Emanuele Borgonovo Market Structural Decision Return Quantitative Methods for Management First Edition Quantitative Methods for Management Emanuele Borgonovo 1 Chapter three: Models Quantitative Methods for Management Emanuele Borgonovo 2 Models • A Model is a mailmatical-logical instrument that the analyst, the manager, the scientist, the engineer develops to: – foretell the behaviour of a system – foresee the course of a market – evaluate an investment decision accounting for uncertainty factors • Common Elements to the Models: – Uncertainty – Assumptions – Inputs • Model Results Quantitative Methods for Management Emanuele Borgonovo 3 Building a Model • To build a reliable model requires deep acquaintance of: – the Problem – Important Events regarding the problem – Factors that influence the behavior of the quantities of interest – Data and Information Collection – Uncertainty Analysis – Verification of the coherence of the Model by means of empiric analysis and , if possible, analysis of Sensitivity Analysis Quantitative Methods for Management Emanuele Borgonovo 4 Example: the law of gravity • We want to describe the vertical fall of a body on the surface of the earth. We adopt the Model: F=mg for the fall of the bodies • Hypothesis (?): – – – – Punctiform Body (no spins) No frictions No atmospheric currents Does the model work for the fall of a body placed to great distance from the land surface? Quantitative Methods for Management Emanuele Borgonovo 5 Chapter II Introductory Elements of Probability theory Quantitative Methods for Management Emanuele Borgonovo 6 Probability • Is it Possible to Define Probability? • Yes, but there are two schools • the first considers Probability as a property of events • the second school asserts that Probability is a subjective measure of event likelihood (De Finetti) Quantitative Methods for Management Emanuele Borgonovo 7 Kolmogorov Axioms U B A P(U) 1 P( A ) 0 If A e B mutually esclusive events , P( A B) P( A ) P(B) Quantitative Methods for Management Emanuele Borgonovo 8 Areas and rectangles? U A B C D and U A B C D E • Suppose one jumps into the area U randomly. Let P(A) be the Probability to jump into A. What is its value? • It will be the area of A divided by the area of U: P(A)=A/U • Note that in this case: P(U)=P(A)+ P(B)+ P(C)+ P(D)+ P(E), since there are no overlaps Quantitative Methods for Management Emanuele Borgonovo 9 Conditional Probability • Consider events A and B. the conditional Probability of A given B, is the Probability of A given the B has happened. One writes: P(A|B) U B AB A Quantitative Methods for Management Emanuele Borgonovo 10 Conditional Probability • Suppose now that B has happened, i.e., you jumped into area B (and you cannot jump back!). B AB A •You cannot but agree that: •P(A|B)=P(AB)/P(B) •Hence: P(AB)=P(A|B) *P(B) Quantitative Methods for Management Emanuele Borgonovo 11 Independence • Two events, A and B, are independent if given that A happens does not influence the fact that B happens and vice versa. B B A AB A Thus, for independent events: P(AB)=P(A)*P(B) Quantitative Methods for Management Emanuele Borgonovo 12 Probability and Information • Problem: you are given a box containing two rings. the box content is such that with the same Probability (1/2) the box contains two golden rings (event A) or a golden ring and a silver one (event B). To let you know the box content, you are allowed to pick one ring from the box. Suppose it is a golden one. – In your opinion, did you gain information from the draw? – the Probability that the oil one is golden is 50%? – Would you pay anything to have the possibility to draw from the box? Quantitative Methods for Management Emanuele Borgonovo 13 In the subjectivist approach, Probability changes with information Quantitative Methods for Management Emanuele Borgonovo 14 Bayes’ theorem • Hypothesis: A and B are two events. A has happened. • Thesis: P(B) changes as follows: P(B) before A P(B A ) Probability of A given B P(B) P( A B) P( A ) New value of the Probability of B Quantitative Methods for Management Probability of A Emanuele Borgonovo 15 Let us come back to the ring problem • Events: • A: both rings are golden • o: the picked up ring is golden • the theorem states: P( A o) P( A ) P(o A ) P(o) • P(A)=Probability of both rings being golden before the extraction =1/2 • P(o)=Probability of a golden ring=3/4 • P(o|A)=Probability that the extracted ring is golden given A=1 (since both rings are golden) • So: Quantitative Methods for Management 1/ 2 1 P( A o) 2/3 3/4 Emanuele Borgonovo 16 Bayes’ theorem Proof Starting point P( AB) P( AB) Conditional Probability formula P( A B) P(B) P(B A ) P( A ) thesis P( A B) Quantitative Methods for Management P(B A ) P( A ) P(B) Emanuele Borgonovo 17 the Total Probability theorem D B Uand A C 1 P( A ) P(B) P(C) P(D) • the total Probability theorem states: given N mutually exclusive and exhaustive events A1, A2,…,AN, the Probability of an event and in U can be decomposed in: P(E) P(E A1 ) P( A1 ) P(E A 2 ) P( A 2 ) ... P(E AN ) P( AN ) • Bayes theorem in the presence of N events becomes : P( A 1 E) Quantitative Methods for Management P(E A 1 ) P( A 1 ) N P(E A ) P( A ) i1 i Emanuele Borgonovo 18 i Continuous Random Variables • Till now we have discussed individual events. there are problems in which the event space is continuous. For example, think of the failure time of a component or the time interval between two earthquakes. the random variable time ranges from 0 to +. • To characterize such events one resorts to Probability distributions. Quantitative Methods for Management Emanuele Borgonovo 19 Probability Density Function • f(x) is a Probability density function (pdf) if: – It is integrable – And – the integral of f(x) over -:+ is equal to 1. f ( x )dx 1 • Note: f(x0)dx is the Probability that x lies in an interval dx around x0. Quantitative Methods for Management Emanuele Borgonovo 20 Cumulative Distribution Function • Given a continuous random variable X, the Probability that X<x is given by: x P(X x ) f ( t )dt dP • If f(x) is continuous, then: f ( x ) dx X2 • Note: P( X1 x X2 ) f ( x )dx P( x X2 ) P( x X1 ) X1 Quantitative Methods for Management Emanuele Borgonovo 21 the exponential distribution • Consider events that happen continuously in time, and with continuous time T. • If the events are: – Independents – With constant failure rates • the random variable T is characterized by an exponential distribution: P(T t ) 1 e • and by the density function: Quantitative Methods for Management λt f ( t ) e t Emanuele Borgonovo 22 Meaning of the Exponential Distribution • We are dealing with a reliability problem, and we must characterize the failure time, T. T is a random variable: one does not know when a component is going to break. All one can say is that for sure the component will break between 0 and infinity. Thus, T is a continuous random variable. • Let us consider that failures are independent. This is the case if the failure of one component does not influence the failure of the other components. • Let us also consider constant failure rates. This is the case when repair brings the component as good as new and when the component does not age during its life. • Under these Hypothesis, the failure times are independent and characterized by a constant failure rate at every dt. What is the Probability distribution of T? • Let us consider a population of N(t) components at time t. If is the failure rate of a component, then N(t)dt is the number of failues in dt around time t. Quantitative Methods for Management Emanuele Borgonovo 23 the Exponential Distribution • Thus the change in the population is: • -N(t)dt=N(t+dt)-N(t)=dN(t) • Where the minus sign indicates that the number of working components has decreased. T T • Hence: dN(t ) dN( t ) N( t ) dt • Which solved leads: 0 N( t ) dt lnN(T) / N(0) T 0 N(T ) T e N(0) • N(T) is the number of components surviving till T. N(0) is the initial number of components. Set N(0)=1. then N(T)/N(0) is the Probability that a component survives till T. Quantitative Methods for Management Emanuele Borgonovo 24 Pdf and Cdf of the Exponential Distribution 1 0.9 0.8 0.7 P(t<T) 0.6 0.5 0.4 0.3 f(t) 0.2 0.1 0 0 5 10 15 T/t Quantitative Methods for Management Emanuele Borgonovo 25 Expected Value, Variance and Percentiles Expected Value : Ex xf ( x )dx Variance : Vx E ( x Ex ) 2 ( x Ex ) f ( x )dx E x Ex 2 2 2 S tan dard Deviation : Vx Percentile p: is the value xp of X such that the Probability of X being lower than xp is equal to p/100 Quantitative Methods for Management Emanuele Borgonovo 26 the Normal Distribution • Is a symmetric distribution around the mean • Pdf: 1 fG ( x ) e 2 1 x 2 ( ) 2 X • Cdf: 1 PG ( x X) e 2 Quantitative Methods for Management 1 x 2 ( ) 2 dx Emanuele Borgonovo 27 Graphs of the Normal Distribution Distribuzione Normale Standard 3000 2500 fG ( x) f(x) 2000 1500 1000 500 0 -4 -3 -2 -1 0 x 1 2 3 4 Cumulative Gaussian Distribution 10000 9000 8000 PG ( x X) 7000 6000 5000 4000 3000 2000 1000 0 -5 -4 -3 -2 -1 0 1 2 3 4 x Quantitative Methods for Management Emanuele Borgonovo 28 Lognormal Distribution • Pdf 1 fL ( x ) e x 2 1 ln x 2 ( ) 2 0 x • Cdf X 1 e 0 x 2 PL ( x X) Quantitative Methods for Management 1 ln x 2 ( ) 2 Emanuele Borgonovo 29 Lognormal Distribution .20 fL ( x) f ( x) 0.1 0 0 0 20 0.07 1 PL ( x X) x 50 1 f2( x) 0.5 0 0 0 0.07 Quantitative Methods for Management 40 20 40 x 50 Emanuele Borgonovo 30 Problem II-1 and solution • the failure rate of a car gear is 1/5 for year (exponential events). • What is the mean time to failure of the gear? t e t dt 1/ 5 • What is the Probability of the gear being integer after 9 years? P( t 9) 1 P( t 9) 1 (1 e T ) e (1/ 5 )9 16.5% Quantitative Methods for Management Emanuele Borgonovo 31 Problem II-2 • You are considering a University admission test for a particularly selective course. the admission test, as all tests test, is not perfect. Suppose that the true distribution of the class is such that 10% of the applicants are really qualified and 90% are not. then you perform the test. If a student is qualified, then the test will admit him/her with 90% Probability. If the student is not qualified he/her gets admitted at 10%. Now, let us consider a student that got admitted: – What is the Probability that the student is really qualified? – Is it a good test? How would you use it? – (Hint: use the theorem of Total Probability) Quantitative Methods for Management Emanuele Borgonovo 32 Problem II-3 • For the example of the two rings, determine: – P(B|o) – P(B|a) – the Probability of being in A given that the picked ring is golden in two consecutive extractions, having put the ring back in the box after the first extraction – the Probability of being in B given that the picked ring is golden in two consecutive extractions, having put the ring back in the box after the first extraction Quantitative Methods for Management Emanuele Borgonovo 33 Problem II-3 • For the example of the two rings, determine: – P(B|o) • Solution: there are only two possible events, A or B. Thus, P(Bor)=1P(Aor)=1/3 – P(Ba) • P(Ba)=1, since B is the only event that has a silver ring. One can also show it using Bayes’ theorem: • P(Ba)=P(aB)*P(B)/[P(aB)* P(B)+P(aA)*P(A)]. Since P(aA)=0, one gets 1 at once. – the Probability of being in A given that the picked ring is golden in two consecutive extractions, having put the ring back in the box after the first extraction • Using Bayes’ theorem: P( A 2o) Quantitative Methods for Management P(2o A ) P1( A ) P(2o A ) P1( A ) P(2o B) P1(B) Emanuele Borgonovo 34 Problem II-3 • where, in the formula, subscript 1 indicates the probabilities after the information of the first extraction has been taken into account: – P1(B)=P(Bor)=1/3 and P1(A)=P(Aor)=2/3. – One can note that P(2oA)=1, and P(2oB)=1/2. P(2oB) is the Probability to pick a golden ring at the second run, given that one is in state B. – Thus, we have all the numbers to be substituted back in the theorem: P( A 2o) P(2o A ) P1( A ) P(2o A ) P1( A ) P(2o B) P1(B) 1* 2 / 3 0 .8 1 * 