Dimetrodon - A square rocket that doesn't quite fit in

The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

boatgeek

Well-Known Member
Joined
Dec 27, 2014
Messages
7,229
Reaction score
7,559
My major winter project this year is a bunch of firsts:

First 54mm motor
First rocket designed from blank sheet for L2 motors
First hand-laid tube
First NC-cut fiberglass fins and bulkheads
First large scratch-built toroidal chute

And what brought you here: First square rocket. Here's a couple of pictures of the Rhino model so far.

Dimetrodon1.PNG Dimetrodon2.PNG

You know when you get a package of plastic dinosaurs and there's the sail-backed one? That's a dimetrodon. They were the top predators of their time, about 200 million years ago, long before most of the other dinosaurs we know and love. Because I really like to do biomimicry in rocketry and I'm basing this on a sail-backed creature, the fin shape is naturally based on a dimetrodon sail (D. gigashomogenes if you care).

The tubes will be laid up on a piece of 4"x4"x1/4" steel square tube that a co-worker scavenged from his dad's shipyard's scrap bin. I have laid up two short test pieces. The first one was a dog's breakfast, but the quality of the second makes me pretty confident that I can make this all work. It turns out that having the proper tools and keeping tension on the glass going on to the mandrel are both extremely important. More on that in a future post when I have the pictures. I'll also have to do a lot of custom work to make couplers and the nose cone. I haven't really figured that all out yet, but it will involve some combination of 3-D printed parts, wood moldings, and fiberglass.

I'll follow along with and some details about simulations to support stability and altitude predictions in future posts. CFD is involved to try to validate an Open Rocket based approach. There's also more backstory on the name to come.
 
Last edited:
Sounds interesting!
How do you get the fiberglass tube off the steel tube? (A question also valid for cylindrical tubes, but it feels even more difficult on a square tube)

Oliver

On the second test piece, I covered the tube with two layers of wax paper, and added paste wax to the top of the second layer as a mold release. After the layup had gotten pretty stiff, I pulled the layup off of the tube, pulled the inside layer of wax paper off, and slid it back on to the tube. The removed layer was enough thickness that it slid pretty well. That said, the test piece didn't slide the entire length of the square tube, so I'll have to watch that on the final builds. On the first test piece, I followed mostly the same procedure, except that the wax paper was taped to the mandrel and I waited longer to pull it off. I nearly didn't get it off the mandrel without cutting it.

I like Dimetrodons and I like that rocket.

Thank you!
 
Here's how I modeled the Dimetrodon in OpenRocket:

1. Calculated the diameter that would give the equivalent cross section area (4.51"). That should get the pressure drag on the nose and tail about right. Friction drag is about 12% low due to surface area, but that's life.
2. Modeled the nose as a conical shape. that's probably a little lower drag than the pyramid, but it should be pretty close.
3. Modeled the 4 fins on that diameter (not further from centerline like they actually are). That should give me a conservative CP.
4. Weights/actual CG will be updated as the build progresses.

Once I get the CFD results in, I'll be able to see if the streamlines do anything weird around the corners, and also a quick check to see if the CP agrees with OpenRocket. The intention is to err on the side of overstable rather than under. I'll also expect that OR will overpredict altitude and adjust delays accordingly.
 
By what process are you determining he center of pressure with CP? Calculations from the pressure/shear data?
 
By what process are you determining he center of pressure with CP? Calculations from the pressure/shear data?

That's basically it. Apparently, our CFD program doesn't give a center of drag, which I would think would be an important thing for most programs. However, the model is split in half down centerline to reduce computation time. The x axis is along centerline, y is perpendicular to the cut plane, and z is parallel to the cut plane. From Fx, Fz, and Mz, I should be able to calculate out a range of x locations where Mz is equal to zero (ie CP), depending on how far Fx is off centerline. Hopefully, this will not be a big range.

If you have a better idea, I would sure love to hear it! I will have Fx, y, z and Mx, y, z at my disposal.
 
