February 9, 2004

A Discussion of Cracks Below Bonanza and Baron Wing Center Sections

by Dick Wilson

I am an owner of an F33A Bonanza, and a retired loads and strength engineer with 36 years of experience at Douglas.  So I am very much interested in what is being done to our airplanes by the so-called "cracks in the spar web" Airworthiness Directives. The following is a detailed analysis of this problem; directed to the attention of the FAA, Raytheon, and the many owners of the affected aircraft. It shows that the cracks occur not in the spar web but in fuselage former panels below the wing front spar that are not related to wing strength, or fuselage strength, or safety of the airplane structure.  I  explain why the cracks occur, and why patching these cracking areas is futile and probably damaging to vital structural components.

The sketch below summarizes the whole discussion following it: the lower cap of the front spar is a very strong member which carries the tension load applied to it by the wing bolts  which causes it to elongate. The thin panel below it is attached to the cap so it must elongate even more than the lower cap, which causes it to crack. The cracks have no effect on strength of the spar or the fuselage.
 

Fig. 101A in the Bonanza 33 parts catalog illustrates the whole structure discussed below.  This structure is separated into 2 parts and they are analyzed separately as "free bodies":  (1) the load-carrying wind carry-through I-beam, and (2) the fuselage former below the wing.

Wing Carry-Through I Beam

The fuselage is supported (lifted) by 4 vertical forces labeled V in Figure A below. These are shears in the 4 front spar bolts. They are shown as applied to the wing carry-through beam, to lift the fuselage, in Figure A below. Also shown are the tension forces applied by the bolts to the lower spar cap, labeled T, and the compression forces into the upper cap, labeled C.

Question: Where are the 4 vertical loads V reacted by the fuselage? Answer: at 4 vertical rows of rivets above the lower spar cap; a total of approx. 32 rivets. Only one of the 4 rivet rows is illustrated; at the right front in Figure A, showing the web flange and the rivets attaching it to the skin. These reactions are represented by 2 arrows labeled S at the sides, not at the rivets where they occur, to avoid clutter. And finally, the 2 down loads labeled B represent the buttocks of the pilot and co-pilot. This is a summary of all the loads applied externally to this structure, and by the axiom of free body analysis they must total zero vertically and horizontally. This is called ΣV = 0 and ΣH = 0.  It is a Law of Physics, not a theory by Dick Wilson.

Fuselage Former Panel Below the Wing

This panel  is illustrated below, separated from the wing as another free body. It has a flange around the bottom riveted to the fuselage skin. For this analysis a strip of "effective skin" is attached (not shown) as part of the free body:                   

The ADs and the SBs state that cracks in this panel could lead to "total failure."  This must mean that the panel and strip of skin are considered necessary to help the wing  lift the fuselage. That assumption can be tested easily:  The sum of all the vertical loads on this free body must equal zero.  These loads are:

Weight of panel and fuselage skin strip  = W
Aerodynamic lift on skin strip = L
Sum of vertical rivet loads along the top = R

Solve the following equation to obtain the value of  R:

R = L - W =   maybe   +/-  20 pounds. 

At 4.4 Gs the wing is pushing up on the fuselage with a force of at least 8,000 pounds, so if the former panel is cracked all to pieces the aircraft strength might be affected by plus or minus 0.25 %.  At this point my analysis has proven that cracks in the Figure B panel do not matter.  But I will go on.

Refer back to Figure A for a moment; this is a proper I beam with an upper cap, a web, and a lower cap. This I beam carries all the loads, and is unaffected by the Figure B former panel below it. But consider what happens to Figure B: the horizontal strain (by classic beam theory) is proportional to distance from the center (neutral axis) of the I beam spar.  If the lower cap of the spar  is  6 inches below the N.A. then a point 12 inches below the N.A. would be strained twice as much as the lower cap.  But to analyze that thin curve- bottomed panel as an elastic extension of the I-beam lower cap is foolishness. I bring it up only to show that we should not expect it to survive high wing loads without yielding, wrinkling or cracking. And it is obvious that patching the cracks will do nothing but cover them up and prevent further inspection.  Two analogs come to mind: "tilting at windmills and "see no evil."

The clusters of 9 huck bolts at each end in Figure B are critical load-carrying attachments of the lower spar cap components, as shown in the Figure D cross-section further on. The integrity of these 9-huck clusters (4 per airplane) depends on the uniformly snug fit of the hucks in all of the holes. These huck bolts should not be removed and replaced on an assembled airplane.

These hucks must pass thru the former panel to be installed. They therefore accidentally stretch the continuous strip of the former panel above the cutouts, and since this thin strip is a lousy tension member, perforated with 9 stress-raiser holes at each end, and strained even more than the lower spar cap, it tends to develop cracks.

Cracks are also occurring in the web flange around the curve below the 9 huck bolts. The web, being stretched as described above, pulls to the right on the skin and fails the flange in shear along the skin. If the web was cut away, or if the skin rivets were removed in the area, the cracks would probably not occur.

