N-Decision: 15 inch x15 foot

The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I'll offer up my modest analysis skills though there's not much I can do that Tim can't do better. How'd you wrangle PML into doing these tubes? I'd spoken to them for another project and couldn't get anywhere even with a much higher budget than this project.
 
botscott08,

Maybe things have changed with the addition of PML’s telescope tube business: https://www.carbonscopetubes.com/Custom_Sizes.php.

I would credit myself with superior powers of persuasion. But that’s unlikely. I did have several telephone calls with Frank where we negotiated about existing tooling and in-stock materials.

Note my first inquiry was for three pieces of 20-inch by 59-inch airframe. Quote was $750 per tube before shipping. Figure $175 materials, $350 labor, $70 insurance and $50 capital / rent. That leaves $105 profit or about 15%. I’m not a PML booster but I do believe that is fair trade.

Feckless
 
That's a fantastic price. I paid more than that for my 8" tubes and I even supplied the carbon fiber and I bought 8 tubes. I'll have to bother them for my next project.
 
TRF,

Attached is a preliminary design. Several things worth noting on this scratchpad:

1. Format is 3FNC.
2. Pad weight is approximately 130 pounds including CTI 20k N-s motor.
3. Thrustplate is estimated as 10 pounds 6061 aluminum.
4. Motor mount is 7.5-inches. Adapters will configure motors progressively. Thanks for the encouragement Todd.
5. Otherwise rings and bulkheads are 0.50-inch birch.
6. Finstock is 0.75-inch end grain balsa sandwiched by single plies of 3k carbon twill fabric.
7. The 11-inch shoulder is the main parachute bay. Volume is 130 cubic-inches.
8. Main parachute is Sky-angle XXXL. That's right, triple X!
9. A 4-inch access tube allows nose-weight adjustments and maybe a tracking device.
10. Coupler is 14-inch centered on the junction. Construction is open lay-up fiberglass.
11. Simulation shows 8k AGL on CTI-N3301. Top speed is 580 MPH on matt finish.
12. Simulation shows 13k AGL on CTI-O8000. Top speed is 850 MPH on matt finish.

Feckless

15-inch Scratchpad.jpg

15 inch scratchpad rendering.jpg
 
7. The 11-inch shoulder is the main parachute bay. Volume is 130 cubic-inches.
8. Main parachute is Sky-angle XXXL. That's right, triple X!

Are you talking about the 15-inch by 11-inch nose cone shoulder? Volume is probably more like 1300 cu. in.

130 cu. in. won't be enough for the XXXL.
 
Tim,

Posted directly to the forum as a .RKT file. Hope that works.

Feckless

Feckless, before I start, I just want to verify a few things...

  1. Are meaning for the rocket to be single deploy as shown in the simulations? Or are you doing drogueless dual deploy and then ejecting the parachute bulkhead at altitude (with alts in nosecone)?
  2. I believe there is a stray 7.5" to 4" adaptor ring in the MMT, please verify that it is "stray."
  3. Finally, you have a "coupler bulkhead" at the aft of the main coupler. Why? I only say this because you are limiting yourself to 52" of motor. This is likely not sufficient for what Todd H. might have planned. A 40kNs 6" motor for instance would typically be 52" to 56" in length. Please reconsider the need for this bulkhead. Or if you keep it put it at the forward end of the coupler to give you 66" of motor length.
  4. I don't see a mass component or sleeve for the fin lamination... Where you planning on adding that after fin simulations to get the correct weight? Or did you already have a planned layup for the fins that I should enter in the simulations?
 
Last edited:
Are you talking about the 15-inch by 11-inch nose cone shoulder? Volume is probably more like 1300 cu. in.

130 cu. in. won't be enough for the XXXL.

This is probably true, but tons of volume can be recovered by reducing the length of the nosecone central tube, so he'll likely be OK (if I understand how his deployment is working).
 
