Abstract
In this paper, a retrial system with multiple working vacations is considered, and a mathematical model of the system is established by supplementary variable method, and a dynamic analysis of the system is carried out. Firstly, the model is transformed into an abstract Cauchy problem in Banach space by introducing the state space, main operator and its definition domain. Secondly, the -semigroup theory in functional analysis and the spectral theory of linear operators are used to prove that the main operator of the model generates a positive contraction -semigroup, which leads to the existence of a unique, non-negative time-dependent solution of the system that satisfies the probabilistic properties. Finally, Greiner’s boundary perturbation idea and the spectral properties of the corresponding operators are used to show that the time-dependent solution strongly converges to its steady-state solution.
Keywords:
k/n(G) retrial system; working vacation; C0-semigroup; asymptotic behavior; Dirichlet operator MSC:
47D03; 47A10; 90B25
1. Introduction
Reliability mathematics, a core branch of reliability theory, employs probability theory, mathematical statistics, and stochastic processes to investigate the quantitative laws governing system functionality. Its fundamental objective is to analyze failure mechanisms, evaluate life distributions, and predict reliability performance through mathematical modeling. In recent years, substantial research attention has been directed toward developing diverse methodological frameworks for reliability analysis, including Markov renewal processes, fault tree analysis, and the supplementary variable method (SVM) for non-Markovian systems [,]. The SVM, introduced to queuing theory by Cox [] and subsequently applied to reliability modeling by Gaver [], has proven particularly valuable for handling general repair time distributions. This technique enables the analysis of systems where repair times do not follow exponential distributions, expanding the applicability of reliability models to more realistic scenarios. Subsequent scholars have extensively utilized this method to investigate various repairable systems [,,,].
The concept of working vacation was first introduced into queuing systems in 2002 by Servi and Finn [], who studied the M/M/1 queuing system with working vacations. In this system, customers are served at a lower rate by servers during the vacation period instead of the service being completely stopped. For example, during off-peak hours, an escalator may run at a lower speed (working vacation) rather than stopping completely. When a passenger arrives, it resumes normal operation. This concept has been extended to various configurations including GI/M/1 [], M/G/1/WV [] and M//1 [] systems. Recent extensions incorporate multiple server breakdowns [], N-policy with server breakdowns [], inventory management considerations [] and server breakdowns with working vacations []. The working vacation strategy has been successfully adapted to repairable systems, offering significant practical advantages. Wang et al. [] investigated machine maintenance problems using working vacation policies, determining optimal maintenance rates through Newton’s method. Liu et al. [] analyzed cold standby systems with multiple working vacations using matrix analysis, while Deora et al. [] proposed optimization methods for dynamic maintenance strategies. Recent work by Wu et al. [] conducted sensitivity analysis of systems with multiple failure modes and working vacations.
Retrial queueing models, originating from telephone exchange systems [], have been systematically developed by Falin and Templeton [], Artalejo and Gómez–Corral []. These models have found applications in voting systems, which are critically important in intelligent transportation, energy systems, medical devices, and communication infrastructure. Krishnamoorthy and Ushakumari [] investigated voting systems with retrial strategies, while Wu et al. [] proposed models combining multiple vacations and replaceable equipment. Subsequent research has extensively developed this area [,,,,]. Recent research has focused on integrating working vacation and retrial strategies to address dynamic optimization requirements in complex engineering scenarios. Li et al. [] constructed the first M/M/1 queue with both retrial and working vacation strategies under feedback mechanisms. Yang and Tsao [] analyzed standby systems with working vacations and component retrial, while Do et al. [] derived steady-state solutions for systems with fixed retrial rates. Kumar et al. [] recently investigated fault-tolerant systems with double trial characteristics and multiple working vacations.
However, a significant theoretical gap persists. Most studies employing the supplementary variable method describe their models through partial differential-integral equations and assume that “time-dependent solutions exist, are unique, and converge to steady-state solutions” without rigorous mathematical verification. This issue remained largely unaddressed until Gupur [] applied -semigroup theory to prove well-posedness for a parallel repairable system with a general repair rate function. Subsequent work [,,,,,,] has extended this approach to various reliability models, establishing existence, uniqueness, and asymptotic stability of the time-dependent solutions.
