Forum Topic

24. abril 2008 10:17:45

The SIL assignment risk graph proposed in Annex A of EN 62061 does not make it clear whether multiple deaths are considered (& if so the max numbers of potential deaths) or not.

There is (in some quarters at least) a belief that machines can only generate a max of a single death. For large machines this is clearly not the case (Ramsgate disaster. 1994, 6 deaths, 7 serious injurys).

As a manufacture of bespoke machines, some very large with this potential, we can either;
1) Use the proposed risk assignement matrix as it stands.
2) Modify the proposed matrix to add one or two tiers higher in the consequenses column
3) Look for a different methology

EN 62061 is clearly the standard to generally apply, and most of our machines will fall into the matrix as proposed, so either (1) or (2) above appears to be best as the use of one methology is going to be more sensible. I have looked at (2) but am very concerned about the potential for "over" assignment as it is too easy (?) to move in to the requirement for SIL 4 (& EN 62061 can't then be used). EN 61508 suggested assignments don't really help.

What do the other daughter standards suggest for potential for multiple but localised deaths, and at what sort of number does it get "ramped up"? Can a few multiple deaths be considered as the same level of consequence as a single death/seriouis injury or is it always worse?

Any other comments?

regards

H Wheeler

26. abril 2008 10:23:31

In other industries than the machinery industry risk is considered a matter of the company itself. Every company needs to decide how much risk it can tolerate. In terms of money or production loss this is easy. If you are a rich company you can increase the limits if you are a poor (small) company you cannot handle large losses and thus you decrease the limits, ie, you add sooner more safety.

Now with people dying that is a different story and also in the machinery that is a different story. In the machinery business many companies take the standard risk method from the standards. In western EU any person dying is already one too many but still we accept that. So you see many times that companies have one level that says 1 person dying and then the next level says "multiple people dying". Of course you can easily make only one level "one or more people" dying it only means that your highest risk levels must take the best possible protection.

My opinion is that you need to forget about the risk criteria in the standard and judge it yourself with every machine required. If only 1 person is around only 1 person can die. If 10 people are around 10 people can die and thus you need to consider that risk. In other words you need to adjust the risk graph / matrix to the situation at hand.