Scalded Cat Steam Rocket Project Report

by William J. Inman, Reaction Research Society

Editor’s Note: This article was originally published in the March 2001 issue of RRS News, an RRS print magazine. It is reprinted here on June 20, 2020, for the RRS.ORG website with permission from the author and from the RRS. Copyright belongs to the author and the RRS.

Some of the products mentioned in the article are still available and links to the company website are provided solely for the reader’s convenience but does not constitute an endorsement of any product by the RRS.

William J. Inman’s Scalded Cat Steam Rocket Project


Water has the ability to hold and store a tremendous amount of energy in the form of heat. Unlike more conventional propellants that store their energy chemically, the steam rocket, or hot water rocket as it’s known, relies on the amount of heat stored in the water. Two other properties of water that make the steam rocket work so well are the vapor pressure developed as the water is heated beyond it’s “normal” boiling point and that when released it will expand to 1700 times the volume it occupied in the liquid state. It can be heated to 700 degrees Fahrenheit (at 3200 psi) before it reaches it’s critical point. Power increases with heat, but so does pressure, so the farther up the scale it goes the stronger the tank needs to be to withstand the pressure. Optimum performance is a balancing act between power of higher pressure and the weight of a stronger tank. Obviously, the tank should be made of the strongest, lightest weight, heat resistant material available… Titanium would undoubtedly be the ultimate if cost were no object.

Vapor pressure of water at increasing temperatures up to critical point of 705 F.

In the nozzle, the water starts flowing as it enters the convergent section. The venturi principle causes the local pressure to drop as velocity increases through the nozzle, and as pressure drops, the water starts flashing to steam. This steam, as it expands, continues to accelerate in the divergent section. The percentage of water that actually becomes steam depends on the amount of stored heat in the water. The temperature will drop all the way to the ambient boiling point at the nozzle exit, being converted to kinetic energy in the form of jet velocity. This velocity can exceed Mach 2 in a well-designed nozzle. As the water level in the tank drops, it boils, keeping the volume above it filled with steam, maintaining the equilibrium. This also consumes some of the heat in the tank, so the tank pressure will drop about 25% during the course of the discharge.


The “Scalded Cat” Motor

At the time I started this project, I knew much less about steam rocket theory than I do now. The motor was based on a piece of surplus 4-inch diameter, type 316 stainless steel, schedule 10 pipe that I found. Wall thickness was 0.120 inches and the burst pressure as stated by the supplier was 4000 psi. I got a pair of stainless steel domed end caps and had them welded on, then a hole bored in the center of one and a 1-inch threaded stainless steel pipe fitting welded in for the nozzle attachment. Three steel fin-mounting tabs were welded to the nozzle end of the tank and a flange for mounting the payload section was welded to the other end. Compared to the 45 pound welding oxygen cylinder I used for most of my static testing, this was a lightweight tank, but at 24 pounds, it’s still pretty heavy for a flight tank. To take advantage of its strength and to partially offset its weight, I ran it at higher pressures than most previous steam rockets that I read about. The flight on December 2, 2000, was at 1500 psi tank pressure (610 degrees Fahrenheit). Theoretical specific impulse (Isp) at this heat is around 75 lb-sec / lb.

Welded-on steel fin mounts with nozzle screws into the bottom tank head.

The nozzle was machined aluminum with a 3/8-inch throat; a figure I arrived at because I wanted a throat area of 0.110 square inches. There was a curved convergent section who’s curve radius was 12 times the throat diameter, then the divergent section had a half-angle of 10 degrees (20 degrees between the walls) and an expansion ratio of 18.3 to 1. This made the exit diameter 1.600 inches. The throat was 1/2-inch long to give a pair of O-rings on the plug a place to seat.

Scalded Cat steam rocket nozzle design

The fins were 0.085-inch thick aluminum and were bolted to the steel fin tabs at the bottom of the tank by running machine screws through the fins and screwing them into the threaded holes in the tabs. The fins extended beyond the back of the tank and also bolted to tabs on the fiberglass boat-tail to help secure it. The boat-tail also had a ring at the back end that slipped over the nozzle to keep it straight.

Boat tail with aluminum fins bolted into the welded steel fin attach points. Nozzle retention points at the nozzle.

The payload section mounting flange was a piece of stainless pipe 1/4-inch smaller in diameter than the tank and 0.030-inch thinner. It was 3-inch long with three semi-circular notches cut in one end leaving three “pedestals” that were welded to the top of the tank. This reduced the steel to steel contact and hopefully the heat transfer rate from the tank to the flange. A total of six holes, three sets of two, were drilled and tapped in this flange for the mounting of the payload section adapter.

Vehicle Specifications

Length = 7.5 feet

Diameter = 4.5 inches

Weight (filled) = 53.2 lbs.

Water capacity = 8.5 liters / 2.25 gallons (80% full)

Operating temperature = 610 degrees Fahrenheit

Tank pressure = 1500 psi

Calculated tank yield point = 1800 psi

Estimated peak thrust = 297 lbs.

