Schneider Electric Cooling Portfolio Jim Tubbesing Heath Wilson APC/Schneider North Texas Rep Tubbesing Solutions
Data Center Cooling Agenda 1. How to size data center, server rooms, IDF rooms, etc in 2015? 2. Changing Demand and how does that effect data center cooling: servers no longer are constant load, they are dynamic and therefore we need dynamic cooling to support the load 3. Different Technologies to solve ever changing loads Schneider Electric 2
Calculating Cooling Requirements 1. IT Equipment (70-80%) 1. Rack loads 2. Non-rack loads 2. UPS & Power Distribution (10-15%) 3. Lighting (10%) 4. People (2%) BUT calculating just the heat load does not solve the entire equation SEC Server Room Equipment List Location Division Status Chassis / Type Brand Model/T ype Power (kw) Frisco ME Active PowerEdge Server Dell R410 417 Frisco ME Active PowerEdge Server Dell R410 417 Frisco ME Active PowerEdge Server Dell R610 435 Frisco ME Active PowerEdge Server Dell R710 474 Frisco ME Active PowerEdge Server Dell R710 474 Frisco ME Active PowerEdge Server Dell R710 474 Frisco ME Active PowerEdge Server Dell R520 366 Frisco ME Active PowerEdge Server Dell R420 303 Frisco ME Active PowerEdge Server Dell R420 303 Frisco ME Active PowerEdge Server Dell R720 608 Frisco ME Active PowerEdge Server Dell R720 608 Frisco ME Active PowerEdge Server Dell R720 608 Frisco ME Active PowerVault StorageArray Dell MD3200 298 Frisco ME Active PowerVault StorageArray Dell MD3200 298 Frisco ME Active KACE KBOX K1100 Dell K1100 257 Frisco ME Active KACE KBOX K2100 Dell K2100 257 Frisco ME Active Network Switch Cisco 3750X 750 Frisco ME Active Network Switch Cisco 3750X 750 You must know the CFM at different loads Frisco ME Active Network Switch Cisco 2960G 370 Frisco ME Active Network Switch Cisco 2960G 370 Frisco ME Active Network Switch Cisco 2960G 370 Frisco ME Active Network Switch Cisco 3750G 147 Frisco ME Active Network Switch Cisco 2960G 370 Frisco ME Active Network Switch Cisco 2960G 370 Frisco ME Active Network Switch Cisco 2960S 370 Frisco ME Active Network Switch Cisco 2960X 370 Frisco ME Active Network Switch Cisco 2960S 370 Frisco ME Active Router Cisco 1800 45 Frisco ME Active Router Cisco 2911 210 Frisco ME Active Router Cisco 2921 210 Frisco ME Active Firewall Barracud a X300 450 Total 12119 Schneider Electric 3
Calculating Cooling Requirements 1. Below is a graph with different cfm for different types of equipment 12 10 Rack Power (kw) 8 6 4 2 0 0 100 200 300 400 500 600 700 800 900 1000 0 47 94 142 189 236 283 330 378 425 472 Effective Cool Airflow Delivered to a Single Rack (cfm) (l/s) Schneider Electric 4
Data Center Cooling Challenges 1 High Density Rack power densities of up to 50kW or more per rack thermally challenge traditional data centers that were designed for an average of 3kW per rack 2 Changing Demand Data center lifetime will span refresh cycles involving unknown load and therefore heat profiles Power/cooling Capacity Virtualized Load Original Load Virtualized Load 3 Dynamic Loads Virtualization and dynamic servers can cause roaming hot spots Virtualized Load Load Scale DOWN Virtualized Load Load Scale UP Load 4 Changing 20+ years of conventional thinking. Only one way to cool the data center Schneider Electric 5
Cooling Solutions: System Performance Characteristics Scalable solutions capable of handling IT equipment heat loads from 1 to 50 kw per rack Intelligent monitoring and dynamic controls system to adjust to fluctuating IT heat loads and airflow requirements Efficiently address data center cooling needs through the implementation of a hybrid architecture Minimize TCO through variable fan technology and right-sized components Heat removal at the source to eliminate mixing and ensure uniform inlet temperatures to the rack Schneider Electric 6
