Beruflich Dokumente
Kultur Dokumente
Design
Reliability is a new concept, barely three decades old and is primarily due to
the complexity, sophistication and automation inherent in modern
technology. It has been established that the reliability, which is a measure of
quality, is an essential element at each stage of the equipment
manufacturing procedure through design and production to final delivery to
the user.
Reliability can be defined in the following ways:
1. The reliability of a component (or a system) is the probability that the
component (or a system) will not fail for a time t.
2. The reliability of a system is called its capacity for failure –free
operation for a definite period of time under given operating
conditions.
3. Reliability is the probability of an item performing its intended
function over a given period of time under the operating conditions
encountered.
F (t) = m/N
It is clear that at any time, t,
R (t) + F(t) = 1
because R (t) & F (t) are mutually exclusive events.
2
v. Density function, f(t)
The probability that a random trial yields a value of t within the interval
from t1 to t2 , is
t2
∫ f (t) dt
t1
then f(t) is the density function for the continuous random variable.
vi. Distribution function, F (t)
The distribution function, F (t) is the probability that in a random trial,
the random variable is not greater than t, thus,
t
F (t) = ∫ f (t) dt
-∞
3
Figure 1.1
Then the eccentricity vector for the left support ēl is determined by the
equation
4
ēl = Σ ē i ………… 1
i=1
where ē1 , ē2 , ē3 and ē4 are the eccentricities for the outer ring, balls, inner
ring and shaft respectively. Similarly the eccentricity for the right support is
determined by
5
ēr = Σ ē i ………. 2
i=1
where ē1 , ē2 , ē3 , ē4 and ē5 are the eccentricities for the bore, outer ring,
balls, inner ring and shaft respectively. (It is assumed here that the housings
for the two bearings cannot be bored with one setup, which means that the
right bore may have a displaced geometric axis with respect to the left). The
statistical means for the eccentricity magnitude in equations 1 and 2 are
related to the tolerance by equations 3 and the Rayleigh distribution
parameters are found from equation 4.
μe = Δ / 4 , ……….. 3
4
Δ - tolerance for each outside diameter of a ring
ē = ēl + ēr ……….. 5
9
σ 2 = Σ σ i2 …………. 6
i=1
*
P(α > α*) = ∫ f(e) de …………. 7
e*
where e* = α*L
+0.016
db = 40
-0
5
Then the corresponding tolerances are
Shaft Δs = 0.016 mm
σ2 = 0.006415 σ = 0.08
2. A single lip seal separating a rotating shaft from the stationary case is
shown. This type of a seal is usually used for retaining lubricants in
machines having rotating and/or oscillating shafts. Various polymer
materials are used for seal elements. The function of the seal is
determined by its ability to maintain a constant radial pressure
between the sealing element and the shaft. Anything causing a
decrease in this pressure affects leakage and thus seal failure.
Consider factors affecting the radial lip seal performance.
6
OIL
2. Wear: Wear at the contact point of the seal directly affects its
functioning. The rate of the wear depends on
• Coefficient of friction at the seal- shaft interface
• Pressure exerted by the garter spring
• Pressure due to the seal-shaft fit
• Pressure caused by the oil
• Shaft surface finish
• Shaft surface hardness
• Speed of rotation
• Shaft oscillations
• Abrasion resistance of the seal material
7
result, a seal cannot follow the shaft displacements, an event
that constitutes seal failure. Factors affecting fatigue are:
• Fluctuating temperature
• Chemical environment
• Oscillating shaft
• Pressure at the seal-shaft interface
Not all the above factors are equally important in seal reliability. To
assign some weight to these factors would require more specific data about
the design and the seal. Note that seal reliability is affected by the process of
assembly as well, since it can simply be damaged while assembled. In
practical situations, experience accumulated in industry is used as a guide.
8
Figure 3.1 Brake Hoist
In the first case, the brake is normally locked, thus both the springs are
exerting forces on the corresponding shoes. To unlock the brake, the
hydraulic actuator that is used must retract both the springs and thereby
both the shoes, so the reliability diagram consists of all the components in
series system. Therefore if one of the components fails, the brake does not
unlock.
