OPPORTUNISTIC MAINTENANCE FOR MULTI-COMPONENT SHOCK MODELS

Size: px
Start display at page:

Download "OPPORTUNISTIC MAINTENANCE FOR MULTI-COMPONENT SHOCK MODELS"

Transcription

1 !#"%$ & '("*) +,!.-#/ ;:=< S TELUJVFXWYJZTEJZR[W6\]BED S H_^`FILbadc6BEe?fBEJgWPJVF hmijbkrdl S BkmnWP^oN prqtsouwvgxyczpq {v~} {qƒ zgv prq Šs Ž Žw š œ š Ÿž Qž žoª«o ««ž «± ²t «³ µ t ª«ª š «žº¹ª» žo Ÿ ¼ ¾½ ³ U š À À šà ³ÁÝÂÄà ÃÆÅ ÇoÈ ÉËÊÆÅ Å«Ê ÌQ Á ²tžš QÍ Î ÃoÉËÊ Ê ÍoÉËÊ Ã Ï Ç ÐYÑo Ò Í Î ÃœÉÓÊ Ê ÍoÉ±Å Å Ô ÔwÐPÕÖ ª«ƒ š «`Ø Ù t `Ø ž Æ Ðd¹ŠÚIÛ Ö Ó ` Ü Ü«Ù ÙÙ(Ø ² ÝØ ÙŠ t `Ø ž«üš ( «

2 OPPORTUNISTIC MAINTENANCE FOR MULTI-COMPONENT SHOCK MODELS Lirong Cui 1 School of Management & Economics Beijing Institute of Technology Beijing , P.R. China lirongcui@bit.edu.cn Haijun Li 2 Department of Mathematics Washington State University Pullman, WA 99164, U.S.A. lih@math.wsu.edu July Supported by the NSF of China grant Supported in part by the NSF grant DMI

3 Abstract This paper is concerned with opportunistic maintenance on a multi-component cumulative damage shock model with stochastically dependent components. A component fails when its cumulative damage exceeds a given threshold, and any such a failure creates a maintenance opportunity, and triggers a simultaneous repair on all the components, including the non-failed ones, such that damages accumulated at various components are reduced to certain degrees. Utilizing the coupling method, stochastic maintenance comparisons on failure occurrences under different model parameters are obtained. Some positive dependence properties of this multi-component shock model are also presented. Key words and phrases: Opportunistic maintenance, multi-component cumulative damage shock model, stochastic comparison, positive dependence, coupling method.

4 1 Introduction A system of components working in a random environment are subjected to wear and damage over time and may fail unexpectedly. The components are replaced or repaired upon failure, and such unpleasant events of failure are at the same time also considered in practice as opportunities for preventive maintenance on other components. The goal of this paper is to introduce a general opportunistic maintenance policy for a multi-component cumulative damage shock model and study its structural properties using stochastic comparison methods. Opportunistic maintenance basically refers to the scheme in which preventive maintenance is carried out at opportunities, either by choice or based on the physical condition of the system. In this paper, we focus on the situation in which the opportunities for preventive maintenance are generated by the failure epochs of individual components. At each failure epoch, the failed components are correctively repaired and other components that are still operational are also preventively serviced so that all the components are maintained and restored to certain conditions. An advantage of this opportunistic maintenance is that corrective repair combined with preventive repair can be used to save set-up costs. Note that by combining both types of repair, one may not know in advance which repair actions should be taken, and thus sacrifices the plannable feature of preventive maintenance. However, there are many situations in which opportunistic maintenance is effective. For example, when corrective repair on some components requires dismantling of the entire system, a corrective repair on these components combined with preventive repair on other or neighboring components might be worthwhile. Another instance is when a certain corrective repair on failed components can be delayed until the next scheduled preventive maintenance. Opportunistic maintenance has been first studied in Radner and Jorgenson 1963, and in McCall Since then, many extensions of opportunistic maintenance have been introduced and studied in the literature. Berg (1976) studies a system with two identical components with exponential distributed lifetimes, for which the non-failed component as well as the failed component are both replaced by a new one if the age of the non-failed component exceeds a threshold. Zheng and Fard (1991) examine an opportunistic maintenance policy based on failure rate tolerance for a system with k different types of components. Pham and Wang (2000) propose two new (τ, T ) opportunistic maintenance policies for a k-out-of-n system. These and other opportunistic maintenance models have been summarized in Dekker, van der Schouten and Wildeman (1997) and in Wang (2002). All these models, however, address the optimization issues for components operating independently. The thrust of the maintenance model introduced in this paper is the general opportunistic 1

5 repairs implemented at correlated failures for a system of components that are stochastically dependent. To compare the effectness of various policies, the maintenance comparisons on the failure counts under basic age, block and other policies have been studied in the literature. It is wellknown (see Barlow and Proschan 1981) that the number of failure of components with NBU (New Better Than Used) lifetimes under age or block replacement policy is stochastically smaller than the number of component failures without any preventive maintenance. Block, Langberg and Savits (1990a) extend the maintenance comparison to general block policies, and obtain stronger results that compare the processes of failure counts under different policies. Li and Shaked (2003) examine various maintenance comparisons of failure counts for imperfect repair models with preventive maintenance. Shaked and Shanthikumar (1989) introduce several maintenance policies for multivariate repairable systems operating in random environments, and study multivariate stochastic comparisons on the counters of failure at various components for the models with minimal repairs at failure and block replacements at predetermined time instants. Li and Xu (2004b) develop a multivariate imperfect repair model with coordinated random group replacements, and establish dependence comparisons on the failure counts at various components. The basic idea for maintenance comparisons of repairable systems is that if component lifetimes possess certain positive aging properties, then it should be beneficial to implement a maintenance policy so that the failure occurrences can be reduced. The details on maintenance comparisons and related techniques can be found in the survey papers by Block, Langberg and Savits (1990b) and by Kijima, Li and Shaked (2000) and the references therein. In this paper, we introduce a shock model for a system of components subject to correlated damages of shocks that arrive at the system according to a point process. A component fails when its cumulative damage exceeds a given threshold, and any such a failure creates a maintenance opportunity, and triggers a simultaneous repair on all the components, including the failed ones, such that damages accumulated at various components are reduced to a certain degree. This cumulative damage shock model ables us to develop a coupling method, using which, we study the structural properties of opportunistic maintenance, and compare stochastically the failure counts for this model under different processes of shock arrivals and repair degrees. This research is motivated from the analysis on opportunistic maintenance of a system of components operating in a random environment, where component lifetimes are dependent, and repair degrees at various components are also correlated. These failure and economic correlations, in addition to stochastic dependence introduced by the opportunistic mainte- 2

6 nance scheme, complicate the dependence structure of failures at various components. A failure simultaneously occurred at some components creates an opportunity of preventive repair for other components. As such, if some components fail more frequently, then the other components could receive more preventive repairs, and thus may be less likely to fail. Despite this negative dependence, our results show that the number of failure occurrences among all the components is positively correlated over time if the damages and repair degrees at various components are positively correlated in the sense of association. Our results also show that a correlated shock model with opportunistic maintenance experiences less failures stochastically at various components than a similar shock model without such a maintenance. The paper is organized as follows. Section 2 introduces our model and its performance measures of interest in this study. Section 3 establishes the main result that compares stochastically the processes of failure instants under different model parameters, and also show that opportunistic maintenance reduces the failure occurrences at various components in our shock model. Finally, Section 4 concludes the paper with a temporal association property for the system failure epochs. Throughout this paper, the term increasing and decreasing mean non-decreasing and non-increasing respectively, and the measurability of sets and functions as well as the existence of expectations are often assumed without explicit mention. Any inequality between two vectors with finite or infinite dimensions means the inequalities component-wise. A product space of partially ordered sets is equipped with the component-wise partial ordering. 2 A Multi-component Shock Model with Opportunistic Maintenance In this section, we introduce a cumulative damage shock model for a system of components with a general opportunistic repair policy, and also discuss the difference of our model with the other repair models in the literature. Consider a system of m components that are subjected to random shocks, whose occurrences are governed by a non-explosive point process τ = {τ n, n 1} that has no multiplicities. The counting process of the shocks is given by A = {A(t), t 0}, where A(t) = max{n : τ n t}. Any arriving shock simultaneously inflicts random damages on all the components in the system. Suppose that for each 1 j m, the damage of the nth shock to component j is 3

7 X n,j, and then X n = (X n,1,..., X n,m ), n 1, is the vector of correlated random damages of the nth shock brought to various components in the system. Suppose that X n, n = 1, 2,..., are non-negative bounded random vectors, that is, there exist positive real numbers x j, 1 j m, such that for 1 j m, n 1, 0 X n,j x j, almost surely. (2.1) In other words, we focus on the situations such as normal wear and fatigue, and exclude huge damages resulted from disasters. Damages accumulate additively at various components, and component j, 1 j m, fails when its cumulative damage up to time t exceeds for the first time its design threshold, say d j > 0. Note that several components may fail at the same time. Suppose that the first failure occurs at components j J 1 {1,..., m}; that is, all the components j J 1 fail simultaneously. Thus the first failure time is given by where T 1,J1 = inf{t > 0 : D (1) j (t) > d j, j J 1 }, (2.2) D (1) A(t) j (t) = X n,j for t 0. n=1 At time T 1,J1, the failed components are correctively repaired, and the other components are preventively repaired, so that the damage levels of all components are reduced. Specifically, let Y 1 = (Y 1,1,..., Y 1,m ) be the vector of repair degrees at various components, with 0 < y 1,j Y 1,j 1, where 0 < y 1,j 1 is a real number, 1 j m. At T 1,J1, the combination of corrective and preventive maintenance reduces the cumulative damage level of component j to (1 Y 1,j )D (1) j (T 1,J1 ), 1 j m. All the repair times are negligible. After time epoch T 1,J1, the damage accumulation at component j, 1 j m, continues. The second failure occurs at components j J 2 {1,..., m}, and the failure time is modeled by where T 2,J2 = inf{t T 1,J1 : D (2) j (t) > d j, j J 2 }, (2.3) D (2) j (t) = (1 Y 1,j )D (1) j (T 1,J1 ) + A(t) n=a(t 1,J1 )+1 X n,j for t T 1,J1. In general, at the kth failure time T k,jk, the failed and operational components are all repaired instantaneously, and the cumulative damage level of component j is reduced to (1 Y k,j )D (k) j (T k,jk ), 1 j m. After that, the damage accumulations at various components continue. The epoch of the (k + 1)th failure occurred at components j J k+1 {1,..., m} is then defined by T (k+1),jk+1 = inf{t T k,jk : D (k+1) j (t) > d j, j J k+1 }, n 0, (2.4) 4

