Chapter 17

Automated Assembly Systems

Chapter Contents
17.1 Fundamentals of Automated Assembly Systems
17.1.1 System Configurations
17.1.2 Parts Delivery at Workstations
17.1.3 Applications
17.2 Analysis of Automated Assembly Systems
17.2.1 Parts Delivery System at Workstations
17.2.2 Multistation Assembly Machines
17.2.3 Single-Station Assembly Machines
17.2.4 Partial Automation
17.2.5 What the Equations Tell Us

The term automated assembly refers to mechanized and automated devices that perform
various assembly tasks in an assembly line or cell. Much progress has been made in the
technology of assembly automation in recent years. Some of this progress has been mo-
tivated by advances in the field of robotics. Industrial robots are sometimes used as com-
ponents in automated assembly systems (Chapter 8). In this chapter, automated assembly
is discussed as a distinct field of automation. Although the manual assembly methods
described in Chapter 15 will be used for many years into the future, there are significant
opportunities for productivity gains in the use of automated methods.
Like the transfer lines discussed in the preceding chapter, automated assembly sys-
tems are usually included in the category of fixed automation. Most automated assembly
systems are designed to perform a fixed sequence of assembly steps on a specific product.

472
Sec. 17.1 / Fundamentals of Automated Assembly Systems 473

Automated assembly technology should be considered when the following conditions

exist:

• High product demand. Automated assembly systems should be considered for

products made in millions of units (or close to this range).
• Stable product design. In general, any change in the product design means a change
in workstation tooling and possibly the sequence of assembly operations. Such
changes can be very costly.
• A limited number of components in the assembly. Riley [11] recommends a maxi-
mum of around a dozen parts.
• The product is designed for automated assembly. In Chapter 24, the product design
factors that allow for automated assembly are explored.

Automated assembly systems involve a significant capital expense. However, the

i­nvestments are generally less than for the automated transfer lines because (1) work units
produced on automated assembly systems are usually smaller than those made on trans-
fer lines and (2) assembly operations do not have the large mechanical force and power
­requirements of processing operations such as machining. Accordingly, in comparing an au-
tomated assembly system and a transfer line with the same number of stations, the assembly
system would tend to be physically smaller. This usually reduces the cost of the system.

17.1 Fundamentals of Automated Assembly Systems

An automated assembly system performs a sequence of automated assembly operations

to combine multiple components into a single entity. The single entity can be a final prod-
uct or a subassembly in a larger product. In many cases, the assembled entity consists of
a base part to which other components are attached. The components are usually joined
one at a time, so the assembly is completed progressively.
A typical automated assembly system consists of the following subsystems: (1) one
or more workstations at which the assembly steps are accomplished, (2) parts feeding
­devices that deliver the individual components to the workstation(s), and (3) a work han-
dling system for the assembled entity. In assembly systems with one workstation, the work
handling system moves the base part into and out of the station. In systems with multiple
stations, the handling system transfers the partially assembled base part between stations.
Control functions required in automated assembly machines are the same as in the
automated production lines of Chapter 16: (1) sequence control, (2) safety monitoring,
and (3) quality control. These functions are described in Section 16.1.3.

17.1.1 System Configurations

Automated assembly systems can be classified according to physical configuration. The prin-
cipal configurations, illustrated in Figure 17.1, are (a) in-line assembly machine, (b) dial-type
assembly machine, (c) carousel assembly system, and (d) single-station assembly machine.
The in-line assembly machine, Figure 17.1(a), is a series of automatic workstations
located along an in-line transfer system. It is the assembly version of the machining transfer
line. Synchronous and asynchronous transfer systems are the common means of transport-
ing base parts from station to station with the in-line configuration.
 474 Chap. 17 / Automated Assembly Systems

Components added at stations (6)

Components added at stations

Asby
y Aut
sb

A ut
A ut

sb
A
Asby Asby Asby Asby Asby Asby

y
Starting Completed Starting A
base parts Aut Aut Aut Aut Aut Aut assemblies base parts
Asby
Aut
Sta Sta Sta Sta Sta Sta
1 2 3 n–2 n–1 n y
sb
(a) A ut
Asby A
Aut

Completed
assemblies
(b)

Components added at stations

Starting
base parts Components added at one station
Asby Asby Asby Asby Asby A ut
Aut Aut Aut Aut Aut
sb
A
y

y
sb Asby
Asby Asby Asby Asby Asby A ut Starting Completed
Aut Aut Aut Aut Aut A base parts Aut assemblies
Completed
assemblies (d)
(c)

Figure 17.1 Types of automated assembly systems: (a) in-line, (b) dial-type, (c) carousel,
and (d) single station.