In the second case, the brake is normally unlocked but becomes locked
when hydraulic pressure is decreased. This is possible even if one brake shoe
fails to lock the other shoe locks the brake. But for the above situation to be
true all the components that lock the individual brake shoes must be in
series.
6 5 4 3 2 1
5 3 1
6 4 2
9
4. Three possible configurations of gear reduction boxes are shown.
Assuming that the input output characteristics of all boxes are
identical,
a. Discuss the reliability advantage and disadvantage of every
configuration and
b. Draw the reliability diagram for each configuration considering only
such components as gears
In the first case, the failure of any one of the elements of the
system will lead to failure of the system. So all the components in the
system (the bearing 1, input shaft, gears, output shaft and bearing 9
are in series. The above system has a reliability advantage over the
others by having the least number of components.
1 2 3 4 5 6 7
2 3
1 6 7
4 5
8
9
10
Figure 4.3 (b)
5 6
1 2 3 4 9
7 8
In the third case , reliability is reduced at the intial stage itself where
there is only one set of gears from the input shaft to the middle shaft. After
this the reliability is increased with the redundant set of gears from middle
shaft to the output shaft.
11
Figure 5.1 Spur Gear Support System
S G
S G
12
6. Two concentric springs make up a support. Both the springs are known
to be in a state of random failure with the rates h1 = 0.005
failures/kilocycle and h2 = 0.008 failures/kilocycle for the inner and
outer springs respectively. Each of the two springs is capable of
withstanding the fluctuating load. However, if the inner spring fails, the
failure rate of the outer one becomes h2* = 0.012 failures/kilocycle; if
the outer spring fails the failure rate of the inner one becomes h 1* =
0.0075 failures/kilocycle. What is the reliability function of the support
from 0 to 400 kilocycle?
13
= 0.002719 * 5
= 0.01359
Reliability function
= 1/15 (2.5+3.5+…………+32+36)
= 15.6 Kilocycles
The chi-square values from tables are
95 percent confidence (α = 0.05), 2n = 30
14
μχ(L) = 2n μχ / { χ2 [(α/2), 2n]}
= (30 *15.6) / 53.632
= 8.72
Then the 99 percent confidence limits for μχ are 8.72 ≤ μχ ≤ 33.94
It is seen how an increase in the confidence level (1-α) widens the confidence
interval. A probability that the true mean is within the specified limits is
higher if the range is wider.
15
9. Consider the data given in problem 7. Do they fit the exponential
distribution and if so, what is the value of the parameter λ? Let us take
Δti = 5 Kilocycles. Note that in this case n = 15, carry out the
calculations and plot the results on the weibull paper.
ti
Ns(ti) F(ti) R(ti) f(ti) h(ti)
Kilocycles
<5 12 0.176 0.824 - -
5-10 9 0.370 0.630 0.0388 0.062
10-15 7 0.5 0.5 0.026 0.052
15-20 5 0.63 0.37 0.026 0.07
20-25 3 0.76 0.240 0.026 0.108
25-30 2 0.825 0.175 0.013 0.074
30-35 1 0.89 0.11 0.013 0.118
35-40 0 0.955 0.045 0.013 0.288
Where F(ti) = 1 – [(Ns(ti)+ 0.7)/(N +0.4)] , here N = 15
R(ti) = 1- F(ti)
f(ti) = d[R(ti)- R (ti+1)]/dt
h(ti) = f(ti) / R (ti)
It is seen that a straight line approximates the failure data fairly accurately,
except for the last point. For the comparison h(t) is calculated in the above
table, and the results are shown in the plot. It is seen that h(t) has a small
upward trend if it was not the last point (the out-of-sequence point can be
due to the small size of the recorded data).The simplest way to find λ is to
use the equation
λ = - t/ (ln R (t))
= - t/ (ln 1-F (ti)),
for any point t =ti that lies on the straight line. Let us take t i =20 Kilocycles
and the corresponding F (ti) = 0.63
Then,
λ = - t/ (ln 1-F (ti)),
= 20/0.994
= 20.1 Kilocycles
This estimation gives λ larger than the mean μ λ* = 15.6 Kilocycles in problem
7. At the same time for μλ* = 15.6, the constant hazard value h* = 0.064 and
approximates only the initial part of the failure times well.