Fin laminate and simulation summary:

  • 0 to 1 layers of Carbon Fiber (CF) => Flutter Likely.
  • 2 layers of CF => Flutter Possible; 0% Flutter Margin.
  • 3 layers of CF => 15% Flutter Margin.
  • 4 layers of CF => 25% Flutter Margin.
  • 4 layers provides nearly a 20% improvement in Angle-of-Attack (AoA) margin versus 3 layers.
Notes:
  1. All analysis using 3/4" end-grain balsa core.
  2. Assumes cross-ply CF layup, e.g., 4-layer: 0, 90, 45, 135, 0 (core), 135, 45, 90, 0.
  3. Large fins with low strength core sensitive to AoA stresses.
  4. Veil layer of fiberglass (FG) provides negligible strength addition and was ignored in laminate analysis.

NDecision_9AoA.JPG

NDecision_9Flut.JPG

NDecision_7Flut.JPG

Capture.JPG
 
Last edited:
Tim,

Responses below:

Feckless, before I start, I just want to verify a few things...

  1. Are meaning for the rocket to be single deploy as shown in the simulations? Or are you doing drogueless dual deploy and then ejecting the parachute bulkhead at altitude (with alts in nosecone)?
    >>Dual deploy with a tiny drogue to control separation during decent. It's not yet included as the design is preliminary scrtachpad.
  2. I believe there is a stray 7.5" to 4" adaptor ring in the MMT, please verify that it is "stray."
    >>It there as a reminder to account for removable adapter weight. In that regard it is a "stray."
  3. Finally, you have a "coupler bulkhead" at the aft of the main coupler. Why? I only say this because you are limiting yourself to 52" of motor. This is likely not sufficient for what Todd H. might have planned. A 40kNs 6" motor for instance would typically be 52" to 56" in length. Please reconsider the need for this bulkhead. Or if you keep it put it at the forward end of the coupler to give you 66" of motor length.
    >>Understood and agree. That part is a centering ring. I will recheck the model to assure it has a 7.5-inch hole.
  4. I don't see a mass component or sleeve for the fin lamination... Where you planning on adding that after fin simulations to get the correct weight? Or did you already have a planned layup for the fins that I should enter in the simulations?
    >>No layup schedule yet. But I see you have some suggestions in the next post.
 
That's a lot of weight, and in the wrong part of the rocket. Why so thick?

I agree. Ten pounds is a lot of "junk in the trunk." The component is not yet designed but will probably be about 4 pounds. That is based on the plate calculations presented earlier in this thread. I would welcome and appreciate your comments on that design process.

Feckless
 
Tim,

Thanks! I recognize you put some thought and effort into this. That is very much appreciated.

You’re a few steps ahead of me. Please help me catch up? Can we start with the panel layup and how you derived the user defined material specifications for FinSim? Did you use “The Laminator” or some other method? If you did use The Laminator are you sure it applies to sandwich panels?

Feckless
 
I would welcome and appreciate your comments on that design process.
As CarVac said, if you used ribs you wouldn't need the plate itself to be so thick. Ribs could even be made out of plywood. Aluminum does make a nice tail plate, but I've generally gone with 1/16" or at most 0.100" thick. Just make sure that there are no long spans without ribs or a plywood backer.
 
Tim,

Thanks! I recognize you put some thought and effort into this. That is very much appreciated.

You’re a few steps ahead of me. Please help me catch up? Can we start with the panel layup and how you derived the user defined material specifications for FinSim? Did you use “The Laminator” or some other method? If you did use The Laminator are you sure it applies to sandwich panels?

Feckless

Feckless,

Let me supply some of this answer and I'm sure there will be more questions which we can address. I want to preface this by saying I am not a materials or structural or mechanical engineer. My use and interpretation of modeling data and software has been acquired over hours/days, dare I say months of study Federal/DoD and various university/industry docs and texts purchased. Saying that, I am far from formally trained and I may absolutely be wrong in my projections and conclusions--caveat emptor. Now, if you are scared enough, I will proceed (with caution)...

