precision

[mstram]

How precisely are planes assembled ?

I'm curious as how to calculate or estimate the effects of imprecise construction / design.

I'm looking for the info for model plane construction, but it could also be used for a homebuilt ... one of these days ;)

Take some extreme examples.

Wingpan 30' . Chord 4'

1) Left wing span is only 29' ..or 28' ...20 .. etc

When does the plane become unflyable ?

2) Back to normal, but left wing chord is now 3.9' 3.5' .. 2' ...

3) Normal but, left wing is further forward/back along the fuselage than the right wing, offset by 1mm, 1cm, 1meter

4) Normal but, left wing has more/less dihedral than the right wing. 1, 5, 10 degrees

5) Normal but left wing has more/less sweep than right


5) Same things for the stab. E.g. left stab span is less than right, ... left stab chord less than right

I'd imagine there could be some heavy math involved, if you know it and feel inclined, lay it on me

Mike

[John Farley]

Mike

I can see why you are concerned with the way a model, homebuilt or regular factory aircraft would fly if it was in any way asymmetrical. But I suggest you need not be.

Such aerodynamic variations as you describe (well most of them)are unlikely to cause a crash because it is almost certain that the controls would be able to exert more influence than any such asymmetry. Bit like running a car with a flat, you might have to use a bit of skill with the steering wheel but one would expect to retain control

No man made object will be perfect, so what we are talking about here are the acceptable dimensional tolerances. All aircraft drawings will include limits on the variation from perfect that are acceptable. It is not sensible for me to quote actual limits here because they will vary so much between aircraft types and will depend on overall size and so on. Also some parameters are much more important than others - for example if one wing is bolted on at an angle of incidence that is say 2 deg different from the other wing then you might find the ailerons could not cope with the subsequent variation in lift between the wings.

Some very badly damaged (asymmetric) aeroplanes have been safely landed following combat damage, bird strike, structural failure and so on. B747s have shed quite a few flap sections on the approach to land over the years – very often the crew did not realise anything was amiss. A bumpy day can inflict very much larger disturbances on the “equilibrium” of an aeroplane than the sort of dimensional variations we are talking about here.

I had a friend who landed a jet fighter after a wing fold mechanism let one wing tip fold up in flight (about 5 feet or more as I recall) It also jammed the ailerons so he had to use the rudder for lateral control. It was a tail wheel aeroplane and he only burst one tyre landing at 220kts on a 2000 yard runway. But he was a good stick. And they did give him a medal.

Why don't you build a model wehere you can actually try out different size wings and so on? I have a 6 ft model where I can put the one piece wings on as much as one inch off centre and it is still controllable.

[Genghis the Engineer]

I'm not sure I entirely agree with you John, although I've not seen much theoretical work to justify my point, a few practical experiences suggest to me that there are variations.

Example #1 was a 2-seat turbo-prop training aeroplane. For reasons that were never satisfactorily explained, about 1 aircraft in 20 would roll inverted if you stalled it with full flaps. We were just starting to get into a deep and meaningful investigation into this when some bright squadron SEngo decided to start swapping rear fuselages around. Within a few weeks, he'd eliminated the problem across the fleet without any significant "highbrow" effort at-all, so we at BDN gave up and got on with something else.

Example #2 is a 2 seat piston-prop trainer / private light aircraft. It has neutral spiral stability, except in the post-build air tests about 1 in 10 has postive spiral stability.

Example #3 is a flexwing microlight, about 1 in 3 on initial post-build air test has a marked tendency to turn which requires adjustment to "trim out".

Example #4 was a high wing 2-seat light aircraft, cleared for deliberate spinning which decided after a service to take 4-6 turns each time to recover from a spin that should have taken ½-1 turn. This was eventually tracked down to tolerances in wing rigging.

Example #5 was a derivative of #2 which turned out to drop a wing about 70° at the stall. This turned out to be down to freeplay (and to an extent manufacturing tolerances) in the linkages between two sides of elevator. Eventually eliminated the problem by introducing a mandatory mod incorporating two penny washers to bring the worst case freeplay down to tolerable limits.