The panel as shown in Figure B has slots for the floor beams with sharp corners at the top, and numerous holes for various things like the nose gear retract rod to pass through. So obviously it is not intended to carry loads. These perforations end at the bottom of the spar behind the panel because the spar cap (under the panel) would interfere with through passage of anything. These cutouts, especially the floor beam slots, demonstrate that the designer did not count on this panel for strength. Ironically, many years later, there are people who think that if it cracks it must be reinforced to make it stronger. This line of thought should be changed to "If it cracks it should be made more flexible."   In Figure C  I have added 6 big holes along the top and 2 triangular cutouts below. These added holes would add the needed flexibility to prevent cracks.  But the 6 top holes would also provide a psychological advantage, that is, the  real load-carrying spar cap inside could then be seen and maybe appreciated.

Figure C 

 _______________________________________________________________

February 16, 2004
 

As previously indicated, the web is a continuous panel over the lower spar cap and it appears to be the spar cap itself, or at least part of it. This illusion is enhanced by that powerful cluster of hucks that seem to help the panel transfer the 3/4 in. wing bolt load into the spar cap. Otherwise why would there be so many hucks?

I believe that Figure D below clearly shows what takes the load.  The 10 huck bolts thru Fitting - 16 (5 aft and 5 forward) shown at the left carry the total bolt load from the fitting to the spar. You can count these 10 attachments by removing the bolt inspection plate and looking inside. The rest of the hucks serve to carry mostly vertical shears at the plane between the spar ant tie plate. Now somebody tell me how could that thin web skin, shown draped over the end of the tie plate, possibly get involved with holding the wing on?  

The following schematic is a cross-section just above the bolt fitting looking down. The various parts are shown with their dash-numbers from Fig. 101A of the parts manual.
 

Figure D

The obvious conclusion from the above sketch is that the web below the spar, cracked or not, cannot possibly have an effect on the strength of the spar cap, or the fitting, or the 10 huck bolts between them. And the "free body" analysis shows that the lower web could be removed completely without impairing the ability of the wing to lift the fuselage.

__________________________________________________

March 16, 2004

April 24, 2004

I recently received copies of the Beech crack repair drawings, 36-4004 for Bonanzas and 58-4008 for Barons.  I was surprised to learn that the added doubler not only covers the web cracks below the spar but it also extends, in one piece, all the way to the upper spar cap.  This gives me a clue into the possible reasoning behind the AD s and the SB s,  maybe the original framers thought that the cracks in the non-structural fairing below the spar might creep up into the very critical wing to fuselage connection between the spar caps. Yet they did not include this critical region in the requirement for dye-check crack inspection!  The mystery continues. In this addition I will further describe the web as separate unrelated functional and non-functional parts. Then I will go into the surprising details of the doubler installation.

First, an overview of the structure as previously discussed. Figure G illustrates the front spar carry-thru bulkhead.  The dark shaded area comprises the upper spar cap, the lower spar cap, and the tie plate. Since these components are very thick, and fastened together with numerous huck bolts, this shaded area should be regarded as a solid unit. (See also Figure D).

          The web portion between the spar caps with the 2 rows of rivets serves as one of four ATTACH ANGLE between the tie plate (shaded area) and fuselage skin, and the 4 angles transfer all of the vertical loads between the wing and fuselage. This ATTACH ANGLE, or web portion, is shown in Sec. A. ( The web to the right of the tie plate rivets has no effect in this load transfer).  Due to the great stiffness of the tie plate, the rivets between it and the ATTACH ANGLE are uniformly loaded in shear, and it follows that the skin rivets, a short distance away, are therefore uniformly loaded.  Likewise the flange corner throughout the length of the ATTACH ANGLE is uniformly loaded.  Shear failures in these areas should not be suspected unless the rivets are removed and then improperly replaced, or if the design load factor is exceeded.

          I have shown in the above paragraph that the web (ATTACH ANGLE) above the spar cap is pulling the fuselage upward.  Figure F shows that the bottom web pulls the fuselage skin inboard and slightly downward. The opposite direction.  The web corner shear changes direction somewhere below the top row of 4 hucks, so a crack cannot be sustained from below to above that point.  

 The Crack Repair Kit

All of the original attachments discussed here were of course installed in the factory when there was reasonable access. Skin rivets were driven with the heads out and bucked ends inside. The hucks that attach the lower bolt fitting to the spar cap were mounted in uniformly reamed holes when there was access for a huck gun to pull them. Now to appreciate some of the great difficulty involved in this job imagine performing all the intricate steps described below, from inside the cabin with one hand down in one of those narrow corners below the seat tracks. Please refer to Figure H. The light grey area is the doubler. It comes as a blank part (no holes) and it has a flange as shown in Section A.  To install the doubler all of the rivets (in this quadrant) shown in Fig. G, which attach the wing to the fuselage are removed. The skin rivets must be drilled out at the bucked end because the heads are inaccessible inside the outer wing, then replaced by blind rivets pulled from inside, in new holes because the old ones are too close to the doubler corner. The control wrinkles in the flange are ground off flush to accommodate the doubler. The ATTACH ANGLE (Fig. G) is largely removed from action and replaced by the doubler (which doubles nothing).  To explain part of what happens in this critical area  I quote Note 1 from the Bonanza drawing:

    "An additional rivet is to be added on each side of an existing rivet hole in the carry thru web flange if the existing hole will fall into radius of repair doubler. Locate added rivets half way between existing rivet holes with 2D edge distance. If an added rivet on one side of an existing rivet hole falls in the area where the control wrinkle is cut away, bond an .040 2024-T3 aluminum shim cut to shape into cutout before installing repair doubler and rivets.  If 2 or more added rivets fall in the areas where the control wrinkles are cut away, remove the carry thru web flange between the outermost added rivet locations and bond a single .040 by .62 wide 2024-T3 alum shim into the total cut away area before installing repair doubler and rivets. Fill open hole in skin spar web flange with EC2216 and smooth with al-oxide paper. Do not damage skin. Bond shims in place with EA 9309 NA adhesive. Adhesive is to fill all gaps between shims, webs, and skin to prevent accumulation of moisture. Before bonding shims, ensure there are no sharp edges where cut away, then use corrosion proofing on edges. Treat the shims with Alodine 1200 prior to instl.