Thrust duration = 4.75 seconds

Estimated power = 5500 Newton-seconds, “M 1155”

Propellant mass fraction = 35%

Parachute (tank) = PML, 84-inches

Parachute (payload) = PML, 54-inches

Electronics = Adept ALTS2 and Blacksky AltAcc2

Deployment charges = 3 (redundant)

Charge igniters = 4 (redundant)

Bridle (shock cord) = Kevlar “muletape”

Fins = 3 (aluminum)

Nozzle throat area = 0.110 square inches

Nozzle expansion ratio = 18.3

Divergent cone taper = 20 degrees between walls

The Payload Section Adapter

This part was used to provide a slip-fit mount for the composite payload section while helping isolate it from the heat. It bolted to the mounting flange with six machine screws and extended 6.5 inches up beyond the end of the flange so the area in contact with the payload section would not be touching a hot steel surface on the other side. I needed something strong, heat resistant, a poor heat conductor, and made of a material I could work with. The only epoxy I could find that claimed to be good to 600 degrees Fahrenheit was J-B Weld, so it was thinned with lacquer thinner and used as the laminating resin for a Kevlar structure.

The top end of the tank and the mounting flange and eye-bolt. The payload section is in front of the motor. Notches in the Kevlar ring are to minimize contact with the payload tabs.
Looking down into the electronics bay. The plywood ring with ends of two “T-nuts” are visible with silicone “form-a-gasket” on the inner edge. AltAcc is mounted to the airframe (left) while a brass charge canister is inserted into the gas baffle for the backup charge (right). When the lid is installed, the altimeter and the other two canisters fit in between this other equipment (a close fit!).

J-B Weld – Adhesive Products

A J-B Weld and Kevlar ring was epoxied to the outside as a stop to keep the bottom edge of the payload section 1.85 inches above the upper edge of the steel flange. A Kevlar and J-B Weld “floor” or bulkhead was added to put another heat barrier between the tank and the payload section. Cellulose insulation was stuffed into the area between the tank and bulkhead.

The Payload Section

For this section, I used an 18-inch length of 4-inch phenolic tubing from LOC Precision with several layers of fiberglass wrapped around it.

LOC Precision – Rocket Kits and Accessories

I was concerned about the heat from the tank damaging it so I added 2-inch of Kevlar and J-B Weld composite to the bottom where it was closest to the metal flange. The bottom 11-inches of the payload section was open and housed the 84-inch PML parachute. The Kevlar “muletape” shock line was attached to a 3/8-inch eye-bolt in the top of the tank. Above this section was a 1/2-inch plywood bulkhead that housed a black powder charge and expansion chamber / stainless steel gas baffle section. There were two igniters in this charge, one connected to the Adept ALTS2 altimeter and the other to the Blacksky AltAcc2 accelerometer. These were to be triggered by the “main” event switches on the two electronic devices to blow the chute out if the 54-inch pilot chute hadn’t already deployed it. I did this for a backup system in case the payload section got soft and sticky from the heat and didn’t slide off easily as planned.

The whole steam rocket assembly with nosecone, payload and the metal propellant tank to hold the water propellant load.

There was a compartment above the bulkhead where the altimeter and accelerometer were housed. The three canisters for the powder charges were also in this compartment, blowing their gases through the bulkheads into the gas baffles. The canisters were 1/2-inch brass pipe nipples with 3/8-inch plugs inserted in one end with pipe threads, sealed with Teflon tape. The igniter wires were inserted through holes in these plugs and sealed with 6-minute epoxy. The AltAcc was attached to the inner wall of the airframe in the usual manner and the ALTS2 was attached to a piece of aluminum box tubing that was epoxied to the removable lid of this compartment.

The compartment lid was also 1/2-inch plywood with a 3/8-inch eye-bolt attached to it’s center and a gas baffle compartment on each side of the eye-bolt. The underside of this lid had a ring of Permatex “blue” silicone form-a-gasket where it sealed to the mounting ring. There were also two threaded holes, one at the base of each gas baffle, for the brass change canisters to screw into. Four stainless steel #6 machine screws held the lid to the mounting ring, which was made of 1-inch plywood epoxied to the inner wall of the airframe tube. “T”-nuts embedded in this built-up ring distributed the load from the screws. On the 3/8-inch eye-bolt was the Kevlar “muletape” shock line to the 54-inch PML pilot chute.

