View
226
Download
0
Category
Preview:
Citation preview
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 1/13
4.2.4Part Feeding Case Study (Asfahl, 1992)
4.2.4.1 Selector Efficiency
A general criterion for evaluating a mechanical part selector is “efficiency”
which is defined as follows:
O
I
F efficiency
F = (4.1)
where O F = output feed rate of correctly orientated parts
I F = total input feed rate
It is possible to calculate the efficiency from experimental data after a
selector system has been designed and fabricated. However, a growing body
of scientific data has made possible the synthesis of efficiencies from
available information on the individual selector mechanism employed by the
system. For each selector mechanism there is a “transition probability
matrix” that reveals the probability that a piece will change to orientation y
when it passes through a selector device when the part is in orientation x
prior to encountering the device.
The most popular mode of selector device is the rejector , which simply
discards an incorrectly oriented piece back into the center of the vibratory
bowl or other hopper. The pure rejector is a devise for which the transition
probability from a given entry state to the rejected state is 100 percent. A
device can be a pure rejector for some entry states and not others. If a
device is a pure rejector for all entry states, it can be call a “total rejector,”
meaning that it rejects every part that encounters it, regardless of
orientation. The concept of “total rejector” is introduced here merely to
clarify the term “pure rejector.” In real automation application, pure rejectors
would have utility; total rejectors would not.
The general case for a part orientation devise recognizes a probability of
transition from any feasible orientation to any feasible orientation, includingthe possibility of rejection. Therefore, an n by n + 1 matrix of probabilities is
generated, in which n represents the number of feasible orientations. For
example, a rectangular block has six feasible orientations as shown below.
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 2/13
Thus, there are 42 probabilities in a 6 x 7 matrix used to describe completely
the general orientation characteristics of a device that orients rectangular
blocks. With most real devices, most of the probabilities in the matrix are
either zero or one (0 or 100 percent). Now let us consider an example
combination of orientation devices and perform probability calculations to
determine the distribution of output orientations.
Example 1:
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 3/13
Solution :
Before proceeding with a quantitative analysis of the probabilities associated
with each selector device, the general characteristics of each selector can be
observed in Table 3.1. The phase one wiper can be seen to be a pure rejector
for the “erect lengthwise” orientation, an intuitively reasonable characteristic
when the table is compare with the diagram in Figure 3.17. The wiper is only
a 90 percent rejector for the “erect crosswise” orientation; the other 10percent are switched to “flat lengthwise,” which is still an incorrect
orientation. The four remaining input orientations are shown by the table to
be unaffected by the wiper. The phase two off-ramp is shown in Table 3.1 to
be pure rejector for orientation e and f, just as was indicated in the diagram,
and has no effect on the other four orientations. The final device is the phase
three inclined rail, which has the function of lifting the “flat lengthwise” block
to “on-edge lengthwise,” the correct orientation. From Table 3.1 it can be
seen that the rail is only 80 percent successful in this function, the other 20
percent being rejected from the track.
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 4/13
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 5/13
4.2.4.2 Efficiency vs. Effectiveness
A 29.8 percent efficiency might seem to be little disappointing, but
remember that the objective in the example was to orient piece parts
automatically into a single correct orientation when presented with six
possible input orientations, and that the input distribution of those six
orientations was very unfavorable, with only 5 percent of the input streamappearing in the correct orientation to begin with. The example of selector
system above had a challenging assignment, and most real industrial
applications of automatic assembly present similar challenges in part
orientation. It is fascinating to watch an industrial vibratory bowl at work,
and it is not uncommon at all to observe a large quantity of rejected
orientations being tossed back into the bowl for a retry.
Another way to view the selector example is in term of “effectiveness”
instead of efficiency.
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 6/13
4.2.4.3 Part Wear and Damage.A final consideration may be that the parts selector will overwork the parts,
oscillating and vibrating them and kicking many of them back for retry after
retry. If the effectiveness is virtually 100 percent, it is possible to use the
system efficiency to calculate the chances that a part will be tossed back k
times before reaching an acceptable orientation. The formula is:
1100 100
k
k
E E p
= −
(4.3)
where k p = = probability that the part will be kick back k times
E= efficiency
k = number of kickbacks
For previous example:
(4.2)
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 7/13
0
0
29.8 29.81 0.298
100 100 p
= − =
1
1
29.8 29.81 0.209
100 100
p
= − =
2
2
29.8 29.81 0.147
100 100 p
= − =
10
10
29.8 29.81 0.009
100 100 p
= − =
Thus, nearly out of hundred parts will be kicked back ten times before
achieving an acceptable orientation. The automation engineer must decide
whether or not this kind of treatment will damage the product.