Engineering data in this kit is FAA approved.
This kit is an FAA Minor Change."       
(End of Note 1.) 

             All of the huck bolts, that hold the lower wing bolt fitting spar cap (via the tie plate) are ground off and driven out.  The following is a quote from part of the intricate installation instructions for this operation. I've been told that a prominent FBO cut an emergency access hole in the fuselage skin to do this job, then attached an external patch over the hole.

           "Remove all doublers, brackets or clips that would cause interference with the installation of the repair doubler. (Retain for reinstallation). Remove all huck bolts inside the 2.0" min. area except for one of the most outboard huck bolts.  Leave this huck bolt installed but grind it off flush against the webs being careful not to mark the web. A piece of thin steel sheet can be used to prevent the grinder from contacting the web. Locate a .128/.133 diameter hole in the repair doubler at the location of and centered on the ground off huck bolt before the repair doubler is bonded in place. Through the .128/.133 diameter hole, push the ground off huckbolt out of the carry thru after the repair doubler is bonded and bolted in place. Insert the shaft and install the cutter through the hole in the bottom of the wing attach fitting. Back drill and ream this last huck bolt hole per procedure given on the applicable kit drawing."

           It is ironic that the concern for, and misunderstanding of the harmless cracks has caused such a drastic intrusion into two un-cracked areas that (1) hold the wing spar cap together and (2) tie the wing to the fuselage.  Instead of being a reinforcement the patch almost certainly degrades these critical parts. Under normal circumstances most of the above procedures would never pass the quality standards of Beech and the FAA. For instance, in note 1 above " . . . remove the carry thru web flange between the outermost added rivet locations and bond a . . . shim into the total cut away area.

          Three IA's have told me that the dye inspections cannot be performed, due to poor access.  Another said about the same, then added “The inspection used to take me a long time, but now I do it in a few minutes.”

           A January ABS magazine article (pg 8214) stated "Many shops that have done spar repair kit installations in the past don=t want to do them anymore."

August 6, 2004

The Crack Repair Kit, continued

        Figure H Section A shows the configuration of the doubler addition in the critical wing-to-fuselage connection between the spar caps. Fig. I below shows this cross-section in detail.  Alongside Fig. H, Note 1 is copied from the Beech Kit Drawing 36-4004. Throughout the Bonanza and Baron Kit Drawings there are numerous notes that seem to have been added as troubles developed on the drawing board or in the field with this hazardous procedure. I believe Note 1 is the most worrisome of these. See Figure I in relation to the following provisions of Note 1, keeping in mind that these are blind cherry-max rivets installed in an area of very tight access, and with no hope of inspecting the back-side.

(1)  "An additional rivet is to be added - - if the existing hole will fall into the radius - -".  Translation: the original rivet holes are probably not usable. They are too close to the doubler radius, and anyway they were drilled out from the bucked ends of the original rivets, probably resulting in figure 8 holes in the fuselage skin inside the wing. And scratch the term "additional rivet".  This means "replacement rivet".

(2) "If an added rivet - - - falls in the area where the control wrinkle is cut away, - - bond a shim - -" Figure I illustrates this situation. The original angle (web flange) that attached the wing to the fuselage is cut away, the original rivets are removed, both from the wing (tie plate) and from the fuselage (skin) and replaced with cherry rivets that are installed with great difficulty and the blind ends cannot be inspected.

        One of the numerous problems with this is that rivet single-shear strength values are based on tests with panels attached together, rather than with a spacer (shim) between them.

        The Beech "doubler" is really a "singler".  It replaces highly critical structural components with a patch that could never meet normal standards of aircraft construction. This "repair" is bonded in place, all gaps and discrepancies are to be filled with adhesive and the whole critical area is covered and cannot be inspected.

        Many mechanics have installed the doublers, and they are experts on the subject. The FAA should talk to them and learn all the details. George Braly has done this, and he recently posted the following on the Beech-Owners site: 

Folks,

Hard prediction here.
       We have talked with several folks who have done a number of these repairs. Based on that information and based on a detailed look at the structure and the minimal access to the structure ... it is virtually certain that a fleet wide performance of these repairs as presently described by Beech will result in a significant number of airplanes with "occult" damage to the wing spar structure FROM THE REPAIR - - damage that is truly harmful to the long term integrity of the structure and which will eventually result in a fatigue crack that will cause the loss of an airframe.
This is a potential disaster for the fleet.
Every airplane that is repaired . . . will almost immediately become "suspect”.

Regards, George
 

October 16, 2004

In the letter from the FAA reproduced in the March 17, 2004 addition herein is the statement "The FAA considers the web(s) as primary structure."  That makes me want to ask "which part?"   Here I will discuss the functional differences between the several areas of the web and show that some parts should not be called "primary".

