534

IEEE TRANSACTIONS ON RELIABILITY, VOL. 56, NO. 3, SEPTEMBER 2007

Reliability and Prot Evaluation of a PLC Hot Standby System Based on a Master-Slave Concept and Two Types of Repair Facilities
Bhupender Parashar and Gulshan Taneja
AbstractProgrammable Logic Controllers (PLC) are frequently used by a good number of companies like steel plants, biscuit manufacturing companies, etc. Various plants/companies use two PLC at a time: one operative, and the other as a hot standby to avoid big losses. Analysis of the reliability, and prot of a hot standby PLC system is of great importance; and hence the present paper examines such a system wherein two PLC are working in master-slave fashion. Initially, the master unit is operative, and the slave unit is in hot standby. The slave unit can also fail, but with a lower failure rate than the master unit. The master unit has the priority of operation & repair over the slave unit. While operating, the latest information from the master unit keeps on transferring to the slave unit. There are three types of failure: minor, major-repairable, and major-irreparable. The ordinary repairman who stays with the system repairs the minor failures. The expert repairman who is available upon demand repairs the major failures. Various measures of the system effectiveness, such as the mean time to system failure, steady-state availability, busy period of the ordinary as well as expert repairmen, expected number of replacements, etc. are obtained by using semi-Markov processes, and regenerative point Techniques. Prot incurred to the system is evaluated, and a graphical study is also made. Real data from an industrial application is used in this study. Index TermsMaster-slave units, programmable logic controller (PLC), regenerative process, semi-Markov process.

ACRONYM1 PLC INR FCFS MTSF Programmable Logic Controller Indian National Rupee First Come First Served Mean Time to System Failure NOMENCLATURE the master unit is operative the slave unit is operative the slave unit is in hot standby constant failure rate of the master unit constant failure rate of the slave unit
Manuscript received September 17, 2006; revised November 11, 2006; accepted December 4, 2006. Associate Editor: L. Cui. B. Parashar is with the Department of Applied Mathematics, JSS Academy of Technical Education, Noida 201301, Uttar Pradesh, India (e-mail: parashar_b@rediffmail.com). G. Taneja is with Department of Statistics, University College, Rohtak 124001, Haryana, India (e-mail : drgtaneja@yahoo.com). Digital Object Identier 10.1109/TR.2007.903151
1The

g(t), G(t) , h(t), H(t) w(t), W(t)

singular and plural of an acronym are always spelled the same.

the master unit is under repair in case of minor failure the continuation of repair of the master unit from the previous state in the case of a minor failure the master unit is under repair by the expert repairman in the case of a major failure repair of the master unit by the expert repairman is continuing from the previous state the master unit is under replacement in the case of a major failure replacement of the master unit is continuing from the previous state a failed master unit is waiting for repair from the ordinary repairman a failed master unit is waiting for repair from the expert repairman a failed master unit is waiting for replacement from the expert repairman the slave unit is under repair in the case of a minor failure the continuation of repair of the slave unit from the previous state in the case of a minor failure the slave unit is under repair by the expert repairman in the case of a major failure the repair of a slave unit by the expert repairman is continuing from the previous state the slave unit is under replacement in the case of a major failure the replacement of a slave unit is continuing from the previous state a failed slave unit is waiting for repair from the ordinary repairman a failed slave unit is waiting for repair from the expert repairman a failed slave unit is waiting for replacement from the expert repairman p.d.f., and c.d.f. of the repair time of a unit having a minor failure p.d.f., and c.d.f. of the repair time of a unit having a major failure p.d.f., and c.d.f. of the replacement time of a unit having a major failure p.d.f., and c.d.f. of the waiting time symbol for Laplace convolution revenue per unit up time cost of the ordinary repairman

0018-9529/$25.00 2007 IEEE

PARASHAR AND TANEJA: RELIABILITY AND PROFIT EVALUATION OF PLC HOT STANDBY SYSTEM

535

Fig. 1. State transition diagram.

MTSF

cost per unit up time for which the expert repairman is busy for repair cost per unit up time for which the expert repairman is busy for replacement cost per visit of the repairman cost per unit replacement this measure is dened as the expected time for which the system is in operation before it completely fails. steady-state availabilitythis is the probability that the system will be able to operate within the tolerances for a specied time period. busy period of the expert repairman for repair; the total fraction of time for which the system is under repair by the expert repairman busy period of the expert repairman for replacement; the total fraction of time for which the system is being replaced by the expert repairman expected number of visits of the expert repairman; it represents the number of times the expert repairman has visited the system in steady-state expected number of replacements; represents the average number of replacements in steady-state

