by David Matuseski, Cummins  -  

Finding the best combination of generator size and system architecture is a goal for all data centre designers. Often, a larger generator can provide unique advantages. When you combine large generators into a paralleling system, you can achieve a whole new level of advantages. The data centre designer should carefully analyse all options to optimise the generator system design.

There are many factors to consider when designing the generator system for a data centre. Two of the most important items to consider are the optimal size of the generator and the architecture of how the generators integrate into the overall power system. In some installations you have no choice but to go with a larger generator due to the size constraints of the site.

This can be especially true when installing additional power generation into an existing building. Using a greater number of smaller generators would not have been a good option in this example. The higher power density of large generators was the better choice. In addition to the unique challenges in metropolitan areas, sometimes the footprint advantages of using large generators are also desired in new construction.

Paralleling generators in a data centre design can have many advantages. It does not fit every situation, but paralleling does provide many benefits over a single generator design. The advantages have varying degrees of importance in a data centre application, but efficiency (reducing your stranded generator power capacity), can be the most attractive. The generator system is one of the more expensive systems to procure, install and maintain. Making this system as efficient as possible will have a significant effect on reducing data centre cost.

An important challenge in a paralleling system architecture is achieving concurrent maintainability. This is due to the common paralleling bus required in a generator paralleling system. Various system architectures have been developed to address this challenge. One of the more common architectures is adding tie breakers (or sometimes tie switches) to the paralleling bus. These tie breakers provide segments in the bus that allow you to partially shut down the switchgear for maintenance.

For example, if you have an N+2 design, you can completely isolate one segment of switchgear for maintenance or repair. This design is also used in an N+1 design with tie breakers in-between each generator on the paralleling bus. The protective relaying on a segmented bus can be designed to eliminate the paralleling bus as a single point of failure. This is most commonly achieved with a current differential scheme.

Another way to achieve concurrent maintainability is that each generator can be switched from one paralleling bus to an alternative paralleling bus. In an N+1 design, there would be one additional generator and corresponding transfer switch. Any one component, including the paralleling bus, can be shut down for maintenance and there will still be the full ‘N’ capacity available from the generator system. There is another architecture for paralleling generators being used in data centres. This is sometimes called ‘3 to make 2’ (or ‘4 to make 3’, etc.)

Contact Sbu Gule, Cummins Africa Middle East, Tel 011 451-3433, sbu.gule@cummins.com