At this time, I'm just using SolidWorks for my simple CFD models. I'd have to read more about its data acquisition to actually work out a way of locating the CP.

Hence my question to you lol.

Sounds like I have a new question for my uncle. If I could nail that process down, I can be free of OpenRockets over conservative estimates and Barrowman reliance.
 
That's basically it. Apparently, our CFD program doesn't give a center of drag, which I would think would be an important thing for most programs. However, the model is split in half down centerline to reduce computation time. The x axis is along centerline, y is perpendicular to the cut plane, and z is parallel to the cut plane. From Fx, Fz, and Mz, I should be able to calculate out a range of x locations where Mz is equal to zero (ie CP), depending on how far Fx is off centerline. Hopefully, this will not be a big range.

If you have a better idea, I would sure love to hear it! I will have Fx, y, z and Mx, y, z at my disposal.

Getting CP from CFD is a bit more involved. You need to work on an elemental basis and integrate. You need some additional CFD post-processing functions, like those found in TecPlot or EnSight software. I haven't done it myself, but this looks like a good start:

https://spaceflightsystems.grc.nasa.gov/education/rocket/cp.html
 
That approach is pretty much exactly what I was planning. However, since my model is split in half, my Fx isn't exactly on centerline, so it will have a contribution to Mz. Integrating over elements would be the ideal approach, but I'm asking a coworker to do this on his off time for beer so I don't want to ask too much! He was planning on a run with airflow 2 degrees off centerline as well. Hopefully everything will correlate nicely, but I don't really expect that once Mr. Murphy gets involved. :)
 
Onwards to cover work already completed. Here is a picture of the two test pieces I laid up. The second one is on the left and is obviously better than the first one to the right. The first test piece is 3 wraps of 10 oz and the second is 6.5 wraps of 6 oz.

IMG_0872.jpg

What went wrong on the first test piece:
* I didn't have enough tension on the fiberglass, so it didn't hold well to the corners. When I tried to smooth it out later, I made the huge bubble you see in the upper left.
* I think it was too heavy a glass so it didn't hold to corners well. The corner radius is about 3/4".
* I couldn't get the bubbles out with the brush I was using.

To fix those problems on the second piece, I did the following:
* Rolled the dry glass on to a piece of PVC pipe, then spring clamped that to another piece of pipe to make a weight that wouldn't unroll the glass and would still tension the dry glass. I'll get a picture of the rig when I glass up a tube.
* Used a fiberglass roller to clear the bubbles. See the picture below. This thing is magic. I highly recommend one to anyone rolling tubes.
* Went to 6 oz cloth and an epoxy (System 3 Silvertip) that is supposed to make it easier to clear bubbles. The epoxy I was using was a general purpose laminating epoxy from Fiberlay, but the System 3 stuff does seem to clear bubbles easier.

IMG_0873.jpg

My layup procedure was as follows:
* Wrap the square tube in one piece of wax paper, scotch taped to itself only (not to the tube) so it can slide on the tube.
* Wrap that wax paper in another layer of wax paper attached the same way, then put paste wax over that.
* Lay up the glass using a chip brush to put epoxy on the tube/glass and the roller to smooth it out. The top layer probably doesn't need more epoxy.
* Wrap the whole thing in one more layer of wax paper as a poor man's peel ply to keep drips from forming. You can see in the roller picture that it left some ridges where there were wrinkles in the paper. I will try to get those out next time.
* When the epoxy is partially cured but not hard, slip it all off the mandrel then pull the inside layer of wax paper out to make a small gap so that the finished piece (with one layer of wax paper still) can slide freely on the mandrel.
* Put the finished piece back on the mandrel to finish curing. I used slow hardener so it took some time.

I plan on using basically the same approach for the final tubes.

For some reason, neither Firefox nor IE are letting me upload pictures. I will get them up as soon as I can. Arrgh. Photos now attached. Thanks TRF IT staff!
 
Last edited:
If you didn't see them above, I got pictures up in the last post.

