The Loss of N2685

Note:  This report is in inverse chronological order...the most recent material immediately follows, and the further you scroll down, the older the reports are.  These older reports are retained to illustrate how the understanding of an event changes as time goes by.
 

Final Update:  19 July 2001

The NTSB Final Report is in:

"The National Transportation Safety Board determines the probable cause(s) of this accident as follows.

An overload failure of both flying wires, on the underside of the right wing, while the airplane was climbing, due to improper balance between the two flying wires, by unknown person(s)."

If you read the full report, you'll see that one of the forward flying wires and the turnbuckle on the other forward wire failed on the right wing.  The analysis didn't find any other reason for the wire or turnbuckle to fail:  No corrosion, no workmanship errors, no apparent mechanical fault.

It is certainly a bit frustrating.  We are left, essentially, with a "default" conclusion:  There wasn't anything wrong with the wire or turnbuckle and the airplane wasn't performing aerobatics.  The only scenario that's left is that the two parallel bracing wires were significantly different in tension, and the airplane hit turbulence that exceeded the capacity of the single cable assembly supporting the load.

The question is, what's the chance it's going to happen to one of OUR airplanes?  How sensitive is the Fly Baby bracing system to uneven tension?

If you look on the online NTSB record database on Fly Baby accidents, you will NOT find a similar case.  There have been other wing failures, but they have all be attributed to other, more obvious causes (spar carry-through rotted away, etc.).  The online reports only go back to 1983, but Pete Bowers isn't aware of any similar instances before that.  In over 40 years, this is the first accident where a wing failure couldn't be traced to a mechanical fault.

It seems like most Fly Baby owners have been able to keep their wire tension close enough.

There are some things in the NTSB report that owners and builders may want to take note of:

  1. The owner bought the airplane a few months prior to the accident
  2. The owner bought it as a source of parts.
  3. The date of the last annual conditional inspection was unknown
  4. The owner did not know the maintenance status of the airplane, nor does he know the location of the aircraft maintenance records (logbooks), nor who had last worked on the rigging of the aircraft.
Everyone has to make their own judgement.  But I'd have to describe this situation as a bit...casual.

We don't know the history of the aircraft prior to the last owner's purchase.  He bought it not long before the accident.  We don't know if that means that the airplane had been sitting in a barn for the last ten years, or whether the previous owner flew it every day, and twice on Sunday.

The owner bought it for "parts," which certainly implies that this 27-year-old homebuilt may not have been airworthy.  But yet, selling a perfectly-good homebuilt as "parts" is a common liability dodge in the homebuilding world.  It may have been a perfectly-good airplane, and the previous owner sold it as "parts" so he wouldn't be held liable for any future accident.

(Side note:  This approach to shake a liability tail is pretty weak.  Do you think Bendix or Slick never get sued for magnetos involved in accidents?)

We don't know if the last owner intended to fly the aircraft itself, nor does the NTSB report say whether this was the first flight after the last owner's acquisition of it.

In short:  I don't think this case closely reflects the situations of most Fly Baby builders and owners.  This is the first cable failure in forty years of Fly Baby flying.  Routine attention to equalizing the tension in the flying wire pairs is probably sufficient to ensure safe flight.

Let me recap Pete Bowers' instructions on how to get the flying wire tension equal:  Tighten both turnbuckles by hand until you can't tighten them any more, then turn them one or two additional turns using a nail for additional leverage.  Paul Bowyer used this system to tighten his wires, then checked them with a cable tensiometer.  They were within 13% of each other.  Given 40 years of success using Pete's method, I'd say it's probably all right.  Read more on Paul's experiment further down on this page.

In talking to folks about this accident, one man reminded me of the old classic way of setting cable tension:  Pluck the cable and listen to the "thrum."  The human ear is a pretty good gauge.  Remove the wire-bracing "arrows" before doing this, though, because they'll tend to damp the vibration.

If you can borrow or buy a cable tensiometer, all the better.  The precision units cost about $800, but I see Aircraft Spruce has economy ones for as low as $32.  I bought their $130 one.  It only took me a few minutes to get my cable sets balanced.  The Economy one would probably work just as well, for balancing the pairs.

The rest of this page contains the previous reports I did on this accident.  Much of the information is obsolete, but there's still some good data there.

Ron Wanttaja.





