In the span of a few seconds, a smooth countdown to launch Space Shuttle Mission STS-93, a five-person crew, and the Chandra X-Ray Observatory abruptly ended twenty years ago just before ignition of Orbiter Columbia’s three Space Shuttle Main Engines (SSME).
An indication that hydrogen gas was leaking into the orbiter’s aft compartment where the engines were mounted popped up on a systems engineer’s computer screen as pre-liftoff water began to be poured into the launch pad flame trench.
At T-10 seconds, the Ground Launch Sequencer (GLS) gave Columbia’s flight computers the go to start the engines and the hydrogen burn off igniters started as another precursor to main engine ignition that occurs just under seven seconds from liftoff. Starting the engines with leaking hydrogen could make the situation worse, so with no time to lose, primary hazardous gas system engineer Ozzie Fish called out on the Shuttle launch team’s main voice loop to stop everything.
It could take a second or two to manually stop a Shuttle launch countdown, which fell on primary GLS engineer Barbara Kennedy to make happen by pushing a button on her console to cutoff the countdown.
The indication of a leak turned out to be a false positive but the quick reaction of the pair was able to stop the launch sequence less than half a second before the first main engine start command, saving the launch team from a weeks-long turnaround after an ‘on-the-pad’ engine shutdown.
Instead, Columbia was ready to try again in two days.
GLS overview
The Ground Launch Sequencer automated all the fast-paced activity during the last nine minutes of the terminal countdown. The software automatically conducted the final sequence of actions required to get the Shuttle ready for ignition of its engines and liftoff. In addition to sending commands to vehicle and ground support systems to perform actions on a specific timeline, the GLS also monitored all of the critical ground and vehicle parameters during that time to enforce a pre-determined set of Launch Commit Criteria.
Liftoff of the STS-93 mission was scheduled overnight from July 19 into July 20, during the local midnight hour at the Kennedy Space Center (KSC). One group of GLS engineers was on console in Firing Room 1 of the Launch Control Center at KSC in the middle of the afternoon on the 19th, as other members of the launch team began remotely loading the External Tank of the Shuttle vehicle with its liquid, cryogenic propellants on Launch Pad 39B.
For STS-93, Kennedy, who was with United Space Alliance at the time, got the assignment to be the primary GLS. “I had worked the prime seat before so it wasn’t new to me in any way,” she said in a recent interview.
“It was a routine flow that we had performed for STS-93 but it was more exciting because it was the first woman commander with Eileen Collins at the lead and Jeff Ashby as the pilot, so that was kind of neat,” Kennedy added. “With doing some pilot training in the past I really thought that that was superb that I was able to work the prime seat for GLS that day to help her get the Chandra X-Ray Telescope out into orbit, so that was quite an honor to be sitting in that seat when Eileen Collins was the first commander.”
“It was pretty much as I recall a standard Ground Launch Sequencer configuration that day and as we moved through the count we would do standard events and I don’t remember anything anomalous that happened early in the flow — it could have been, it was a lot of years ago,” she noted. “We were just excited to be there for that day.”
“We would have never known close to T-0 that we would have had an indication that we would have to react very quickly to that anomalous condition that we had not seen before.”
Credit: NASA.
(Photo Caption: Firing Room 1 in 1998. The back row of consoles, including where the GLS group was stationed, are not visible from this point of view.)
The GLS group worked at the Integration Console. “We’re a part of what they call the Integration Console and they have several of the senior engineers working there and they’re designed to integrate all problems and say ‘we’re go for launch,’ Kennedy said.
Their console area was in the back of the firing room. “All the way in the back, farthest from the windows,” she noted.
“The Ground Launch Sequencer folks are three people on the prime team and then three people on the earlier team. When they start tanking at T-6 hours and filling up the External Tank with the LOX (liquid oxygen) and LH2 (liquid hydrogen) fuel basically they have a early pre-setup and monitoring of all parameters in the launch sequencer and then the prime seat or ‘hot-seat’ people that are going to do the terminal count come in at T-3 hours so they’re a little more fresh and ready to react,” she noted.
“So that’s basically what we did when we worked the prime seat, it was T-3 hours, walk in the door and the team goes ‘all right, GLS is here!’ And about that time the crew is also getting suited up and ready to board the spaceship.”
Credit: NASA.
(Photo Caption: A composite diagram showing the layout of the consoles in Launch Control Center Firing Rooms for Space Shuttle launches (upper left), and the seating chart for the Integration Console, which was in the far back/right in the Firing Room (Firing Room 1 was used for STS-93). The GLS engineers were seated with Shuttle test and project engineers.)
The sequencer is not active until later in the countdown, and while the crew is driven out to the launch pad to board the Orbiter, the GLS Group is setting up the software for the specific mission. “When GLS Prime team comes in at T-3 hours we would initiate the launch sequencer, do the pre-sequencing events, set up what was going to be the parameters for that day for that specific vehicle, and then we would move on to activate critical monitoring,” Kennedy explained.
“Around the two-hour time period we would activate thousands of parameters and what their limits were,” she said. “You begin with pre-sequencing events and activating parameters and constantly verifying a group of parameters and sometimes doing a ‘one-shot’ on other parameters just to see how other things are going and these are all in what they call the Description Document, the GLSDD.”