Yes, I did use The Laminator and the product specs you provided for the core material. The product specs provided just covered the strength specs and not the physical/mechanical specs. So I supplemented the strength specs with the physical properties of light Balsa found in a number of government/forest industry reference documents. The aggregate specs that were used are in the file below:

View attachment EndGrainBalsa.docx

From that point I built a user-defined laminate material aptly named End Grain Balsa. It was the core of a stack up that was surrounded by equivalent of T300-type carbon fiber / epoxy matrix. I progressively added CF layers and simulated (see pic below, core material not shown). I then subsequently fed the aggregate laminate properties into FinSim with the appropriate fin geometry given in your Rocksim file and fin thickness as defined by the layup. Summary results were given previously, but I had to get up to 2 layers of CF on each side (5 layers total) to get above Mach 1.1/1.2 on flutter and divergence velocities. Seven layers total provided 15% margin, nine layers total 25% margin. In addition, I did try a few fiber cross-ply variations on the seven and nine layer stacks to provide optimum results. As expected it is the cross-ply stack shown in the previous post. Of course, in addition, in each layering step the flutter results benefited both by a stiffer skin and an overall thicker fin.

ply_stacking (2).jpg

For completeness resultant aggregate properties of the 7 and 9 layer stack, respectively, are given below (i.e., The Laminator output).

View attachment Laminate15incher_v2.txt View attachment Laminate15incher_v5.txt

Now for your last question... I believe the real question is, "Can The Laminator software adequately model a composite structure that is "dominated" by a vertical honeycomb structure?" My answer is, "I'm not sure." To defend that answer, I will post an excerpt from a well written paper on honeycomb material characterization (Section 4.1.2, Macromodeling, apparently a Master's Thesis from a student at the Imzar Institute of Technology in Turkey). It basically summarizes what is true of laminate analysis in general, it is a simplification and averaging of the characteristics of each of the composite materials, not only in their individual layers, but also across the whole 3D structure. Under that definition, yes, I feel the results are reasonable although I do suspect that the Angle-of-Attack characteristics are conservative (but that maybe should be left for another post). Saying the previous, it is clear that a honeycomb structure with its fiber strength in the vertical direction is a different beast than carbon or fiberglass cloth. At this link is the full paper to show you how they modeled the subcomponents of each of the honeycombs and the facesheets. So are we close, I believe so. Are we exact, HECK NO!

The macromechanical approach is concerned with the contributions of each ply
to the overall properties, therefore the properties of the fiber and matrix are averaged to
produce a set of homogenous, orthotropic properties. In the case of composite laminate
there is an additional level of complication which arises as a result of stacking several
layers of composites with different orientation and properties. For a given stacking
sequence, the stress-strain relations of a composite laminate can be derived and the
various coupling mechanisms between in-plane and out of plane deformation modes can
be explored. In macromechanical modeling, prediction of failure of a unidirectional
fiber reinforced composite is usually accomplished by comparing some functions of the
overall stresses or strains to material strength limits. Several failure criteria such as
maximum stress, maximum strain, Tsai-Hill, Tsai-Wu have been suggested to predict
the failure. These criteria are based on the average composite stress strain states.
Macromechanical modeling does not consider the distinctive behavior of the fiber and
matrix materials. Although the macromechanical approach has the advantage of
simplicity, it is not possible to identify the stress-strain states in the fiber, matrix and
their interface. In contrast, in the micromechanical approach, the constituents and their
interface can be definitely considered to predict the overall response of the composite as
well as the damage initiation and propagation in the composite. (Chen 2000)


Cheers,
Tim
 
Last edited:
Tim,

Was showing my concerns at fin design to a colleague. His comment was hobbies don't include math. Point taken and I promise we'll get back to building big @$$ rockets. But the fin design is important.

For the weary here is a summary. For insomniacs please read on.

1. FinSim is an "aeroelastic" analysis application for rocket fins. It is now out of print.
2. I simulated my fin design using 0.75 inch thick "balsa-only" as first pass to carbon-balsa-carbon sandwich panels.
3. FinSim predicts excessive flapping near 600MPH for "balsa-only"
4. Twisting for "balsa-only" appears to be acceptable.
5. Sandwiches improve bending (flapping) properties. Hope abides.
6. I'm calling some friends to help simulate the materials properties of the proposed sandwich finstock.
7. Comparison of the material simulation to certain parameters in FinSim will stimulate confidence towards flutter calculations.

