Speaker
Description
The European Spallation Source, now being commissioned in Lund, will be the most powerful neutron source in the world. The volume of operations data it generates ranks among the largest at any accelerator facility. Temperatures, vacuum pressures, magnet currents, RF cavity phases, beam positions, interlock states — the control system records roughly 1.2 million distinct signals, and, under formal stewardship directives, keeps them for the operating lifetime of the facility. The archive rests on about 3 PB of long-term storage fed by a tiered NVMe and disk infrastructure, and it grows continuously while the machine is still ramping toward full power.
This talk describes what that infrastructure looks like — EPICS, the archiver appliance, the event-based Synchronous Data Service, and the operator tools built on top of them — and what the data is actually used for today. The honest answer is that it is used to run the machine. Shift leaders log faults through a tracker that has become part of the daily routine. Machine-protection events are retrievable as complete synchronised snapshots rather than scattered log lines. Live data feeds the physics models used to commission the linac.
Interest in doing more is real and organised, through research collaborations with universities and industry and a growing set of machine-learning projects working against control-system data. The wider field has already shown the shape of what an archive like this can support: classifying faults, predicting errant beam before it is lost, and moving operation from reactive intervention toward prediction and, in time, a degree of autonomy. Almost none of that has yet been done with this archive as a whole — and that is not an accident.
ESS is commissioning a machine that has never been built before, expected to run at the limits of what is physically possible, driving protons into a tungsten target with 95% availability. When you cannot know in advance what will fail or why, you instrument everything. The archive is what that necessity leaves behind: built to answer the questions its experts already knew to ask, which is why it is not yet shaped for the ones nobody has asked.