The web panels, 0.040 and 0.050 inches thick (depending on serial no.) cover the much heavier wing carry-thru spar structure, fore and aft sides, and these thin panels extend down to the bottom skin. Here is a photo of an entire aft-web (click on it). Also turn back to Figure E where I have divided this total web into distinct functional parts.

In the photo, two horizontal rows of pre-drilled rivet holes can be seen. After the webs are installed 6 huck bolt holes at each end will extend these rivet rows in a continuous line across the fuselage. I will refer to this long straight row as the LSR. There will be 9 additional holes for huck bolts below the LSR as shown in the lower left in Figure E, but pretend they are not there for now. These 9 bolts hold the real heavy duty spar parts together underneath the web. For this discussion of the web, pretend that they have flush heads in the 0.17 in. thick tie plate under the web and are out of sight.

During assembly of the airframe the whole aft-web (likewise the front one) was required to define a precise shape and location of the local fuselage in relation to the rest of the airframe. It is hard to imagine the preformed skin panels being brought together to form the fuselage without a framework in place first. But in flight the various parts of the web have distinct characteristics, and it is inaccurate to lump them all together as one primary structural component. These various parts of the web are labeled in Figure E, and their boundaries are defined by rows of rivets. (Remember, the lower 9 hucks do not go thru the web for this discussion).

Area X is in the proper position as spar I-beam web, fastened with a straight row of rivets to the straight lower spar cap underneath it. Note that it can’t do much shear-carrying with that very large cutout for the landing gear retraction parts, and there is no shear in area X during a symmetrical pull-out; but call it the spar I-beam shear web anyway, even primary structure if you wish.

Area Z

Area Z of the web shown bounded by hucks and rivets top bottom and sides (Figure E) serves only as a dust cover over the Tie-plate (click) beneath it. Note the vertical joggles in the aft-web photo to accommodate the much thicker tie plate underneath which is real structure. . Also see Figure D for these details. Obviously the designers did not extend the 0.050" web over area Z to reinforce the 0.17" tie plate. Click on the photo back-side for a reassuring view of the tie plates and spars attached behind the forward web which comprise the real load-carrying structure.. Incidentally, the odd shaped holes in the tie plates are there, I believe, to provide formed vertical stiffeners (see the formed flanges) and also to reduce weight. Everywhere I look in all of this structure I am much impressed by the skill of the designers.

The Wing To Fuselage Attachment Area

In Fig. E the web portion between area Z and the fuselage skin, is what I call the ATTACH ANGLE. It is bounded by the row of tie plate rivets and the row of skin rivets (not shown). There are 4 of these ATTACH ANGLES (left and right, fore and aft) that attach the fuselage to the wing. They are the only attachments of the fuselage to the wing. I am capitalizing ATTACH ANGLES for emphasis to distinguish the only parts of the web that are real load carrying primary structure.

Area Y

Area Y is the web portion below the long straight row of rivets ( LSR) shown in Figure Y or in photo Aft-web. Please continue to consider the lower 9 huck bolt holes missing. This canoe- shaped figure is attached to the lower spar cap by the LSR and by its flange to skin rivets.

In the discussions of Figure A and Figure B of this report I proved that Figure A lifts the fuselage and holds the wings on and that it is impossible for Figure B to contribute to either of these tasks. I used the free-body theorem, which I believe to be the foundation of all structural loads analysis. I think this started with a theorem by Archimedes in about 210 BC: "For every action there must be an equal and opposite reaction". Someone back in the colonial days stated a corollary of that: "Thou canst not clap with only one of thy hands". For present purposes, I used "Nothing is pushing up on figure B, therefore the fuselage cannot push down on it."   But this classical time-honored reasoning has been rejected by the FAA as "simplifying assumptions and generalized discussion."   I will keep trying.

The Web Flange - Top to Bottom

It appears that Beech and the FAA have made no distinction between the web flange below the LSR and the portion above the LSR. In the following I will explain the vast difference between the skin attachment above the LSR and that below.

The tie-plate should be regarded as indefinitely stiff in compression or tension, because of its thickness. The 7 or 8 rivets connecting the attach angle to the tie plate all push uniformly upward on the skin rivets, one to one less than 1" away, transferring equal shear force per rivet to the fuselage skin. Without this uniform distribution, afforded by the rigidity of the tie plate, an analyst could not multiply the handbook strength of one rivet by the number of rivets and compare the result with the weight of the fuselage x G to obtain a margin of safety.

Below the stiff tie plate and the LSR the connection of the web flange to the skin looks the same as that above but it is entirely different for the following reasons:

(1) It is a connection between a thin curved skin and a thin un-stiffened panel

(2) The web here is severely strained horizontally by the lower spar cap, resulting in unpredictable shear and tension stresses in the rivets and web flange.

(3) The skin connections above the LSR provide the entire lifting force to the fuselage. The skin connections below cannot and do not contribute to lifting the fuselage.

Now I will put the 9 huck bolts below the LSR back into the discussion. Cracks have occurred around the 9 hucks as well as in the web flange at the skin below. If these bolts did not protrude thru panel Y, obviously cracks would not have occurred there. Furthermore, I believe that the panel might have been sufficiently flexible in shear to accommodate the change of shape caused by the spar cap extension without cracking the flange at the bottom anywhere. I discussed the severe stress concentration in the web at the 9 hucks in connection with Figure F.