8 where D (k+1) j (t) = (1 Y k,j )D (k) j (T k,jk ) + A(t) n=a(t k,jk )+1 X n,j for t T k,jk. (2.5) Here and in the sequel, i n=i X n,j is understood as zero if i < i. Thus, prior to the repair at T k,jk, the damage accumulated at component j is D (k) j (T k,jk ), and after the repair at T k,jk, the damage at component j reduces to D (k+1) j (T k,jk ). After T k,jk, the damage accumulation and failure instances continue. Let Y = {Y n, n 1} be a stochastic process with Y n = (Y n,1,..., Y n,m ), with 0 < y n,j Y n,j 1, 1 j m, (2.6) where 0 < y n,j 1, n 1, 1 j s, are real numbers. Here Y n,j represents the degree of the random repair on component j at the nth failure epoch. We assume that A, X and Y are mutually independent. We also assume that all the repairs on failed components are successful; that is, we assume that x j d j + x j y n,j, 1 j m, n 1. (2.7) Clearly, (2.6) and (2.7) imply that after the repair at any failure epoch T n,jn, 1 j m, D (n+1) j (T n,jn ) = (1 Y n,j )D (n) j (T n,jn ) (1 y n,j )(d j + x j ) d j, almost surely. Therefore, a corrective repair on any failed component always restores that failed component back to working condition. In other words, we do not allow that failed components are left unattended for a period of time. Note that x j d j +x j is increasing in x j, and becomes very small when x j is substantially smaller than d j. Thus, this assumption (2.7) is especially applicable in the situations such as wear and fatigue. Let N(t, A, X, Y) denote the number of failure epochs for the system up to time t, namely, N(t, A, X, Y) = max{n : T n,jn t}. (2.8) Also let N j (t, A, X, Y) denote the number of failures occurred at component j up to time t, namely, N j (t, A, X, Y) = max{n : T j n t}, 1 j m, (2.9) where T j n is the nth failure time of component j, 1 j m. Note that every failure occurrence may consist of a set of simultaneous failures, and as such, m N(t, A, X, Y) N j (t, A, X, Y). j=1 5

9 The performance measures of interest for our model with the shock arrival process A = {A(t), t 0}, the shock damage process X = {X n, n 1} and repair degree process Y = {Y n, n 1} are following two counting processes. N(A, X, Y) = {N(t, A, X, Y), t 0}, and, (2.10) N(A, X, Y) = {(N 1 (t, A, X, Y),..., N m (t, A, X, Y)), t 0}. (2.11) Li and Xu (2001) discuss a cumulative damage shock model with no maintenance and obtain various dependence comparison results on the first failure times of the system components. Kijima (1991) study a univariate cumulative damage shock model with block preventive maintenance, where each planned maintenance reduces the damage level by 100(1 b)%, 0 b 1. Li and Xu (2004a) introduce a condition-based group maintenance policy for multi-component cumulative damage shock models and extend Kijima s model to the multivariate case. In contrast, our cumulative damage shock model introduced here combines corrective and preventive repairs, and implements opportunistic repairs on all the components at failure. It is also worth mentioning that Çinlar and Özekici (1987) introduce a replacement model for multi-component devices operating in random environments, in which component j, 1 j m, fails when its intrinsic age (random cumulative hazard) runs over its intrinsic lifelength. Shaked and Shanthikumar (1989) study several general multi-component replacement systems (some with preventive block replacements) operating in random environments. They assumed that the state of the component is unobservable and obtained structural properties using correlated random cumulative hazard processes of the components. However, we here assume that the damage level of each component is observable and preventive maintenance actions on the components are dependent upon the failure of other components. 3 Stochastic Comparison of Shock Models with Opportunistic Maintenance In this section, we use the coupling technique to obtain some a strong comparison result for shock models with opportunistic maintenance, which will be used later to analyze its temporal dependence properties. We also show that the failure counts at various components for the model with opportunistic repairs are stochastically less than that for the model without any opportunistic repair. 6

10 Let E be a partially ordered Polish space (that is, a complete, separable metric space) with a closed partial ordering. An E-valued random variable X is said to be stochastically larger than another E-valued random variable Y (denoted by X st Y ) if Ef(X) Ef(Y ) for all real-valued increasing functionals f on E. It can be shown that X st Y if and only if one can construct two random variables X and Y on the same probability space such that X and X have the same distribution, and Y and Y have the same distribution, and X Y, almost surely. (3.1) We refer the reader to Shaked and Shanthikumar (1994) for details on this and other properties on stochastic orders. Let D([0, )) be the Polish space of all right-continuous functions that have left-hand limits, equipped with the following partial order, f g if f(t) g(t), for all 0 t <, where f, g D([0, )). The space [D([0, ))] m is also Polish with the component-wise partial order. By a counting process we mean a stochastic process whose sample paths are non-negative, right-continuous step functions, starting at 0 and only increasing by jumps of size 1. Definition 3.1 Let N = {(N 1 (t),..., N m (t)), t 0} and N = {(N 1(t),..., N m(t)), t 0} be two multivariate counting processes. N is said to be stochastically larger than N, denoted by N st N, if Ef(N) Ef(N ) for all real-valued increasing functionals f: [D([0, ))] m R. Note that the ordering in Definition 3.1 can also be characterized by means of stochastic ordering of finite dimensional distributions of N and N. It follows from (3.1) that N st N if and only if one can construct M = {(M 1 (t),..., M m (t)), t 0}, M = {(M 1(t),..., M m(t)), t 0}, on the same probability space such that M and N have the same distribution, and M and N have the same distribution, and (M 1 (t),..., M m (t)) (M 1(t),..., M m(t)), for all t 0, almost surely. (3.2) For a detailed discussion on stochastic comparisons of counting processes, we refer the reader to Kijima, Li and Shaked (2000) and the references therein. Our first result states that increasing repair capacity and decreasing shock damage reduce the number of system failure occurrences. 7

11 Theorem 3.2 Let N(A, X, Y) and N(A, X, Ȳ) be two counting processes of failure instants for two shock models with the same shock arrival process A, but different shock damage processes X = {X n, n 1} and X = { X n, n 1}, and different repair degree processes Y = {Y n, n 1} and Ȳ = {Ȳ n, n 1}, respectively. If X st X, and Y st Ȳ then N(A, X, Y) st N(A, X, Ȳ). Proof. Since X st X, we have X st X. Also because Y st Ȳ, and X and Y are independent, and X and Ȳ are independent, we obtain that ( X, Y) st ( X, Ȳ). Because of (3.1), there exist two stochastic processes ( X, Y ) = {( X n, Y n), n 1} and ( X, Ȳ ) = {( X n, Ȳ n), n 1}, on the same probability space, such that ( X, Y ) and ( X, Y) have the same distribution, and ( X, Ȳ ) and ( X, Ȳ) have the same distribution, and ( X n, Y n) ( X n, Ȳ n), almost surely, for all n 1. That is, X n X n, Y n Ȳ n, almost surely, for all n 1. We can also construct a common shock arrival process A on the same probability space. Let S ( S) denote the shock model with shock arrival process A, random damage process X ( X ), and random repair degree process Y (Ȳ ). Let T n,jn and T n, Jn be, respectively, the nth failure epoch in S and S, at which several components may fail simultaneously. Let N(A, X, Y ) = {N(t, A, X, Y ), t 0} and N(A, X, Ȳ ) = {N(t, A, X, Ȳ ), t 0}, where N(t, A, X, Y ) = max{n : T n,jn t} and N(t, A, X, Ȳ ) = max{n : T n, Jn t}. (3.3) Following our construction, it is evident that N(A, X, Y) and N(A, X, Y ) have the same distribution and N(A, X, Ȳ) and N(A, X, Ȳ ) have the same distribution. N(A, X, Y ) and N(A, X, Ȳ ) are constructed in the same probability space. follows from (3.2) that it suffices to show that In addition, Then, it N(t, A, X, Y ) N(t, A, X, Ȳ ), for all t 0, almost surely, (3.4) or equivalently, by (3.3), we need to show for each n, T n,jn T n, Jn, almost surely. (3.5) Let D (k) (k) j (t) ( D j (t)) be the cumulative damage on the jth component of S ( S) at time t such that T k 1,Jk 1 t T k,jk ( T k 1, Jk 1 t T k, Jk ), 1 j m, k 1. We now establish, 8

12 via induction, (3.5) and the following: If T k 1, Jk 1 < T k,jk, for some k n, then, prior to the repair at T k,jk in S, D (k) j (T k,jk ) D (k) j (T k,jk ), for any 1 j m. (3.6) Initial Step: Since S receives more damage than S at any (common) shock arrival epoch, S experiences an earlier failure. Thus, T 1,J1 T 1, J1, where J 1, J 1 {1,..., m}. Note that T 0, J0 = 0 < T 1,J1, but we obviously have D (1) (1) j (T 1,J1 ) D j (T 1,J1 ) for any 1 j m, prior to the repair in S at the first failure time epoch. Inductive Step: Suppose that (3.5) and (3.6) hold for n 1 We need to prove them for n+1. There are two cases. 1. If T n+1,jn+1 T n, Jn, then clearly, T n+1,jn+1 T n+1, Jn Suppose now that T n+1,jn+1 > T n, Jn. From the inductive hypothesis (3.6), there exists a largest index 0 k n, such that, prior to the repair in S at T k,jk > T k 1, Jk 1, D (k) j (T k,jk ) D (k) j (T k,jk ), for any 1 j m. (3.7) From T k,jk to T n, Jn, S and S experience the same number (n k + 1) of failure occurrences (Figure 1). Arrange all these failure instants in an increasing order, T k,jk = S 1 S 2... S 2(n k+1) 1 S 2(n k+1) = T n, Jn. Let Z l ( Z l ) be the amount of damage S ( S) received from the environment over time interval (S l, S l+1 ], l = 1,..., 2(n k + 1) 1. Because of the coupling construction, S receives more damage than S at any (common) shock arrival epoch, Z l Z l, l = 1,..., 2(n k + 1) 1. Each damage Z l ( Z l ), reduced somewhat at repairs after S l, is added to the cumulative damage at T n, Jn. We now argue that after the repair in S at T n, Jn, the cumulative damage at any component in S is smaller than that in S. For this, we consider the following cases for S l < S l+1. T k,jk T n+1,jn+1 S S T k 1, Jk 1 Tn, Jn Tn+1, Jn+1 Figure 1: Case 2: Time points of interest 9

