Reliability and Availability - Coursework Example

 Reliability and Availability Complex systems such as communication systems, aircraft, printers, automobiles and medical diagnostics systems are repaired and not replaced when they fail. When the systems are subjected to a customer use environments, it is indispensable to determine the performance characteristics and reliability under these conditions (BAUER, 2012). It is becoming increasingly evident that these systems can exhibit unpredictable behavior faced with unexpected perturbations on their operating environments. Such alarms can be safe, but due to the latent flaws in the design of the systems combined with widespread coupling between its components, the effects of small agitations may be large and destructive, possibly rendering the inoperative. Other areas of interest may include assessing the expected number of failures during the warranty period, evaluating the rate of wear out and determining when to replace the whole systems. When addressing the reliability characteristics of the complex repairable systems, a process used instead of a distribution, the most common process model is the power law model (STAPELBERG, 2009). This model is popular for a few reasons; one is that truly practical foundation in terms of minimal repair. In this situation, when the repair of the system is enough to get the system operation again, secondly, if the period of the first period follows the distribution, then each succeeding the failure is governed by the power law model in case of minimal repair. A system failure occurs when the system does not meet the requirements, failing to designate the target. A system failure analysis is an investigation conducted to determine the underlying reasons for the non-conformance to systems requirements. The performance analysis is performed in order to recommend the appropriate corrective actions. System reliability is substantial challenges in system proposal unreliable systems are not only a significant source of user frustration, and also expensive (BLOCH, 2009). Avoiding downtime and the actual downtime make more than forty percent of the total cost of ownership for the modern systems. Unfortunately, with the introduction of large constituent total in today’s large scale systems are becoming the rule rather than exemption. System reliability has been a key concern since the first computer systems were built 50 years ago, we know little about the basic characteristics of the failures in real systems. Research in industries, as well as academia, is based on suppositious and often basic assumptions are considered. The period between the failures is exponentially distributed failures are independent. The cause is that there are virtually no data failures in real large systems publically available which could be used to derive more realistic models, the long term goal of the plan is to allow the conception of more reliable systems through a deeper understanding of real world failures. In our recent work, we compose and analyzed failure data on protuberance outages in a large number of HPC groups and data storage failures on several production systems. The system analysis starts with a clear understanding of the problem. Once the problem has been accomplished, all the potential failure causes are identified using the fault tree analysis. The process evaluates potential failures using several techniques. The techniques help in converging on the cause of the failure among many identified potential causes. Once it has been recognized, the slant outlined herein changes a range of curative actions and then selects the tracks optimum corrective action implementation. When confronted with a system failure, there’s always a natural tendency to begin assembling hardware to search for the cause. Hardware failure can reveal the valuable information and safeguards are vital in preventing losing the information from careless teardown actions. One of the most known is looking for prior to disassembling failed hardware. This is where the fault tree analysis enters the picture. Statistical methods of reliability are based on data analysis and test plan for industrial production systems. Statistical methods of reliability data updates and improves the Established methods as it proves how to smear the new numerical, graphical or simulation based methods for planning reliability studies and analyzing likelihood based methods of handling censored data and truncated data and more. The data collected for 9 years at a national laboratory which included 100 failures recorded on more than twenty different systems. Our findings included; average failure rates which differ across systems which range from 20 to more than 700 failures, and that time between failures is modeled well by a weibull Distribution with abating danger rate. From one system to another, the mean time varies from less than one hour to promote than a day, and the repair time are well modeled by a lognormal distribution. Failures Years A graph of failures against years Significant parts of the work focus on the analysis of the statistical properties of failures, as it has been recorded in our data. A better knowledge about the statistical properties of system failures processes is not only necessary for reliability evaluation of the new system design, but also empowers developers, more reliable and available systems. The common assumptions about the statistical onsets of failures are that they form a Poisson procedure this implies two key main properties; exceptionally distributed time between independence failures. We found that our analysis of characteristics of failures and cluster bulges outages that the assumption is not extremely realistic. System