2 / 3 1/ 2 * 1 / 3 – It is the same problem as in the example, but with adjourned probabilities. • the Probability of being in B given that the picked ring is golden in two consecutive extractions, having put the ring back in the box after the first extraction – Solution: 1-P(A2o)=0.2 Quantitative Methods for Management Emanuele Borgonovo 35 Chapter III: Introductory Decision theory Quantitative Methods for Management Emanuele Borgonovo 36 An Investment Decision • At time T, you have to decide whether, and how, to invest $1000. You face three mutually exclusive options: – (1) A risky investment that gives you $500 PV in one year if the market is up or a loss of $400 if the market is down – (2) A less risky investment that gives you $200 in one year or a loss of $160 – (3) the safe investment: a bond that gives you $20 in one year independently of the market Quantitative Methods for Management Emanuele Borgonovo 37 Decision theory According to Laplace • “the theory leaves nothing arbitrary in choosing options or in making decisions and we can always select, with the help of the theory , the most advantageous choice on our own. It is a refreshing supplement to the ignorance and feebleness of the human mind”. • Pierre-Simon Laplace • (March 28 1749 Beaumont-en-Auge - March 5 1827 Paris) Quantitative Methods for Management Emanuele Borgonovo 38 Decision-Making Process Steps Problem identification Alternatives identification Model implementation Alternatives evaluation Sensitivity Analysis Yes Further Analysis? No Best Alternatives implementation Quantitative Methods for Management Emanuele Borgonovo 39 Decision-Making Problem Elements • Values and Objectives • Attributes • Decision Alternatives • Uncertain Events • Consequences Quantitative Methods for Management Emanuele Borgonovo 40 Decision Problem Elements • Objectives: – Maximize profit • Attributes: – Money • Alternatives: – Risky – Less Risky – Safe • Random events: – the Market • Consequences: – Profit or Loss Quantitative Methods for Management Emanuele Borgonovo 41 Decision Analysis Tools • Influence Diagrams • Decision Trees Market up Less Risky Market How should the invest $1000? Risky Structural Decision Return Safe Quantitative Methods for Management prob_up Market down 1-prob_up Market up prob_up Market down 1-prob_up Emanuele Borgonovo 42 Influence Diagrams • Influence diagrams (IDs) are… “a graphical representation of decisions and uncertain quantities that explicitly reveals probabilistic dependence and the flow of information” • ID formal definition: – ID = a network consisting of a directed graph G=(N,A) and associated node sets and functions (Schachter, 1986) Quantitative Methods for Management Emanuele Borgonovo 43 ID Elements NODES ARCS • Informational Arcs = Decision • probabilistic Dependency Arcs = Random Event • Structural Arcs = utility Quantitative Methods for Management Emanuele Borgonovo 44 ID Elements Informational Arc Decision Node Decision Node Sequential Decisions Structural Chance Node Conditional Arc Chance Node Value Node probabilistic Dependency Quantitative Methods for Management Emanuele Borgonovo 45 Influence Diagram Levels 1. Physical Phenomena and Dependencies 2. “Function level”: node output states probabilistic relations (models) 3. “Number level”: tables of node probabilities Quantitative Methods for Management Emanuele Borgonovo 46 Case Study 2 - Leaking SG tube • Influence Diagram for Case Study 2 Leakage Rate shutdown_cost Leakage from primary to secondary, maximum rate of 20 l/hr time_to_repair Primary Decisions I - Normal Makeup II - Shutdown III - Reduce Power IV - Isolate SG Cooling Chemical Volume Control System Value Secondary Cooling days_to_shutdown Deterministic Information core_damage_cost Quantitative Methods for Management Emanuele Borgonovo 47 Influence Diagram Market Structural Decision Return Quantitative Methods for Management Emanuele Borgonovo 48 Decision Trees • Decision Trees (DTs) are constituted by the same type of arcs of Influence Diagrams, but highlight all the possible event combinations. • Instead of arks, one finds branches that emanate from the nodes as many as the Alternatives or Outcomes of each node. • With respect to Influence Diagrams, Decision Trees have the advantage of showing all possible patterns, but their structure becomes quite complicated at the growing of the problem complexity. Quantitative Methods for Management Emanuele Borgonovo 49 the Decision Tree (DT) Market up Less Risky Market down How should the invest $1000? 1-prob_up Risky Market up Market down Safe Quantitative Methods for Management Emanuele Borgonovo 50 Decision Tree Solution • Alternative Payoff or utility: E[Ui ] Pi (C j ) UC j j • j=1…mi spans all the Consequences associated to alternative the • Uj is the utility or the payoff of consequence j • Pi(Cj) is the Probability that consequence Cj happens given that one chose alternative the • In general, we will get: P(Cj) =P(E1E2… EN), where E1E2… EN are the events that have to happen so that consequence Cj is realized. Using conditional probabilities: • P(Cj) =P(E1E2… EN)=P(EN| E1E2… )*…*P(E2| E1)*P(E1) Quantitative Methods for Management Emanuele Borgonovo 51 example Market up Blue Chip Stock How should the invest $1000? Risky investment P.up Market down C1 C2 1-P.up Market up P.up Market down 1-P.up C3 C4 CD paying 5% C5 Quantitative Methods for Management Emanuele Borgonovo 52 Problem Solution • Using the previous formula: 2 E[URisky ] P(C j ) C j P.up C1 (1 P.up) C2 j1 2 E[ULessRisky ] P(C j ) C j P.up C3 (1 P.up) C4 j1 2 E[ULessRisky ] P(C j ) C j 1 C5 j1 Quantitative Methods for Management Emanuele Borgonovo 53 the Best Investment for a Risk Neutral Decision - Maker Market up Blue Chip Stock 0.600 $56 Market down $200 ($160) 0.400 Market up How should the invest $1000? Risky investment $500; P = 0.600 $60 0.600 Market down ($600); P = 0.400 0.400 CD paying 5% return = $50 Quantitative Methods for Management Emanuele Borgonovo 54 Run or Withdraw? You are the owner of a racing team. It is the last race of the season, and it has been a very good season for you. Your old sponsor will remain with you for the next season offering an amount of $50000, no matter what happens in the last race. However, the race is important and transmitted on television. If you win or end the race in the first five positions, you will gain a new sponsor who is offering you $100000, besides $10000 or $5000 praise. However there are unfavorable running conditions and an engine failure is likely, based on your previous data. It would be very bad for the image of you racing team to have an engine failure in such a public race. You estimate the damage to a total of -$30000. What to do? Run or withdraw? • A) Elements of the problem: – – – – – What are your objectives What are the decision alternatives What are the attributes of the decision What are the uncertain events What are the alternatives Quantitative Methods for Management Emanuele Borgonovo 55 Example of a simple ID Decision Quantitative Methods for Management Engine failure Final Classification Profit Emanuele Borgonovo 56 From IDs to Decision Trees Engine failure failure Decision Out of first five $20,000 0.500 Run $20,000; P = 0.500 1.000 Win $57,250 $110,000; P = 0.250 0.500 No failure Decision 0.500 Run : $57,250 pfailure=0.5 In first five $94,500 0.300 Out of first five $50,000; P = 0.100 pfive=0.30 0.200 pout=0.2 pwin=0.5 $105,000; P = 0.150 Withdraw Engine_failure=0 Quantitative Methods for Management Old sponsor $50,000 $50,000 1.000 Emanuele Borgonovo 57 Sequential Decisions • Are decision making problems in which more than one decisions are evaluated one after the other. • You are evaluating the purchase of a production machine. Three models are being judged, A B and C. the machine costs are 150, 175 and 200 respectively. If you buy model A, you can choose insurance A1, that covers all possible failues of A, and costs 5% of A cost, or you can choose insurance policy A2, that costs 3% of A cost, but covers only transportation risk. If you buy model B, insurance policy B1 costs 3% of B cost and covers all B failures. Insurance B2 costs 2% of B and covers only transportation. For model C, the most reliable, the insurance coverages cost 2% and 1.5% respectively. Based on this information and supposing that the machines production is the same, what will you choose? • (failure Probability of A in the period of interest=5%) • (failure Probability of B in the period of interest=3%) • (failure Probability of C in the period of interest=2% Quantitative Methods for Management Emanuele Borgonovo 58 Influence Diagram Decision Quantitative Methods for Management Assicurazione Ruttura Costo Emanuele Borgonovo 59 Decision Tree Assicurazione Decision 1 -150-5%*(150) = (£158); P = 1.000 A 1 : (£158) 2 Sì 0.050 -150-2%*150-150 = (£303) (£161) No 0.950 -150*(1+2%) = (£153) 1 -(175+3%*(175)) = (£180) B A : (£158) pA=0.05 pB=0.03 pC=0.02 1 : (£180) Sì 2 0.030 -175-2%*175-175 = (£354) (£184) No 0.970 -175-2%*175 = (£179) 1 -200-2%*200 = (£204) C 1 : (£204) 2 Sì 0.020 -200-1.5%*200-200 = (£403) (£207) No 0.980 Quantitative Methods for Management -200-200*1.5% = (£203) Emanuele Borgonovo 60 the Expected Value of Perfect Information • Data and information collection is essential to make decisions. Sometimes firms hire consultants or experts to get such information. But, how much should one spend? • One can value information, since it is capable of helping the decision-maker in selecting among alternatives • the value of information is the added value of the information. • the expected value of perfect information (EVPI) assumed that the source of information is perfect, and then: EVPI E[Knowing ] E[BeforeKnow ing] • the definition is read as follows: how much is the decision worth with the new information and without • N.B.: we will refer only to aleatory uncertainty Quantitative Methods for Management Emanuele Borgonovo 61 Example: investing Decision Value Market Market Decision Up RISKY 0.500 £500; P = 0.500 £50 Down 0.500 (£400); P = 0.500 Up LESS: £50 RISKY RISKY P _UP =0.5 0.500 £200 £20 Down 0.500 (£160) SAFE £20 Market=0 Quantitative Methods for Management Emanuele Borgonovo 62 EVPI for the Example Market Value Decision Decision Market Up 0.500 RISKY £500; P = 0.500 LESS RISKY RISKY : £500 £200 SAFE £20 £260 RISKY P _UP =0.5 (£400) Down 0.500 LESS RISKY SAFE : £20 (£160) SAFE £20; P = 0.500 Quantitative Methods for Management Emanuele Borgonovo 63 EVPI Result EVPI E[Knowing ] E[r ] 260 50 210 Quantitative Methods for Management Emanuele Borgonovo 64 Problems Quantitative Methods for Management Emanuele Borgonovo 65 How much to bid? • Bob works for an energy production company. Your company is engaged in the decision of how much to bid to salvage the wreckage of the SS.Kuniang, a carbon transportation boat. If the firm wins, the boat could be repaired and could come back to its transportation activity again. Pending on the possible winning and on the decision is the result of a judgment by Coast Guard, which will be revealed only after the opening of the bids. That is, if the Coast Guard will assign a low value to the ship, this would mean that the ship is considered as recoverable. Otherwise, the boat will be deemed unusable. If you do not win, you will be forced to buy a new boat. • Identify the decision elements • Structure the corresponding ID and DT Quantitative Methods for Management Emanuele Borgonovo 66 Influence Diagram with three events • Given the following elements: – Alternatives 1 and 2 – Events: A=(up, down); (B=high, low);(C=good, bad); – Consequences Ci (one distinct consequence for each event combination) – If A=Down happens, then CAdown is directly realized • Draw the ID corresponding to the problem • Draw the corresponding Decision Tree • If C now depends on both A and B outcomes, how does the ID become? • How does the DT change? Quantitative Methods for Management Emanuele Borgonovo 67 Solution • Influence Diagram the Skip Arc A C Decision Consequences B Quantitative Methods for Management Emanuele