REG NASA-LLIS-0840-2000 Lessons Learned Mean Time To Repair Predictions.pdf

1、Best Practices Entry: Best Practice Info:a71 Committee Approval Date: 2000-04-20a71 Center Point of Contact: JSCa71 Submitted by: Wil HarkinsSubject: Mean Time To Repair Predictions Practice: Predict the mean time to repair (MTTR) of avionics and ground electronics systems at any level of maintenanc

2、e (on orbit, intermediate or depot level) using analytical methods . This technique assumes a constant failure rate, and should be used accordingly.Programs that Certify Usage: This practice has been used on International Space Station Program.Center to Contact for Information: JSCImplementation Met

3、hod: This Lesson Learned is based on Maintainability Technique number AT-2 from NASA Technical Memorandum 4628, Recommended Techniques for Effective Maintainability.Benefit:The Mean Time To Repair (MTTR) predictions can be used to highlight those areas of a system that exhibit poor maintainability i

4、n order to justify improvement, modification, or a change of design. They also permit the user to make an early assessment of whether the system predicted downtime and logistic requirements are adequate and consistent with the system operational requirements and allocations.Provided by IHSNot for Re

5、saleNo reproduction or networking permitted without license from IHS-,-,-Implementation Method:In general, the MTTR of a system is an estimated average elapsed time required to perform corrective maintenance, which consists of fault isolation and correction. For analysis purposes, fault correction i

6、s divided into disassembly, interchange, re-assembly, alignment and checkout tasks. The repair time of a maintainable unit generally consists of both a large number of relatively short-time repair periods and a small number of long-time repair periods. The former would correspond to the more usual c

7、ase where the failed unit is replaced by a spare at the operational site on detection of a failure. The long downtimes would occur when diagnosis is difficult or removing a defective part is complicated due to, for instance, rusted/stripped mounted nuts. Having a collection of such field data provid

8、es the design engineer an opportunity to assess the Mean Time To Repair (MTTR) of the current system as it matures, or to predict the MTTR of a new system according to its features with the current system.MTTR is a useful parameter that should be used early in planning and designing stages of a syst

9、em. The parameter is used in assessing the accessibility/locations of system components; for example, a component that often fails should be located where it can easily be removed and replaced. The estimated MTTR may also dictate changes in system designs in order to meet the turn-around time criter

10、ia for critical systems, such as communication and life support systems on the Space Station. In addition, the parameter helps in calculating the life cycle cost of a system, which includes cost of the average time technicians spend on a repair task, or how much Extravehicular Activity (EVA) time is

11、 required for astronauts to repair a system.MTTR is defined as the average time necessary to troubleshoot, remove, repair, and replace a failed system component. An interval estimator for MTTR can be developed from the mean of the sample data, within a lower and a upper limit with a confidence bound

12、. For example, from a sample data set, one can find with 90-percent confidence that the range 3.2 to 4.2 will contain the population mean. Unfortunately, the exact MTTR of a system can never be found due to data uncertainties.Log-Normal DistributionThe distribution most commonly used to describe the

13、 actual frequencies of occurrence of system repair time is the log normal because it reflects short duration repair-time, a large number of observations closely grouped about some modal value, and long repair-time data points. The general shape of log normal distribution is shown in Figure 1.Provide

14、d by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-refer to D descriptionD Figure 1: Lognormal Distribution Without getting involved in the derivation of the distribution equations which can be found in any statistical textbook, the following example will illu

15、strate how MTTR of a replaceable unit may be calculated from a finite observed set of data.Example 1: The repair times tifor an orbital replaceable unit (ORU) are observed to be 1.3, 1.5, 1.7, 1.8, 2.2, 2.6, 3.0, 3.1, and 3.9 hours. Using log normal distribution to estimate the MTTR of the unit.Solu

16、tion:refer to D descriptionD Utilizing statistical methods, the Maximum Likelihood Estimator (MLE), or the best estimated value of the mean is:Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-refer to D descriptionD The Maximum Likelihood Estimator of

17、 the variance is:refer to D descriptionD Therefore, the mean of the log normal distribution of this example is:refer to D descriptionD and its variability of time to repair is:refer to D descriptionD Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Ho

18、w to Implement the MTTR ProcessAccurately estimating the MTTR of a new system is more than applying the derived formulas on field data of any existing systems. The designer must know the overall maintenance concept and operating conditions of the new system; for example, how and where the system is

