Riding a Beam of Numbers
A flawless launch puts the unmanned VDL-Cargo into a 182x183km Earth orbit and translunar injection is completed just over an hour later. This time, the ECPS performs slightly better than expected, pushing the 34,027kg VDL-Cargo towards the Moon with 4.9t of propellant remaining at shutdown.
A course correction 24 hours out lowers the anticipated perilune and in doing so successfully tests the VDL's main engine. It proves to be sufficiently accurate that no further corrections are needed. Lunar orbit injection begins at T+87:38 under automatic control, lasting 7m 5s and resulting in a 41x243km lunar orbit, far lower than expected. A 3s burn an hour later raises perilune to a safer 66km.
Plans for further manoeuvers are put on hold while the error is analysed. It is clear from models of the orbit’s parameters that the injection burn was made at slightly the wrong angle (about a degree out). Over the next two orbits, controllers diagnose a misalignment of a gimbal offset when the system was last updated shortly before the LOI burn.
A new offset is uploaded to the VDL's computer, but it is thought safest to attempt a relative minor manoeuver to test the settings before proceeding further. At T+96:02 the orbit is changed to a near circular 95x100km.
With the ship now in a stable configuration, the Mission Controller delays the descent to give controllers time to rest and allow for further checks. Lessons have been learned from the early “Explorateur” flights, when controllers made quick (often too quick) adjustments and then continued with the mission as if nothing had happened. Modern flight plans allow time to solve some of these unexpected problems and flight rules now demand that controllers stop and diagnose faults that might threaten the mission, rather than relying on ad-hoc adjustments.
As part of these checks, engineers conduct a complete ground based computer simulation of the upcoming descent. The parameters of the ship's orbit, fuel loads, centre of gravity, inertia, control response times, engine performance and an exact copy of all the settings stored in the on-board computer’s memory are used to check how the VDL will behave during descent. The powerful ICL mainframes at Mission Control take the hundreds of numbers that define the properties of the VDL and use them to simulate every pitch, yaw, thruster firing, throttle response and dozens of other parameters of the ship’s behaviour as it travels down to the surface.
Trajectory designers have programmed the ship with an “ideal path” to follow; this is a smooth curve, gently bending down towards a vertical touchdown at the landing site. The path is carefully designed to allow the VDL to balance thrust, attitude and lunar gravity at all stages of flight, without the need for any quick changes.
Of course, this ideal solution will never occur in reality. The ship will always be a bit high, too slow, a different mass or not quite as responsive as the calculations predict. Consequently, the most important task of the guidance algorithm is to allow for these errors and steer the ship towards the ideal path, while also accepting input from pilots and updates from the landing radar.
It must also ensure that all manoeuvers happen quite slowly and gently; they cannot be allowed to call for say, a “snap turn” of 60 degrees. It takes the RCS thrusters time to torque the ship in any particular direction and to limit the buildup of any excessive pitch rates, the system is electronically restricted to a maximum rate of 5 degrees/second.
In addition to the ship’s position, velocity and acceleration, the computer has to consider two further parameters – “Jerk”, the rate-of-change of acceleration (essentially due to pitch rate) and “Snap”, the rate-of-change of rate-of-change of acceleration (dominated by the rate-of-change of pitch rate due to the action of the attitude control thrusters). Each of these parameters is a vector and is the differential of the previous one, and so it is possible for the computer to solve them numerically.
On Selene 4 the caution at Mission Control pays off. Controllers discover an erroneous zero setting of a target point in one of the guidance programs. If this zero had gone unnoticed, the effect would have been to cause the VDL to descend too quickly in the middle part of the landing. It would have crashed shortly after switching to the "Targeting phase”. A straightforward change of a number in one word of the computer's memory is all that is required to correct the problem.
The flight plan is resumed at T+104:12 and lunar descent orbit is entered at T+105:35, bringing the VDL to within 16.3km of the lunar surface when at perilune. After two orbits to refine the ground-based tracking solution, the computer is commanded to proceed with the landing.
At T+110:25:02, the "Pilotmode 5" guidance routine ignites the engine to start the 700km-long braking burn and descent towards the lunar surface.
For the first 20s, the engine fires at its minimum 25% throttle setting, before going to 100% for 10s, then back to 25% for a further 15s. Although not strictly necessary on an unmanned flight, these throttle changes serve to test the engine's performance early in the landing. If something went wrong at this stage, a crew could quickly and safely abort the descent and stay in lunar orbit.
The throttling and early phase of descent proceeds smoothly, albeit with the ship pitching under its control thrusters more than was expected. At 400s (6m 40s) into the descent, a manned flight would reach the Landing Decision Gate (LDG - the infamous "point of no return"). The VDL is performing well and if there had been a crew on board they would no doubt have been able to continue to attempt a landing.
90km out from the landing site, ground tracking suggests the ship will land short and the Targeting Pilot, Guy Larosse, is advised he will have to extend his glide slope (as it is still called, despite not being in any way a glide) in the final stages of the landing.
