Posted on August 30, 2010 by Greg More
Today’s high power demands, increasing power costs and constantly changing equipment require data center and facilities managers to look for power solutions that can deliver a lot of power but are flexible. A typical data center has both medium density (5-10 kW/rack) and high density (15-30 kW/rack) equipment racks. For example, some racks may have 2U/4U application and database servers, others may have 1U Web servers, still others may support blade servers, storage networks, or high density networking fabric.
The power requirements for this variety of equipment can range from single phase to three phase, rack power (kW) may be anywhere from 5 kW to 30 kW per rack, and there could be dozens of different physical receptacles. The key is to efficiently utilize the available power capacity and to maintain uptime to the greatest extent possible while changes are made.The traditional approach uses pedestals at several locations in the data center with under-floor conduit or whips from the pedestal to racks. All the cable under that raised floor can impede air flow and cooling. The fixed feeds lead to decreased flexibility and challenging serviceability. In fact, it is not uncommon for some cables to be abandoned in place.
Another approach worth considering is 400V three-phase power provided by overhead track busways with plug in modules which in turn feed intelligent rack PDUs. This deployment frees up more useable floor space than the pedestal approach. It also reduces the cable clutter under the raised floor for better air flow and cooling.
Both single and three-phase loads can be supplied from one busway so varying power densities can be fed with the same infrastructure. Some systems allow data center staff to add and remove circuits without de-energizing loads. This makes it easy to change voltage, current, and receptacle configurations as needs change. Power connections can be added and changed when needed which defers costs.
Running 400V from the busway directly to 400V rack PDUs eliminates transmission losses (480V–208V / 120V (4% loss) vs. 400V (0%)). Processors at 120V have 18% loss vs. processors at 230V (Line - Neutral at the PDU) which have a 15% loss. This deployment provides a net improvement of 7% (4% + (18%-15%)).
The rack PDU provides branch circuit overload protection at the rack and addresses the practical physical limitations to delivering multiple circuits such as for blade servers with multiple power supplies.
If an intelligent rack PDU (iPDU) is used, sophisticated energy metering, monitoring and management as well as environmental monitoring can be easily installed. This allows data center managers to monitor temperature and follow the latest ASHRAE guidelines to lower cooling costs. The power metering allows managers to determine actual power requirements so they can do informed capacity planning. Armed with information from iPDUs and the flexibility of busway power, changes can be made as needs and layouts change. This means efficient use of available power including freeing up “stranded” power. The environmental monitoring ensures changes don’t lead to undetected hot spots.
For more detailed information about flexible 400V power distribution join Raritan September 30, 2010 at the Datacentre Dynamics show in Chicago.