To summarize the discussion above, the total web serves as a forming bulkhead during assembly. During flight with ailerons deflected, area X may serve as a spar shear web. Area Z has no load function. The area I referred to as the ATTACH ANGLE holds the airplane together and must not be tampered with. Area Y is just part of the bulkhead former and it is not primary structure.

The Crack Repair Doubler Revisited

An installed Beech doubler on the right side of of an aft web is shown in two photos: Doubler and Doubler-looking-down . This doubler was later removed an sent to me. It was originally installed by a Beech shop, when the airplane was all together and the wings were on. This Baron is being restored following an accident, and fortunately presents a rare opportunity to see both sides of all the parts. These photos show how thoroughly all the areas discussed are hidden from future inspectors. Ironically the new Beech SB 53-2360 that requires a permanent cover to be bonded and bolted over the whole area also requires for inspection for cracks at 1500 hrs. after the installation, and every 500 hrs. afterward. . Incidentally, click on Outside which is a photo of the outside skin, normally inside of the wing. In conjunction with photo Doubler it can be seen that 3 original rivets in the lower portion of the ATTACH ANGLE were removed and not replaced. One hole was drilled but left empty (see Doubler). I believe a cherry rivet could not be pulled here because the nearby bolts/nuts were in the way. (These threaded bolts replaced hucks).

I previously discussed the strength problem created by extending the doubler up into the critical ATTACH ANGLE area.. Now I have additional information. Refer to the kit instruction Note 1 (in italics) beside Figure H and also click on Wrinkles , a photo showing the "control wrinkles" that must be ground away to accommodate the doubler . Little finger-nail-shaped shims are fitted into the ground away voids and the new blind rivets are installed thru these shims or maybe thru their edges. (Remember, with the doubler covering everything inside, the mechanic cannot see the old holes or the shims, where to drill the new holes). None of the original rivet holes are usable because they are too close to the doubler bend radius. The new blind cherry rivets are fewer in number than the original bucked AD rivets, they cannot be inspected for integrity inside the the wing, and worst of all the flange and the skin are not together, defying the rules for rivets in single shear.

This morning I talked to an A&P/AI who has installed about 6 of the kits. He said that the doubler has a somewhat different shape than the ATTACH ANGLE flange (with wrinkles ground off) and after the doubler is attached to the web face with 15 bolts there are gaps under the flange in places as wide as about 0.08 inches. He said that this is primarily a problem on an aft installation. I have found from the drawings that the left rear doubler blank is the same as the right front and right rear is the same as the left front. The fuselage is tapered slightly between front and rear, so I suspect that the gaps were caused by making the front and rear blanks the same shape.

The instructions call for filling all discrepancies at the flange with 0.040 shims and adhesive after grinding the control wrinkles off. A note on the Baron drawing permits use of adhesive alone with no shims. The drawing notes emphasize prevention of corrosion by filling gaps with adhesive. Probably the adhesive is allowed to harden before the blind rivets are installed in the web flange/skin holes. The result at the worst places could therefore be a sandwich consisting of: (1) the 0.065" doubler, (2) an 0.080" misfit gap filled with adhesive, (3) either a little finger-nail-shaped piece of 0.040" aluminum or an 0.040" layer of adhesive, and finally the 0.040" fuselage skin. The fuselage weight x G is supposed to be transferred via cherry rivets cantilevered across this sandwich with a plastic filling.

The doubler should be considered as 3 separate (dysfunctional) areas, starting at the bottom:

(1) The replacement web flange below the LSR that covers (obscures) previous harmless cracks

(2) The portion over the hucks that covers (obscures) harmless cracks in that area

(3) A haphazard replacement for the wing-to-fuselage attachment.

If I were not the highly diplomatic person that I am, I would say that the Beech doubler kit was a cynical response to a blanket order from the FAA, categorically requiring cracks to be repaired.

Recommendations

This monologue is not a good place for me to discuss my recommendations, it would be far better to have a discussion with the rule makers, which probably won’t happen. But I think it is obvious from all of the above that the requirement for "repair" of the cracks be canceled.  For aircraft with existing doublers installed I suggest a periodic inspection of the flange above the LSR, and also a borescope inspection of rivets and skin above the LSR thru the openings either side of the lower wing bolt fittings.

All photos were furnished by David Peterson

November 12, 2004

Thrust and Braking

So far, in this analysis, I have dealt only with loads in the plane of the front spar, in particular the vertical load applied by the wing to the fuselage in a symmetrical 4.4 G pullout.  I believe that Beech and the FAA have been concerned over the strength of the wing to fuselage connection for that loading condition.

The other day George wrote in an e-mail that he was tooling up to show that "other loading mechanism" may be causing the cracks, and I believe he was referring to fore and aft loads applied to the fuselage by the wing.  So in this addition I will address Thrust and Braking. I believe that this will complete all of the possible loading conditions that have alarmed Beech, the FAA, and George Braly.

Figure J below is a diagram showing a tire, a propeller, and the front and rear spars with their numerous rivet attachments to the skin. I am dealing with only the right wing and the right side of the fuselage. 

The propeller thrust applies, of course, only to the Baron, and the brake force applies to either the Baron or the Bonanza if the brakes are on. Please visualize that the Baron prop and the taxiway (or runway) apply forward and aft forces (respectively) to the wing, which transfers these forces to the 4 wing bolts shown in the background. I haven=t yet gotten to the fuselage but I will.

