sides of the centerline anywhere else.
 
rogers M2 F2.jpg
NOTE: THIS PHOTOGRAPH IS A DIRECT RESPONSE TO THE PROMPT. IT IS NOT RANDOM.
 
wanttaja said:
...
Any more thoughts?
...
Click to expand...

1. Aim small, miss small?

2. My gut feeling is perhaps it should be framed (runway excursions/total accidents)/total evolutions. But I know that would be darn near impossible (or at least well beyond the ambition of an armchair quarterback such as myself).:biggrin:
 
Flying Keys said:
No solid conclusions can be drawn from these data. Needs to be normalized for usage within each width range.

Even if you normalize by the amount of runways in each range, it changes the ratio. There are a lot more 71-85’ wide runways than any other width.

Nonetheless, I’d hypothesize that use, especially by flight training, would increase the ratio in that range.
Click to expand...
This. To test the null hypothesis--that excursions are equally likely on any size runway, you have to know the percentage of operations on each runway width. An excursion might be 100x more likely on a 25' wide runway, but if there are 10,000x operations on 150' wide runways, the data could look like that.

Plus the reporting issue.
 
Lindberg said:
This. To test the null hypothesis--that excursions are equally likely on any size runway, you have to know the percentage of operations on each runway width. An excursion might be 100x more likely on a 25' wide runway, but if there are 10,000x operations on 150' wide runways, the data could look like that.

Plus the reporting issue.
Click to expand...
I think you are missing what the chart is showing. It is showing runway excursions as a percentage of total accidents. This should normalize the data for total number of flights, unless of course airports with large runways are safer in every way except for runway excursions.

I wonder how often there are runway excursions that would have been prevented by another 10 feet of width? I know very little about runway excursions, thank God, but YouTube is full of examples where it was going to take a lot more than a few more feet ...
 
dcat127 said:
I think you are missing what the chart is showing. It is showing runway excursions as a percentage of total accidents. This should normalize the data for total number of flights, unless of course airports with large runways are safer in every way except for runway excursions.

I wonder how often there are runway excursions that would have been prevented by another 10 feet of width? I know very little about runway excursions, thank God, but YouTube is full of examples where it was going to take a lot more than a few more feet ...
Click to expand...
AMEN and of course!
 
I have some thoughts but would need to look at this more in depth. How would I recreate the dataset myself?
 
wanttaja said:
How does the width of the runway factor into runway excursions? Seems like the narrower the runway, the more excursions should occur.

The weird thing is, the data doesn't seem to support that hypothesis
Click to expand...
I don’t think it’s weird at all. I would expect that. Plenty of width means you can afford to be looser. So we are. Narrow naturally focuses us. There are probably hundreds if not thousands of examples both in and out of aviation.
 
dmspilot said:
I have some thoughts but would need to look at this more in depth. How would I recreate the dataset myself?
Click to expand...
1. Download the NTSB data from:


"avall.zip" is the file you want.

2. Inside the zip file is "avall.mdb", a Microsoft Access file. Load that into access.

3. The file contains a set of tables that have to be assembled in queries to extract the data

4. Here's the query I used in the runway excursion study:
1778857949796.png
The boxes, of course, are the tables that were included in the original download.

The "Event ID" is the common factor between the tables (this is different from the NTSB number). You can see which parameter from which table is being included in the final output. The matrix at the bottom shows what parameter from what table is included in the output.

The key thing for this particular study is the "Events_Sequence" file, which includes an "Occurrence_code" with an accompanying "Occurrence_description". You can search for "Excursion" in the description to extract only the nine types of excursions, or you can search by the "Code"...a six-digit figure. If it ends in "300", it's an excursion; if it starts with a "3" it happened during takeoff, and starting with a "5" means it happened during landing.

You can sort this in Access, but I usually just get the raw data out and import it to Excel.

Ron Wanttaja
 
wanttaja said:
You can sort this in Access, but I usually just get the raw data out and import it to Excel.
Click to expand...
As a short-cut, I've uploaded my basic Excel excursion-data output to:


It's 43 megs.

(Edit: Note that each accident in the database may have more than that one occurrence entry. So while there are ~65,000 rows in the spreadsheet, some accidents have two or more entries.)

Ron Wanttaja
 
Last edited:
dcat127 said:
I think you are missing what the chart is showing. It is showing runway excursions as a percentage of total accidents. This should normalize the data for total number of flights, unless of course airports with large runways are safer in every way except for runway excursions.
Click to expand...
Yes, that's my basic process...show how often a particular event occurs in relation to the total number of accidents. It's like results showing that 15% of all Cessna 172 accidents involve loss of engine power, vs. 29% of the Lancair accidents. There are three times as many Cessna 172s in the registry than Lancairs, but looking at the percentage of accidents gets around that. Unless there are too few accidents; I generally won't go public on a particular type unless I have at least 50 accidents to look at.

However, in retrospect, I'm not sure this approach is the best one for excursion data. My process includes only accidents where the runway data is available, which is only about half the total number of accidents for the time period. I wonder if that's over-emphasizing the excursion data; are the NTSB investigators not bothering to record the runway dimensions in cases where it didn't matter, which would affect the results.

This is why I post results like this to POA; gives me a bit of a peer review and I get suggestions of how the analysis could be improved or where problems might exist. Rather than use this is as a basis for a later formal article, I'll just leave it here as a curiosity.