I laid up one of the body tube sections today, with the help of my daughter. Mistakes were made, but it looks like catastrophe was avoided. Pictures will follow later. Lessons learned:

* Even if you have a slow hardener, mix up two smaller batches instead of one big batch of epoxy.
* Check to see how hard the layup has cured BEFORE you pull it all the way off the mandrel. There was a moment when I really didn't think I was going to get it back on to finish curing.
* If the wax paper mold release isn't really tight to the mandrel, the layup wants to pull away from the flat sides of the mandrel.
 
I still don't have pictures of the layup, but I promise to bring them. Before cutting the tube to length, I weighed it so I could see what percent of the weight was in glass vs. epoxy. I came in at 57% glass, 43% epoxy, which seems pretty good for a hand layup. I'd be happy to get comparison info from other builds and especially vacuum bagging.

The main event for this post is that I got the CFD results in. CP from CFD with the rocket 2 degrees off of oncoming airflow was almost exactly where OR placed it when I calculated it as described above. I'm a lot more comfortable now that I've got a good margin. Here's some eye candy, all at Mach 1.2 with the airflow 2 degrees off of rocket centerline.

CFD 1.jpg
Streamlines, with the rainbow colors showing velocity from slow (blue) to fast (red)

CFD 2.jpg
Pressure, with two really nice sets of shock waves forward and aft. You can see how the grid spacing gets closer in the shock waves. That's from my coworker writing a routine that adds cells to the model where there are big transitions in mach number across a cell.

CFD 3.jpg
Velocity, same as the pressure one above.

In other news, the fin can tube is cut to length (44"), the motor mount is cut, and the fins and bulkheads are out to my other coworker's NC router for cutting. I'm changing plans on the chute. Instead of a scratch-built toroidal, I'm going to do a scratch X-form. A friend mentioned that a square rocket demands a square chute.
 
I note that top flight recovery lists x-form chutes up to 90". this might be useful if your fabric shop doesn't carry rip-stop nylon.
Rex
 
I still don't have pictures of the layup, but I promise to bring them. Before cutting the tube to length, I weighed it so I could see what percent of the weight was in glass vs. epoxy. I came in at 57% glass, 43% epoxy, which seems pretty good for a hand layup. I'd be happy to get comparison info from other builds and especially vacuum bagging.

The main event for this post is that I got the CFD results in. CP from CFD with the rocket 2 degrees off of oncoming airflow was almost exactly where OR placed it when I calculated it as described above. I'm a lot more comfortable now that I've got a good margin. Here's some eye candy, all at Mach 1.2 with the airflow 2 degrees off of rocket centerline.

View attachment 308503
Streamlines, with the rainbow colors showing velocity from slow (blue) to fast (red)

View attachment 308504
Pressure, with two really nice sets of shock waves forward and aft. You can see how the grid spacing gets closer in the shock waves. That's from my coworker writing a routine that adds cells to the model where there are big transitions in mach number across a cell.

View attachment 308505
Velocity, same as the pressure one above.

In other news, the fin can tube is cut to length (44"), the motor mount is cut, and the fins and bulkheads are out to my other coworker's NC router for cutting. I'm changing plans on the chute. Instead of a scratch-built toroidal, I'm going to do a scratch X-form. A friend mentioned that a square rocket demands a square chute.

Nice. What was your CFD code? How many cells and how long to run?
 
I think its pretty amazing that your friend's results match your OpenRocket geometry approximation. So far, my Solidworks Torque/F-resultant results have been kicking the CP slightly towards the rear of what OR calculates.
 
I note that top flight recovery lists x-form chutes up to 90". this might be useful if your fabric shop doesn't carry rip-stop nylon.

Not only does my local outdoor fabric store carry ripstop, it has 8-10 colors each in three weight/coating options! I have a 6-yard piece of fluorescent orange and some white that will make something both obnoxious and highly visible.

Nice. What was your CFD code? How many cells and how long to run?

Code was STAR CCM+, and I'll ask how many cells. Run time was in the 30-60 minute timeframe for each option on a 12-core machine. The number of cells is a little fuzzy because of the dynamic remeshing he rigged up.