 I have decided that the sum of the 4 fore-aft bolt loads are maximum in these conditions::

 Baron - Maximum static thrust at the beginning of takeoff roll, no brakes please.

 Bonanza - Brakes locked, maximum static thrust at fuselage centerline, resisted by an equal and opposite braking force at the pavement, one half per each main tire.

 Start with the baron; someone might imagine that the top bolts carry all of the fore-aft load because of the high position of the thrust. Time now for a brief lesson in vector analysis. That thrust vector should be moved down to the center of the 4 bolt pattern, that is, halfway between the top and bottom bolts, if you add a clockwise moment equal to the thrust x distance moved. Now where does that moment go? Answer: to an up load on the rear spar equal to the moment divided by the distance between the spars, and an equal down load on the front spar. Likewise the brake load at the pavement causes an up load component on the rear spar, an equal down load component on the front spar, and a fore-aft force equal to the pavement load applied at the center of the 4 bolt pattern.

 Now before George says that choosing the center of the bolt pattern for the fore-aft force reaction is just my arbitrary opinion, I will state the reason. My best estimate is that the fore-aft line of fuselage stiffness is at that vertical location. If you say that the thrust goes to the top bolts and the drag goes to the bottom bolts you are applying what I call either the "hydraulic" theory or the "termite" analogy; or maybe both, that loads flow like water or creep like termites.

  So a fore-aft force is shared equally by the top and bottom bolts, and any torque or twist is resisted by a vertical couple at the spars. A brake load twists the whole airplane so that the nose gear load is increased and the main gear load is decreased. Can you imagine that twist coming from an aft load on the lower front bolt and a forward load on the upper front bolt?

 Now I am ready to make an estimate of the aft brake load on the Bonanza. That's easy. Michael Derby told me that a 550 will produce 1734 lbs. neglecting prop losses. So take it down to about 1600 lbs. real life and apply 800 lbs. per side about midway between the upper and lower bolts on the Bonanza.

 As stated above, the Baron load condition is beginning of takeoff roll. The right engine thrust is reacted by half of the acceleration force of the airplane (F = M*A). The right wing bolts accelerate half of the fuselage weight. I will make a wild guess that a loaded Baron fuselage weighs 60 percent of the GW.  Therefore the right engine applies a forward force to the 4 bolts of 60 % of its thrust to accelerate half of the loaded fuselage. For a 550 that comes to 60 % of 1600 lbs. = 960 lbs. applied about midway between the upper and lower bolts on the Baron. Regarding the distribution between the front spar bolts and the rear spar bolts I would say they may be about equal. In the case of the brake load, the main gear trunnion is bolted snugly to both spars. A static test setup, in the absence of a wing, would require deciding how to apportion the load between the bolts. Please look at the photo link Doubler-looking-down discussed in the October 16, 2004 addition; this shows a keel member or stringer down near the lower bolt.  It might be tempting for a test planner to say "You, brake load, go there!"                           

 Now I will get inside the fuselage. See section A-A in Figure J, at the front spar. The upper and lower bolt loads are transferred to the thick tie plates, which are clamped together top and bottom at the bolt fittings. This forms a rigid rectangular assembly, almost but not quite like a solid block of aluminum. This block receives the thrust load or the brake load about halfway between the upper and lower bolts. This load is then transferred by the attach angles to the fuselage. Each tie plate rivet transfers a small load to its rivet partner at the skin by a complex pattern of sheet metal bending stresses that could not be measured with strain gages even if you thought you had a reason to do so.

 I stated that the max. fore-aft load on the Bonanza is 800 lbs. and on the Baron it is 960 lbs. To give you a hint of what=s coming, one AD-5 rivet will carry 574 lbs. in single shear and there are about 28 rivets between the attach angles, both spars, and the fuselage skin. That would be, for the Baron an average of 960/28 = 34 lbs. per rivet.. No one working on this crack project will know how the load is apportioned among those 28 or so attachments. So I will make a very wild guess and say the rivet loads vary somewhere between zero and twice the average, or 78 lbs. maximum. That results in a factor of safety of 574/78 = 7.4.  I have no idea what the bending stresses are in the corner of the attach angle, but I'll bet they are too small to crack the skin.

 George has a project going to develop and market an AMC. He and I have talked at length about his test on the phone, where to put strain gages, etc. I suggested that he apply a number of gages to the lower corner of the former panel, load the bolts and take readings. Then remove certain rivets and maybe make certain cutouts in the panel, as I recommended at Figure C. Then load the bolts see if the readings are reduced. That might prove my "beneficial relaxation" theory and it might enable him to get an AMC.  He may be planning to do that, but his recent enigmatic statement about "other loading mechanisms" has made me worry.  I very much want his project to succeed.

 According to the e-mails, George is staging quite a dog and pony show.  Steve Oxman has suggested a web cam.  Maybe there should be a blond in a leather mini-skirt with a laser pointer

December 7, 2004

Response to Dwerlkotte and Harradine Reports

On about Feb. 20, 2004 the ABS asked a "highly credentialed aerospace engineer" to review my report and discuss it with me. The discussion did not take place. Shortly afterward that engineer was joined by another "highly credentialed aerospace engineer" and they were contracted by the ABS to "develop a definitive (final) understanding of what causes spar web cracking - - "   (See Update 8.25.04).