19、going to be operated and how its failed units will be swapped out. With this background, the designer can proceed to approximate the maintenance procedure of the new system, then select an existing system that has been exposed to similar operating conditions and that has a mature set of operating da

20、ta. After the similarity between the two systems is assessed, the designer then can determine certain conversion factors needed to make the existing system data more applicable to the new system. Once this is done, the predictions for the new system are more meaningful and accurate.Elements of MTTRT

21、he MTTR prediction of a system begins at the replaceable unit level (RUL) where a defective unit is removed and replaced in order to restore the system to its original condition. Then the system MTTR predictions are accomplished by integrating the MTTRs of maintainable units. The following defines t

22、he elements used in the MTTR prediction of a system:Fault Isolation: Time associated with those tasks required to isolate the fault to the item.Disassembly: Time associated with gaining access to the replaceable item or items identified during the fault correction process.Interchange: Time associate

23、d with the removal and replacement of a faulty replaceable item or suspected faulty item.Reassembly: Time associated with closing up the equipment after interchange is performed.Alignment: Time associated with aligning the system or replaceable item after a fault has been corrected.Checkout: Time as

24、sociated with the verification that a fault has been corrected and the system is operational.Constant failure rates: The rate of failures that result from strictly random or chance causes. This type of failure occurs predominantly in the useful life period of a unit.K factor: For on-orbit tasks, a c

25、onversion factor may be applied to convert elemental task times performed in 1-g environment to Micro-gravity environment. The conversion factor may be derived from data of past similar programs or from the neutral buoyancy testing.Provided by IHSNot for ResaleNo reproduction or networking permitted

26、 without license from IHS-,-,-Ground Rules and AssumptionsIn the prediction, certain ground rules and assumptions apply:a71 Mean Time To Repair (MTTR) does not include the maintenance overhead, which is generally non-related task time such as time to fill out a requisition, time to go get tools, bre

27、ak-time, time waiting for parts, etc.a71 Worksite time is the only variable considered.a71 All equipment experiences a constant failure rate.a71 All tasks are performed sequentially by one crew member unless otherwise noted.a71 Maintenance is performed in accordance with established maintenance proc

28、edures and appropriately trained personnel.a71 The prediction depends upon the use of recorded reliability and maintainability data and experience that have been obtained from comparable systems and components under similar conditions of use and operation.System Level PredictionAt the system level,

29、MTTR is calculated by summing the product of the replaceable items MTTRs and their corresponding failure rates; the result is then divided into the sum of all replaceable items failure rates.Mathematically, it can be expressed as:Provided by IHSNot for ResaleNo reproduction or networking permitted w

30、ithout license from IHS-,-,-refer to D descriptionD As an example, assume the three ORUs of a system have the following MTTRS, Variance (V), and failure rates L:Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-refer to D descriptionD Table 1: ORU / MT

31、TR Apply the above formula to calculate the system MTTR:The results of the above example indicate that the most often failed unit will essentially drive the MTTR and variance of a system.refer to D descriptionD Overall, the prediction is a straight forward process and is useful in estimating a syste

32、ms MTTR. Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Even with a limited set of data, if the prediction is used early in the design phase, the derived value should help in shaping a preliminary design guideline for the system. In addition, the pr

33、ediction can also verify logistics and maintainability requirements at some later stage.References:1. Lamarre, B. G., Mathematical Modeling, Reliability and Maintainability of Electronic Systems, Edited by: J.E. Arsenault and J.A. Roberts, Computer Science Press, p372 - 373.2. Miller, Irwin, Probabi

34、lity and Statistics for Engineers, Prentice Hall Inc., Englewood Cliffs, p116.3. MIL-HDBK-338-1A, Electronic Reliability Design Handbook, Department of Defense.Impact of Non-Practice: This MTTR prediction technique is a fast, simple, accurate and effective approach for providing a design baseline fo

35、r repair times. Design and product assurance engineers can use the MTTR data to effectively define sparing, logistics and maintenance programs for a pending design.Related Practices: N/AAdditional Info: Approval Info: a71 Approval Date: 2000-04-20a71 Approval Name: Eric Raynora71 Approval Organization: QSa71 Approval Phone Number: 202-358-4738Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-