The Nosecone and Parachute Arrangement

The nosecone was a 4-inch LOC Precision unit with a wrap of 1.8 ounce Kevlar on the inside of the neck to help reinforce it after the bottom had been removed to gain access to the interior space. A 3-foot length of Kevlar “muletape” was attached to the inside of the tip of the nosecone by having a loop go around an aluminum cross-rod inserted through holes on each side of the nosecone tip. This whole assembly was then encased in a solid mass of epoxy, then the cross-rod cut off flush with the outside surface of the nosecone. On the other end of this line was a loop sewn in with Kevlar thread from Edmund Scientific. The 15-foot main shock line and parachute shroud lines were all attached at this point. The main shock line had accordion folds sewn into it with Kevlar thread. The stitches were not heavy duty so they would break when a load was applied. The first six folds had a single stitch holding them, the second set of six folds had a double stitch, the third set had a triple stitch and the fourth had a quadruple stitch. The idea was that the singles would break first, letting out 3 inches of line out at each break and adding tension. Then the doubles would start breaking, increasing tension and still letting out 3 inches per fold. By the time all the stitches were broken (which they were), hopefully things would be slowed down enough to keep the final shock from being too severe. (Kevlar does not stretch.) The 25-foot line from the tank to the 84-inch parachute was stitched up in this same manner.

Accordion folds were sewn into “muletape” Kevlar bridle line. Breaking the stitches allows the line to lengthen while adding increased tension and slowing the rate of separation, reducing shock when the line pulls tight.
The “recovery module” was built onto a 1/2-inch plywood disc that also served as the electronics compartment lid. The two gas baffles on the left of the disc with an eye-bolt between them are reinforced with Kevlar. The two brass change canisters (the far one is behind the altimeter) can be seen with the igniter wires in their end plugs. The Adept altimeter is bolted to a piece of aluminum box tubing.
The first flight of the Scalded Cat used the nose cone tot the left.


The Launch Tower

Somehow I got the bright idea to build a tower with six longitudinal tubes of 1/2-inch electrical metal tubing (EMT). There would be one on each side of each of the three fins, just far enough apart to let the fin pass without binding. The reason was so I could pop rivet the three burner shrouds to these tubes, allowing each shroud to cover the entire tank surface between fins. “U” shaped strap steel brackets were welded to each set of tubes to hold them together and allow the fins to pass through. The three “U” brackets were held together by other pieces of steel strap welded to the outside corners, making a triangle shape at each of these brace points. The braces were spaced every 47-inches along the length of the 25-foot tower. For the real support, three lengths of 1-inch EMT were welded to the outside of the points of these “triangles”, also running the full length of the tower. In retrospect, diagonal cross braces should have been used and the second set of 1/2-inch tubes should have ended right above the tank where there was no longer any need for them. Anyway, it worked well enough. Three guy wires ran from the 12-foot point to anchors in the ground and another three ran from the 24-foot point to a second set of anchors 2 feet past the first set. Turnbuckles on all six ends made adjustment precise and easy.

The tower could be raised and lowered by pivoting on a stand which was a 3/4-inch galvanized pipe sitting in mounts on two 37-inch high welded steel “A-frames”. A flat attachment point was welded to two of the 1-inch EMT main supports and “U-bolts” went around the 3/4-inch pipe and through holes in the flats. To raise it, a couple of guys would get under the top end, raise it over their heads and start walking towards the base. After it was raised a certain amount, a third guy would start pulling a rope tied to the 12-foot point. A bolt on the bottom of the lower tower extension went through the base to hold it in position while the guy wires were being adjusted, and then help lock everything together.

The detached lower tower extension with the plug/release mechanism sitting in its notches. The release lever arm is sticking out in back. The fiberglass-covered styrofoam steam deflector is epoxied into the bottom. The hole in the flat bottom plate on the right is to bolt it to the plywood base.

The lower tower extension was a 17-inch piece of 7-inch diameter well casing with slots and access holes cut in it. A bottom plate was welded on for a place to bolt the plywood base, and three 3/8-inch headless bolts were welded to the upper end to bolt to the bottom of the tower. There was a fiberglass-covered styrofoam steam deflector in the bottom of this piece to direct the steam flow away from the electric actuators and the gas valve.

The Tower Base

This was what the tower sat on and what held all the peripheral ground support equipment. It was a 30-inch square piece of 3/4-inch thick plywood with two galvanized “Telespar” box sections bolted along the bottom of two opposite edges. These box sections were 36-inch long so they protruded 3-inches past each end of the plywood. Each of these four ends had a hole drilled in it to accept a 5/8-inch steel hold-down stake. The two welded “A-frame” tower supports bolted to the edges of the plywood base and had cross-bracing on the back side. A pipe coupler was welded to the top of each of these “A-frames” so the 3/4-inch tower support pipe could slide through.

A bracket to hold the release actuator was attached to one side of the tower and a bracket to hold the gas valve actuator was attached near the back of the lower tower extension. Then there was a third bracket to hold the clamp that secured the gas manifold near the back edge of the base. The plywood was thoroughly primed and painted to ward off the effects of the elements and the steam blast.

