Reliability of A System

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

THE RELIABILITY OF A SYSTEM BY LAME, GARBA USMAN PRINCIPAL TECHNICAL OFFICER, CENTRE FOR GEODESY AND GEODYNAMICS TORO,

NATIONAL SPACE RESEARCH AND DEVT. AGENCY NIGERIA. Email: lame.garba@yahoo.com siddeeqlame@yahoo.com
for enquiries contact: +2348032105430, +2348050552545, +2348096764662

RELIABILITY OF A SYSTEM
Definition: Reliability is defined as the probability of a system that will operate to an agreed level of performance for a specified period of time, subject to specified environmental conditions. In general, the definition means that each value coated for some equipments relates only to the operating period concern, and the specified level of performance and working conditions. Reliability is but one characteristics of electronic device or system or machines which must be considered when selecting one of a number of alternative of designs. From the users point of view, the most rational criteria for deciding which design is best is that of minimal total line cost. This procedure involves estimating the cost of purchasing equipments and initial spare units, the cost of routine maintenance and spare parts for replacement. In some cases the cost of being unable to operate if the equipments fails without warning and must be withdrawn from service for repairs . the total estimated cost for all these activities during the total life of the equipments is then used as a criteria of value and the arrangement given the least total life cost is adopted. Equipment reliability then becomes an important parameter in the design of such equipments. As the reliability increases , the cost of buying equipments to be held in reserve in case of failure, the cost of
2

spare replacements, and the cost of maitaining staff all decreases. However, the cost of design and development and the initial purchase price all increases very rapidly as more effort is devoted to increase the reliability. CAUTION: there is thus a stage beyond which no economic benefit can be obtained from any increase in reliability.

MEAN TIME BETWEEN FAILURE (MTBF) Reliability suffers from one disadvantage and this arises from the need to specify a particular operating period for the equipment. If the same equipment is operated for a different period, its realibility will be different. Thus, one piece of equipment may have a large range of different values of realiibility, depending upon the period for which it is used. The most useful measure of performance which does not involve the period of observations is the mean time between failure (MTBF). This applies only to repairable equipments and is obtained by runnining an item for a stated length of time under stated stress conditions and calculating the mean value of the time between the consecutive failure, that is by measuring the total test period T, during which N faults occur.

Another way of expressing equipment reliability is the failure rate, that is the number of faults per unit time. For many electronic systems, the failure rate is approximately constant for much of the working life of the equipment, where this is the case, the failure rate is the reciprocal of the MTBF, thus
3

 = Where nf is the number of failure which occur in time t and Nsis the number of survivors. MTBF is usually expressed in hours and corresponding units of are faults /hour. For components , is extremely smalland the units may be altered to give more convenient numbers. Thus, failure rates may be quated as a percentage per 1000hrs, failure per 10 to the power of 6 hours 106 or failure of 109hours. For example, a system with an MTBF of 2000hrs has =1/MTBF But MTBF is 2000hrs =1/2000hrs =0.00005faults/hour And in percentage =0.00005 x 100% =0.05% failure/hr Therefore in 1000hrs = 0.05 x 1000% =50% failure/1000hrs.

The MTBF is a concept applicable to any type of equipment which can be repaired by the replacement of faulty components or units and other things being equal, the equivalent with the greatest MTBF will be

the most reliable regardless the period of observation. MTBF thus provides a most convenient index of reliability.

THE MEAN TIME TO FAILURE (MTTF) The MTBF is a measure of reliability for a repairable equipment thus a similar measure is useful for component such as valves, resistors, capacitors, transistors, inductors e.t.c. which are thrown away items that cannot be repaired. The correct measure for this component is the mean time to failure (MTTF). This may be calculated from the results of life testing as follows . let a set of N items be tested untill all has failed, the times of failure being :T1, t2, t3, t4, t1- - - - - - - - - - tn.

Then the observe MTTF is given by

As with the measurement of MTBF,the life test must be conducted on only a sample of the production batch, so that the observed MTTF will be subject to sampling errors. The failure rate will be given by  If is independent of time. For example, if 6units were tested until failure , and the times to failure were 320, 250, 380, 290, 310 and 400 hrs. The total test in hours is 1,950hrs.
5

MTTF = Total test hours (time)/number of units =1950/6 =325hrs. = failure rate = 1/325 = 0.003076923

FAILURE RATE As far as design is concern, the life the life of a component can be greatly improved if it is operated within the full rated values of the current, voltage and power. This technique is called DERATING, which is used extensively to reduce failure rate. The failure rate of a component may be found by operating large numbers, of the components for a long period and noting the number of failures which occur. For example, a batch of small signal diodes taken straight from the production line and placed on test at their maximum rated voltage. Within a short a period it is quite likely that several diodes will fail, due to manufacturing faults and material imperfections. Gradually, as the diodes that inherit weakness fail, and are withdrawn, the rate of failure will fall. This initial period of high failure rate, known as BURN-IN or early failure, is followed by a period where the rate of failure levels off to an almost constant value. This time is known as RANDOM FAILURE OR USEFUL LIFE. Using the failure rate over this period enables predictions made of reliability by means of probability theory. If the test were continued beyound the useful life period, a gradual increase in failure rate will be observed as the diodes fails one
6

by one because of the ageing process which is called WEAROUT PERIOD. The variation in failure rate (FR) with time is shown graphically below and because of its shape it is often called the BATH-TUB curve.

WEAR-OUT FAILURE

If diodes are to be used in a design where high reliability is required, the early failure effect could be eliminated by pre-ageing all productions for a short period say 100hrsand using those that survive. Wear out could be delayed by improving the design, by changing materials and by a closer control of production, though this may be very costly. The random failure rate during use could be reduced by
7

FAILURE RAT E

AGEING B URN IN USEFUL LIFE PERIOD


FIGURE: BATH-TUB CURVE. FAILURE RATE VARIATION WITH TIME.

Y EARL FAILURE

RANDOM FAILURE

operating the diode at lower power. Returning to the test, suppose that after burn-in period, out of 400 diodes left on test, 5 fails over 1000hrs period, then the average failure rate is :

FR=5/400X100%/1000HRS =5/4%/1000HRS =1.25%/1000HRS But this expression can also be expressed as failures/hour. FR=5/400X1/1000/HR=12.5/hr This failure rate obtain from a sample of production is called the best estimate. Because sample sizes are relatively small to reduce too favorable in figure, a higher failure rate is giving. The failure rate is then said to have a confidence level of 60, 50, and 90%. For example a confidence level of 80% means that if an infinite no. of sample were taken, 80% of the failure rate obtain will be below the figure for the 80% confidence. To adjust the best estimate figure of failure rate, a constant is added to give a confidence level figure. These constants c60, c80, c90 can be applied when the failure rate is itself constant. Hence the failure rate is calculated using the formula:

FR= NO. OF FAILURES+C60/ NO. OF COMPONENTS/HR.

You might also like