In the typical application of the dial-type machine, Figure 17.1(b), base parts are loaded
onto fixtures or nests attached to the circular dial. Components are added and/or joined to
the base part at the various workstations located around the periphery of the dial. The dial-
indexing machine operates with a synchronous or intermittent motion, in which the cycle
consists of the service time plus indexing time. Dial-type assembly machines are sometimes
designed to use a continuous rather than intermittent motion. This is common in beverage
bottling and canning plants, but not in mechanical and electronics assembly.
The operation of dial-type and in-line assembly systems is similar to the operation
of their counterparts for processing operations described in Section 16.1.1, except that
assembly operations are performed. For synchronous transfer of work between stations,
the ideal cycle time equals the operation time at the slowest station plus the transfer time
between stations. The production rate, at 100% uptime, is the reciprocal of the ideal cycle
time. Owing to part jams at the workstations and other malfunctions, the system will al-
ways operate at less than 100% uptime.
As seen in Figure 17.1(c), the carousel assembly system represents a hybrid between
the circular work flow of the dial-type assembly machine and the straight work flow of the
in-line system. The carousel configuration can be operated with continuous, synchronous,
or asynchronous transfer mechanisms to move the work around the carousel. Carousels
with asynchronous transfer of work are often used in partially automated assembly sys-
tems (Section 17.2.4).
Sec. 17.1 / Fundamentals of Automated Assembly Systems 475

In the single-station assembly machine, Figure 17.1(d), assembly operations are per-
formed on a base part at a single location. The typical operating cycle involves the place-
ment of the base part at a stationary position in the workstation, the addition of components
to the base, and finally the removal of the completed assembly from the station. An impor-
tant application of single-station assembly is the component placement machine, widely
used in the electronics industry to populate components onto printed circuit boards. For
mechanical assemblies, the single-station cell is sometimes selected as the configuration for
robotic assembly applications. Parts are fed into the single station, and the robot adds them
to the base part and performs the fastening operations. Compared with the other three
system types, the cycle time is longer in a single-station assembly system because all of the
assembly tasks are performed sequentially instead of simultaneously. Hence, production
rates are inherently slower. Single-station assembly systems are analyzed in Section 17.2.3.

17.1.2 Parts Delivery at Workstations

In each of the configurations described earlier, a workstation accomplishes one or both of

the following tasks: (1) a component is delivered to the assembly work head and added to
the existing base part in front of the work head (at the first station in the system, the base
part is often deposited onto a work carrier or pallet fixture), and (2) a fastening or joining
operation is performed at the station to permanently attach components to the existing
base part. In the case of a single-station assembly system, these tasks are carried out mul-
tiple times at the single station. Task (1) requires the parts to be delivered to the assembly
work head. The parts delivery system typically consists of the following hardware:

1. Hopper. This is the container into which the components are loaded at the worksta-
tion. A separate hopper is used for each component type. The components are usu-
ally loaded into the hopper in bulk. This means that the parts are randomly oriented
in the hopper.
2. Parts feeder. This is a mechanism that removes the components from the hopper
one at a time for delivery to the assembly work head. The hopper and parts feeder
are often combined into one operating mechanism. A vibratory bowl feeder, pic-
tured in Figure 17.2, is a very common example of the hopper-feeder combination.

Feed track

Outlet
Bowl

Bowl support frame

Suspension springs
Electromagnet

Base

Support feet

Figure 17.2 Vibratory bowl feeder.

476 Chap. 17 / Automated Assembly Systems

Wiper blade (to wipe

upright or stacked parts
back into hopper)

Cutout (to drop

To feed cup-shaped parts facing
track down back into hopper)
Parts enter
from hopper
(a)

Rail (to reorient parts

To feed Parts enter
from flat orientation)
track from hopper
(b)

Figure 17.3 (a) Selector and (b) orientor devices used with
parts feeders in automated assembly systems.

3. Selector and/or orientor. These elements of the delivery system establish the proper
orientation of the components for the assembly work head. A selector is a device
that acts as a filter, permitting only parts in the correct orientation to pass through.
Incorrectly oriented parts are rejected back into the hopper. An orientor is a device
that allows properly oriented parts to pass through, and reorients parts that are not
properly oriented initially. Several selector and orientor schemes are illustrated in
Figure 17.3. Selector and orientor devices are often combined and incorporated into
one hopper-feeder system.
4. Feed track. The preceding elements of the delivery system are usually separated
from the assembly work head by a certain distance. A feed track moves the com-
ponents from the hopper and parts feeder to the location of the assembly work
head, maintaining proper orientation of the parts during the transfer. There are
two general categories of feed tracks: gravity and powered. Gravity feed tracks
are most common. In this type, the hopper and parts feeder are located at an
elevation above that of the work head. Gravity is used to deliver the components
to the work head. The powered feed track uses vibratory action, air pressure, or
other means to force the parts to travel along the feed track toward the assembly
work head.
5. Escapement and placement device. The escapement removes components from the
feed track at time intervals that are consistent with the cycle time of the assembly
work head. The placement device physically places the component in the correct
location at the workstation for the assembly operation. These elements are some-
times combined into a single operating mechanism. In other cases, they are two
separate devices. Several types of escapement and placement devices are pictured
in Figure 17.4.
Sec. 17.1 / Fundamentals of Automated Assembly Systems 477

Rotary indexing
table Stack of parts
Feed track
Feed track

Empty nest

Flow of
parts
Empty nest Rotary indexing
table
(a) (b)

Pick-and-place
device
Parts Feed
track

Work carriers

From
feeder
(c)
(d)
Work
carriers

From feeder

Work carriers

Pick-and-place
device
(e)

Figure 17.4 Various escapement and placement devices used in automated assembly
systems: (a) and (b) horizontal and vertical devices for placement of parts onto dial-
indexing table; (c) escapement of rivet-shaped parts actuated by work carriers; (d) and
(e) two types of pick-and-place mechanisms that use suction cups to pick up parts.

