Speaker
Description
Following discussions at ARW2024 and continued exchanges with HIPA users at PSI, we modernized the reliability and availability reporting framework of the PSI High Intensity Proton Accelerator (HIPA) to better reflect experiment‑relevant performance and improve trust in the published statistics. Traditionally, availability was reported only for beam currents above 1000 µA, and outages were logged in detail only when exceeding five minutes.
User feedback highlighted strongly differing sensitivities: some experiments depend mainly on delivered charge, others require sustained high current, and many are particularly impacted by interruptions longer than ~30 minutes. In response, starting in 2024 we refined outage classification by introducing a dedicated category for interruptions longer than 30 minutes to distinguish high‑impact events from the broader set of >5‑minute outages. Beginning in 2025, we expanded reporting beyond fixed thresholds by collecting and analysing the full distribution of trip and outage durations, enabling advanced visualizations (e.g., Pareto analyses) that reveal frequency and operational relevance across timescales.
For 2026, the framework converges toward standardized, user‑aligned definitions and reporting logic:
(i) two availability figures at MHC1 will be published for low current (1000 µA) and high current (1450 µA),
(ii) the reporting “day” follows the operational scheduling boundary (08:00–08:00), and (iii) a single documentation threshold of 15 minutes is used to reduce reporting overhead while preserving the dominant downtime and cause picture.
Users also pointed out that rapid sequences of consecutive trips, with time‑to‑failure intervals below ~30 seconds, are effectively indistinguishable to their instrumentation and behave as a single extended interruption. Starting in 2026, such clusters are aggregated and reported as one outage, and short intervals of nominal operation between them are no longer counted as productive uptime.
This presentation will introduce the updated definitions, reporting logic, and latest operational results, and will report progress toward automated cause classification of short interruptions using interlock system information and enhanced logic.