4.2.5 Analysis on Production and Throughput (Asfahl,
1992)
The automatic assembly machine, whether it be rotary or in-line, produces a
completed assembly every time the machine indexes, regardless the number
of stations in the assembly process. To compute the (ideal) production rate
of an automatic assembly machine, one needs to know only the indexing
cycle time; the number of stations is immaterial. It is true that throughputtime, the time required to assemble a given assembly from start to finish, is
dependent upon the number of stations and the index time, but production
rate is not.
Example 2:
An automatic assembly machine is of the dial-indexing configuration, has 8
stations, and is driven by a geneva mechanism in which the driver has a
rotational speed of 30 rpm. What is the production and throughput time of this machine?
Solution:
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 8/13
In geneva mechanism, every revolution of the driver constitutes one
indexing of the assembly machine. Therefore, the production rate is 30
units/min.
Production time = 1/production rate = 1/30 units/min x 60sec/min
= 2s/unit
Throughput time = prod. Time x no. of stations
= 2s x 8 stations
= 16 s
4.2.5.1 Machine Jamming
It is easy to overlook the pitfalls of automation, and one of the most
notorious examples can result from the setting of automatic multistationassembly machines without due consideration to the potential effects of
station malfunction or jamming. Consider an eight-station rotary indexing
machine driven by a geneva mechanism in which the index time is three
seconds and the dwell time is five seconds. The reader may want to check
the validity of the ratio of index to dwell time. Under ideal operating
conditions (no malfunctions), this eight-station rotary indexing machine will
produce a completed assembly every eight seconds and will achieve a
corresponding production rate of 450 units/ hour.
In more realistic case, suppose that each station malfunctions on theaverage of once every 100 cycles, a seemingly tolerable rate of work
stoppage. One or more station malfunctions will immediately jam the
indexing machine, requiring an operator to make adjustments to restart the
machine. For our example, let us say that this adjustment and restart
process requires a mere ten minutes. The result is a drastic reduction in the
productivity of the automated assembly machine, as can be seen in the
following series of calculations.
If the chances of malfunction is one in 100, the chance that a given stations
will not malfunction in a given cycle is 99% or 0.99. But all eight stationsmust operate without malfunction to produce a completed assembly
successfully. So the probability of no malfunction in a given cycle is a
product of the chances of no malfunctions at each station during that cycle.
Multiplying the station success probabilities;
80.99 0.9227=
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 9/13
Therefore, out of 10,000 machine cycles, 9227 assemblies will be produced
without malfunction, with a cycle time of eight seconds per assembly. This
will consume
9227 8sec 73816sec 20.50hr × = =
In the other 773 cycles (10,000-9227), at least one station will malfunction
and require a 10-minute repair. This will consume
773 10 min/ 7730 min 128.83beakdown hr × = =
The total time to produce the 9227 assemblies is then
20.50 128.83 149.33hr hr hr + =
for the total of operating time plus malfunction downtime. But note which
figure is the larger ! The percent of downtime is
128.830.863 86.3%
149.33= =
of the total production time! The production rate has been reduce from the
ideal 450 units per hour calculated earlier to
922761.8 /
149.33
unitsunits hr
hr =
The efficiency of the assembly machine would be the ratio of the actualproduction rate to the ideal production rate, calculated as
61.8 /0.137 13.7%
450 /
units hr efficiency
units hr = = =
an amazing 86.3% drop in efficiency from introduction of only a slight(one
percent) station malfunction rate. Such is the world of automation, and the
automation engineer should be prepared to deal with the realities of system
malfunction and downtime when planning for a new automated assembly
machine.
4.2.5.2 Component Quality Control
The previous section discussed the disastrous effects of station malfunction
in an automatic assembly machine, but failed to explain why these
malfunctions might occur. The predominant cause of assembly station
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 10/13
malfunction is some random variation in the components being assembled—
variation of a magnitude that cannot be handled by the assembly machine. If
tighter specifications can be applied or closer quality control can be exacted
upon the components produced to existing specifications, the automation
engineer may be able to achieve astonishing improvements in assembly
machine production rates. Suppose in the example of the previous section
that 90 percent of the assembly malfunctions were due to faulty
components. Elimination of the component quality problem would then
reduce the station malfunction rate from one out of 100 cycles to one out of
10,000 cycles. Such station reliability corresponds roughly to “four sigma”
performance.
