Documente Academic
Documente Profesional
Documente Cultură
of
production / Job shop Manufacturing cells Transfer line
Characteristic
2
Course programme
Analysis of the characteristics, scope, design, management and
organizational issues of the different configurations of production
systems.
Criteria for the design of manufacturing systems (job shop,
manufacturing cells, transfer lines).
Criteria for the design of manual assembly systems (fixed position
assembly, paced and unpaced lines, continuous flow lines, multi-
model and mixed-model lines, assembly shop).
Criteria for the design of automated assembly systems (lines,
flexible assembly systems).
Criteria for the design of process plants.
Analysis of some industrial technologies and related management
and organizational issues.
3
Course structure
M1 Introduction M2 Design of
to production manufacturing
systems systems
M3 Design of M4 Process
assembly plants and
systems industries
4
Teaching method
5
Course schedule
Time schedule
Monday 8.15-12.15
Friday 11.15-15.15
6
Exam
7
Exam Post Scriptum
8
References
9
Contacts
Prof. Dr. Sergio Terzi
Room: DIG 2.09 (second floor)
Phone number: 02-23992803
E-mail: sergio.terzi@polimi.it
Meeting: Friday, 09:30-11:00, based on request
10
Introduction to production systems
Tools
2
How production systems work
3
How production systems work
4
How production systems work
5
How production systems work
6
How production systems work
Production
systems
Manufacturing
Process plants
plants
Continuos Parts
Assembly
flow production
Fixed
Job shops positions
Batch
Assembly
Cells shops
Transfer Assembly
lines lines
7
How production systems work
8
How production systems work
9
How production systems work
Continuous
Batch /
Intermittent
High vol.
- Transfer line
Quantity of each product
Single
Medium vol.
- line / cell
11
The concept of lead time
12
The concept of lead time
Production Lead
Time
13
The concept of lead time
MATERIAL QUEUING BEFORE WORK SYSTEM Trial run for WAITING AFTER
PROCESSING SET UP
PROCESSING PROCESSING
HANDLING processing
May also include quality control, setting up CNC programmes,
machine breakdown, waiting for repair, repair, operator breaks,
quality issues, shortages of parts, detailed planning and control,
etc.
The concept of lead time
TIME
(DD - DO) Total Lead Time
15
Performances of Production Systems
16
Performances of Production Systems
A B C D
2 6 5 3
6 0,17
5 0,20
3 0,33
17
Performances of Production Systems
A B C D
2 6 5 3
18
Performances of Production Systems
A B C D
2 6 5 3
q LT min = 2 + 6 + 5 + 3 = 16 minutes
q Whichwill be the highest production rate (or
throughtput) achievable in such a line?
19
Performances of Production Systems
A B C D
2 6 5 3
20
Performances of Production Systems
21
Performances of Production Systems
Honeywell make
controls to order and
dispatch on the same
day for some products
22
Performances of Production Systems
23
Performances of Production Systems
Lower WIP
Less uncertainty
Easier to control
Better delivery
performance
24
Performances of Production Systems
26
Performances of Production Systems
Babcock BES refit warships.
They are also working on
Heathrow Terminal 5
27
Performances of Production Systems
Line Very short Low Variable Very high Very low Risky
Performances of Production Systems
Percentage of
waste intercepted
by quality control
= internal
Production System performance of
interest for the
firm
Level of machines
obsolescence =
Delivery lead time to operative condition
produce a product with
a quality ok = external
performance that can
be measured by the
client
30
Costs of Production Systems
31
Costs of Production Systems
32
Costs of Production Systems
q Installation costs in CAPEX
Feasibility study (preliminary economic analysis of the project)
Development of the project
Acquisition of the ground;
Building construction
Installation of plant services
Acquisition and installation of machinery and equipment
Intangible cost of knowledge assets (know-how, patent acquisition, payment of
royalties)
Interest payable on any mortgages or loans for investment
33
Costs of Production Systems
35
Costs of Production Systems
36
Costs of Production Systems
37
Costs of Production Systems
38
How to evaluate an investment in production
q Cost analysis
q Revenue / savings analysis
q Cash Flow analysis
q Discounted Cash Flow Analysis
q IRR / Payback time
39
Thought questions
3
A common example of a traditional industry in which interference problems have been
studied is the textile industry. It is the case of the weaving shed in a textile mill,
where a worker may attend several looms, depending on the type of loom installed and
the characteristics of the cloth being woven. The operator/worker's tasks include
changing the supply cone holding the yarn when it is empty, tying together strands of
yarn that break, etc. Occasionally, many looms may be stopped simultaneously
because the operator/worker can attend one machine at a time, and the machine
interference is the subsequent result (remark: the weaving shed in a textile mill is a
functional department in a job-shop).
