Six Reasons Why Modular Power and Cooling Plants Will Make Traditional Data Center Designs Obsolete

Posted by on April 22, 2012

Neil Rasmussen (Senior Vice President of Innovation, IT Business for Schneider Electric) delivers keynote presentation that asserts that many traditional data centers are inefficient, costly and incompatible with high density IT deployments, and that pre-engineered and pre-manufactured power and cooling subsystems based on standardized modules can provide optimized and predictable performance faster and at lower cost.

Uptime Institute Symposium 2011 Video Clip (Click Here To View)

Six Reasons Why Modular Power and Cooling Plants Will Make Traditional Data Center Designs (Click Here To Download Green Data Center Conference Presentation)

Presentation Highlights Delivered by Neil Rasmussen:

Traditional data center defined:

  • Power and cooling devices from various manufacturers are integrated for a project
  • System performance is predicted by analysis
  • Controls are created for the project
  • Management software is customized for the project
  • Cooling is by CRAC/CRAH units located in the IT room
  • Air is distributed under floor via vented tiles
  • Outdoor heat rejection via dry cooler, condenser, or cooling tower

Modular power and cooling plants defined:

  • End-to-end power and cooling systems are pre-engineered and pre-manufactured
  • Standard building blocks are available
  • Modules may be internally fault tolerant, and can be combined to achieve redundancies
  • Equipment arrives in pre-packaged modules, such as skids, containers, or kits
  • System performance is guaranteed by spec
  • Controls are standard
  • Management software is standard for the modules

Reason #1: Economizer regulations

  • ASHRAE, LEED, and local regulations require economizer modes
    • Old model: Economizer assists mechanical plant, when possible
    • New model: Mechanical plant assists economizer, when needed
  • Maximization of free cooling is a complex design and control problem, difficult to achieve in a unique design
  • An integrated design, that considers IT supply temperatures and airflows, load factors, and ambient conditions is best achieved in a standardized, pre-engineered system

Reason #2: Dynamic power variation

  • Power management functions in IT equipment will play a major role in reducing overall energy use of data centers
    • Old model: Long term adaptations to slow changes in load
    • New model: Cooling plant optimizes for wide swings in IT load
  • Traditional plant design responds through interventions (moving tiles, turning equipment on and off, adding equipment)
  • An optimizing cooling plant adapts to changing load and airflow requirements by design

Reason #3: Speed of deployment

  • Modularity is not automatically faster
  • Modularity allows standardization. It is standardization that makes cycle time faster
    • Old model: 18 month design-build-commission cycle
    • New model: 4 month order-install cycle. Design is off-the-shelf. Systems mainly pre   commissioned.
  • Eventually standard power and cooling modules will be inventory items

Reason #4: Scaling requirements

  • Oversizing remains a major drag on the data center industry
  • It is the dominant contributor to energy inefficiency
  • It causes waste of CAPEX and OPEX
    • Old model: Build it now because it is too painful – slow – risky – burdensome – costly to adapt   later
    • New model: Modular design for scalability
  • Capacity can also be scaled to meet changes in power density and redundancy

Reason #5: Control & management system costs

  • Controls and management system cost in traditional data center is around $.50 -$1.50 per watt (under 10% of system cost)
  • To actually correctly engineer such unique systems for a traditional data center should cost about $5 per watt (nearly equal to expected system cost) if we
    • Optimized for energy savings
    • Did full testing under all operating and fault conditions
    • Documented the system fully
    • Embedded effective diagnostics
    • Did appropriate fault-tree and event-tree analysis
    • Designed to accommodate expected changes

Reason #6: Lower installed and operating costs
Extra installed costs

  • Container / skid / package costs
  • More smaller devices replace fewer larger devices

Installed cost savings

  • One-time engineering
  • Defer costs of capacity not yet required
  • Programming & configuration
  • Rigging
  • Documentation
  • Shipping / installing damage
  • Factory vs. field labor
  • Less field testing

Extra operating costs

  • None identified

Operating cost savings

  • Reduced expertise requirements
  • Energy costs
  • No maintenance costs on capacity not yet required

Conclusion

  • Standardized modular power and cooling plants have lower cost and better performance
  • Challenges such as dynamic power, economizer optimization, and high density will accelerate the end of traditional design approaches
  • Modular approaches to cooling plants just as effective for indoor IT rooms as they are for IT containers

Resources

“Economizer Modes of Data Center Cooling Systems” Full White Paper 132 (Download Here)

“Containerized Power and Cooling Modules for Data Centers” Full White Paper 163 (Download Here)

“Hot Aisle vs. Cold Aisle Containment” Full White Paper 135 (Download Here)

APC White Paper Library (Click Here)

APC TradeOff Tools™ Library (Click Here)

Universal Networking Services is proud to partner with Datapod to deliver an unique alternative to the traditional bricks and mortar data center installation. With Datapod we can provide the data center community an alternative solution that maximizes their investment and increases the reliability and availability of their mission-critical facility.  Datapod is an unique, modular data center system that incorporates innovative design and cutting edge mechanical and electrical engineering. It has extended the concept of containerized data centers to include critical site infrastructure such as modular generators, chillers, and deployment services thereby providing a complete infrastructure solution for data centers. By enabling data center users to deploy when they like, where they like and for how long they like, the Datapod system offers performance superior to that of  a “bricks and mortar” data center facility, deploys faster and at a more cost-effective price point.

Please feel free to contact us to learn more.

Comments are closed.