My own effort begins with FinSim's default balsa material. According to the application's instructions a lower bound of flutter (flapping) and divergence (twisting) are calculated using Pine's "quasi-steady" analysis. Note I've use the default lift curve of 2*pi. Data are, according to my simulation for generic 0.75- inch balsa, given in figure 1. Also note I've relocated the center of gravity to 60% root-cord based strictly on the geometry of a rectangle married to a triangle.

These data indicate flapping resonance at 632MPH and twisting resonance at 750MPH. This is the resonant condition alone. It says nothing about material stress and strain bracketing resonance.

The situation is illustrated with detail using FinSim's "unsteady" U-g analysis illustrated by figure 2. There "g" is the dampening factor. Where "g" is negative vibrations tend to dampen out. Zero and positive values of "g" mean vibrations will grow in amplitude at each "harmonic" cycle. There the fin fails.

My observation of the U-g analysis in "balsa-only" is low damping in the flapping, bending mode. The twisting mode appears sufficiently "stiff" to the application.

Since sandwich panels excel at redistribution of bending forces I believe there may exist a single skin, light weight design as suggested.

I have some friends in the materials simulation business. I'm going to call favors and, hopefully, validate the NACA data circled in figure 2. As CAE is expensive I can only ask for simulation of the proposed fin. However, if FinSim's NACA parameters can be made to match the CAE material data then confidence should be placed in the resonance prediction. Material stress / strain is another conversation.

Comments?

ND flutter figure 1.jpg

ND flutter figure 2.jpg
 
Fleckless,

A couple of comments (some obvious):

  1. Thanks for the input on CG. As you might note in my initial results, I moved it back to 0.5, but 0.6 is more likely given the geometry.
  2. Clearly the Balsa-only numbers don't meet your expected velocity limits, falling more than 20% and 15% short of 1.1 Mach on divergence and flutter respectively (in Pines model), and thus the need for the CF lamination layers.
  3. You provided an altitude in which the analysis should be conducted. Typically I just use Sea Level because the results are more stringent (providing some safety margin), but to be accurate based on your Rocksim results, you reach maximum velocity at ~3Kft, not 10Kft and thus this is the pertinent altitude that should be used (see attached graph).
  4. U vs g results tend to be more stringent for Flutter velocity (being more than 20% lower in the case you show -- 578mph vs 750mph). I tend to rely on the Pines model over U vs g. They seem to be equally referenced in the literature, but just based on my experience, when I have had a fin shear from flutter, it has been at velocities closer to the upper limit on Pines, not U vs g.
-Tim

RS9Result.JPG
 
Last edited:
Fleckless,

Uh, oops... In reviewing my last post I did some review of the sims starting yesterday evening and realized in my haste (and doing first round late at night) that I screwed up and used the Modulus of Elasticity in the Y-direction for FinSim input versus the X-direction. Too many numbers on that columned output--sorry! So taking a step back I re-simmed everything again in The Laminator and FinSim. Here are the corrected FinSim inputs and critical velocity results:

FinSimData.JPG

So, suffice it to say, I am now dialing back and saying two layers of CF is adequate to get the safety margin you need (Divergence @ 1.44 Mach vs expected velocity of 1.1 Mach; 30% margin). FinSim analysis was done at the 3Kft altitude level when approximate max flight velocity is reached.

-Tim
 
Last edited:
TRF,

Attached are photos of my feckless work at making a coupler for these airframes.

First step is to make a release surface from 10 mil Mylar film. That thicker grade of Mylar is used for rigidity. It holds it shape within the tube where thinner sections of Mylar would "flop" and collapse. Because this Mylar is rigid it can be formed and taped to ITSELF along the inside seam. That is important as taping the seam to airframe impedes extraction.

Next the Mylar ends are taped to the airframe. That is to seal edges and to prevent possibility of Mylar's collapse under the load of wet layup. This set-up is shown in the first picture. Note several dark lines are shadows visualized through the translucent tube.

Cloth selected for this job is a 16-ounce satin weave. Thickness is 0.018-inch. That selection is made for both strength and for rapid build up to desired wall thickness. It is cut to half yard sections that cover approximately 200 degrees circumference. I did not want to make full turns about the circumference for fear of sag and collapse at lay-up.

