Uber Ultra-light tear

[kennyrayandersen]

trackstripper,

The aluminum tongue will be mechanically fastened to the bottom of the trailer cabin at at least two locations. The first is at the intersection of the forward wall at the start of the overhang. At this location the trailer is pushing down onto the tongue. I'll use something like a U-shaped strap and two fasteners that go through the floor (at this location there is a 2x2 pine/fir board buried in the composite panel) and is countersunk so it will be flush with the floor. Standard head hex bolts with washers installed thread down.

The other attachment (there will be at least one other attach point but there may be more. This will be based on the finite element model results) will be at the end of the aluminum beam that is furthest away from the hitch. This one will be in tension. The overhang is 30 inches and this reaction point is 42 inches away from the front of the cabin -- overall beam length is 72 inches (preliminary -- this could go up or down), which means that the load to balance the 300 lb maximum hitch load is:

Reaction aft point = 300*30/42 = 214 lb
This is the tension load on the aft fastener

Reaction at the front of the cabin = 300 + 214 = 514 Lb
This is the load at the front of the cabin, but it is in compression and doesn't put these attaching fasteners in tension.

That's not much load in the aft bolt and this could easily be reacted by a single 1/4 inch bolt. At the point this through bolt attaches, I will likely have another 2x2 buried in the composite floor, but really a load this low could easily be reacted by the panel itself with just a small plywood insert, a couple inches in diameter, buried in the floor panel (and it may some to that!).

I'll try modeling a couple of different schemes and based on what shows the least stress/best stiffness I'll make a final choice. What I try and do is start out simple, and then add a little something if it's necessary.

I'll probably also put just a couple of angles on either side of the aluminum beam where it meets the front of the cabin to support side loads.

[Trackstriper]

What would the effect be to go with a round section tube? Here's what I'm thinking. Modern hang gliders often use 7075 alloy for the frame. Leading edge spars are put under a bending stress perhaps similar to a trailer tongue. I happen to have a few old (flown but not broken, from a dismantled glider) spars at home...where I'm not at the moment. I may get back by the end of the weekend. I'd have to measure what I have to see if there is enough length, which I think there is, and diameter and wall thickness....but think perhaps 2.50 diameter by .049 wall, perhaps metric dimensions, some of the good stuff was imported from Europe. Typically the tubing is anodized. Might this work for you? Price is right....shipping.

Bruce

Second thoughts. This stuff is fairly "springy", might you run a narrow angle A frame to all but eliminate the horizontal effect and halve the vertical?

[angib]

Kenny, assorted thoughts:

- 3g bump on both trailer wheels and the tow vehicle simultaneously is pretty severe. But it's interesting that it comes so close to the Aussie rules. The 20% difference between the two (using your assumptions) I don't count - getting in the same order of magnitude is the important thing!

- When working with steel tubes of these sizes, you will never achieve yield as the tube will buckle before that - but the difference shouldn't be great, maybe 10% or so.

- When working with aluminium, using yield or proof stress isn't satisfactory unless you've investigated the fatigue loading as well. Unlike steel, aluminium has no fatigue limit, so every tiny stress cycle contributes to fatigue failure. For that reason, using ultimate strength for something that is continuously loaded isn't adequate. Of course a fatigue analysis requires more data than we've got, but even if you just apply a 'factor of safety' (out of the history books, that) to it, it shouldn't be ignored.

- You suggest that the deflections are acceptable and I think vertical ones generally are. Transverse deflections tell us the stiffness and here the evidence from this forum is that you can build a single tongue with insufficient transverse stiffness, so that it shimmies at highway speeds. What the acceptable stiffness is, I've got no idea, but it ain't zero!

Andrew

[kennyrayandersen]

Bruce, Andrew
Andrew points out an interesting conditional difference between aluminum and steel, and that’s fatigue. As I address your question I’ll naturally drift into a response to Andrews observations, which BTW, I always appreciate (nothing like a second set of eyes looking at a problem). I didn’t get into the fatigue aspect because the rest of the number crunching seemed to be getting plenty thick already. I’ll elaborate on that, but let me address your problem first.

