Monday, November 30, 2009

Aerodynamics

The MLAS aero team invested considerable effort in the development of a database suitable for predicting vehicle flight performance. Data collected in wind tunnel tests and through Computational Fluid Dynamics (CFD) analysis provided the insight required by the broader team to predict structural loads, size fins and drag plates, position the CG, and do pre-flight trajectory analyses. The location of vehicle Center of Pressure (Cp) -- the piece of aero data of most interest to modelers concerned with longitudinal stability -- was established through this work.

Cp is not measured directly but is calculated from the pitching moment and normal force acting on a vehicle in flight. We used an 8% scale model of the MLAS flight vehicle in the Calspan 8-foot Transonic Wind Tunnel to collect force and moment data. A model component build-up approach was used in testing which allowed us to derive the longitudinal and lateral forces and moments contributed by each model component. The wind tunnel model was itself a work of art: a precision instrument, milled from solid aluminum blocks, with stainless steel fins. This replica of the flight vehicle could be assembled and oriented as desired to test each of the three separate flight configurations: boost, coast and reorientation. A clever system of access hatches and a rotatable sting mount allowed the reorientation configuration to be tested at all pitch angles between 0 and 180 degrees. The model is shown here mounted on the Calspan tunnel's test apparatus. Data gathered in the tunnel was nearly identical to that predicted by CFD and ultimately used to fix position of the MLAS Cp in the boost configuration and zero angle of attack at X=180.5 in ... slightly forward of the Coast Skirt's center.

Scaling for µMLAS puts the tunnel predicted CP 10.36" from the tip of the model's nose. Interestingly, RockSim's stability equations put it at 10.28" ... very close, and well within the margin of error for the tunnel predictions. Of course, this could just be coincidental -- one data point is hardly enough to validate a piece of software -- but it gave me a warm fuzzy feeling about RockSim. And at $120, RockSim was significantly less expensive than those wind tunnel tests.

Friday, November 27, 2009

Coordinates

Another pause for the mundane but necessary: coordinates.

Coordinate axes provide a convenient method for communicating the orientation of a flight vehicle and describing the location of key features. We followed traditional missile/aircraft conventions when establishing the MLAS coordinate system. Visualize a set of 3-dimensional coordinates with the X-axis running vertically, "Y" 90-degrees to that, horizontally, and "Z" coming out of the page. The MLAS X-axis is the longitudinal axis of the vehicle .... the -X direction is towards the nose and the +X direction is towards the tail. "Y" and "Z" are a little more arbitrary. We based them on the vehicle orientation at the pad .... an observer facing the ship as shown to the left would be looking straight down the Z-axis in the -Z direction with the +Z axis coming directly towards him. In other words, the -Z side of the vehicle would face the ocean and +Z the land. The -Y-axis would be to the observer's left and +Y to the right.

The "zero" reference depends on the axis. We set Y-zero and Z-zero at the vehicle centerline, where those axes intersect X. We based the X-zero reference on the Crew Module apex: if you extend lines down the sides of the CM, the point at the top where they intersect is X-zero. The rest of the ship is referenced from there. That's why the tip of the nose is at X minus 26.8" and why you have to add 26.8" to all dimensions if you wish to measure from the Forward Fairing nose cap instead of from Xo as shown in the configuration drawings. We labeled directions from the point of view of an observer looking up from below, the ocean at his feet. The -Z (ocean) side is thus the "bottom;" +Z (land) is the "top;" +Y (the observer's right) is "right;" and -Y (the observer's left) is "left." This all looks odd on drawings done from the perspective of one looking down from the nose of the vehicle, but makes perfect sense to one working on the hardware from underneath.

The roll patterns on the fairing and CM correspond to the vehicle axes. The +Z face is the one with two black bars as shown, -Z (the ocean side) has a single long black bar, +Y (right per our convention and from the point of view of the picture above) has a single short black bar, and -Y (left per convention) a pattern of three bars.

You'll find all this in the configuration drawings ... I carried it through to µMLAS with the added twist of angles numbered in a counterclockwise direction looking up from the bottom with -Z at 0°, -Y at 90°, and so forth. Naturally, that's differs from the MLAS convention which calls that angle omega and sets 0 at the +Z axis. The confusion just adds to the challenge.

Wednesday, November 18, 2009

Motor Mount

While the MLAS MK-70 motors were mounted in a spidery-looking cage, the mount for the µMLAS model is a more conventional 3 x 29mm cluster centered in a 3" phenolic airframe. The centering rings came from Giant Leap and the tubing from PML. The tight cluster didn't leave much room for centering ring web or epoxy fillets to carry the thrust of any individual motor, so I added a single wrap of fiberglass to the motor tubes to transfer loads between them. In the original design, the forward end of the assembled motor mount extended beyond the front of the Boost Skirt through a fiberglass sleeve in the Coast Skirt centering rings to provide support to the base of the CM and keep it from dropping out of the Forward Fairing in flight. Unfortunately, this arrangement also provided a place for the Coast Skirt to bind at separation. I shortened the motor tube after-the-fact and epoxied the section of 3" tubing removed from it into the Coast Skirt sleeve to support the CM. Drawings in the µMLAS package are of the updated design.

Motor retention is provided by a screw and washer centered between the three motors. The screw threads into a brass insert embedded in Epoxy Clay in the cavity between the motors.

