The use of modular components is nothing new to the mission-critical industry. For years, we have been building systems out of individual identical components, such as paralleling generators and UPS modules to create system capacity or redundancy. For example, as systems have grown larger, we have been required to replicate them to achieve the total capacity or the redundancy to meet the project requirements. Many times an engineer takes on an unnecessary risk when working on a new project thinking he or she has an obligation to create an entirely new topology when, in fact, the engineer may have been hired because of the success of past designs.
I believe strongly in keeping things simple. Unless I’m directed otherwise, I have an obligation to my clients to give them the benefit of my experience to make their systems reliable, cost-effective, simple to operate, and easy to maintain. My clients should feel comfortable with their systems and be able to understand how and why things operate the way they do. I do not want them to be apprehensive because they don’t understand the system or they feel it is overly complex. I believe that the way to achieve these goals is to use combinations or arrangements of standard components that I know are simple, reliable, and understandable for clients, and that will do the job.
While it is fun and challenging to come up with new designs using the latest components and creations from any number of different manufacturers, the exposure we have as the creative engineering force behind the design is enormous. In these situations, we all stay up late at night (or even wake up in the middle of the night) obsessing over every possible “what if.” In the end, we usually find the new configuration failed because there were component interactions we had no way of predicting, or there were undiscovered fatal flaws in equipment.
Trailblazing is never easy. When we decide to make the recommendation to a client to use new components that have few, if any, operational years (or even hours) of reliable service history behind them, we are placing our reputations in the hands of faceless and nameless equipment designers who will be of little help should the system drop the load.
Now, I am not suggesting that all is lost or that we should all simply give up and stay home next week. What I am suggesting is that it is perfectly acceptable to take the “if it ain’t broke, don’t fix it” approach to our designs. If you have come across a configuration of components that is reliable, simple, and that can be adapted to alternate topologies then, by all means, stick with your proven concept.
Years ago, data centers were purpose-built facilities designed and constructed to meet the specific needs of a client. We did not think of the design as being modular, even though we would replicate individual systems topologies side by side to meet the capacity and redundancy requirements of the project. When raised-floor space became a saleable commodity, the concept of replicating facilities in city after city across America and around the world took off. Different sites had different requirements depending on the size of the facility or the concentration of Internet-savvy entrepreneurs in the area. We began referring to sites as “3+1” or “4+1” locations. We were, of course, referring to the number of primary and reserve or redundant systems for the sites. But more importantly, the topology had become a combination of standardized components arranged in a standardized fashion. Our designs were modular, and each site did not have a specific custom box of parts.
The standardization of components and topologies allowed us to run a set of calculations on the availability or reliability of our systems. We were able to create, use, and reuse the same system-commissioning documents. Our wiring diagrams and interconnection details became standard and, more importantly, our clients knew in advance what they were getting. They knew how the system would work and how to optimize system performance. In terms of optimizing the design, we were able to adjust equipment sizes and eliminate multiple contingency allowances. Optimizing performance meant optimizing the cost model. Our goal was to be efficient on every front—in the use of space, energy, construction time, and capital funds.
For us as designers, the goal moving forward is to come up with our own modular designs that can be adapted easily to achieve different topologies to meet different client needs. The three simple single lines shown as Figures 1, 2, and 3 are examples of this concept. The basic topology is a single utility source and generator supplying a main bus. Each main bus in turn supplies both UPS and essential support loads. This building block (or module) is the same in each figure.
Once we have established a modular design that can be adapted to different infrastructure designs, we can move toward both the standardization and optimization of components. The standardization of components would involve working the specifications with selected vendors to establish a foundation for consistent, reliable performance. The specification needs to establish minimum component quality and performance, yet remain open and flexible enough to allow multiple acceptable offerings. Each vendor should be able to provide meantime between failures (MTBF) and meantime to recovery (MTTR) numbers for its products, which will allow you to create a matrix of availability calculations based on the various vendor combinations. Cross-referencing availability with cost may allow you to determine the preferred combination of components that creates the best value proposition for your specific design.
Perhaps the biggest area for optimization that I see is in the mechanical cooling strategies currently being employed. Current configurations incorporate cooling with direct exchange (DX) units, with air-cooled chillers, with water-cooled chillers, with outside air and water spray, or perhaps water-cooled cabinets. Different distribution configurations may present different loss profiles, which, in turn, will influence the net output capacity of a given UPS selection. Each of these options presents a different demand for power that alters the critical capacity for a given utility and diesel combination.
With a little time, effort, creativity, and some computer skills you can create a matrix, or several versions of the same matrix, with different variables to potentially help you and your client select the most cost-effective configuration for a critical application. The key, however, before creating any type of matrix to determine the optimum solution for the problem, is to create a simple, robust, cost-effective modular design that can be applied to a variety of system topologies. Your design should use standard, reliable, readily available components. Once the design is in place, applying a consistent methodology of calculating system availability may help you to revise, modify, and improve the design of your modular infrastructure concept.
Bergthold is senior vice president and chief technology officer of Cupertino Electric Inc. He is a licensed engineer and is registered to practice electrical engineering in 13 states. Bergthold is a member of IEEE and represents Cupertino Electric in the International Assn. of Electrical Inspectors, the 7x24 Exchange, AFCOM, the Assn. of Facilities Engineers, and the Uptime Institute.