The actual moment of inertia for the tube size you are talking about might hit the theoretical stress numbers for bending and there is no problem with a round section per se (though you might have trouble integrating it into the 2 inch coupler); but, thin sections are worse in fatigue AND additionally there is a bearing load from the coupler, which ain’t huge, but holes and small radii, and abrupt changes in load paths are generally what cause fatigue problems. Therefore, I wouldn’t use material that thin for the tongue. If you scrutinize the second set of calculations (sometimes things actually get a little dicier with real light-weight, lowly loaded structure as opposed to something that has a nice big load that can generate some robust sizing that doesn’t get into a lot of these secondary failures), you’ll notice that by the numbers, the sizing is quite light gage. The fact that you can’t even buy the material this thin is probably an indicator that either the demand is not there, or it’s just too hard to extrude material that thin, or even that when people use it they run into problems in service. Regardless, for those reasons there is probably a lower limit to how thin we can get with the aluminum.

OK, there’s a limit – how do we figure that one out? The best source for data is the trusty Military handbook -- METALLIC MATERIALS AND ELEMENTS FOR AEROSPACE VEHICLE STRUCTURES. They publish curves for the allowable stress for a given number of cycles. Once the stress is low enough, you can have infinite cycles. The fatigue loading is going to be different that the ultimate static loading because you don’t go around smacking 3g sinkholes ever time you take the tear out of the garage (I’m thinking our bodies probably wouldn’t put up with that!). The trailer sees 1g just sitting in the driveway. We could probably assume something conservative and use that. My guess is that even a 1.5g loading isn’t seen too often (especially if you suspension is set up correctly to suck up the bumps); so, let’s assume out fatigue stress is 1.5. Also, when calculating fatigue, it’s important to know how reversible the load is as generally, fatigue is only a problem for the tension areas of the part – stuff, as a rule, doesn’t fail due to fatigue in compression. In the case of a trailer, we would expect that the least amount of load the tongue would see is zero – I wouldn’t expect to see up loads on the hitch, like the down loads. What we are talking about is the loads that happen over and over again – these are pretty much in one direction; so, to assume that the load returns back to zero for every cycle is in fact quite conservative. So, we’ve been conservative for load reversal, and conservative for the magnitude, so the final result should have a bit of safety built in.

The fatigue curve is below:




The stress for infinite life is a around 22Ksi for 6061 (Mil-Hdbk doesn’t have data for the 6063-T52). It would be lower, but the load doesn't fully reverse (just back to zero) which is better of fatigue (this gives a high allowable fatigue stress).

Since our previously calculated stress is for a 3 g condition, the actual fatigue stress would be half of that. Originally, for a 16 gage section (.065 wall), the stress was just under 29 Ksi, so the fatigue stress would be 14.5 Ksi. We can see from the curve above that this is below the infinite life stress therefore it should be fine. Having said that, my earlier suggestion that thinner sections are not as robust with respect to fatigue, leads me to want to use at least the 14 gage (.085 walls) material. Since that’s not available (and actually neither is the 16 gage for that matter), I settled on the 11 gage (.12 walls). At this point, following through the previous calculations, the stress has dropped considerably as with each gage increase we’ve added area and stiffness. The final stress (3g) was 16.8 Ksi, with a fatigue stress of 8.4. Now at this stress level, even if the stresses were fully reversible (which they are not) we would easily have an infinite life. Additionally the aluminum tongue is just a little more robust and stronger than a steel one AND it weighs 3 lb less (even for such a small light-weight trailer – for a larger trailer, the weight-savings are even more significant). It's also less likely to cripple as the walls are thicker.

As a cautionary note: if you have to put in holes, put them in the side near the center, if it’s a filled hole (tight tolerance hole with a bolt in it or a squeeze rivet that fills the hole), then you can put it on top of the tongue. The bottom of the beam is a place where you would try and avoid putting a hole (stress concentration).

Andrew-specific
The fact that it is the same magnitude is quite comforting, but, I was mostly trying to suggest two things. If you have a high tongue weight is may be significantly unconservative, and especially so if you are going off-road. In that case, I would highly recommend just calculating the moment based on 5gs and using the tongue strength chart from there.

Though it might appear to be a general instability failure (buckling), I suspect that the failure mode, especially for the thinner gages, is in fact crippling, which is a local instability. I could be wrong -- I'd have to check.

Your point about the lateral stiffness is especially well taken -- that's why I gravitated toward the 2x2 section as intuition tells me the 1x2 could get a little whippy in the lateral direction. Also, the standard couple is 2 inches wide which integrates into the 2x2 tube nicely.