Interference also exists in more modern job shops, especially automated
manufacturing systems with computer numerically controlled (CNC) machine tools.
In this specific case, the operator/workers tasks normally include loading / unloading
materials to / from the CNC machine (this is another example of the workers role in a
functional department in a job-shop).
Another case can be, e.g., again in the context of metal working processes > think e.g.
of a transfer line starting its operations from coils of rolled sheet metal and, after
some workstations along the line, ending with finished product (after automatic cut
operations, etc.). Similarly to other cases, there is the need for the man. In this case,
typically the man is required to load other coils because the previous one has been
consumed (i.e. loading task for the worker/operator assigned to the transfer line).
The SCAMP case can be another example (see previous lecture on manufacturing
cells), especially relating to minor stoppages / interruptions > if there is some
problem (problem at a raw material, causing the minor stoppage), each workstation can
be independently managed > the repairman can make an intervention in order to fix the
problem; then, the workstation is restarted (remark: the intervention is of course carried
out in safe conditions > the robot is stopped and the workstation is down, independently
from others, i.e. after safely turning off the power to just the segmented circuit of the
single workstation, until the workstation itself is restarted).
Generalizing, in all the above mentioned cases, considering its utilization rate, the
worker/repairman is partly in a standby state (i.e. machine should be the bottleneck),
until operations (i.e. unmanned operations) at a workstation are completed / ends up
(also due to unexpected events) >>> when this happens, it requires a service from the
operator / worker (i.e. a service request is issued to the worker/repairman and some
delays / waiting time may occur waiting for the delivery of such a service > not to say,
but to have in mind: if the worker is very much utilized the waiting time of the machine
would be high > worth remark to avoid this risk).
3
Problem setting occurrence of the worker-induced machine interference
> analyzing this problem can be considered as part of the design of a
manufacturing system
4
some operations are finished/triggered at a machine/workstation.
Problem setting questions for the analysis and design of the manufacturing system,
having a specific view on how operators are interacting (and interfering) with machines /
workstations:
In the remainder some basic questions are shortly discussed > questions Q1-Q2 can
represent the final aims of the design study, and they can be answered through
answering to Q3, at analysis step, that is: (Q3) How much is the occurrence of operator-
induced machine interference affecting the capacity required to meet demand? ---> (Q1)
How many operators do we need to attend machines? + (Q2) Where are the
bottlenecks: machines or operators?
(Q1) How many operators do we need to attend machines? (> implicitly related to
Q3 analysis step, @ background)
In order to answer this question, two are the main entities that should be
studied/analyzed: manual work operations and machine operations (+ their
relationships) > the operating cycles of respective resources (workers and
machines).
Some data are needed to characterize these entities. In particular, the amount of
manual work contained within the overall machine operating cycle can be defined
as first rough time work study > a rough first indication can be calculated by means of
a workload / load factor > this can be defined as a proportion / share of the overall
machine operating cycle time when the worker is required to carry out at standard
performance (standard performance of own task he / she has been assigned), otherwise
the machine cannot resume production.
