“RELIABILITY, SAFETY AND RISK ANALYSIS”

“RELIABILITY, SAFETY AND RISK ANALYSIS”

Lecture 1: Introduction to the Course

22/02/2021Piero Baraldi ([email protected])

Piero Baraldi

2

Introduction to the course:• Reliability• Safety• Risk Analysis

演示者

演示文稿备注

The first half of the course will be about the concept of reliability You graduate at polimi with a very high mark. You work for a very famous engineering company… you design a turbine. You are one of the best turbine designer. Then the turbine is made with very good materials…. And start working in a energy production plant…

Piero Baraldi

Degradation & Failure3

Healthy Degradation initiation

Evolution to… failure

Failure

演示者

演示文稿备注

What is our problem? We consider industrial component. In the example a glass of red wine . Even if you are very good at designing the component and you use the best material and technologies available, the industrial component, sooner or later will start to degrade. Here the component is represented by a glass of wine. You see at this time you have the onset of a degradation process (a crack). The crack becomes bigger and bigger and at a given time you will have the collapse of the component.

44Piero Baraldi

Degradation (some examples)

Creeping of turbine blades

Erosion of chokevalves

Crack propagation

演示者

演示文稿备注

OK, now let’s change the glass of wine with three components from three very different industrial systems These are the blades of a gas turbine. They can undergo a process of creeping. Creeping is a plastic deformation, the blade becomes thinner and at the end can break. The loss of a turbine blade is one of the most feared failure modes of turbomachinery. The turbine becomes unbalanced, this causes the failure of the turbine bearing which will stop the turbine in a very short time, the consequence is the failure of the rotor with parts of the system (missiles) that can reach hundreds of meters from the turbine Oil industry. A valve in an offshore oil plant.due to the sand we have a process of erosion of this choke valve. Which we have to periodically replace. Nei pozzi ad alta pressione il flusso può raggiungere 400/500 m/s. Tempi di vita intorno anche all’anno. Difficoltà nel rimpiazzo fermo attività MOTORS: crack propagation in bearings of rotating machinery, which often results in damage to the bearings and consequent reduced efficiency, or even severe damage, of the entire motor system.

Consequences

Failure: an Example from the Nuclear Industry

• Davis-Besse accident (February 2002): Refueling outage a cavity of the size of an American football in the reactor

pressure vessel head. Only a layer of cladding of 7.6 mm thick was left by the corrosion.

5

Repair Action: new lid (600M $) Stop operation until March 2004 (2 years)

Possible consequences in case of rupture Loss of coolant accident (LOCA)

Emergency safety procedures to protect from core damage or meltdown

The jet of reactor coolant might have damaged adjacent control rod drive mechanisms,

hampering or preventing reactor shut-down.

Core melt down

演示者

演示文稿备注

This is the top of the pressure vessel of NPP, these are the control bars. Here we should have 15 cm of steel. During a routine inspection, workers felt a wiggle of the control rode drive mechanisms. It should be impossible since it should be sorrounded by the 15 cm thick steel. I this case , they were lucky: they detected the cavity just in time, but what about if we had the ropture? … it was a problem of Boric Acid Corrosion. As with all environmentally-assisted damage modes there is an expected distribution in the extent of damage reflecting the stochastic processes occurring early in the damage development, and the uncertainties associated with modeling and system definition. The practical question here (quite apart from addressing the failure to address the problem in a timely manner) is: "Can we expect other similar incidents, for which the Davis Besse case is part of a wider distribution?" In fact, there have been many similar though less serious incidents including at least one case where almost the entire thickness of a low alloy steel pipe was corroded away to the internal stainless steel cladding due to a boric acid leak from a valve above). Other incidents have also occurred including failures of low alloy steel bolting of flanged joints exposed to high velocity steam jets (steam cutting) from primary side leaks where the degradation was probably mainly caused by erosion with a contribution from BAC where condensation occurred. Unfortunately, we do not know with certitude the specific system conditions that can lead to extensive BAC under boric acid deposits; in many cases relatively little corrosion of carbon or low alloy steel occurs under such deposits. We cannot readily explain, for instance, the fact at Davis Besse that there was extensive BAC at one penetration (#3), but no BAC damage at an adjacent and nominally identical penetration other than as a possible consequence of different elapsed times from the first occurrence of a leak in the two penetrations. Consequently a defensible proactive mitigation action is difficult to define at this time, and degradation management can be accomplished by reliable inspection and timely repair or replacement