As long as the trailer only see 3'gs, the tongue should see it. Of course, in Florida, they got sink holes big enough for the car and trailer

[kennyrayandersen]

OK, back from the dead.

Lots of things got in the way -- you know -- busy work, busy home, and then there was the I realized that the cutout for the AC was too close to the roof, and that the kitchen wasn't deep enough, so I had to completely redo the first model, run everything again etc. There are several conditions and not all plots for all conditions are shown. Sometimes an area would be critical for one condition, like forward floor parked, but the aft floor would be critical for the 3g road (inertia) condition, etc. Anyway, it's probably way more than anyone wants to know, but I'm trying to show that the concept is valid and the result is a very light-weight tear. Except for a few local reinforcements, which are entirely expected, the stresses on a tear whether parked or going down the road are quite small. Therefore, if we strengthen these local areas only, the rest can be quite light yet quite strong enough using these composite building techniques. It will be a bit before I can start the build so it's a bit academic at this point, but this is how I plan to build so hopefully this will be backed up at some point with some empirical data. Also, as I continue with the design I will post additional updates.



This is the deflection of the floor in the 3g condition (road)


this is the same condition as the floor above, but this plot shows the strains in the floor fiberglass skins.


this is the same condition as the floor above, but this plot shows the strains in the interior fiberglass skins.


this is the same condition as the floor above, but this plot shows the strains in the side wall fiberglass skins.


this is the same condition as the floor above, but this plot shows the strains in the top fiberglass skins.


transverse shear in the floor (30 is allowed)


transverse shear in the floor in the other direction (30 is allowed)


transverse shear in the interior (30 is allowed)


transverse shear in the interior other direction (30 is allowed)


This is the parked deflection (load case -- two people co-located in the middle of the trailer (See the classic Finn's motto -- i.e. do you sleep in that thing?... not only!)


Parked strains -- trailer supported at the tongue


top core shear (30 is allowed)


top core shear other direction (30 is allowed)


Transverse shear in sides (30 allowed)


Transverse shear in sides other direction (30 allowed)

[/img]
transverse shears near axle attach. Note that this area will likely have to be reinforced locally, though the allowable is 30.

[bobhenry]

Man oh man oh man. and all the pretty colors too.

While I have all the great minds in one corner I would like an opinion on tongue design.

I see a concentrated tongue load at the front of the trailer in the pictures above . My thoughts have always been to extend the main member of the tongue ( with or without diagionals) from the ball mount clear to the rear most member of the trailer. Any up lift at the ball mount creates a positive load in the front but a negative load at the rear. With the rear member pulling downward this would reduce the load at the front. By distributing the load more evenly it would seem a lighter tongue would be possible because you have reduced the bending moment at the front most point of the trailer. We have see the utility trailres that have support only to the second member and the member is distorted ( bent) downward . Had this load been distributed across the full length of the trailer and had been shared with all the cross members I do not see it happening. It seems the extra 6 to 8 foot of tongue under the rear of the trailer would spread the loads more evenly whether positive ~negative ~ right or left. Is there any merit to this thinking ?

[angib]

Kenny, can you describe the loading and boundary conditions - how did you apply the loads (as self-weight of elements?) and how is the model restrained?

The very low stresses at the ends of the door cutouts surprises me - I've always had a gut concern about those points.

Andrew

[angib]

Is there any merit to this thinking ?

Bob, in general, I'd say 'no'! Extending the tongue under the trailer does not change the load on the tongue at all - it just alters how that same load is transferred from the tongue into the main frame. If you look at, say, the critical point where the tongue passes the front of the main frame, then changing the structural design behind that point does not change the loads in front of that point.

I'm not suggesting that trailers with a single tongue fixed to two closely spaced cross members is a good idea. The simplest improvement is to move the cross members apart, which will push the rear one closer to the axle.

Andrew

[bobhenry]

Is there any merit to this thinking ?

Bob, in general, I'd say 'no'! Extending the tongue under the trailer does not change the load on the tongue at all - it just alters how that same load is transferred from the tongue into the main frame. If you look at, say, the critical point where the tongue passes the front of the main frame, then changing the structural design behind that point does not change the loads in front of that point.

