FLOOD08791

• County: Statewide
• Basin: Statewide
• Title: ANCOLD Guidelines on Risk Assessment
• Date: 8/1/1998
• Prepared By: ANCOLD Working Group on Risk Assessment
• Floodplain - Doc Type: Educational/Technical/Reference Information

The 1994 Guidelines proposed that Individual Risk would be computed as the average Individual Risk
over the population at risk, or as the Individual Risk for the person most at risk.

The average conditional probabltity of deatb, given dam failure, is the expected loss of life (LOL) divided
by the population at risk (PAR). A difficulty with tbe average risk concept is that LOL/PAR declines as the
study reach is extended downstream, The fact that average Individual Risk is a function of the extent of the
study reach was recognised in the 1994 Guidelines. In an effort to ensure consistency of approach, tbe
"Dambreak Affected Zone" was defined in the 1994 document, in an attempt to have practitioners select the
study reach in a uniform way (at least for purposes of computing average Individual Risk).

X The Individ~aLRisk.fQrJh.eperson most at risk is the value that really matters. In 1994 no consistent way of
ide[;tifying that person, and then of estimating the conditional probability of death, was recognised and so no
real guidance was given. It would be fair to say tha~ the focus was on average Individual Risk in the 1994
Guidelines.

Against this background, the Individual Risk criteria of the 1994 Guidelines were effectively:

. Limit value of average Individual Risk: IE-05 per annum (2% of young person background risk)
. Limit value of Individual Risk for person most at risk: IE-04 (20% of young person background risk)
. Objective value of average Individual Risk: IE-06 per annum (0.2% of young person background risk)
. Objective value of Individual Risk for person most at risk: 1 E-05 (2% of young person background risk).

Societal Risk

Societal Risk criteria aim to take account of society's. aversion to disasters that involve multiple fatalities.
The general principle is that the greater the expected loss of life in the event of dam failure, the lower the
acceptable chance of dam failure should be. The present approach to Societal Risk criteria seems to have
originated with Farmer (1967) who proposed criteria for the siting of nuclear power plants in Britain. A
constant expected value of loss of life (that is the product of the probability of dam failure and the loss of
life given failure) as the number of lives lost (N) increases, would require that if the number of lives lost
doubled, the acceptable probability of failure would halve. Such an approach is termed "risk neutral" (Rowe,
1981). Those who would make choices that require a reduction in expected value of life loss as the number
of lives lost increases, are termed "risk averse" by Rowe. A majority of people seem to be risk averse, based.
on Rowe's research.

Two features of Societal Risk criteria are:

. they are concerned only with the number of lives lost, and not with the identities of the persons
involved. Thus itinerant campers count equally with permanent residents (contrary to the situation with
Individual Risk)
. they are event based. Thus each individual dam failure scenario (for example, piping due to earthquake,
storage 85% full, night time, summer holiday season) is considered separately in judging whether a dam
complies (unlike Individual Risk where the focus is on the total risk posed by the dam).

In the 1994 Guidelines, the Societal Risk criteria were expressed as FIN curves, where N is the number
of lives lost and F is the probability of failure that results in loss of N or more lives. Thus F is a cumulative
distribution function. In the text of the 1994 document, the symbol f was used in the text for the cumulative
function, but by convention F is more appropriate. The FIN curves were plotted with both axes to the same
log scale, so that a line with a slope of minus I represents a constant expected value of life loss. A decrease
in expected value as N increases can be achieved by steepening a straight line to a larger constant negative
slope or by having a curved line that shows a progressive increase in slope. Farmer (1967) argued for a slope
steeper than minus I, although he was plotting magnitude of radiation release rather than loss of life. In the
Netherlands, the approach of a steeper straight line has been adopted in two jurisdictions {see Figures 52 and