On February 10, 2022, we launched Launch Vehicle 0008 (LV0008). This was our first launch with a deployable customer payload and our first time launching from Cape Canaveral. After a nominal first stage flight, an anomaly occurred during the stage separation process which resulted in the upper stage not reaching orbit and the end of the mission. We immediately initiated our investigation process to determine the root cause of the anomaly. Now, we can share more about what we’ve learned to date.
Our investigation verified that the payload fairing did not fully deploy prior to upper stage ignition due to an electrical issue. The separation mechanisms (our fairing has 5 of these) were fired in an incorrect order, which resulted in off-nominal movement of the fairing that caused an electrical disconnection. Due to the disconnection, the last separation mechanism never received its command to open, which prevented the fairing from separating completely before upper stage ignition.
Separately, we discovered a software issue that resulted in the upper stage engine being unable to use its Thrust Vector Control system. This led to the vehicle tumbling after the off-nominal stage separation, and caused the end of the mission.
What We Learned:
The root cause of the fairing separation issue was an error in an electrical harness engineering drawing. This harness was built and installed onto the vehicle exactly as specified by our procedures and the engineering drawing, but the drawing error led to two harness channels being swapped. Prior to the LV0008 flight, we had conducted an end-of-line signal test to verify the separation system and ensure that the system was wired correctly. This test would have been able to detect an error in the harness build or installation, but it was unable to detect an error in the design. The swapped separation channels caused a different deployment sequence than we expected, and this led to the failure to open the fairing. We’ve been able to recreate the failure mode by conducting several experiments at our factory with real flight hardware, one of the benefits of having an active production floor with several launch vehicles in various states of production at the same time.
After determining the root cause of the software issue, we found that our flight control software was vulnerable to a specific “packet loss” failure mode. A missed series of signals resulted in a chain of events, resulting in the upper stage’s inability to recover from its tumble. Although we had designed our software suite to be resilient to packet loss, an unlikely combination of factors caused a failure that we didn’t predict. We have been able to use our hardware-in-the-loop simulator to step through exactly what happened and diagnose the root cause with high confidence.
How We Fixed It:
Through the investigation process we had identified two problems that needed fixing: the harness issue and the software issue. Soon after discovering the harness drawing error, we fixed the drawing and incorporated the change on previously built harnesses. We also implemented a new end-of-line signal test that will allow us to identify this class of issue in the future, if it were to occur, prior to launch. On the software side, we’ve introduced a trio of upgrades designed to make our system even more resilient to packet loss and other similar failure modes. Through constant iteration and extensive testing, we have been able to demonstrate that the changes eliminate the failure mode we saw on LV0008, while making the software suite much more robust.
Here at Astra, iteration and learning are core parts of our culture. I’ve been continuously impressed with the speed, passion, and diligence that the team showed as they worked through these complex issues to identify exactly what occurred and determine the right path forward to resolve each problem. With the root causes identified and corrective measures in place, we’re preparing to return to the launch pad with LV0009 soon — stay tuned!