Borgonovo 68 Solution • Corresponding Decision Tree C B A good (No Payoff) high bad up Decision (No Payoff) good (No Payoff) low 1 bad (No Payoff) down (No Payoff) B=0 C=0 good high (No Payoff) bad (No Payoff) up good (No Payoff) low 2 bad (No Payoff) down (No Payoff) B=0 C=0 Quantitative Methods for Management Emanuele Borgonovo 69 Solution • Influence Diagram II A C Decision Consequences B Quantitative Methods for Management Emanuele Borgonovo 70 Solution • Decision Tree II: C B A good (No Payoff) high bad up (No Payoff) good Decision (No Payoff) low 1 bad (No Payoff) good (No Payoff) down bad B=0 (No Payoff) good (No Payoff) high bad (No Payoff) up good (No Payoff) low 2 bad (No Payoff) good (No Payoff) down B=0 Quantitative Methods for Management bad (No Payoff) Emanuele Borgonovo 71 Sales_Costs • Given the following Influence Diagram and Decision Tree, given P_High and P_High|High, P_high|low, find the value of the Alternatives as a function of the assigned probabilities. Supposing P_high=0.5 and P_high|high=P_high|low=0.3, find the preferred alternative. Sales Cost Vendite Payoff Decisione high Decision P_high Costo Invest High P_Alte|high Low 1- P_Alte|high High Basso high=0.5 P_Alte=0.3 P_high=0.5 • • 1-P_high -10 20 P_ P_Alte|high Low 1- P_Alte|high Do not Invest 0 0 5 What would be the preferred decision if to a higher investment cost there would correspond a better sale result? Set: P_high|high=0.6 and P_high|low=0.2 Quantitative Methods for Management Emanuele Borgonovo 72 Solution Sales_Costs Vendite Costo Decisione Alte Alto 0.500 0.300 (£7) Basse Investo 0.700 (£1) alto=0.5 P _Alte=0.3 P _alto=0.5 (£10) Alte Basso Non-Investo : £5 £0 0.500 0.300 £6 £20 Basse 0.700 £0 Non-Investo £5; P = 1.000 Quantitative Methods for Management Emanuele Borgonovo 73 Breakdown in Production • An industrial system composed from two lines has experience a breakdown in one line. Production, therefore, is reduced by 50%. the management asks you collaboration on the following decision. It is explained to you that there are two ways to proceed: 1) an intermediate repair, of the duration of two days, with a repair cost of EUR500000. For every day of production loss of EUR25000 for day is sustained (Full production amounts at EUR50000). From the engineer estimates, the Probability of perfect repair in two days is equal to P_2g. In the case in which the repair it is not perfect (partial repair), the line will come back with a loss of 15% of the productive ability; 2) a more incisive intervention, of the duration of 10 days, with a cost of repair of EUR1000000. With Probability P_10g the line will be as before the breakdown. – – – – According to you, the residual life of the system is important for the decision? Suppose that there are still three years of life for the system. Which strategy should you carry out? Determine the decision problem elements. Draw the Influence Diagram and the corresponding Decision Tree. Find the value or values of the probabilities for which a complete repair is more convenient than a partial one. – What would you would advise to the director of the system to do based on the engineer estimates? Quantitative Methods for Management Emanuele Borgonovo 74 EVPI Problems • Determine the EVPI for the random event nodes in the previous IDs and DTs of the following problems: • Sales_Costs (lez. 2) • Production break-down (lez.2) Quantitative Methods for Management Emanuele Borgonovo 75 Troubles in Production • • • One of the two production lines of the plant you manage has broke down. the plant production capacity is therefore halved. the management faces the following decision and asks you a collaboration. Technically one can a: 1) perform an temporary repair, lasting two days, and costing €500000. For every lost production day one has a revenue loss of €25000 for day (the total daily production value is €50000). Based on the Engineer estimates, the Probability of perfect repair in two days is P_2g . In the case of an imperfect repair, the production capacity will be lowered by 15%. 2) perform a more incisive repair, lasting 10 days, and costing €1000000. With Probability P_10g the line will be as good as new. In your opinion, the residual plant life is relevant to this decision? Suppose that there are still three years of life for the plant. What should one decide? – Identify the decision making elements – Draw the Influence Diagram for the problem – Find the values of the probabilities for which one or the other intervention is more convenient – What would your suggestion to the plant director be? – What would happen if the plant life were 2 and 4 years instead of 3? Quantitative Methods for Management Emanuele Borgonovo 76 Influence Diagram Riparazione_10g_Perfetta Perdite Decisione Riparazione_2g_perfetta Quantitative Methods for Management Emanuele Borgonovo 77 Decision Tree Riparazione_2g_perfetta Decisione Riparazione_2g_perfetta -50000-500000 Intervento_2g P_2g Riparazione_2g_non_perfetta Riparazione_10g_Perfetta=0 -50000-500000-25000*0.15*365*years 1-P_2g Riparazione_10g_Perfetta P_10g=0.9 P_2g=0.3 years=3 Riparazione_10g_Perfetta -250000-1000000 Intervento_10g Riparazione_2g_perfetta=0 P_10g Riparazione_10g_non_Perfetta -1000000-250000-years*.05*25000*365 Riparazione_2g_perfetta=0 1-P_10g Quantitative Methods for Management Emanuele Borgonovo 78 Probability Values • Three years Quantitative Methods for Management Emanuele Borgonovo 79 2 and 4 years • 2 years • 4 years Quantitative Methods for Management Emanuele Borgonovo 80 Chapter IV Elements of Sensitivity Analysis Quantitative Methods for Management Emanuele Borgonovo 81 Sensitivity Analysis • Various Types of SA – One Way SA – Two Way SA – Tornado Diagrams – (Differential Importance Measure) • Uncertainty Analysis – Monte Carlo – (Global SA) Quantitative Methods for Management Emanuele Borgonovo 82 How do we use SA? • a) To check model correctness and robustness • b) To Further interrogate the model – Questions: • What is the most influential parameter with respect to changes? • What is the most influential parameter on the uncertainty (data collection) Quantitative Methods for Management Emanuele Borgonovo 83 Sensitivity Analysis (Run or withdraw) • Underline the critical dependencies of the outcome Tornado Diagram at Decision Sensitivity Analysis on pfailure $62K pwin: 0.3 to 0.7 $59K pfive: 0.2 to 0.4 $56K Expected Value pfailure: 0.25 to 0.75 Run Withdraw Threshold Values: pfailure = 0.597 EV = $50K $53K $50K $47K $44K $41K $49K $55K $61K $67K Expected Value Quantitative Methods for Management $73K $38K 0.450 0.525 0.600 0.675 0.750 pfailure Emanuele Borgonovo 84 Summary • Sensitivity Analysis – One way sensitivity – Two way sensitivity – Tornado Diagrams • Uncertainty Analysis – Aleatory Uncertainty – Epistemic Uncertainty – Bayes‘ theorem for continuous distributions – Monte Carlo Method Quantitative Methods for Management Emanuele Borgonovo 85 Sensitivity Analysis • By sensitivity analysis one means the study of the change in results (output) due to a change in one of the model parameters (input) • the simplest Sensitivity Analysis types are: – One way sensitivity – Two way sensitivity – Tornado diagrams Quantitative Methods for Management Emanuele Borgonovo 86 One-way Sensitivity Analysis • A one way sensitivity is obtained changing the Model input variables one at a time, and registering the change in the decision value. • It enables the analyst to study the change in value of each of the alternatives with respect to the change in the input parameter under consideration Sensitivity Analysis on pfailure $62K Run Expected Value $59K Withdraw $56K Threshold Values: pfailure = 0.597 EV = $50K $53K $50K $47K $44K $41K $38K 0.450 0.525 0.600 0.675 0.750 pfailure Quantitative Methods for Management Emanuele Borgonovo 87 Two-way Sensitivity Analysis • In a Two-way Sensitivity Analysis two parameters are varied at the same time. • Instead of a line, one obtains a plane, in which each region identifies the preferred alternative that correspond to the combination of the two parameter values Quantitative Methods for Management Emanuele Borgonovo 88 Tornado Diagrams • the analysis is focused on the preferred decision • An interval of variation for each input parameter is chosen • the parameters are changed one at a time, while keeping the oilrs at their reference value • the change in output is registered • the output change is shown by means of a horizontal bar • the most important variable is the one that corresponds to the longest bar. Quantitative Methods for Management Emanuele Borgonovo 89 Example of a Tornado Diagram Tornado Diagram at Decision pfailure: 0.25 to 0.75 pwin: 0.3 to 0.7 pfive: 0.2 to 0.4 $49K $55K $61K $67K $73K Expected Value Quantitative Methods for Management Emanuele Borgonovo 90 Upsides and Downsides • Upsides – Easy numerical calculations – Results immediately understandable Quantitative Methods for Management • Downsides – Input range of variation not considered together with the output range: should not be used to infer parameter importance – One or two parameters can be varied at the same time Emanuele Borgonovo 91 Sensitivity Analysis and Parameter Importance • Parameter importance: – Relevance of parameter in a model with respect to a certain criterion • Sensitivity Analysis used to Determine Parameter Importance • Concept of importance not formalized, but extensively used – Risk-Informed Decision Making – Resource allocation • Need for a formal definition Quantitative Methods for Management Emanuele Borgonovo 92 Process • Identify how sensitivity analysis techniques work through analysis of several examples • Formulate a definition • Classify sensitivity analysis techniques accordingly Quantitative Methods for Management Emanuele Borgonovo 93 Sensitivity Analysis Types • Model Output: U f(x1, x2,..., xn ) • Local Sensitivity Analysis: – Determines model parameter (xi) relevance with all the xi fixed at nominal value • Global Sensitivity Analysis: – Determines xi relevance of xi’s epistemic/uncertainty distribution Quantitative Methods for Management Emanuele Borgonovo 94 the Differential Importance Measure • Nominal Model output: – No uncertainty in the model parameters – and/or parameters fixed at nominal value • Local Decomposition: f f f dU dx1 dx 2 ... dx n x1 x 2 x n • Local importance measured by fraction of the differential attributable to each parameter DIM(x i ) Quantitative Methods for Management dUxi dU xo Emanuele Borgonovo 95 Global Sensitivity Indices • Uncertainty in U and parameters is considered • Sobol’’s decomposition theorem: n U f( x) f0 fi ( xi ) i1 • Sobol’Indices f ( x , x ) ... f ij 1i jn i j 12...n x i1 Si1...is (x i ) Di1...is D ... f i1...in dx i1 ...dx n x i1 f 2 ( x )dx f0 2 Ω Quantitative Methods for Management Emanuele Borgonovo 96 ( x) Formal Definition of Sensitivity Analysis (SA) Techniques • SA technique are Operators on U: x1 x2 xn or I(x)^ [U f(x1, x2,..., xn )] or I(xn) Quantitative Methods for Management I(x1) I(x2) Emanuele Borgonovo 97 Importance Relations • Importance relations: – X the set of the model parameters; – Binary relation xi xj iff I(xi)I(xj) xi~xj iff I(xi)I(xj) xi xj iff I(xi)I(xj) xi xj iff I(xi)I(xj) • Importance relations induced by importance measures are complete preorder Quantitative Methods for Management Emanuele Borgonovo 98 Additivity Property • In many situation decision-maker interested in joint importance: I( x i , x j ) I( x i x j ) • An Importance measure is additive if: I( x i , x j ) I( x i ) I( x j ) • DIM is additive always • Si are additive iff f(x) additive and xj’s are uncorrelated Quantitative Methods for Management Emanuele Borgonovo 99 Techniques that fall under the definition of Local SA techniques IMPORTANCE MEASURE EQUATION TYPE ADDITIVE DIM dU xi Local Yes Local No Local No Local No Local No Local No dU L Tornado Diagrams One Way Sensitivity Fussell-Vesely U x i x0 x i U U U U( x 0 ) x i U( x 0 ) Risk Achievement Worth U xi x0 U0 Quantitative Methods for Management Emanuele Borgonovo 100 Global Importance Measures