Examples #6, #7 were two different aeroplanes I had to look into that owners complained "floated" forever on landing causing them to use far too much runway. Both, as it turned out, had - for reasons never fully established but somewhere in the construction of the pitot-static systems, moderately (10-15kn) underreading ASIs. So, pilots were approaching and landing 10-15kn too fast, and using too much runway.


Having said that, I think the issue is primarily one of consistency, #1 is the only such case I've seen that was genuinely dangerous. So, the issue is more (in my opinion) of whether every aeroplane in a fleet is the same, and whether the published handling and performance material (POH, AFM, etc.) truly reflect the individual aeroplane. That, I suspect, is not looked into enough - how many manufacturers or authorities really insist on post build test on investigating the differences from "standard" and documenting them for that aircraft?

G

[mstram]

John,

I guess I should re-phrase the question.

I didn't mean that I was worried how planes *are* currently being assembled.

I was thinking more the design stage / experimental (i.e. models).

A perfectly symetrical plane is of course ?? desireable.

" Also some parameters are much more important than others - for example if one wing is bolted on at an angle of incidence that is say 2 deg different from the other wing then you might find the ailerons could not cope with the subsequent variation in lift between the wings. "

That point was made on another forum I asked this on (rcgroups.com). The precision of the wing shape was mentioned as being far more important than say the length of the fuselage.

I *am* curious as the effects of the specific examples I invented.

E.g. With a "shorter" wing on one side, is the added weight of the longer wing going to be more of an effect, or will the added lift of the larger wing-half have more influence?

Will a reduced chord on one side have the same effect as a shorter wing-half ?

I have Flightwise by C. Carpenter, and Aircraft Design by Raymer, but both those books seem to assume you are building a "sensible" plane ;)

Mike

[John Farley]

Mike you said

E.g. With a "shorter" wing on one side, is the added weight of the longer wing going to be more of an effect, or will the added lift of the larger wing-half have more influence?

You should already know the answer to that. If the bigger wings weight had a greater effect than its aerodynamics then what would be the point of wings period?

I told you that I moved my model wing across 1 inch and it had little effect. That was a complete answer to your quote above. Please take time to read and think before writing.

[John Farley]

Sorry Genghis. Not sure I get your point. Mike was talking of very big dimensional differences was he not? I was suggesting crashes would not occur until dimensional differences were huge

I reckon all aircraft off a prod line are different and adjustments are made accordingly during prod flt test. Certainly at Dunsfold we documented the L&D trim adjustments needed to make the aircraft fly straight in conventional flight very pedantically. The trim gauges were then re-zeroed with the trims in the necessary flight position and the whole thing reflown again. In PJB it was harder to get consistency as we were up against engine related variations in swirl and from nozzle trimming to get required RPM/JPT relationships. But all this meant was that we had to set the tolerances wider. The max we accepted was 0.06 lateral g out of trim at mid transition speeds and if it was more than that then really big measures were taken from changing nozzles to changing engines.

Re stalling behaviour (not performance) I happen to know that each HS125 was fitted with individual breaker strips as required following flight checks in order to to get consistent handling across the fleet.

One could go on but I suspect you get my point

[VP959]

I'm inclined to agree with you, John, about production tolerances. I recall picking up a Jetstream after a re-paint, checking the docs, querying the fact that the aileron trim was now supposedly neutral (previously this A/C had always been neutral with two notches left trim, to correct what I assume was a manufacturing tolerance problem) and being reassured that all the rigging checks were OK and the trimmer really should be now in the middle for neutral. Oh what a surprise we had on taking off for the check flight and finding that the aircraft had a fairly significant inbuilt roll, corrected by trimming back to the "normal" setting.........

[Genghis the Engineer]

JF - I think your point was that these variations, once an aircraft design is basically sorted, are rarely catastrophic. My point was that aircraft do vary in small ways, such as to make standardised performance or (especially) handling data incorrect - and thus degrade, but not catastrophically so, safety.