The Nozzle Plug / Release

Based loosely on the release mechanism designed by Bob Truax for his steam rockets in the 1950’s and 1960’s, this multi-talented device serves several purposes. First, it plugs the nozzle throat so no water or steam will escape before it’s released. Second, it provides a connection to the pressure gauge so it can be monitored during heating. Third, it has the integral clamping system that holds the rocket on the plug until released, and (provide) the means of releasing it.

The central plug is machined out of steel and has a long narrow taper to the 3/8-inch tip that goes into the nozzle throat. This tip is 0.60 inches long and has two O-ring grooves to accept Parker fluorocarbon or “Viton” O-rings to make the seal. The part below the taper is threaded with 7/8-inch bolt threads. A hole is drilled through the center to provide access to the tank pressure.

The plug/release mechanism locked onto the nozzle (boat-tail removed). The brass fitting connects to a steel brake line to the pressure gauge. Rotating the release cam (holding the red bars out) allows the bars to pivot off the nozzle.

The bottom end of the plug is drilled and threaded to accept a 1/8-inch brass pipe fitting. This fitting is an adapter that allows a 5/16-inch automotive steel brake line to be used to connect the pressure gauge, which sits on the tower’s 3/4-inch support pipe on the end facing the blockhouse.

A support structure with three “spokes” is built around a 7/8-inch nut that screws onto the plug. The “spokes” are steel box tubing long enough to reach past the wall of the lower tower extension and sit in the bottom of three dedicated notches in the extension. Each of the spokes has a rectangular hole cut in it’s top and bottom to allow a smaller piece of square steel bar to pass through. This bar is pinned to the spoke by a 1/4-inch bolt running through it crossways, allowing it to pivot. When the three bars are brought together at the top, they contact the tapered outer walls of the nozzle like fingers.

Below the structure with the spokes and bars is a cam plate made from a round piece of 1/8″ steel sheet welded to a bored-out 7/8-inch nut that slips onto the plug. Three equally-spaced half-round notches are cut into the edge of this plate. The spacing between the plate or cam and the “spokes” structure is adjusted with washers between the two. When adjusted correctly, the “cam” edges of the plate will hold the bottom edges of the three bars out at a distance that positions the other end of the bars so they hold the nozzle firmly onto the plug, with the O-rings seated in the throat by “gripping” the tapered outer walls like fingers holding a knob. Rotating this cam by pulling on an attached lever arm with a 12-volt DC electric linear actuator allows the bottoms of the three bars, or fingers, to fall into the three notches, pivoting around the 1/4-inch bolts and releasing the nozzle from it’s grip. A 7/8-inch “keeper nut” with a nylon insert is screwed onto the plug below the cam and give it something to ride on and keep the spacing so it turns freely but doesn’t have excess play.

The Burners and Gas Delivery System

At the bottom of the tower are three sheet metal burner shrouds that are as long as the tank (48 inches). In the bottom of each of these shrouds is a 30,000 BTU propane gas log lighter for a fireplace attached by two “U-bolts”. There are adapters for flexible appliance gas lines on each burner to attach to the manifold. The manifold is a 1/2-inch pipe nipple and “L’s” on each side, creating three points to attach the flex-lines. A clamp with three notches fits over these three lines at the manifold, holding it to the plywood base. On the other end of the feed nipple is a brass ball valve with a union on the other end. The rubber hose from the propane bottle is connected to the manifold at this union.

Launch tower erected with the power cables running back to the blockhouse, propane bottle to the right feeding the burners

Attached to the ball valve is an aluminum extension that is painted bright red so the valve position can be determined visually from the blockhouse. Also attached to the valve handle is the end of a 24-volt DC electric linear actuator attached to the control panel in the blockhouse. This actuator is used to open and close the gas supply to the main burners.

Electric linear actuators used in the launch process.
4.5-inch pressure gauge with a red plate behind it for easier visibility.
Sheet metail placed around the three burner shrouds for better heat retention. Propane torch igniters seen at the bottom. Steam pressure gauge is reading tank pressure.

Three small handheld propane torches are positioned around the base of the tower pointed up into the shroud burner areas. These act as pilot lights for the main burners should they need to be turned off and then back on again. They also add additional BTU’s to the heating effort but don’t put out enough to maintain heat (and pressure) by themselves.

The Control Devices and Panel

Pressure is monitored visually by watching a 4.50-inch diameter pressure gauge with binoculars from the blockhouse. Heating is controlled by a gas valve in the line to the main burners. A 24-volt DC linear actuator is attached to the handle of the gas valve and opens and closes it by pushing and pulling. It is wired to a double pole-double throw (DPDT) toggle switch on the control panel so that pushing it one direction opens the valve and pushing it the other direction closes it. It is a three-position momentary switch so releasing it allows it to spring back to the center “off” position. The power comes from two 12-volt batteries wired in series in the box. The control panel is actually the lid of the battery box.

The directional control switch and “launch button” are located under the safety flap on the control panel. Both the 24-volt system cord to the gas valve actuator and the 12-volt system cord to the launch actuator extending from the right side of the box.

