Presentation on theme: "1 Setting Target Reliabilities by Marginal Safety Returns Rolf Skjong Strategic Research Det Norske Veritas JCSS Workshop on Code."— Presentation transcript:
1 Setting Target Reliabilities by Marginal Safety Returns Rolf Skjong Strategic Research Det Norske Veritas Rolf.Skjong @dnv.com JCSS Workshop on Code Calibration, March 21-22 2002
2 Traditional Approach - SRA Typical Example:DNV Classification Note 30.6 (1992) on Structural Reliability Analysis of Maritime Structures Explains why SRA does not produce Probabilities with a frequency interpretation –No gross error –Epistemic uncertainty & model uncertainties included –SRA talk of “notional” reliabilities
3 Traditional Approach - SRA Target should depend on consequence Calibration against known cases (that are acceptable good/best practices in the industry) Calibration against similar cases with similar consequences Based on accepted decision analysis techniques Based on tabular values (presented as a last resort)
4 Traditional Approach - SRA Target should depend on consequence Calibration against known cases (that are acceptable good/best practices in the industry) Calibration against similar cases with similar consequences Based on accepted decision analysis techniques Based on tabular values (presented as a last resort)
5 Traditional Approach - SRA Based on tabular values (presented as a last resort)
6 Traditional Approach - QRA Quantitative risk assessment is the basis for regulations in many industries –PSA/PRA - Nuclear –Hazardous Industries (Seveso I/II) –Offshore (Safety Case) –Shipping (FSA) –Etc.
7 Quantitative Risk Assessment Two uses of Risk assessment Use as a basis for receiving and maintaining a licence to operate (the plant, platform etc.) –Safety Case Use as a basis for implementing risk reducing measures for “populations (all cars, all ships, all planes etc.) –Formal Safety Assessment
8 Traditional Approach - QRA Present Risk Results in terms of –Individual risk (fatalities) –Individual risk (Health and Injuries) –Societal Risk (group Risk) –Environmental risk –Economic risk (not necessarily a regulatory issue)
9 Traditional Approach - QRA Example Individual risk
10 Traditional Approach - QRA Example Health Risk
11 Traditional Approach - QRA Example Societal Risk
12 Traditional Approach - QRA Low Risk High Risk Intolerable ALARP Negligible Not acceptable Acceptable Acceptable if made ALARP
13 Traditional Approach - QRA The As Low As Reasonably Practicable Area implies that cost effectiveness assessment may be used Risk is made As Low As Reasonably Practicable, when all cost effective safety measures have been implemented Implies that a decision criteria for cost effectiveness will be required
14 Methods for deriving criteria Human capital approach Willingness to pay Comparing to well informed (risk informed) decisions in democratic forum (a willingness to pay) Comparing to previous decision (a willingness to pay) Societal Indicators (a willingness to pay) Individual decisions
15 Human capital approach Value of man as a resource in economic production Has discredited cost effectiveness & cost benefit assessment Contradicts ethical principle (Protagoras: “Homo mensura” and later formulations, e.g. Kant) Same reason as many governments ban research on human stem-cells
16 Willingness to pay Many forms of willingness to pay studies –Questionnaires –Observed behaviour (e.g. insurance) –Implicit in previous decisions –Implicit in existing regulations –Etc.
17 Comparing to well informed decisions Example from IMO: UN Organisation for maritime safety and environmental protection regulations
18 Comparing to previous decision By reallocation 40.000 lives could be saved annually in the US $ 42.000 35 = $ 1.5 million
19 Societal Indicators Societal Indicators used to rate “quality of life” in countries Published by UN (UNDP) Many different indictors exist Include such parameters as: GDP/Capita, Life Expectancy at Birth, literacy etc.
22 Individual Decisions Also individuals take decision that increase life expectancy and reduces accident frequencies For example: –Buy safer cars –Buy more healthy food –Go to the doctor more frequently –Etc. How much increase in purchasing power is necessary to increase the life expectancy in a population by “e”
26 Criterion? For an OECD member country (excluding the newest members) the criteria is somewhere in the range $ 1.5 -3.0 million Some uncertainties relates to: Fatalities as indicator or actual fatalities NCAF or GCAF Despite uncertainty different methods give surprisingly consistent results In a QRA a factor of 2 is not much compared to the uncertainty in the analysis In SRA the design variables are continuos, and we do not know how sensitive the resulting dimensions are to the criterion
27 Advantages in SRA Consistency with QRA The decision is based on the derivative of P F with respect to design variables only Not reliant on probability (the absolute number) Example: Mid ship bending moment