So, I suspect that we were making different points, and aren't actually disagreeing with each other.

In this harsh commercial world we live in, there is much pressure nowadays not to do the sort of fly-fix-fly work on new-build aeroplanes that you did at Dunsfold. One tries to resist it and get aircraft as standardised as possible, but it can be a battle to be allowed to. Company accountants want to keep the testing down to the actual minimum, and at times these days I think they're actually supported by people at the authority, who rarely see a badly sorted aeroplane and assume that they all come out of the box spot-on.

G

[John Farley]

Genghis

I’m very sorry to hear that aircraft today may not be flown until they are right. Sounds like a bad company culture at work. The test pilots carrying out production flights are the final quality control function and not to allow them to continue until they are happy represents a very slippery slope so far as I am concerned. Sure 20 years ago the local prod manager would rant and rave to my face and worse behind my back but he never got his way, because if he did we would all soon be out of a job. Aeroplanes that have a good in-service reputation sell better and crash less often. But you don’t need me to tell you that.

Regards
John

[Genghis the Engineer]

YES

[ICT_SLB]

One other effect is much-improved manufacturing techniques. CADCAM, Numerical Control machining & riveting and laser alignment all contribute to a much more uniform product - sorry, aircraft. Whether that aircraft is good or bad still depends on the original design - CFD doesn't always beat the practised eye of a good Chief Designer.

Similarly active flight controls & sensor systems allow the aircraft's response (including stall) to be tailored to a much more uniform standard. It's common now to do check stalls to 1kt & 0.1 degree AoA on the first production test flight with no stall strip required.

The difference in construction methods & standards could not be better illustrated by the Nimrod saga in today's Telegraph. See

http://www.telegraph.co.uk/news/main.jhtml?xml=/news/2002/11/03/nraf03.xml&sSheet=/news/2002/11/03/ixhome.html

[northwing]

It is rare to find that all the aircraft in a fleet are exactly the same because they tend to get built in quite small batches and there are inevitable changes in personnel and working practices. The dimensional variation is normally quite small but will give rise to individual idiosyncracies in trim etc.

When the Tristars were converted to tankers the UK contractor built an elaborate rig to install the underfuselage hose-drum unit. It failed to work because none of the aircraft were true to drawing, and moreover were all different to one another. At one stage the 80-year old head of the company was laying on his back on the hangar floor trying to kick the thing into position. Apparently Lockheed had been through a phase of hiring labour of dubious immigration status at the time and quality control had suffered.

[ICT_SLB]

Again, Northwing, you're talking about a forty-plus year-old aircraft. Contrast that with the current Canabeer CRJ or Boring 737-NG where 100s of aircraft are built on a true production line. On a recent visit to Renton, we were informed that the standard installation provision for the HUD (either of two flavours) was done in Wichita and then did not need any realignment when the system was finally fitted in Renton i.e. repeatable measurements to optical standards!

[Genghis the Engineer]

It strikes me that those who say there isn't significant difference these days are talking about part 25 or larger military aircraft. Certainly where I earn my living (part 23 or smaller) these differences are as important as John describes (I recall putting a +30°C perf write-down on a twin 2 years ago). I'm guessing because down here in the weeds we can't afford the sort of production / QC technology that has become essential for the likes of Boeing / Airbus / BAE(?) over the last 10 years or so.

G

[Few Cloudy]

Once, an aeroplane which Swissair had leased to another company was wrongly taxied through the de-icing rig at MUC. The result was that the tail got banged on the rig (it was an MD-80) and the stab, screw, whole horizontal stab and the left elevator had to be changed.

The pieces came from Australia, Canada and China. The work force was a Swissair bunch with the McD. rep in attendance. We did the test flight on it and, as always had some delays before the off. The last one was because the right tip was measured to be 1/2" behind and 1/4" above the left one on the rigging check.

A cable went to Long Beach and back came one saying "Well done - that's better than we usually manage!" The thing flew well too.