Launch is initiated with another electric actuator, this one a 12-volt DC unit, also wired through a DPDT toggle switch in the battery box. Three 12-volt batteries wired in parallel power this actuator. One lead goes through a momentary red pushbutton switch wired in series with the DPDT switch. The DPDT is a two position, one for extend, the other for retract. This allows the cam to be rotated back to the “reset” position easily, which is good because we had to move it back and forth once to release the rocket for it’s maiden flight. The red “launch” pushbutton and the DPDT toggle switch controlling the direction of the release actuator are both under a spring-loaded safety flap made from an outside electrical box outlet cover.

Another view of the control panel with cable feed-through glands. Use of 12-gauge extension cords works well.

To connect the control box to the actuators at the pad, color-coded 12-gauge extension cord is used. Two 25-foot cords were bought, one yellow, the other blue with an orange stripe. Yellow is for the 24-volt gas valve actuator while blue-and-orange is for the 12-volt release actuator. These 25-foot cords were cut a few feet from the “female” end and attached to their respective switches in the box with the ends dangling outside a foot or so. The other long piece was wired to the actuators with the “male” end like a regular power cord on any appliance or power tool. Two 100-foot cords with the same color coding bridged the distance from the blockhouse to the pad.


Setup and Preparation

The tower base already leveled and staked down on launch day and the tower was waiting nearby. The guy wire anchors were driven in at the pre-determined and marked spots and the peripherals were all attached to the base and tower. The igniters and deployment charges were already set up earlier in the motel room so once it was time to start the heating, the altimeter and AltAcc were turned on. After the tower was raised vertical and secured, the burners were lit and the AltAcc armed. We did not time the heating, but it went fairly quickly once a piece of sheet metal was wrapped around the tower at the position of the heaters. It was carefully bent so the fins would pass inside it during launch. When the pressure reached 1400 psi and then launch hopefully at a point where it had dropped to 1350 psi, the target pressure. Instead, the pressure continued to climb to 1500 psi, where it stayed until launch.

Bill Inman makes a minor adjustment to the fins of his steam rocket, the Scalded Cat
“Team Steam” shortly before the launch. From left to right: Jeanne Hoover, Bill Inman and Tim Clifford.

The Flight

When the release actuator was retracted, nothing happened. This had happened before, but when checked again during my last static test, it worked fine. Here we were sitting at 1500 psi with the release cam turned and the rocket just sitting there. So I had Tim Clifford, my partner and launch officer, switch the directional control to “reset”, work the actuator, then flip it back to “launch” and try it again. This time, after a couple seconds of hesitation, it took off on the most beautiful plume of steam I’ve ever seen. From the blockhouse it is not possible to visually follow a rocket very far into it’s flight through the small windows, so we just stood there listening to the roar as the sound came from farther and farther away. Finally, it stopped and for a brief moment there was no sound, until there was some cheering from the bunkers. The command was given over the P-A system to “quiet down”, and to “listen for an impact”. A few seconds later there was cheering again, and this time a more irritated repeat command was given only to be answered by shouts. “What was that?” … “A paraachute?” … “Two parachutes?” … “O.K. Keep an eye on it and stay under cover until the heavy piece is down.”

Launch of the Scalded Cat at approximately 2:30pm, December 2, 2000. Not the clearest view of the launch from behind the weathered blast windows of the blockhouse.
Another view of the steam rocket launch, photo courtesy of Richard Butterfield

Knowing it was under canopy was the best feeling of all. I have seen so many rockets crash because of recovery system failure that it makes that part especially critical. There was also the satisfaction of knowing that along with being the first successful steam rocket launch in the 57-year history of the RRS (at that time in the year 2000), it was also going to be one of the very few RRS flights to make a soft landing under parachute. I was able to squeeze out through the blockhouse door enough to actually see the parachutes coming down in the southwestern sky, the tank falling slightly faster than the payload section.

Parachute recovery was successful
Jeanne Hoover stands over the safely recovered motor.

Post-Recovery Examination

The only damage found was where the ring at the base of the boattail got broken in two spots from being driven into the ground from the weight of the tank. Otherwise, everything was all right and the altimeter was reporting 4,479 feet. That evening, Bill Seiders was kind enough to download the AltAcc on his computer. It showed a maximum acceleration of 128 feet per second (4 G’s) to a velocity of 506 feet per second (345 MPH), a coast time of 15 seconds, and a peak altitude of 4,400 feet.

Blast pattern after the steam rocket launch. Very little equipment damage after the launch.

Editor’s post-script: Bill Inman has decided to rejoin the RRS after being away for many years. We enjoyed talking with him at our virtual meeting on June 12, 2020. He spoke by teleconference as we are still unable to hold our meetings in person at the Ken Nakaoka Community Center in Gardena, California, due to the COVID-19 restrictions from the city of Los Angeles. Bill has decided to start a new steam rocket build based on the many lessons learned over the years and we hope he’ll teach some of us how to make this unique form of rocket fly.