IMPORTANCE MEASURE Sobol’ Indices EQUATION TYPE ADDITIVE Di1...is Global No Global No Global No Global No Global No Global No D Extended Fast Si 2 A p2w i Bp2w i p 1 2 A 2j B 2j j1 Morris Pearson Smirnov Standardized regression coefficients Quantitative Methods for Management d(x i ) i f ( x1,..., x i ,..., x n ) cov(U, x i ) i U sup Y1( Xi ) Y2 ( Xi ) bk k Emanuele Borgonovo 101 Sensitivity Analysis in Risk-Informed Decision-Making and Regulation • Risk Metric: R f(x , x ,..., x ) 1 2 n • xi is undesired event Probability • Fussell-Vesely fractional Importance: (R, x ) FV(x ) i i R • Tells us on which events regulator has to focus attention Quantitative Methods for Management Emanuele Borgonovo 102 Summary of the previous concepts • Formal Definition of Sensitivity Analysis Techniques • Definition of Importance Relations • Definition enables to: – Formalize use of Sensitivity Analysis – Understand role of Sensitivity Analysis in Riskinformed Decision-making and in the use of model information Quantitative Methods for Management Emanuele Borgonovo 103 Chapter V Uncertainty Analysis Quantitative Methods for Management Emanuele Borgonovo 104 Uncertainty Analysis Monte Carlo Simulation at Decision 1.000 0.900 Probability 0.800 0.700 0.600 0.500 0.400 0.300 0.200 0.100 0.000 $10K $40K $70K $100K $130K Value Quantitative Methods for Management Emanuele Borgonovo 105 Summary • Distinction between Aleatory Uncertainty ed Epistemic Uncertainty • Epistemic Uncertainty and Bayes‘ theorem • Monte Carlo Method for uncertainty propagation Quantitative Methods for Management Emanuele Borgonovo 106 Uncertainty • Aleatory Uncertainty: – From “Alea”, die: “Alea jacta est” It refers to the realization of an event. – Example: the happening of an earthquake • Epistemic Uncertainty: – From GreeK “Eit”, knowledge it reflects our lack of knowledge in the value of the Aleatory Model input parameters. the aleatory model or model of the world is the model chosen to represent the random event. Quantitative Methods for Management Emanuele Borgonovo 107 Example: Model of the World • the Probability of Earthquakes is usually modeled through a Poisson model: n P(n, t ) e t ( t ) n! • that rappresents the Probability that the number of earthquakes between 0 and t is equal to n. • the Poisson Distribution holds for independent events, in which next events (arrivals) are not influenced by previous events and the Probability of an event in a given interval of time is the same independently of the time where the interval is located • the Model chosen to describe the arrivals of earthquakes is given the non-humble name of "model of the world" (MOW). Quantitative Methods for Management Emanuele Borgonovo 108 Some useful information on Poisson Distributions • the Poisson Probability that n events happen on 0-t is: e t (t )n P(n, t ) n! • the sum on n=0... of P(n,t) is, obviously, equal to 1. et (t )n (t )n t e et et 1 n! n 0 n0 n! • the Probability of k>N is given by: N e t (t )n et (t )n 1 n! n! nN1 n 0 • E[n]=t Quantitative Methods for Management ne t (t )n (t )n (t )n1 t t e e t et t et n! n 0 n0 n 1! n0 n 1! Emanuele Borgonovo 109 the Corresponding Epistemic Model • Now,in spite of all the efforts and studies, it is unlikely that a scientist would tell you: the rate ( ) of arrivals of earthquakes is exactly xxx. More likely, he will indicate you a range where the “true value” of lies. For example cuold be between 1/5 and 1/50 (1/years). Suppose that the scientist state of knowledge on can be expressed by a uniform distribution u( ): Epistemic distribution for the frequency of earthquakes 8 u( ) 0 1/ 50 1/ 5 u( ) 1/ 5 1/ 50 1/ 50 1/ 5 7 6 f(lambda) 5 4 3 2 1 0 Quantitative Methods for Management 0 0.02 0.04 0.06 0.08 0.1 0.12 lambda 0.14 0.16 0.18 0.2 Emanuele Borgonovo 110 Combining the Epistemic Model and MOW • We have been dealing with two Models: • MOW: the events happen according to a Poisson Distribution • Epistemic Model: Uniform Uncertainty Distribution • then, what is the Probability of having 1 earthquake in the next year? • Answer: there is no unique Probability, but a p(n,t, ) for all values of . • Thus, we have to write: e t (t )n p(n, t, )d u( )d n! Quantitative Methods for Management Emanuele Borgonovo 111 …. • This expression tells us that not necessarily all Poisson distributions weight the same. Thus: t e (t )n P(n, t ) p(n, t, )d u( )d n! • In our case: u()=c; λt E[P(n, t )] p(n, t , λ)dλ c e (λt ) n dλ n! • Hence, there is an expected Probability! Quantitative Methods for Management Emanuele Borgonovo 112 In General • the MOW will depend on m parameters , ,…: MOW (t ,,.... ) • the event Probability (P(t)) will be: P( t ) MOW ( t , ,.... )f (, ,....) d d..... Quantitative Methods for Management Emanuele Borgonovo 113 An problem • the failure time of a series of components is characterized by the exponential Probability function : t df e dt • From the available data, it emerges that: 1/ 5 p 0.5 1/ 8 p 0.3 1/ 10 p 0.2 • What is the mean time to failure? Quantitative Methods for Management Emanuele Borgonovo 114 Solution • E[t]= 0 0 0 ( t 1e 1t dt ) 0.5 ( t 2e 2t dt ) 0.5 ( t 3e3 t dt ) 0.1 1/ 1 0.5 1/ 2 0.3 1/ 3 0.1 6.9 Quantitative Methods for Management Emanuele Borgonovo 115 Continuous form of Bayes Theorem • Epistemic Uncertainty and Bayes’ theorem are connected, in that we know that we can use evidence to update probabilities. • For example, suppose to have a coin in your hands. will it be a fair with, i.e., will the Probability of tossing the coin lead to 50% head and tails?. • How can we determine whether it is a fair coin? • ….let us toss it…. Quantitative Methods for Management Emanuele Borgonovo 116 Formula • the Probability density of a parameter, after having obtained evidence and, changes as follows: π( λ ) L(E λ ) π 0 ( λ ) L(E λ) π 0 ( λ )dλ • L(E) = MOW likelihood • 0() is the pdf of before the evidence, called Prior Distribution • () is the pdf of after the evidence, called Posterior Distribution Quantitative Methods for Management Emanuele Borgonovo 117 From discrete to continuous • Let us take Bayes‘ theorem for discrete events: P( A j E) P(E A j ) P( A j ) n P(E A ) P( A ) i i i1 • Let us go to continuous events: our purpose is to know the Probability that a parameter of the MOW distribution assumes a certain value, given a certain evidence • Thus, event Aj is: takes on value * • Hence: P(Aj)0()d 0()=prior density • therefore: P(EAj) has the meaning of Probability that the evidence and is realized given that equals * . One writes: L(E ) and it is the likelihood function • Note: it is the MOW!!! Quantitative Methods for Management Emanuele Borgonovo 118 From discrete to continuous • the denominator in Bayes’ theorem expresses the sum of the probabilities of the evidence given all the possible states (the total Probability theorem). In the case of epistemic uncertainty these events are all possible values of . Thus: n P(E A ) P( A ) L(E ) ()d i1 i i 0 • Substituting the various terms, one finds Bayes‘ theorem for continuous random variables we have shown before Quantitative Methods for Management Emanuele Borgonovo 119 Is it a fair coin? • What is the MOW? • It is a binomial distribution with parameter p: n k P(k,n k ) p (1 p)nk k • What is the value of p? • Suppse we do not know anything about p. Let us assume a uniform prior distribution between 0 and 1: π 0 (p) 1 if 0 p 1 π 0 (p) 0 otherwise • Let us get some evidence. • At the first tossing it is head • At the second tail • At the third head Quantitative Methods for Management Emanuele Borgonovo 120 Result • • • First tossing – Evidence: h. – MOW: L(hp)=p – Prior: 0 Second tossing: – Evidence: t – MOW: L(tp)=(1-p) – Prior: 1 Third tossing: – Evidence: h – MOW: L(hp)=p – Prior: 2 • Equivalently: – Evidence: h, t, h – L(hthp)=p2(1-p) – Prior: 0 Quantitative Methods for Management π1 (p) L( h p) π 0 ( p ) p 1 2p 1 L(h p) π (p)dp pdp 0 π 2 ( p) 0 L( t p) π1 (p) p (1 p) 1 L(t p) π (p)dp (p p 1 2 )dp 0 6(p p ) 2 π 3 ( p) L( h p ) π 2 ( p ) L(h p) π 2 (p)dp p 2 (1 p) 1 2 p (1 p)dp 0 12(p p ) 2 π 3 ( p) 3 L(hth p) π 0 (p) L(hth p) π 2 (p)dp p 2 (1 p) 1 1 2 p (1 p) 1dp 0 12(p 2 p3 ) Emanuele Borgonovo 121 Graph 2 1.8 3 1.6 2 1.4 1.2 1 1 0 0.8 0.6 0.4 0.2 0 0 Quantitative Methods for Management 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Emanuele Borgonovo 122 Conjugate Distributions • Likelihood • Prior distribution – Poisson – Gamma t e ( t ) P(n, t ) n! n • Posterior: gamma β α λα 1 βλ π 0 (λ, α, β) e Γ(α) • with: '' ' 1 ' (, ' ,' ) e (' ) Quantitative Methods for Management ' r ' t Emanuele Borgonovo 123 Conjugate Distributions • Likelihood • Prior distribution: – Normal 1 fX ( x ) e σ x 2π – Normal 1 x μx 2 ( ) 2 σx • Posterior: Normal 1 π 0 (m) e σμ 2π • with: μ ' 1 fG ( x ) e σ' x 2π Quantitative Methods for Management 1 x μ' ( ' )2 2 σx σ 'x 1 mμx 2 ( ) 2 σμ μ(σ x )2 nx(σμ0 )2 (σ x )2 n(σμ0 )2 (σ x / n)2 (σ μ )2 ( σ μ ) 2 ( σ x )2 / n Emanuele Borgonovo 124 Conjugate Distributions • Likelihood – Binomial n k p (1 p)nk k • Posterior, Beta: • Prior: – Beta π 0 (p) p( q1) (1 p)r 1 • with: q' q k π1(p) p( q' 1) (1 p)r ' 1 Quantitative Methods for Management r' r n k Emanuele Borgonovo 125 Summary of Conjugate Distributions MOW - Likelihood Prior Distribution Posterior Distribution Binomiale Beta Beta Poisson Gamma Gamma Normal Normal Normal Normal Gamma Gamma Negative binominal Beta Beta Quantitative Methods for Management Emanuele Borgonovo 126 Epistemic Uncertainty in Decision-Making Problems • Investment: 2 E[URisky ] P(C j ) C j P.up C1 (1 P.up) C2 j1 2 E[ULessRisky ] P(C j ) C j 1 C5 2 j1 E[ULessRisky ] P(C j ) C j P.up C3 (1 P.up) C4 j1 • Suppose that P.up is characterized by a uniform pdf between 0.3 and 0.7 2 E[URisky P.up ] P(C j ) C j P.up U1 (1 P.up) U2 j1 • How does the decision changes? • It is necessary to propagate the uncertainty in the model Quantitative Methods for Management Emanuele Borgonovo 127 Analytical Propagation of Uncertainty • It is the same problem of the MOW … E URisky E[URisky P.up ] f (P.up ) dP.up (P.up U 1 (1 P.up) U2 )f (P.up )dP.up • Repeating for the other decisions and comparing the resulting mean values, one gets the optimal decision. • Recall that: E[g( x1, x 2 ,..., x n )] g( x ) f ( x )dx Quantitative Methods for Management Emanuele Borgonovo 128 the Monte Carlo Method • Sampling a value of P.up • For all sampled P.up the Model is re-evaluated. • Information: – Frequency of the preferred alternative – Distribution of each individual Alternative Quantitative Methods for Management Emanuele Borgonovo 129 the core of Monte Carlo • 1) Random Number Generator “u” between 0 and 1 0 1 u • 2) Numbers u are generated with a uniform Distribution • 3) Suppose that parameter is uncertain and characterized by the cumulative distribution reported below: Distribuzione cumulativa esponenziale 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Quantitative Methods for Management 0 0.1 0.2 0.3 0.4 0.5 x 0.6 0.7 0.8 0.9 1 Emanuele Borgonovo 130 Inversion theorem 1 Distribuzione cumulativa esponenziale 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 • Inversion theorem: 0 0 0.1 0.2 0.3 0.4 0.5 x 0.6 0.7 0.8 0.9 1 1 F (u) • the values of sampled in this way have the Probability distribution from which we have inverted Quantitative Methods for Management Emanuele Borgonovo 131 Example • Let us evaluate the volume of the yellow solid through the Monte Carlo method. V0 V nin V lim V0 n N Quantitative Methods for Management Emanuele Borgonovo 132 Application to ID and DT • For every Model parameter one creates the corresponding epistemic distribution • Run nr. 1: • One generates n random numbers between 0 and 1, as many as the uncertain variables are • One samples the value of each of the parameters inverting from the corresponding distribution • Using these values one evaluates the model • One keeps record of the preferred alternative and of the value of the decision • the procedure is repeated N times. Quantitative Methods for Management Emanuele Borgonovo 133 Results • Strategy Selection Frequency • Decision Value