66Piero Baraldi6

Explosion of the platform Deepwater Horizon (Gulf of Mexico), April 2010 (source: sciencesetavenir.fr)

Probable cause: leakage in the oil pumping system

Kamal MEDJAHER and Noureddine ZERHOUNI, workshop (Politecnico di Milano, 2015)

11 fatalities 4.9 billion barrels of oil spilled into Gulf of

Mexico Reputation Reputation Repair efforts:

1 billion dollars spent in Google, Adwords, and Youtube advertising

New CEO $20 billion claims fund $52 million to federal and state health

organizations to fund behavioural healthsupport and outreach programs across theUS Gulf Coast region

Consequences

Failure: an Example from the Oil and Gas Industry

Piero Baraldi

7

Cause: failure of a fishplate (source: Bureau d’enquêtes sur les accidents de transport terrestre)

Train derailment in Brétigny-sur-Orge in July 2013

7 fatalities 21 people were seriously injured 180 people were injured

Failure: an Example from Railway

Consequences

Sources: Keynote of Pierre Dersin at PHM Europe 2014, Muller et al. 1996, Sovacool 2008, Tzanakakis 2013, Kamal MEDJAHER and Noureddine ZERHOUNI workshop (Politecnico di Milano, 2015)

演示者

演示文稿备注

Metal bar that is bolted at the end of 2 rails to join them

Piero Baraldi

Failure: an Example from Civil Infrastructures 8

演示者

演示文稿备注

Morandi Bridge collapse tower

Piero Baraldi

Failure 9

演示者

演示文稿备注

Our life is based on critical infrastructures, such as an electric grid or a transportation systems. They can fail due to external events, cyclone, avalanche,…

Piero Baraldi10

Failure costs (Some Numbers)

According to Network Rail (UK), rail infrastructure failures and defects are responsible for 14 million minutes of delay per year

In automobile domain, failures cost around 288 millions US $ per day

Sources: Keynote of Pierre Dersin at PHM Europe 2014,

Piero Baraldi

Failure Definition 11

Failure definition: the termination/loss of ability of an item to perform its required function

Failure examples: Total cessation of function

• An engine stops running• A structure collapses

Deterioration/instability of function• a motor that is no longer capable of delivering a specified torque• a structure that exceeds a specified deflection

演示者

演示文稿备注

For example, a pump to pump a flow between minimum and maximum, a valve to stop a flow in a given amount of time…

Piero Baraldi12

Failure prevention

FailuresPrevented by

Design for Reliability Maintenance

Time

NormalDegradation

onsetRepair

演示者

演示文稿备注

Add a safety system to your industrial plant in order to improve its reliability.

Piero Baraldi

Reliability

Reliability (ISO8402): ability to perform an assigned task for a given time under given environmental and operational conditions • Always present in human activities

• From reasonable to rational solutions

13

演示者

演示文稿备注

ISO=International Standard Organization (continuously from the start of the operation until the system mission time) Environmental and Operationl Conditions … Vessel of a NPP = 40 years F1 engine = 1 season (21 races) carriege Concept of reliability always present in the human history. However, until the 2nd war war, people were not investing the concept of reliability. After a failure they were just replacing the failed component, maybe with one of better quality, design,… With the 2nd war war and the complexity of the industry, this was not possible (airplane).

Piero Baraldi

Reliability engineering: When?

II World War: USA

Radar Vacuum tubes

GERMAN

V1 Missile

14

• lot of failures• poor system performance• high maintenance costs

first reliabilitystudies

first 10 launcheswere allfiascos

first reliabilitystudies

Lusser (German Mathematician):“the reliability of a chain of components is determined by the reliability of the weak link”

演示者

演示文稿备注

5 times as the other electronic components V1; after the launches they were trying to improve all the components, but they were not able to improve the system…

Piero Baraldi

Reliability Engineering

• Why do systems fail? (reliability physics to discover causes and mechanisms of failure and to identify consequences)

• How to develop reliable systems?• How to measure/test reliability (in design and operation)? • How to maintain systems reliable (maintenance)?

