SYNOPTIC ABSTRACT
The article discusses truncated inactivity lifetimes for systems with cold standby units. The most common bivariate lifetime distribution was used to clarify these systems. The essay concentrates on the Farlie-Gumbel-Morgenstern family for building a dependence structure between components and to study the properties of suggested systems. The article also compares systems with independent identically distributed components and dependent identically distributed components having the same survival function by using various stochastic orders. Moreover, some applications of the theoretical results of reliability theory are studied.
KEY WORDS AND PHRASES:
1. Introduction
Suppose T is a positive random variable that denotes the lifetime of a unit or equipment, having density function f, absolutely continuous distribution function F, and survival function with respect to the Lebesgue measure. In many reliability problems, it is interesting to consider variables of the following types. denote the time elapsed after failure until time t, given that the unit has already failed at time t, having distribution function , and which is known in literature as inactivity time or past lifetime. In recent times, the random variable has received considerable attention in the literature; see El-Bassiouny and Alwasel (Citation2003), Lai and Xie (Citation2006), Nair and Sudheesh (Citation2010), Mahdy (Citation2012, Citation2016), Rezaei, Gholizadeh, and Izadkhah (Citation2015), and Kundu and Sarkar (Citation2017).
Consider a probability density function f(t) for a lifetime random variable X with distribution function and survival function In addition, we assume that the mean life and variance are finite. Likewise, let Y be the second lifetime random variable with density function , distribution function , and survival function ; . Furthermore, the mean life and variance are both assumed to be finite. Let and . Let X have a reversed hazard rate function , and Y have a reversed hazard rate function Then, the mean inactivity lifetime functions and variance past lifetime functions are defined by, and respectively.
The following definitions are essential for this study.
Definition 1.
The distribution function of the random variable X is said to have the following characteristics.
A decreasing reversed hazard rate , if is a decreasing function in t, or if is logarithmically concave in
An increasing strong mean past lifetime class, if is non-decreasing for all .
The following stochastic orders are defined in Nanda, Singh, Misra, and Paul (Citation2003), Shaked and Shanthikumar (Citation2007), Mahdy (Citation2009, Citation2012), and Kayid and Izadkhah (Citation2014).
Definition 2.
Let X1 and X2 be two non-negative and absolutely continuous random variables, with distribution functions and , density functions and , reliability functions and reversed hazard rate functions and , mean past lifetime functions and , and variance past lifetime functions and , respectively. Hence, X1 is smaller than or equal to X2 in the following cases as,
A usual stochastic order ( if for all .
A hazard rate order () if is decreasing in .
A reversed residual lifetime order () if for all or for all .
A mean past lifetime order () if for all .
A variance past lifetime order () if for all .
A strong mean past lifetime order () if for all .
A likelihood ratio order ( if for all .
Suppose is the lifetime of a coherent system, and are dependent lifetimes with a common survival function . Then, the system survival function can be presented as, where depends only on ψ and on the survival copula of Lehmann (Citation1966) considered the definition of the bivariate dependence of two random variables and as, (1) (1)
A solution methodology is provided and demonstrated to determine optimal design configurations for nonrepairable coherent systems with three types of redundancies: cold standby, warm standby, and hot standby. The cold standby redundancy method involves having one system as a backup for another identical primary system where in the response time is of minimal concern. In the warm standby redundancy method, the secondary system runs in the background of the primary system, and it is used when time is somewhat critical. Finally, hot redundancy is used when the process must not go down for even a brief moment under any circumstance and where the primary and secondary systems run simultaneously.