13 (a) Suppose that S l = T p,jp, k p n. Let q = min{i : T i, Ji T p,jp }. Since T k,jk > T k 1, Jk 1, we must have k q. Then, from the inductive hypothesis (3.5), q p. Thus, the contribution of Zl toward the cumulative damage on component j in S at T n, Jn, after the repair at T n, Jn, is given by n (1 Ȳ i,j) Z l, 1 j m. i=q While the contribution of Z l toward the cumulative damage in S at T n, Jn by n i=p+1 (1 Y i,j)z l, 1 j m. Since Y i,j Ȳ i,j for 1 j m and i 1, and Z l Z l, we have is given n (1 Ȳ i,j) l n (1 Y i,j)z l, 1 j m. i=q i=p+1 Thus, the contribution of Z l toward the cumulative damage on component j in S at T n, Jn, after the repair at T n, Jn, is smaller than the contribution of Z l toward the cumulative damage on component j in S at T n, Jn. (b) Suppose that S l = T q, Jq, k q n. Let p = min{i : T i,ji > T q, Jq }. Then, from the inductive hypothesis (3.5), q + 1 p. Thus, the contribution of Z l toward the cumulative damage in S at T n, Jn, after the repair at T n, Jn, is given by n i=q+1 (1 Ȳ i,j) Z l, 1 j m. While the contribution of Z l toward the cumulative damage in S at T n, Jn by n (1 Y i,j)z l, 1 j m. i=p Since Y i,j Ȳ i,j for 1 j m and i 1, and Z l Z l, we have n i=q+1 (1 Ȳ i,j) Z l n (1 Y i,j)z l, 1 j m. i=p is given 10

14 In either case, for any l = 1,..., 2(n k + 1) 1, the contribution of Zl toward the cumulative damage in S at T n, Jn, after the repair at T n, Jn, is smaller than the contribution of Z l toward the cumulative damage in S at T n, Jn. In addition, because of (3.7) and Y i,j Ȳ i,j for 1 j m and i 1, we have n (1 Ȳ i,j) i=k (k) D j (T k,jk ) n (1 Y i,j)d (k) j (T k,jk ), 1 j m. i=k Thus the contribution of D(k) j (T k,jk ) at T k,jk toward the cumulative damage on component j in S at T n, Jn, after the repair at T n, Jn, is smaller than the contribution of D (k) j (T k,jk ) at T k,jk toward the cumulative damage on component j in S at T n, Jn. Therefore, after the repair at T n, Jn, D (n+1) j ( T n, Jn ) D (n+1) j ( T n, Jn ), 1 j m. Since S receives more damage than S at any (common) shock arrival epoch, then for any t such that T n, Jn t min{t n+1,jn+1, T n+1, Jn+1 }, D(n+1) j (t) D (n+1) j (t), 1 j m. Thus, T n+1,jn+1 = inf{t T n,jn : D (n+1) j (t) > d j, j J n+1 } inf{t T n, Jn : In addition, D (n+1) j (T n+1,jn+1 ) The induction concludes the proof. D (n+1) j (t) > d j, j J n+1 } = T n+1, Jn+1. D (n+1) j (T n+1,jn+1 ), for any 1 j m. Theorem 3.2 can be used to obtain some bounds for the failure count N(t, A, X, Y), as the following example shows. Example 3.3 Consider a cumulative damage shock model S, in which, the shocks arrive at the system according to a Poisson process A with rate λ. The shock damage process X = {X n, n 1} consists of independent and identically distributed (i.i.d.) random vectors. The repair degree process Y = {Y n, n 1} satisfies the assumption (2.7). To find a lower bound for E(N(t, A, X, Y)), we introduce another cumulative damage shock model S, in which the shock arrival process is Poisson process A, the shock damage process is X, but the repair degree process Ȳ = {Ȳn, n 1} is given by Ȳ n,j = 1, for all n 1, 1 j m. That is, such a shock model replaces all the components upon any component failure. Since Y st Ȳ, then, by Theorem 3.2, we have for any t, N(t, A, X, Y) st N(t, A, X, Ȳ). 11

15 Note that N(A, X, Ȳ) = {N(t, A, X, Ȳ), t 0} is a renewal process with inter-failure times that have the same distribution as the following first passage time T = inf{t 0 : A(t) n=1 X n,j > d j, for some j}. (3.8) Let W(t) = A(t) n=1 X n, t 0. Since A(t) is a Poisson process, and X n s are i.i.d., then W = {W(t), t 0} is a Markov process starting at 0, with the following properties. 1. The sample paths of W are almost surely increasing. 2. The discrete-time Markov chain { n i=1 X i, n 1} has a stochastically monotone transition kernel, that is, for any increasing function f : R m R, is increasing in x. n n 1 E[f( X i ) X i = x] = Ef(x + X n ), i=1 i=1 It follows from Shaked and Shanthikumar (1987) that T defined in (3.8) has an IFRA property; that is, its average cumulative hazard rate is increasing. This IFRA property and a result from Barlow and Proschan (1981, page 171) imply that t E(T ) Thus, we obtain a lower bound, E(N(t, A, X, Ȳ)) t E(N(t, A, X, Y)) t E(T ) 1. E(T ) 1. Last and Szekli (1998) study a repairable system with general imperfect repairs, and obtain the comparison results for this model with IFR (Increasing Failure Rate) component lifetimes. However, under the Markovian assumption as in Example 3.3, the time to failure in our opportunistic maintenance model is only IFRA, which is weaker than IFR. Thus, the results in Last and Szekli (1998) can not be applied to the case in Theorem 3.2. In general, it is not true that increasing repair capacities at various components, as in Theorem 3.2, would reduce the vector of failure counts in the stochastic sense. However, since the component lifetimes in a cumulative damage shock model possess a natural multivariate positive aging property, it should be beneficial to implement opportunistic maintenance for multi-component cumulative damage shock models. In fact, we are able to show, in some cases, that the vector of failure counts in a system without opportunistic maintenance is 12

16 stochastically larger than the vector of failure counts in the system equipped with opportunistic maintenance. Consider a shock model in which Y n,j = 1 for all 1 j m, and n 1. For simplicity, let N j (t, A, X) denote the number of failure occurred at component j up to time t in this opportunistic replacement system S with shock arrival process A and shock damage process X. Let N j (t, A, X), 1 j m, denote the number of failure occurred at component j up to time t in a system S with the same shock arrival process A and same shock damage process X, but no opportunistic replacement. Note that this latter shock model S only replaces failed components with new ones upon failure. Theorem 3.4 Let N(A, X) = {(N 1 (t, A, X),..., N m (t, A, X)), t 0}, and N(A, X) = {( N 1 (t, A, X),..., N m (t, A, X)), t 0}. Then N(A, X) st N(A, X). Proof. We construct a common shock arrival process A and a common shock damage process X in the same probability space. Without loss of generality, let S ( S) denote the above-mentioned shock model with shock arrival process A, random damage process X, and opportunistic replacement (no opportunistic replacement). Let Tn j and T n j be, respectively, the nth failure epoch of component j in S and S, 1 j m. Let M j (t) = max{n : Tn j t} and M j (t) = max{n : T n j t}. (3.9) Thus, processes M = {(M 1 (t),..., M m (t)), t 0} and M = {( M 1 (t),..., M m (t)), t 0} are constructed in the same probability space. Following our construction, it is evident that N(A, X) and N(A, X) have the same distributions as M and M respectively. Then, it follows from (3.2) that it suffices to show that (M 1 (t),..., M m (t)) ( M 1 (t),..., M m (t)), for all t 0, almost surely, or equivalently, by (3.9), we need to show for each n, 1 j m, Tn j T n, j almost surely. (3.10) 13

17 We prove (3.10) via induction. Let D j (t) ( D j (t)) be the cumulative damage on the jth component of S ( S) at time t 0. Note that after the replacement of component j at any failure, the damage level reduces to zero. Initial Step: Suppose that all the components j J {1,..., m} fail simultaneously first. From our construction, we have T j 1 = T j 1, for all j J. We now show that T1 i T 1, i for i {1,..., m} J. At the first failure epoch, S replaces all the components with new ones, whereas S performs no opportunistic replacement for the operational components. Thus, after the replacement at the first failure epoch, D i (T j 1 ) = 0 D i ( T j 1 ) for any j J, and i {1,..., m} J. Since the both systems receive the same amount of damage at any shock epoch from the environment, we have D i (t) D i (t) for any 0 t min{t1, i T 1}, i i {1,..., m} J. Thus, T1 i = inf{t : D i (t) > d i } inf{t : D i (t) > d i } = T 1, i for i {1,..., m} J. Inductive Step: Suppose that (3.10) holds for n, and we now argue that it is also true for n + 1. For any 1 j m, if T j n+1 Tn j then T j n+1 Tn+1. j If T j n+1 > Tn, j then we have, after the replacement at Tn, j D j (Tn) j = 0 D j (Tn). j Since the both systems receive the same amount of damage at any shock epoch from the environment, we have D j (t) D j (t) for any Tn j t min{tn+1, j T n+1}. j Thus, T j n+1 = inf{t > Tn j : D j (t) > d j } inf{t > Tn j : D j (t) > d j } = T n+1. j for 1 j m. The induction concludes the proof. 4 Positive Dependence Properties of System Failures As we mentioned before, stochastic dependence introduced by opportunistic maintenance complicates the dependence nature of failure counts at various components. We are able to show in this section that, under certain conditions, the process of system failure occurrences is positively associated over time. We also present a positive dependence property of the failure counts at various components. 14