Distribution Monte Carlo Simulation at Decision 1.000 0.900 Probability 0.800 0.700 0.600 0.500 0.400 0.300 0.200 0.100 0.000 $10K $40K $70K $100K $130K Value Quantitative Methods for Management Emanuele Borgonovo 134 Problem V-1 • the mean time to failure of a set of components is characterized by an exponential distribution with parameter . Suppose that is described by a uniform epistemic distribution between 1/100 and 1/10. – Which is the MOW? Which is the epistemic model? – What is the mean time to failure? • Suppose you registered the following failure times: t=15, 22, 25. – Update the epistemic distribution based on the new data – What is the new mean time to failure? Quantitative Methods for Management Emanuele Borgonovo 135 Problem V-2: Investing • • • • • • We are again thinking of how to invest. Actually, we were not aware of the bayesian approach before. Thus we start using data about P_up in Bayesian way. After 15 working days we get the evidence: up,down, down,down,down,up,down,up,down,up,down,up,up,up. Assuming that each day is independent of the previous one: a) Which are the MOW and the epistemic model? b) What is the best decision without incorporating the evidence? c) What is the distribution of P_up after the evidence? d) What do you decide when the new information is incorporated in the model? Solution: – – • the MOW is the model of the events that accompany the decision. It is our ID or DT. More in specific, there is a second mode which is the one utilized for modeling the fact that the market can be up or down. This is a binomial distribution with parameter P_up. the epistemic model is the set of the uncertainty distributions used to characterize the lack of knowledge in the model parameters. In this case, it is the distribution of P_up. We need to choose a prior distribution for P_up. We choose a uniform distribution between 0 and 1. b) We write the alternative payoffs as a function of P_up. Quantitative Methods for Management Emanuele Borgonovo 136 Prob. 5-2 E[ULessRisky P _ up ] C5 2 E[URisky P_up ] P(C j ) C j P_up C1 (1 P_up) C2 j1 2 E[ULessRisky P _ up ] P(C j ) C j P_up C3 (1 P_up) C4 j1 2 E[ U LessRisky P _ up ] P(C j ) C j 1 C5 j1 • Substituting: E[URisky]=50, E[USafe ]= 20, E[ULess Risky ]= 20 E[ULessRisky ] E[ULessRisky P_up ]f (P _ up )dP _ up EP_up (C3 C4 ) 0.5(C1 C2 ) 1 0 1 1 0 0 E[URisky ] E[URisky P_up ]f (P _ up )dP _ up P_up C1 (1 P_up) C2 f (P _ up ) dP _ up Quantitative Methods for Management EP_up (C1 C2 ) 0.5(C1 C2 ) Emanuele Borgonovo 137 Investment • c) Let us use Bayes’s theorem to update the prior uniform distribution – – – • evidence: up,down, down,down,down,up,down,up,down,up,down,up,up,up L(E|P_up): 15! L(E | P_up) (P _ up )7 (1 P _ up )8 7!8! Prior: 0 uniform bewteen 0 and 1 Bayes’theorem: L(E | P_up) 15! (P _ up )7 (1 P _ up )8 1 7!8! 1 15! 7 8 0 7!8! (P _ up ) (1 P _ up ) 1 dP _ up (P _ up )7 (1 P _ up )8 1 (P _ up ) (1 P _ up ) 7 8 dP _ up 0 3.5 3 2.5 • Posterior Distribution 2 p0 1.5 p1 1 0.5 0.98 0.91 0.84 0.7 0.77 0.63 0.56 0.49 0.42 0.35 0.28 0.21 0.14 0 • • 0.07 0 E[p_up]=0.47 d) Posterior Decision: E[URisky]=23, E[USafe ]= 20, E[ULess Risky ]= 9.2 Quantitative Methods for Management Emanuele Borgonovo 138 Problems • Apply the one way, two way and Tornado Diagrams SA to the IDs and DTs of the previous chapters: • Discuss your results Quantitative Methods for Management Emanuele Borgonovo 139 Bayesian Decision • • • • • You are the director of a library shop. To improve the sales, you are thinking of hiring additional sale personnel. This should, in your opinion, improve the service level in the shop. If this happens, you expect an increase in costumer number, and correspondingly, an increase in revenue sales. Suppose that the number of people entering the shop is, any day, distributed according to a Poisson distribution with uncertain. the prior distribution of is a gamma with mean equal to 55 and standard deviation equal to 15. the cost increase due to the hiring is 5000EUR for month. If the service quality improves and the library receives more than 50 customers per day, revenues increase would amount at 15000EUR (on the average). If less than 50 customers visit the shop, then revenues would not increase (and you loose the 5000EUR). What to you decide? You decide to monitor the number of customers on the next 6 working days: 75,45,30,80,72,41. You update the Probability. What do you decide now? How much do you expect to gain now? Perform a sensitivity analysis on the probabilities. What information do you get? Quantitative Methods for Management Emanuele Borgonovo 140 Influence Diagram Decisione Servizio Clienti Guadagno Clienti Service Decision Improves Pmigl Invest Quantitative Methods for Management 1-Pmigl Not Invest Clienti=0 Servizio=0 P_50_up Less than 50 1-P_50_up Does not improve P=0.1 Pmigl=0.5 P_500=0.5 P_500_down=0.5 P_500_up=0.7 More than 50 10000 -5000 -5000 0 Emanuele Borgonovo 141 Chapter VI Introduction to Decision theory Quantitative Methods for Management Emanuele Borgonovo 142 Summary • Preferences under Certainty – Indifference Curves – the Value Function [V(x)]: properties – Preferential independence • Preferences under Uncertainty – Axioms of rational choice – utility Function [U(x)] in one dimension – Risk Aversion • Preferences with Multiple Objectives – Multi-attribute utility Function Quantitative Methods for Management Emanuele Borgonovo 143 Preferences Under Certainty • Example: you are choosing your first job. You select your attributes as: location (measured in distance from home), starting salary and career perspectives. You denote the attributes as x1, x2, x3. you have to select among five offers a1, a2,…,a5. Every offer gives you certain values of x1, x2, x3 for certain. How do you decide? • It is a multi-attribute decision problem in the presence of certainty, since once you decide you will receive x1,x2,x3 for certain. • In this case you have to establish how much of one attribute to forego to receive more of anoilr attribute. Quantitative Methods for Management Emanuele Borgonovo 144 Preferences under Certainty Opzione Valore • Here is a diagram for the Choice Opction 1 2 3 X1=0.0 X2=0.0 X3=0.0 X4=0.0 X5=0.0 Quantitative Methods for Management 4 5 X1 X2 X3 X4 X5 Emanuele Borgonovo 145 Structuring Preferences • Indifference Curves: x2 x1 • Points on the same curve leave you indifference Quantitative Methods for Management Emanuele Borgonovo 146 the Value Function • You can associate a numerical value representing you preferences to each indifference curve: x2 x1 V( x) v( x1, x2,..., xn ) • V(x) is the function that says how much of xi one is willing to exchange for an increase or decrease in xk Quantitative Methods for Management Emanuele Borgonovo 147 V(x) • V(x) is a value function if it satisfies the following properties: • a) x' x' ' v( x' ) v( x' ' ) • b) Quantitative Methods for Management x' x' ' v( x' ) v( x' ' ) Emanuele Borgonovo 148 Example • For the “first job choice”, suppose that you value function is as follows: v( x) 3 / x 4x2 x3 2 1 2 • where x1 measures the distance from home in 100km, x2 is the career perspective measured on a scale from 0 a 10 and x3 the starting salary in kEUR. • Suppose to have received the following offers: – (1, 5, 20), (5, 4, 10), (8,3,60), (10, 5, 20), (10,2,40) • Which one would you pick? Quantitative Methods for Management Emanuele Borgonovo 149 Preferences under Uncertainty Opzione Utilità Evento Casuale Suppose one has to choose between lotteries that offer a mix the previous job offers: to choose one does not use the value function, but must resort to the utility function (U(x)) 1 2 P11 U1 P12 U2 P13 U3 P14 U4 P41 U1 P42 U2 P43 U3 P44 U4 Decision 3 4 Quantitative Methods for Management Emanuele Borgonovo 150 utility Function • the utility function is the appropriate one to express preferences over the distributions of the Attributes. • Given two distributions 1 and 2 on the Consequences , Distribution 1 is more or as much desirable than Distribution 2 if and only if: EU( x1) EU( x2) Quantitative Methods for Management Emanuele Borgonovo 151 Utility vs. Value – One attribute Problem. Suppose that alternative 1 produces x1 and the 2 x2, then 12 if x1>x2 – Let us take two Alternatives 1 and 2, with x1>x2, given with certainty. – the value function will give us: v(x1)>v(x2) – Let us now consider the following problem: 1 P1 2 1-P1 X1 X2 XI – To choose one need u(x1) and u(2) Quantitative Methods for Management Emanuele Borgonovo 152 Stochastic Dominance Distributions over attribute x 1 2 0.9 1 0.8 Probability distributions over x 0.7 Distribution 1 is dominated by distribution 2, if obtaining more of x is preferable. Vice versa, if less of x is preferable, then Distribution 2 is dominated by distribution 1 0.6 0.5 0.4 0.3 0.2 0.1 0 0 1 Quantitative Methods for Management 2 3 4 5 x 6 7 8 9 10 Emanuele Borgonovo 153 One Attribute Utility Functions Quantitative Methods for Management Emanuele Borgonovo 154 Certainty Equivalent • Given the lottery: 1 X1 P1 2 XN 1-P1 X3 • the value of x such that you are indifferent between x* for certain and playing the lottery. 1 X1 P1 2 X2 1-P1 X* • In equations: u( x*) E u( x) • N.B.: if you are risk neutral, then x*=E[x] Quantitative Methods for Management Emanuele Borgonovo 155 definition of Risk Aversion • a decision-maker is risk averse if preferisce sempre the expected value of a lottery alla lottery 1 2 : £10 2 £10 0.500 0.500 £40 £(£20) 20 £10; P = 1.000 • Hp: increasing utility function. Th: You are risk averse if the Certainty Equivalent of a lottery is always lower than the expected value of the lottery • You are risk averse if and only if your utility function utility is concave Quantitative Methods for Management Emanuele Borgonovo 156 Risk Premium and Insurance Premium • the Risk Premium (“RP”) of a lottry is the difference between the expected value of the lottery and your Certainty Equivalent for the lottery: RP Ex x * • Intuitively, the Risk Premium is the quantity of attribute you are willing to forego to avoid the risks connected with the lottery. • Suppose now that E[x]=0. the insurance premium is how much one would pay to avoid a lottery: IP : x * Quantitative Methods for Management Emanuele Borgonovo 157 Mailmatical Definition • the Risk Aversion function is defined as: r( x ) : u" ( x ) u' ( x ) d r( x ) ln( u' ( x )) dx • Or, equivalently: • Supposing a constant risk aversion one gets an exponential utility function: d ln( u' ( x )) ln( u' ( x )) x u' ( x ) e x dx u( x ) x 1 x u' ( t )dt e dt u( x ) u( x 0 ) (e e x 0 ) u( x 0 ) x0 t u( x ) a e x b Quantitative Methods for Management Emanuele Borgonovo 158 Risk Preferences • Constant Risk Aversion U 1 e r • Compute constant through Certainty Equivalent (CE): 0.5 e x1 Quantitative Methods for Management 0.5 e x1 / 2 e CE Emanuele Borgonovo 159 Investment Results with Risk Aversion Market Decision Market up 1-exp(-200/70) = 1 Blue Chip Stock 0.600 -3 Market Down 1-exp(-(-160)/70) = -9 0.400 Market up TwoStock 1-exp(-500/70) = 1 Risky Investment 0.600 prob_up=0.6 -2,110 Market Down 1-exp(-(-600/70)) = -5,278 0.400 Bond=1 1-exp(-50/70) = 1; P = 1.000 Quantitative Methods for Management Emanuele Borgonovo 160 A quale value accetterei l’investimento rischioso Sensitivity Analysis on prob_up 400.0 Blue Chip Stock 0.0 Risky Investment Expected Value -400.0 Bond -800.0 Threshold Values: -1,200.0 prob_up = 0.96 EV = 0.5 -1,600.0 prob_up = 1.00 EV = 0.9 -2,000.0 -2,400.0 -2,800.0 -3,200.0 0.40 0.52 0.64 0.76 0.88 1.00 prob_up Quantitative Methods for Management Emanuele Borgonovo 161 Esempi of funzioni utility • Linear: u=ax – Risk Properties: • Risk Neutral • Exponential: u( x) a ecx b – Risk Properties: • - sign: Constant Risk Aversion, + sign: Constant Risk Proneness • Logarithmic: – Risk Properties: u( x ) a ln( x ) b • Decreasing Risk Aversion Quantitative Methods for Management Emanuele Borgonovo 162 Problems Quantitative Methods for Management Emanuele Borgonovo 163 problem VI-1 • For the following three utility functions, u( x ) a x u( x ) a e x b u( x ) a