Backend
Profile Screening and Frequency-Secure Operating Limit Derivation for Data Center Microgrids
arXiv cs.SY0 views
arXiv:2611.23504v1 Announce Type: new Abstract: Rapid power variations from artificial intelligence (AI) workloads pose frequency-security challenges for islanded data-center microgrids. This paper proposes a generator step- response-based framework to rapidly screen candidate load pro- files and derive frequency-secure operating limits for IT workload scheduling. Identified generator responses are superimposed to assess the cumulative frequency effects of successive load changes without repeated dynamic simulations. The framework identifies potentially unsafe intervals and determines allowable load magni- tudes, ramp rates, and minimum event spacing, while accounting for prediction uncertainty and governor limits. Validation on a 2.5 MVA islanded microgrid with eight generator configurations and 14 synthetic AI workload profiles shows that 94.75% of 40,208 assessment windows receive decisive screening labels, with no frequency-violating windows classified as secure.
Source:https://arxiv.org/abs/2611.23504PreviousAdaptive Line Switching for Cascade Mitigation in Power Systems via Graph Reinforcement LearningNextReinforcement Learning Optimized Flight Control for Interception of High-Altitude Maneuvering Targets
Related posts
I Built an Instagram Competitor Monitoring Workflow After Scraping Rate Limits Became Unmanageable
0dev.to
Back up an OpenLDAP directory: two exports, one reload, four checks
0dev.to
n8n error handling that actually wakes you up (and doesn't wake you up for nothing)
0dev.to
Google’s Mac Agent Puts Gemini in the Driver’s Seat
0dev.to
The button stayed put. An animation made our Android automation reject the tap anyway.
0dev.to

My AI Agent Passed 10 of 12 Tests. The 2 Failures Taught Me More Than the 10.
0dev.to