I. INTRODUCTION ELIABILITY models for systems have widely been discussed under various assumptions by a number of researchers. Investigations for evaluating various measures of system effectiveness for such systems have been carried out by various authors [1][7]. Cost-benet analysis, and graphical study have also been done by a good number of researchers [8][11]. For graphical study, they have taken assumed values for failure and repair rates, and not used the observed values. Taneja [12] analyzed the reliability, and prot of a system with a PLC used in hot standby conguration, collecting real data on failure and repair rates of PLC wherein FCFS pattern is followed for the repair of the failed units, and also for the operation of the repaired units. There may also be the systems wherein the priority for repair/operation is given to one unit over the other. Such systems include a hot standby PLC system with a master-slave conguration. The master unit has a priority for repair/operation over the slave unit. Keeping the above in view, the present study is an attempt to collect real data on failure, repair, and replacement rates, the cost of replacement, and probabilities of failure for 126 PLC; and analyze a model for evaluating the reliability, and the prot of a PLC hot standby system in a master-slave conguration, with two types of repair facilities. The situation considered in the present study is the same as that from which these data have

536

IEEE TRANSACTIONS ON RELIABILITY, VOL. 56, NO. 3, SEPTEMBER 2007

been collected. Using these data, various values of rates, costs, and probabilities are obtained as: Probability of Minor Failure, Probability of Major Failure (repairable), Probability of Major Failure (irreparable), Estimated value of failure rate Estimated repair rate (minor), Estimated repair rate (major), Estimated replacement rate (major), The system is analyzed by making use of semi-Markov processes, and regenerative point techniques; and the following measures of the system effectiveness are obtained: The prot incurred to the system is evaluated, and a graphical study is also done. II. MODEL DESCRIPTION AND ASSUMPTIONS 1) Initially the master unit is operative, and the slave is in hot standby. 2) The master and slave units can talk to each other, and the latest information is transferred from the operative unit to the hot standby unit in real time. 3) Upon failure of the master unit, the slave unit becomes operative immediately. 4) The slave unit while in standby may also fail, but its failure rate is lower than that of the master unit.

5) There are three types of failures: minor, major-repairable, and major-irreparable. 6) Failure times are assumed to follow an exponential distribution, whereas the other times follow arbitrary distributions. 7) PLC may fail due to the following reasons: power supply failure/software corruption (minor) component failure (major-repairable) the unit is burnt (major-irreparable) 8) The minor failures are to be repaired by the ordinary repairman, while the major-repairable failures, and majorirreparable failures are to be repaired by the expert repairman. The ordinary repairman stays with the system, while the expert repairman is available on demand. 9) Priority for repair is given to the master unit. 10) Priority for operation is also given to the master unit. 11) After each repair, the system works as good as a new one. III. TRANSITION PROBABILITIES AND MEAN SOJOURN TIMES A transition diagram showing the various states of transition of a system is shown in Fig. 1. The epochs of entry into states 0, 1, 2, 3, 4, 5, 6, and 16 are regenerative points, and thus these states are regenerative states. The transition probabilities are can given at the bottom of the page. The non-zero elements be obtained as

PARASHAR AND TANEJA: RELIABILITY AND PROFIT EVALUATION OF PLC HOT STANDBY SYSTEM

537

The mean sojourn times in the regenerative states (where , 1, 2, 3, 4, 5, 6, 16) are given by

The unconditional mean time taken by the system to transit to any regenerative state when time is counted from the epoch of entrance into state is mathematically stated as

Thus,

Fig. 2. MTSF vs. failure rate ().

IV. PROFIT ANALYSIS Pulling these elements together, we dene prot as

V. PARTICULAR CASE

Using the values estimated from the data collected (i.e., , , , , , , , , , , , , , ) the following values of various measures of system effectiveness are obtained: Mean time to system failure Availability Busy period of ordinary repairman for repair Busy period of expert repairman for repair Busy period of expert repairman for replacement Regarding the failed states as absorbing states, and employing the arguments used for regenerative processes, we obtained the recursive relations for the mean time to system failure, the availability, the busy period for an ordinary repairman, the busy periods for the expert repairman for the repair and replacement separately, the expected number of visits of the expert repairman, and the expected number of replacements. Then, taking Laplace/ Laplace Stieltjs Transforms of these recursive relations, we obtain steady-state solutions for the above mentioned measures of system effectiveness. Expected no. of visits per unit time Expected no. of replacements VI. GRAPHICAL INTERPRETATION Fig. 2 represents the behavior of the Mean Time to System . It can be Failure (MTSF) with respect to the failure rate concluded from the graph that the MTSF decreases as the failure increases. rate Fig. 3 represents the behavior of the availability with respect . It is obvious from the graph that system to the failure rate

538

IEEE TRANSACTIONS ON RELIABILITY, VOL. 56, NO. 3, SEPTEMBER 2007

Fig. 3. Availability vs. failure rate (). Fig. 5. Prot (P ) vs. cost per visit (C ) for different values of revenue per unit up time (C ).