18 Let E be a partially ordered Polish space with a closed partial ordering. An E-valued random variable X is said to be positively associated if for any two functions f, g that are both increasing (or both decreasing) with respect to, E[f(X)g(X)] [Ef(X)][Eg(X)] or Cov(f(X), g(x)) 0. (4.1) Association of a random vector (X 1,..., X m ) implies the positively orthant dependence of (X 1,..., X m ), that is, m m E( f j (X j )) Ef j (X j ), (4.2) j=1 j=1 for any non-negative functions f 1,..., f m that are all increasing or all decreasing. This positively orthant dependence implies that for any x 1,..., x m, m P (X 1 > x 1,..., X m > x m ) P (X i > x i ), i=1 m P (X 1 x 1,..., X m x m ) P (X i x i ). i=1 The association was first introduced by Esary, Proschan and Walkup (1967), and by Fortuin, Kastelyn and Ginibre (1971), and its general theory was developed by Lindqvist (1988). The association enjoys many desirable properties and has been applied to various areas in probability and statistics. The following basic properties, taken from these papers, can be easily verified. Theorem 4.1 Let E 1 and E 2 be partially ordered Polish spaces. Let X be an E 1 -valued random variable and Y be an E 2 -valued random variable. 1. If X is associated, (Y X = x) is associated for all x, and E[f(Y ) X = x] is increasing (or decreasing) in x for all increasing functional f, then Y is associated. 2. If X is associated and f : E 1 E 2 is increasing (or decreasing), then f(x) is associated. 3. If X is associated, Y is associated, and X and Y are independent, then (X, Y ) is also associated. 4. Let Z = {Z n, n 1} is a discrete-time process of E 1 -valued random variables. Z is associated if and only if Z is associated in time, that is, for any n, (Z 1,..., Z n ) is associated. 15

19 It follows from Theorem 4.1 that in order to obtain the association property, we first need to establish some monotonicity properties of the shock model with opportunistic maintenance. Theorem 3.2 describes the monotonicity properties of our opportunistic maintenance model with respect to the shock damage and repair degree, and our next result gives the monotonicity property with respect to the shock arrival process A = {A(t), t 0}. Let τ = {τ n, n 1} denote the sequence of the shock arrival times, that is, A(t) = max{n : τ n t}. (4.3) A sequence {a n, n 1} of real numbers is said to be increasing if a n a n+1, for all n 1. Lemma 4.2 Let {a n } and {ā n } be two increasing sequences of non-negative real numbers. Consider N({a n }, x, y) = [N(A, X, Y) τ n = a n, X n = x n, Y n = y n, n 1], N({a n }, x, y) = [N(A, X, Y) τ n = a n, X n = x n, Y n = y n, n 1], where x = {x n, n 1}, y = {y n, n 1} are two sequences of non-negative real vectors. If a n ā n for all n 1, then 1. N({a n }, x, y) N({ā n }, x, y). 2. N({a n }, x, y) N({ā n }, x, y). Proof. We only prove (1), and (2) can be similarly proved. Let S ( S) denote the shock model with deterministic shock arrival process {a n } ({ā n }), damage process x, and repair degree process y. Let T n,jn and T n, Jn be, respectively, the nth failure epoch in S and S, at which several components may fail simultaneously. Let N(t, {a n }, x, y) = max{n : T n,jn t} and N(t, {ā n }, x, y) = max{n : T n, Jn t}. (4.4) Note that T n,jn and T n, Jn are all deterministic and any failure in both systems can only occur at the shock arrival epochs. Suppose that the nth failure in S occurs at a kn, for some k n, we have, T n,jn = a kn, n 1. Since S receives the same amount of shock damage at a n as that S receives at ā n, and S implements the opportunistic repair of same degree at a n as that S implements at ā n, the nth failure in S must occur at ā kn. Thus for any n 1, T n,jn = a kn ā kn = T n, Jn. This and (4.4) imply that N(t, {a n }, x, y) N(t, {ā n }, x, y) for any t 0, 16

20 and thus (1) holds. Lemma 4.2 immediately implies the following theorem, which describes the impact of the shock arrival process A on the failure occurrences of our opportunistic maintenance model. Theorem 4.3 Let N(A, X, Y) and N(Ā, X, Y) be two counting processes of failure instants for two shock models with the same shock damage process X, and same repair degree process Y, but different shock arrival processes A = {A(t), t 0} and Ā = {Ā(t), t 0}, respectively. If A st Ā, then 1. N(A, X, Y) st N(Ā, X, Y). 2. N(A, X, Y) st N(Ā, X, Y). Proof. Again, we only prove (1), and (2) can be similarly proved using Lemma 4.2 (2). Let τ = {τ n, n 1} ( τ = { τ n, n 1}) denote the sequence of the shock arrival times of A (Ā). If A st Ā, then τ st τ. It follows from Lemma 4.2 (1) that N({a n }, x, y) is decreasing in {a n } with respect to the component-wise ordering, then we have, N(A, x, y) st N(Ā, x, y). Then (1) follows from unconditioning on X and Y. Consider in the remainder of this paper a shock model with opportunistic maintenance, in which the damage process X = {X n, n 1} is a sequence of independent and identically distributed random vectors, the repair degree process Y = {Y n, n 1} is also a sequence of independent and identically distributed random vectors, and shock arrival process A is a renewal process. Theorem 4.4 If both X n and Y n are associated, then N(A, X, Y) is associated in time; that is, for any 0 t 1 t 2... t n <, (N(t 1, A, X, Y),..., N(t n, A, X, Y)) is associated. Proof. Let τ = {τ n, n 1} be the sequence of shock arrival times of A. Let M(τ, X, Y) = (N(t 1, A, X, Y),..., N(t n, A, X, Y)), and let M({a n }, x, y) denote M(τ, X, Y) given that τ = {a n }, X = x, Y = y. Association of M(τ, X, Y) follows from three steps. 17

21 1. Since A is a renewal process, then τ n = n k=1 Z k where {Z n, n 1} is a sequence of independent and identically distributed random variables. From Theorem 4.1 (2) and (3), we obtain that (τ 1,..., τ n ) is associated for any n 1. It then follows from Theorem 4.1 (4) that τ is associated. From Lemma 4.2, M({a n }, x, y) is decreasing in {a n } with respect to the component-wise order. It follows from Theorem 4.1 (2) that M(τ, x, y) is associated. 2. Since X n is associated for all n, then it follows from Theorem 4.1 (3) and (4) that X is associated. From Theorem 3.2, M(τ, x, y) is stochastically increasing in x. Thus, by (1) we just proved above and Theorem 4.1 (1), we have M(τ, X, y) is associated. 3. Since Y n is associated for all n, then it follows from Theorem 4.1 (3) and (4) that Y is associated. From Theorem 3.2, M(τ, X, y) is stochastically decreasing in y. Thus, by (2) we just proved above and Theorem 4.1 (1), we have M(τ, X, Y) is associated. Using Lemma 4.2 (2), we can also establish a positive dependence property of failure counts at various components. Theorem 4.5 If both X n and Y n are vectors of independent random variables, then the vector of failure counts at various components satisfies the positive orthant dependence property, that is, for any t 0, m m E( f j (N j (t, A, X, Y))) Ef j (N j (t, A, X, Y)) (4.5) j=1 j=1 for any non-negative functions f 1,..., f m that are all increasing or all decreasing. Proof. Let τ = {τ n, n 1} be the sequence of shock arrival times of A. Let (N 1 (t, {a n }, x, y),..., N m (t, {a n }, x, y)) = [(N 1 (t, A, X, Y),..., N m (t, A, X, Y)) τ n = a n, X n = x n, Y n = y n, n 1]. It follows from Lemma 4.2 that (N 1 (t, {a n }, x, y),..., N m (t, {a n }, x, y)) is decreasing in {a n } with respect to the component-wise ordering. From the proof of Theorem 4.4, the renewal process τ is associated, and thus, by Theorem 4.1 (2), (N 1 (t, A, x, y),..., N m (t, A, x, y)) is associated, which implies that m m E( f j (N j (t, A, x, y))) Ef j (N j (t, A, x, y)) (4.6) j=1 j=1 18

22 for any non-negative functions f 1,..., f m that are all increasing or all decreasing. Let X n = (X n,1,..., X n,m ) and Y n = (Y n,1,..., Y n,m ) for n 1. Also let, for 1 j m, Z j = {X n,j, Y n,j, n 1}. Note that X n s are i.i.d, and Y n s are also i.i.d.. Since, for each n, both X n and Y n are vectors of independent random variables, we have Z 1,..., Z m are mutually independent. Let P Zj denote the probability measure induced by Z j, 1 j m, it then follows from (4.6) that m E( f j (N j (t, A, X, Y))) j=1 m j=1 [ Ef j (N j (t, A, x, y))p Zj (dz j )], where z j is a realization of Z j, 1 j m. Note that N j (t, A, X, Y) only depends A and Z j, 1 j m, we obtain that for 1 j m, Ef j (N j (t, A, x, y))p Zj (dz j ) = Ef j (N j (t, A, X, Y)), and thus (4.5) holds. In general, Theorem 3.2 can not be extended to the vector of failure counts at various components, and thus it is not clear whether (4.5) still holds if both X n and Y n are associated vectors. 5 Conclusions In this paper, we introduce a new multi-component cumulative damage shock model to study opportunistic maintenance for a system of stochastically dependent components. The components natural positive aging properties and the coupling method able us to show that if the shock damages are stochastically smaller, and the repair degrees are stochastically larger, then the number of failure occurrences is stochastically smaller. We also show that a shock model with opportunistic maintenance experiences less failures stochastically at various components than a similar shock model without such a maintenance. This finding is consistent with the general understanding that preventive maintenance should be beneficial if component lifetimes possess certain positive aging property. Furthermore, we show that the number of failure occurrences is positively correlated over time if the damages and repair degrees at various components are positively correlated in the sense of association. We also establish a spatial dependence property for the failure counts of various components. The results we obtained in this paper shed new light on understanding the complex structural behavior of opportunistic maintenance for the components operating in a common random environment. 19

