Data centers require large amounts of power to operate and are ideally suited to use microgrid technology to cut costs and improve efficiency.  One significant challenge in implementing a power strategy is to be able to perform model-validated real-time diagnostics of the entire electrical system – from the substation to the servers.  One of the great promises of the emerging microgrid opportunity is to use the cost of onsite power generation as an asset and market participant. 

There are two key aspects to merging high-availability data center requirements in the microgrid.  First, and most important, is that nothing can compromise availability.  Second is the infrastructure.  The microgrid must be transparent to the data center operations team.  In other words, microgrid participation should be a “check” that comes to you in the mail, not a new investment and expanded staffing or capability.

A microgrid can be defined as a grouping of electrical generation sources and loads that have various modes of operation, including the ability to disconnect from the macrogrid and function autonomously as physical or economic conditions dictate.  Data centers have specific key requirements that fit perfectly into the definition of microgrid.  They must have high reliability, provide redundant backup power systems and be capable of operating in an islanding mode of operation disconnected from the macrogrid, continuing to support mission-critical loads.

The microgrid is typically considered to be on the customer side of the meter. Elements important to consider on this side include generation options, such as diesel engines, batteries, solar cells, etc. It’s also important to coordinate with the utility, since disconnecting from the main grid can cause significant issuess for the utility macrogrid. Another key component is a macrogrid management system, the software responsible for interfacing with a variety of devices and systems, including a data-acquisition system and a building-management system. The software master controller must be able to communicate across multiple vendor hardware platforms and protocols.

In order to implement a successful microgrid as an electrical power network, three important steps must take place:

Create a model of the electrical network with all electrical characteristics of each component as shown in Figure 1

Collect the data in real-time for the entire system as shown in Figure 2

Perform analytics and diagnostics of the model compared to actual real-time data from the system as shown in Figure 3. This information can be used to optimize the power usage and loading to identify issues prior to problems occurring.

Data center operators can use this information for a single building, campus-wide network or for multiple locations across the country. Regardless of the power strategy, the economics is secondary to the loss of power. Ensuring that power is available is critical for data center operators. A microgrid places new technology capabilities on how electricity can be supplied and used. The fact that mission critical networks must maintain power shows that microgrids are a perfect technical solution for data centers.

Operating the microgrid while separated from the utility is typically known as “islanding.”  In this operating mode the primary concern is whether the local generation can support the local load in a stable fashion. Just meeting the load demands is only part of this solution. It is necessary to ensure that the starting of a large load does not cause dynamic instability in the microgrid.  

Figure 4 shows a typical decision tree for this type of operating mode.

Capabilities such as reduced cost energy management, implementation time, downtime, alarm management and advanced monitoring lead to reduced power infrastructure management direct and indirect costs through unsurpassed power availability and reliability.

Volumes can also be written about how microgrids meet energy efficiency and environmental incentives. Add these to the operational advantages we have described, and you have an ideal solution to many of the recurring power management challenges data center managers are dealing with today.

About the authors: Kevin Meagher is CTO at Power Analytics and John Jennings is the company’s director of product marketing

Disclaimer: Views expressed in this article are those of the author and do not necessarily reflect the views of DatacenterDynamics.