A large number of studies have discussed standby systems and their extensions and applications, including Yang and Dhillon (Citation1995), Wang, Lai, and Li (Citation2003), Meng, Yuan, and Yin (Citation2006), Eryilmaz (Citation2011, Citation2012, Citation2013), Tannous, Xing, Rui, Xie, and Ng (Citation2011), Wu and Wu (Citation2011), Xing, Tannous, and Dugan (Citation2012), Zhai, Peng, Xing, and Yang (Citation2013), Levitin, Xing, and Dai (Citation2014a, Levitin, Xing, & Dai, Citation2014b, Levitin, Xing, Johnson, & Dai, Citation2015), Liu, Cui, Wen, and Guo (Citation2016), Kaur and Gupta (Citation2017), and Mendes, Ribeiro, and Coit (Citation2017).
Throughout this study, SU is used to indicate the single unit system, is used to indicate the cold standby unit, AU is used to indicate active equipment, SDU is used to indicate standby equipment, is used to indicate the conditional distribution function, and RL is used to indicate the residual lifetime of the system.
Let SU be equipped with CU, and suppose denotes the lifetime of AU and denotes the Obviously, the lifetime of the whole system corresponds to the random variable . Let the system have survived to the age of s1 and already failed by time s2; then, the inactivity lifetime of a system can be defined as, for and Here, implies only the survival of the system at time s1 and failure by time s2, but no information is included about which unit survives at time s1. Note, that the system has benefited from the alternative element (standby unit). Therefore, it could exceed the age of s1, but it fails before time s2 even after we use the standby unit.
To the author’s knowledge, the inactivity lifetime for a system with SDU has been scarcely investigated from a theoretical point of view. Therefore, we define it by analyzing and studying the properties of the following conditional random variables. (2) (2) and (3) (3) where represents the inactivity lifetime of the system, considering that AU has a breakdown before time s1; on the other side, the full system works with the effect of the standby unit and fails before time Similarly, represents the inactivity lifetime of the system under the condition that the system works at time s1 with the active component and fails before time s2. Clearly, and are more helpful than .
For study and analysis of systems and with identically distributed (ID) components, we consider two cases wherein the relationship between the two components and is either dependent or independent. The most common bivariate lifetime distribution (bivariate generalized exponential distribution; BGE) was used to clarify these systems. Moreover, the Farlie-Gumbel-Morgenstern (FGM) family is extensively used in building a dependence structure between the marginal distributions; we use it to build the distribution functions of and when and are dependent. Furthermore, we derive and by using the FGM Copula. We compare the systems with independent identically distributed components and dependent identically distributed components having the same survival function by using various stochastic orders. In addition, we provide some applications of theoretical results using reliability theory.
The remainder of the article is organized as follows. Section 2 describes the main results, and Section 3 discusses and includes some reliability applications of the new lifetime’s classes.
2. The Main Results
Suppose and are two dependent lifetime random variables with joint cumulative distribution function and marginal distributions and , for .
Now, we can derive the distribution function of by the following theorem.
Theorem 1.
The conditional distribution function of given, can be represented as follows.
For , (4) (4)
For , and , (5) (5)
Proof.
By considering the conditioning on and letting or , we have the following results. (6) (6)
For and , we have , and by EquationEquation (6)(6) (6) we have, (7) (7) and for and , we get, (8) (8)
In addition, (9) (9)
It follows from EquationEquations (7–9) that the required result (i) is provided. When,
and , we have, (10) (10)
Similarly, (11) (11) is also valid for the following relation.
Substituting EquationEquations (9–11) into EquationEquation (6)(6) (6) , we obtain (ii).
Based on relation EquationEquation (1)(1) (1) , we can say that there is positive dependence between and if,
However, and then, (12) (12)
In addition, based on the alternative positive dependence introduced by Lai and Xie (Citation2006), we can say that is left-tail decreasing in in if,
Under this definition, we can rewrite relation EquationEquation (12)(12) (12) as,
By using Theorem (1), we can determine the average value of as, (13) (13)
Corollary 1.
