1. Principles for risk and uncertainty analysis and management, in a production assurance setting Roger Flage PhD student on the RAMONA project (research tasks 1 and 4)

2. 2 Basis for this presentation Aven, T. & Flage, R. Use of decision criteria based on expected values to support decision-making in a production assurance and safety setting. Reliability Engineering and System Safety, to appear. Flage, R. & Aven, T. On treatment of uncertainty in system planning. Reliability Engieering and System Safety, to appear.

3. 3 Expected value decision criteria How should we use the E[NPV] approach, with adjustments, in the decision-making process? NPVNet Present Value

4. 4 The E[NPV] Expected cash flow at year t = E[B t ] – E[C t ] Analysis period Discount rate at year t B t benefits (revenues) at year t C t costs at year t

5. 5 Uncertainty reduction in system planning Should we use predefined uncertainty interval categories to direct uncertainty reduction processes?

6. 6 Uncertainty reduction structure Prediction interval categories Feasibility phaseConcept development phase Engineering phase E 1 [Y]± 40% Y ± 30% ± 20% E 2 [Y] E 3 [Y] Y Performance measure (e.g production downtime)

7. 7 Uncertainty reduction structure Prediction interval categories Feasibility phaseConcept development phase Engineering phase E 1 [Y]± 40% Y ± 30% ± 20% E 2 [Y] E 3 [Y] Y Performance measure (e.g production downtime) Use with care

8. 8 Some E[NPV] adjustments Risk-adjusted discount rate Downward revision of benefits, upward revision of costs Safety margin Negative safety margin Cut-off periods

9. 9 Some E[NPV] adjustments Risk-adjusted discount rate Downward revision of benefits, upward revision of costs Safety margin Negative safety margin Cut-off periods Considerable arbitrariness

10. 10 Wrong assumptions All aspects of risk and uncertainty have been taken into account in the formulae There is no need for seeing beyond the formulae

11. 11 The main problem Risk and uncertainties are represented by probabilities, but probabilities are not perfect tools for expressing risks and uncertainties.

12. 12 Decision situations 1.Known, “objective” probability distributions can be established 2.More or less complete ignorance 3.A situation between the two extremes 1) and 2)

13. 13 Approaches to arriving at a good decision Decision criterionDecision Analyses Managerial review and judgement Decision Basis for the analyses Other concerns Analyses

14. 14 Uncertainty reduction in system planning Should we use predefined uncertainty interval categories to direct uncertainty reduction processes?

15. 15 Uncertainty reduction structure Prediction interval categories Feasibility phaseConcept development phase Engineering phase E 1 [Y]± 40% Y ± 30% ± 20% E 2 [Y] E 3 [Y] Y Performance measure (e.g production downtime)

16. 16 Uncertainty reduction structure Prediction interval categories Feasibility phaseConcept development phase Engineering phase E 1 [Y]± 40% Y ± 30% ± 20% E 2 [Y] E 3 [Y] Y Performance measure (e.g production downtime) Use with care

17. 17 Structural levels and planning guidance Structural level Plant or system System Subsystem, equipment and component

18. 18 Structural levels and planning guidance Structural levelForm of planning guidance Plant or systemOverall ideal goals SystemRequirements related to expected performance Requirements related to uncertainty about performance Specifications related to design or operation Subsystem, equipment and component Requirements related to expected performance Requirements related to uncertainty about performance Specifications related to design or operation

19. 19 Structural levels and planning guidance Structural levelForm of planning guidanceSpecific examples and examples of attributes Plant or systemOverall ideal goals“No production losses” SystemRequirements related to expected performance Throughput availability Demand availability Requirements related to uncertainty about performance Prediction interval limits Limit for variance of lost throughput Specifications related to design or operation Capacity Size and weight Operating temperature range Maintenance - Spare part needs - Manpower needs Subsystem, equipment and component Requirements related to expected performance Mean time to failure (MTTF) Mean time to repair (MTTR) Requirements related to uncertainty about performance Reliability/availability at a specified time Specifications related to design or operation Capacity Size and weight …

20. 20 Structural levels and planning guidance Structural levelForm of planning guidanceSpecific examples and examples of attributes Plant or systemOverall ideal goals“No production losses” SystemRequirements related to expected performance Throughput availability Demand availability Requirements related to uncertainty about performance Prediction interval limits Limit for variance of lost throughput Specifications related to design or operation Capacity Size and weight Operating temperature range Maintenance - Spare part needs - Manpower needs Subsystem, equipment and component Requirements related to expected performance Mean time to failure (MTTF) Mean time to repair (MTTR) Requirements related to uncertainty about performance Reliability/availability at a specified time Specifications related to design or operation Capacity Size and weight …

21. 21 Structural levels and planning guidance Structural levelForm of planning guidance Specific examples and examples of attributes “No production losses” Throughput availability Demand availability Prediction interval limits Limit for variance of lost throughput Capacity Size and weight Operating temperature range Maintenance - Spare part needs - Manpower needs Mean time to failure (MTTF) Mean time to repair (MTTR) Reliability/availability at a specified time Capacity Size and weight … Plant or systemOverall ideal goals SystemRequirements related to expected performance Requirements related to uncertainty about performance Specifications related to design or operation Subsystem, equipment and component Requirements related to expected performance Requirements related to uncertainty about performance Specifications related to design or operation Avoid unless a rationale can be given May be stated when level of detail in planning becomes high Do not treat as absolute limits “Optimisation” (in a broad sense) Feasibility and concept development phases Engineering phase

22. 22 Conclusions Be careful in using adjustments of expected values to reflect risk and uncertainties Use predefined uncertainty interval categories with care

23. 23 Situation 1 Known, “objective” probability distributions can be established Net benefit0110 Probability0.900.090.1 E[NPV] = 0.09 Net benefit-10000110 Probability0.000010.900.090.1 E[NPV] = 0.09 + safety concerns + environmental issues

24. 24 Situation 2 More or less complete ignorance Net benefit-10 8 010010 8 Probability0.25 E[NPV] = 25 1.E[NPV] poor prediction  consider distribution 2.Poor basis for probabilities (knowledge based) 3.Even more extreme outcomes could occur 4.Uncertainties related to non-economic outcome dimensions e.g. health problems for workers in 20-30 years Little is gained by introducing a specific formula reflecting uncertainty and risk aversion

25. 25 Situation 3 A situation between the two extremes (1) and (2) How much weight to give to risk and uncertainties? The company’s attitude towards risk and uncertainty The frame conditions, for example defined through taxes, regulations etc. Compensation schemes (monetary or in kind) Insurance and liability Sustainability. Does the project assist in sustaining vital ecological functions, economic prosperity and social cohesion? Ethical concerns, for example related to equity and fairness Political concerns … A balanced perspective is required