Fig. 4. Prot (P ) vs. failure rate () for different values of the revenue per unit up time (C ).

availability decreases with the increase in the value of the failure . rate Fig. 4 shows the behavior of prot (P) with respect to failure for different values of the revenue per unit up time . rate It can be concluded from the graph that prot (P) decreases with , and has higher an increase in the values of the failure rate . Notice also values for higher revenue per unit up time

accordingly as . So, for the model to be benecial, , the failure rate should be less than 0.001965. for , and 65, the value of the failure rate Similarly, for should be less than 0.00214, and 0.00232 respectively. Fig. 5 reveals the behavior of prot (P) with respect to the of the expert repairman for different values cost per visit . Prot (P) decreases as of revenue per unit up time increases. Further, for , accordingly as . So, for , the repairman should not be paid an amount greater than 107500 INR per visit. , and , the repairman should not Similarly, for be paid an amount greater than 125050, and 143100 per visit respectively. Fig. 6 represents the behavior of prot (P) with respect to revfor different values of failure rate . enue per unit up time It can be concluded from the graph that the prot (P) increases with an increase in the value of , and has higher values for , lower failure rates. It can also be noticed that, if accordingly as . then So, for , for the model to be benecial, the should , and 0.0005, the values for be 4.19. Similarly, for should be greater than 9.85, and 15.15 respectively. Fig. 7 depicts the behavior of prot (P) with respect to the for different values of the repair revenue per unit up time . It can be concluded from the graph that the prot (P) rate , and has higher increases with an increase in the value of values for higher repair rates. It can also be noticed that, if , then accordingly as . So, for the model to be benecial for , the should be 11.7. Similarly, for , and 0.008, the should be greater than 8.3, and 6.8 respectively. values for that, if

, then

PARASHAR AND TANEJA: RELIABILITY AND PROFIT EVALUATION OF PLC HOT STANDBY SYSTEM

539

REFERENCES
[1] M. Yamashem, A repairable multiple state device with general repair time, IEEE Trans. Reliab., vol. R-29, p. 276, 1980. [2] K. Murari and C. Maruthachalam, 2-unit parallel system with periods of working and rest, IEEE Trans. Reliab., vol. R-30, 1981. [3] A. Dharmadhikari and Y. Gupta, Analysis of 1-out-of-2:G warm standby repairable system, IEEE Trans. Reliab., vol. R-34, p. 550, 1981. [4] L. R. Goel, G. C. Sharma, and R. Gupta, Reliability analysis of a system with preventive maintenance and two types of repair, Microelectron. Reliab., vol. 26, pp. 429433, 1986. [5] C. Ibe Oliver and S. Wein Anne, Availability of systems with partially observable failures, IEEE Trans. Reliab., vol. 41, pp. 9296, 1992. [6] G. S. Mokaddis, S. W. Labib, and A. M. Ahmed, Analysis of a twounit warm standby system subject to degradation, Microelectron. Reliab., vol. 37, no. 4, pp. 641648, 1997. [7] S. Attahiru, W. L. Alfa, and Y. Q. Zhao, Stochastic analysis of a repairable system with three units and repair facilities, Microelectron. Reliab., vol. 38, no. 4, pp. 585595, 1998. [8] V. Goyal and K. Murari, Prot consideration of a 2-unit standby system with a regular repairman and 2-fold patience time, IEEE Trans. Reliab., vol. R-34, p. 544, 1985. [9] G. Taneja, Stochastic and prot analysis of some reliability models with different types of failure and repair, Ph.D. Thesis, Department of Statistics, M.D.University, Rohtak, India, 1992. [10] V. R. Prasad and M. Raghavachari, Optimal allocation of interchangeable components in a series-parallel system, IEEE Trans. Reliab., vol. 47, pp. 255260, 1998. [11] R. K. Tuteja, G. Taneja, and U. Vashistha, Cost-benet analysis of a system where operation and sometimes repair of main unit depends on sub-unit, Pure and Applied Mathematika Sciences, vol. LIII, no. 1/2, pp. 4161, 2001. [12] G. Taneja, Reliability and prot analysis of a system with PLC used as hot standby, in Proc. Of International Conference on Reliability and System Engineering (INCRESE 2005), India, December 2005, pp. 455464.

Fig. 6. Prot (P ) vs. revenue per unit up time (C ) for different values of failure rate ().

Bhupender Parashar is with the faculty in the Department of Applied Mathematics at JSS Academy of Technical Education, Noida, India. He received his M.Sc. (Operations Research) in 1997 with rst division, and Ph.D. in 2005 both from M.D.University, Rohtak, India. He has published 10 research papers in journals, and 4 research papers in the proceedings of National/International conferences. His research domain includes reliability modeling, and information theory.

Fig. 7. Prot (P ) vs. revenue per unit up time (C ) for different values of the repair rate ( 2).

Gulshan Taneja is a Lecturer (Selection-grade), and Head at the Department of Statistics at University College, M.D.University, Rohtak, India. He received his M.Sc. (Mathematical Statistics) in 1988 securing the rst position, and hence was honored by the Jawahar Lal Memorial Award; and he received the Ph.D. in 1992 both from M.D.University, Rohtak, India. He has published over 30 journal research papers, and over 10 research papers in the proceedings of National/ International conferences. He has supervised 5 Ph.D. dissertations. His research interests include reliability modeling, and reliability statistics.