Guideline 2.5-- Treatment of Uncertainty.

DOE-DP-STD-3023-98

produce benefits that accrue over time and (2) decision options that must be

repeated or extended to produce lasting benefits.

9.2.5

Guideline 2.5-- Treatment of Uncertainty. The RBP system should include a means to

address uncertainties in the prioritization results. Documentation should include a

discussion of how uncertainties affect the prioritization results. The sophistication of

the method chosen to address uncertainties and the level of effort devoted to

assessing the impact of uncertainties should be commensurate with the value of the

information as well as the scale of the effort.

Discussion. When addressing uncertainty, it is important to ensure inclusion of a

range of technical interpretations and viewpoints, and avoid forcing technical

consensus where it does not exist. An example helps to illuminate many of the

issues involved. Defense-in-depth is a common practice in limiting the risks of

hazardous operations that could threaten public health and safety. A prioritization

model that captures only known accident risks would value the prevention or

mitigation of these accidents but would accord little or no value to defense-in-depth.

To capture fully the value of defense-in-depth, it is necessary to undertake the difficult

task of evaluating all the uncertainties: some kinds of accidents may have been

missed, evaluations of the reliability of safety provisions may have been unduly

optimistic, or some accidents might proceed down paths not anticipated. Defense-in-

depth is intended to protect the public from risks that remain unmitigated due to just

such regulatory limitations.

The treatment of uncertainty may-- at its simplest-- be entirely qualitative. In many

cases, a mix of quantitative and qualitative methods are employed, with the qualitative

treatment reserved for those kinds of uncertainty that are difficult to evaluate

quantitatively, such as those that give value to defense-in-depth. It is also possible to

attempt a fully quantitative treatment of all uncertainties, though doing so for our

ignorance of all risk contributors inevitably entails a highly subjective approach. It is

not uncommon to be able to develop bounding estimates of what might have been

missed or distorted in a risk assessment, and these bounds are often useful in

illuminating the potential range of values that defense-in-depth strategies might

warrant.

There are four key aspects of uncertainty in prioritization. A discussion of each is

provided below.

a. Variability and Uncertainty. Some inputs to the RBP system are predictable only in

a statistical fashion, i.e., there is a stochastic or random component. For example,

the RBP system may consider a risk due to accidents that affect public health and

safety. The number of individuals affected by such accidents is a variable quantity

(in the mathematical sense, a random variable that is described by a probability

distribution) that depends on the nature of the accidents considered, the weather

at the time of the accident, and individual response to such accidents. A common

way of expressing variabilities in risk assessments is through the use of

complementary cumulative distribution functions. In the above example, one could