23 References [1] Barlow, R. E. and F. Proschan (1981). Statistical Theory of Reliability and Life Testing, To Begin With, Silver Spring, MD. [2] Berg, M. (1976). Optimal replacement policies for two unit machines with increasing running costs. Stochastic Processes and their Applications [3] Block, H. W., Langberg, N., and Savits, T. H. (1990a). Maintenance comparisons: Block policies. Journal of Applied Probability 27, [4] Block, H. W., N. Langberg and T. H. Savits (1990b). Stochastic comparisons of maintenance policies. In Topics in Statistical Dependence (Eds. H. W. Block, A. R. Sampson, and T. H. Savits), IMS Lecture Notes-Monograph Series [5] Çinlar, E. and S. Özekici (1987). Reliability of complex devices in random environments. Probability in the Engineering and Informational Sciences 1, [6] Dekker, R., R. Wildeman and F. A. Van der Duyn Schouten (1997). A review of multicomponent maintenance models with economic dependence. Mathematical Methods of Operations Research [7] Esary, J. D. F., F. Proschan and D. W. Walkup (1967). Association of random variables, with applications. Ann. Math. Statist [8] Kijima, M. (1991). A cumulative damage shock model with imperfect preventive maintenance. Naval Research Logistics [9] Kijima, M., H. Li and M. Shaked (2000). Stochastic processes in reliability, Handbook of Statistics, Vol. 19, D. N. Shanbhag and C. R. Rao, eds [10] Fortuin, C.M., P.W. Kastelyn and J. Ginibre (1971). Correlation inequalities on some partially ordered sets. Commu. Math. Phys [11] Last, G. and Szekli, R. (1998). Stochastic comparison of repairable systems by coupling. Journal of Applied Probability 35, [12] Li, H. and M. Shaked (2003). Imperfect repair models with preventive maintenance. Journal of Applied Probability [13] Li, H. and S. H. Xu (2001). Stochastic bounds and dependence properties of survival times in a multi-component shock model. Journal of Multivariate Analysis

24 [14] Li, H. and S. H. Xu (2004a). A multivariate cumulative damage shock model with preventive maintenance. Technical Report, Washington State University. [15] Li, H. and S. H. Xu (2004b). On the coordinated random group replacement policy in multivariate repairable systems. Operations Research [16] Lindqvist, H. (1988). Association of probability measures on partially ordered spaces. Journal of Multivariate Analysis 26, [17] McCall, J. J. (1963). Operating characteristics of opportunistic replacement and inspection policies. Management Science, [18] Pham, H. and H. Wang (2000). Optimal (τ, T ) opportunistic maintenance of a k-outof-n: G system with imperfect PM and partial failure. Naval Research Logistics [19] Radner, R. and D. W. Jorgenson (1963). Opportunistic replacement of a single part in the presence of several monitored parts. Management Science, [20] Shaked, M. and J. G. Shanthikumar (1987). IFRA properties of some Markov processes with general state space. Mathematics of Operations Research [21] Shaked, M. and J. G. Shanthikumar (1989). Some replacement policies in a random environment. Prob. in the Eng. and Infor. Sciences [22] Shaked, M. and J. G. Shanthikumar (1994). Stochastic Orders and Their Applications, Academic Press. [23] Wang, H. (2002). A survey of maintenance policies of deteriorating systems. European Journal of Operational Research, [24] Zheng, X. and N. Fard (1991). A maintenance policy for repairable systems based on opportunistic failure rate tolerance. IEEE Transactions on Reliability

Coherent Systems of Components with Multivariate Phase Type Life Distributions

Coherent Systems of Components with Multivariate Phase Type Life Distributions !#"%$ & ' ")( * +!-,#. /10 24353768:9 ;=A@CBD@CEGF4HJI?HKFL@CM H < N OPc_dHe@ F]IfR@ ZgWhNe@ iqwjhkf]bjwlkyaf]w

More information

Markov Repairable Systems with History-Dependent Up and Down States

Markov Repairable Systems with History-Dependent Up and Down States Markov Repairable Systems with History-Dependent Up and Down States Lirong Cui School of Management & Economics Beijing Institute of Technology Beijing 0008, P.R. China lirongcui@bit.edu.cn Haijun Li Department

More information

MULTIVARIATE COMPOUND POINT PROCESSES WITH DRIFTS

MULTIVARIATE COMPOUND POINT PROCESSES WITH DRIFTS MULTIVARIATE COMPOUND POINT PROCESSES WITH DRIFTS By HUAJUN ZHOU A dissertation submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY WASHINGTON STATE UNIVERSITY Department

More information

On the Convolution Order with Reliability Applications

On the Convolution Order with Reliability Applications Applied Mathematical Sciences, Vol. 3, 2009, no. 16, 767-778 On the Convolution Order with Reliability Applications A. Alzaid and M. Kayid King Saud University, College of Science Dept. of Statistics and

More information

Monotonicity and Aging Properties of Random Sums

Monotonicity and Aging Properties of Random Sums Monotonicity and Aging Properties of Random Sums Jun Cai and Gordon E. Willmot Department of Statistics and Actuarial Science University of Waterloo Waterloo, Ontario Canada N2L 3G1 E-mail: jcai@uwaterloo.ca,

More information

MULTIVARIATE DISCRETE PHASE-TYPE DISTRIBUTIONS

MULTIVARIATE DISCRETE PHASE-TYPE DISTRIBUTIONS MULTIVARIATE DISCRETE PHASE-TYPE DISTRIBUTIONS By MATTHEW GOFF A dissertation submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY WASHINGTON STATE UNIVERSITY Department

More information

Tail Dependence of Multivariate Pareto Distributions

Tail Dependence of Multivariate Pareto Distributions !#"%$ & ' ") * +!-,#. /10 243537698:6 ;=@?A BCDBFEHGIBJEHKLB MONQP RS?UTV=XW>YZ=eda gihjlknmcoqprj stmfovuxw yy z {} ~ ƒ }ˆŠ ~Œ~Ž f ˆ ` š œžÿ~ ~Ÿ œ } ƒ œ ˆŠ~ œ

More information

On Some Properties of the Discrete NBU Class Based On Generating Function Order

On Some Properties of the Discrete NBU Class Based On Generating Function Order Journal of Statistical Theory and Applications Volume 11, Number 3, 2012, pp. 209-223 ISSN 1538-7887 On Some Properties of the Discrete NBU Class Based On Generating Function Order I Elbatal Institute

More information

A novel repair model for imperfect maintenance

A novel repair model for imperfect maintenance IMA Journal of Management Mathematics (6) 7, 35 43 doi:.93/imaman/dpi36 Advance Access publication on July 4, 5 A novel repair model for imperfect maintenance SHAOMIN WU AND DEREK CLEMENTS-CROOME School

More information

Stochastic Renewal Processes in Structural Reliability Analysis:

Stochastic Renewal Processes in Structural Reliability Analysis: Stochastic Renewal Processes in Structural Reliability Analysis: An Overview of Models and Applications Professor and Industrial Research Chair Department of Civil and Environmental Engineering University

More information

Reliability of Coherent Systems with Dependent Component Lifetimes

Reliability of Coherent Systems with Dependent Component Lifetimes Reliability of Coherent Systems with Dependent Component Lifetimes M. Burkschat Abstract In reliability theory, coherent systems represent a classical framework for describing the structure of technical

More information

CHAPTER 1 A MAINTENANCE MODEL FOR COMPONENTS EXPOSED TO SEVERAL FAILURE MECHANISMS AND IMPERFECT REPAIR

CHAPTER 1 A MAINTENANCE MODEL FOR COMPONENTS EXPOSED TO SEVERAL FAILURE MECHANISMS AND IMPERFECT REPAIR CHAPTER 1 A MAINTENANCE MODEL FOR COMPONENTS EXPOSED TO SEVERAL FAILURE MECHANISMS AND IMPERFECT REPAIR Helge Langseth and Bo Henry Lindqvist Department of Mathematical Sciences Norwegian University of

More information

Conditional Tail Expectations for Multivariate Phase Type Distributions

Conditional Tail Expectations for Multivariate Phase Type Distributions Conditional Tail Expectations for Multivariate Phase Type Distributions Jun Cai Department of Statistics and Actuarial Science University of Waterloo Waterloo, ON N2L 3G1, Canada Telphone: 1-519-8884567,

More information

Time-varying failure rate for system reliability analysis in large-scale railway risk assessment simulation

Time-varying failure rate for system reliability analysis in large-scale railway risk assessment simulation Time-varying failure rate for system reliability analysis in large-scale railway risk assessment simulation H. Zhang, E. Cutright & T. Giras Center of Rail Safety-Critical Excellence, University of Virginia,

More information

Some Recent Results on Stochastic Comparisons and Dependence among Order Statistics in the Case of PHR Model

Some Recent Results on Stochastic Comparisons and Dependence among Order Statistics in the Case of PHR Model Portland State University PDXScholar Mathematics and Statistics Faculty Publications and Presentations Fariborz Maseeh Department of Mathematics and Statistics 1-1-2007 Some Recent Results on Stochastic

More information

Optimal Time and Random Inspection Policies for Computer Systems

Optimal Time and Random Inspection Policies for Computer Systems Appl. Math. Inf. Sci. 8, No. 1L, 413-417 214) 413 Applied Mathematics & Information Sciences An International Journal http://dx.doi.org/1.12785/amis/81l51 Optimal Time and Random Inspection Policies for

More information

Chapter 2 SOME ANALYTICAL TOOLS USED IN THE THESIS

Chapter 2 SOME ANALYTICAL TOOLS USED IN THE THESIS Chapter 2 SOME ANALYTICAL TOOLS USED IN THE THESIS 63 2.1 Introduction In this chapter we describe the analytical tools used in this thesis. They are Markov Decision Processes(MDP), Markov Renewal process

More information

Optimal maintenance for minimal and imperfect repair models

Optimal maintenance for minimal and imperfect repair models Optimal maintenance for minimal and imperfect repair models Gustavo L. Gilardoni Universidade de Brasília AMMSI Workshop Grenoble, January 2016 Belo Horizonte Enrico Colosimo Marta Freitas Rodrigo Padilha

More information

On Computing Signatures of k-out-of-n Systems Consisting of Modules

On Computing Signatures of k-out-of-n Systems Consisting of Modules On Computing Signatures of k-out-of-n Systems Consisting of Modules Gaofeng Da Lvyu Xia Taizhong Hu Department of Statistics Finance, School of Management University of Science Technology of China Hefei,

More information

Politecnico di Torino. Porto Institutional Repository

Politecnico di Torino. Porto Institutional Repository Politecnico di Torino Porto Institutional Repository [Article] On preservation of ageing under minimum for dependent random lifetimes Original Citation: Pellerey F.; Zalzadeh S. (204). On preservation

More information

An Integral Measure of Aging/Rejuvenation for Repairable and Non-repairable Systems