Let denote the lifetime of an equipment with distribution function , and survival function . Similarly, let be the standby random lifetime with distribution function and survival function When and are independent, we have the COD of given , which can be obtained as,
When , and we have, (14) (14) where,
The cold standby redundancy method involves having one system as a backup for another identical primary system wherein the response time is of minimal concern. When and are independent and we consider the cold system, it is evident that the inactivity time of the system is the inactivity time of . Since the system will fail before time and there is residual lifetime after s1, the inactivity time of the system is the inactivity time of at time ; consequently, the inactivity time of system is obtained as follows. (15) (15)
The residual of system after s1 can be represented as, (16) (16)
The BGE is discussed in Kundu and Gupta (Citation2009) and has been utilized in many applications, such as analysis of lifetime data. In the following application, we clarify the above results by using BGE.
If indicates bivariate random variables and , then the joint probability density function of for and u2 is, where,
Theorem 2.
If indicates bivariate random variables and , we obtain the mean the inactivity lifetime of the system given that the active unit has failed before time s1, which is denoted by as follows. when where
Proof.
Since if and , it follows the series given below (Nadarajah & Kotz, Citation2004, pp. 324, Equation (1.7)). (17) (17) and (18) (18)
Then, by using EquationEquation (18)(18) (18) , we can obtain (19) (19)
Therefore, EquationEquation (19)(19) (19) implies that, (20) (20)
In addition,
We write the above expression as, (21) (21) where and
According to EquationEquations (20)(20) (20) and Equation(21)(21) (21) , after some simplifications, we have, where
By using Theorem 2.3 in Kundu and Gupta (Citation2009), we have, where for and where and are the singular and absolute continuous parts, respectively. With and by using EquationEquations (13)(13) (13) , Equation(17)(17) (17) , and Equation(19)(19) (19) , we can complete the proof.
Proposition 1.
If indicates bivariate Weibull random variables with the joint probability density function of, if and otherwise. By EquationEquation (15)(15) (15) , we can compute the average value of by the transformation technique as follows.
By using EquationEquation (16)(16) (16) , we get the residual of system after s1 as, where and is the upper incomplete gamma.
In the following result, we study the distribution function of defined by EquationEquation (3)(3) (3) .
Theorem 3.
Suppose and are dependent lifetime random variables with the joint cumulative distribution function and marginal distributions and , for . Hence, the conditional distribution function of given is, where
Proof.
Consider or , where . Then, we have, and
Thus, the required result is provided.
By using Theorem (3), the average value of can be presented as the following formula.
Corollary 2.
Let denote the lifetime of an equipment with distribution function and survival function . Similarly, let be a standby random lifetime with distribution function and survival function When and are independent, we have the average value of as follows. (22) (22) where
The Farlie-Gumbel-Morgenstern (FGM) family is extensively used in building the dependence structure between marginal distributions. It is defined as, where This section establishes explicit expressions for systems by using the FGM Copula. Now, the density function of the FGM Copula can be obtained as,
Now, we derive by using the FGM Copula as follows. where
In addition,
This can be rewritten as, where
Next, we provide the main results related to stochastic orders. First, we compare a system with and having the same survival function.
Theorem 4.
Let = be a random variable of system based on lifetimes with the same distribution function . Suppose = is a random variable of system based on lifetimes with the same distribution function Let and be their respective domination functions. Thus, the following properties are obtained.
for all F if and only if is a decreasing function in , and for all
for all F if and only if is a decreasing (increasing) function in x, and for all
for all F if and only if is a decreasing (increasing) function in x, and for all
for all F if and only if is decreasing (increasing) for all t, and for all
Proof.
According to EquationEquation (4)(4) (4) and EquationEquation (14)(14) (14) and as mentioned earlier in Definitions (1 and 2), we can get the complete proof.
Proposition 2.
Let = be a random variable of system EquationEquation (1)(1) (1) based on lifetimes with the same distribution function F. Suppose is a random variable of system () based on lifetimes with the same distribution function Let and be their respective domination functions. Thus, the following properties are obtained.
for all if and only if is convex in
Proof.