Launching Rockets – Adventures at Spaceport America

by RRS member, Joseph Maydell

It was the Saturday of Memorial Day weekend. My flight was experiencing rough turbulence as it flew over the mountains on final approach to El Paso, TX. I was traveling to Spaceport America as a sponsor on four upcoming space-shot attempts. After collecting my luggage, I picked up my rental truck and headed north on the two-hour drive to Spaceport America. The only other way to access Spaceport America is to fly into Albuquerque and make the three-hour drive south. I had decided to fly into El Paso to save some time. Texas had actually been my home ten years earlier while working at NASA’s Johnson Space Center as an International Space Station (ISS) flight controller.

Getting through security earlier that day had been an adventure. My carry-on only contained mission critical hardware and was flagged for inspection. Everyone in the security line stared as TSA agents pulled antennas, circuit boards, a soldering iron, hot air rework station, trays of SMT (surface mount) components, wiring, ground control units, and weather balloon inflation equipment out of my carry-on. Everything was thoroughly swabbed for explosive residue and a lot of questions were asked.

The reason I had been asked to sponsor the next four launches at Spaceport America was because I had led the development of a new set of avionics for professional rocketry. It consists of a flight computer called the Eagle and a handheld ground control station. It was developed as part of a program for safely launching and recovering rockoons. It has the ability to launch, stage, and recover a multi-stage rocket as well as other proprietary features unique to rockoon flight. It has a very accurate barometric sensor and an aviation-grade inertial measurement unit (IMU). However, what space-shot teams find especially appealing is the global positioning system (GPS) receiver that can obtain GPS lock at any altitude.

The first launch was set for Monday morning at 6:00 AM. The rocket was a two-stage rocket built by Coleman Merchant from Princeton University as part of his master’s thesis.

It had the energy and propellant mass fraction to easily pierce the von Karman line (100 km of altitude). A group of cadets from West Point were also on site to assemble and align the launch rail on loan from Kevin Sagis, Virgin Orbit’s chief engineer. My responsibility, in addition to monitoring the health of the Eagle avionics package, was to launch weather balloons in the hours leading up to the rocket launch. This was critical for obtaining the upper level winds for calculating the firing solution. In the coming week, I would be launching a new type of radiosonde I had developed that would help lower the cost of obtaining upper level wind analysis prior to rocket flights.

Two rockets made by Coleman Merchant at the Princeton Rocket Laboratory

At 3:30 AM Monday morning, the team assembled for final launch preparations. Radiosonde operations were going well. Preparations at the launch site were also progressing smoothly. However, there were concerns that the brackets used to bolt the 1010 launch rail to the main launch rail structure could make contact with the carbon fiber fins on the booster during launch. This hadn’t been apparent earlier since the launch rail was still being prepared the previous day. The decision was wisely made to delay the launch. An hour later, all the brackets had been trimmed using a hacksaw.

Launch preparations would typically begin at 3:30AM

After aligning the launch rail with the final firing solution obtained from my radiosonde data, the rocket was armed. We all moved to Mission Control to complete final checks. This is when we discovered another technical issue. Since so many electronics, transmitting at various frequencies, were crammed into the nosecone, and since the nosecone was in such close proximity to the large launch rail structure, it was taking longer for the electronics to obtain GPS lock. We had done a radio-frequency (RF) test of the avionics package with all electronics running the previous day, including GPS lock testing, but not on the launch rail since it was still being assembled. It took about five minutes, but eventually all electronics with GPS receivers had GPS lock. After getting a “go” from White Sands and Spaceport America, the final countdown resumed and Chase Lewis, the West Point pyro-lead, sent the signal that launched Coleman’s rocket.

Coleman’s rocket accelerating off the pad at 46 G’s

Even from a mile away, it was difficult for the eye to catch it as it accelerated off on its way to space. Acceleration during boost reached 46 g. At booster burnout, the rocket was traveling Mach 2.4. A charge fired which separated the sustainer from the booster. A few seconds later, the sustainer engine fired and the sustainer once again experienced a peak acceleration of 46 g along its X (vertical) axis. However, as speeds approached Mach 3.8, the rocket became unstable and began to fly in a large upward spiral. Acceleration on both the Y and Z axis, which should ideally be zero, hit 42 g. Somehow the rocket managed to hold together before exiting the earth’s atmosphere at which point all acceleration loads went to zero. A few minutes later, the rocket re-entered the earths’ atmosphere under drogue. The booster had landed much earlier. It didn’t have any electronics and its recovery method was ballistic.

The booster recovery method was simple ballistics. Someone stumbled upon it a couple of weeks later.

After analyzing the data recorded by my avionics system (there were two altimeters by a different vendor but we couldn’t access the data), the leading theory for the upward spiral was inertial roll coupling. This is an aerodynamic phenomenon that can happen to both rockets and high-speed aircraft at a critical roll rate. Symptoms include divergence of angle of attack, large side-slip angle, and violent accelerations and loads. Air-frames with a low roll moment of inertia are particularly prone.

