STIG B-I Mission Report
STIG B-I Mission, Saturday October 6th 2012
Update by Neil Milburn, posted 02/22/2013
If you are having problems viewing photos in Internet Explorer, it appears to be resolved by updating to the IE10 Beta version. Firefox, Chrome and Safari all worked without issues.
This was our first licensed flight ever, all others has been under permit or a waiver of some description, even the three minute long Lunar Lander Challenge Level II flights. It had been a long time coming; our last mission was STIG A-II back on February 22nd, almost nine months ago. Building an entirely new vehicle from scratch in nine months would normally be considered stellar performance, but for us it was an eternity. This flight was significant on an industry front in that this was the first ever FAA/AST licensed commercial flight of a reusable launch vehicle!
We opted for a three day on the pad protocol being a virgin flight (sorry Sir Richard) and our first licensed flight. Given the distance from our home base in TX to Spaceport America, roughly 750-850 miles depending on southern or northern entrance, the crane truck with the GSE has to set off four days before launch day and the rocket trailer and crew van three days before. The commercially regulated crane truck cant quite make the distance legally so Joseph usually camps up in glamorous Van Horn, TX. The rocket trailer and crew van can make it in one long day.
One other logistics problem that has now been resolved is getting through the NM port of entry just across the TX border. Some may remember our first eventful trip for the first X-Prize Cup when Joseph was held at the inspection point for trucking violations at this same port of entry. Peter Diamandis had to come to our rescue, and save the X-Prize Cup, by calling in assistance from Governor Bill Richardson who in turn had the Secretary of Transportation call a rather surprised state trooper who then escorted us to Las Cruces. Its good to have friends in low places.
More recently I had been stopped for bypassing the port of entry with the pick-up drawn rocket trailer for a STIG A launch. Apparently ALL commercial (and rental!) vehicles that weigh more than a Smart Car have to stop and get a fuel sticker. This involves standing in line with the truckers for an indeterminate period of time to pay $12 for gas / diesel taxes that you might not pay if you dont fill up in NM! That wasnt the only problem though and my ticket cited another half dozen assorted offences
not having a CDL medical driver examination on file
even though it doesnt require a CDL to drive the vehicle, and speaking with a funny English accent was another. This time we had our proverbial ducks in a row and other than a lengthy delay at the port of entry, we sailed through without issue.
We chose Las Cruces as our base of operations primarily for the better facilities (we need a hotel where I can park a 65-ft rig close by) and also to be close to the Spaceport America offices for the MRR. Thursday is our set-up day and we had a lot of new GSE to unload plus some modifications to the launch stool on the pad to accommodate the larger airframe of STIG B. We had also observed some bounce back of the launch stand arms on retraction and we had made some last-minute changes that we wanted to test out prior to flight. Preparations went so well that we actually had the vehicle on the launch stand for a dry run, including the pull-test to check the retraction mechanism, before the end of the day.
Friday was for tear down ready to install the payloads and to work on the recovery system installation. Wamore still had to program the AGU (Automatic Guidance Unit) for the requisite landing spot and then all the hardware had to be reinstalled. The new tripod built for lifting the vehicle when recovering in the field proved invaluable for installing and removing the modules. During this period we also pumped up the HP helium tank to 5,000-psig and TU Braunschweig pumped down their payload with a vacuum pump which ran all night. The team departed the launch site around 4:00 p.m. to attend the mandatory MRR (Mission Readiness review) which this time was held at the NMSA offices in Las Cruces. It was a bit cramped with so many attending but this looks like it will become the venue of choice for future MRRs.
The increased helium storage pressure of 5,000-psig on STIG B means that we cant easily cascade fill without the use of 6,000-psig cylinders which are both expensive and not readily available. The solution to this was an air operated Haskel gas booster which is capable of pumping up to nearly 10,000-psig using shop air pressure.