Cloth sections are laid dry into the airframe atop fitted Mylar and then saturated with 8-ounces of neat resin. The mass is worked by squeegee followed by a resin roller. Each "hemisphere" is allowed 3 hours cure before the next is laid. Two layers or four hemispheres are laid as shown in the second picture.

After an overnight cure those end tapes are removed and the coupler is removed. It slides right out as shown in the third picture.

I removed the coupler after two layers to more easily work cloth and resin. Working inside a 60-inch long airframe is awkward. Another 4 layers will be applied for a final thickness of 0.100-inch.

Feckless

ND Mylar Release.JPG

ND Coupler Glassed.JPG

ND Coupler Extract.JPG
 
TRF,

For those interested I’ve received preliminary results from CAE on the proposed fin: ¾-inch end grain balsa, two layer of 6-ounce carbon twill and US composites epoxy. First task was to compare the natural frequencies of those fins among simulations. Please refer to the attached thumbnails as I discuss observations?

1. First three bending modes are illustrated
2. Relative stress goes from blue to red, red being the highest stress.
3. Note blue contours define nodal modes.
4. Bending / flapping is the first mode. Note single blue contour at root’s edge.
5. Torsion is the second mode. Note the fin’s cord is split across a blue contour.
6. Divergence, both bending and flapping, is the third mode. Note the bifurcated blue contour.

These results are consistent with the literature discussed previously. First bending, then torsion followed by divergence. Divergence is destruction assured but not the necessarily the LOWER BOUND of fin failure.

Now to compare the natural frequencies as predicted by CAE and by FinSim:

1. Bending according to FinSim balsa only = 118Hz. Bending composite fin according to CAE = 72Hz. Difference = (40%) relative to FinSim.
2. Torsion according to FinSim balsa only = 180Hz. Torsion composite fin according to CAE = 188Hz. Difference = 4% relative to FinSim.
3. Divergence according to FinSim balsa only = 184Hz. Divergence composite fin according to CAE = 348Hz. Difference = 47% relative to FinSim.

Note these result are inconsistent in the mixed materials input data. Also confusing is why the composite’s natural frequency is not consistently above that of the raw balsa.

Next steps:

1. Confirm the natural frequency simulation using test coupons and accelerometer data.

Feckless Again

ND Fin mode 1.jpg

ND Fin mode 2.jpg

ND Fin mode 3.jpg
 
Last edited:
Considered Foam Core ?


High density foam, 3 layers 5.5 Carbon vacuum bagged. Stiff as steel all 6 only 7.2lbs

325lb rocket. Flew on O's& P to M-1.2

ready to fly again.


Another way: High density foam sandwiched between 1/16 sheet G-10.
Scary stiff.
Super Slick Finish to paint.

100_4955.jpg

100_4957.jpg

100_4958.jpg

100_6164.jpg
 
Last edited:
What does the rocket weight have to do with the thrust plate calc?
Thrust is Thrust.
Does NOT matter if it's fighting gravity or drag.....100% of the motor is 100% of the motor....the thrust plate takes it all.

Also - the thrust plate USUALLY pushes on the aft end of the fins....
Your fancy strength calc's are probably not modeling the support correctly.
 
Jim,

No, I hadn’t. But I want to thank you for this data point. It points exactly to where I want to go. May I ask what foam did you select? And from your experience how important is the core? I mean, do you believe you could have picked out the core and been left with a hollow fin just as robust?

Feckless
 
Fred,

Thanks for your comment on that subject. My habits favor error so I will appreciate your insight.

I understand you comments about load sharing with fins and through motor mounts. Could we please consider the case where that is not the design and thrust plate transfers the entire load to airframe?

I included weight of the rocket after drawing a free body diagram. That would illustrate the sum of forces acting on the plate. That is to say it must support both the weight of rocket pushing down on its flanges and the thrust of motor pushing up through its center. Both forces act in concert to deform the plate to a concave shape. Would you agree?

Otherwise may I please ask your help? If you were asked to design such a minimalist thrust plate how would you approach the numbers?

Feckless
 
Back
Top