The hardware elements of the parts delivery system are illustrated schematically
in Figure 17.5. A parts selector is illustrated in the diagram. Improperly oriented parts
are returned to the hopper. In the case of a parts orientor, improperly oriented parts are
reoriented and proceed to the feed track. A more detailed description of the various ele-
ments of the delivery system is provided in Boothroyd, Poli, and Murch [3].
478 Chap. 17 / Automated Assembly Systems

Parts improperly
oriented fed
back into hopper
Selector
Hopper

Feed track Assembly

work head

Work carrier
Escapement
and
placement

Figure 17.5 Hardware elements of the parts delivery system at

an assembly workstation.

17.1.3 Applications

Automated assembly systems are used to produce a wide variety of products and subas-
semblies. Table 17.1 presents a list of typical products made by automated assembly.
Table 17.2 provides a representative list of assembly processes that are performed
on automated assembly machines. It should be noted that certain assembly processes are
more suitable for automation than others. For example, threaded fasteners (screws, bolts,
and nuts), although common in manual assembly, are a challenging assembly method to
automate. This issue is discussed in Chapter 24, which also provides some guidelines for
designing products for automated assembly.

Table 17.1   Typical Products Made by Automated Assembly

Alarm clocks Fuel injectors Pumps for household appliances
Ball bearings Gearboxes Small electric motors
Ball point pens Lightbulbs Spark plugs
Cigarette lighters Locks Wrist watches
Computer diskettes Mechanical pens and pencils
Electrical plugs and Printed circuit board
sockets assemblies

Table 17.2   Typical Assembly Processes Used in Automated Assembly Systems

Automatic dispensing of adhesive Snap fitting
Insertion of components (electronic assembly) Soldering
Placement of components (electronic assembly) Spot welding
Riveting Stapling
Screw fastening (automatic screwdriver) Stitching
Sec. 17.2 / Analysis of Automated Assembly Systems 479

17.2 Analysis of Automated Assembly Systems

This section provides mathematical models to analyze the following topics in auto-
mated assembly: (1) the parts delivery system at workstations, (2) multistation auto-
mated assembly systems, (3) single-station automated assembly systems, and (4) partial
automation.

17.2.1 Parts Delivery System at Workstations

In the parts delivery system, Figure 17.5, the parts feeding mechanism is capable of
removing components from the hopper at a certain rate f. These components are
­assumed to be randomly oriented initially, and must be presented to the selector or
orientor to establish the correct orientation. In the case of a selector, a certain propor-
tion of the components will be correctly oriented initially and these will be allowed to
pass through. The remaining components that are incorrectly oriented will be rejected
back to the hopper. In the case of an orientor, incorrectly oriented components will be
­reoriented, resulting ideally in a 100% rate of components passing through the device.
In many delivery system designs, the functions of the selector and the orientor are com-
bined. Let u be the proportion of components that pass through the selector-orientor
process and are correctly oriented for delivery into the feed track. Hence the effective
rate of delivery of components from the hopper into the feed track is fu. The remaining
proportion, 11 - u2, is recirculated back into the hopper. Obviously, the delivery rate
fu of components to the work head must be sufficient to keep up with the cycle rate of
the assembly machine.
Assuming the delivery rate of components fu is greater than the cycle rate Rc of the
assembly machine, the system needs to have a means of limiting the size of the queue in
the feed track. The usual solution is to place a sensor (e.g., limit switch or optical sensor)
near the top of the feed track to turn off the feeding mechanism when the feed track is full.
This sensor is referred to as the high-level sensor, and its location defines the active length
Lf 2 of the feed track. If the length of a component in the feed track is Lc, then the number
of parts that can be held in the feed track is nf 2 = Lf 2/Lc. The length must be measured
from a point on a given component to the corresponding point on the next component in
the queue to allow for possible overlap of parts. The value of nf 2 is the capacity of the feed
track.
Another sensor placed along the feed track at some distance from the first sen-
sor is used to restart the feeding mechanism. If the location of this low-level sensor
is defined as Lf1, then the number of components in the feed track at this point is
nf 1 = Lf 1/Lc.
The rate at which parts in the feed track are reduced when the high-level sensor
is actuated (which turns off the feeder) = Rc, which is the cycle rate of the automated
assembly work head. On average, the rate at which the quantity of parts will increase
upon actuation of the low-level sensor (which turns on the feeder) is fu - Rc. This rate of
increase will not be uniform due to the random nature of the feeder-selector operation.
Accordingly, the value of nf1 must be large enough to virtually eliminate the possibility of
a stock out after the low-level sensor has turned on the feeder. The following example il-
lustrates how these rates of decrease and increase can be used to determine the depletion
and replenishment times in the feed track.
480 Chap. 17 / Automated Assembly Systems