6 November Update

Just a minor update, with some other thoughts.  I talked to the NTSB investigator this week.  The metalurgical tests are in on the cables and turnbuckles in the Jones accident, but he declined to tell me what the result was, as he didn't want me to pre-empt the official report..  He said the official report will be out in about 60 days.  It's his call, and he was kind enough to discuss the accident with me earlier.

So we won't know if the metalurgical tests indicated a flaw in either the cable or the turnbuckles for a month or two.  In the interim, though, I've received some interesting information from folks. The accident airplane was about twenty years old, and two folks have brought up the subject of cable longevity:

Stephan Holger had an interesting insight:

"The German hang gliding association (DHV, http://www.dhv.de) is
well known for it's profound technical knowledge of ultralight aircrafts
that mostly use wire cables, too. The DHV requires the German hang gliders
to be checked by the manufacturer every second year. Part of the standard
procedure is the replacement of the lower cables. Costs around $150. The DHV
might have research information about wire cable aging you could request."
Also, I met Tom Staples again at Arlington back in July.  He says he replaces his bracing wires every five years.

Let me reiterate right now:  We don't know what the result of the metalurgy test is.  If the hardware got a clean bill of health, unequal tension of the flying wires is probably the leading suspect.  However, there is some question regarding the cable longevity.  The turnbuckles and shackles probably be inspected and re-used, and to replace the cables every five years probably will cost $60 or so...just $12 a year.  Cheap insurance, that.

Look for an update and summary when the final report is released.


6 August Update

Since it's been several months, I thought people would appreciate some sort of update.  I talked the NTSB investigator this week.  He has not received he results of the metalurgy tests on the wire and turnbuckle, but he's going to do some checking on them.

I received a few interesting email from fellow builder/ownes and another aircraft designer with some thoughts on the accident.  I'll quote excerpts here, and add my own thoughts.

Paul Bowyer on Cable Tension

Paul Bowyer writes:
After reading the disturbing news of another Flybaby wing separation I decided to try and get a more accurate tension reading on the parallel flying wires to make sure they were equal in tension. Thought you might be interested.

First I "tuned" the cables in the normal way and tightened using Pete Bowers recommendation. I then went on a hunt for a proper cable tensioning meter. One of our local helicopter AMEs loaned me his cable
tension measuring tool. "$1,000 if you drop it Paul or just  a box of donuts if it still works!"  Great guy and obviously more concerned about people than equipment- thanks Alan.

The results of the testing  were somewhat interesting and surprising. Static tension on the cables ranged from 150pounds to 170pounds approx.... What surprised me the most was how sensitive the meter was to the smallest amount of turnbuckle adjustment, reading 50 to 60 pounds variation with very small
turnbuckle adjustments.

Interesting data.  First, I'm happy that tightening the cable using Pete's method resulted in tensions as close as Paul measured.  Worst case, that's only a 13% difference (between min and max...the difference between the AVERAGE tension would be less).  I don't think a moderate amount of error is going to be a problem....after all, the cables stretch a bit, and that should help balance things out.

However, some people might say:  "If the cable is already under 200 pounds of tension, doesn't that reduce the total carrying capability of the cable?  If the cable is designed for a 1600 pound load, won't it now break at 1400 pounds?"

The answer, according to some engineers I've spoke to, is no:  This initial tension (called "Preload") does NOT subtract from the total carrying capability.  The tension in the cable won't exceed the preload until the load itself exceeds the preload.

It's a hard concept to understand.  I couldn't get a good "Wire Tension for Dummies" analogy out of either of the guys I talked to.  But here's one way to get a glimmering of understanding:

Bolt a table to the floor, and drill a 1/8" hole through it.  Attach one end of a 1/8" cable to a scale, then use a short bit of cable to a turnbuckle.  Hang the turnbuckle from the ceiling.  Pass the cable through the hole in the table and nicopress a thimble into the end under the table.

Start tightening the turnbuckle.  The nicopress fitting won't go through the hole, so the cable gets tight.  Turn the turnbuckle until the scale reads 200 pounds.

Now:  Hang a 1000 pound weight from the thimble under the table.  You look under the table, and find the cable has stretched enough that the nicopress fitting is no longer in contact with the table.

Does the scale read 1000 pounds plus the 200-pound preload?  Of course not...it's hanging free.  The cable is only under the tension supplied by the 1000-pound weight.

If you'd added weight slowly to the cable, the scale would have read 200 pounds until the first ounce OVER 200 pounds had been added.  Then the scale would have increased normally.