Probability of no station malfunction in a given typical cycle:
8[ ] 0.9999 0.9992 probability nostationmalfucntion = =
thus, in 10,000 machine cycles, 9992 assemblies will be produced with no
malfunction with a cycle time of eight seconds per completed assembly. This
will consume
9992 8 sec 79936 22.20hr × = =
In the other 8 cycles (10,000-9992), at least one station will malfunction and
require a 10-minute repair. This will consume
8 x 10 min = 80 min = 1.33 hr
From here try to calculate the total production time, downtime percentage,
production rate and the efficiency.
Example 3:
An in-line automatic transfer and assembly machine has 30 consecutive
assembly stations. The line is under control of a walking beam with an index
time of four seconds and a dwell of 20 seconds. Each station along the line
will operate without malfunction with a reliability of 0.999 when its hopper issupplied with quality components. Any defective component will cause a
station to jam, which in turn will precipitate a line jam because there is no
provision for in-process assembly storage along the line. A jam or
malfunction requires 10 minutes to correct.
(a) No station malfunction or jamming
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 11/13
Assuming no station malfunction or jamming at all, what is the ideal
production capability of this line? What is throughput time?
Solution:
Cycle time
4 sec 20 sec 24 secC
T index dwell = + =
Ideal production rate
3600 /150 /
24sec/
sc hr R unit hr
unit = =
Throughput time
30 24 sec/ 720sec 12 minT T station station= × = =
(b) Assuming station malfunction
Assuming ideal component quality, what is the production rate
considering station malfunction? What is the percent downtime? What is
the throughput time?
Solution:
Prob [a cycle will not malfunction] = 300.999 0.9704=
Prob [a cycle will malfunction] = 1 0.9704 0.0296− =
Time to produce 9704 assemblies
Total time = total successful cycle time + malfunction
correction time
=24 sec 1 10 min 1
9704 2963600 sec 60 min
hr hr units malfucntions
unit malfunction× × + × ×
=64.69hr+49.33hr = 114.02hr
Production rate
R = 9702 units/ 114.02hr = 85 unit/hr
Percent downtime
D = 49.33 hr/114.02 hr = 0.43 = 43%
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 12/13
Throughput time
1 60 min30
85 /T
T stationunits hr hr
= × ×
=21.18 min
(c) Assuming defective part jamming
Assuming component lot quality is at the level of 12 of one percent
defective, what is the effect of defective components upon production
rate?
Solution:
We now have two ways that the station can jam:
1. General station malfunction (reliability = 0.99)
2. Defective component (1
2of one percent =0.005)
For the situation in which neither of these possibilities occurs we must
compute the product of probabilities:
Prob[a given station will not jam at a given cycle] = (0.999) X Pob[no
defective component]
= (0.999)(1 – 0.005) = 0.999 X 0.995
= 0.994
Since the system has 30 stations, any one of which is capable of bringing
down the entire line, we must compute the product of probabilities that
individual stations will not jam to obtain the overall system success
probability for a given cycle:
Prob [a cycle will not jam] = 300.994 0.8349=
And the probability that the system will jam in a given cycle is the
complement:
Prob [a cycle will jam] = 1 0.8349 0.1651− =
Time to produce 8349 assemblies
8/7/2019 Selector efficiency
http://slidepdf.com/reader/full/selector-efficiency 13/13
Total time = total successfully cycles time + unjam time
=24sec 1 10 min 1
8349 16513600sec 60 min
hr hr units jams
unit jam× × + × ×
= 330.82hr
Production rate
R = 8349 units/ 330.82hr = 25.24 units/hr
Percent downtime
D = 275.16 hr/ 330.82 hr = 0.83 = 83%
Efficiency
E =25.24 / 0.17 17%150 /
actualproductionrate units hr
idealproductionrate units hr = = =
Throughput time
1 60 min30
25.24 /T
T stationsunits hr hr
= × ×
71.32 min 1.19hr = =
Thus, a component quality level of only ½ of one percent defectives in this
case reduces the production rate from 95 units/hr to 25.24 units/hr,
increases downtime from 43 percent to 83 percent, and throughput time
increases from a little over 20 minutes to over an hour.
So important is the quality and uniformity of components to the success of
assembly automation that some firms pay several times the standard price
for such items as screws, nuts, and bolts to purchase the finest quality
available for use in automated assembly.
Recommended