An Integral Measure of Aging/Rejuvenation for Repairable and Non-repairable Systems An Integral Measure of Aging/Rejuvenation for Repairable and Non-repairable Systems M.P. Kaminskiy and V.V. Krivtsov Abstract This paper introduces a simple index that helps to assess the degree of aging

More information

arxiv: v1 [math.pr] 1 Jan 2013

arxiv: v1 [math.pr] 1 Jan 2013 The role of dispersal in interacting patches subject to an Allee effect arxiv:1301.0125v1 [math.pr] 1 Jan 2013 1. Introduction N. Lanchier Abstract This article is concerned with a stochastic multi-patch

More information

Integrating Quality and Inspection for the Optimal Lot-sizing Problem with Rework, Minimal Repair and Inspection Time

Integrating Quality and Inspection for the Optimal Lot-sizing Problem with Rework, Minimal Repair and Inspection Time Proceedings of the 2011 International Conference on Industrial Engineering and Operations Management Kuala Lumpur, Malaysia, January 22 24, 2011 Integrating Quality and Inspection for the Optimal Lot-sizing

More information

Calculating the Expexted Time to Control the Recruitment When System Fails

Calculating the Expexted Time to Control the Recruitment When System Fails Available online at www.scinzer.com Scinzer Journal of Engineering, Vol 2, Issue 1, (2016): 1-5 Calculating the Expexted Time to Control the Recruitment When System Fails Kannadasan K*, Pandiyan P, Vinoth

More information

Filtrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition

Filtrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition Filtrations, Markov Processes and Martingales Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition David pplebaum Probability and Statistics Department,

More information

A new condition based maintenance model with random improvements on the system after maintenance actions: Optimizing by monte carlo simulation

A new condition based maintenance model with random improvements on the system after maintenance actions: Optimizing by monte carlo simulation ISSN 1 746-7233, England, UK World Journal of Modelling and Simulation Vol. 4 (2008) No. 3, pp. 230-236 A new condition based maintenance model with random improvements on the system after maintenance

More information

One important issue in the study of queueing systems is to characterize departure processes. Study on departure processes was rst initiated by Burke (

One important issue in the study of queueing systems is to characterize departure processes. Study on departure processes was rst initiated by Burke ( The Departure Process of the GI/G/ Queue and Its MacLaurin Series Jian-Qiang Hu Department of Manufacturing Engineering Boston University 5 St. Mary's Street Brookline, MA 2446 Email: hqiang@bu.edu June

More information

Preservation of Classes of Discrete Distributions Under Reliability Operations

Preservation of Classes of Discrete Distributions Under Reliability Operations Journal of Statistical Theory and Applications, Vol. 12, No. 1 (May 2013), 1-10 Preservation of Classes of Discrete Distributions Under Reliability Operations I. Elbatal 1 and M. Ahsanullah 2 1 Institute

More information

Failure rate in the continuous sense. Figure. Exponential failure density functions [f(t)] 1

Failure rate in the continuous sense. Figure. Exponential failure density functions [f(t)] 1 Failure rate (Updated and Adapted from Notes by Dr. A.K. Nema) Part 1: Failure rate is the frequency with which an engineered system or component fails, expressed for example in failures per hour. It is

More information

A cost-based importance measure for system components: an extension of the Birnbaum importance

A cost-based importance measure for system components: an extension of the Birnbaum importance A cost-based importance measure for system components: an extension of the Birnbaum importance Shaomin Wu 1 Kent Business School, University of Kent, Canterbury, Kent CT2 7PE, United Kingdom Frank P.A.

More information

Integrated Production Scheduling and Preventive Maintenance Planning for a Single Machine Under a Cumulative Damage Failure Process

Integrated Production Scheduling and Preventive Maintenance Planning for a Single Machine Under a Cumulative Damage Failure Process Integrated Production Scheduling and Preventive Maintenance Planning for a Single Machine Under a Cumulative Damage Failure Process Yarlin Kuo, Zi-Ann Chang Department of Industrial Engineering and Management,

More information

Matching via Majorization for Consistency of Product Quality

Matching via Majorization for Consistency of Product Quality Matching via Majorization for Consistency of Product Quality Lirong Cui Dejing Kong Haijun Li Abstract A new matching method is introduced in this paper to match attributes of parts in order to ensure

More information

Gamma process model for time-dependent structural reliability analysis

Gamma process model for time-dependent structural reliability analysis Gamma process model for time-dependent structural reliability analysis M.D. Pandey Department of Civil Engineering, University of Waterloo, Waterloo, Ontario, Canada J.M. van Noortwijk HKV Consultants,

More information

IEOR 6711: Stochastic Models I, Fall 2003, Professor Whitt. Solutions to Final Exam: Thursday, December 18.

IEOR 6711: Stochastic Models I, Fall 2003, Professor Whitt. Solutions to Final Exam: Thursday, December 18. IEOR 6711: Stochastic Models I, Fall 23, Professor Whitt Solutions to Final Exam: Thursday, December 18. Below are six questions with several parts. Do as much as you can. Show your work. 1. Two-Pump Gas

More information

STOCHASTIC REPAIR AND REPLACEMENT OF A STANDBY SYSTEM

STOCHASTIC REPAIR AND REPLACEMENT OF A STANDBY SYSTEM Journal of Mathematics and Statistics 0 (3): 384-389, 04 ISSN: 549-3644 04 doi:0.3844/jmssp.04.384.389 Published Online 0 (3) 04 (http://www.thescipub.com/jmss.toc) STOCHASTIC REPAIR AND REPLACEMENT OF

More information

AN INTEGRAL MEASURE OF AGING/REJUVENATION FOR REPAIRABLE AND NON REPAIRABLE SYSTEMS

AN INTEGRAL MEASURE OF AGING/REJUVENATION FOR REPAIRABLE AND NON REPAIRABLE SYSTEMS R&RAA # 1 (Vol.1) 8, March AN INEGRAL MEASURE OF AGING/REJUVENAION FOR REPAIRABLE AND NON REPAIRABLE SYSEMS M.P. Kaminskiy and V.V. Krivtsov Abstract his paper introduces a simple index that helps to assess

More information

Some stochastic inequalities for weighted sums

Some stochastic inequalities for weighted sums Some stochastic inequalities for weighted sums Yaming Yu Department of Statistics University of California Irvine, CA 92697, USA yamingy@uci.edu Abstract We compare weighted sums of i.i.d. positive random

More information

CHAPTER 1. Stochastic modelling of the effect of preventive and corrective. maintenance on repairable systems reliability

CHAPTER 1. Stochastic modelling of the effect of preventive and corrective. maintenance on repairable systems reliability CHAPTER 1 Stochastic modelling of the effect of preventive and corrective maintenance on repairable systems reliability Laurent Doyen and Olivier Gaudoin Institut National Polytechnique de Grenoble Laboratoire

More information

A conceptual interpretation of the renewal theorem with applications

A conceptual interpretation of the renewal theorem with applications Risk, Reliability and Societal Safety Aven & Vinnem (eds) 2007 Taylor & Francis Group, London, ISBN 978-0-415-44786-7 A conceptual interpretation of the renewal theorem with applications J.A.M. van der

More information

On the Decreasing Failure Rate property for general counting process. Results based on conditional interarrival times

On the Decreasing Failure Rate property for general counting process. Results based on conditional interarrival times arxiv:1209.1256v1 [math.pr] 6 Sep 2012 On the Decreasing Failure Rate property for general counting process. Results based on conditional interarrival times F. G. Badía, C.Sangüesa Departamento de Métodos

More information

A Minimal Repair Model With Imperfect Fault Detection

A Minimal Repair Model With Imperfect Fault Detection A Minimal Repair Model With Imperfect Fault Detection Hendrik Schäbe TÜV Rheinland InterTraffic e-mail: schaebe@de.tuv.com Igor Shubinski Closed company "IB Trans", e-mail: igor-shubinsky@yande.ru Abstract

More information

Expected Time Delay in Multi-Item Inventory Systems with Correlated Demands

Expected Time Delay in Multi-Item Inventory Systems with Correlated Demands Expected Time Delay in Multi-Item Inventory Systems with Correlated Demands Rachel Q. Zhang Department of Industrial and Operations Engineering, University of Michigan, Ann Arbor, Michigan 48109 Received

More information

314 IEEE TRANSACTIONS ON RELIABILITY, VOL. 55, NO. 2, JUNE 2006

314 IEEE TRANSACTIONS ON RELIABILITY, VOL. 55, NO. 2, JUNE 2006 314 IEEE TRANSACTIONS ON RELIABILITY, VOL 55, NO 2, JUNE 2006 The Mean Residual Life Function of a k-out-of-n Structure at the System Level Majid Asadi and Ismihan Bayramoglu Abstract In the study of the

More information

f X (x) = λe λx, , x 0, k 0, λ > 0 Γ (k) f X (u)f X (z u)du

f X (x) = λe λx, , x 0, k 0, λ > 0 Γ (k) f X (u)f X (z u)du 11 COLLECTED PROBLEMS Do the following problems for coursework 1. Problems 11.4 and 11.5 constitute one exercise leading you through the basic ruin arguments. 2. Problems 11.1 through to 11.13 but excluding

More information

A CONDITION-BASED MAINTENANCE MODEL FOR AVAILABILITY OPTIMIZATION FOR STOCHASTIC DEGRADING SYSTEMS

A CONDITION-BASED MAINTENANCE MODEL FOR AVAILABILITY OPTIMIZATION FOR STOCHASTIC DEGRADING SYSTEMS A CONDITION-BASED MAINTENANCE MODEL FOR AVAILABILITY OPTIMIZATION FOR STOCHASTIC DEGRADING SYSTEMS Abdelhakim Khatab, Daoud Ait-Kadi, Nidhal Rezg To cite this version: Abdelhakim Khatab, Daoud Ait-Kadi,

More information

Abstrct. In this paper, we consider the problem of optimal flow control for a production system with one machine which is subject to failures and prod

Abstrct. In this paper, we consider the problem of optimal flow control for a production system with one machine which is subject to failures and prod Monotonicity of Optimal Flow Control for Failure Prone Production Systems 1 Jian-Qiang Hu 2 and Dong Xiang 3 Communicated By W.B. Gong 1 This work is partially supported by the national Science Foundation

