Efficiency of Data Center cooling Comparison of data room layouts and cooling systems Bohumil Cimbal Product manager for Cooling Systems in CONTEG Company
Objective Basic questions between investor and cooling designer What DC arrangement is preferred? What is the requested cooling capacity? What temperatures in DC? How to increase Cooling efficiency?
Content Datacenter as a heat source Cooling effectiveness Methods to increase effectivity DC room layouts and cooling systems Case study comparison
Data center cooling
Data center cooling
Increasing effectivity Measure and calculate efficiency awareness is a good start Application of power saving components fans, compressors, pumps, EC motors, etc. Use of sophisticated logic control software, communication of all parts in system, free-cooling, pressure control, etc. Set the correct temperatures air, water, refrigerant
Data center components Power consumption
Power usage effectiveness Power usage effectiveness TOTAL FACILITY POWER Power delivery components (UPS, generators, PDUs, batteries, and distribution losses external to the IT equipment) Cooling system components (chillers, air conditioners, pumps, and cooling towers) Compute, network, and storage nodes Other miscellaneous component loads (lighting, cleaning...) IT EQUIPMENT POWER Load associated with all of the IT equipment Computer, storage, and network equipment Supplemental equipment (monitors, and workstations/laptops) used to monitor or otherwise control the datacenter
Power usage effectiveness
Ideal Temperature in DC Different points of view IT equipment Comfort x Cooling Energy Efficiency
Ideal Environment in DC From IT point of view New ASHRAE Classes 2011 2011 Thermal Guidelines for Data Processing Environments Expanded Data Center Classes by Technical Committee 9.9 Ti = 18 27 C ϕ = 30-60%
Ideal Temperature for IT Server power consumption increases at higher ambient temperatures
Ideal Temperature for IT
the colder the better...
Ideal Temperature in DC From Energy point of view Benefits of higher temperature: Higher capacity of heat exchanger Thermal losses to the surround of DC Friendly compressor circuit conditions Higher efficiency of cold source Long free-cooling utilization Air temperature increase AC energy savings 1 K 4 % 2 K 8 % 3 K 12 % 4 K 16 % 5 K 20 % Source: Schweizer Bundesanstalt für Energiewirtschaft
Ideal Temperature in DC Temperature impact on Cooling capacity (kw) Return air Temperature Water temperature 6/12 C 10/16 C 12/18 C 15/21 C 45 C 81,8 72,4 67,7 60,7 40 C 70,1 60,7 55,9 48,9 35 C 58,3 48,9 44,2 37,1 30 C 46,5 37,1 32,4 25,3 25 C 37,1 27,7 23,0 15,9
Ideal Temperature in DC 22% annual energy consumption difference Ti = 25 C Ti = 18 C
the hotter the better...
Hot and Cold Air Separation Right conditions for computers + Maximal energy efficiency
Blanking panels and Separation frames SIDE VIEW BLANKING PLATES
Racks in open aisle
Not enough air
Too much air
Optimal air flow Impossible to keep steady Servers change the air-flow continually
Mechanical Air Separation with air flow control The only possibility!