Although reliability models incorporating multiple working vacations and retrial strategies have been explored in previous studies, their application to a k/n(G) voting repairable system remains unaddressed. Building on our pevious work [], which analysed a three-component standby system with these features, the present study significantly extends the scope. Compared to [], we introduce a multiple working vacations strategy, assume a general distribution for repair times, and posit that non-failed components remain intact after system failure. Relative to [], our model incorporates both working vacations and a retrial strategy, assumes reliable repair equipment, and adopts general repair time distributions.
This paper makes three key contributions: First, we develop a novel mathematical model for the k/n(G) repairable system using the SVM, formulating it as an abstract Cauchy problem on a Banach space. Second, through -semigroup theory and Gupur’s methods [], we prove the existence of a unique, non-negative time-dependent solution that satisfies probability conditions. Finally, by applying Greiner’s boundary perturbation theory [] and spectral analysis of the associated operators, we demonstrate the strong convergence of time-dependent solutions to steady-state solutions, thereby providing rigorous mathematical justification for models in this domain.
2. System Description and Assumptions
2.1. Assumptions and Descriptions of the Model
The mathematical model satisfies the following assumptions.
- (1)
- This system consists of n components and a repairer.
- (2)
- The system works when components out of n or more than k components are operating normally; it fails when the number of failed components is greater than or equal to .
- (3)
- During a system failure, normal components stop working and no new failures occur until the component under repair is restored. Thereafter, all k working components begin operating at the same time, and the system restarts.
- (4)
- When the repairer is idle, and a component failure, then it is immediately repaired. if the repairer is busy when a component failure, the failed component enters the retrial orbit and attempts to request a repair again after a period of time. This process continues until the repair is successful.
- (5)
- The repairer in this system follows a policy of multiple working vacations. Specifically, the repairer can enter consecutively working vacation states. During these vacation states, the repairer continues to perform repairs on failed components, with a reduced rate compared to normal working conditions. Upon the completion of each working vacation, if no component failures are present, a new working vacation can be initiated immediately without waiting for new repair tasks. Furthermore, working vacations may be triggered under certain system states, such as when some components have failed but the system remains operational.
- (6)
- When the system is operational and no component fails or the failed component in the orbit is not retried, the repairer takes a round of working vacation. If a component fails or a failed component in the orbit is retried during a working vacation, the repairer will perform the repairs at a reduced rate.
- (7)
- If the system is in a failed state upon completion of a working vacation, the repairer immediately begins to repair the failed components. Once the repair is completed and the system is operational again, the repairer starts a new working vacation.
- (8)
- At the completion of a working vacation, the repairer initiates a new working vacation immediately if the system is operational with no failed components and no active retrials from the orbit. However, if a component failure is detected or a retrial occurs at this point, the repairer first performs the repair before starting the next vacation. This cycle of consecutive vacations continues until a failed component is detected upon the completion of a vacation.
- (9)
- The life distribution of each component, the repairer’s vacation time and the retrial time of the failed component in the orbit are assumed to follow an exponential distribution with parameters and , respectively. Furthermore, each failed component in the orbit has an equal probability of being selected for retrial.
- (10)
- Let denote the conditional probability that a failed component is repaired in the interval during a working vacation, given that it has not been repaired by time x. denotes the conditional probability that a failed component is repaired during the normal working period in the interval , given that it has not been repaired by time x. Let the repair time follow a general distribution with probability density function (PDF) . Sincethen, from the property of the conditional probability and , we haveConsequently, we have
- (11)
- All random variables are independent and any repaired component has a lifetime distribution identical to that of a new one.
Let denotes that, at time t, there are failed components in the orbit, and be a random variable representing the state of the system. According to the previous assumptions, this system has the following possible states:
- : At time t, the system is operational with failed components in the orbit. The repairer is on a working vacation and is currently idle.
- : At time t, the system is operational with failed components in the orbit. The repairer is on a working vacation and is currently busy.
- : At time t, the system is down with failed components in the orbit. The repairer is on a working vacation and is busy.