By using the main results related to the likelihood ratio order introduced in Müller and Stoyan (Citation2002), Hu and Zhuang (Citation2005), and Khaledi and Shaked (Citation2007), and according to EquationEquation (4)(4) (4) and EquationEquation (14)(14) (14) as mentioned in Definitions (1 and 2), we can determine whether and are two systems with common survival functions as well as hazard rate functions and respectively. Therefore, we can obtain the complete proof.
Theorem 5.
Under the assumptions of Proposition 2, when and are differentiable, if and only if / is increasing for all .
Proof.
The results can be provided by using the same steps in Theorem 2.5 in Navarro et al. (Citation2013).
Proposition 3.
(i) If where is a continuous strictly decreasing function on (
(ii) If for any such function
(iii) If and if , which is independent of and , we obtain
Proof.
From Theorem (2.2) in Ahmad, Kayid, and Pellerey (Citation2005), Corollary (3.1) and Theorems (3.2 and 3.3) in Kayid and Ahmad (Citation2004), and Nanda and Shaked (Citation2001), we deduce the results.
3. Applications
Based on the approach presented in previous sections, the purpose of this study is to introduce some applications of theoretical results to dependence in reliability. We consider the independence relationship between the main system and the standby system. There is a dependence relationship between the components of the same system that are used in the same environment or that share the same load while supposing that components of the standby system are independent.
We assume that and are lifetimes of the components of two systems and , respectively, and we suppose that and , for The joint distribution functions of and are written as, and respectively. By using Sklar’s theorem (see, e.g., Nelsen, Citation2006), we get and , which can be represented, respectively, as, and where κ is the connecting copula. We use the following lemma, which was introduced by Nelsen (Citation2006).
Lemma 1.
If κ is a copula, then where and for all . Functions κL and κU are called Frèchet–Hoeffding bounds or minimal and maximal copulas, respectively. Now, the average of is, and the mean function of κU can be written as for all and .
Application 1. Suppose a system with lifetimes and Therefore, from EquationEquations (15)(15) (15) , Equation(16)(16) (16) , and Equation(22)(22) (22) , we get the average value of as, where (23) (23) where and is the mean function of
Now, we obtain RL after s1, by using EquationEquation (16)(16) (16) as follows. where and where is the mean function of κ while considering random variable .
As mentioned in Navarro and Spizzichino (Citation2010), a vector () is said to be majorized by another vector () (()) if and for , once are order statistics of random variable Analogously, for random variable , we have as order statistics.
Application 2. Suppose is a main system containing lifetimes and a standby system , with joint distribution functions of and that are written as, and respectively. Now, let there be another system with lifetimes and standby system , with joint distribution functions of and that are written as, and respectively. EquationEquation (23)(23) (23) implies that, where
Similarly, distribution of the second system can be represented as, where
We use the same distribution copula κ with the same mean for the first and second main systems. According to Theorem 3.1 in Navarro and Spizzichino (Citation2010), if κ is an increasing function, we get if and only if for all s, where with distribution function Analogously, with distribution function Then, we have which leads to By Proposition 2, we have , and
Application 3. Suppose Z1, Z2 and Z3 are mutually independent components with the following density functions. where denotes the generalized exponential distribution with parameters , while denotes the exponential distribution with parameter α0. For more details about these distributions see Gupta and Kundu (Citation2001). Let a system have two systems Assume that all of them are reformed and maintained independently and,
According to Sarhan and Balakrishnan (Citation2007), the joint distribution function of and is, (24) (24) where By Lemma (3.2) in Sarhan and Balakrishnan (Citation2007), we have the case of the conditional bivariate exponential density function. Upon setting in EquationEquation (24)(24) (24) , we get, (25) (25) where and the marginal probability density function of in this case is
From this, we deduce that, (26) (26) and (27) (27) where
and
Similarly, we get, (28) (28) and (29) (29) where
EquationEquations (25)–(29) demonstrate that the COD of given for is as follows.