This is really a poor analogy to the aircraft installation...but I think it illustrates how the preload might work.  CIRCA Nieuport designer Graham Lee doesn't agree that the preload does not reduce the total carrying capacity...see the next section.

The important point is that cables like preloads.  If a wing cable will be normally operating at 500 pounds of load, a 500-pound preload means the tension in the cable stays constant.  Cycling induces wear and metal fatigure.  The preload eliminates cycling, and the cable lasts longer.  Even if you don't think I'm right about the preload, DON'T slack off your cables to "increase their carrying capacity."

Graham Lee on Design Issues

Graham Lee is the designer of the CIRCA Nieuport replica, a 7/8th scale aluminum-tube replica of the WWI Nieuport fighter.  I used to correspond with Graham back in the days I was building one of his aircraft (until I got lured away into the Fly Baby club...got to be hard to go work on building a plane when I had a perfectly good flying one out at the airport).

Anyway, Graham has some interesting points:

I just heard about this incident and was directed to the web page.  There was an incident here in Alberta a lot of years ago and the MoT  published photos of the cable, snapped at the nico and within about 9 inches for the rest of the strands.
My reason for writing is the 75% figure mentioned as the possible load on the front spar.  It could be 75% in this instance, but at high angle of attack I use 100% on the front spar. The lift vector is well forward in a pull up. The load in a hard pull up can be over 3 G! Next, I use I think the same 30 degree (counted as 60 degrees from the vertical). 60 degrees is double the actual vertical load. So we have ~1000 pounds. Divided by 2 (wings) is ~500 pounds. Multiplied by ~3 (G) is 1500 pounds and multiplied by the load angle 2, = 3000 pounds on the front wires.. If the wires aren't evenly tensioned the load is on one wire and it is only good for ~2000 less the tension put into it on assembly.

If the plane is truely designed for 5.7G, the lift wires should have 2850 pounds tensile each and should be a single wire, free to self align at the single very large turnbuckle on the end. (ie: looped around and then back to the anchor point.) The rear spar load can be roughly half of the total load with 5.7 g. This similar incident has happened a couple of times and I think the lift wire system could be re-designed to avoid it.  There is one other scenario that can cause failure. If the lift wires are not in tension on the ground, and the plane changes from nil to positive load, the whipping of the lift cable can induce several numbers of G more than the airframe G forces. This might be from even a "lump" in the air, or a sudden stick movement.

In the first paragraph, Graham mentions a similar Canadian accident.  Can any of our Canadian friends chase down an accident report on this one?

Graham's suggestion about using single lift wires is valid...my engineering friends have made the same suggestion.  All aircraft designs are compromises; in this particular case, I think, Bowers went with two cables for buildability.  Going to a single cable system would require 3/16" cable, which is much harder to work with than 1/8".  One way around this using 1/8" or 5/32" cable would be to use a triangular plate, with the two lift wires attached to two vertices and the remaining vertex attached to the wing.  That way the loads in the two cables would balance out.  You'd have to come up with a good way to attach the system to the spar.

It's ironic indeed that the one simple solution to the two-cable problem...solid bracing rods...were themselves probably to blame for Steve Hinton's accident last year.  Can't win for losing, I guess....

Finally, note that Graham says the cables are good for their rating "...less the tension put into it on assembly."  Graham feels that the preload should be subtracted from the total load-carrying ability.  He may be right, not withstanding my writeup in the earlier section...I'm neither an expert, nor have the people I've talked to take a serious look at the wing structure.  But Graham does agree with the need for preload.

Tom Staples on Cable Age

Tom Staples of Victoria BC has been flying his Fly Baby (with the stock 1/8" cables) for about twenty years.  We chatted at the Arlington Fly In this summer.  He feels that the age of the cables on the accident might be a factor.  He replaced his a few years back.

Wrapup

The metalurgy tests should tell the tale.  If a flaw or corrosion is found, we'll at least have a cause.  If not, we'll be pretty much left with speculation.  .

First Follow-Up:  19 May 2000

First off, the NTSB Preliminary Report has been added to their web page.  In addition, I spoke to the NTSB investigator, Robert Hancock.  Preliminary investigation was performed by FAA Accident Investigation, and the data was forwarded to Mr. Hancock.

As you'll see from the NTSB report, two flying wires were found broken.  Mr. Hancock reports they were of the stranded type (as called for in the plans), not solid like the Hinton case.  He said that they were the forward pair on the underside of the wing on the right-hand side.  The cable itself broke on one assembly, and the turnbuckle on the other.  The parts have been sent out for analysis.

