報告_OCP2026_Schneider_800VDC保護與協調_20261015

PDF 原檔：報告_OCP2026_Schneider_800VDC保護與協調_20261015_original.pdf

講者公開簡報（OCP 官方 Recorded Content 頁提供下載）。文字以 PyMuPDF 逐頁抽取，### p<n> 為頁碼；圖表內容見原檔。

圖片清單（已驗證 2026-10-19）

原始內容

p1

800 VDC POWER SYSTEMS 800 VDC changes the protection strategy. Coordinate protection, fault isolation, and system safety—from source to load. PROTECTION • COORDINATION • SYSTEM SAFETY Manav Patel VP, Strategy · Schneider Electric Low Voltage Products & Systems Tony Landry NAM Leader · Schneider Electric Power Systems Competency Center AI OPEN DATA CENTER

p2

DATA CENTER FACILITY - POWER DISTRIBUTION Sub-Project

p3

AGENDA Three lenses for predictable 800 VDC protection Architecture sets the structure. Coordination defines the response. Safety validates the system. 01 ARCHITECTURE System-Level Protection Architecture Coordinate fault detection, performance targets (e.g clearing time), and system recovery across the full 800 VDC system. → DESIGN SYSTEM BEHAVIOR 02 COORDINATION DC-Specific Coordination Challenges Address absent zero-crossing, persistent DC arcs, and selective isolation with validated coordination. → COORDINATE FAULT RESPONSE 03 SAFETY Integrated Safety Frameworks Extend standards to interoperability, fault behavior, and resilience under transient conditions. → VALIDATE END-TO-END SAFETY

p4

01 System-Level Protection Architecture Moving beyond device-level protection to coordinated system design

p5

The design question has changed From “which device is fastest?” to “which system response is predictable?” Yesterday’s question • Which breaker clears fastest? • Which device has the highest rating? • Which curve coordinates on paper? 800 VDC reality • Fault current is shaped by converters. • Stored energy dominates the first microseconds. • Architecture determines what must remain online. The right question • Can the entire source, protection, grounding and load system respond predictably under fault? Predictable outcomes come from co-design and system validation, not device speed alone.

p6

System-first by design From “which device is fastest?” to “which system response is predictable?” 1 ARCHITECT • Set redundancy & recovery strategy • Grounding scheme • Source/load behavior 2 CO-ENGINEER Match protection to converter fault behavior and system energy • Optimized: Current controlled behavior integrated within the converter, paired with proven electromechanical breakers • Design-minimized: external SS protection distributed without that co-design 3 VALIDATE Use digital models, HIL or laboratory to prove selectivity, stability and recovery before deployment.

p7

System-first by design From “which device is fastest?” to “which system response is predictable?” CO-ENGINEERED SYSTEM • Protection matched to converter behavior • System-level selectivity proven before deployment • Conventional and solid-state devices used where each adds value DESIGN-FREE ASSEMBLY • Devices added after architecture decisions • Speed optimized locally, not necessarily system-wide • Higher risk of over-specification or mismatched behavior • Conventional EMCB and ETU technologies can deliver full selectivity with controlled or current limiting sources • SSCB become decisive when self-protecting sources demand fast isolation and rapid shutdown if architecture topology do not allow enough hold up time

p8

02 DC-Specific Coordination Challenges and Examples Why AC coordination logic doesn't transfer directly to 800VDC

p9

Protection coordination is a system outcome Source behavior, protection technology, and distribution architecture must work together. 01 SOURCE BEHAVIOR Converter topology & impedance • Current limiting—level or duration • Under-/overvoltage self-protection 02 PROTECTION TECHNOLOGY Fuse • Electromechanical circuit breaker • Hybrid circuit breaker • Solid-state circuit breaker 03 ARCHITECTURE TOPOLOGY Grounding scheme • Redundancy—N+1, 4-makes-3, ATS, STS… OPTIMIZED PROTECTION Coordination • Cost Integration • Readiness Maintenance ONE COORDINATED SYSTEM Design the system—not the device in isolation. Controlled fault current (DC UPS, some SST): traditional and solid-state protection can both react in time.