The real MLAS' motor nozzles were just visible beneath the edge of the Boost Skirt, so I added four dummy nozzles to the model's aft centering ring to duplicate the look. I cut these from pieces of McMaster-Carr 1" OD, thick-walled phenolic tubing with Estes BT-20 "crayon" nose cones inserted to duplicate the motor nozzles. I could have omitted that last detail, I suppose, but details like that are half the fun of scale models. And besides, I had plenty of weight margin.

So I thought, anyway.

Sunday, November 15, 2009

Progress

A pause from the MLAS story ....

While at KSC for the STS-129 launch, I snagged this shot of the Ares Launch Umbilical Tower (LUT) being constructed atop its Mobile Launch Platform. It has a "back to the future" sort of look to it, reminiscent of the Saturn LUT. Thinking of that prompted me to pay a visit to the past and drive over to the Cape Canaveral Air Force Station (CCAFS) Merc/Gemini/Apollo pads too - what's left of them, anyway. Those shots are posted in Photo Collections.

It's just as hard to get off the planet now as it was 50 years ago: Vo still equals sqrt(GM/r).

Wednesday, November 11, 2009

Internal Structure

The real MLAS flight vehicle wasn't much more than a fiberglass shell. The MK-70 motor cluster was mounted in a frail-looking "cage" (shown at left with two of four inert motor casings installed) that served only to keep the motors pointed in the right direction. It was mounted inside the Boost Skirt with a series of struts connected to the shell with turnbuckles. Adjustments to the turnbuckles allowed the motors to be aligned with respect to one another and the centerline of the vehicle. They also provided a means for pre-loading the forward end of the cage against the bottom of the CM simulator, so the 280,000 pounds of thrust generated by the motors would be transferred directly to the CM. The CM was itself connected to the Forward Fairing (FF) with four "Y-Fittings" (so-called because of their shape) and held in place with frangible nuts. The Coast and Boost Skirts were connected to one another and the FF with frangible joints. In flight, the motors applied thrust to the base of the CM and propelled it forward; it then carried the Fairing and Skirts along with it. The fins were installed through slots in the Skirt shells and bolted to brackets inside. Aero loads (drag and the lift of the fins) were taken by the shell.

The only other significant internal structure was at the top of the Forward Fairing where the turnaround drogue chutes were connected. After the motors burned out and the frangible joints fired to separate the Boost and Coast Skirts, the drogues were deployed to re-orient and stabilize the Fairing so the CM could be released. Chute loads were taken through an eight-sided "Octagon Fitting" and some support members installed just below the nose cap and transferred through the Fairing's motor troughs to the CM Y-Fittings. When the Y-Fitting frangible nuts were fired, the CM dropped free of the Fairing.

I briefly considered mounting the µMLAS motors in some sort of open structure similar to the MLAS motor cage - emphasis on "briefly." Thrust vector alignment would have been just as critical as it was to MLAS and without benefit of the laser metrology used on the real ship, much less certain. And I doubted I could build anything beefy enough to take motor loads without self-disassembling. So I built µMLAS with traditional centering rings cut from 3/16" and 1/4" plywood. I did the cuts with a DeWalt DW660 cutout tool, then turned the stack of rings on the drill press and sanded them to fit the tube sections. The Coast and Boost Skirts each have two rings - the picture to the right is of the ring stack on the drill press taken before the center holes were finished with a sanding drum.

I've started posting additional pics both of the real MLAS and the model under "Photo Collections." When you're browsing the MLAS shots, keep in mind what we were trying to do: this was a quick-turnaround, low-cost concept demonstrator. You won't see clean rooms, bunny-suited technicians, and precisely machined handling fixtures here - we didn't need them. In many ways, MLAS was just a big model rocket. Emphasis this time on "big."

Monday, November 9, 2009

Airframe Woes

Cutting the airframe and filling spirals proved to be more difficult than I expected. I had cut 6" tubing on my table saw in the past but was never happy with the results. Turning the tube seemed like a reasonable option - I thought I could make clean cuts that way and quickly dispense with the spirals. The 10" µMLAS airframe was too big for my lathe, so I built a mandrel to turn it vertically in my drill press using the lathe's drive spur and a dead center sold by Penn State Industries mounted to the drill press table. At right you can see the entire assembly on the drill press as it appeared just before I made the cuts. This worked, but constructing, balancing, and trimming the mandrel turned out to be a time-consuming chore of its own and the final tubing cuts were not as clean as I'd hoped. I did get the spirals filled and sanded, but if I had the whole thing to do again I'd probably do the cuts by the old "tape and razor saw" method and fill the spirals by hand. Live and learn.

The real MLAS airframe proved to be a bigger challenge than expected too. The test article was not significantly weight-constrained, so we opted to build the fairing and skirts from a simple fiberglass-over-foam composite manufactured by Northrop-Grumman at their Gulport, Mississippi, shipyard. Balsa was used for the core instead of foam in locations where loading dictated (seemed strange to build real flight hardware out of traditional modeling materials - foam and balsa - but there we were). The fairing and skirts were manufactured in sections, shipped cross-country by truck, and assembled at Wallops. Our excitement on taking delivery of the first sections was dampened when we took some core samples to test the strength of the composite layup: the first cuts reeked of uncured epoxy. As it turned out, the adhesive used to layup some of the sections reacted adversely with a pre-applied adhesive in the cloth and the composite didn't cure properly. The whole episode cost us several months while Northrop manufactured new parts, and provided some lessons in "unintended consequences" ... the professional version of live and learn.