availability is calculated by modeling the systems as an interconnection Of parts in parallel and series, they are several rules, which detects whether the components are placed in parallel or series. If the failure of the part leads to a combination which becomes inoperable, the two parts are regarded to be functioning in series. If failure of the part clues to the other portion taking over the operation of the failed part, the two parts are measured to be functioning in parallel. Series Parallel The system consists of input transducer, which receives the signal and converting them into a data stream which is suitable for signal processing. The output is fed to a concluded pair of signal processors (LOCKS, 2009). The dynamic signal processor turns on the input while the reserve signal processor ignores the data from the input transducer. The active signal processor drives the data lines. The standby keeps the output, and input transducers are passive devices with no microprocessor control. The application of the concept of system reliability was protracted from the parallel system to the partly terminated or k-out of-n: G (F). An expression of yielded acceptable results for a wide range of values of k, n and component reliability, for systems of interest characterized by good components, the expression becomes highly liable to round off errors, and disastrous cancellations took place (BIROLINI, 2009). The numerical difficulties seemed inevitable as they were integrally associated with the meaning of the reliability metric. We introduce another metric; the cost of electricity of the mean time to failure (MTTF) the metric is tangible. However, this metric measures the relative change in reliability which can be obtained for a given comparative change in the failure. The methodology used combines the simulation and analysis. We derive an expression for ER.C for a parallel system based on a continuous limit. The spread sheet computes the following measures of reliability between the consecutive pairs. The data in the spreadsheet are from the study of the reliability of the sum of seven skinfolds of the systems. The change ΔR in reliability due to a change Δ n in The number of components is: ΔR = (ΔR / Δn) Δn ≈ (∂R / ∂n) Δn = - (1 - R0) n ln ( 1 - R0 ) Δn, = (1-R0)n ln ( The change in cost ΔC due to a change Δ n in the Number of components is ΔC = C0 Δ n As discussed when generating non homologous passion process, the can be generated very efficiently and in a fair straightforward fashion. This is the contrast with no homologous passion processes, where group methods incline to be much less frank. We can categorize the available methods for generating non homogeneous passion processes into three broad groups, which are; order statistics, inversion methods and acceptance rejection. In generating homologous, we use the simplest and useful model for M (t) = λ t and the repair rate is the constant m (t)= λ. The model comes about when the interracial times between failures are independent and identical dispersed according to the exponential distribution with stricture. This forms the basis of the following formulae: F(t) = 1 –e I λ t= CDF of the waiting time which is next to the failure N(T) = cumulative number of the failures from 0 to time P{N(t)=k } = λ tke-XT K! M(t)= λ expected number of failures M’ (t) = m(t) λ = repair rate I/λ = mean time between the failures System reliability is determined by the development of reliability models. The complexities of the models are dependent upon the various factors such functions, missions and redundancy characteristics. The general approach is to capture the modeling effort with reliability block diagram. The economic benefits of the block diagram are that they assist in identifying problems in case a system fails. They ease the work when it comes to problem solving by breaking it into blocks making it easy to repair the system. In conclusion, the concepts of MTTF, reliability and failure, rate have been defined, justified and discussed. In general, the failure can be interpreted as the frequency with which failures rate can be expected to occur (CHAPMAN, 2004). The description works well with the experimental estimation of the unknown parameters and provides an intuitive perspective. However, the reliability estimation is, in essence, probability equations. The reliability reports supply MTTF estimates scaled over range failures of the systems. The metric model assists in rectification of failures in systems. Due to the experience of the system failure form the various projects, I recommend these systems should repaired and we research on how we can increase the efficiency of the systems and reduce the failure often. The current work is identifying the real life instances in the industry and the methodology being developed to cope with the results emerging. This is to guarantee that the systems are fully functioning throughout the period. References BAUER, R. (2012). Reliability and availability of cloud computing. Hoboken, N.J., Wiley-IEEE Press. STAPELBERG, R. F. (2009). Handbook of reliability, availability, maintainability and safety in engineering design. London, Springer. BLOCH, H. (2012). Compressor troubleshooting and repair. New York, Tab. VILLEMEUR, A. (1992). Assessment, hardware, software and human factors. Chichester [u.a.], Wiley. LOCKS, M. O. (1995). Reliability, maintainability, and availability assessment. Milwaukee, Wis, ASQC Quality Press. BAUER, E., ZHANG, X., & KIMBER, D. A. (2009). Practical system reliability. Piscataway, NJ, IEEE Press. BIROLINI, A. (2010). Reliability engineering theory and practice. Berlin, Springer. CHAPMAN, J. (2004). System failure: why governments must learn to think differently. London, Demos. Read More