Example 17.1 Parts Delivery System in Automatic Assembly

The cycle time for a given assembly work head = 6 sec. The parts feeder has
a feed rate of 50 components per min. The probability that a given component
fed by the feeder will pass through the selector is u = 0.25. The number of
parts in the feed track corresponding to the low-level sensor is nf 1 = 6. The
capacity of the feed track is nf 2 = 18 parts. Determine (a) how long it will
take for the supply of parts in the feed track to go from nf 2 to nf1, and (b) how
long it will take on average for the supply of parts to go from nf1 to nf 2.
Solution: (a) Tc = 6 sec = 0.1 min. The rate of depletion of parts in the feed track
starting from nf 2 will be Rc = 1>0.1 = 10 parts>min
18 - 6
Time to deplete feed track1time to go from nf 2 to nf1 2 = Tde = = 1.2 min
10
(b) The rate of parts increase in the feed track when the low-level sensor is
reached is fu - Rc = 150210.252 - 10 = 12.5 - 10 = 2.5 parts>min
18 - 6
Time to replenish feed track1time go from nf1 to nf 2 2 = Tre = = 4.8 min
2.5

17.2.2 Multistation Assembly Machines

In this section, the operation and performance of automated assembly machines that
have several workstations are analyzed. These include dial-indexing machines, many in-
line assembly systems, and certain carousel systems. Assumptions underlying the analysis
are similar to those in the analysis of transfer lines: (1) assembly operations at the stations
have constant element times, although the times are not necessarily equal at all stations;
(2) synchronous parts transfer is used; and (3) there is no internal storage.
The analysis of an automated assembly machine with multiple stations shares much
in common with the approach used for transfer lines in Section 16.3. Some modifications
in the analysis must be made to account for the fact that components are being added at
workstations in the assembly system, whereas no components are added in transfer lines.
The operations of multistation assembly systems are depicted in Figures 17.1(a), (b), and
(c). The equations that describe these operations are based on the approach developed
by Boothroyd and Redford [2].
The typical operation at a workstation consists of a component being added and/or
joined in some fashion to an existing assembly. The existing assembly consists of a base
part plus the components assembled to it at previous stations. The base part is launched
onto the line either at or before the first workstation. The components added at each sta-
tion must be clean, uniform in size and shape, of high quality, and consistently oriented.
When the feed mechanism and assembly work head attempt to join a component that
does not satisfy this specification, the station can jam. When a jam occurs, it results in the
shutdown of the entire system until the fault is corrected. Thus, in addition to the other
mechanical and electrical failures that interrupt the operation of an automated produc-
tion line, the problem of defective components is one that specifically plagues the opera-
tion of an automatic assembly system.
Sec. 17.2 / Analysis of Automated Assembly Systems 481

The Assembly Machine as a Game of Chance. Defective parts occur in man-

ufacturing with a certain fraction defect rate q 10 … q … 1.02. In the operation of an
assembly workstation, q is the probability that the component to be added during the
current cycle is defective. When an attempt is made to feed and assemble a defective
component, the defect might or might not cause the station to jam. Let m = probability
that a defect results in a jam at the station and consequential stoppage of the line. Since
the values of q and m may be different for different stations, these terms are subscripted
as qi and mi, where i = 1, 2, cn, and n is the number of workstations on the assembly
machine.
At a particular workstation, say station i, there are three possible events that might
occur when the feed mechanism attempts to feed the next component and the assembly
device attempts to join it to the existing assembly at the station.

1. The component is defective and causes a station jam. The probability of this event is
the fraction defect rate of the parts at the station 1qi 2 multiplied by the probability
that a defect will cause the station to jam 1mi 2. This product is the same term pi as in
the previous analysis of transfer machines in Section 16.3. For an assembly machine,
pi = miqi. When the station jams, the component must be cleared and the next com-
ponent allowed to feed and be assembled. It is assumed that the probability of two
consecutive defects is very small, equal to q2i .
2. The component is defective but does not cause a station jam. This has a probability
11 - mi 2qi. With this outcome, a bad part is joined to the existing assembly, per-
haps rendering the entire assembly defective.
3. The component is not defective. This is the most desirable outcome and the most
likely by far (it is hoped). The probability that a part added at the station is not
­defective is equal to the proportion of good parts 11 - qi 2.

