Definition:
Mean Detection Time (MDT) is a loose term in the the space of Functional Safety. Some call it latency, diagnostic latency, or mean time to detect (MTTD). It is not a formal term defined in IEC 61511-1. Some books use MDT in another way which is discussed below.
SIL Safe, in our literature and calculations, uses MDT as the average time required to detect a failure in a component with automatic diagnostics. It is a product of its diagnostics and is not something the facility can change. Newer components have near 0 sec MDT, but older ones can be 1 hour or more.
- a – time to detect the failure
- b – time spent before starting the repair
- c – effective time to repair
- d – time after repair to put back in service.
- MTTR = a + b + c + d –> from the failure itself to the restoration
- MRT = b + c + d –> meaning MRT does not include time to detect.
- MTTR = a + MRT
- MDT = a –> mean detection time
- DTI ≈ 2 * MDT
A related term is diagnostic test interval (DTI), which is the time between diagnostics being run. MDT is often 1/2 of the DTI.
MDT is not a term that shows up by itself in PFDavg and STR calculations. This is part of MTTR so it is included in that.
Alternative Definition:
The other way MDT is defined or used is the total detection latency for all SIFs, even those without diagnostics. Thus, it would include proof test interval TI. This is not wrong, but it does change how some of the equations are portrayed. But that is not how our literature and calculational tools are setup. Doing it this alternative way is acceptable as long as everything is aligned and the equations are updated.
Key Points:
- A portion of MTTR
- A loose word that is used differently in different spaces.
- Only applies to systems with diagnostics.
Example:
A plant has a SIF with 2oo3 instruments. The instruments are fifteen years old with diagnostics. The documentation states that the diagnostics run every 1 hour. That would be your diagnostic test interval. The MDT would be half that or 30 minutes.
Cited Source:
- Limble article – Guide to Understanding Failure Metrics