Raised Floor Architecture for Air Distribution Schneider Electric 7
Redundant Cooling Legacy Architecture Raised Floor Design Parameters: 4kW per rack / 40 racks total Total power 160kW Room size: 38' x 38' x 14' high HOT AIR HOT AIR COLD AIR/ PERFORATED TILES COLD AIR/ PERFORATED TILES COLD AIR/ PERFORATED TILES Schneider Electric 8
Redundant & Predictable Cooling Legacy Architecture - CFD Redundancy: Initial temperature gradient diagram shows all 5 CRAC units on Any unit failure results in loss of cooling to an area N+1 at room level does not always provide adequate cooling in failure modes Sectional Plane @ 5-6 from Raised Floor Schneider Electric 9
New Cooling Architecture Capable of handling high density racks Fundamentally not require a raised floor Modular Simple to design Predictable Quick to deploy High efficiency Delivering cold air is not the problem. Getting rid of the hot air is! Preventing hot air recirculation into equipment inlets prevents the servers from over heating. Schneider Electric 10
Drivers for Development of HACS/CACS The separation of hot and cold air is one of the most promising energy-efficiency measures available to new and legacy data centers today. Bruce Myatt of EYP Mission Critical The drivers are: High energy costs Accelerated energy consumption rates Schneider Electric 11
Benefits of Containment The Efficiency Benefits Cooling systems set to a higher supply temperature while still supplying the load with safe operating temperatures The elimination of hot spots Increased economizer hours Reduced humidification/dehumidification costs Better overall physical infrastructure utilization Schneider Electric 12
Cold-Aisle Containment Cold-aisle containment system (CACS) deployed with a room-based cooling approach A homegrown cold-aisle containment system Schneider Electric 13
Hot-Aisle Containment Hot-aisle containment system (HACS) deployed with row-based cooling Example of a hot-aisle containment system operating as an independent zone Hot-aisle containment system ducted to a remote air conditioner Schneider Electric 14
Containment s Effect on Work Environment Cold-Aisle Containment With cold-aisle containment, the uncontained area becomes the same temperature as the Hot Aisle CRAH CRAH Tape library Cold aisle containment Cold aisle containment Storage Uncontained Work area becomes hot aisle Cold aisle containment Cold aisle containment Non-racked equipment ingesting higher temperature air CRAH CRAH With CACS, high temperatures can be problematic for IT personnel stationed in the data center Hot-Aisle Containment With hot-aisle containment, the uncontained area becomes the same temperature as the Cold Aisle CRAH CRAH Hot aisle containment Hot aisle containment Tape library Storage Uncontained work area becomes cold aisle Hot aisle containment Hot aisle containment Non-racked equipment ingesting air at same temperature as servers CRAH CRAH With HACS, high temperatures stay confined to the hot aisle and do not affect IT personnel stationed in the data center Schneider Electric 15
Predictable Cooling Eliminate Mixing Close Coupling of Row Cooling keeps the hot air in the hot aisle Cooling Units Traditional Approaches Allow cold and hot air streams to mix Cooling Units Target the Heat to Eliminate Hot Spots which can cause servers to overheat Schneider Electric 16
Predictable Cooling Intelligent Control Active Response Controls Increase Availability by Actively Responding to thermal Changes Temperatures change entering IT Equipment InRow temperature probes Sense changes and send signal to controller Controller adjusts cooling capacity to balance with the heat load Active Response Controls ensures IT equipment is kept at the proper temperature Schneider Electric 17
Close Coupled Cooling 30% More Efficient Component Power InRow Air Handler Computer Room Air Handler Units Assumptions: 750 kw IT Load All Systems at 100% Sensible Cooling Load Only Chilled Water Pump Cooling Tower Pump Fans 30.6 88.0 kw 10.2 11.0 kw Chiller 83.9 94.7 kw 18.5 18.5 kw Cooling Tower 16.2 18.3 kw Total Power 159.3 230.5 kw 65 % 7 % 11 % - 11 % Efficiency Metric Annual Operating Cost 0.21 0.31 139,572 201,887 $ USD Energy Efficient Cooling for Data Centers: A Close-Coupled Row Solution Schneider Electric 18
Flexible - Install in any IT Environment With APC Racks Without APC Racks New Data Center (Green Field) Existing Data Center (Brown Field) Schneider Electric 19
Fan Power (%) Energy Efficient Variable Capacity Control 750 kw Data Center Example Operating at 80% Variable Speed Fans Constant Speed Fans 40 % Reduction Fan Power Component Power InRow Air Handler Computer Room Air Handler Units 20% Fans (750 kw IT Load) 30.6 88.0 kw IT Load (kw) Variable Cooling Capacity Follows The IT Load Fans 18.4 88.0 kw (600 kw IT Load) SAVINGS 12.2 0 kw Reduces Load on Chiller Schneider Electric 20
Conclusion HACS vs. CACS The key to efficient cooling strategies is the prevention of hot and cold air mixing Both HACS and CACS offer improved power density and efficiency when compared with traditional cooling approaches HACS is more efficient than a CACS HACS can save 43% in annual cooling system energy cost with a 15% reduction in annualized PUE compared to CACS while holding the uncontained data center area to 24 C/75 F HACS should be the default containment strategy for all new data center designs Schneider Electric 21
Schneider Cooling Offering Uniflair CRAH/CRAC InRow Self-Contained Room Air Distribution EcoBreeze Chillers Heat Rejection Room Air Removal Units Rack Air Distribution InRow Chilled Water InRow DX Pumped Refrigerant Thermal Containment Schneider Electric 22
Thermal Containment Schneider Electric 23
Traditional Cooling Best-In-Class efficiency Standard Features Electronically Commutated (EC) fans Low coil impedance High Sensible Heat Ratio (SHR) Fan speed optimization Active stand-by mode 60-70% power reduction compared to legacy CRAH/CRAC units Schneider Electric 24
The Uniflair Difference High efficiency in all operating conditions Optimized air flow management Compatibility with high-density cooling solutions Compact dimensions Innovative solutions Maximum reliability Schneider Electric 25
Features & Benefits Total Cost of Ownership Fans controlled to minimum speed required to achieve target cooling EC fans higher efficiency than even VFD driven fans Compact dimensions minimize cooling footprint in IT space Flexibility Available in upflow and downflow with multiple airflow options Adaptive controls for various types of installation Availability Dehumidification operates without reduction in airflow DDC modulation of 2 or 3-way CW valve for capacity control Serviceability Front panels can be opened without special tools Normal maintenance requires front access only Push-button electrical panel catch opens without airflow disruption Schneider Electric 26
Efficient - Components 1 Backward Curved EC Fans (EBM Pabst) 11 2 Immersed Electrode Humidifier (Carel) 3 30% Efficiency Air Filters (MERV 7) 5 3 6 8 7 4 2 or 3-Way Valve & Actuator (Schneider Electric) 5 Cooling Coil (Heatcraft/Luvata) 6 Temperature and Humidity Sensor (Carel) 4 9 7 8 Manual Disconnect Switch (Schneider Electric) User Interface (Carel) 1 2 10 9 Microprocessor Controller (Carel) Pictured: TDCV2500 (25-Ton est. capabity) 10 Integrated Airflow Switches 11 Front Service Access Schneider Electric 27
Summary 1. Sizing, need to ask for the server information, not just the watt load. Call TSC if you need help. 2. As data centers continue to evolve, new technology is going to be needed to insure the most effiecent solution is created. 3. Remember, APC has all the different technologies to solve these issues. Schneider Electric 28