“Those were all agreed upon for that flow, and so as you moved through the sequencing as some things might not be crucial for that launch, so you would mask out or bypass anything that was not [relevant] for that particular mission, so you had a configuration that was already set and agreed upon by NASA. As the launch progressed sometimes those parameters were changed to a different phase of the launch countdown and we would update those parameters and always field any issues coming in, so as you’re walking through the launch countdown, getting down into T-20 minutes, different configurations were required and basically that’s how the sequencer worked is that you did certain things in an order that was already pre-approved.”
“It was very routine many times, so it was not a surprise on any of these things that you had to change a configuration, it was pre-approved usually,” Kennedy said. In addition to mission-specific parameters, mission or orbiter or situation-specific data may need to be ignored.
“For an out-of-tolerance you would mask that for the day or you know if you had four parameters and you knew one was bad there was a way in the sequencer you could go into the table and simply mask out one of them to vote good if you wanted it or vote bad, so we had the capability in the software to do that,” Kennedy explained. “Also for the sequencing, you could skip over line items in the sequencer; for example, if you had a powered payload or if you didn’t have a powered payload, you would want to activate the things being monitored on a powered payload or conversely not being monitored.”
“So there were different configurations in the software and that was what was so neat about the Ground Launch Sequencer, we had a million lines of code but you had a lot of flexibility.”
Credit: NASA.
(Photo Caption: Images of the fully-loaded STS-93 Shuttle vehicle during the countdown for the first launch attempt taken by the Final Inspection Team from the Pad 39B service structure. The team went to the pad after the External Tank was filled and in stable replenish mode to inspect the vehicle.)
The GLS was activated at T-45 minutes. “You would move through the countdown and various things on the vehicle would be done, cycling vent doors and as you moved through the GLS milestones going down into T-9 minutes we would verify that all systems were go before we picked up the count at nine minutes,” Kennedy said. “So there was a built-in hold, both at [T-minus] 20 minutes and at T-9 minutes to ensure that you had finished troubleshooting any anomalous conditions and that you were ready to go.”
The GLS auto-sequence started at T-9 minutes and counting and was in control of the terminal launch sequence until T-31 seconds when it handed over primary control to the vehicle’s flight computers located in the Orbiter. “As you’re progressing and doing all these sequenced events, they’re in order, they’re run the same way every countdown and that’s how it works in all of aerospace is that you have a known sequence and it should be pretty much hands-off, you shouldn’t have a human in the loop for standard events like that, it’s more efficient for the computer and the software to actually work per the countdown clock and sequence those events,” Kennedy said.
The prime GLS operator was one of the voices Shuttle launch viewers would hear during those final nine minutes of the countdown; as the sequencer passed through the different GLS milestones and verified that all the Launch Commit Criteria was being met, the GLS operator would make that call on the voice loop to inform the rest of the launch team. “You’re moving through T-9 minutes and the auto-sequence has been initiated and of course you’re going to go for the OAA (Orbiter Access Arm) retract at seven minutes and thirty seconds, then the APU start happens at five minutes,” Kennedy said.
“If any anomalous condition occurs before that time you would stop at five minutes before you initiated APU start.” The GLS engineers were monitoring their screens for the nominal progression through the milestones and anything not-so-nominal.
“We had another main screen that the other GLS operator [watched]– there was a prime seat and a backup seat,” Kennedy explained. “So that was the screen where the anomalous conditions came in, it was called the lookup.”
“As the lookups would display any anomalous condition you would be monitoring to say OK, that one voted good because we had two of the four that were good, and that one voted good because we had one of the three that were good. We had a lot of redundancy in Shuttle for protection of both the crew and the vehicle.”
Credit: NASA.
(Photo Caption: Another image of the fully-loaded STS-93 Shuttle vehicle (Orbiter, External Tank, Solid Rocket Boosters) during the countdown for the first launch attempt taken by the Final Inspection Team on the evening of July 19, 1999.)
The Orbiter’s Redundant Set Launch Sequencer (RSLS), running on its four primary General Purpose Computers (GPC), is in charge of the last half minute of the countdown. Although the RSLS was now issuing commands and enforcing the Launch Commit Criteria, the GLS continues to monitor that criteria and is still issuing some commands.
The automatic hand-off from GLS to RSLS occurs with the “Go for Auto-Sequence Start” milestone at T-31 seconds and the GLS operator announces that milestone on the voice loops and arms a manual cutoff capability. “Once we sent the ‘go for auto-sequence start’ to the on-board, the on-board actually launches the vehicle, so at that point, you’re just monitoring.”
“You’re working through T-31 seconds, you’re prepared for a cutoff any time after that.” Below the console displays was a programmable function panel, with several buttons. “Once the 31 second [milestone] comes through you arm the manual cutoff, so you would arm it with your left hand and then you would have your thumb on the button for the execute which would send the vehicle the stop.”
“The arm is on the top and the execute is on the right bottom,” she noted. While the GLS and Columbia’s computers are busy starting the Solid Rocket Booster (SRB) hydraulic power units (HPU), finishing configuration of the Main Propulsion System (MPS) for engine start, and continuing to verify all the criteria are being met, NASA Test Director (NTD) Doug Lyons made calls on the launch team voice loop as the countdown passed through twenty-five, twenty, fifteen, and ten seconds for an on-time liftoff at 12:36 am Eastern Daylight Time.