One of the payloads was from our old friends at the University of Purdue led by Professor Steve Collicott. The students for this launch were Christopher Riemer, Joseph Whitman and Austen Wildberger and the payload was the venerable SPEAR , "Students of Purdue Experimenting on Armadillo Rockets." This payload is triggered by an integral G-sensor switch and watches bubble formation, not only during the period of micro-G, but also during the boost phase during which the near 2-G initial boost slowly rises to more than 4-G at MECO.
We decided on a 6:00 a.m. departure from the basecamp hotel which put us on the pads at ~7:00 a.m. entering from the southern route. Sunrise was at 7:10 so we had good dawn light on arrival. Weather conditions were acceptable with very light winds but there was scattered to significant cloud cover. This is no longer a problem unless the ceiling drops below 5,000-ft AGL or visibility is less than 10-nm. Our launch license allows us to launch through cloud at our discretion. Cloud ceiling is checked both visually and with the closest AWOS in El Paso.
Earlier that morning, much earlier, I had collected the weather data and run a TAOS (Trajectory Analysis & Optimization Software) trajectory analysis of where we anticipated recovering the vehicle and drogue / nose cone. The winds aloft were almost exclusively out of the West which was ideal for direction although they were stronger than we would have liked aloft peaking at more than 80-mph. Wind speed at ground level was imperceptible and likely a constant given the sheltered nature of the desert bowl that the Spaceport is located in, surrounded by distant mountains to the West and East, with lesser mountains to the North.
After discussions with Wamore we opted to stick with a main deployment altitude of 10,000-ft AGL (3,050-m) and see just how far we could fly-glide back towards the landing site. The nose cone and ballute we expected to land some 10-km East and barely South of the launch pads, still well inside our Flight Hazard Area. The vehicle main would deploy when the vehicle was about 7-km downwind and the hope was that, with its 2:1 glide ratio, it would be able to fly its way back to the launch area.
Vehicle preparations took a little longer than we had hoped, about 90-minutes, but we had planned accordingly and even had time to take a few obligatory photographs of the observers next to the launch vehicle on the stand before hazardous ops started. Topping off the HP helium tank took next to no time because the pressure had dropped very little overnight. We had decided to locate the launch control trailer directly facing the launch stand down the dirt track and have the observers located near the UP Aerospace center more than a couple of hundred meters distant. The only people other than the launch team at the Armadillo Launch Control area would be Dave Gerlach from FAA/AST and Bill Gutman the Spaceports RSO (Range Safety Officer.) This worked well and kept the uninvolved well away from the busy ALC area and we will do this again for the next mission.
The hazardous launch prep was uneventful except for a problem with the RTT (Remote Thrust Termination System) not operating during the pre-flight checks. After a few minutes troubleshooting Russ tracked this down to a bad USB cable and a drop in replacement resolved that problem. The total time from going on checklist to punching the launch button was 40-minutes and this included 10-minutes for resolving the RTT problem. Now that we dont have to load helium with the GOX heat exchanger we have nearly halved the amount of time on checklist. STIG A used one of the incredibly lightweight Scorpius helium tanks but we learned, from bitter experience, that we have to chill the helium down during loading or else risk losing vessel integrity from heating effects. The STIG B tank is a more robust, but extremely heavy, NGV tank that we pump up to 5,000-psig using a Haskel pump prior to heading out to the pads.
The above sequence of launch photos shows the early visual evidence of a non-nominal trjactory.
The 15-second countdown and launch was flawless. The launch stand retracted without drama and the boost, although definitely not as brisk as with STIG A, still got off the launch stand in a hurry. It was immediately evident though, even to the naked eye that she was drifting slightly to the West. There is always an optical illusion that she is coming directly overhead which can be disconcerting for neophytes, but this was definitely drifting away from the pad. At T+22.5 seconds (this is from ignition; the time to lift-off is roughly 2-seconds after ignition) the thrust was terminated by the MFC (Main Flight Computer) on Groundspeed exceeding the flight rule limit of 50-m/sec.