Besides, the basic problem still exists: The non-intuitive result that says the rate of excursions decreases with runway width. Don't get me wrong, I *love* non-intuitive results, but I need some logic that leads to the right conclusion.

Ron Wanttaja
 
Back in the early 80's I was flight instructing at KBED (Hanscom Field, Bedford, MA). The joint use Airport had long wide runways (still does). My students found it difficult to land on the centerline, often due to a lack of confidence in their ability to do so. In those days the Tewmac airport in Tewksbury, MA still existed. As I recall it was about 3000' long and 29' wide. A quick trip to Tewmac would inevitably result in a landing on the centerline. Followed by my saying - if you can do it at Tewmac you can do it at Hanscom. It was a successful technique. Now that I instruct in a level D FFS at a 142 school I rarely get clients who don't land on the center of the runway. I do get stories though. One Captain who used contract co-pilots from time to time told me that when a co-pilot did not land on the centerline she would say "that's okay, the centerline is for the professionals." Harsh but effective.
 
I’d think that narrow (and short) runways tend to be used by more experienced pilots. Airparks, back country, cropduster strips aren’t usually the destination for novices.
 
Another factor this dataset may overlook is whether the excursion was from running off the side of the runway or going off the ends. Width may not have been a factor in at least some.
 
wanttaja said:
Yes, that's my basic process...show how often a particular event occurs in relation to the total number of accidents. It's like results showing that 15% of all Cessna 172 accidents involve loss of engine power, vs. 29% of the Lancair accidents. There are three times as many Cessna 172s in the registry than Lancairs, but looking at the percentage of accidents gets around that. Unless there are too few accidents; I generally won't go public on a particular type unless I have at least 50 accidents to look at.

However, in retrospect, I'm not sure this approach is the best one for excursion data. My process includes only accidents where the runway data is available, which is only about half the total number of accidents for the time period. I wonder if that's over-emphasizing the excursion data; are the NTSB investigators not bothering to record the runway dimensions in cases where it didn't matter, which would affect the results.

This is why I post results like this to POA; gives me a bit of a peer review and I get suggestions of how the analysis could be improved or where problems might exist. Rather than use this is as a basis for a later formal article, I'll just leave it here as a curiosity.

Besides, the basic problem still exists: The non-intuitive result that says the rate of excursions decreases with runway width. Don't get me wrong, I *love* non-intuitive results, but I need some logic that leads to the right conclusion.

Ron Wanttaja
Click to expand...
The issue is that the data aren’t normalized for usage. You could have one 40 foot runway and twenty 75 foot runways in the US and it’d stand to reason that because more 75’ runways exist, you’ll see a higher incidence count on those runways because it’s reasonable to presume there’s more overall use. That’s not an accurate measure, either, because the one 40’ runway might actually be used 100x as much as the 75’ runways. I’m just pointing out the flaw in the study.

In reality, the width range that has the highest amount of runways in the US happens to be the one that includes 75’, by quite a bit of margin. Something like 1726 hard-surfaced runways in that 71-85’ range compared to 41-55’ (619) and 56-70’ (947). Could include soft runways but one might presume there’s a lot less usage and/or reporting on those.
 
dcat127 said:
I think you are missing what the chart is showing. It is showing runway excursions as a percentage of total accidents.
Click to expand...
There's more than that in the table though, including the "percent of excursions" column.
dcat127 said:
This should normalize the data for total number of flights, unless of course airports with large runways are safer in every way except for runway excursions.
Click to expand...
Nope. It normalizes the data for the number of accidents on that width of runway. The question this math would attempt to answer is, "If there's an accident on an x' wide runway, what is the likelihood that it is an excursion?" But there's a lot of confounding, even to answer that question.
 
wanttaja said:
One of the features of the NTSB accident database is that it lists the various occurrences that happen during each accidents. Here are the flags for a number of accidents where the aircraft departs the runway:
After landing Runway excursion​
Landing Runway excursion​
Landing-aborted after touchdown Runway excursion​
Landing-flare/touchdown Runway excursion​
Landing-landing roll Runway excursion​
Other Runway excursion​
Takeoff Runway excursion​
Takeoff-rejected takeoff Runway excursion​
Unknown Runway excursion​
Recently, I realized that the dimensions of the runways involved were often included the database. Which leads to the question:

How does the width of the runway factor into runway excursions? Seems like the narrower the runway, the more excursions should occur.

The weird thing is, the data doesn't seem to support that hypothesis. I divided the runway width into six categories, counted the excursions involved in each category, and divided it by the number of total accidents for each category.
View attachment 152027
Note that the percentage is the percentage of accidents at that runway width...yes, there are more landings (and accidents) on 75-foot-wide runways, but we're looking at the percentage of events at that size.

For me, this is a non-intuitive result... it doesn't seem reasonable. I can think of a few reasons for this...for instance, Flight Schools are probably more likely to be using the wider runways, so that student excursions make up more of the activity on the 70+ runways. Seems like classic taildraggers would probably have more experienced pilots, but the results for the production taildraggers is practically the same as above. The analysis doesn't take the runway *length* into consideration, either...some of those excursions could be off the end, rather than off the side.

Any more thoughts?

Ron Wanttaja
Click to expand...
Maybe people getting a bit lazy on speed and centerline at bigger runways? I have done all my training at a 12,000-foot, 200-foot wide runway. My first instructor said we could do touch and goes without turning. But I’ve seen people go off the sides in GA planes.
 
You must log in or register to reply here.
Back
Top