I think its pretty amazing that your friend's results match your OpenRocket geometry approximation. So far, my Solidworks Torque/F-resultant results have been kicking the CP slightly towards the rear of what OR calculates.

By "almost the same" I mean "within a couple of inches" and on the conservative side. In other words, it was close enough that I'm not worried about stability if I have ~3 calibers of stability, since I'm not pushing Mach on a plausible motor. If you're getting something way different, I'd first suspect CFD setup. Second, with the runs with the wind direct from the nose, the apparent CP was quite a bit further forward. I suspect that this was basically the center of the frictional drag on the rocket with little or no steadying influence from the fins. As soon as it had a small angle of attack, the CP pulled back to about the OR predicted position. I suspect that even a rocket flying apparently dead straight is always at a very small angle of attack due to unsteady air, so the latter version is more accurate to what you see in reality.
 
By "almost the same" I mean "within a couple of inches" and on the conservative side. In other words, it was close enough that I'm not worried about stability if I have ~3 calibers of stability, since I'm not pushing Mach on a plausible motor. If you're getting something way different, I'd first suspect CFD setup. Second, with the runs with the wind direct from the nose, the apparent CP was quite a bit further forward. I suspect that this was basically the center of the frictional drag on the rocket with little or no steadying influence from the fins. As soon as it had a small angle of attack, the CP pulled back to about the OR predicted position. I suspect that even a rocket flying apparently dead straight is always at a very small angle of attack due to unsteady air, so the latter version is more accurate to what you see in reality.

I guess I have a tighter tolerance for "almost the same" lol.

Oh, my setup is highly suspect even by my novice standards :). Having said that though, when I had the Rocksim trial, I made several of my rockets and compared them to my OpenRocket models. Invariably, rocksim gave a CP location noticeably further back than OR (less conservative, but possibly still within your range of "almost the same").

I definitely need to start adjusting the flow angle though. Fortunately North AL is likely getting snow tomorrow which means I will hopefully have extra time to set up runs :eek:.

Also, nice mach cones! I haven't simulated anything over mach because none of my builds have gotten close yet.
 
I suspect that your chute will be a good bit larger than my 36" job (I am guessing around 72"). so methinks that you should really think about a minimum of 12 lines if not 16 (probably would be a good idea to ask 'crossfire' to see if he can offer some advice).
Rex
 
I guess I have a tighter tolerance for "almost the same" lol.

Oh, my setup is highly suspect even by my novice standards :). Having said that though, when I had the Rocksim trial, I made several of my rockets and compared them to my OpenRocket models. Invariably, rocksim gave a CP location noticeably further back than OR (less conservative, but possibly still within your range of "almost the same").

I definitely need to start adjusting the flow angle though. Fortunately North AL is likely getting snow tomorrow which means I will hopefully have extra time to set up runs :eek:.

Also, nice mach cones! I haven't simulated anything over mach because none of my builds have gotten close yet.

Yeah, I go back to my old materials professor, who said that engineering is the science of good enough. :) I figure that a couple of inches is a pretty reasonable margin when you're thinking of a simplified hand calculation method (Barrowman) compared to a CFD-type detailed analysis. Since I don't want to be anywhere near marginal and I'm not trying to push the envelope, I'm OK with a little more uncertainty. I also work in an industry where +/- 10% is a totally reasonable margin for things like power required and weight. There is a lot of art mixed with the science of naval architecture, so I'm more comfortable with a handwave of "close enough."

Yeah, I was surprised how well the mach cones came out. I've flown one machbuster, but it was kind of ridiculous. Someday, I'll dust it off again and put a smaller motor in.

Rex R, thanks for the pointer. I'm guessing 48-60" or so, but will recheck once I get more weights in from building.
 