15

演示者

演示文稿备注

These are the questions to which reliability engineering is trying to answer

Piero Baraldi

Our Big Problem…

lamp of my wife’s bedside table

lamp of mybedside table

演示者

演示文稿备注

This is making this course a little different from most of the other courses you have seen until now. We have to deal with random event Heat transfer, there is an equation, maybe is differential, with partial derivatives,… difficult to solve…., but there is an exact solution…Here is different…

Piero Baraldi

Our Big Problem…

t

t

X

X



http://www.centennialbulb.org/photos.htm

演示者

演示文稿备注

This is making this course a little different from most of the other courses you have seen until now. We have to deal with random event Heat transfer, there is an equation, maybe is differential, with partial derivatives,… difficult to solve…., but there is an exact solution…Here is different…

Piero Baraldi

Our Big Problem…

t

t

X

X

The failure time is a random variable!

How to represent the failure time?

Probability distribution: fT(t)



演示者

演示文稿备注

Random - uncertain variable – a variable for which we do not know the true value in advance. What is the consequence in practice of this? We must have few lectures to recall the basic concepts of probability and statistics (I promise , just few lectures at the beginning of the course).

Piero Baraldi

Reliability and Probability

Definition of reliability (ISO8402): the ability to perform an assigned task for a given time under given environmental and operational conditions

Operative definition of reliability: Probability that an item performs its assigned task for a given time under given environmental and operational conditions

Piero Baraldi

Reliability, Performance and Cost: the Trade-Off

• Objective: design and build product for improved performance Faster aircraft Thermodynamically more efficient energy conversion

devices

• Increase ‘load’ Aircraft decrease weight Energy conversion devices work at larger temperature

• Approach the physical limit of the system Aircraft Increase stress level in its components Energy conversion devices heat-induced losses of

strengths and more rapid corrosion

• Number of failures increases (reliability decreases)

• Countermeasures should be taken (cost increases) Purer material Tighter dimensional tolerance Monitoring & improved maintenance

20

演示者

演示文稿备注

When we consider the reliability of a component, we should take into account that the concept of reliability is deeply connected with the performance and the cost of the system. At an industrial level, the objective is always that of improving the performance of the system. To obtain better performance, what we typically do is to increase the load on the system. What is the problem of this? We will approach the … limits of the system, … s at the nd the results will be to increase … or, it is the same, to decrease the reliability of the system. But this is typically not acceptable, so we should take countermeasures. … The results of this countermeasures is an increase of the costs.

Piero Baraldi

Reliability, performance and cost: the trade-off

• Objective: design and build product for improved performance Faster aircraft Thermodynamically more efficient energy conversion

devices

• Increase ‘load’ Aircraft decrease weight Energy conversion devices work at higher temperature

• Approach the physical limit of the system Aircraft Increase stress level in its components Energy conversion devices heat-induced losses of

strengths and more rapid corrosion

• Number of failures increases (reliability decreases)

• Countermeasures should be taken (cost increases) Purer material Tighter dimensional tolerance Monitoring & improved maintenance

21

Use new-design components, thanks to new technologies(e.g. iron instead of wood in Structures)

Potentially, in the long term:• Better performance• Lower costs• Larger reliability

But in the early stage of introduction of the new technology:• Lower reliability

e.g. iron instead of wood in structures:• Problem of brittle fractures

演示者

演示文稿备注

The concept of reliability should not be considered alone, but intrinsically connected with those of performance and cost. The objective that we have as engineer is always that of improve the performance of our system.

Piero Baraldi

• Maintenance (IEC60300): set of actions that ensure the ability of an item to be retained in (preventive maintenance) or restored to (corrective maintenance) the functional state required by the purpose for which it was conceived.

Maintenance 22

Failure Maintenance

preventivemaintenance

演示者

演示文稿备注

Set of actions that we do to: 1) maintain the equipment working (able to perform…) 2) In case of failure: to repair/replace

Piero Baraldi

Maintenance Costs

Derived from M. Garetti

G$/year

演示者

演示文稿备注

More work done by machine, less by personnel. Machines may have failure.

Piero Baraldi

24


Piero Baraldi

Safety

• SAFETY ≡ freedom from unaffordable harm

演示者

演示文稿备注

State of being "safe", the condition of being protected from harm or other non-desirable outcomes

Piero Baraldi

26Hazard = Source of Potential Damage

演示者

演示文稿备注

Hazards can have very different sources…

Piero Baraldi

Hazard

barrier

No Hazard

Safety

• SAFETY ≡ freedom from unaffordable harm

• The ‘parmesan cheese’ model

演示者

演示文稿备注

Hazard = Source of potential damage 1) Radioactive material in a NPP 2) Oil and gas in the well reservoir We try to put a barrier (represented by a slice of cheese) to avoid that the potential damage (the hazard) becomes a real damage For example in a Nuclear Power Plant we have the fuel rod, the pressure vessel and the containment

Piero Baraldi

Not all barriers work...