More information

Availability and Reliability Analysis for Dependent System with Load-Sharing and Degradation Facility

Availability and Reliability Analysis for Dependent System with Load-Sharing and Degradation Facility International Journal of Systems Science and Applied Mathematics 2018; 3(1): 10-15 http://www.sciencepublishinggroup.com/j/ijssam doi: 10.11648/j.ijssam.20180301.12 ISSN: 2575-5838 (Print); ISSN: 2575-5803

More information

Network Reliability Assessment in a Random Environment

Network Reliability Assessment in a Random Environment Network Reliability Assessment in a Random Environment S. Özekici, 1 R. Soyer 2 1 Department of Industrial Engineering, Koç University, 80910 Sarıyer-İstanbul, Turkey 2 Department of Management Science,

More information

Bounds for reliability of IFRA coherent systems using signatures

Bounds for reliability of IFRA coherent systems using signatures isid/ms/2011/17 September 12, 2011 http://www.isid.ac.in/ statmath/eprints Bounds for reliability of IFRA coherent systems using signatures Jayant V. Deshpande Isha Dewan Indian Statistical Institute,

More information

Some New Results on Information Properties of Mixture Distributions

Some New Results on Information Properties of Mixture Distributions Filomat 31:13 (2017), 4225 4230 https://doi.org/10.2298/fil1713225t Published by Faculty of Sciences and Mathematics, University of Niš, Serbia Available at: http://www.pmf.ni.ac.rs/filomat Some New Results

More information

Abstract This paper deals with the problem of scheduling imperfect preventive maintenance (PM) of some equipment. It uses a model due to Kijima in whi

Abstract This paper deals with the problem of scheduling imperfect preventive maintenance (PM) of some equipment. It uses a model due to Kijima in whi Modelling and Optimizing Sequential Imperfect Preventive Maintenance Michael Bartholomew-Biggs 1, Ming J. Zuo 2 and Xiaohu Li 3 1 School of Physics Astronomy and Mathematics, University of Hertfordshire,

More information

Optimal Control of PDEs

Optimal Control of PDEs Optimal Control of PDEs Suzanne Lenhart University of Tennessee, Knoville Department of Mathematics Lecture1 p.1/36 Outline 1. Idea of diffusion PDE 2. Motivating Eample 3. Big picture of optimal control

More information

Recap. Probability, stochastic processes, Markov chains. ELEC-C7210 Modeling and analysis of communication networks

Recap. Probability, stochastic processes, Markov chains. ELEC-C7210 Modeling and analysis of communication networks Recap Probability, stochastic processes, Markov chains ELEC-C7210 Modeling and analysis of communication networks 1 Recap: Probability theory important distributions Discrete distributions Geometric distribution

More information

Some results on the ageing class

Some results on the ageing class Control and Cybernetics vol. 34 (25) No. 4 Some results on the ageing class by Leszek Knopik Faculty of Mechanical Engineering, University of Technology and Agriculture Bydgoszcz, ul.kaliskiego 7, Poland

More information

Lecture 10: Semi-Markov Type Processes

Lecture 10: Semi-Markov Type Processes Lecture 1: Semi-Markov Type Processes 1. Semi-Markov processes (SMP) 1.1 Definition of SMP 1.2 Transition probabilities for SMP 1.3 Hitting times and semi-markov renewal equations 2. Processes with semi-markov

More information

MS&E 321 Spring Stochastic Systems June 1, 2013 Prof. Peter W. Glynn Page 1 of 7

MS&E 321 Spring Stochastic Systems June 1, 2013 Prof. Peter W. Glynn Page 1 of 7 MS&E 321 Spring 12-13 Stochastic Systems June 1, 213 Prof. Peter W. Glynn Page 1 of 7 Section 9: Renewal Theory Contents 9.1 Renewal Equations..................................... 1 9.2 Solving the Renewal

More information

Solutions to Homework Discrete Stochastic Processes MIT, Spring 2011

Solutions to Homework Discrete Stochastic Processes MIT, Spring 2011 Exercise 1 Solutions to Homework 6 6.262 Discrete Stochastic Processes MIT, Spring 2011 Let {Y n ; n 1} be a sequence of rv s and assume that lim n E[ Y n ] = 0. Show that {Y n ; n 1} converges to 0 in

More information

Stochastic Comparisons of Two-Units Repairable Systems

Stochastic Comparisons of Two-Units Repairable Systems Stochastic Comparisons of Two-Units Repairable Systems Josué M. Corujo a, José E. Valdés a and Juan C. Laria b arxiv:1804.03098v [math.pr] 5 Sep 018 a Facultad de Matemática y Computación, Universidad

More information

RELATING TIME AND CUSTOMER AVERAGES FOR QUEUES USING FORWARD COUPLING FROM THE PAST

RELATING TIME AND CUSTOMER AVERAGES FOR QUEUES USING FORWARD COUPLING FROM THE PAST J. Appl. Prob. 45, 568 574 (28) Printed in England Applied Probability Trust 28 RELATING TIME AND CUSTOMER AVERAGES FOR QUEUES USING FORWARD COUPLING FROM THE PAST EROL A. PEKÖZ, Boston University SHELDON

More information

CHAPTER 3 MATHEMATICAL AND SIMULATION TOOLS FOR MANET ANALYSIS

CHAPTER 3 MATHEMATICAL AND SIMULATION TOOLS FOR MANET ANALYSIS 44 CHAPTER 3 MATHEMATICAL AND SIMULATION TOOLS FOR MANET ANALYSIS 3.1 INTRODUCTION MANET analysis is a multidimensional affair. Many tools of mathematics are used in the analysis. Among them, the prime

More information

Mathematics Department, Birjand Branch, Islamic Azad University, Birjand, Iran

Mathematics Department, Birjand Branch, Islamic Azad University, Birjand, Iran REVSTAT Statistical Journal Volume 14, Number 3, June 216, 229 244 PROPERTIES OF n-laplace TRANSFORM RATIO ORDER AND -CLASS OF LIFE DISTRIBUTIONS Authors: Jalil Jarrahiferiz Mathematics Department, Birjand

More information

Lecture 4a: Continuous-Time Markov Chain Models

Lecture 4a: Continuous-Time Markov Chain Models Lecture 4a: Continuous-Time Markov Chain Models Continuous-time Markov chains are stochastic processes whose time is continuous, t [0, ), but the random variables are discrete. Prominent examples of continuous-time

More information

arxiv:math/ v4 [math.pr] 12 Apr 2007

arxiv:math/ v4 [math.pr] 12 Apr 2007 arxiv:math/612224v4 [math.pr] 12 Apr 27 LARGE CLOSED QUEUEING NETWORKS IN SEMI-MARKOV ENVIRONMENT AND ITS APPLICATION VYACHESLAV M. ABRAMOV Abstract. The paper studies closed queueing networks containing

More information

T. Liggett Mathematics 171 Final Exam June 8, 2011

T. Liggett Mathematics 171 Final Exam June 8, 2011 T. Liggett Mathematics 171 Final Exam June 8, 2011 1. The continuous time renewal chain X t has state space S = {0, 1, 2,...} and transition rates (i.e., Q matrix) given by q(n, n 1) = δ n and q(0, n)

More information

Distributions of failure times associated with non-homogeneous compound Poisson damage processes

Distributions of failure times associated with non-homogeneous compound Poisson damage processes A Festschrift for Herman Rubin Institute of Mathematical Statistics Lecture Notes Monograph Series Vol. 45 (2004) 396 407 c Institute of Mathematical Statistics, 2004 Distributions of failure times associated

More information

Design of multi-component periodic maintenance programs with single-component models

Design of multi-component periodic maintenance programs with single-component models Design of multi-component periodic maintenance programs with single-component models Joachim Arts 1 and Rob Basten 2 1 University of Luxembourg, Luxembourg Centre for Logistics and Supply Chain Management,

More information

Reliability Analysis in Uncertain Random System

Reliability Analysis in Uncertain Random System Reliability Analysis in Uncertain Random System Meilin Wen a,b, Rui Kang b a State Key Laboratory of Virtual Reality Technology and Systems b School of Reliability and Systems Engineering Beihang University,

More information

Mahdi karbasian* & Zoubi Ibrahim

Mahdi karbasian* & Zoubi Ibrahim International Journal of Industrial Engineering & Production Research (010) pp. 105-110 September 010, Volume 1, Number International Journal of Industrial Engineering & Production Research ISSN: 008-4889

More information

N.G.Bean, D.A.Green and P.G.Taylor. University of Adelaide. Adelaide. Abstract. process of an MMPP/M/1 queue is not a MAP unless the queue is a

N.G.Bean, D.A.Green and P.G.Taylor. University of Adelaide. Adelaide. Abstract. process of an MMPP/M/1 queue is not a MAP unless the queue is a WHEN IS A MAP POISSON N.G.Bean, D.A.Green and P.G.Taylor Department of Applied Mathematics University of Adelaide Adelaide 55 Abstract In a recent paper, Olivier and Walrand (994) claimed that the departure

More information

LIMITS FOR QUEUES AS THE WAITING ROOM GROWS. Bell Communications Research AT&T Bell Laboratories Red Bank, NJ Murray Hill, NJ 07974

LIMITS FOR QUEUES AS THE WAITING ROOM GROWS. Bell Communications Research AT&T Bell Laboratories Red Bank, NJ Murray Hill, NJ 07974 LIMITS FOR QUEUES AS THE WAITING ROOM GROWS by Daniel P. Heyman Ward Whitt Bell Communications Research AT&T Bell Laboratories Red Bank, NJ 07701 Murray Hill, NJ 07974 May 11, 1988 ABSTRACT We study the

More information

A General Bayes Weibull Inference Model for Accelerated Life Testing

A General Bayes Weibull Inference Model for Accelerated Life Testing A General Bayes Weibull Inference Model for Accelerated Life Testing J. René Van Dorp & Thomas A. Mazzuchi The George Washington University, Washington D.C., USA Submitted to: European Safety and Reliability

More information

1 Delayed Renewal Processes: Exploiting Laplace Transforms

1 Delayed Renewal Processes: Exploiting Laplace Transforms IEOR 6711: Stochastic Models I Professor Whitt, Tuesday, October 22, 213 Renewal Theory: Proof of Blackwell s theorem 1 Delayed Renewal Processes: Exploiting Laplace Transforms The proof of Blackwell s

More information

Anew index of component importance

Anew index of component importance Operations Research Letters 28 (2001) 75 79 www.elsevier.com/locate/dsw Anew index of component importance F.K. Hwang 1 Department of Applied Mathematics, National Chiao-Tung University, Hsin-Chu, Taiwan

More information

Are Base-stock Policies Optimal in Inventory Problems with Multiple Delivery Modes?

