Hybrid Model for Long-Term and Short-Term Power Demand 2 Forecasting in HPC Data Centers

Authors: Stefano Rinaldi, Chiara Franzoni, Salvatore Dello Iacono, Lavinia Chiara Tagliabue, Robert Birke, Silvia Meschini

Abstract: The power demand of High-Performance Computing (HPC) infrastructures exhibits both stable weekly regularities and rapid workload-driven fluctuations, which are difficult to capture reliably with a single modeling paradigm. Achieving more sustainable HPC op- eration requires accurate forecasts at multiple horizons: short-term predictions support operational control (e.g., proactive power capping and energy-aware scheduling), whereas long-term forecasts are essential for planning activities (e.g., capacity provision- ing and energy procurement). Together, these capabilities reduce operational cost and risk while enabling more efficient and sustainable datacenter management. This paper investigates multi-horizon forecasting of aggregated active power consump- tion in an operational HPC datacenter utilizing a four-month dataset (five-minute time intervals) from the University of Turin. We propose a hybrid residual-learning framework that integrates a long-term structural forecaster with a short-term residual corrector uti- lizing a Temporal Convolutional Network (TCN) to address the simultaneous presence of weekly regularities and short-term workload-induced fluctuations. Assessment utilizing a rolling-origin protocol covers a timeframe of 15 minutes to 6 hours and extends 1 to 3 weeks into the future. Performance of the proposed approach has been compared against the SARIMA baseline.

Keywords: HPC datacenters; Power demand forecasting; multi-horizon time series fore- casting; Residual learning; Temporal convolutional networks (TCN).