p10

There is no single “800 VDC fault” The waveform depends on the source and the energy already stored in the system Controlled fault current (DC UPS, some SST): traditional and solid-state protection both react in time. Red, Blue, Green curves “Non-Limiting Source” « Inductive » type of current waveforms following an exponential growth which is linked a time constant of L/R. This type of current is provided by sources such as PV, Battery, Fuel Cell, Rectifier, and is the type of waveform UL standard are requesting for short-circuit tests Purple and Orange curves “Limiting Source” « Capacitive » type of waveform which can be found at the output of an active converter (DC UPS, SST, …). The waveform is usually divided in 2 steps, a capacitor discharge with a high rate of rise of current following by an infeed current from the upstream source, limited or not. Currently, Fuse and Electromechanical Circuit Breaker are tested and certified only following Non-Limiting type of circuit. This may lead to misunderstanding of protection performances. Implication: Test waveform and source control behavior must be explicit in every coordination study.

p11

Protection choice is a four-way trade-off — not a speed race Faster interruption improves isolation, but cost, integration effort and readiness determine what can be deployed at scale FUSE Traditional / aR EMCB Magnetic or ETU HCB Hybrid breaker SSCB Solid-state breaker BREAKING TIME COST INTEGRATION EFFORT TECHNOLOGY READINESS 0.5–5ms 1–20ms 0.2–1 ms < 10 μs FASTER ISOLATION → Low Low Low Low Medium–high High High ESTABLISHED Standards gap for branch protection above 600 Vdc Limited protection features ESTABLISHED Performance remains source- and waveform- dependent DEVELOPING Topology, controls and fault behavior must be qualified ADVANCING Cooling and economics still getting shaped Select at design phase & optimize the system for key factors: Match the device to source behavior, protected boundary and rack architecture — not breaking time alone Medium Breaking times mentionned in this table refers to branch protection device

p12

Architectural Redundancy Decides What's at Stake Protection speed changes recovery in every topology — but only one architecture puts the IT load itself at risk. REDUNDANT PATHS Multi-corded / redundant racks IT load stays online Redundancy guarantees load continuity; protection speed determines how quickly the affected powertrain recovers. TRANSFER-DEPENDENT PATH Single-corded racks with transfer schemes IT load is on the clock With no rack-level redundancy, clearing must complete inside the transfer and rack hold-up window. Favors uptime by design Fault clearing < transfer < rack hold-up Miss the window → IT load loss; simplicity comes at uptime's expense.

p13

The load stays online — even if one source shuts down Redundancy preserves load continuity; coordination isolates the fault and restores the source. DC DC P1 D 50 V C B A IT Rack #1 D 50 V C B A IT Rack #2 P Low-medium-rated OCPD Fuse, EMCB, SCCB(CR), SCCB, SCHCB P High-rated OCPD Fuse, EMCB, SCCB(CR), SCCB, SCHCB 800 V DC source #A 800 V DC source #B 1 2 3 P1 P1 P1 PR PR PR PR PR PR PR PR SOURCE MAY SHUT DOWN —THE LOAD STAYS ON The critical sequence is to isolate the fault, then restart the affected source. A Source shuts down Without coordination, the source may trip while the redundant path keeps the load energized. B Isolate the fault • Open the correct protection device and contain the faulted section. • Size the upstream OCPD to trip—even if the source isolates first—to speedily identify fault location and repair. C Restart the source Restore automatically after isolation where the control scheme supports it. SEQUENCE → isolate • verify • restart OUTCOME → load remains online Load Continuity: Ensuring IT load is supported (not dropped) *Powertrain Continuity: Localizing faults and minimizing disruption and time-to-repair within the affected/faulty powertrain