16
interpretation of data on graph, to the assumption that the data can be
described by exponential distribution (in fact, it cannot be), or both.
17
Figure 9.1 Weibull Plot
18
10. Failures of 20 shafts operating at a constant stress level were reported.
The decimal logarithm, log K, of the no. of cycles to failure was distributed as
follows: 4.2, 4.4, 4.58, 4.6, 4.7, 4.76, 4.81, 4.86, 4.88, 4.96, 4.98, 4.99, 5.02,
5.12, 5.17, 5.25, 5.36, 5.4, 5.49 & 5.78. 1. Take the interval ∆t = ∆(log K) =
0.2 and plot the reliability, failure density and hazard functions. Does it look
like log K is normally distributed? 2. Assume that log K is a normally
distributed variable and that the mean and standard deviation can be
approximated by the average and an unbiased variance, respectively. What
no. of failures should be expected in a population of 100 products with similar
shafts in similar conditions during the first 100 kilocycles? 3. What are the 95
percent confidence limits for the no. of failures during the first 100 kilocycles?
The test and reliability data are given in the table as follows. Plots of R (t i),
f (ti), h (ti) are as shown in the figure below. Although h(t) function is
increasing, which means that qualitatively a normal distribution for log K
would not be contradictory, the f(t) function for the assumed ∆t does not
give an indication that the normal distribution is a good fit.
= 4.96
= 0.146
19
Where F(ti) = 1 – [(Ns(ti)+ 0.7)/(N +0.4)] , here N = 20
R(ti) = 1- F(ti)
f(ti) = d[R(ti)- R (ti+1)]/dt
h(ti) = f(ti) / R (ti)
Where
For α = 0.05 and v = n-1 = 19 the t value from the tables( B -3) is t 0.025,19
= 2.093, then
then,
0.084 ≤ σ2 ≤ 0.311
Thus there is a 95 percent confidence that the no. of failed shafts may be
within the limits
40 ≤ N f ≤ 78
20
Figure 10.1 Reliability, failure, and hazard functions for shafts
21
11. Consider the data in problem 9 for 20 failed shafts, and (1) Make a
weibull probability plot, (2) estimate the parameters of the distribution, (3)
find out the nature of the failure rate, and (4) find out whether the
distribution is close to normal, an assumption made in problem 9.
(1) The plot is constructed on the basis of data in table 10.1 assuming δ =
0. It looks like a straight line fits the experimental points, and thus the
Weibull distribution can be assumed to be appropriate in this case.
Then
Taking from table 10.1(F1, t1) =(0.083,4.3) and (F2, t2) = (0.966,5.7) gives
22
Figure 11.1 Weibull Plot
23
12. Failures of dc motors in specified intervals of days are given in table
12.1. Assuming that the dc motors were in a state of random failure,
what is the 95 percent, two-sided confidence interval for the mean life?
Determine the number of days at which 10 percent motors will have
failed.
Interval
0-200 200-400 400-600 600-800 800-1000
Number of 5 3 2 2 1
failures
Table 12.2 Probability of Failure:
Days Mid point Number of Probability of
failures failure (Pi)
0-200 100 5 0.384
200-400 300 3 0.230
400-600 500 2 0.153
600-800 700 2 0.153
800-1000 900 1 0.077
= 676.1 days
= 223.2 days
Then the 95 percent confidence limits for μχ are 223.2 ≤ μχ* ≤ 676.1
24
2. 10% failure P =0.1
= 360 ln (1-01) –1
= 37.92 days
25