I'm not suggesting that trailers with a single tongue fixed to two closely spaced cross members is a good idea. The simplest improvement is to move the cross members apart, which will push the rear one closer to the axle.
Andrew

and that is all we are trying to do alter how the load is transfered to the frame. As the tongue is lifted the load is transfered as a downward pull on the frame members aft of the axle relieveing so of the weight seen at the very front of the frame. Would this not be the same as adding weight aft of the axle to relieve some tongue weight. I realized the tongue would have to handle the same total weight it is simply spread throughout the frame members.[color=blue][/color]

so if we move the tongue to the centerline of the axle we have improved the loading condition! So why not extended it beyond the axle ?

[kennyrayandersen]

Kenny, can you describe the loading and boundary conditions - how did you apply the loads (as self-weight of elements?) and how is the model restrained?
The very low stresses at the ends of the door cutouts surprises me - I've always had a gut concern about those points.
Andrew

The constraints are simply supported at two points along either side where the axle would attach and simply supported at the coupler ball. There are no other constraints in the model. I used RBE3s (a multipoint constraint which has no stiffness – load point is dependent, others are independent) to apply the loads. Loads were 65 Lb ice chest, 44 Lb water, 40 Lb AC, 35 Lb mattress, 30 Lb misc kitchen gear. I should plot a picture of that (it would be worth a thousand words!). All other elements had mass based on the material card physical properties. For the 1g parked condition the inertia was 1g, and for the 3g condition it was obviously 3g. Since it’s a linear model, the results from either can be directly scaled.

There’s not much to get wrong with the model as the boundary conditions are quite simple. I have the tongue attached at 3 locations to the body with only the aft location having an x-member present. I was a little surprised by the stresses/strains around the door as well, and that’s why I put a generous radius at the door corner as I was expecting a bit more there myself. Evidently, the total load being applied at the front wall of the tear isn’t insignificant of course, but the ‘beam’ carrying the load is the full depth of the sidewall (shear) and the roof and floor (upper and lower caps in the overall ‘beam’). This beam is quite capable it appears. This couples with the large cutout radii result in quite low stresses in the corners.

Not included in the model are the doors, which should add a few pounds per piece, some hinges and miscellaneous hardware, but I don’t think any of those will change the results so much. There are still some areas that IMO it would be wise to reinforce locally, but none of that looks to be anything crazy. I think an extra ply on either side of the sandwich where the front vertical wall meets the floor, and maybe a ¼ inch half-round of plywood in the side walls at the rear axle attach plus a couple extra plies locally should pretty much do it for that area. Perhaps the outboard portion of the lower vertical galley bay might get and additional ply in the facesheets. In addition to those, I think a plywood plug at the two additional tongue attach points. Also, right now the tongue is about 8 feet. – I’m thinking to extend it full-length, not for strength, but because I want to put a receiver to put a bike rack out at the rear. Also, though I showed that it was possible to save a little weight to go with the Aluminum tongue, I think that the receiver and the front coupler will integrate better with steel and the weight penalty isn’t too bad considering that I could go to 16 gage (.065 wall) with the steel and still meet the strength requirements – it would be 5-6 extra pounds. The aluminum alloy comes in 11 gage, and so is heavier than required to begin with.

The big thing I noticed is that if you build with composite sandwiches like this, the body weight is in fact much lighter; so, the axles, apparently, need to be shifted back even further toward the rear than usual – even without so much in the kitchen. In addition to that, if I hang the bikes off the back, it will exacerbate that situation further. I might have to come up with a bit of a counter-weight that I can stick on the tongue when I carry the bikes – something that mounts just aft of the coupler that acts as ballast. I wouldn’t want the tongue to be lifting off the ball!

[kennyrayandersen]

and that is all we are trying to do alter how the load is transferred to the frame. As the tongue is lifted the load is transferred as a downward pull on the frame members aft of the axle relieving so of the weight seen at the very front of the frame. Would this not be the same as adding weight aft of the axle to relieve some tongue weight? I realized the tongue would have to handle the same total weight it is simply spread throughout the frame members.[color=blue][/color]

so if we move the tongue to the centerline of the axle we have improved the loading condition! So why not extended it beyond the axle ?