About 9 months later they completed and published 2 Discussions. The first engineer turned out to be Mr. Harradine. His analytical background emphasized large pressurized fuselage structures. The second, Mr. Dwerlkotte, was a Senior Design Engineer for Beech from '53-'65, he is capable of performing finite element analyses, and he is CEO of Dwerlkotte Associates which consults in obtaining FAA approvals. He is an FAA Designated Engineering Representative.

The ABS has just issued a Request for Proposal to companies capable of performing a definitive finite element analysis, DER rating required. It seems clear to me that Dwerlkotte Associates is ideally suited and positioned to perform this task. Mr. Dwerlkotte has the added advantage of being able to blend the Beech point of view into the conclusions of the study.

The D/R Discussions both contain a brief stress analysis, which I believe Mr. Harradine wrote, of the area around the skin cutout at the lower wing bolt.  To summarize this analysis:

The ultimate strength of the aluminum skin is 60 ksi (2400 lbs. per in.).  If there is a longitudinal stress in the skin of 10 ksi (400 lbs. per in.) then the bending stress in the flange at the huck will be about 16 ksi.  That therefore gives a margin of safety of (60/16-1)*100 = 275%, but this is not sufficient to prevent early cracking if this loading is frequently repeated.   

The perimeter of the bottom skin is about 55 inches from mid spar to mid spar.  Mr. Harradine's wild guess of 400 lbs. per in. is the equivalent of about 22,000 lbs. tension in the bottom skin. He doesn't derive this load or say where it came from, but he lists the probable loading conditions as being nose gear loads and wheels up landings.   

Unlike "large pressurized fuselage structures" the Bonanza/Baron skins generally do not carry tension stresses. Some of the fuselage skin panels carry shear, but most are there simply to form the shape of the fuselage, to provide a smooth aerodynamic surface, and to provide a firm support for a cool paint job. Certainly the panel analyzed in the reports has no tension in Bonanzas; it ends at the opening provided for the exhaust stack, instead of marching on forward to somehow support nose gear loads or gear-up landing crunches on the forward fuselage.   Likewise, the "J-stringer" has no tension load in it.  It serves 2 purposes: to stiffen the skin and to support the floor boards. 

I therefore would like to correct the Harradine analysis to the following:

The ultimate strength of the aluminum skin is 60 ksi (2400 lbs. per in.).  If there is a longitudinal stress in the skin of 0 ksi (0 lbs. per in.) then the bending stress in the flange at the huck will be about 0 ksi.  0 ksi is insufficient to cause early cracking, even if frequently repeated.

The Attempted Takeover By the ABS Magazine Board

In February of this year I began my one-man project of informing the FAA of the AD problems, which were adversely affecting many thousands of airplanes. I established this contact with the FAA with a new web site created for this purpose. Shortly afterward Tom Turner of the ABS sent me an e-mail asking if I had a Structural Engineer's License.  I replied that I did not, and explained to him what an SEL is, namely a license to design buildings issued by state building regulators. At that point the ABS went underground with a 9 month "independent" study. It now appears, based on 9 months of innuendo and on the now revealed D&H Discussions, that the ABS study has been based on the ground rule of contradicting and supplanting my report to the FAA. Furthermore it is apparent that the future of the airplanes has not been given top priority by the ABS and that they have no regard for the success of my project to try to save the airplanes.

Hardware and Engineering Principles Denied

In my analysis I took advantage of a fortunate feature of the front spar carry-thru-I-beam-bulkhead-frame- former assembly (how's that for multi-culturalism?), namely that it could be neatly separated into two finite elements: Figure A (ibid) which carries the loads and Figure B which does not.  Then in Figure E, I separated the bulkhead panel into 4 finite elements, X, Y, Z, and the ATTACH ANGLE.  That's two finite element analyses (FEAs). I showed the great importance of the ATTACH ANGLE, I explained that the thin finite element Z is useless because under it lays the 4 times thicker TIE PLATE.  Turner, Dwerlkotte, Harradine and Braley have systematically refused to recognize that the TIE PLATE exists, and they eschew the separation of the bulkhead into 4 finite elements.  Yet they are planning to dissect the whole damn fuselage and part of the wing into maybe 10,000 little cubes! Tom Turner, you can learn a lot more about this project you are planning with a little homework exercise. Go to:

      http://dattaraj_rao.tripod.com/FEM3D/index.html

And perform a small FEA.  You can increase the number of nodes as you wish, drag the nodes around to model, say, a small strip of fuselage skin attached with one huck (keep it simple). Just guess at a load as Harradine did. Keep repeating the load until the strip shows signs of fatiguing. Then stop and perform a definitive evaluation

December 10, 2004

Earl M. Conditt Jr.
Associate ACO Manager
Wichita Aircraft Certification Office

Dear Mr. Conditt

In the October 16, 2004 addition to this report I identified 4 “ATTACH ANGLES” which are the sole structural connections between the wing and the fuselage.  One is illustrated in Figure G of the May 8 addition.

The October 16, 2004 section titled “The Crack Repair Doubler Revisited” discusses the damage done to these vital connections by the doubler installation.  From scant data available I estimate that doublers are being installed in about 17 aircraft per month.

