More
- Determine the amount of redundancy you really need to reach your reliability/availability requirements. Often, 100 percent redundant A-B bus designs are proposed because it is an easy and obvious way to provide redundancy, but it is also the most costly form of redundancy by almost any measure you want to use including initial cost, space, efficiency and maintenance cost.
- Reliability and availability come from the system architecture, not the reliability and availability of individual pieces of equipment. If system reliability/availability depends strongly on the values of a single piece of equipment, the system cannot be highly reliable. That is why, for example dual corded loads are far preferable to single corded loads connected to dual power paths through a static transfer switch.
- Avoid paralleling UPS (uninterruptible power supply) whenever possible. Just because you need a lot of power doesn't mean you need to parallel UPS modules. Often, the most reliable and cost effective way is to use a distributed or iso-redundant configuration using multiple, independent power paths using large individual UPS building blocks. Individual UPS building blocks of 1000 kVA or more are now available. Whenever you parallel, single point failures are introduced and they extend far beyond the portions of the system that appear to be common on the one line diagram. Often, failures in a single UPS of a parallel system can cause failure of the entire system and any parallel system must contain some degree of common control elements that can cause complete failure of the parallel system. Further, common control elements may not be obvious to someone who didn't design the controls.
- Rather than paralleling, consider breaking the load into smaller pieces. With UPS modules in the megawatt range, the difficulty of load management is not greatly increased.
- Maintain a level of redundancy that provides consistent reliability/availability throughout the system. For example, it is probably a waste of money to have a highly redundant UPS system if everything upstream and downstream is in a single path. Unfortunately, you see this repeatedly in many system designs. Redundancy alone doesn't make a system concurrently maintainable. How the redundant system elements are arranged in the system does. Can one redundant element be removed and replaced while you operate on the other without hot work?
- If you are considering the use of sealed batteries because you need a lot of power in a small space, why not consider flywheel UPS? They consume even less space than conventional UPS and sealed batteries, and if you are not installing battery monitoring, flywheel UPS can be much more reliable.
Consider the following: Sealed batteries fail open circuit which means a single failure in string of 40 12V jars leave the UPS with no ride through. Then, let’s make an extremely optimistic assumption the mean time between failure (MTBF) of a single jar is 10 years (four years is a more realistic number). This equates to 87,600 hours MTBF for a single jar. However, you have 40 jars in series which gives a string MTBF of 2,190 hours (87,600/40), which is approximately three months. In addition, consider the mean time to repair (MTTR) of a failed battery string? It isn't the 10 to 20 minutes it takes to replace a battery jar. That is only a small portion of the time. The real MTTR is the time it takes to discover the failure has occurred plus the time replace the jar. If the UPS only checks the battery string once per month (which is common), the MTTR is two weeks (336 hours). These two numbers taken together mean there is a very significant probability a second battery string in the system will fail before the first is repaired. The probability worsens proportionally if multiple battery strings are required to support the load of a single UPS module and the factorial of the number of UPS modules rather than proportionally.
ED+C's Green Product Buzz Guides