Twelve minutes 9 seconds into the descent, the ship reaches a point 10km from its programmed landing site and the on-board computer automatically switches to Pilotmode 4, the Targeting Phase. The VDL pitches down by about 40 degrees, bringing the landing site into view of the LPI. For these unmanned cargo flights a TV camera is mounted on this computer-controlled sight, which always points towards the current landing site and feeds images back to a controller on Earth.
This controller (who will always be a Selene astronaut) stands in a duplicate of a VDL cockpit and can instruct the VDL's computer to update its landing site based on the TV images being fed back to him from the Moon.
The fuzzy black-and-white image takes a moment to stabilise, but by 12m35s Larosse confirms he sees the surface. He recognises a small crater to the north of the landing site a few moments later. After a second or two he nudges his control stick eight times to re-designate the landing site about 800m downrange. Thinking he is still short of the site, he waits for nearly 15s to let the VDL properly stabilise onto its new trajectory, then enters 5 more clicks, 4 downrange and one to the south, before quickly clicking one back (to move uprange). Satisfied with the images he sees through the LPI, Larosse’s job is done. Although he continues to monitor the ship's progress down to the surface, ready to make further updates or even attempt manual control if the on-board systems fail, he enters no further updates and the VDL settles on its new course.
At 15m52s the computer switches to its final landing mode “Pilotmode 2” - to stop any residual horizontal motion and gently lower the VDL to the surface.
Contact probes and accelerometers indicate that the VDL touches down 16 minutes 11 seconds after Descent Engine Ignition, at a speed of just 1.2m/s. The engine shuts itself off 0.6s later and no further movement is detected after DPI+16:13.2.
At T+110:41:18, cheers and applause erupt around the control room at Biscarosse as the telemetry indicating touchdown is received. It quickly settles as mission engineers focus on the status of their new lander. Two minutes later, a mast carrying the panoramic camera is raised and starts to send higher quality images of the landing site back to Earth.
Earth based radio location and images from the camera soon confirm that the ship touched down a mere 70m from the planned landing site, which is about 90km southeast of the crater Copernicus. The performance of the systems and Larosse’s corrections during the descent were almost perfect. Calculations and telemetry agree that the VDL had 820-830kg of fuel remaining on board at landing, slightly higher than the expected margin. It could have flown for another 138s, sufficient to allow it to land (at most) 5km further downrange or 2km to either side. The final landed mass of 15,702kg is greater than all previous Selene, NASA and Soviet lunar landers put together.
The ship has successfully ridden its beam of numbers from 4,000mph down to zero, from ten miles up and 450 miles away to land within a few hundred feet of the pre-planned site.
Ten years after the Project was begun, the first lander large enough to carry a crew has reached the lunar surface.
Selene 4
A flawless launch puts the unmanned VDL-Cargo into a 182x183km Earth orbit and translunar injection is completed just over an hour later. This time, the ECPS performs slightly better than expected, pushing the 34,027kg VDL-Cargo towards the Moon with 4.9t of propellant remaining at shutdown.
A course correction 24 hours out lowers the anticipated perilune and in doing so successfully tests the VDL's main engine. It proves to be sufficiently accurate that no further corrections are needed. Lunar orbit injection begins at T+87:38 under automatic control, lasting 7m 5s and resulting in a 41x243km lunar orbit, far lower than expected. A 3s burn an hour later raises perilune to a safer 66km.
Plans for further manoeuvers are put on hold while the error is analysed. It is clear from models of the orbit’s parameters that the injection burn was made at slightly the wrong angle (about a degree out). Over the next two orbits, controllers diagnose a misalignment of a gimbal offset when the system was last updated shortly before the LOI burn.
A new offset is uploaded to the VDL's computer, but it is thought safest to attempt a relative minor manoeuver to test the settings before proceeding further. At T+96:02 the orbit is changed to a near circular 95x100km.
With the ship now in a stable configuration, the Mission Controller delays the descent to give controllers time to rest and allow for further checks. Lessons have been learned from the early “Explorateur” flights, when controllers made quick (often too quick) adjustments and then continued with the mission as if nothing had happened. Modern flight plans allow time to solve some of these unexpected problems and flight rules now demand that controllers stop and diagnose faults that might threaten the mission, rather than relying on ad-hoc adjustments.
As part of these checks, engineers conduct a complete ground based computer simulation of the upcoming descent. The parameters of the ship's orbit, fuel loads, centre of gravity, inertia, control response times, engine performance and an exact copy of all the settings stored in the on-board computer’s memory are used to check how the VDL will behave during descent. The powerful ICL mainframes at Mission Control take the hundreds of numbers that define the properties of the VDL and use them to simulate every pitch, yaw, thruster firing, throttle response and dozens of other parameters of the ship’s behaviour as it travels down to the surface.
Trajectory designers have programmed the ship with an “ideal path” to follow; this is a smooth curve, gently bending down towards a vertical touchdown at the landing site. The path is carefully designed to allow the VDL to balance thrust, attitude and lunar gravity at all stages of flight, without the need for any quick changes.
Of course, this ideal solution will never occur in reality. The ship will always be a bit high, too slow, a different mass or not quite as responsive as the calculations predict. Consequently, the most important task of the guidance algorithm is to allow for these errors and steer the ship towards the ideal path, while also accepting input from pilots and updates from the landing radar.