p14

Grounding Scheme: Uptime vs. Simplicity The grounding scheme sets the first-fault behavior before protection technology is selected. UNGROUNDED (IT) First fault → keep operating Uptime advantage, with active monitoring SOLIDLY GROUNDED (TN-S) First fault → trip immediately Simple clearing, with an uptime trade-off • Keeps running through the first insulation fault, creating time for a planned repair. • Requires continuous insulation monitoring to detect a second fault before it compounds. • Clears the first fault immediately with familiar protection behavior. • Trips the powertrain on any first fault; architectural redundancy must carry the load. Choose grounding first: IT favors uptime; TN-S favors simplicity. Y Y X X DC Rack Branch Distribution Side DC Z Y Y X X DC Z Z Z Z Z Source to be modified for SE DC UPS for specific schemes

p15

Grounding Scheme – Impact on Protection Selection Grounding topology determines whether an SSCB can be applied — and how many poles it must protect SSCB Topology Solidly Grounded Unipolar · 2 Wires Solidly Grounded Bipolar · Midpoint grounded distributed (3 wires) or not (2 wires) Ungrounded / HRG Unipolar · 2 Wires Ungrounded / HRG Bipolar · 3 Wires One Protected Pole Better cost, footprint & efficiency Two Protected Poles Higher cost, footprint & losses CAN BE USED NOT SUITABLE Must protect and/or disconnect 2 wires (+ and −) CONDITIONAL Opens only after 1st-fault detection by monitoring device to detect and open at first fault; needs validation & interoperability certification NOT SUITABLE Must protect and/or disconnect all wires CAN BE USED CAN BE USED CAN BE USED If one pole can operate at system voltage level NOT SUITABLE Must protect and/or disconnect all wires Can be used Conditional Not suitable Two protected poles unlock the widest grounding compatibility but has implications to cost, footprint and efficiency Grounding Scheme Variations

p16

03 Integrated Safety Frameworks Expanding standards toward end-to-end system safety

p17

Device standards are necessary. System standards make safety predictable. Safety must be verified through whole-system behavior under faults, transients and component interactions. PRODUCT STANDARDS ALREADY COVER A strong device-level baseline SYSTEM-LEVEL QUESTIONS REMAIN Predictability is still the gap • Liquid-cooling acceptance and test protocols remain unresolved • Converter–protection interactions are not certified as one system • Standards bodies need direct engagement while requirements are still forming SYSTEM STANDARDS MUST DEFINE THREE OUTCOMES INTEROPERABILITY Converters and protection coordinate predictably across the full architecture. FAULT RESPONSE The system response is defined and testable from the instant a fault occurs. TRANSIENT RESILIENCE Safe operation holds through switching transients, load steps and abnormal conditions. Shift the certification question from “Does the device pass?” to “Does the system behave predictably?” • Fuses & breakers are not tested according to system behavior • DC ATS/ STS standards need to be evolved to match application requirements • Switchboard & busway limited 1000 Vdc

p18

What the ecosystem needs next A shared evidence model for 800 VDC system safety 1 COMMON FAULT PROFILES Agree source and stored-energy waveforms for qualification. 2 DECLARED RESPONSE CONTRACTS Specify detection, limiting, isolation and recovery behavior. 3 SYSTEM-LEVEL VALIDATION Prove coordination with models, HIL and representative hardware. 4 OPEN INTERFACES + DATA Enable interoperable protection by enabling convergence between conversion & protection parameters and intedependencies Open standards should define the behavior to prove, while allowing multiple technologies to compete.

p19

CALL TO ACTION Validate the system before you scale it Move from device selection to coordinated architecture, ecosystem readiness and field-proven deployment. 1 DEFINE System KPIs Set system-level performance targets, then optimize the architecture to deliver them. 2 ALIGN & UPSKILL The ecosystem Amend codes-and-standards, upskill consultants, contractors and AHJs for safe deployment. 3 MATCH Readiness Stage the rollout so technology maturity and personnel capability advance together. 4 PILOT Before scale Validate the complete system in a pilot and resolve practical challenges before deployment. Define → align → match readiness → pilot → scale At 800 VDC, the winning protection strategy is not the fastest device. It is the most predictable system.

p20

DATA CENTER FACILITY – LOW VOLTAGE DIRECT CURRENT POWER DISTRIBUTION