- : At time t, the system is operational with failed components in the orbit. The repairer is in a normal working period and is currently idle.
- : At time t, the system is operational with failed components in the orbit. The repairer is in a normal working period and is currently busy.
- : At time t, the system is down with failed components in the orbit. The repairer is in a normal working period and is currently busy.
The assumption of a general repair time distribution implies that the process does not possess the Markov property in continuous time. This limitation is resolved by introducing the supplementary variable , representing the elapsed repair time. Consequently, the combined process becomes Markovian (see [,]).
Set
The supplementary variable technique yields the following system of partial differential equations describing the system:
with the boundary and the initial conditions:
where
2.2. Reset the Model
Take the following Banach space:
as a state space.
The maximal operator with its domain is defined as:
where
Take the following boundary space and operators:
and
where
If we define the operator by
3. Well-Posedness of the System (17)
In this section, we first prove that generates a positively contaction -semigroup in X, and then we determine the conjugate space of X to show that is a conservative operator. is a conservative operator, and then apply Fattorini’s theorem [] to obtain that is an equidistant operator, which leads to the Well-posedness of the system (17).
Theorem 1.
Let then generates a positive contraction –semigroup .
Proof.
The proof proceeds in the following four steps:
Step 1. Establish the resolvent estimate for .
Step 2. Verify .
Step 3. Demonstrate that generates a -semigroup via the Hille-Yosida theorem, and confirm that and are bounded linear operators. Applying the perturbation theorem then allows us to conclude that also generates a -semigroup .
Step 4. Prove that is a dispersive operator, which implies that is a positive contraction semigroup.
The subsequent detailed calculations and technical estimates follow standard arguments analogous to those in [] and are therefore omitted for brevity. □
We can easily determine that , the dual space of X, is as follows:
Obviously, is a Banach space. In X we define the subset
Then Theorem 1 implies that For we take
then and
Which shows that is conservative for the set
Since we have the following result by the Fattorini theorem [].
Theorem 2.
is isometric operator for , i.e.,
From above theorems we have the conclusion of this section.
Theorem 3.
4. Asymptotic Behavior of the Time–Dependent Solution of the System (17)
Since the asymptotic behaviour of the time-dependent solution of the system (17) is determined by the spectral distribution of the main operator on the imaginary axis, this section aims to characterize this behaviour through a spectral analysis of the operator. A key challenge in this analysis is posed by the system’s boundary conditions. To address this, we adopt Greiner’s boundary perturbation technique [], originally introduced in the context of population equations, to examine the spectrum of the operator associated with this system. Using this approach, we first identify the regular points of the main operator along the imaginary axis. Specifically, we show that 0 is an eigenvalue of the main operator with geometric multiplicity one. We then demonstrate that all nonzero points on the imaginary axis belong to the resolvent set of the operator. Furthermore, the adjoint operator of the main operator is derived, and it is verified that 0 is also an eigenvalue of this adjoint operator. Finally, by combining these spectral results with Theorem 1.96 in [], we establish the strong convergence of the time-dependent solution of the system (17) to its steady-state solution.
If we define as
then for any given , consider equation , i.e.,
From the definition of the resolvent set, we can obtain the following result.
Lemma 1.
Let are measurable and
Then
which indicates that contains all points on the imaginary axis.
Proof.
For any
we estimate
Since is dense in by Adams [], Equations (48)–(51) hold for all .
Which completes the proof. □
Lemma 2.
Let are measurable and
If
then
Proof.
If , then which is equivalent to
By the imbedding theorem [] with , , it implies
Similarly, we estimate
The above formulas show that □
Observe that the operator is surjective. Furthermore, for any ,
is invertible. If for any , then the Dirichlet operator can be defined as
Lemma 2 shows the explicit form of for ;
where
By (98) and the expression of we have
Lemma 3.
Let and . Then and GM(0) = 1, i.e., geometric multiplicity is 1.
Proof.
Consider the equation
Furthermore, due to
where
Lemma 4.
Let are measurable and
Then contains all points on the imaginary axis except 0.
Proof.
If and From the Riemann-Lebesgue lemma
There exists for ,
which applying the fact and
For , we estimate
This shows
Equation (141) implies that as , and from that together with [] (Characteristic Equation (3.5)), we know that for i.e.,
According to Theorem 1, -semigroup is a positive contraction. Additionally, ref. [] (Theorem 1.12) shows that is imaginary additively cyclic, consequently for any positive integer k, we have
From this equation together with (142), we obtain that □
Finally, we present the adjoint operator of , which is , and then prove that and GM(0) = 1.
Lemma 5.
is as follows:
where χ in is a constant which is irrelevant to .
Lemma 6.
Let and . Then and GM(0) = 1.
Proof.
Consider the equation
Then take the limit at we have
Substituting (171), and (172) into (165)–(166), (168)–(169), respectively, and then combining it with (152) and we deduce
Combining (4.133) and (4.154), we have
By combining Theorem 1, Lemmas 3, 4 and 6 with [] (Theorem 1.96), we obtain the main result.
Theorem 4.
Let are measurable and
Then the time–dependent solution of the system (2.13) strongly converges to its steady-state solution, that is,
here P is an eigenvector corresponding to 0 in Lemma 3 and ϑ is determined by an eigenvector and the initial value in Lemma 6.
5. Conclusions
In this paper, a k/n(G) voting repairable system with retrial strategy and multiple working vacations strategy is developed by using the SVM. The reliability model of the system is first transformed into an abstract Cauchy problem on a chosen Banach space by defining the system’s main operator and its domain. By applying the -semigroup theory from functional analysis and spectral theory of operators, we prove that the main operator generates a positive, conttaction -semigroup, which ensures the existence of a unique, non-negative time-dependent solution consistent with probability constraints. Furthermore, we demonstrate that zero is an eigenvalue of both the main operator and its adjoint, each with geometric multiplicity one. Using Greiner’s boundary perturbation approach, we characterize the spectrum of the main operator on the imaginary axis. These results collectively establish the strong convergence of the time-dependent solution to the steady-state solution.
A key direction for future research will be a detailed analysis of the exponential stability and asymptotic expressions of the time-dependent solution.
Author Contributions
Conceptualization, C.L., R.E. and E.K.; methodology, C.L. and R.E; validation, C.L., R.E. and E.K.; writing—original draft preparation, C.L.; writing—review and editing, R.E. and E.K.; funding acquisition, R.E. All authors have read and agreed to the published version of the manuscript.
Funding
This research was funded by the Natural Science Foundation of Xinjiang Uygur Autonomous Region, 2025D01C40, and the Natural Science Foundation of Xinjiang University.
Data Availability Statement
Data are contained within the article.
Acknowledgments
The authors would like to thank the editor and referees for their valuable comments.
Conflicts of Interest
The authors declare no conflicts of interest.
References
- Cao, J.; Cheng, K. Introduction to Reliability Mathematics; Higher Education Press: Beijing, China, 2012. [Google Scholar]
- Gupur, G. Mathematical Methods in Reliability Theory; Science Press: Beijing, China, 2020. [Google Scholar]
- Cox, D.R. The analysis of non-Markovian stochastic processes by the inclusion of supplementary variables. Math. Proc. Camb. Philos. Soc. 1955, 51, 433–441. [Google Scholar] [CrossRef]
- Gaver, D.P. Time to failure and availability of paralleled systems with repair. IEEE Trans. Reliab. 1963, R-12, 30–38. [Google Scholar] [CrossRef]