It must also ensure that all manoeuvers happen quite slowly and gently; they cannot be allowed to call for say, a “snap turn” of 60 degrees. It takes the RCS thrusters time to torque the ship in any particular direction and to limit the buildup of any excessive pitch rates, the system is electronically restricted to a maximum rate of 5 degrees/second.
In addition to the ship’s position, velocity and acceleration, the computer has to consider two further parameters – “Jerk”, the rate-of-change of acceleration (essentially due to pitch rate) and “Snap”, the rate-of-change of rate-of-change of acceleration (dominated by the rate-of-change of pitch rate due to the action of the attitude control thrusters). Each of these parameters is a vector and is the differential of the previous one, and so it is possible for the computer to solve them numerically.
On Selene 4 the caution at Mission Control pays off. Controllers discover an erroneous zero setting of a target point in one of the guidance programs. If this zero had gone unnoticed, the effect would have been to cause the VDL to descend too quickly in the middle part of the landing. It would have crashed shortly after switching to the "Targeting phase”. A straightforward change of a number in one word of the computer's memory is all that is required to correct the problem.
The flight plan is resumed at T+104:12 and lunar descent orbit is entered at T+105:35, bringing the VDL to within 16.3km of the lunar surface when at perilune. After two orbits to refine the ground-based tracking solution, the computer is commanded to proceed with the landing.
At T+110:25:02, the "Pilotmode 5" guidance routine ignites the engine to start the 700km-long braking burn and descent towards the lunar surface.
For the first 20s, the engine fires at its minimum 25% throttle setting, before going to 100% for 10s, then back to 25% for a further 15s. Although not strictly necessary on an unmanned flight, these throttle changes serve to test the engine's performance early in the landing. If something went wrong at this stage, a crew could quickly and safely abort the descent and stay in lunar orbit.
The throttling and early phase of descent proceeds smoothly, albeit with the ship pitching under its control thrusters more than was expected. At 400s (6m 40s) into the descent, a manned flight would reach the Landing Decision Gate (LDG - the infamous "point of no return"). The VDL is performing well and if there had been a crew on board they would no doubt have been able to continue to attempt a landing.
90km out from the landing site, ground tracking suggests the ship will land short and the Targeting Pilot, Guy Larosse, is advised he will have to extend his glide slope (as it is still called, despite not being in any way a glide) in the final stages of the landing.
Twelve minutes 9 seconds into the descent, the ship reaches a point 10km from its programmed landing site and the on-board computer automatically switches to Pilotmode 4, the Targeting Phase. The VDL pitches down by about 40 degrees, bringing the landing site into view of the LPI. For these unmanned cargo flights a TV camera is mounted on this computer-controlled sight, which always points towards the current landing site and feeds images back to a controller on Earth.
This controller (who will always be a Selene astronaut) stands in a duplicate of a VDL cockpit and can instruct the VDL's computer to update its landing site based on the TV images being fed back to him from the Moon.
The fuzzy black-and-white image takes a moment to stabilise, but by 12m35s Larosse confirms he sees the surface. He recognises a small crater to the north of the landing site a few moments later. After a second or two he nudges his control stick eight times to re-designate the landing site about 800m downrange. Thinking he is still short of the site, he waits for nearly 15s to let the VDL properly stabilise onto its new trajectory, then enters 5 more clicks, 4 downrange and one to the south, before quickly clicking one back (to move uprange). Satisfied with the images he sees through the LPI, Larosse’s job is done. Although he continues to monitor the ship's progress down to the surface, ready to make further updates or even attempt manual control if the on-board systems fail, he enters no further updates and the VDL settles on its new course.
At 15m52s the computer switches to its final landing mode “Pilotmode 2” - to stop any residual horizontal motion and gently lower the VDL to the surface.
Contact probes and accelerometers indicate that the VDL touches down 16 minutes 11 seconds after Descent Engine Ignition, at a speed of just 1.2m/s. The engine shuts itself off 0.6s later and no further movement is detected after DPI+16:13.2.
At T+110:41:18, cheers and applause erupt around the control room at Biscarosse as the telemetry indicating touchdown is received. It quickly settles as mission engineers focus on the status of their new lander. Two minutes later, a mast carrying the panoramic camera is raised and starts to send higher quality images of the landing site back to Earth.
Earth based radio location and images from the camera soon confirm that the ship touched down a mere 70m from the planned landing site, which is about 90km southeast of the crater Copernicus. The performance of the systems and Larosse’s corrections during the descent were almost perfect. Calculations and telemetry agree that the VDL had 820-830kg of fuel remaining on board at landing, slightly higher than the expected margin. It could have flown for another 138s, sufficient to allow it to land (at most) 5km further downrange or 2km to either side. The final landed mass of 15,702kg is greater than all previous Selene, NASA and Soviet lunar landers put together.
The ship has successfully ridden its beam of numbers from 4,000mph down to zero, from ten miles up and 450 miles away to land within a few hundred feet of the pre-planned site.
Ten years after the Project was begun, the first lander large enough to carry a crew has reached the lunar surface.
Last edited:
