Received 29 February 2016; accepted 18 July 2016; published 21 July 2016
The reliability analysis is an essential practice for the installations where failure may turn out hazardous either in terms of huge financial loss or threat to human life. These causes inspired the literature to a greater extent  -  . Further, with the pioneer works of Smith  and Pyke  , the use of semi-Markov regenerative processes became popular for developing reliability models of probabilistic systems  -  . To assure stable system performance both in terms of reliability as well as availability, the proviso of standby redundancy is widely considered in the literature  -  . The repairable systems are described with the feature that as soon as any component/unit fails it is either repaired or replaced by the service facility. So the service facility plays a key role in keeping a repairable system operational for longer period of time. In such cases, the failure of service facility interrupts the system performance in terms of availability, reliability and profit  -  . So the modeling of arrival and treatment times of server becomes essentially important to marginalize the loss due to system down time. Keeping these facts in view, the present paper investigates a cold-standby system taking account of waiting time for arrival and treatment of server subject to failure. The semi-Markov processes and regenerative point technique are used to obtain following measures of system performance in steady state:
a) Transition probabilities and mean sojourn times in different states.
b) MTSF and reliability of the system.
c) System availability.
d) Server busy period.
e) Expected number of repairs and treatments.
f) And expected profit.
2. System Assumptions & States Description
To provide ease to the computational work, the model is developed using the following set of assumptions:
a) The model consists of two identical units. Initially, one unit is in operation and another as cold-standby.
b) The unit in standby mode can’t fail.
c) Upon failure of the operative unit the standby becomes operative instantly.
d) All the failures are repairable to be repaired by the server but the server takes some time to arrive.
e) The server may fail while working but curable.
f) The server restoration subjects to treatment with some elapsed time.
g) All the repairs, treatments and switching are perfect.
h) The system works as long as at least one unit remains working.
i) All the random variables are assumed to be statistically independent.
j) All the random variables follow general probability distribution with different distribution functions.
2.2. States of the System
The system model comprises of regenerative and non-regenerative states. The states are regenerative whereas the states are non-regenerative. The detailed description of all possible states is as follows:
3. Notations & Acronyms
: pdf/cdf of failure time of the unit.
: pdf/cdfof failure time of the server.
: pdf/cdf of repair time of the failed unit.
: pdf/cdf of the treatment time of the server.
: pdf/cdf of the waiting time of the server for treatment.
: pdf/cdf of the arrival time of the server.
: pdf/cdf of direct transition time from a regenerative state i to a regenerative state j without visiting any other regenerative state.
: pdf/cdf of first passage time from a regenerative state i to a regenerative state j or to a failed state j visiting state k once in (0, t].
: pdf/cdf of first passage time from regenerative state i to a regenerative state j or to a failed state j visiting state k, r once in (0, t].
: pdf/cdf of first passage time from regenerative state i to a regenerative state j or to a failed state j visiting state k, r and s once in (0, t].
: Probability that the server is busy in the state Si up to time ‘t’ without making any transition to any other regenerative state or returning to the same state via one or more non-regenerative states.
: Contribution to mean sojourn time (µi) in state Si when system transit directly to state j.
: Contribution to mean sojourn time (µi) in state Si when system transit to state j via k and n times between
(s)/(c): Stieltjes convolution/Laplace convolution.
: Laplace Stieltjes Transform (LST).
Laplace Transform (LT).
: Inverse Laplace Transform.
4. The Model Development
4.1. The State Transition Diagram
Taking account of all possible transitions and the re-generative points a system schematic state transition diagram is constructed as given in Figure 1. The solid dots denote the regenerative epochs for various states of the model. The probability density functions of various random variables are also shown in Figure 1.
4.2. State Transition Probabilities
Simple probabilistic considerations, yields the following expressions for the non-zero elements
Figure 1. The schematic system state transition diagram.
For these Transition Probabilities, it can be verified that
4.3. Mean Sojourn Times
The Mean sojourn time µi in state Si are given by:
The unconditional mean time taken by the system to transit from any state Si when time is counted from epoch at entrance into state Sj is stated as:
5. Stochastic Analysis
5.1. Reliability Measure
Let be the c.d.f of the first passage time from regenerative state Si to a failed state. Regarding the failed state as absorbing state, we have the following recursive relations for:
where Sj is an un-failed regenerative state to which the given regenerative state Si can transit and Sk is failed state to which the state Si can transit directly.
Taking LST of Equation (3) and solving for, we get MTSF as follows:
The reliability R(t) is given by
5.2. Economic Measures
Let the system entered the regenerative state Si at t = 0. Considering Sj as a regenerative state to which the given regenerative state Si transits, the recursive relations for various profit measures in (0, t] are given as follow:
Using LT/ LST, of Equations (4)-(7) and solving we get the results in steady state as below:
Further, using the values of above performance measures, the profit incurred to the system model in steady state is given as below.
6. Example (Particular Case of Exponential Distribution)
For the sake of convenience, let us suppose all the random variables follow the exponential distribution with the probability density functions given below.
We assume some particular values for various time rates and costs i.e.
Failure rate of server (g) = 0.02 per unit time, Failure rate of unit (λ) = 0.008 per unit time.
Repair rate of unit (α) = 0.3 per unit time, Treatment rate of server (β) = 0.05 per unit time.
Server arrival rate (y) = 0.08 per unit time, Waiting treatment time (x) = 0.08 per unit time.
For this, we obtained the values for different measures of system performance as follows:
MTSF = 1110.909 unit time, Availability = 0.977238, Busy period of server = 0.023716,
Expected number of repairs = 0.000918, Expected number of treatments =0.000364 and
System profit = 19532.31.
The detailed results are given in tabular form. Here Tables 1-3 respectively, illustrate the effect of server treatment rate for various combinations of parameters on Mean Time to System Failure (MTSF), availability and profit assuming.
Table 1. Effect of various parameters on MTSF.
Table 2. Effect of various parameters on system availability.
Table 3. Effect of various parameters on system profit.
7. Discussion and Concluding Remark
A stochastic model for a repairable cold standby system, with waiting arrival and treatment times of server, is discussed in this paper. The theory of semi-Markov process and regenerative point technique is used to derive expressions for measures of reliability and profit. An example is given under the setup of exponential distribution by assigning distinct values to various parameters and costs considered for the system model. The further numerical results (as given in Tables 1-3) indicate that MTSF, availability and the profit rise with increasing server treatment rate (β), repair rate of units (α) and server arrival time (ψ) but the trend reverts with increasing the failure rates of server (g) and unit (λ). The persisting trend reveals that the waiting arrival and treatment times impact a lot on the system performance. Therefore, the study re-iterates the practicalities that a cold standby system served by a repairable server can be kept reliable and profitable by:
1) Using standard units with low failure rates.
2) Deploying efficient server with high repair rates.
3) Planning for higher arrival rate of server and,
4) Arranging rapid after failure treatment of the server.
The re-iteration of the true facts evidently proves the acceptability of the probabilistic model developed in this paper. The study may be inspiring and useful for system planners and reliability engineers for developing highly reliable and profitable systems to earn users’ satisfaction.
The study finds its application in diverse areas such as power generating systems with standby reservoirs, communication systems with redundant channels, remote sensing systems with alternate power backups etc.
This work is a part of Major Research Project F. N. 42-34/2013(SR) financially supported by UGC under MHRD, Govt. of India. The authors are grateful to anonymous referee for their valuable comments and suggestions.