Data Center Room Arrangements Plenum Feed With Room Return Cold aisle Containment w. Plenum Feed or In-row Cold aisle Containment w. Top cooling Room Feed with Plenum Return Closed aisle Containment w. In-row Units Modular Closed Loop w. In-row Units
Data room Cooling systems Comparison CRAC x In-Row x Topcooling
DC arrangement Cold and Hot Aisles
CRAC = Computer Room Air Conditioner
Air path length CRAC solution Cold air delivered under floor Long way to servers
In-Row Air-conditioning units integrated into rows of racks
Air path length In-ROW solution Deliver cold air where required front of servers
Top cooling Air-conditioning units on top of racks
Air path length Topcooling solution Deliver cold air where required front of servers
Temperature zones In-ROW solution Easy to plan different power and temperature zones CRAC solution Only one temperature in all zones in one room
Easy future enlargement In-ROW solution Possibility of DC enlarging in steps - minimizing initial investment CRAC solution Big initial investment Low initial efficiency
Basic features comparison Short air path to servers Easy to plan different power and temperature zones DC enlarging in steps - minimizing initial investment Open or Closed Architecture Cold or Hot Containments Technical service out of DC room Water in DC room Total power consumption of indoor units CRAC In-row Top cooling
Case study Financial effect: 1. Floor area savings 2. Energy savings Example: - 16 racks 600mm, depth 1000mm, height 42U - contained cold aisle - 35 C in hot zone, 25 C in cold zone - Chilled water system (10/15 C) - heat load 6kW/rack (total demanded cooling capacity 96 kw) - requested redundancy n+1
Case study CRAC (CW) In-Row (CW) CoolTop (CW) 3 CRAC units cooling capacity 53 kw air flow 9.000 m 3 /h dimensions 950 x 900 mm consumption 1,8 kw Occupied floor area = 2,6 m 2 Total consumption 3,6 kw (2 running units) 6 in-row units cooling capacity 21 kw air flow 3800 m 3 /h dimensions 300 x 100 mm consumption 0,77 kw max (0,3 kw at capacity 96/6=16 kw per unit) Occupied floor area = 1,8 m 2 Total consumption 1,8 kw (6 low-speed running units) 4 Topcooling units cooling capacity 38 kw air flow 7.700 m 3 /h) dimensions 2400 x 600 mm consumption 0,7 kw max (0,2 kw at capacity96/4=24 kw per unit) Occupied floor area = 0 m 2 Total consumption 0,8 kw (4 running units)
Case study CRAC (CW) In-Row (CW) CoolTop (CW) 3 CRAC units cooling capacity 53 kw air flow 9.000 m 3 /h dimensions 950 x 900 mm consumption 1,8 kw Occupied floor area = 2,6 m 2 Total consumption 3,6 kw (2 running units) 6 in-row units cooling capacity 21 kw air flow 3800 m 3 /h dimensions 300 x 100 mm consumption 0,77 kw max (0,3 kw at capacity 96/6=16 kw per unit) Occupied floor area = 1,8 m 2 Total consumption 1,8 kw (6 low-speed running units) 4 Topcooling units cooling capacity 38 kw air flow 7.700 m 3 /h) dimensions 2400 x 600 mm consumption 0,7 kw max (0,2 kw at capacity96/4=24 kw per unit) Occupied floor area = 0 m 2 Total consumption 0,8 kw (4 running units)
Case study Financial effects: 1. Floor area savings Floor area per unit Number of units Occupied floor area Floor area price Financial loss (m 2 ) (pcs) (m 2 ) ( /m 2 /year) ( /year) Top cooling 0 4 0 10 000 0 In-Row 0,3 6 1,8 10 000 18 000 CRAC 0,9 3 2,7 10 000 27 000 Expected price of one footprint (0,6m 2 ) is 500 per month.
Case study Financial effects : 1. Energy savings Indoor units consumption Number of units Annual power consumption Energy price Annual costs (kw) (pcs) (kwh) ( /kwh) ( /year) Top cooling 0,2 x 4 = 0,8 4 7008 0,15 1 051 In-Row 0,3 x 6 = 1,8 6 15768 0,15 2 365 CRAC CW* 0,7 x 3 = 2,1 3 18396 0,15 2 759 CRAC CW** 1,8 x 2 = 3,6 3 31536 0,15 4 730 * Redundant units are working (partial operation). ** Redundant units are stand-by.
Annual savings ( ) 30000 25000 20000 15000 10000 5000 0 Top cooling In-Row CRAC
Summary DC concept Business plan in time, sizes, capacities Cooling design cooling system, CFD simulation, critical situations, TIER classes, safety points Building process product quality, piping system, monitoring Operation air separation, temperature setting, efficiency tracing
Thank you for your attention Dipl.Ing. Bohumil Cimbál B.Cimbal@conteg.cz www.conteg.com