We still had one more launch window the following day. The decision was made to launch the second rocket and see if the problem repeated itself. No two rockets have the same roll rate due to tolerances in fin can and nozzle manufacturing processes. The hope was that the second stage would either stay below or above the critical roll rate during sustainer engine burn.

The launch window for Tuesday had also been scheduled from 6:00 to 10:00 AM. However, White Sands Missile Range informed us shortly before 6:00 AM that our launch window would close at 6:35 AM. This was unexpected and placed a lot of pressure on the team as we prepared the second rocket for launch. The rocket was armed just a few minutes before the launch window would close and we didn’t have time to allow the electronics to acquire GPS lock. The decision was made to launch with the hope GPS lock would be acquired during flight away from the interference caused by the launch rail. The booster once again flew flawlessly, but the sustainer never ignited. It coasted up to 19.7 km (64,600 ft) before coming back down under drogue. None of the electronics obtained GPS lock during the flight. The chance of us ever finding the sustainer and determining why its engine never ignited seemed unlikely. In the distance, we watched a missile soar into space over White Sands Missile Range. Now we knew why our launch window had been cut short.

I knew my avionics system had line-of-sight range, so in theory, as long and I could get my hand-held ground station high enough above the terrain, I would be able to receive telemetry. One idea I had was to mount my handheld ground unit to the top of the launch rail. We lowered the hydraulically-actuated launch rail and taped my ground control unit to the tip before raising it back up. The ground control unit was now sitting 40 feet above the desert. We lowered it a few minutes later and were disappointed to see that the ground controller had not logged any telemetry packets. This meant the rocket had to be in a gully or valley at a distance greater than a few miles. The next idea I had was to drive back and forth across Spaceport America along the expected flight path. I knew that if I came within a mile or two of Coleman’s rocket, I would receive packets and we could then locate the rocket. After driving for about an hour down some very rough roads, my ground controller started to log packets. An hour later we all hiked out to the sustainer which was lying in a valley. The sustainer was very close to where the booster had actually been targeted to impact as calculated by the upper wind analysis and firing angle solution.

On the drive back from the recovery area, I got a flat tire from an old fence-post nail. I tried to speed up through the cloud of dust from the truck in front of me to flag for help, but once my rim was hitting the ground I had to stop. I could have been out there for hours by myself if I hadn’t been able to break the lug nuts free with the inadequately short tire wrench I found under the truck’s passenger seat. Fortunately, I did make sure I had plenty of water, snacks, first aid kit etc., before heading out to try to find the sustainer.

So what went wrong on Coleman’s second space-shot attempt? It appears both altimeters rebooted when they fired the booster/sustainer separation charges. Because they were both rebooting, neither one fired the sustainer igniter. Since Coleman had only reached out to me two weeks earlier about integrating my avionics package into his rocket, my system hadn’t been approved by Spaceport America for initiating any flight events on his rocket. All it could do was go along for the ride while saving and transmitting flight data.

Coleman’s rockets had both flown amazingly well. The first space shot had come amazingly close to space. You could tell that a lot of experience and engineering analysis went into the design of his two rockets. I asked Coleman what he enjoyed most about the project:

“After spending months, coming out with a really nice final product that you are really proud of. Everything on this came out exactly the way I wanted it to. I don’t really have any regrets about how it was made.”

They truly were both impressive rockets. I asked Coleman what his biggest takeaway was:

“Spend more time on the electronics than you think you should. Don’t leave it until the last minute. It’s almost the most important part of the rocket. It’s something a lot of teams get wrong. They’re so focused on making sure it won’t rip apart.”

As an avionics systems developer, I couldn’t agree more. Coleman flew home and I had to start preparing for the next two space-shot attempts with Operation Space.

Operation Space was a project started by 18-year-old Joshua Farahzad. It was collaboration of students from multiple universities that had joined forces through the internet to design and build a space-capable two-stage launch vehicle. They had reached out to me a few months earlier about sponsoring their space-shot attempt and flying my avionics package into space on their rocket. I saw it as an opportunity to get additional testing and data on the Eagle system. Test it they did, in ways I could have never imagined!

The first launch attempt was scheduled for Thursday morning at 6:00 AM. However, assembly of the first rocket wasn’t completed until late Thursday afternoon. Parts designed and manufactured in different parts of the country didn’t fit together the way they were expected to fit. Last minute modifications were required including additional machining of fins and other critical components. The avionics bay was completely redesigned on Wednesday and rebuilt on Thursday. The first deployment test didn’t occur until Thursday evening.