Similarly, for and , we get the COD of given as, where
Acknowledgements
The author is deeply thankful to the editor and reviewers for their time and efforts.
References
- Ahmad, I. A., Kayid, M., & Pellerey, F. (2005). Further results involving the MIT order and the IMIT class. Probability in the Engineering and Informational Sciences, 19(3), 337–395.
- El-Bassiouny, A. H., & Alwasel, I. A. (2003). A goodness of fit approach to testing mean residual times. Applied Mathematics and Computation, 143(2–3), 301–307.
- Eryilmaz, S. (2011). The behavior of warm standby components with respect to a coherent system. Statistics & Probability Letters, 81(8), 1319–1325.
- Eryilmaz, S. (2012). On the mean residual life of a k-out-of-n:G system with a single cold standby component. European Journal of Operational Research, 222(2), 273–277.
- Eryilmaz, S. (2013). Reliability of a k-out-of-n system equipped with a single warm standby component. IEEE Transactions on Reliability, 62(2), 499–503.
- Gupta, R. D., & Kundu, D. (2001). Generalized exponential distribution: Different method of estimations. Journal of Statistical Computation and Simulation, 69(4), 315–337.
- Hu, T., & Zhuang, W. (2005). Stochastic properties of p-spacings of generalized order statistics. Probability in the Engineering and Informational Sciences, 19(02), 257–276.
- Kaur, G., & Gupta, P. (2017). Availability improvement of a hot standby system by inclusion of a substitute system. International Journal of Statistics and Reliability Engineering, 4(2), 105–109.
- Khaledi, B. E., & Shaked, M. (2007). Ordering conditional lifetimes of coherent systems. Journal of Statistical Planning and Inference, 137(4), 1173–1184.
- Kayid, M., & Ahmad, I. A. (2004). On the mean inactivity time ordering with reliability applications. Probability in the Engineering and Informational Sciences, 8, 395–409.
- Kayid, M., & Izadkhah, S. (2014). Mean inactivity time function, associated orderings and classes of life distributions. IEEE Transactions on Reliability, 63(2), 593–602.
- Kundu, D., & Gupta, R. D. (2009). Bivariate generalized exponential distribution. Journal of Multivariate Analysis, 100, 581–593.
- Kundu, C., & Sarkar, K. (2017). Characterizations based on higher order and partial moments of inactivity time. Statistical Papers, 58(3), 607–626.
- Lai, C. D., & Xie, M. (2006). Stochastic Aging and Dependence for Reliability. Springer Science and Business Media, Inc., New York, USA.
- Lehmann, E. L. (1966). Some concepts of dependence. The Annals of Mathematical Statistics, 37(5), 1137–1153.
- Levitin, G., Xing, L., Johnson, B. W., Y., & Dai, Y. (2015). Mission reliability, cost and time for cold standby computing systems with periodic backup. IEEE Transactions on Computers, 64(4), 1043–1057.
- Levitin, G., Xing, L., & Dai, Y. (2014a). Minimum mission cost cold–standby sequencing in non-repairable multi-phase systems. IEEE Transactions on Reliability, 63(1), 251–258.
- Levitin, G., Xing, G. L., & Dai, Y. (2014b). Optimal component loading in 1-out-of-N cold standby systems. Reliability Engineering & System Safety, 127, 58–64.
- Liu, B., Cui, L., Wen, Y., & Guo, F. (2016). A cold standby repairable system with the repairman having multiple vacations and operational, repair, and vacation times following phase-type distributions. Communications in Statistics - Theory and Methods, 45(4), 850–858.
- Mahdy, M. (2009). On Some New Stochastic Orders and their Properties in The Statistical Reliability Theory (Ph.D. Dissertation). Benha University, Egypt.