ln( x ) b • compute: – the risk aversion function r(x) – the risk premium for 50/50 lotteries – the insurance premium Quantitative Methods for Management Emanuele Borgonovo 164 Problem VI-2 • Consider a 50/50 lottery. Determine your Risk Aversion constant, assuming an exponential utility function. • Reexamine some of the problems discussed till now utilizing instead of the monetary payoff the corresponding exponential utility function with the constant determined above. How do the decisions change? Quantitative Methods for Management Emanuele Borgonovo 165 Problem VI-3 • You are analyzing some alternatives for your next vacations: – A guided tour through Italian cultural cities (Rome, Florence, Venice, Siena …an infinite list..), duration 10 days, cost 500EUR, for a total of 1500km by bus. – A journey to the Caribbean, lasting 1 week, cost 2000EUR, by plane. – 15 days in a wonderful mountain in Trentino, for a cost of 2000EUR, with 500km of promenades. • Do you need a utility or a value function to decide? • Suppose that, after some thinking, you discover to have the following three attribute utility function: 1 2 x1 x ( x ) s( x ) 1 e a 2 3 b c • where x1 is the vacation cost in kEUR, x2 is distance in km and x3 is a merit coefficient regarding relax/amusement to be assigned between 1 and 10. • What do you choose? Quantitative Methods for Management Emanuele Borgonovo 166 Chapter VII the Logic of Failures Quantitative Methods for Management Emanuele Borgonovo 167 Elements of Reliability theory Quantitative Methods for Management Emanuele Borgonovo 168 Safety and Reliability • Safety and Reliability study the performance of systems. • Reliability and safety study cover two wide areas: – System Failures and Failure Modes • Structure Function – Failure Probability • Failure Data Analysis • the approach can be static or dynamic. Static approach is analytically simpler and is more diffuse. Quantitative Methods for Management Emanuele Borgonovo 169 Systems • A system is a set of components connected through some logical relations with respect to operation and failure of the system • More simple structures are: – Series – Parallel Quantitative Methods for Management Emanuele Borgonovo 170 Series 1 2 n • Every component is critical w.r.t. the system being able to perform its mission. • the fault of just one component is sufficient to provoke system failure • Redundancy: 0 Quantitative Methods for Management Emanuele Borgonovo 171 Parallel Systems 1 In 2 Out n • Each of the components is capable of assuring that the system accomplish its tasks. • Thus, to provoke the failure of the system, all the components must be contemporarily failed • Redundancy: n-1 Quantitative Methods for Management Emanuele Borgonovo 172 Elements of System Logics Quantitative Methods for Management Emanuele Borgonovo 173 Boolean Logic • An event (and) can be True or False • State Variable or Indicator: 1 Xj 0 if E has happened if E has not happened • Properties: – (XJ)n=Xj – X j X j 0 where X j is the complementary of XJ • This simple definition enables one to use algebraic operations to describe the logical behavior of systems. Quantitative Methods for Management Emanuele Borgonovo 174 Series Systems • • • • Let Ei denote the event “the i-th component failed.” Let XT denote the event: “the System failed”. XT takes the name of Top Event. For the system failure, by definition of series, it is enough that one single component failes. Thus it is enough that E1 or E2 or …. or En is true. • From a set point of view: E1E2 ... En E1E3E2 • From a logical p.o.v., we get the following expression: n XT 1 (1 X1 ) (1 X2 ) ... (1 Xn ) 1 (1 Xi ) i1 Quantitative Methods for Management Emanuele Borgonovo 175 Parallel Systems • Let Ei denote the event “the i-th component has failed.” • Let XT denote the event: “the system has failed”. • the condition for failure of the system is that all component fail. This happens if E1 and E2 and … En are true at the same time. • From a Set point of view: E1E2 ... En E1E3E2 • the logical expression is: n X T X1 X 2 ... Xn Xi i1 Quantitative Methods for Management Emanuele Borgonovo 176 the Structure Function • In general, a system will be formed by a combination of series or parallel elements, or other logics (as we will see next). • One defines the Structure Function of a system the logical expression that expresses the top event (XT) as function of the individual failure events. Quantitative Methods for Management Emanuele Borgonovo 177 the Logic of Performance • Let Ai denote the event “the i-th component is working (=Not failed).” • Let YT denote the event: “the system is working”. • For a series system: all the components must be working for the system to work. Thus: A1 , A2 , … and An must be true at the same time YT Series n Yi i 1 • In parallel: for the system to work it is sufficient that just one component is working. Thus: A1 or A2 or An must be true. YT Parallel n 1 (1 Yi ) i 1 Quantitative Methods for Management Emanuele Borgonovo 178 n/N Logics • n/N logics are intermediate logics between series and parallel. • N represents the total number of components in the system and n the number of components that must contemporarily fail to break the system. • As an example, a system has a 2/3 logic if it has 3 components and when two components have failed the system failes. 2/3 1 In 2 Out 3 Quantitative Methods for Management Emanuele Borgonovo 179 Example: 2/3 System Logics • Let us find XT for a 2/3 system • Events: E1, E2, E3, Indicators: X1, X2, X3 • Events that provoke a failure: E1E2 E3, E1 E2 , E1E3, E3 E2 . • Let us denote E1E2 E3=Z1, E1 E2 =M1, E1 E3= M2, E3 E2 = M3. • For XT to happen: Z1(M1 M2 M3). • the structure function expression is: XT 1 (1 Z1 ) (1 M1 ) (1 M2 ) (1 M3 ) • Let us go to a level below: XT 1 (1 X1 X2 X3 ) (1 X1 X2 ) (1 X1 X3 ) (1 X2 X3 ) • Let us solve the calculations, noting that (Xi)n=Xi : XT X1X2 X1X3 X2 X3 2X1X2 X3 Quantitative Methods for Management Emanuele Borgonovo 180 Probability Sum Rules • We recall that, for generic Events: n P( Ei ) P(Ei ) P(EiE j ) i1 j ij,i j P(E E E ) P(E E E E ) ijk ,i jk i j i ijkh,i jhk i j k h ... ( 1)n P(E1E2 ...En ) • We recall that, if the events are independent: n P(EiE j ,..., En ) P(E s ) s 1 • Rare Event Approssimation: neglect all terms corresponding to multiple events Quantitative Methods for Management Emanuele Borgonovo 181 Golden Rule • In practice: the System Failure Probability is computed from the solved Structure Function, substituting to indicator Xi the corresponding Event Probability. XT X1X2 X1X3 X2 X3 2X1X2 X3 P( XT ) P(E1E2 ) P(E1E3 ) P(E2E3 ) 2P(E1E2E3 ) Quantitative Methods for Management Emanuele Borgonovo 182 Proof • the System Failure Probability is: • P(XT)=P[(Z1(M1 M2 M3)]= P[(Z1 Z2]=P(Z1)+P(Z2)-P(Z1Z2) • where: – P(Z1)=P(E1E2 E3) – P(Z2)=P(M1 M2 M3)= P(M1)+ P(M2)+P(M3)-P(M1M2)- P(M1M3)P(M3 M2)+P(M1 M2 M3). Ma: M1= E1E2, M2= E3E1, M3= E3E2. Noting, that: M1 M2= M1 M3= M2 M3= M1 M2 M3 = E1E2E3. Substituting: P(Z2)=P(E1E2)+ P(E3E1)+P(E3E2)-P(E1E2E3)- P(E1E2E3)P(E1E2E3)+P(E1E2E3). Thus: • P(Z2)=P(E1E2)+ P(E3E1)+P(E3E2)-2P(E1E2E3) – P(Z1Z2)=P(E1E2E3 E1E2 E3E1 E3E2)=P(E1E2E3 ) • Thus: P(XT)= P(E1 E2 E3)+P(E1E2)+ P(E3E1)+P(E3E2)-2P(E1E2E3)-P(E1E2E3)= P(E1E2)+ P(E2E1)+P(E3E2)-2P(E1E2E3) Quantitative Methods for Management Emanuele Borgonovo 183 Problems Quantitative Methods for Management Emanuele Borgonovo 184 Problem VII-1 • For the following systems compute: – the Structure Function for System Failure – the Structure Function for System Operation – the Failure Probability – the Operation Probability 1 1 2/3 3 2 Quantitative Methods for Management 2 3 1 2 3 4 4 Emanuele Borgonovo 185 Problem VII-2 • for the following system: 1 2 3 4 • Compute the Failure Probability supposing independent events and denoting the component failure probability by p. • Repeat the computation starting with the system success function, YT. Verify that the two results coincide. Quantitative Methods for Management Emanuele Borgonovo 186 Chapter VIII Elements of Reliability Quantitative Methods for Management Emanuele Borgonovo 187 Cut and Path sets • • • • Failure Logic By cut set one means an event/set of events whose happening causes system failure By minimal cut set one means a cut set that does not have other cut sets as subsets Success Logic By path set one means an event/set of events whose happening causes system to work By minimal path set one means a path set that does not have other path sets as subsets Quantitative Methods for Management Emanuele Borgonovo 188 Even Trees • Event Trees: represent the sequence of events that lead to the event top. Initiating Event Event 1 Event 2 Top Event Sì No Sì No Quantitative Methods for Management Emanuele Borgonovo 189 Example • One has to establish the sequence of events that lead to leakage of toxic chemicals from a production plant. High pressure in one of the pipes can cause a breach in the pipe itself, with leakage of toxic material in the room where the machine works. the filtering of the air conditioning could prevent the passage of the toxic gas to the outside of the room. A fault on the air circulation system due to air filter fault or maintenance error, would lead to the diffusion of the gas to the entire firm building. At this point, public safety would be protected only by the building air circulation system, last barrier for the gas going to the outsides. • Draft the event tree for this sequence. Quantitative Methods for Management Emanuele Borgonovo 190 Gas Leakage Fault - Tree High Pressure Pipe Room Building Top Event Yes No No No Quantitative Methods for Management Emanuele Borgonovo 191 Fault Trees • Fault Trees: represent the logical connection among failures that lead to the failure of a barrier • they are characterized by a set of logic symbols that connect a series of events And Or event Base • Basic Event: is the event that represents the base of the fault-tree. From a physical point of view, it represents the failure of a component or of part of it. From a modeling point of view, it represents the lowest level of detail. Quantitative Methods for Management Emanuele Borgonovo 192 Example • Let us consider the failure of the aeration system. Suppose that the system is composed by two main parts: an suction engine and a static filter. the failure of the aeration system, thus, happens either due to engine failure or for filter fault. Aeration 1 Engine Quantitative Methods for Management Static Filter Emanuele Borgonovo 193 Example • We could however realize that the level of detail could be Further increased. In fact, we discover that the engine can brake for a failure of its mechanical components and, in particular, of the fan or for a fault of the electric feeder. the filter can break because of wrong installation after maintenance or for an intrinsic fault. • the fault tree becomes as follows: Quantitative Methods for Management Emanuele Borgonovo 194 Level II Aeration 1 Engine Elettr. Quantitative Methods for Management Mech. Static filter Fault Install. Emanuele Borgonovo 195 From Fault Trees to Structure Functions • Engine=A • Static filter =B • Electr.=1, Mech=2, Fault=3, Install.=4 XT 1 (1 A ) (1 B) 1 1 [1 (1 X1 ) (1 X 2 )] 1 [1 (1 X3 ) (1 X 4 )] 1 1 ( X1 X2 X1X2 )] [1 ( X3 X 4 X3 X 4 ) X1 X2 X3 X 4 X1X3 X1X 4 X1X2 X3 X 4 X2 X3 X 2 X 4 X1X3 X 4 X2 X3 X 4 X1X 2 X3 X1X2 X 4 X1X2 X3 X 4 • What are the minimal cut sets? Quantitative Methods for Management Emanuele Borgonovo 196 Rare Events Approximation • If the event probabilities are low (rare events), then lower the event intersection probabilities will be. • One neglects the probabilities of intersections. • the Failure Probability is computed as sum of the minimal cut sets Probabilities: P( XT ) P( X1 ) P( X2 ) P( X3 ) P( X4 ) Quantitative Methods for Management Emanuele Borgonovo 197 Event Trees & Fault Trees High pressure Pipe Aeration 1 Aeration 2 Top Event yes No No No Aeraz. 