In a room illuminated by blinking lights and glowing monitors, more than 2,000 synchronized computers are triggered to run 5 million lines of code. The intricate code language is responsible for aligning and firing 192 laser beams — and carrying some 800 channels of target diagnostic data — efficiently and reliably several times a day. This isn’t a scene from a science fiction movie: it’s the control room of LLNL’s National Ignition Facility (NIF).
The Computational Engineering Division (CED) within LLNL’s Engineering Directorate, is a critical but largely invisible element of the effort toward achieving nuclear fusion ignition. CED’s role in supporting NIF’s fusion energy goals is writing the algorithms that direct the lasers toward the hohlraum target and for various diagnostics that help determine how to make sense of NIF shots and to keep the fusion apparatus running.
The Automatic Alignment team, led by Vicki Miller Kamm within CED’s Image Analysis and Controls Group, builds automated software that aligns the 192 NIF lasers to the target before every shot. Critical to running NIF, this team develops automated image-processing routines that calibrate the laser’s position using thousands of images taken throughout the facility. Automatic alignment algorithms analyze those images to provide beam positions, which the automatic alignment control system uses to move motorized mirrors that direct the laser beams.
Each of the images helping to align the lasers may require multiple algorithms to account for different disruptions that the image may be subjected to, including noise, intensity fluctuations or intensity gradients. If algorithm 1 fails, algorithm 2 may take over; if algorithm 2 fails, algorithm 3 takes over; if algorithm 3 fails, then the automatic alignment system declares a failure. Multiple beam failures may derail a planned NIF shot, so the reliability and speed with which the algorithms run is important for the day-to-day operation of NIF.
And, as the NIF is a dynamic system whose operating conditions are constantly changing, the automated alignment team’s adaptability is crucial to NIF’s continued functioning, too.
“When we think of NIF, we think of 192 beams impinging on a target that is the size of a millimeter,” noted CED’s Abdul Awwal, one of the top experts on the automatic alignment algorithms for the NIF beam lines. “But when I joined the automatic alignment team in 2003, NIF only had four beams. In those days, if the alignment failed on any beam, you were called to the control room to look at the failure and make any code changes on the fly.”
Now, Awwal and his team have a systematic process involving development, testing, release and online testing, giving engineers the opportunity to anticipate and correct any code errors before they affect the shot environment. In addition to refining testing processes, Miller Kamm’s team has over the last several years sped up the alignment process, and the latest power improvements were instrumental in the success of the Dec. 5 ignition shot.
CED engineers’ work controls photographic processes not just for alignment, but in the area of diagnostics as well. These photographic diagnostics chronicle what happens during a given shot at the NIF by using high-speed complementary metal-oxide semiconductor (CMOS) sensor technology that can take multiple frames for a combined exposure time of one nanosecond. The team responsible for creating these diagnostics is the detectors team within the Target Science and Engineering (TASE) group.
One of the diagnostics that the TASE group is responsible for — the Gated Laser Entrance Hole image (or GLEH-2 — takes pictures of the hohlraum at an angle to help scientists understand what is happening inside the target chamber during a shot. What makes this especially challenging in the case of high-yield shots is that electromagnetic interference and neutron radiation can interfere with the functionality of a camera. As electrical engineer Brad Funsten of the TASE group puts it: “Neutrons can cause what is called displacement damage whereby digital circuitry in the camera might switch to an unintended logic state due to neutrons building charge in the electronics. So, the objective is to design the camera’s hardware electronics to attempt to mitigate these effects.”
Building in some redundancy in the digital circuitry and using older-generation electronics are some of the ways in which this is accomplished. Funsten designs hardware code for the camera system’s single master digital controller, the Field Programmable Gate Array, which communicates with a CMOS sensor to understand the behavior of the target.
While NIF’s many diagnostics filter, amplify, shield or transport the raw data in the process of making it discrete and meaningful, the presence of these diagnostics can itself affect the data collected. Judy Liebman of CED leads the NIF Automated Analysis Signal and Image Processing Team, which builds automated analysis codes that provide consistent and fast experimental results after every NIF shot. The team also writes custom mathematical analysis codes that strip the data of these effects and recombine it to approximate the state of the data minus any interference from the diagnostic apparatus.
As with many aspects of a NIF shot, a wide variety of hardware on its diagnostics needs to be precisely calibrated to produce accurate diagnostic results. This challenge is heightened by the frequency with which hardware is swapped out and replaced to record data from various experiments.
Considering the complexity of the undertaking, achieving ignition had long seemed like a distant goal for many at the Lab. As Awwal noted: “People retiring from NIF would say that they’d be watching the news for when ignition happens. I used to think, ‘Who knows if we will achieve it before my own retirement?’” An answer in the affirmative was a welcome career milestone.
thomas244 [at] llnl.gov