The probabilities of the three possible events must sum to unity for any workstation;
that is,
miqi + 11 - mi 2qi + 11 - qi 2 = 1 (17.1)
For the special case where mi = m and qi = q for all i, this equation reduces to the
following:
mq + 11 - m2q + 11 - q2 = 1 (17.2)
Although it is unlikely that all mi are equal and all qi are equal, the equation is nev-
ertheless useful for computation and approximation purposes.
To determine the complete distribution of possible outcomes that can occur on an
n-station assembly machine, the terms of Equation (17.1) are multiplied together for all
n stations:

q 3 miqi + 11 - mi 2qi + 11 - qi 2 4 = 1
n
(17.3)
i=1

In the special case where mi = m and qi = q for all i, this reduces to

[mq + 11 - m2q + 11 - q2]n = 1 (17.4)
Expansion of Equation (17.3) reveals the probabilities for all possible sequences of events
that can take place on the n-station assembly machine. Regrettably, the number of terms
482 Chap. 17 / Automated Assembly Systems

in the expansion becomes very large for a machine with more than two or three stations.
The exact number of terms is equal to 3n, where n = number of stations. For example,
for an eight-station line, the number of terms = 38 = 6561, each term ­representing the
probability of one of the 6,561 possible outcome sequences on the assembly machine.

Measures of Performance. Fortunately, it is not necessary to calculate every

term to use the description of assembly machine operation provided by Equation (17.3).
One of the performance characteristics of interest is the proportion of assemblies that
contain one or more defective components. Two of the three terms in Equation (17.3)
represent events in which a defective component is not added at the given station. The
first term is miqi, which indicates that a station jam has occurred, preventing a defec-
tive component from being added to the existing assembly. The other term is 11 - qi 2,
which means that a good component has been added at the station. The sum of these
two terms represents the probability that a defective component is not added at station i.
Multiplying these probabilities for all stations provides the proportion of acceptable
product coming off the line:

Pap = q 11 - qi + miqi 2
n
(17.5)
i=1

where Pap can be thought of as the yield of good assemblies produced by the assembly
machine. If Pap = the proportion of good assemblies, then the proportion of assemblies
containing at least one defective component Pqp is given by

Pqp = 1 - Pap = 1 - q 11 - qi + miqi 2

n
(17.6)
i=1

In the case of equal mi and equal qi, these two equations become, respectively,
Pap = 11 - q + mq2 n (17.7)

Pqp = 1 - 11 - q + mq2 n (17.8)

The yield Pap is an important performance metric of an assembly machine. To have a cer-
tain proportion of assemblies with one or more defective components in the final output
is a significant problem. These assemblies must be identified by inspection and sortation,
or they will be mixed in with the good assemblies, which could lead to undesirable conse-
quences when the assemblies are placed in service.
Other performance measures of interest are the machine’s production rate, the pro-
portion of uptime and downtime, and the average cost per unit produced. To calculate
production rate, the frequency of downtime occurrences per cycle F is first determined.
If each station jam results in a machine downtime occurrence, F is found by summing the
expected number of station jams per cycle:

F = a pi = a miqi
n n
(17.9)
i=1 i=1

In the case of a station performing only a joining or fastening operation and not adding a
part at the station, then the contribution to F made by that station is pi, the probability of
a station breakdown, where pi does not depend on mi and qi.
Sec. 17.2 / Analysis of Automated Assembly Systems 483

If mi = m and qi = q for all stations, i = 1, 2, p , n, then the above equation for F

reduces to the following:
F = nmq (17.10)
The average actual production time per assembly is given by

Tp = Tc + a miqiTd
n
(17.11)
i=1

where Tc = ideal cycle time of the assembly machine, which is the longest assembly task
time on the machine plus the indexing or transfer time, min; and Td = average downtime
per occurrence, min. For the case of equal mi and qi,
Tp = Tc + nmqTd (17.12)
The production rate is the reciprocal of average actual production time:
1
Rp = (17.13)
Tp

This is the same relationship as Equation (16.9) for transfer lines. However, the operation
of assembly machines is different from processing machines. In an assembly machine,
­unless mi = 1.0 for all stations, the production output will include some assemblies with
one or more defective components. Accordingly, the production rate should be corrected
to give the rate of acceptable product, that is, those that contain no defects. This is simply
the yield Pap multiplied by the production rate

q 11 - qi + miqi 2
n

Pap i=1
Rap = PapRp = = (17.14)
Tp Tp

where Rap = production rate of acceptable product, units/min. When all mi are equal and
all qi are equal, the corresponding equation is
Pap 11 - q + mq2 n
Rap = PapRp = = (17.15)
Tp Tp
Equation (17.13) gives the production rate of all assemblies made on the system, includ-
ing those that contain one or more defective parts. Equations (17.14) and (17.15) give
production rates for good product only. The problem still remains that the defective
products are mixed in with the good units. This issue of inspection and sortation is con-
sidered in Chapter 21.
Line efficiency is calculated as the ratio of ideal cycle time to average actual produc-
tion time. This is the same ratio as defined in Chapter 16, Equation (16.11),
Rp Tc
E = = (17.16)
Rc Tp

where Tp is calculated from Equation (17.11) or Equation (17.12). The proportion down-
time D = 1 - E, as before. No attempt has been made to correct line efficiency E for the
yield of good assemblies. The efficiency of the assembly machine and the quality of units
produced by it are treated here as separate issues.
484 Chap. 17 / Automated Assembly Systems

