Condition-Based Maintenance Services Address Data Center Challenges
Vertiv
Vertiv argues that time-based preventive maintenance alone cannot keep up with the scale and density of AI-era data centres and proposes adding condition-based maintenance driven by equipment telemetry and machine-learning analytics.
On this page
Overview
The paper lists limits of current practice. Generic hardwired or IP alarms collected via a BMS give little condition information beyond output anomalies. MTBF or time-based replacement of capacitors, fans and filters cannot catch early-life failures and can replace parts earlier than necessary.
On sensing it favours factory-fitted sensors: retrofit third-party sensors lack history from factory and site testing, and machine-learning models only learn from their installation date. Secure IoT connectivity is a prerequisite, and offline data gathering is offered where secure connectivity is unavailable.
The analytics produce health scores, remaining useful life, anomaly detection and quality assessment. Alerts are triaged into three classes: trend, investigate at next visit, or immediate intervention, so that telemetry becomes tiered service actions.
The text is a service pitch and makes no numeric performance claim; its value lies in spelling out which data outputs and connectivity are prerequisites for condition-based maintenance.
Key points
- Generic BMS-collected alarms give little condition information beyond output anomalies.
- MTBF or time-based replacement of capacitors, fans and filters misses early-life failures and can replace parts early.
- Retrofit third-party sensors lack factory and site test history; ML only learns from the installation date.
- Analytics produce health scores, remaining useful life, anomaly detection and quality assessment.
- Alerts are triaged as trend, investigate at next visit, or immediate intervention.
- Secure IoT connectivity is required; offline data gathering is the fallback.
Using it in design
- Record telemetry capability (protocols, factory sensors, remote monitoring support) in the equipment catalog.
- Add to the design checklist an OT network segment and a secure remote-monitoring path for UPS, cooling and PDU fleets.
- Specify data output requirements at procurement.
- Support health-score and alert-tier fields per asset in the operations module.
Monthly newsletter · free
Keep your reading list current
Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.
SoSo CAD
Assess redundancy through scenarios
Compare normal operation and equipment failures in SoSo CAD, with supplied loads and capacity shortfalls in one view.
Request a demo
Who should read it
Operations managers setting maintenance strategy and designers writing equipment telemetry requirements.
Related standards
- IEC 62443
- ISO 55000
- EN 50600-3-1
Publication details
- Authors
- Vertiv
- Publisher
- Vertiv
- Year
- 2024
- Document number
- SL-71367
- Document language
- English
- Source type
- Manufacturer
- Access
- Free to read
- Licence
- Copyright retained; citation only — the publisher’s terms apply to the original document
Citation
Vertiv (2024). Condition-Based Maintenance Services Address Data Center Challenges. Vertiv. SL-71367. https://www.vertiv.com/48ee75/globalassets/documents/vertiv-data-driven-services-wp-en-71367-web.pdf
Events on this subject
Browse the events calendarAll content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.