The first frame in which the Groundspeed exceeded 50.0-m/sec is shown on the above graph. The engine terminated thrust within less than one second but during which interval we picked up another 4-m/sec in vertical velocity (215-m/sec at MECO) and just over 2-m/sec Groundspeed. Had the Groundspeed limit been set higher or even non-existent, the flight would still have terminated on 7-km IIP some 40-seconds later at ~T+60 seconds (see dotted line extrapolation of boost period.)
At this point the vehicle weather-cocked, coasted into the prevailing wind and did a gravity-turn with an apogee of 4,430-m AGL at T+44.56 seconds.
There had been a slow negative roll during the boost phase of 4-deg/second. Post boost, the roll reversed at a slightly higher rate of 13-deg/second. It is uncertain if this roll would have continued because as soon as the nose cone and ballute deployed there were wild oscillations although, it does appear that as the oscillations died down the roll reversed once again. This observation is based on the straight line transition from +180 to -180 just prior to drogue release but -180 to +180 post drogue release.
The drogue did not deploy fully for about 4-5 seconds likely because it was in dirty trailing air behind the vehicle upon initial deployment. At apogee the vehicle was at an alpha of circa 66-degrees (63-deg yaw and 26-deg pitch) and traveling at 46-m/sec relative to the ground in still reasonably dense air (just less than 20,000-ft MSL). It wasnt until T+56-secs that the drogue fully deployed and put the brakes on. The vehicle reached terminal velocity of 68-m/sec in just 10 seconds.
At this stage the vehicle still had the residual fuel on board and we estimate the mass at close to 900-kg (2,000-lbm) at an altitude of 4,500 5,000-m MSL. Working backwards from these numbers we come up with a CD value of 27.6 relative to the vehicle cross sectional area and 0.8 based on the nominal chute diameter (9.75-ft). The value we had been using in TAOS for subsonic velocities was 25.7 so we can now adjust based on test results. However, we still do not have any good data in the supersonic regime which we are anticipating will be considerably higher. The good news is that in the regime for main chute deployment we are good.
The main chute deployed (on a timer basis because GPS was down; see recovery system notes later in document) at T+83-seconds and the opening was very rapid with indications of close to a 10-G shock. There were signs of ineffective operation of the slider and Strong is working on a solution.
Once fully opened the estimated CD using the same approach as for the drogue was 198 based on the body diameter (3.87 based on the chute nominal area). This is an improvement over the estimated value of 143 used in TAOS.
Looking at the groundspeed track, the chute makes a fairly consistent ~9-m/sec throughout its stable terminal velocity descent. There are a few points at which there are noticeable drops but they occur when the AGU is putting the vehicle through a sharp turn to track to the landing point.
Two points to draw from this; (1) the glide-descent ratio is about 2:1 and did not appear to be sensitive to direction and (2) consequent to this it is reasonable to assume that the winds aloft at the range are not the same as those we obtain from the weather data services.
We were anticipating winds as high as 10-m/sec from the West in this range which would have led to rapid shifts in groundspeed as the vehicle turned upwind and downwind. The surrounding mountains not only provide protection at ground level, they appear to do so to some significant altitude aloft. This works to our advantage in terms of both downwind drift during descent and ability to recover from significant distances.
Opening the main at 3-km AGL should allow us to glide-fly back from distances of 6-km. The potential downside to this is that we could possibly see a rapid wind shear as we come out of the shadow of the mountain range during boost.
Bottom line is that if we give the Wamore chute a chance to operate, it does a remarkable job of finding its way home!
Above shows location of the actual landing location (white) and the desired (yellow) about 50-m distant. In retrospect, this was a little too close the trailer and next time we will chose a location somewhat more distant.