Safety

演示者

演示文稿备注

Poor mouse…, this barrier the helmet is not going to work

Piero Baraldi

Not all barriers work...

Safety

演示者

演示文稿备注

Protective mask

Piero Baraldi

No Hazard

Hazard

Safety: Multiple Barriers

Piero Baraldi

Geological Barrier

Technical BarriersEmbeddingStorage caskOver packsBackfill

Waste

Safety: Multiple Barriers - Example

演示者

演示文稿备注

Hazard = Spent nuclear fuel Radioactive waste repository

Piero Baraldi

Hazard

HumanErrors

ProceduralErrorsFaults in

Redundancies

Safety: the Swiss Cheese Model

演示者

演示文稿备注

The problem is that the safey barriers that we design can have weaknesses, the holes in the figure. So in some particular conditions they can be ineffective. Depending on the hazards, the environmental conditions and so on, these weaknesses are continuosly changing position and size. According to this model, we have a system failure when all the holes align, permitting a trajectory, an accident opportunity, the hazard passes through all

Piero Baraldi

33


Piero Baraldi

The Concept of Risk

Hazard

SafeguardsEnvironment

People

UNCERTAINTY

演示者

演示文稿备注

A risk requires the presence of an hazard = potential source of damage. A Tank filled with a potentially exposive solution. Damage can be to the worker of the plant, to the population or to the environment. If there is an hazard, safeguards are typically used to prevent the possible undesired consequence of the hazard. Typically we can try to prevent the occurrence of the hazard (good control of the solution conditions) or mitigate its consequence (build a containment around the tank). Notice that not all the hazards becomes damage, there is an uncertainty behind the fact that the hazard translates from potential to actual damage. So, the idea of risk is connected with uncertainty and damage.

Piero Baraldi

The Concept of Risk 35

RISK = POTENTIAL DAMAGE + UNCERTAINTY

Dictionary: RISK = possibility of damage or injuryto people or things

Piero Baraldi

Probabilistic Risk Assessment 36

Piero Baraldi

Probabilistic Risk Assessment 37

1. What undesired conditions may occur? Accident Scenario, S

2. With what probability do they occur? Probability, p

3. What damage do they cause? Consequence, x

RISK = { Si, pi, xi }

Piero Baraldi

S p x

S1 p1 x1

… … …

SN pN xN

{Si, pi, xi}

Probabilistic Risk Assessment (PRA): Results

Piero Baraldi

RISK = p∙xRisk Measures

演示者

演示文稿备注

Let us start for simplicity from a system with a single possible accidental scenario. A measure of risk should combine the probability of occurence with a measure of the consequences. It can be natural to consider the product. So RISK is equal to probability * consequence. Tipically our perception is that the consequences are more relevant than probability power of k… Prevention add redundancies, improve reliability of the components Mitigation emergency system that reduces the consequences Protection

Piero Baraldi

RISK = p∙x k(>1)

Risk Measures

演示者

演示文稿备注

Let us start for simplicity from a system with a single possible accidental scenario. A measure of risk should combine the probability of occurence with a measure of the consequences. It can be natural to consider the product. So RISK is equal to probability * consequence. Thipically our perception is that the consequences are more relevant than consequences power of k… Prevention add redundancies, improve reliability of the components Mitigation emergency system that reduces the consequences Protection emergency system that reduces the consequences outside

Piero Baraldi

RISK = Σipixik(>1)

WARNING:

RISK (A)= RISK (B) A=(P, x); B=(p, X)

RISK REDUCTION:

A: Prevention B: Mitigation, Protection

Risk Measures

演示者

演示文稿备注

Prevention add redundancies, improve reliability of the components Mitigation emergency system that reduces the consequences (mitigate the accident) e.g. a fire protection system Protection containment system that limits the consequence outside

Piero Baraldi

The level of risk is not acceptableand risk control measures arerequired to move the risk figure tothe previous regions

The level of risk is broadlyacceptable and generic controlmeasures are required aimed atavoiding deterioration

The level of risk can be tolerableonly once a structured review ofrisk-reduction measures hasbeen carried out

Risk Evaluation: Risk Matrix

Piero Baraldi

Risk Assessment & Management Procedure 43

Piero Baraldi

Global Risk 44

Piero Baraldi

45

Course Syllabus

Piero Baraldi

Topics

• Basics of probability• Reliability of simple systems• Markov processes for reliability and availability

analysis of more complex systems• Monte Carlo simulation method for reliability and

availability analysis• Estimation of reliability parameters from

experimental data• Maintenance in the energy industry• Probabilistic Risk Assessment• Bayesian networks, fault and event tree analysis

for identification and quantification of accidentalsequences

• Dependent Failures• Importance Measures

Part 1: Reliability

Part 2: Risk

Assessment

Piero Baraldi

Topics (Part 1)

• Basics of probability

t

t

X

X

The failure time is a random variable!