The spar plates (which were the cause in the Hinton accident) are apparently not involved in this accident.

Mr. Jones was a recent purchaser of the aircraft.  According to the FAA database, it was built in 1973 by someone named "Hing" (hence the name on the NTSB report).

Mr. Hancock does not have any information on the history of this particular aircraft.  If you know anything about it, email me and I'll pass you his telephone number.

Speculation follows:

There has been at least one previous accident (before the period covered by the online NTSB database, alas) which was traced to cable failure.  In this case the Nicopress fittings were incorrectly installed (using a pair of pliers instead of the Nicopress tool).

It is unknown whether the accident aircraft had standard 1/8" cable flying wires or whether the aircraft had been upgraded to a larger size.  However, the 1/8" cables are adequate for "normal category" flying; Bowers himself only recommends upgrading to 5/32" if aerobatics will be flown.

The flying-wire assemblies on the Fly Baby use stranded 1/8" aircraft cable and AN130-16S turnbuckles.  The cable itself is good for about 2000 pounds, and the turnbuckles are rated at 1600 lbs (hence the "-16").  A cable assembly is only as strong as its weakest element, so each flying wire assembly can reliably support 1600 pounds.

Of course, there are are total of eight flying wires under the two wings (remember, the "flying wires" are the ones UNDER the wing; they support the aircraft weight in flight).  Since a single wire can support more than 150% of the Fly Baby's total weight, eight of them must give the plane a heck of a G-capability, right?

Sadly, that's untrue.  If you look at the Fly Baby, you'll see the flying wires attach to the bottom of the wing at a rather steep angle.  Here's a simplified sketch:

The lift is going straight up, but the wire is only about 30 degrees to the wing.  This tends to increase the tension in the bracing wire.  Tension in the wire is equal to the force upward (the lift) divided by the sine of the wire angle:  T = F / Sin(Angle).

So, let's take a look at Fly Baby.  Let's assume a 925 pound airplane, with stock cables rated at 1600 pounds.  How many "G"s is it good for?  In this case, we're going to turn the above equation around, and solve for the Force (lift).  The equation is now Force = Tension x Sin (Angle).

Ah, there's one thing we forgot:  We've got eight flying wires.  So the equation now becomes:

Force = Number of Wires x Tension x Sin(Angle)

To get the Gs this is equal to, we just divide by the aircraft's gross weight.

G = N x T x Sin (Angle) /Gross Weight

The angle is about 30 degrees, the number of wires is eight, and the maximum allowed tension is 1600 pounds.  Our formula becomes 8 x 1600 x Sin (30 degrees) which comes out to 6400 pounds.  Divide by the gross weight of the aircraft (1000 pounds) and we learn that the stock wire system is theoretically capable of 6400/925 or 6.9 Gs.

If you look on page 8-33 of your plans, you'll see a Finnish Fly Baby, upside down, with 5286 pounds on his wing...a weight corresponding to about 5.71 Gs.  This was performed in front of government representatives, and there was no damage to the aircraft.  5.7 Gs is equal to the ultimate limit for Normal category.

Alas, things aren't this simple.  The four bracing wires on each wing attach two to each spar...and the weight isn't evenly distributed between these spars.  I'm told that, at high G loading, the center of pressure shifts aft and DOES tend to equalize the loading between the two pairs of flying wires per wing.  But...if we take a worst case and assume the forward pair is taking 75% of the load.  We'll assume four flying wires (just the two forward pairs) and assume they're supporting 75% of the total gross weight:  G = 4 x 1600 x Sin(30) /(0.75 x 925) or 4.6 Gs.   If this seems low, remember that we're making an assumption that the forward wires get three-quarters of the load.  I'm sure they get more than half, but 75% might be too high.   In any case, the components themselves have margin.  The previous owner of my Fly Baby actually had a set of bracing wires tested to destruction.  They failed at about 1975 pounds... which, if you use the equation above, comes out to about 5.7Gs, even with the 75%-load-on-forward pair assumption.

[Update 9 June 2001:  The NTSB investigator discussed this issue with Pete Bowers.  Pete says the normal distribution is 60% forward/40% aft (vs. the 75/25 I used above).  He says the load may shift as much as 5% under load.]

All right...in my long-winded way, how does this relate to the recent accident?