- Linton, D.G. Some advancements in the analysis of two-unit parallel redundant systems. Microelectron. Reliab. 1976, 15, 39–46. [Google Scholar] [CrossRef]
- Adachi, K.; Kodama, M.; Ohashi, M. k-out-of-n: G system with simultaneous failure and three repair policies. Microelectron. Reliab. 1979, 19, 351–361. [Google Scholar] [CrossRef]
- Cheng, S.; Dhillon, B. Reliability and availability analysis of a robot-safety system. J. Qual. Maint. Eng. 2011, 17, 203–232. [Google Scholar] [CrossRef]
- Yuan, L.; Xu, J. A deteriorating system with its repairman having multiple vacations. Appl. Math. Comput. 2011, 217, 4980–4989. [Google Scholar] [CrossRef]
- Servi, L.D.; Finn, S.G. M/M/1 queues with working vacations s (M/M/WV). Perform. Eval. 2002, 50, 41–52. [Google Scholar]
- Baba, Y. Analysis of a GI/M/1 queue with multiple working vacations. Oper. Res. Lett. 2005, 33, 201–209. [Google Scholar] [CrossRef]
- Wu, D.A.; Takagi, H. M/G/1 queue with multiple working vacations. Perform. Eval. 2006, 63, 654–681. [Google Scholar] [CrossRef]
- Jain, M.; Agrawal, P.K. M/Ek/1 queueing system with working vacation. Qual. Technol. Quant. Manag. 2007, 4, 455–470. [Google Scholar] [CrossRef]
- Jain, M.; Jain, A. Working vacations queueing model with multiple types of server breakdowns. Appl. Math. Model. 2010, 34, 1–13. [Google Scholar] [CrossRef]
- Yang, D.Y.; Wu, C.H. Cost-minimization analysis of a working vacation queue with N-policy and server breakdowns. Comput. Ind. Eng. 2015, 82, 151–158. [Google Scholar] [CrossRef]
- Jeganathan, K.; Reiyas, M.A. Two parallel heterogeneous servers markovian inventory system with modified and delayed working vacations. Math. Comput. Simul. 2020, 172, 273–304. [Google Scholar] [CrossRef]
- Yang, D.Y.; Chung, C.H.; Wu, C.H. Sojourn times in a markovian queue with working breakdowns and delayed working vacations. Comput. Ind. Eng. 2021, 156, 107239. [Google Scholar] [CrossRef]
- Wang, K.H.; Chen, W.L.; Yang, D.Y. Optimal management of the machine repair problem with working vacation: Newton’s method. J. Comput. Appl. Math. 2009, 233, 449–458. [Google Scholar] [CrossRef]
- Liu, B.; Cui, L.; Wen, Y.; Shen, J. A cold standby repairable system with working vacations and vacation interruption following markovian arrival process. Reliab. Eng. Syst. Saf. 2015, 142, 1–8. [Google Scholar] [CrossRef]
- Deora, P.; Kumari, U.; Sharma, D. Cost analysis and optimization of machine repair model with working vacation and feedback–policy. Int. J. Appl. Comput. Math. 2021, 7, 1–14. [Google Scholar] [CrossRef]
- Wu, C.H.; Yang, D.Y.; Ko, M.H. Performance sensitivity analysis for machine repair problem with two failure modes and working vacation. Int. J. Reliab. Qual. Saf. Eng. 2024, 31, 2350040. [Google Scholar] [CrossRef]
- Fayolle, G. A simple telephone exchange with delayed feedbacks. In Proceedings of the International Seminar on Teletraffic Analysis and Computer Performance Evaluation, Amsterdam, The Netherlands, 2–6 June 1986; pp. 245–253. [Google Scholar]
- Falin, G.; Templeton, J.G. Retrial Queues; Chapman and Hall: London, UK, 1997. [Google Scholar]
- Artalejo, J.R.; Gómez-Corral, A.A. Retrial Queueing Systems. Math. Comput. Model. 1999, 30, 13–15. [Google Scholar]
- Krishnamoorthy, A.; Ushakumari, P.V. Reliability of a k-out-of-n system with repair and retrial of failed units. Top 1999, 7, 293–304. [Google Scholar] [CrossRef]
- Wu, W.; Tang, Y. A Study of the k/n(G) Voting Repairable System with Repairers on Multiple Leave and Repairable Equipment Replacement. Syst. Eng.-Theory Pract. 2013, 33, 2604–2614. [Google Scholar]