1 engine Electr. Quantitative Methods for Management filter Mech. Fault Installaz. Emanuele Borgonovo 198 Probability of the Top Event • From the Event Tree: P(Top ) P(Condutt . Aeraz.1 Aeraz.2 AP) • Expanding: P(Top ) P( Aeraz.1 Aeraz.2 Condutt .) * P(Condutt . AP) P( Aeraz.1Aeraz.2,Condutt ., AP) P( Aeraz.2 Cond., AP) P(Cond AP) • the conditional probabilities are found solving the corresponding Fault Trees Quantitative Methods for Management Emanuele Borgonovo 199 Definitions Quantitative Methods for Management Emanuele Borgonovo 200 Failure Density • Given a system, tet us denote with fs ( t )dt the Probability that the system fails between t and t+dt • It must hold that: f (t)dt 1 s 0 Quantitative Methods for Management Emanuele Borgonovo 201 Reliability • The Reliability of a system between 0 and t is the Probability that the system fulfills its function between 0 and t • The Unreliability of a system between 0 and t is the Probability that the system breaks withint time T: F( t ) Pr(T t ) f ( t)dt 0 • Thus the Reliability [R(t)] is related to the failure time pdf as follows: t R( t ) 1 Pr(T t ) 1 f ( t)dt • Note that, if f(t) is continuous: 0 R' ( t ) f ( t ) Quantitative Methods for Management Emanuele Borgonovo 202 General Fault rate (t) Infant Mortality Useful Life Aging t Quantitative Methods for Management Emanuele Borgonovo 203 Hazard/Failure Rate • Failure rate, (t), is the Probability that the system si rompa between t and t+dt, given that is sopravvissuto fino a t. • Dalla definition segue immediatamente the relaction with the Reliability and the function densità: ( t ) R( t )dt f ( t )dt • Thus: Quantitative Methods for Management f (t ) R' ( t ) ( t ) R( t ) R( t ) Emanuele Borgonovo 204 Legami between R(t), f(t) and (t) • From the above definition, there follows: f (t ) R' ( t ) ( t ) R( t ) R( t ) • Relationship R(t)- (t): R( t ) e t ( t ) dt 0 • Relationship f(t)- (t): t f ( t ) ( t ) R( t ) ( t ) e Quantitative Methods for Management ( t ) dt 0 Emanuele Borgonovo 205 time medio of failure (MTTF) • The mean time to failure is defined as: MTTF t f ( t )dt 0 Quantitative Methods for Management Emanuele Borgonovo 206 System Reliability Quantitative Methods for Management Emanuele Borgonovo 207 Reliability of systems in Series • Series: P(TS t ) R S (t ) 1 FS (t ) P(T1 T2 ... TN t ) • if independence is assumed: P(TS t ) P(T1 t ) P(T2 t ) P(TN t ) n • Thus: R s ( t ) Ri ( t ) i 1 • Faulure rate: n S (t) i (t) i 1 Quantitative Methods for Management Emanuele Borgonovo 208 Reliability of systems in Parallelo • Failure Probability of the system: P(TS t ) FS ( t ) 1 R S (t ) P(T1 T2 ... TN t ) • if independent: P(TS t ) P(T1 t ) P(T2 t ) P(TN t ) • Thus: n N Fs ( t ) Fi ( t ) RS ( t ) 1 Fs ( t ) 1 1 Ri ( t ) i1 • Failure Rate: i1 1 1 1 1 1 1 ... s ( t ) 1( t ) 2 ( t ) n ( t ) 1( t ) 2 ( t ) 1 ( t ) 3 ( t ) 1 1 .... 1( t ) 2 ( t ) 3 ( t ) 1 ( t ) 2 ( t ) 3 ( t ) ( 1)n Quantitative Methods for Management 1 1( t ) 2 ( t ) 3 ( t ) ... n ( t ) Emanuele Borgonovo 209 Reliability of Standby Systems • A standby system is a system where a subsystem is operational and the other subsystems become operational only after the failure of the system which is operating at the time of failure. • An example is the fifth wheel of a car. • In this case the System Reliability is given by: – 1) Two components: P(TS2 t) P(T1 t) P(T1 t T2 t T1) t – Thus: 2 RS ( t ) R1( t ) f1( t1 ) R2 ( t t1 )dt1 0 – where 2 indicates that there are two components in standby, while the subscript denotes the second component P(TS t ) P(T1 t ) P(T1 t T2 t T1 ) 3 – 2) Three components: – Thus: P(T1 t T2 t T1 T3 t T1 T2 ) t t t1 0 0 RS ( t ) RS ( t ) f1( t1 ) f2 ( t 2 ) R3 ( t t1 t 2 ) dt1 dt 2 3 Quantitative Methods for Management 2 Emanuele Borgonovo 210 Standby Systems with const. failure rates • For a standby system, it holds that: T T1 T2 ... TN • then P(t<T) is given by the convolution of the fi(t). • If these distributions are exponential and the failure rates identical: n1 R e n s Quantitative Methods for Management i 0 t ( t ) i! i Emanuele Borgonovo 211 Failure on Demand • If a system is called in function and does not respond (i.e. does not begin to work), one talks about a “failure on demand”. • For a standby system, one denotes with q the failure on demand probability : t RS ' ( t ) R1( t ) (1 q) f1( t1 ) R2 ( t t1 )dt1 2 0 • and t t t1 0 0 RS ( t ) RS ' ( t ) (1 q)2 f1( t1 ) f2 ( t 2 ) R3 ( t t1 t 2 ) dt1 dt 2 3 Quantitative Methods for Management 2 Emanuele Borgonovo 212 Problems Quantitative Methods for Management Emanuele Borgonovo 213 problem VIII-1 • Write the Fault Trees for the following systems and derive the structure function: 1 1 2/3 3 2 Quantitative Methods for Management 2 3 1 2 3 4 4 Emanuele Borgonovo 214 Problema VIII-2 • Una delle sequenze incidentali di un piccolo reattore di ricerca prevede la spaccatura della conduttura principale del circuito idraulico primario. Se la conduttura si rompe, si ha perdita immediata di raffreddamento del nocciolo - la zona del reattore dove avviene la reazione nucleare. L’incidente si può evitare se il sistema di raffreddamento ausiliario interviene per tempo e se il sistema di spegnimento del reattore interviene con successo. L’insuccesso dello spegnimento può avvenire se uno dei seguenti avvenimenti si realizza: mancata lettura del segnale per un guasto al software [P(Sof|alta press.)=10-4], mancato arrivo del segnale per un guasto del sistema elettrico [P(E|alta press.)= 10-5], mancato sganciamento delle barre per un guasto meccanico [P(Bar|alta press.)= 10-3]. Il sistema di raffreddamento ausiliario è costituito da due pompe in parallelo, con rateo di guasto 1/10000 [1/h] e probabilita’ di guasto on demand di 10-3. Le pompe devono funzionare per 100 ore affinche’ l’impianto sia fuori pericolo. Determinare: – L’albero degli eventi – Gli alberi dei guasti – La probabilità di fondere il reattore dato che si è verificato l’incidente in un anno dato che la frequenza di eventi di alta pressione e’ 0.0001 per anno. Quantitative Methods for Management Emanuele Borgonovo 215 Problema VIII-3 • • • • Un test di polizia per la determinazione del grado di alcool nei guidatori, ha probabilità 0.8 di essere corretto, cioè di dare risposta positiva quando il contenuto di alcool nel sangue è elevato o negativa quando il contenuto è basso. Coloro che risultano positivi al test, vengono sottoposti ad un esame da parte di un dottore. Il test del dottore non fa mai errori con un guidatore sobrio, ma ha un 10% di errore con guidatori ebbri. I due test si possono supporre indipendenti. 1) Determinare la frazione di guidatori che, fermati dalla polizia subiranno un secondo test che non rivela alto contenuto di alcool 2) Qual è la probabilità a posteriori che tale persona abbia un alto contenuto di alcool nel sangue? 3) Quale frazione di guidatori non avrà un secondo test? Quantitative Methods for Management Emanuele Borgonovo 216 Problema VIII-4 • • Un impianto elettrico ha due generatori (1 e 2). A causa di manutenzioni e occasionali guasti, le probabilità che in una settimana le unità 1 e 2 siano fuori serivizio (eventi che chiamiamo E1 ed E2 rispettivamente) sono 0.2 e 0.3 rispettivamente. C’è una probabiltà di 0.1 che il tempo sia molto caldo (Temperatura>30 gradi) durante l’estate (chiamiamo H questo evento). In tal caso, la domanda di elettricità potrebbe aumentare a causa del funzionamento dei condizionatori. La prestazione del sistema e la potenzialità di soddisfare la domanda può essere classificata come: – Soddisfacente (S): se tutte e due le unità sono funzionanti e la temperatura è inferiore a 30 gradi – Marginale (M) : se una delle due unità è funzionante e la temperatura è maggiore di 30 gradi – Insoddisfacente (U): se tutte e due le unità sono non funzionanti 1) Qual è la probabilità che esattamente una unità sia fuori servizio in una settimana? • 2) Definire gli eventi S, M e U in termini di H, E1 ed E2 • 3) Scrivere le probabilità: P(S), P(U), P(M) • Suggerimenti: Utilizzate alberi degli eventi e dei guasti per determinare la funzione di struttura e poi passate alle probabilità Quantitative Methods for Management Emanuele Borgonovo 217 Problema VIII-5: Distribuzione Weibull • Dato un componente con rateo di guasto: t ( t ) 1 • con e 0t calcolare: • R(t), f(t), il MTTF e la varianza del tempo medio di guasto • R(t) è detta distribuzione di Weibull • Disegnare (t),f(t) ed R(t) per =-1,1, 2. Dedurne che la Weibull può essere utilizzata per descrivere il tasso di guasto di componenti in tutta la vita del componente. Quantitative Methods for Management Emanuele Borgonovo 218 Problema VIII-6 • Dato un componente con il tasso di guasto (t) seguente: 10 9 8 7 6 l(t) 1 se 0 t 1 t 1 λ ( t ) 1 se 1 t 10 1 t 2 se t 10 100 5 4 3 2 1 0 0 5 • calcolare: • R(t), f(t), e il MTTF del componente 10 t Quantitative Methods for Management Emanuele Borgonovo 219 15 Problema VIII-7 • Calcolare l’espressione dell’affidabilità [R(t)] di un sistema k su n con rateo di guasto generico. • Calcolare la stessa espressione con distribuzioni esponenziali Quantitative Methods for Management Emanuele Borgonovo 220 Problema VIII-8 • Calcolare l’affidabilità annuale di un sistema con 4 componenti in serie con ratei di guasto [1/h]: (1/6000, 1/8000, 1/10000, 1/5000). • Confrontatela con quella di un sistema in cui i componenti sono messi in: – Parallelo – In logica 3/4 – In logica 2/4 Quantitative Methods for Management Emanuele Borgonovo 221 Problema VIII-9 • Due componenti identici, con tasso di guasto =3x10-7 [1/h] devono essere messi in parallelo o standby. Determinate la configurazione migliore e il guadagno in affidabilità (in percentuale). • Supponete ora che il sistema di switch sia difettoso, con probabilità q=0.01. Quale delle due configurazioni è più conveniente? Quantitative Methods for Management Emanuele Borgonovo 222 Problema VIII-10 • Considerate un sistema in standby di due componenti diversi, con densità di guasto esponenziali. Il MTTF del primo componente è 2 anni, quello del secondo è 3 anni. Calcolate: • La densità di guasto del sistema • Il MTTF • Cosa succede se i due componenti sono identici con MTTF di 2.5 anni? Quantitative Methods for Management Emanuele Borgonovo 223 Prob. VIII-2 Soluzione Rottura Primario Spegnimento Raffreddamento Top Event No Si’ Si’ Raffreddamento Spegnimento On Demand Sof. El. Bar. Pompa 1. Quantitative Methods for Management Emanuele Borgonovo 224 Pompa 2 Prob. VIII-2 Soluzione • Assumiamo eventi rari. • La frequenza si calcola dalla combinazione degli eventi: F frottura P(Raff Rottura Pr im) P(Spegn Rottura Pr im.) • dove: frottura=.000001 per anno • P(Spegn|rottura prim.)=P(Sof|rottura prim.)+P(E|rottura prim.)+P(Bar|rottura prim.)=0.00111 2 1 ( 100 ) • P(Raff| rottura prim.)= q 1 e 10000 • Quindi: =.0010995 F 1/ 1000000 .00109 .0011 2.2 109 Quantitative Methods for Management Emanuele Borgonovo 225 Problema VIII-8 Soluzione • Calcolare l’affidabilità annuale di un sistema con 4 componenti in serie con ratei di guasto [1/h]: (1/6000, 1/8000, 1/10000, 1/5000). – R( t ) e t e t e t e t – Ore in un anno: 8760. 8760 8760 8760 8760 – Sostituendo i numeri: R(t ) e 6000 e 8000 e 10000 e 5000 0.006 1 2 3 4 • Confrontatela con quella di un sistema in cui i componenti sono messi in: 8760 6000 8760 8000 ) (1 e ) (1 e – Parallelo: R( t ) 1 (1 e – ¾ supponendo I ratei di guasto =1/8000. 