Simple:  The bracing design had more than enough margin for the flight in question.  Based on the currently-available data, my opinion is that there was probably a flaw in one of the cable assemblies, weakening it.  The turnbuckle or cable may have been damaged or corroded. The aircraft was almost 30 years old; you can see why the NTSB investigator would like to see a history on it.

There's another factor that's VERY important:  Equalizing the tension in the pairs of bracing wires to each spar.  Here's what Pete says, on Page 8-33 of the plans:

"One possible reason for wire "weakness" is so obvious that it should not have to be mentioned.  When two parallel wires are installed, as on Fly Baby, both should be rigged to the same tension to that each one is carrying an equal part of the load.  If one is allowed to slack off significantly, the safety factor of having two wires is cancelled.  If the single wire that is carrying all the load under such conditions breaks from over-stress, all the load at that moment is transferred to the backed-off wire.  Just the instantaneous taking-up of slack with add an impact load to the force that just broke the other wire.

"This is not just a Fly Baby problem; is applies to all designs with double-wire bracing.  Actually, by using wires, which have a small degree of stretch even under tension, the Fly Baby is at a slight advantage over designs that use "hard" streamline aircraft tie-rods.  If one of a pair of these has any slack at all, it is not carrying its part of the load.  Proper tension is a MAINTENANCE function, and should not be neglected."

One question I often get asked is what is the proper tension for the flying wires.  The answer is that the amount of tension doesn't really matter...as long as both of each pair of wires is approximately the same.  Pete's instructions are to tighten both turnbuckles by hand until you can't tighten them any more, then turn them one or two additional turns using a nail for additional leverage.  As long as you use this technique to tighten your turnbuckles, you should be OK.

As a final note, this does point out the brilliance of the Fly Baby design, as far as wing-folding is concerned.  You don't have to touch a single cable's burnbuckle when you fold or unfold the wings.  The sets of wires are grouped and the final tightening is applied by a single turnbuckle that tightens them all.
 

Initial Report:  18 May 2000


On May 13th, a Fly Baby flown by Woody Jones crashed near Taneytown, Maryland.  Mr. Jones was killed.

I received this email from a friend of his:

"I returned from a fatal crash site on saturday evening where a friend of  mine was flying his fly baby.  His wing went up at an extreme angle then folded back against the fuselage.  The plane was in a climb attitude at the time.  After returning home I started investigating safety info. on the fly baby aircraft and was chilled to the spine when I read "The Loss of N96MG" .  I have very little data on my friends planes history.  I plan to report back after more has been gathered.  But,  in the meantime I am asking you to please report that yet another Fly Baby has lost a wing in flight."
Here's the initial FAA report:
B. Reg.No.: 2685        M/M: EXP            Desc: EXP:  1973 BOWERS FLY BABY 1A
Activity: Pleasure   Phase: Unknown   GA-A/C: General Aviation
Descr: EXPERIMENTAL ACFT CRASHED UNDER UNKNOWN CIRCUMSTANCES AND WAS
       DESTROYED, THE PILOT/SOLE OCCUPANT SUFFERED FATAL INJURIES,
       TANEYTOWN, MD.

WX: HGR 132053Z 25006KT 8SM BKN080 29/22 A2965  Damage: Destroyed
C2. Injury Data: # Crew:   1     Fat:   1     Ser:   0     Min:   0    Unk:
                 # Pass:   0     Fat:   0     Ser:   0     Min:   0    Unk:
                 # Grnd:         Fat:   0     Ser:   0     Min:   0    UNK:
D. Location  City: TANEYTOWN                 State: MD
E. Occ Date: 05/13/2000   Time: 21:10
F. Invest Coverage.  IIC:   Reg/DO: EA07  DO CTY: BALTIMORE
DO State: MD  Others: NTSB
G. Flt Handling.  Dep Pt: UNKN  Dep Date:   /  /       Time:
Dest: UNKN Last Radio Cont: UNKN             Flt Plan: UNK
Last Clearance: UNKN            WX Briefing:
Other:


The FAA Aircraft Registration Database shows this airplane was built in 1973.  The manufacturer is listed as "Hing," and my 1997 database shows the aircraft registered to Michael Teichman.  It appears that Mr. Jones bought the aircraft in the last three years.   From comments in rec.aviation.homebuilt, it appears he was very active in the Maryland-area general aviation world.  No details are yet know regarding the construction of the aircraft.

I'll add updates to this page as I get information.

 Ron Wanttaja.