Friday morning, after 48 hours of round-the-clock work, the first rocket was finally on the launch rail. Chase once again sent the signal that ignited the first stage. Everything went well until the sustainer engine ignited. It was obvious from the smoke trail that the sustainer had gone completely unstable. Once it landed, we lost all communication. Our search in the desert for the sustainer at the last received GPS coordinates proved futile. At the time the leading theory was that the sustainer had lost one of its fins.

The second rocket was launched Saturday morning. Its flight path also went unstable about two seconds after sustainer ignition. It also abruptly stopped transmitting all data once it landed. Once again, we went out to the last received GPS location. We never found the sustainer. However, to our surprise we did find the avionics bay with a short length of parachute tether and a wad of carbon fibers from the nose cone. When it hit the ground the battery tray inside broke loose and crushed my avionics system. Most of the SMT components had popped off the motherboard. Fortunately, the avionics bay was in a clearing only a few feet from where I had received the last packet during flight. Otherwise, we may have never found it since there was a lot of thick brush and we were all looking for a large rocket. We could have easily overlooked the small avionics bay hidden in a thicket. This is probably what had happened when we searched for the first sustainer the previous day. We had been warned not to poke around in the bushes because of the rattle snakes. We hadn’t considered looking for something as small as an avionics bay.

Chase Lewis inspecting the interstage of the Operation Space rocket.

Once we returned to Mission Control, I was able to solder the SMT memory chip to a good Eagle motherboard using my hot air rework station. This made it possible to download the flight data. This is what the flight data revealed: two seconds after sustainer engine ignition, the rocket started to go unstable and then it drastically altered its angle of attack. One tenth of a second later, the avionics bay separated from the rest of the rocket. It did a 180-degree turn and coasted backwards to an altitude of 15.5 km (51,000 ft) with the parachute tether trailing behind it before coming back down. Most likely, aerodynamic loads at Mach 3.5 caused the carbon fiber nosecone to fail. This released the drogue which was housed inside the nosecone. The force of the drogue opening and shredding broke the altimeter bay free from the rest of the rocket. Later, I learned that the nosecones had a major manufacturing defect. There wasn’t enough time to manufacture new nosecones and those who knew about the issue had hoped for the best.

The avionics bay from the second Operation Space launch as we found it. Notice the black wad of carbon fiber presumably from the nosecone shoulder.

The Operation Space Team put in a lot of effort to reach space. It was disappointing to see them only reach 15.5 km. However, I have no doubt that with more experience, an improved design, and better preparation, they can be successful. They had a lot of fun, worked well together, and certainly learned many lessons. One in particular that I would like to emphasize:

You should never underestimate the amount of time, effort, and diligence required for successful space flight. Among other things it requires thorough engineering analysis, diligent acceptance testing of all manufactured parts, exhaustive vehicle integration testing, and well-written operation procedures.

It was now Saturday afternoon. After downloading the flight data, I left Spaceport America with just enough time to drive back to El Paso and catch my flight. I only had one concern. With all the work helping Operation Space machine, wire, assemble, test, and prep their two rockets, I never did get my flat tire fixed. I was on my cellphone telling my wife how excited I was to see her and the kids that evening when a warning light went off. My adventures were not over: I had another flat tire!

About the author

Joseph Maydell has over a decade of both space flight and high-altitude ballooning experience. He is a former ISS Flight Controller and NASA spacecraft systems instructor. He has started multiple successful aerospace businesses and is passionate about inspiring students to pursue careers in space exploration. If you have any questions or comments, you can reach me here.

Apogee Aeronautics, SuperLoki launch rail system

On November 14, Osvaldo and I had the chance to meet Robert Kleinberger of Apogee Aeronautics. Robert runs a small engineering company specializing in sounding rockets. We had a chance to see one of the SuperLoki launcher systems that he had assembled and mounted on a trailer system in Orange County.

SuperLoki launcher from Apogee Aeronautics

view of the control panel, SuperLoki launch trailer

The SuperLoki is a sounding rocket design that was started in the late 1960’s and were used in the US and other western countries up until the supply was exhausted in the 1990’s.

In short, this two-stage rocket system consists of a first-stage solid propellant booster with an instrumented “dart” that coasts up to high altitudes, typically to make upper atmospheric weather measurements. The four-fin booster and four-fin dart are rail-guided to impart a controlled spin for stabilization.

Super Loki PWN-10A model

SuperLoki launch rails, side view on rails has specific details on the SuperLoki launch vehicle – SuperLoki

Wikipedia has a fuller article on the subject of the Loki series of rockets including the Super Loki.
Loki rocket – Wikipedia

Robert was assembling one of his SuperLoki rail launchers for a client, but ultimately, the launch trailer will reside at the Chico Air Museum in Chico north of Sacramento, California. The launcher will be there on display to the public. Although the SuperLoki launch trailer is not yet at the museum, I have put a weblink to the museum’s page.

Chico Air Museum

Although he may not make his presentation personally, Robert has put together a series of notes for our upcoming monthly RRS meeting. We’re looking forward to January 13th. Hope to see everyone there.