On the other hand, the cost per assembled product must take account of the output
quality. Therefore, the general cost formula given in Equation (16.14) in the previous
chapter must be corrected for yield, as
Cm + CoTp + Ct
Cpc = (17.17)
Pap

where Cpc = cost per good assembly, $>pc; Cm = cost of materials, which includes the
cost of the base part plus components added to it, $>pc; Co = operating cost of the assem-
bly system, $>min; Tp = average actual production time, min>pc; Ct = cost of disposable
tooling, $>pc; and Pap = yield from Equation (17.5). The effect of the denominator is to
increase the cost per assembly; as the quality of the individual components deteriorates,
the average cost per good quality assembly increases.
In addition to the traditional ways of indicating line performance (production rate,
line efficiency, cost per unit), there is the additional metric of yield. While the yield of
good product is an important issue in any automated production line, it can be explicitly
included in the formulas for assembly machine performance by means of q and m.

Example 17.2 Multistation Automated Assembly System

A 10-station in-line assembly machine has an ideal cycle time = 6 sec. The
base part is automatically loaded prior to the first station, and components are
added at each of the stations. The fraction defect rate at each of the 10 stations
is q = 0.01, and the probability that a defect will jam is m = 0.5. When a jam
occurs, the average downtime is 2 min. Cost to operate the assembly machine
is $42.00/hr. Other costs are ignored. Determine (a) average production rate of
all assemblies, (b) yield of good assemblies, (c) average production rate of good
product, (d) uptime efficiency of the assembly machine, and (e) cost per unit.
Solution: (a) Tc = 6 sec = 0.1 min. The average production cycle time is Tp = 0.1 +
1102 10.5210.01212.02 = 0.2 min. The production rate is therefore
60
Rp = = 300 total assemblies/hr
0.2
(b) The yield is given by Equation (17.7):
Pap = 51 - .01 + 0.510.0126 10 = 0.9511
(c) Average actual production rate of good assemblies is determined by
Equation (17.15):
Rap = 30010.95112 = 285.3 good assemblies/hr
(d) The efficiency of the assembly machine is
E = 0.1/0.2 = 0.50 = 50%
(e) Cost to operate the assembly machine Co = $42/hr = $0.70/min
Cpc = 10.70/min210.2 min/pc2/0.9511 = $0.147/pc
Sec. 17.2 / Analysis of Automated Assembly Systems 485

Example 17.3 Effect of Variations in q and m on Assembly System Performance

This example shows how the performance measures in Example 17.2 are affected
by variations in q and m. First, for m = 0.5, determine the production rate, yield,
and efficiency for q = 0, q = 0.01, and q = 0.02. Second, for q = 0.01, deter-
mine the production rate, yield, and efficiency for m = 0, m = 0.5, and m = 1.0.
Solution: Computations similar to those in Example 17.2 provide the following results:

q m Rp (pc/hr) Yield Rap (pc/hr) E Cpc

0 0.5 600 1.0 600 100% $0.07
0.01 0.5 300 0.951 285 50% $0.15
0.02 0.5 200 0.904 181 33.3% $0.23
0.01 0 600 0.904 543 100% $0.08
0.01 0.5 300 0.951 285 50% $0.15
0.01 1.0 200 1.0 200 33.3% $0.21

The results of Example 17.3 show that as fraction defect rate q increases (meaning
that component quality gets worse) all five measures of performance suffer. Production
rate drops, yield of good product is reduced, proportion uptime decreases, and cost per
unit increases.
The effect of m (probability that a defect will jam the work head and cause the as-
sembly machine to stop) is less obvious. At low values of m 1m = 02 for the same com-
ponent quality level 1q = 0.012, production rate and machine efficiency are high, but
yield of good product is low. Instead of interrupting the assembly machine operation and
causing downtime, all defective components pass through the assembly process to become
part of the final product. At m = 1.0, all defective components are removed before they
become part of the product. Therefore, yield is 100%, but removing the defects takes
time, adversely affecting production rate, efficiency, and cost per unit.

17.2.3 Single-Station Assembly Machines

The single-station assembly system is depicted in Figure 17.1(d). It consists of a single

work head, with several components feeding into the station to be assembled to a base
part. Let ne = the number of distinct assembly elements that are performed on the
­machine. Each element has an element time, Tej, where j = 1, 2, p , ne. The ideal cycle
time for the single-station assembly machine is the sum of the individual element times of
the assembly operations to be performed on the machine, plus the handling time to load
the base part into position and unload the completed assembly. The ideal cycle time can
be expressed as

Tc = Th + a Tej
ne
(17.18)
j=1

where Th = handling time, min.