The engine appeared to perform nominally based on the static tests undertaken at Caddo Mills. Boosting the HP helium to nominal starting pressure the morning of the launch should also be adopted as standard practice. The current NGV tank is incapable of keeping us out of the blowdown phase without pushing the operating pressure to what could be dangerous limits. We are currently operating at 5,000-psig MAWP on an NGV tank that is Class 3 (flammable) DOT rated to 3,600-psig. DOT rated NGV pressure vessels have a burst safety factor of 2.35-Carbon Fiber / 3.00 Aramid Fiber / 3.50 Glass Fiber. The tank we have should be good for a 10,000-psig minimum burst so we are currently operating with a 2:1 Safety Factor. Pushing it beyond the current pressure would warrant a burst test on one of our tanks and a proof test of the flight article to a 1.5x desired MAWP.
The initial HP helium is at 5,050-psig but drops substantially with ullage pressurization and the ignition test to 4,600-psig. This would warrant boosting the initial pressure a little higher so that we actually see 5,000-psig at ignition.
The average tank pressure during the boost phase was 370-psig with a resultant chamber pressure of 278-psig. This is slightly lower than we expected and it didnt appear that we picked up any bonus because of the static head G-boost. The engine is also a little buzzier than we have previously seen at these pressures but there was no audible sense of this from the ALC position.
Even with this slight shortfall in thrust, the altitude and velocity just before MECO were within a couple of percentage points of predicted and this before we allow for the energy lost in accelerating the vehicle laterally because of the position offset anomaly. That would not appear to be substantial but, in the same 20-second time period we accelerated the vehicle vertically to 190-m/sec, we accelerated her laterally to 40-m/second.
Ignoring drag effects this implies a loss of a little more than 2% of the vertical thrust which would actually compensate for the perceived shortfall in altitude and velocity. Bottom line is that, at least in the subsonic range, our theoretical drag coefficient estimate in TAOS is very close to actual. The PTs (pressure transducers) appear to be reading low because of the temperature and we are adding some insulation to reduce the effects.
Ignoring the first two seconds of the time scale above, which is the ignition and throttle up, and comparing T+21 seconds with T+19 seconds on TAOS, the comparison values are;
Altitude MSL (m) 3,069 3,150
Velocity (m sec-1) 190.8 194.7
Notably, this is even with the chamber pressure some 10-psi below the predicted because of the strange lack of a G-boost?! Correcting TAOS for the thrust vectoring and eliminating the pressure G-boost resulted in predicted values slightly below those actually achieved.
Indications are, extrapolating the pressure decays, that the vehicle would go into blowdown some 50-seconds into the burn. To prevent this would require that we boost the initial helium pressure to circa 6,000 to 6,500-psig with the current pressure vessel. Longer term, changing to a larger Scorpius style vessel would both eliminate the blowdown phase and, importantly, reduce the dry mass by 50 to 60-lbm. This should add 10-km to our altitude capability!
The following is a verbatim cut-and-paste from Phils Recovery System Analysis for STIG B-I
The recovery system was activated at approximately 14,500 feet. The ballute came out appropriately on cue, but the angle of the rocket was near horizontal with a lateral velocity of greater than 50 mph. The south side of the rocket was facing upward and that is the side of the rocket that the ballute streamed out. When the ballute inflated, all of the initial forces were approximately 150 degrees opposed to the ideal direction of force for the recovery system. This would pull the main parachute container back against the recovery section can directly to the south. The G forces were 2.92Gs, 2.02 Gs, and 1.50 Gs.
On the south side of the main parachute container is the pin housing tie wrapped to the pin pull sensor wire. Both the pin pull housing and the pin pull sensor wire show evidence of being severely pinched between the main parachute container and the can. There is also a mesh pattern on the edge of the can that appears to match the weave pattern of the braided hose on the outside of the pin housing.
