Losing power in a major data storage application can cause catastrophic damage. The results can range from the inability to process customer transactions or access data to delays in critical, time-sensitive operations, like stock trades. Additionally, power cuts and surges can damage or even destroy essential equipment. The risk of monetary loss and other liability is great — so maintaining access to a reliable source of power is of the highest consequence.
Industry standards recognize this, and they have actually begun to consider generator power as the primary power source for data centers. Electrical grid connections used for day-to-day power are considered alternative, which highlights the importance of on-site power generation.
The Uptime Institute published a white paper called “Tier Requirements for Power.” It states that “On-site power generation is the only source of reliable power. The high level of performance available from Tier-certified data centers stems from power distribution designs that rely on this reliable power.”
So, if municipal power grids are generally not considered reliable enough for critical applications, how can facilities make sure they are meeting the demand?
Keeping backup generators fueled
While the ultimate goal is to eliminate outages and achieve maximum uptime, it’s understood that 100% uptime is unrealistic. Many data centers provide “five nines” availability, which means that the service availability is 99.999% per year. This is considered to be the highest level of service availability, and it means the system will not be down more than five minutes and 15 seconds per year. “Four nines” would mean the system is unavailable 52 minutes and 36 seconds per year, and “three nines” equates to eight hours and 46 minutes per year.
One of the keys to maintaining the highest possible service availability is redundancy. Design engineers must work with service providers to design systems that meet whatever redundancy level is required. For example, a major data center in California utilizes an N+3 redundancy level, with “N” being the minimum number of generators required to serve the site’s power requirements. In this case, N = 5, so the facility has eight total generators available.
To fuel these backup generators, many data centers employ a fuel oil system that can be automated and monitored remotely to prevent shutdowns. These automated processes might include pump rotation, tank filling and draining, fuel filtration and water separation, suction prime testing, and line leak test cycles. Monitoring operations include tank fuel and water levels, temperatures, leaks, and valve positions.
A device called a fuel system controller (FSC) can perform these and other critical automation and monitoring processes. The FSC is a general-purpose, programmable logic controller with integral I/O. A single FSC can control and/or monitor a simple process, and multiple FSCs can be networked together for coordinated, distributed control of larger systems. Each FSC has its own logic processor for independent distributed control. FSC-to-FSC communication occurs over dual-redundant, masterless NodeNet networks.
Both networks communicate continuously. If one network goes down, the other network is already in operation with no loss of function. Additionally, if any FSC processor fails, all other nodes continue to function. Reliable communication between FSCs eliminates the need to run numerous line and low-voltage wires.
Each FSC node can be equipped with a 4-inch color touchscreen HMI for setup and operator interface. Each HMI can “see” all the FSC nodes. A single RS485 or Ethernet connection to any HMI allows the plant-wide monitoring system to monitor/control all FSC nodes. The most common application of the FSC is to monitor and control emergency generator fuel oil supply pumping and storage systems.
FSCs in Action
One case study involves a large, mission critical facility in Virginia that utilized two identical, redundant systems. Each system consisted of one filtration system with an integral main tank monitor for level and leak indication. The monitor also controlled filling operation. Additionally, each system had one duplex pump set and two 100-gallon day tanks with integral control cabinets. There were four fuel system controllers per system.
Projects, such as data centers or other mission critical facilities, often require unique designs and custom specifications. In this case, the site required all equipment to be inside an electromagnetic pulse (EMP) barrier to block electromagnetic fields. The problem presented by the EMP barrier was that the generators, day tanks, and pump set were contained inside it, but the main storage tank and filtration were outside the barrier. Control signals needed to be transferred between control panels, so fiber optic transceivers were added.
Another case involved a critical need for redundant generators. In this application, each generator has its own FSC day tank panel, but each controller is capable of controlling a second, paired generator in the event its “partner” fails. Every generator is paired to work like this, providing multiple levels of secure redundancy.
Having a reliable source of on-site power generation is not optional for critical data centers. Loss of power due to an outage can result in financial loss, equipment damage, and harm to reputation, so proper fueling and monitoring of backup generator systems is job No. 1. The fuel system controller is one of the simplest, most reliable tools available to accomplish this goal.