Many of the assembly elements involve the addition of a component to the existing
subassembly. As in the analysis of multiple-station assembly, each component type has a
486 Chap. 17 / Automated Assembly Systems

certain fraction defect rate, qj, and there is a certain probability that a defective compo-
nent will jam the workstation, mj. When a jam occurs, the assembly machine stops, and
it takes an average Td to clear the jam and restart the system. The inclusion of downtime
resulting from jams in the machine cycle time gives

Tp = Tc + a qj mj Td
ne
(17.19)
j=1

For elements that do not include the addition of a component, the value of qj = 0 and mj
is irrelevant. This might occur, for example, when a fastening operation is performed with
no part added during element j. In this type of operation, a term pjTd would be included in
the above expression to allow for a downtime during that element, where pj = the prob-
ability of a station failure during element j. For the special case of equal q and equal m
values for all components added, Equation (17.19) becomes
Tp = Tc + nmqTd (17.20)
Determining yield (proportion of assemblies that contain no defective components) for
the single-station assembly machine makes use of the same equations as for the multiple
station systems, Equations (17.5) or (17.7). Uptime efficiency is computed as E = Tc >Tp
using the values of Tc and Tp from Equations (17.18) and (17.19) or (17.20).

Example 17.4 Single-Station Automatic Assembly System

A single-station assembly machine performs five work elements to assemble
four components to a base part. The elements are listed in the table below,
together with the fraction defect rate (q) and probability of a station jam (m)
for each of the components added (NA means not applicable).

Element Operation Time (sec) q m p

1 Add gear 4 0.02 1.0
2 Add spacer 3 0.01 0.6
3 Add gear 4 0.015 0.8
4 Add gear and mesh 7 0.02 1.0
5 Fasten 5 0 NA 0.012

Time to load the base part is 3 sec and time to unload the completed ­assembly
is 4 sec, giving a total load/unload time of Th = 7 sec. When a jam ­occurs, it
takes an average of 1.5 min to clear the jam and restart the machine. Determine
(a) production rate of all product, (b) yield of good product, (c) production
rate of good product, and (d) uptime efficiency of the assembly machine.
Solution: (a) The ideal cycle time of the assembly machine is
Tc = 7 + 14 + 3 + 4 + 7 + 52 = 30 sec = 0.5 min
Frequency of downtime occurrences is
F = 0.0211.02 + 0.0110.62 + 0.01510.82 + 0.0211.02 + 0.012 = 0.07
Sec. 17.2 / Analysis of Automated Assembly Systems 487

Adding the average downtime due to jams,

Tp = 0.5 + 0.0711.52 = 0.5 + 0.105 = 0.605 min
Production rate is therefore Rp = 60/0.605 = 99.2 total assemblies /hr
(b) Yield of good product is the following, from Equation (17.5):
Pap = 51 - 0.02 + 1.010.022 651 - 0.01 + 0.610.0126
51 - 0.015 + 0.810.0152651 - 0.02 + 1.010.0226
= 11.02 10.996210.997211.02 = 0.993
(c) Production rate of only good assemblies is
Rap = 99.210.9932 = 98.5 good assemblies /hr
(d) Uptime efficiency is
E = 0.5/0.605 = 0.8264 = 82.64%

As the analysis suggests, increasing the number of elements in the assembly ­machine
cycle results in a longer cycle time, decreasing the production rate of the machine.
Accordingly, applications of a single-station assembly machine are limited to lower ­volume,
lower production rate situations. For higher production rates, one of the multistation
­assembly systems is generally preferred.

17.2.4 Partial Automation

Many assembly lines in industry contain a combination of automated and manual work-
stations. These cases of partially automated production lines occur for two main reasons:

1. Automation is introduced gradually on an existing manual line. Suppose demand for

the product made on a manually operated line increases, so the company decides to
increase production and reduce labor costs by automating some or all of the stations.
The simpler operations are automated first, and the transition toward a fully auto-
mated line is accomplished over a long period of time. Until then, the line operates as
a partially automated system. (See Automation Migration Strategy, Section 1.4.3.)
2. Certain manual operations are too difficult or too costly to automate. Therefore,
when the sequence of workstations is planned for the line, certain stations are
­designed to be automated while the others are designed as manual stations.