The ballute experienced some minor damage. One D-ring has a broken weld and will be repaired, and several seams were strained a small amount. There was no tearing, so the seams will be reinforced in their strained state by additional Kevlar strap material. It is surmised that the damage to the D-ring occurred when the ballute trailed behind the vehicle and possible made contact with the fins. In a nominal flight this would not be a problem. As it is, the minimal damage for an off-nominal flight is acceptable and easily repairable.
The nose cone, containing the pressurization system that blows off the nose cone and a tracking transmitter to indicate location when the ballute separates from the vehicle, was connected to the main riser leading to the ballute just above the swivel intended to allow rotation of the ballute and nose without inducing a roll into the main vehicle. The connection point was sewn into the main riser such that it could handle the forces of deployment in one direction in shear, but had a reduced strength if the forces moved in the opposite direction in peel. From previous testing we know that the sewing with that particular Kevlar strap using the 70lb Kevlar thread is good for 20,000 lbs in shear when overlapped 16. Unfortunately no matter how long the joint is, the strap is very weak when in the peel direction.
Again there is no video record of the detachment of the nose from the riser, but evidence suggests that after the deployment of the ballute, the oscillation of the system could have caused a force in a direction (peel) that was previously unanticipated and which tore through the threads of the attach point on the riser which in turn released the nose from the recovery system unintentionally. The remaining portion of the riser remained intact and was able to withstand the forces of the recovery process after the release of the nose.
The nose impacted without a deceleration device approximately 1000 meters northwest of the launch point at 32.9460, -106.93202. The nose falls flat when not attached to a recovery device. No impact mark was found on the ground, and the energy of the impact was absorbed by the fiberglass construction causing it to crack and bounce back into the air over 10 meters before coming to rest. The ballute floated down with very little ground wind and landed approximately 475 meters North East of the launch pad. At 32.94545, -106.91811
The strap will be modified so that the nose connection will be attached to the strap with a 12+ overlap that will be in shear in both directions. The attach point will be separated from the main strap and be a separate loop that terminates at the connection to the swivel so that regardless of which direction the nose pulls it will pull in shear and thus be appropriately rated for this load in any direction. The attachment can be made to the nose with either a drag link or a soft link.
The soft links have been tested to over 6000 lbs, and the forces between the nose and the link are much less than this. The connections holding the main body of the rocket to the main straps must be made with Drag Links rated at 15,000 lbs. These have been tested up to 20,000 lbs without deformation.
The main parachute was triggered at T+51.3 seconds after ignition, but did not deploy at this time. At T+82.0 seconds, the main actually released confirmed by the pin pull sensor, at a vertical velocity of 64 meters per second or 143 mph according to the IMU and GPS. The peak accelerations are as follows, 10.74 Gs, 3.84 Gs, and 4.25 gs. The velocity is a little bit high but is within the reasonable expectations for the recovery system to function without damage. This system is tested by the manufacturer, Strong Enterprises, to be dropped out of an aircraft at 135 knots using a 13 foot diameter ring slot parachute as a drogue with a 3 second delay.
Unfortunately none of the video cameras pointing straight up activated so there is no video record from on board to analyze the deployment of the main completely. There are a few photos of the opening, and the data has been correlated to see the speed at which the main opened. It appears that the main chute opened in approximately 100 milliseconds. The photos indicate that the far left and far right sides of the main caught air early in the deployment blowing out a large portion of 2 cells on the right side of the parachute. The slider was severely strained on the grommets by the A and B line sets and is deformed beyond usability for the next flight. The slider must be rebuilt.
There is also evidence on the main parachute of slight burning from the velocity of the shroud lines pulling out of the packing. It burned through both the top and bottom layers of fabric. Visible inspection by Lee Hardesty indicates that the back panels blew out by the force of the air hitting it, not by burning coming out of the deployment bag.
Russ attempted to dump the remaining propellants under the drogue, but there was a residual ignition source still in the engine that caused a relight. Upon seeing chamber pressure the valves were immediately closed. The dump attempts were resumed successfully after the main had opened.