8760 10000 ) (1 e 8760 5000 ) 0.75 • Risultato: 0.11 – 2/4 supponendo I ratei di guasto =1/8000 • Risultato: 0.41 Quantitative Methods for Management Emanuele Borgonovo 226 Problema VIII-9 Soluzione • Due componenti identici, con tasso di guasto =3x10-7 [1/h] devono essere messi in parallelo o standby. Determinate la configurazione migliore e il guadagno in affidabilità (in percentuale) per t=7 anni (61320hs). Rp (t ) 1 (1 e 61320 30000000 ) (1 e 61320 30000000 ) 0.99967 t RS ( t ) R1( t ) f1( t1 ) R2 ( t t1 )dt1 e t (1 t ) 0.99983 2 0 – Il guadagno di affidabilita’ e’ dell’ordine del 10^-2% (0.0002), quindi trascurabile • Supponete ora che il sistema di switch sia difettoso, con probabilità q=0.01. Quale delle due configurazioni è più conveniente? Rp (t ) 1 (1 e 61320 30000000 ) (1 e 61320 30000000 ) 0.99967 t RS ( t ) R1( t ) (1 q) f1( t1 ) R2 ( t t1 )dt1 et (1 (1 q)t ) 0.99965 2 0 Quantitative Methods for Management Emanuele Borgonovo 227 Capitolo IX Decisioni Operative: Ottimizzazione delle Manutenzioni Quantitative Methods for Management Emanuele Borgonovo 228 Decisioni Operative • Decisioni di Affidabilita’ o Reliability Design • Decisioni di Optimal Replacement • Decisioni di ispezione ottimale • Decisioni di riparazione ottimale Quantitative Methods for Management Emanuele Borgonovo 229 Indisponibilita’ • Sistemi riparabili o manutenibili: il sistema puo’ ritornare a funzionare dopo la rottura • Indisponibilta’ istantanea: – q(t):= P(sistema indisponibile per T=t) • Indisponibilita’ limite: lim q( t ) t T • Indisponiblita’ media in T: q q(t )dt 0 T T • Indisponibilta’ media limite: qlim lim T Quantitative Methods for Management q(t )dt 0 T Emanuele Borgonovo 230 Disponibilita’ • La disponibilita’ istantanea e’ il complementare della indisponibilita’. Le altre definizioni seguono immediatamente • Note: la disponibilita’/indisponibilita’ non e’ una densita’ di probabilita’ e l’indisponibilita’ media non e’ una probabilita’. • Interpretazione: la disponibilita’/indisponibilita’ media e’ la frazione media di tempo in cui il sistema e’ disponibile in [0 T]. • Le riparazioni/manutenzioni introducono periodicita’ nel problema Quantitative Methods for Management Emanuele Borgonovo 231 Effetto delle manutenzioni Quantitative Methods for Management Emanuele Borgonovo 232 Calcolo della Indisponibilita’: un unico componente, una sola modalita’ di guasto • Evoluzione temporale: t tr t tr t • A t=0 il sistema entra in funzione dopo la manutenzione. Dopo un tempo t= t torna di nuovo in manutenzione. La manutenzione dura tr. t e’ il tempo in cui il componente e’ soggetto a rotture casuali con (t). • Si nota che il problema e’ periodico, di periodo T= tr+t • Durante t il sistema ha una indisponibilita’ istantanea pari alla sua probabilita’ di rottura, se, come da ipotesi, non ci sono riparazioni: ( t ')dt ' t q( t ) FS ( t ) 1 R( t ) 1 e 0 Quantitative Methods for Management Emanuele Borgonovo 233 Calcolo della Indisponibilita’: un unico componente, una sola modalita’ di guasto ( t ')dt ' 0 0tt • L’indisponibilta’ istantanea risulta quindi: q( t ) 1 e 1 t t t tr t t t • Da cui l’ind. Media: q ( t ')dt ' 0 1 e 0 t tr dt tr t tr • Supponiamo cost e 1. Quindi utilizziamo approssimaz. Taylor: t ( t ')dt ' 1 e 0 1 e t 1 (1 t ...) t • Sostituiamo nella ind. Media, e assumiamo tr<< t : q Quantitative Methods for Management 1 t T r 2 T Emanuele Borgonovo 234 Modi di Guasto • Guasto in funzionamento: f(t) [1/T] • Guasto in hot standby: s(t) [1/T] • Guasto a seguito di manutenzione errata: 0, 1, 2…. Dove: – 0=incondizionale, – 1=dato che 1 manutenzione errata, – 2= dato che 2 manutenzioni errate • Guasto on demand: Q0,Q1 etc. Quantitative Methods for Management Emanuele Borgonovo 235 Indisponibilita’ istantanea con piu’ modi di guasto • Consideriamo per un componente i modi di guasto indicati in precedenza. • A t=0 il componente puo’ essere gia’ guasto se disabilitato dall’erronea manutenzione. Questo evento ha probabilita’ 0. Con probabilita’ (1- 0) il componente invece potra’ invece aver superato con successo la manutenzione. In questo caso il componente potra’ rompersi “on demand” (E1) o con tasso di guasto (t) (E2). Si ha: P(E1E2)=P(E1)+P(E2)-P(E1E2)=Q0+F(t)-Q0F(t). • Riassumendo, tra 0 e t si ha: q(t)= 0+(1- 0)*[Q0+F(t)-Q0F(t)]. • Introduciamo ora una probabilita’ esponenziale per le rotture. Utilizziamo la approssimazione di Taylor. Abbiamo: q(t)= 0+(1- 0)*[Q0+(1-Q0)t]. • Quindi l’indisponibilita’ istantanea e’: 0 (1 0 )Q0 t Q0t 0 t t q( t ) t t t tr 1 Quantitative Methods for Management Emanuele Borgonovo 236 Indisponibilita’ media con piu’ modi di guasto • L’indisponibilita’ media sull’intervallo 0 t+ tr e’: t2 t2 0 t (1 0 )Q0 t Q0 tr 2 2 q t tr • Due assunzioni: 1) Eventi rari 2) t+ tr t t tr q 0 Q0 2 t Quantitative Methods for Management Emanuele Borgonovo 237 Rappresentazione equivalente Componente t Q0 0 t • La funzione struttura e’: • XC=1- (1-Xt) (1-XQ0)(1-X0)(1-Xt)= • = Xt +XQ0+X0+Xt-termini di ordine superiore…. • Approssimazione eventi rari: • XC= Xt +XQ0+X0+Xt Quantitative Methods for Management Emanuele Borgonovo 238 Il caso di due componenti • Sostituzioni successive t tr1tr2 t tr1 tr2 • Sostituzioni distanziate tr t t+2t trtr r t t+2tr r r • Periodo: t+2tr Indisponibilita’ media e’ la somma di piu’ termini: q qR qC qD qM “R”: random, “C” common cause, “D” demand e “M” maintenance Quantitative Methods for Management Emanuele Borgonovo 239 Modi di guasto • Causa comune: sono quei guasti che colpiscono il sistema come uno e rendono inutili le ridondanze e/o annullano indipendenza condizionale dei guasti. • Es.: difetto di fabbrica in parallelo di componenti identici • Errori in manutenzione: human errors • Human Reliability • CC e HR sono due importanti rami dello studio dell’affidabilita’ dei sistemi Quantitative Methods for Management Emanuele Borgonovo 240 Modelli decisionali corrispondenti • Come stabilire una politica di replacement ottimale? • Costruzione della funzione obiettivo – i) Individuazione del Criterio – ii)Costruzione della funzione obiettivo o utilita’ – iii) Ottimizzazione Quantitative Methods for Management Emanuele Borgonovo 241 Esempio 1 • 1 componente soggetto replacement periodico e manutenzione periodica • Criterio = disponibilita’ media • Funzione obiettivo: q(t) • t ottimale: t 2 tr – tr=24 h, =1/10000 (1/h) tott=700hr • Con =1/100000 (1/h) tott=2200hr Quantitative Methods for Management Emanuele Borgonovo 242 Esempio 2 • Ottimizzazione in considerazione del costo di sostituzione e della disponibilita’ • Funzione obiettivo: B( t) Cq q( t) C( t) • t ottimale: dB( t) dt 0 2 d B( t) 0 d2 t • Occorre introdurre vita del’impianto L. L • Si ha: C( t) c 0 dove c0 e’ il costo unitario di t riparazione Quantitative Methods for Management Emanuele Borgonovo 243 Esempio 2 • Introduciamo poi il costo della indisponibilita’: Cq a c0 • definito come multiplo del costo singola riparazione. • Funzione energia: E( t) Cq q( t 0 Q0 tr ) C( t) 1 tr L E( t) a c0 c0 2 2 2 dt t t d 1 • Intervallo ottimale: Quantitative Methods for Management tr L 1 t 2 a 2 Emanuele Borgonovo 244 Esempio 2 c0 5 34000 3.51210 L 70000 a 10 tr 24 1 10000 E( t ) 2000 59.363 0 0 0 5 10 t 4 1 10 100000 5 0 0.001 Q0 0.001 Quantitative Methods for Management 4 t 1.185 10 Emanuele Borgonovo 245 Applicazione del modello • Il modello si applica al meglio a componenti in standby o sistemi di sicurezza passivi. • Infatti si ipotizza che il componente sia rimpiazzato secondo un intervallo di tempo prestabilito t. • Si valuta percio’ la convenienza rispetto alla minimizzazione del costo di replacement e/o alla massimizzazione della disponibilita’ • Per sistemi in funzionamento occorre considerare invece la possibilita’ di riparare il sistema Quantitative Methods for Management Emanuele Borgonovo 246 Riparazioni Quantitative Methods for Management Emanuele Borgonovo 247 Il tasso di riparazione (t) • Analogamente alla rottura, anche il processo di riparazione di un componente ha delle caratteristiche di casualita’. Per esempio, non si sa il tempo necessario alla individuazione del guasto, cosi’ come puo’ essere non noto a priori il tempo necessario all’arrivo delle parti di ricambio o il tempo richiesto dall’esecuzione della riparazione. Tutto cio’ viene condensato in una quantita’ analoga al rateo di guasto, e, precisamente, il tasso di riparazione (t) . E’ uso comune assumere un tasso di riparazione costante - e spesso questa assunzione non e’ peggiore di quella di assumere (t) costante.- Ne seguono: rip( t ) e t t Rip( t ) 1 e MTTR t e t 1 0 • Dove rip(t) e’ la densita’ di riparazione, ovvero la probabilita’ che la riparazione avvenga tra t e t+dt e Rip(t) e’ la probabilita’ che la riparazione avvenga entro t. Notiamo che (t) e’ la probabilita’ che il componente sia riparato tra t e t+dt dato che non e’ stato ancora riparato a t. Quantitative Methods for Management Emanuele Borgonovo 248 Esempio • Consideriamo un sistema composto da due componenti, di cui uno in standby. • Per modellizzare questo problema occorre un approccio diverso sia dai due casi precedenti. • Occorre introdurre gli stati del sistema • Nell’esempio. Il sistema puo’ essere: in funzione con il componente 1 funzionante (stato 1), in funzione ma con il componente 2 funzionante e il componente 1 in riparazione (stato 2), (stato 3) con entrambe i componenti rotti. Da 3 puo’ tornare a 2 e da 2 ad 1. Puo’ passare da 1 a 3 se c’e’ failure on demand. 1 Quantitative Methods for Management 2 3 Emanuele Borgonovo 249 Assunzioni • Stato del sistema al tempo t e’ indipendente dalla storia del sistema. • Questa assunzione e’ alla base dei processi stocastici di Markov. • In particolare, supponiamo che il sistema possa avere M stati e denotiamo con Xt lo stato del sistema al tempo t. Allora Xt potra’ assumere valori 1,2,….M. • Cosa accade in dt? • Il sistema puo’ transitare in un altro stato (eventualmente con dei vincoli): i j Quantitative Methods for Management Emanuele Borgonovo 250 Matrice di transizione • Indichiamo con Pij la probabilita’ che il sistema passi dallo stato i allo stato j P11 P12 P M Pij 21 PM1 P1M PMM • Proprieta’: M • 1) Pij 1 i1 • 2)Se Pii 1 Quantitative Methods for Management allora stato i e’ detto assorbente Emanuele Borgonovo 251 Esempio • Applichiamo uno schema a stati per il sistema in standby. Otteniamo: P13 1 P12 P21 Quantitative Methods for Management 2 P23 3 P32 Emanuele Borgonovo 252 Equazioni di Markov/Kolmogorov dP A P dt • Dove A e’ la matrice di transizione del sistema, P e’ il vettore delle probabilita’ degli stati del sistema. Quantitative Methods for Management Emanuele Borgonovo 253 Costruzione della matrice di transizione P12 1 2 P21 1 2 • Esempio: componente soggetto a rottura e riparazione. 2 stati: in funzione o in riparazione, con tassi di guasto e riparazione. • Chi sono P12 e P21? Sono le probabilita’ di transizione in dt. Quindi: P12= e P21= • La matrice di transizione e’ costruita con le seguenti regole: • (+) se il salto e’ in entrata allo stato, (-) se il salto e’ in uscita • Prendiamo lo stato 1: si entra in 1 da due con tasso (+), si esce con tasso (-). dP1 • Quindi: P1 P2 dt Quantitative Methods for Management Emanuele Borgonovo 254 La matrice di transizione • Analogamente: 1 2 • Quindi: 1 2 dP2 P1 P2 dt A • La matrice di transizione e’: • Il sistema di equazioni differenziali dP A P diventa: dt dP1 P1 P2 dt dP2 P1 P2 dt Quantitative Methods for Management Emanuele Borgonovo 255 Disponibilita’ asintotica e media • E’ la probabilita’ che a t il componente sia nello stato 1. Occorre risolvere il sistema di equazioni differenziali lineari precedente. Modo piu’ usato in affidabilita’ e’ mediante trasformata di Laplace. • Con trasf. Laplace, le equazioni da differenziali diventano algebriche. Dopo aver lavorato con equazioni algebriche, occorre poi antitrasformare. • Si ottiene dunque la disponibilita’ come funzione del tempo. A questo punto due disponibilita’ interessano: quella asintotica e quella media. Il risultato per un componente singolo soggetto a riparazioni e rotture e’ il seguente: Quantitative Methods for Management Emanuele Borgonovo 256 Risultati per un componente • Disponibilita’ istantanea: μ λ P1 ( t ) = + e μ+λ μ+λ ( μ+ λ ) t • Disponibilita’ asintotica: μ MTTF lim P1 ( t ) = = t →∞ μ + λ MTTF + MTTR • Interpretazione: tempo che occorre in media alla riparazione diviso il tempo totale • Disponibilita’ media su T: q( T) Quantitative Methods for Management ( ) 2 expT ( ) ( ) 2 Emanuele Borgonovo 257 Problema IX-1 • Calcolare l’ indisponibilita’ media di un componente in standby soggetto a sostituzione periodica con le seguenti probabilita’ di guasto per t=5000: 0 0.002 Q0 0.002 tr 24 1 15000 • (Soluzione: q=.175) • Calcolare l’intervallo di sostituzione ottimale e l’indisponibilita’corrispondente, con L=70000, a=10 e a=. • (Soluzione: t=14500, q=0.5; t=849, q=0.06) Quantitative Methods for Management Emanuele Borgonovo 258