Examples of operations that might be too difficult to automate are assembly pro-
cedures or processing steps involving alignment, adjustment, or fine-tuning of the work
unit. These operations often require special human skills and/or senses to carry out. Many
inspection procedures also fall into this category. Defects in a product or part that can be
easily perceived by a human inspector are sometimes difficult for an automated inspec-
tion device to detect. Another problem is that the automated inspection device can only
check for the defects for which it was designed, whereas a human inspector is capable of
sensing a variety of unanticipated imperfections and problems.
488 Chap. 17 / Automated Assembly Systems

To analyze the performance of a partially automated production line, the follow-

ing assumptions are made: (1) workstations perform either processing or assembly opera-
tions, (2) processing and assembly times at automated stations are constant, though not
necessarily equal at all stations, (3) the system uses synchronous transfer of parts, (4) the
system has no internal buffer storage, and (5) station breakdowns occur only at automated
stations. Breakdowns do not occur at manual stations because the human workers are
flexible enough, it is assumed, to adapt to the kinds of disruptions and malfunctions that
would interrupt the operation of an automated workstation. For example, if a human op-
erator were to retrieve a defective part from the parts bin at the station, the worker would
immediately discard the part and select another without much lost time. Of course, this
assumption of human adaptability is not always correct, but the analysis is based on it.
The ideal cycle time Tc is determined by the slowest station on the line, which is
generally one of the manual stations. If the cycle time is determined by a manual station,
then Tc will exhibit variability, simply because there is random variation in any repetitive
human activity. However, it is assumed that the average Tc remains constant over time.
Given the assumption that breakdowns occur only at automated stations, let na = the
number of automated stations and Td = average downtime per occurrence. For the auto-
mated stations that perform processing operations, let pi = the probability (frequency)
of breakdowns per cycle, and for automated stations that perform assembly operations,
let qi and mi equal, respectively, the defect rate and probability that the defect will cause
station i to stop. The average actual production time can now be defined as:

Tp = Tc + a piTd (17.21)
i ∈ na

where the summation applies to the na automated stations only. For those automated sta-
tions that perform assembly operations in which a part is added,
pi = miqi
If all pi, mi, and qi are equal, respectively, to p, m, and q, then the preceding equations
reduce to
Tp = Tc + na pTd (17.22)
and p = mq for those stations that perform assembly consisting of the addition of a part.
Given that na is the number of automated stations, then nw = the number of sta-
tions operated by workers, and na + nw = n, where n = the total station count. Let
Casi = cost to operate automatic workstation i, $/min; Cwi = cost to operate manual
workstation i, $/min; and Cat = cost to operate the automatic transfer mechanism. Then
the total cost to operate the line is given by

Co = Cat + a Casi + a Cwi (17.23)

i ∈ na i ∈ nw

where Co = cost of operating the partially automated production system, $/min. For all
Casi = Cas, and all Cwi = Cw, then
Co = Cat + naCas + nwCw (17.24)
Now the total cost per unit produced on the line can be calculated as
Cm + CoTp + Ct
Cpc = (17.25)
Pap
Sec. 17.2 / Analysis of Automated Assembly Systems 489

where Cpc = cost per good assembly, $/pc; Cm = cost of materials and components being
processed and assembled on the line, $/pc; Co = cost of operating the partially automated
production system by either of Equations (17.23) or (17.24), $/min; Tp = average actual
production time, min/pc; Ct = any cost of disposable tooling, $/pc; and Pap = proportion
of good assemblies by Equations (17.5) or (17.7).

Example 17.5 Partial Automation

The company is considering replacing one of the current manual workstations
with an automatic work head on a 10-station production line. The current line
has six automatic stations and four manual stations. Current cycle time is 30 sec.
The limiting process time is at the manual station that is proposed for replace-
ment. Implementing the proposal would allow the cycle time to be reduced to
24 sec. The new station would cost $0.20/min. Other cost data: Cw = $0.15/min,
Cas = $0.10/min, and Cat = $0.12/min. Breakdowns occur at each automated
station with a probability p = 0.01. The new automated ­station is expected to
have the same frequency of breakdowns. Average downtime per occurrence
Td = 3.0 min, which will be unaffected by the new station. Material costs and
tooling costs will be neglected in the analysis. It is desired to compare the cur-
rent line with the proposed change on the basis of production rate and cost per
piece. Assume a yield of 100% good product.
Solution: For the current line, Tc = 30 sec = 0.50 min
Tp = 0.50 + 610.01213.02 = 0.68 min and Rp = 1/0.68 = 1.47 pc/min = 88.2 pc/hr
Co = 0.12 + 410.152 + 610.102 = $1.32/min and Cpc = 1.3210.682 = $0.898/pc
For the proposed line, Tc = 24 sec = 0.4 min
Tp = 0.40 + 710.01213.02 = 0.61 min and Rp = 1/0.61 = 1.64 pc/min = 98.4 pc/hr
Co = 0.12 + 310.152 + 610.102 + 110.202 = $1.37/min and Cpc = 1.3710.612 = $0.836/pc
Even though the line would be more expensive to operate per unit time, the
proposed change would increase production rate and reduce piece cost.

17.2.5 What the Equations Tell Us

The equations derived in this section reveal several practical guidelines for the design and
operation of automated assembly systems and the products made on such systems.

• The parts delivery system at each station must be designed to deliver components
to the assembly operation at a net rate (parts feeder multiplied by pass-through
proportion of the selector/orientor) that is greater than or equal to the cycle rate of
the assembly work head. Otherwise, assembly system performance is limited by the
parts delivery system rather than the assembly process technology.