- Gao, S.; Wang, J. Reliability and availability analysis of a retrial system with mixed standbys and an unreliable repair facility. Reliab. Eng. Syst. Saf. 2021, 205, 107240. [Google Scholar] [CrossRef]
- Wang, Y.; Hu, L.; Yang, L.; Li, J. Reliability modeling and analysis for linear consecutive-k-out-of-n: F retrial systems with two maintenance activities. Reliab. Eng. Syst. Saf. 2022, 226, 108665. [Google Scholar]
- Kumar, S.; Gupta, R. Working vacation policy for load sharing K-out-of-N: G system. J. Reliab. Stat. Stud. 2022, 15, 583–616. [Google Scholar] [CrossRef]
- Hu, L.; Liu, S.; Peng, R.; Liu, Z. Reliability and sensitivity analysis of a repairable k-out-of-n: G system with two failure modes and retrial feature. Commun. Stat.-Theory Methods 2022, 51, 3043–3064. [Google Scholar] [CrossRef]
- Yu, X.; Hu, L.; Ma, M. Reliability measures of discrete time k-out-of-n: G retrial systems based on Bernoulli shocks. Reliab. Eng. Syst. Saf. 2023, 239, 109491. [Google Scholar] [CrossRef]
- Li, J.T.; Li, T.; An, M. An M/M/1 retrial queue with working vacation, orbit search and balking. Eng. Lett. 2019, 27, 97–102. [Google Scholar]
- Yang, D.Y.; Tsao, C.L. Reliability and availability analysis of standby systems with working vacations and retrial of failed components. Reliab. Eng. Syst. Saf. 2019, 182, 46–55. [Google Scholar] [CrossRef]
- Do, N.H.; Do, T.V.; Melikov, A. Equilibrium customer behavior in the M/M/1 retrial queue with working vacations and a constant retrial rate. Oper. Res. 2020, 20, 627–646. [Google Scholar]
- Kumar, P.; Jain, M.; Meena, R.K. Transient analysis and reliability modeling of fault–tolerant system operating under admission control policy with double retrial features and working vacation. ISA Trans. 2023, 134, 183–199. [Google Scholar] [CrossRef]
- Gupur, G. Well–posedness of the system consisting of two repairable units. Acta Anal. Funct. Appl. 2001, 3, 188–192. [Google Scholar]
- Gupur, G. Well-posedness of a reliability model. Acta Anal. Funct. Appl. 2003, 5, 93–209. [Google Scholar]
- Gupur, G. Asymptotic stability of the time-dependent solution of a reliability model. Acta Anal. Funct. Appl. 2005, 7, 299–316. [Google Scholar]
- Haji, A.; Radl, A. A semigroup approach to the Gnedenko system with single vacation of a repairman. Semigroup Forum 2013, 86, 41–58. [Google Scholar]
- Habil, E.B. Asymptotic behavior of repairable device systems under warranty and beyond warranty periods. Positivity 2019, 23, 875–889. [Google Scholar] [CrossRef]
- Kasim, E.; Gupur, G. Dynamic analysis of a complex system under preemptive repeat repair discipline. Bound. Value Probl. 2020, 2020, 71. [Google Scholar] [CrossRef]
- Yumaier, A.; Kasim, E. Dynamic Analysis of the Multi–state Reliability System with Priority Repair Discipline. Acta Math. Appl. Sin. Engl. Ser. 2024, 40, 665–694. [Google Scholar]
- Li, Y.; Xu, G.; Wang, Y. Reliability Analysis and Numerical Simulation of Industrial Robot Drive System with Vacation. Axioms 2025, 14, 275. [Google Scholar] [CrossRef]
- Lai, C.; Kasim, E.; Muhammadhaji, A. Dynamic Analysis of a Standby System with Retrial Strategies and Multiple Working Vacations. Mathematics 2024, 12, 3999. [Google Scholar] [CrossRef]
- Gupur, G. Functional Analysis Methods for Reliability Models; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2011. [Google Scholar]
- Greiner, G. Perturbing the boundary–conditions of a generator. Houst. J. Math. 1987, 13, 213–229. [Google Scholar]
- Fattorini, H.O. The Cauchy Problem; Cambridge University Press: Cambridge, UK, 1983. [Google Scholar]
- Adams, R.A.; Fournier, J.J. Sobolev Spaces; Elsevier: Amsterdam, The Netherlands, 2003. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).