- Mahdy, M. (2012). Characterization and preservations of the variance inactivity time ordering and the increasing variance inactivity time class. Journal of Advanced Research, 3(1), 29–34.
- Mahdy, M. (2016). Further results related to variance past lifetime class and associated orderings and their properties. Physica A: Statistical Mechanics and Its Applications, 462(15), 1148–1160.
- Mendes, A. A., Ribeiro, J. L. D., & Coit, D. W. (2017). Optimal time interval between periodic inspections for a two–component cold standby multistate system. IEEE Transactions on Reliability, 66(2), 559–574.
- Meng, X., Yuan, L., & Yin, R. (2006). The reliability analysis of a two–unit cold standby system with failable switch and maintenance equipment. IEEE International Conference on Computational Intelligence and Security, 2, 941–944.
- Müller, A., & Stoyan, D. (2002). Comparison Methods for Stochastic Models and Risks. Chichester: Wiley.
- Nair, N. U., & Sudheesh, K. K. (2010). Characterization of continuous distributions by properties of conditional variance. Statistical Methodology, 7(1), 30–40.
- Nadarajah, S., & Kotz, S. (2004). The beta gumbel distribution. Mathematical Problems in Engineering, 2004(4), 323–332.
- Nanda, A. K., & Shaked, M. (2001). The hazard rate and the reversed hazard rate orders, with applications to order statistics. Annals of the Institute of Statistical Mathematics, 53(4), 853–864.
- Nanda, A. K., Singh, H., Misra, N., & Paul, P. (2003). Reliability properties of reversed residual lifetime. Communications in Statistics - Theory and Methods, 32(10), 2031–2042.
- Navarro, J., del Águila, Y., Sordo, M. A., & Suárez-Llorens, A. (2013). Stochastic ordering properties for systems with dependent identically distributed components. Applied Stochastic Models in Business and Industry, 29(3), 264–278.
- Navarro, J., & Spizzichino, F. (2010). Comparisons of series and parallel systems with components sharing the same copula. Applied Stochastic Models in Business and Industry, 26(6), 775–791.
- Nelsen, R. B. (2006). An Introduction to Copulas (2nd ed.). New York: Springer.
- Rezaei, M., Gholizadeh, B., & Izadkhah, S. (2015). On relative reversed hazard rate order. Communications in Statistics – Theory and Methods, 44(2), 300–308.
- Sarhan, A. M., & Balakrishnan, N. (2007). A new class of bivariate distributions and its mixture. Journal of Multivariate Analysis, 98(7), 1508–1527.
- Shaked, M., & Shanthikumar, J. G. (2007). Stochastic Orders. New York: Springer.
- Tannous, O., Xing, L., Rui, P., Xie, M., & Ng, S. H. (2011). Redundancy allocation for series-parallel warm-standby systems. In Industrial Engineering and Engineering Management (IEEM), 2011 IEEE International Conference. IEEE, 1261–1265.
- Wang, D. Q., Lai, C. D., & Li, G. Y. (2003). Life distribution of series under the successive damage model. Journal of Systems Science and Complexity, 16(2), 191–194.
- Wu, Q., & Wu, S. (2011). Reliability analysis of two–unit cold standby repairable systems under poisson shocks. Applied Mathematics and Computation, 218(1), 171–182.
- Xing, L., Tannous, O., & Dugan, J. B. (2012). Reliability analysis of nonrepairable cold–standby systems using sequential binary decision diagrams. IEEE Transactions on Systems, Man, and Cybernetics – Part A: Systems and Humans, 42(3), 715–726.
- Yang, N., & Dhillon, B. S. (1995). Stochastic analysis of a general standby system with constant human error and arbitrary system repair rates. Microelectronics Reliability, 35(7), 1037–1045.
- Zhai, Q., Peng, R., Xing, L., & Yang, J. (2013). BDD-based reliability evaluation of k-out-of-(n + k) warm standby systems subject to fault-level coverage. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, 227(5), 540–548.