To stop this damage from continuing, I strongly recommend that you request RAC to add the following note to the Service Bulletins:

IF NO CRACKS HAVE OCCURRED ABOVE THE TOP ROW OF HUCK BOLTS, REMOVE THAT PORTION OF THE DOUBLER ¼ INCH ABOVE THOSE BOLTS PRIOR TO INSTALLATION

Sincerely,

 Dick Wilson              

December 29, 2004

     The main theme of my web site analysis has been to prove that the lower former panel does not help to hold the wing together, nor does it help to attach the wing to the fuselage. Therefore I believe we should conclude that it is not a primary structural component.

     Mr. Conditt stated the following in his Mar. 17 letter to me:  "- - for most circumstances - - - cracks in primary structural components should not be allowed."  That was encouraging to me, because now the basic problem seems to be narrowed down to two terms; "most circumstances" and "primary structure".

I will keep trying. 

January 11, 2005
 

Earl M. Conditt Jr.
Associate ACO Manager
Wichita Aircraft Certification Office

Dear Mr. Conditt:

In my letter of December 10 I recommended that the Beech doublers be cut off just above the huck bolt cluster, prior to installation, as an emergency first step toward preventing continued damage to the wing/fuselage attachments. In this letter I will describe in greater detail the urgency of this recommendation.

The October 16, 2004 paragraph titled "The Wing to Fuselage Attachment Area", I referred to the critical structural connection of the fuselage skin to the flange of the web (see Fig. E, section A-A). In the original design the skin and the flange were together as in the left of the 2 figures to the right. This is called a "single shear" riveted connection, and to obtain a TC Beech should have proven, via analysis and tests, that these attachments conformed to standard handbook strength values for rivets in single shear. Likewise, when Beech sought approval for the doubler their DER must have implied that the mod was as strong or stronger than the original structure. But as I have explained earlier the original flange and rivets are out of action, and the result of the doubler installation is a transformation from the left figure to the right one. To repeat from the October 16, 2004 addition:

The instructions call for filling all discrepancies at the flange with 0.040 shims and adhesive after grinding the control wrinkles off. A note on the Baron drawing permits use of adhesive alone with no shims. The drawing notes emphasize prevention of corrosion by filling gaps with adhesive.. The result at the worst places could be a sandwich consisting of: (1) the 0.065" doubler, (2) an 0.080" or more misfit gap filled with adhesive, (3) either a little finger-nail-shaped piece of 0.040" aluminum or an 0.040" layer of adhesive, and finally the 0.040" fuselage skin. The fuselage weight x G is supposed to be transferred via cherry rivets cantilevered across this sandwich with a plastic filling.

In case there is any doubt about a big loss of strength between the left figure and the right one, I have analyzed the comparison. The triangular areas represent bearing stress patterns. In each case the shear load P equals the area of the triangle. If we define failure of the joint in both cases as occurring at equal angles of rivet rotation (with attendant distortion of the rivet ends), then the relation between the allowable loads P1 and P2 can be established by equating the twisting moments on the rivets. Assuming 0.04 thickness of the skin and original flange, 0.050 thickness of the doubler flange and a gap thickness of G, and equating the twisting moments:

        0.040* 0.66* P1 = (0.33* 0.04 + G + 0.33* 0.05) * P2

Solving this for the ratio of P2/ P1 for various values of gap width G yields:

                    G                  P2/ P!
                    0.00"             1.00
                    0.04"             0.38
                    0.08"             0.24

The minimum gap of the doubler installation is 0.04", the thickness of what remains of the original flange after grinding off the control wrinkles. Therefore the shear strength of about 6 skin attachments above the lower huck bolts is reduced by at least 62% of the original values.

Sincerely,
Dick Wilson

January 17, 2005

February 5, 2005

 "Thou Must Journey Forth to Ada, That Thou Believeth"

On January 20 the Witness posted to Beech-Owners that he just got back from the Church, where in a 10 min. demonstration George brought forth unto him the Truth with a simple business card. I beseeched the Witness for the Word via e-mail and he finally relented. I believe that all Beech owners should perform the card trick at home, to save George from more pesky visitations in Ada. Please click on WITNESS for details.

May 5, 2006

I called George Braly on July 15 2004 to propose that he perform static tests before and after making cutouts in the former (see Fig. C early in my web site) and to compare strain gage readings.  I suggested that these tests might convince the FAA to grant him an AMOC for cutouts and/or stop drilling of either cracked or uncracked lower panels, thus saving the majority of the airplanes from the disastrous Beech doublers.  We talked for 58 minutes.  He declined, believing that the J-stringer at the cutout in the fuselage skin at the lower wing bolt (to see the area, click on CUTOUT ) was responsible for the cracks.  Later I learned that he was placing strain gages in the area on a Baron fuselage (salvaged from a wreck) and would soon "pull, push, bend and twist" the fuselage to demonstrate to FAA personnel that a patch around the cutout would prevent the cracking".

George's project which he began 2 years ago has been held up.  Several owners in late 2004 asked him "what is the delay?"  He replied with various reasons, among them were: his high-dollar Computer Aided Design instruments were tied up on another project, he had yet to formalize a game plan with the FAA, and one of his workers had been in an auto accident. (click on NO-MAJOR-DELAY) John Deakin asked those complainers to "Give him a break, willya? This man is SERIOUSLY overloaded!" and "He WILL come up with an elegant fix, at reasonable cost. - - - I'd stake my life on it." (Click DEAKIN)