GT Events & Conferences

Data Center GT Webinar Japan

Data Center GT Webinar Japan

Event Date(s)

Wednesday, December 2, 2026

Location

Online Webinar + QA

Time Zone

Asia/Kolkata

最新のAIクラスターでは、負荷がミリ秒単位で急変することがあり、突入電流、過渡的な電圧降下、系統側のフリッカなどによって、液冷ループと電力系統の双方に大きな負荷がかかります。このような現象は定常状態の容量設計では捉えられず、PUE目標の達成も難しくなります。本ウェビナーでは、チップ・コールドプレートからCDU、チラー、UPS/BESS、上流の電力系統までを一つのGT-SUITEモデルで連成する方法をご紹介します。負荷の急増・急停止時にCDUの熱応答、制御、電気過渡がどのように相互作用するかを解析し、稼働を守りつつPUEを目標内に保つ制御戦略と非常時対応を設計する手法を解説します。

_________________________________________

In modern AI clusters, workloads can shift abruptly—often within milliseconds—triggering phenomena such as inrush currents, transient voltage drops, and power grid flicker. These events place significant stress on both liquid-cooling loops and electrical power systems. Because such dynamics cannot be captured through steady-state capacity design, they also make it difficult to meet PUE (Power Usage Effectiveness) targets. This webinar introduces a method for integrating components—from chip cold plates and CDUs (Coolant Distribution Units) to chillers, UPS/BESS units, and the upstream power grid—into a single, coupled GT-SUITE model. We will analyze how thermal response, control systems, and electrical transients interact during sudden load spikes or rapid shutdowns, and demonstrate how to design control strategies and emergency response protocols that ensure operational stability while keeping PUE within target limits.

The webinar will be held in Japanese.