Are Base-stock Policies Optimal in Inventory Problems with Multiple Delivery Modes? Are Base-stoc Policies Optimal in Inventory Problems with Multiple Delivery Modes? Qi Feng School of Management, the University of Texas at Dallas Richardson, TX 7508-0688, USA Guillermo Gallego Department

More information

A New Extended Mixture Model of Residual Lifetime Distributions

A New Extended Mixture Model of Residual Lifetime Distributions A New Extended Mixture Model of Residual Lifetime Distributions M. Kayid 1 Dept.of Statistics and Operations Research, College of Science, King Saud University, P.O. Box 2455, Riyadh 11451, KSA S. Izadkhah

More information

On the Pathwise Optimal Bernoulli Routing Policy for Homogeneous Parallel Servers

On the Pathwise Optimal Bernoulli Routing Policy for Homogeneous Parallel Servers On the Pathwise Optimal Bernoulli Routing Policy for Homogeneous Parallel Servers Ger Koole INRIA Sophia Antipolis B.P. 93, 06902 Sophia Antipolis Cedex France Mathematics of Operations Research 21:469

More information

Conditional independence of blocked ordered data

Conditional independence of blocked ordered data Conditional independence of blocked ordered data G. Iliopoulos 1 and N. Balakrishnan 2 Abstract In this paper, we prove that blocks of ordered data formed by some conditioning events are mutually independent.

More information

Examination paper for TMA4275 Lifetime Analysis

Examination paper for TMA4275 Lifetime Analysis Department of Mathematical Sciences Examination paper for TMA4275 Lifetime Analysis Academic contact during examination: Ioannis Vardaxis Phone: 95 36 00 26 Examination date: Saturday May 30 2015 Examination

More information

1. Reliability and survival - basic concepts

1. Reliability and survival - basic concepts . Reliability and survival - basic concepts. Books Wolstenholme, L.C. "Reliability modelling. A statistical approach." Chapman & Hall, 999. Ebeling, C. "An introduction to reliability & maintainability

More information

Data analysis and stochastic modeling

Data analysis and stochastic modeling Data analysis and stochastic modeling Lecture 7 An introduction to queueing theory Guillaume Gravier guillaume.gravier@irisa.fr with a lot of help from Paul Jensen s course http://www.me.utexas.edu/ jensen/ormm/instruction/powerpoint/or_models_09/14_queuing.ppt

More information

Framework for functional tree simulation applied to 'golden delicious' apple trees

Framework for functional tree simulation applied to 'golden delicious' apple trees Purdue University Purdue e-pubs Open Access Theses Theses and Dissertations Spring 2015 Framework for functional tree simulation applied to 'golden delicious' apple trees Marek Fiser Purdue University

More information

STAT331 Lebesgue-Stieltjes Integrals, Martingales, Counting Processes

STAT331 Lebesgue-Stieltjes Integrals, Martingales, Counting Processes STAT331 Lebesgue-Stieltjes Integrals, Martingales, Counting Processes This section introduces Lebesgue-Stieltjes integrals, and defines two important stochastic processes: a martingale process and a counting

More information

Identifiability Issues in Dynamic Stress-Strength Modeling

Identifiability Issues in Dynamic Stress-Strength Modeling Identifiability Issues in Dynamic Stress-Strength Modeling Technical Report No. ASU/215/2 Dated: 4 June, 215 Prajamitra Bhuyan Applied Statistics Unit, Indian Statistical Institute, Kolkata praja r@isical.ac.in

More information

Non-Essential Uses of Probability in Analysis Part IV Efficient Markovian Couplings. Krzysztof Burdzy University of Washington

Non-Essential Uses of Probability in Analysis Part IV Efficient Markovian Couplings. Krzysztof Burdzy University of Washington Non-Essential Uses of Probability in Analysis Part IV Efficient Markovian Couplings Krzysztof Burdzy University of Washington 1 Review See B and Kendall (2000) for more details. See also the unpublished

More information

Budapest University of Tecnology and Economics. AndrásVetier Q U E U I N G. January 25, Supported by. Pro Renovanda Cultura Hunariae Alapítvány

Budapest University of Tecnology and Economics. AndrásVetier Q U E U I N G. January 25, Supported by. Pro Renovanda Cultura Hunariae Alapítvány Budapest University of Tecnology and Economics AndrásVetier Q U E U I N G January 25, 2000 Supported by Pro Renovanda Cultura Hunariae Alapítvány Klebelsberg Kunó Emlékére Szakalapitvány 2000 Table of

More information

Stochastic Models. Edited by D.P. Heyman Bellcore. MJ. Sobel State University of New York at Stony Brook

Stochastic Models. Edited by D.P. Heyman Bellcore. MJ. Sobel State University of New York at Stony Brook Stochastic Models Edited by D.P. Heyman Bellcore MJ. Sobel State University of New York at Stony Brook 1990 NORTH-HOLLAND AMSTERDAM NEW YORK OXFORD TOKYO Contents Preface CHARTER 1 Point Processes R.F.

More information

(b) What is the variance of the time until the second customer arrives, starting empty, assuming that we measure time in minutes?

(b) What is the variance of the time until the second customer arrives, starting empty, assuming that we measure time in minutes? IEOR 3106: Introduction to Operations Research: Stochastic Models Fall 2006, Professor Whitt SOLUTIONS to Final Exam Chapters 4-7 and 10 in Ross, Tuesday, December 19, 4:10pm-7:00pm Open Book: but only

More information

EE 445 / 850: Final Examination

EE 445 / 850: Final Examination EE 445 / 850: Final Examination Date and Time: 3 Dec 0, PM Room: HLTH B6 Exam Duration: 3 hours One formula sheet permitted. - Covers chapters - 5 problems each carrying 0 marks - Must show all calculations

More information

The strong law of large numbers for arrays of NA random variables

The strong law of large numbers for arrays of NA random variables International Mathematical Forum,, 2006, no. 2, 83-9 The strong law of large numbers for arrays of NA random variables K. J. Lee, H. Y. Seo, S. Y. Kim Division of Mathematics and Informational Statistics,

More information

Random Walk on a Graph

Random Walk on a Graph IOR 67: Stochastic Models I Second Midterm xam, hapters 3 & 4, November 2, 200 SOLUTIONS Justify your answers; show your work.. Random Walk on a raph (25 points) Random Walk on a raph 2 5 F B 3 3 2 Figure

More information

Lecture 20: Reversible Processes and Queues

Lecture 20: Reversible Processes and Queues Lecture 20: Reversible Processes and Queues 1 Examples of reversible processes 11 Birth-death processes We define two non-negative sequences birth and death rates denoted by {λ n : n N 0 } and {µ n : n

More information

Minimizing response times and queue lengths in systems of parallel queues

Minimizing response times and queue lengths in systems of parallel queues Minimizing response times and queue lengths in systems of parallel queues Ger Koole Department of Mathematics and Computer Science, Vrije Universiteit, De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands

More information

7 Convergence in R d and in Metric Spaces

7 Convergence in R d and in Metric Spaces STA 711: Probability & Measure Theory Robert L. Wolpert 7 Convergence in R d and in Metric Spaces A sequence of elements a n of R d converges to a limit a if and only if, for each ǫ > 0, the sequence a

More information

Lecturer: Olga Galinina

Lecturer: Olga Galinina Renewal models Lecturer: Olga Galinina E-mail: olga.galinina@tut.fi Outline Reminder. Exponential models definition of renewal processes exponential interval distribution Erlang distribution hyperexponential

More information

Introduction to repairable systems STK4400 Spring 2011

Introduction to repairable systems STK4400 Spring 2011 Introduction to repairable systems STK4400 Spring 2011 Bo Lindqvist http://www.math.ntnu.no/ bo/ bo@math.ntnu.no Bo Lindqvist Introduction to repairable systems Definition of repairable system Ascher and

More information

Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek

Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek J. Korean Math. Soc. 41 (2004), No. 5, pp. 883 894 CONVERGENCE OF WEIGHTED SUMS FOR DEPENDENT RANDOM VARIABLES Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek Abstract. We discuss in this paper the strong

More information

COMPOSITE RELIABILITY MODELS FOR SYSTEMS WITH TWO DISTINCT KINDS OF STOCHASTIC DEPENDENCES BETWEEN THEIR COMPONENTS LIFE TIMES

COMPOSITE RELIABILITY MODELS FOR SYSTEMS WITH TWO DISTINCT KINDS OF STOCHASTIC DEPENDENCES BETWEEN THEIR COMPONENTS LIFE TIMES COMPOSITE RELIABILITY MODELS FOR SYSTEMS WITH TWO DISTINCT KINDS OF STOCHASTIC DEPENDENCES BETWEEN THEIR COMPONENTS LIFE TIMES Jerzy Filus Department of Mathematics and Computer Science, Oakton Community

More information

Introduction to Reliability Theory (part 2)

Introduction to Reliability Theory (part 2) Introduction to Reliability Theory (part 2) Frank Coolen UTOPIAE Training School II, Durham University 3 July 2018 (UTOPIAE) Introduction to Reliability Theory 1 / 21 Outline Statistical issues Software

More information

A Note on Closure Properties of Classes of Discrete Lifetime Distributions

A Note on Closure Properties of Classes of Discrete Lifetime Distributions International Journal of Statistics and Probability; Vol. 2, No. 2; 201 ISSN 1927-702 E-ISSN 1927-700 Published by Canadian Center of Science and Education A Note on Closure Properties of Classes of Discrete

More information

Reading: Karlin and Taylor Ch. 5 Resnick Ch. 3. A renewal process is a generalization of the Poisson point process.

Reading: Karlin and Taylor Ch. 5 Resnick Ch. 3. A renewal process is a generalization of the Poisson point process. Renewal Processes Wednesday, December 16, 2015 1:02 PM Reading: Karlin and Taylor Ch. 5 Resnick Ch. 3 A renewal process is a generalization of the Poisson point process. The Poisson point process is completely

More information