Andrew is nearly 100% theoretically correct and fully 100% practically correct. If there are two attach points on the trailer – one at the front wall intersect and one somewhere aft, the reactions on the tongue are up at the front, down at the vertical wall intersect and up at the aft (note that the reactions at the trailer are opposite with respect to the trailer, the aft attachment wants to pull itself away from the trailer body). The reactions at the vertical wall and aft attach points vary a bit as the tongue overhang and the spacing between the vertical wall intersect and the aft attach points change. I can go into the math, but it’s the same as used in the tongue stress calculation earlier in the post. Considering that the overhang will be constant (once we decide where it should be to function properly, that’s where it stays), the only other variable is the aft attach point. As Andrew was implying by saying he didn’t recommend putting them close together, if these two points were close, they would generate very large reactions and thus stresses. We don’t want that. Of course, we don’t want them really long either necessarily due to weight etc. As the aft attach point moves further from the vertical wall intersect point the reactions on the body decrease. The force at the front of the trailer is: the sum of the load at the ball; plus, the load at the aft attach point. Practically, if you are anywhere near the rear axle you are already so far aft that the reaction is quite small and manageable and there is not really much point in taking it any further. Actually having a tongue about 6 feet in total length also gave pretty reasonable results.
Now, if we attach the tongue to the rear axle, we actually increase the bending moment in the axle beam because the tongue will be applying a down load to the axle, which causes a positive bending moment. At the same time the wheels pushing up and the body edge pushing down also cause a positive bending moment in the axle so those bending moments are additive (not that it would be bad necessarily, but they are in the same direction). I think Andrew is right though, and you just as well connect it to the body and since the load, by this point has already dropped off significantly, there is no point in making the tongue longer.

Except, in my case, I’ve decided to add a bike rack since I mountain and road bike and could see that putting a receiver back there would be a slick easy way of adding a bike rack. I’m not exactly sure of the configuration where both the tongue and the axle are together, because I haven’t thought about that. I could notch the tongue, since it is a deeper section, or I could notch weld both of them together, or I could bolt them both together using a shear clip (should be lighter to weld). If that’s not clear I could make a little freebody diagram on a simple beam – but I’m guessing this is already more than you want to know.

[aggie79]

The big thing I noticed is that if you build with composite sandwiches like this, the body weight is in fact much lighter; so, the axles, apparently, need to be shifted back even further toward the rear than usual – even without so much in the kitchen. In addition to that, if I hang the bikes off the back, it will exacerbate that situation further. I might have to come up with a bit of a counter-weight that I can stick on the tongue when I carry the bikes – something that mounts just aft of the coupler that acts as ballast. I wouldn’t want the tongue to be lifting off the ball!

I will have the same situation with bikes. On short trips (one-night stays), we probably won't carry the bikes. Because I will have a small battery, I've thought about using it as ballast and being able to move the battery from the galley location to the tongue box. Of course, I'd have to wire for both locations and have a selector switch.

[angib]

so if we move the tongue to the centerline of the axle we have improved the loading condition! So why not extended it beyond the axle ?

"Declining returns" is the answer!

To extend the tongue to the back of the frame adds at least 4 feet of extra tongue and doesn't help much with reducing the loads on the main frame. You would get a stronger trailer if you took the weight of that 4 feet of tongue and used it to make other bits of the frame stronger.

Andrew

[angib]

In addition to that, if I hang the bikes off the back, it will exacerbate that situation further. I might have to come up with a bit of a counter-weight that I can stick on the tongue when I carry the bikes.

For stability, adding ballast at the front won't necessarily help. Just increasing tongue weight does not guarantee stability.

As you understand the technical side, it's polar moment of inertia that is the main cause of trailer instability and so putting bikes on the back and a counterweight on the front is the perfect way of increasing the moment of inertia and reducing the stability. Putting just the bikes on the front is a bit better, but not much.

Andrew

[kennyrayandersen]

Andrew,
You're ahead of me on that one. I was only referring to the trailer balance. Any idea what the natural frequency for the trailer should be? I could run a NASTRAN solution and solve for the lower frequency modes; but, once I got the answer I'm not sure what I'd do with it. Maybe I'll do it anyway and just see what I get in the with and without the bicycle configuration. If the frequency is high -- higher than the wheels are rotating, then maybe there won't be so much to excite the mode other then PIO (pilot induced oscillation)?

The mass and especially the mass moment of inertia go up like you say -- I just hadn't thought about it. That's another reason your first light-weight A-frame idea might not be such a bad deal. Bike addition would be tougher with the A-frame though. I guess you could run some guide wires to the side or something?

The steel is a bit stiffer than the aluminum -- I just don't know if it's stiff enough. As Arnold would say... I'll be Baak