A realistic evaluation of service life for durable, electromechanical components like fans is an important deciding factor for users. Because manufacturers cannot test for decades before supplying customers, they rely on a mix of theoretical approaches and realworld values. The calculation and test procedures and the mix of both varies widely from manufacturer to manufacturer. This is why it is important to be able to evaluate the resulting specifications and values correctly.
Two commonly used values are service life and reliability. However, these cannot be converted back and forth, as they have different impact on the failure performance of components. Thus the service life specifies the time period in hours up to the point where ten percent of the devices have failed. In contrast, reliability indicates what is known as the mean time between failures (MTBF) value – the average time when one device fails out of a group started simultaneously. So-called classic failure performance says that a few components can fail at the beginning of operation due to faulty parts or installation errors. In the subsequent period, the devices endure long operating times with only a few, random failures. The MTBF value describes this range. Towards the end, wear then becomes noticeable and the failure rate increases again. The service life is delimited this way.
In order to reduce the test period, manufacturers often operate a large number of devices over a period of six to twelve months. Then the service lifeis extrapolated from the result using different methods. However, these methods provide incorrect results if the test does not include cases of wear. In that case the service life information turns out too optimistically. The test period is often shortened by achieving accelerated ageing using external influences such as increased temperatures, temperature changes or shocks. The often unrealistic, ascertainable effects of temperature influences and their retroactive projection to normal operation are a disadvantage compared to real, long-term tests.
For example, many computing models assume a doubling of service life at a temperature drop of 10 to 15 kelvin. If manufacturers use this extrapolation multiple times, absurdly high service life values quickly result. Here it is useful for the user to compare the service life information at high temperatures. If these are similar, but differ greatly at low temperatures, then the service life is not different, just the mathematical model that was used.
Despite similar results in an accelerated service life test, the specifications of various manufacturers can differentiate in multiple ways. Thus, a conservative estimate of all influencing factors is essential for realistic specifications. However, long-term experience and constantly optimised arithmetic operations are absolutely necessary for such practical evaluations.