How to represent the failure time?

Probability distributions: fT(t|λ)

Piero Baraldi

Topics (Part 1)

• Basics of probability• Reliability of simple systems

pumping system

Piero Baraldi

Topics (Part 1)

• Basics of probability• Reliability of simple systems• Markov processes for reliability and availability analysis of more

complex systems

Piero Baraldi

Topics (Part 1)


complex systems• Monte Carlo simulation method for reliability and availability analysis

Monte Carlo Simulationfor reliability and availability analysis

Piero Baraldi

Topics (Part 1)


complex systems• Monte Carlo simulation method for reliability and availability analysis• Estimation of reliability parameters from experimental data

(Accelerated) degradationtests

Failure times15.8 h15.9 h15.1 h17.2 h…14.5 h

fT(t|𝜃𝜃)

Piero Baraldi

Topics (Part 1)


complex systems• Monte Carlo simulation method for reliability and availability analysis• Estimation of reliability parameters from experimental data• Maintenance

?

Piero Baraldi

Topics (Part 2)

• (Probabilistic) Risk Assessment• Bayesian networks, fault and event tree analysis for identification and

quantification of accidental sequences• Importance Measures• Dependent Failures

53

Piero Baraldi

Topics

• (Probabilistic) Risk Assessment

54

S p x

S1 p1 x1

… … …

SN pN xN

Piero Baraldi

Topics

• Probabilistic Risk Assessment• Bayesian networks, fault and event tree analysis for identification and

quantification of accidental sequences

55

Blowout Accident in Oil and Gas Wells during drilling

Blowout

Kick

Piero Baraldi

Topics (Part 2)

• Probabilistic Risk Assessment• Fault and event tree analysis for identification and quantification of

accidental sequences• Importance Measures

56

Which is the most«critical» component?

Piero Baraldi

Topics (Part 2)


accidental sequences• Importance Measures• Dependent Failures

57

electric grid

Piero Baraldi

Topics (Part 2)


accidental sequences• Importance Measures• Dependent Failures

58

Piero Baraldi

Teaching Organization

• Lectures (traditional + flipped classes)• Exercise Sessions• Homework

development of a matlab/python code for the estimation of reliability and availability of an energy system

multidisiplinary teams (energy, manamegement, nuclear students)• Prepare a presentation and deliver a speech on a specific topic

related to the latest advancement in the field of reliability, safety and risk analysis.

• Seminars

Piero Baraldi

Teaching Organization (energy and nuclear engineering students)

RELIABILITY, SAFETY AND RISK ANALYSIS C

RELIABILITY, SAFETY AND RISK ANALYSIS A+B

June4th

Part 1 Part 2

May 19th

Piero Baraldi

Final Evaluation

33 Points Written exam ([30%;40%] of the final mark):

• Exercises (and questions?) on all the course topics Oral exam ([40%,50%] of the final mark):

• 2 questions (on all the course topics) Homework Monte Carlo (12% of the final mark) Presentation (6%) Bouns Exercise (2%)

1) It is necessary to have at least 18/30 in the written exam to be admitted to the oral2) It is necessary to have 30/30 in the written or oral exam to have 30 cum Laude

Piero Baraldi

Course Material

• Lecture slides (http://www.lasar.polimi.it/)• Zio E., An introduction to the basics of reliability and risk analysis,

World Scientific, 2007.• Zio E., Computational methods of reliability and risk analysis, World

Scientific, 2009.• Zio E., The Monte Carlo Simulation Method for System Reliability and

Risk Analysis• Zio E., Baraldi P., Cadini F., “Basics of Reliability and Risk Analysis:

Worked Out Problems and Solutions”. World Scientific, Singapore, 2011

Piero Baraldi

63

Piero Baraldi

64

Piero Baraldi

65

“RELIABILITY, SAFETY AND RISK ANALYSIS”

Documents

Transcript of “RELIABILITY, SAFETY AND RISK ANALYSIS”