Yeah, I go back to my old materials professor, who said that engineering is the science of good enough. :) I figure that a couple of inches is a pretty reasonable margin when you're thinking of a simplified hand calculation method (Barrowman) compared to a CFD-type detailed analysis. Since I don't want to be anywhere near marginal and I'm not trying to push the envelope, I'm OK with a little more uncertainty. I also work in an industry where +/- 10% is a totally reasonable margin for things like power required and weight. There is a lot of art mixed with the science of naval architecture, so I'm more comfortable with a handwave of "close enough.

And That brings my old fluids/Heat transfer (and part-time boss) to mind: "Heat transfer (and sometimes fluids) can be alot of guesswork. Pick your assumptions and neglect what you can so you get in the ballpark without going insane. One model may be a better guess than another model. And if you become a good enough guesser, you can get someone to pay you for it!"

I'm really excited by the square rocket. Hopefully the RSOs will be too!
 
One more thing on fluid dynamics. My office has two unofficial rules for hydrodynamics:

Don't surprise the water
Put it back nicely when you're done

They work pretty well for all kinds of fluids, although air can be more forgiving than water on the putting it back nicely front.

I think I'll be OK with the RSO. Regardless of the squareness, it's a long body tube with big fins so stability isn't that big an issue. I do have to make sure my coupler points are good and solid. If nothing else, I think the RSOs will want to see it fly!
 
Don't surprise the water?

I feel like there's a story there.
.
.
.
UNless this is some euphemism for flow separation
 
Last edited:
Sharp transitions = angry fluids?

Yeah, this is pretty much it. Sharp transitions, really blunt entrances or exits, or anything else where you make the water change direction robs efficiency. When you're burning a few hundred to a few thousand gallons of diesel a day, that turns into real money really quickly. The very worst thing you can do is drag a lot of transom (very stern of a boat) through the water. The next worst thing is big changes in width at the waterline. One of our clients had a boat that did that (basically, the front of the boat was replaced with a wider section, and there was a short straight-line transition between the wider new section and the narrower old section). They said it set up a vortex on each side that was so stable you could throw a paper cup in it as you left Seattle and it would still be there 4000 miles later arriving in Dutch Harbor, AK.

That said, there are some cool things you can do with setting up vortices on rockets. I've seen an egglofter style that had a square transition from the nose to a narrower body tube. The flow only reattached near the fins, dramatically reducing drag. Air is a little more forgiving than water, being much less dense and much less viscous.
 
That's a crazy vortex story. I wonder if it'd work for a guy on an inner tube.....

Any luck working out how to fabricate the couplers?
 
Okee dokay, Interesting project. I will recount a "triangular" rocket constructed out of a U.S.P.S triangular shipping tube I witnessed fly one time about 10 years ago. I was very new and green at this hobby as a starting out BAR.
Filer used a motor tube up the middle with triangular bulkheads. Three fins that were at the apices of the triangle at the aft end of the rocket along with a triangular nosecone. I don't know if the NC was built up or solid material.

It was a single motor deploy rocket that the flier stated had 12 prior nominal flights without problem and this time he was trying a "K" for the first time. Biggest motor he'd attempted. Button was pushed and the rocket did a disappearing
act except................ there was a puff of confetti after a second or two up pretty high. Looked like smoke but we could see chunks of the cardboard "fluttering" down along with the fins! Where's the rest of the rocket? Who knows we thought.

Well, we're laughing, picking up pieces of cardboard and some time later a yell goes out, "There's a chute!" Sure enough, the motor tube is coming in under the deployed chute with what looked like most of the triangular bulkheads intact!
This was before spool rockets made their appearance on the scene and I scratched my head and assumed since the "tube" was going so fast it just continued going where it was headed. Straight up.

When the spool rockets made an appearance, I surmised the intact bulkheads did something to stabilize the rocket and kept it going straight up. The "smoke screen" of the disintegration obscured visualizing the continued ascent of the tube. I also think the propellant was low smoke plus the fact the sucker really got up high. It seemed like a long time before the deployed chute and "remains" were seen. Nonetheless, the flier was happy to get their hardware back seeing as how little was invested in the bulk of the rocket. Kurt
 
Back
Top