1. Snatch force implications on opening?
2. Material in the deployment bag too rough?
3. Flag on the Slider needed?
4. Cutter on the slider needed?
5. Does excessive snatch force negate nose rolling?
First of all, a nominal flight will not see these kinds of forces, but contingencies will be accounted for in an attempt to recover the vehicle without damage on sub optimal trajectories.
Possible solutions for the next flight:
1. Alter the dump sequence to have a purge to ensure there is no remaining ignition source in the engine at the dump time.
2. Delay the main deployment if possible until after the dump is complete, or at least partially done.
3. Add a flag to the slider to slow the opening of the main.
4. Add a cutter to the slider mechanism to hold the main closed longer after the initial deployment.
5. Add some sort of break cord/rubber band mechanism to hold the slider in the appropriate place for high snatch force openings.
ACS Roll Control: The ACS system remained dormant until MECO as planned and then began to fight the significant aerodynamic forces encountered during the ballistic trajectory through nosecone and drogue deployment. The graph below shows the roll angle, rate and acceleration. The active period is from 22.5-seconds through 45-seconds.
After the initial kick at lift-off which is quite evident from roll acceleration, the roll rate drops back to zero after ten seconds with the vehicle rotated -70 to -80 degrees. The .rprog file indicates that the ROLL- thruster is firing almost continually which, given that the vehicle does not spin-up in any direction but rather stays within a narrow band of 0 to 10-deg/sec with a short peak of 20-deg/sec at T+40-seconds, indicates the control is in the correct direction and in the absence of large aerodynamic forces should result in reasonable roll rates.
If the roll rate could be kept to less than 20-deg/sec (reasonable expectation exoatmospheric) then the acceleration at the periphery of the vehicle would be 0.03-m/sec2 or 0.003-G. There is, of course, a linear reduction approaching the centerline axis of the vehicle.
Detailed analysis of the telemetry revealed that the drift problem which caused the early flight termination was because of an IMU configuration change between STIG A and STIG B. We had debated at length about tether testing STIG B, quite literally hovering a missile! We had done this with STIG A and I have to say it was a very tense period. This could have been accomplished using the largest road going crane available, or perhaps a pair of them, but more importantly, the thrust to weight ratio has the engine working at the bottom end of its range and we were concerned that we might lose an engine and possibly the vehicle.
Since the STIG B flights we have had further discussion about how we might implement this test in future using a large ballast section (to reduce the thrust to weight ratio) and commercial load arrestors in place of our existing strap & bungee cord arrangement (to decrease the shock loading on the crane). However, we actually were able to confirm our IMU configuration issue with a simple swing test on our crane truck in which we swung the vehicle in an arc as fast as could safely be done.
Return to Flight:
1. The boat tail had taken the brunt of the landing load and was deformed beyond reasonable reuse.
2. The engine nozzle had taken load next and it too was deformed but repairable.
3. The two gimbal linear actuators needed either repair or replacement.
4. One gimbal support arm was broken.
5. The heim joints needed to be replaced.
6. Two cells on the main chute needed repair.
7. The drogue had some stretching and broken stitches.
8. The slider grommets were deformed and the slider needs replacing.
9. Reconfigure IMU for STIG B.
10. The nose cone attachment loop needs redesign and replacement.
11. The nose cone itself is damaged beyond rapid immediate repair.
12. The lifting strap stitching needs some reinforcement.
Although this was a somewhat disappointing result for the inaugural flight, there was cause for some celebration. The vehicle was recovered intact enough that it would fly again and our flight safety systems retained their 100% record.
This was not only our first licensed flight, it was the first licensed flight of a fully reusable, liquid bipropellant sounding rocket and the first licensed launch from Spaceport America.
In the true Armadillo tradition, we were learning by flying hardware. Moreover, although nine months seemed like an eternity to us, building a totally new complex vehicle from scratch with our small team seems astounding in the aerospace industry.