A numerical example is worth for discussion: overall machine operating cycle time is
10 minutes (remark: the operating cycle is not only the processing cycle, it is also
including setups, load time, etc.); the amount of manual work 1 minute (@ standard
task performance) > Workload / load factor = 1/10 = 10 % > the reciprocal of the load
factor could give the number of machines that can be theoretically attended by the
worker (leading to a first rough assignment of workers to machines) > however, this
is really a rough analysis > other factors should be taken into account. Is this analysis
enough? Not yet, of course.
This is not enough. The following reasoning helps to motivate that it is not enough.
Sometimes, it could happen that the workers tasks consist only of i) unloading finished
work-pieces from machines which have stopped automatically; ii) loading fresh work-
pieces and iii) restarting the machines > this is clearly a simple workers operating
cycle. Further on, the system may be composed by machines that are alike and/or are
working exactly similar/same work-pieces. In these specific conditions, it may be
possible to achieve an ideal sequence of operations (within the operating cycles),
wherein the worker is capable to operate the theoretical number of machines (indicated
4
by the reciprocal of the load factor) w/o causing particular delays / waiting time, if not
none at all. But this is an ideal case.
Much more commonly, differences occurs in the machines or in work-pieces being
processed. Moreover, frequently attention has to be given when the machines are
concurrently running, also because tasks assigned to the workers are not just part of a
simple operating cycle (different operating cycles are superimposing, thus
complicating the dynamics of the operator) > the result is that the worker cannot
always get to a machine at the exact moment when attention / task is needed, thus
causing interference time.
Concluding, a detailed study of the operating cycles (workers machines) is
required to understand the effective number of operators that should be considered.
This should be done when designing / configuring a system, at least as a prospect, with
more details than studying a simple work load / load factor.
(Q2) Where are the bottlenecks: machines or operators? (> implicitly related to Q3
analysis step, @ background)
4
Method for interference analysis introduction to a simple method
5
different operations is done for workers and machines, also the utilization rate of
workers and machines can be calculated.
A worker is assigned to three machines > the activity cycle diagrams show a reference
scenario, which is now treated as determined scenario (at the moment, as a
deterministic scenario as first hypothesis).
(looking at the specific symbols used in the diagrams, in order to interpret it) grey bars
represent the operating time when the machine is occupied (processing) w/o the need of
the presence of the worker + red lines represent the operating time when the machine
and the worker are both occupied (in this case, main activities may be i) unloading
finished work-pieces from machines which have stopped automatically; ii) loading
fresh work-pieces and iii) restarting the machines).
Interference analysis > herein, there are simple operating cycles, characterized by
the same processing times @ machines of same types + only a load / unload time
required by the worker > the best schedule results is the sequence herein shown (>
ideal sequence) > (in this case) this is resulting in the occurrence of no interference
time.
If we look only at these activities as those realized to the plant, and considers the grey
bar length as 10 units of time and the red lines 2 units, the total length of the schedule
is 28 units, the utilization rate of the single worker is about 43 % (12/28), the
utilization rate of the machines is on average around 86 % (considering a gross
utilization, comprised of the load / unload activities and restart) > bottleneck is the
machine.
(Extending the concept of work load factor of the operator), it is worth remarking that,
during the operating cycle of the machines, in particular during the share of operating
time when the operator is not required (around 57 % in the example), the operator
could make other activities, if free > these then lead to his / her overall utilization
rate >>> e.g. he/she could be preparing some raw material to be processed by the
machine (some material handling tasks are assigned to him / her) >>> such activities
are not directly related to the processing cycle of the machine (still on going), but
indirectly (as a preparation, etc. ): when the operator is directly required by the
machine, we normally assume that a priority is given to serve the machine service
demand (so to the activities directly related to the machines and shown in the red bars).
Normally, as already commented, it is wise to keep a low utilization rate of the
workers, in order to avoid that the machines are waiting too much and, in the worst
case, the bottleneck becomes the worker.
5
Method for interference analysis example 2
The activity cycle diagrams refer now to a new scenario, not as simple as
example #1 (i.e. different processing times @ machines maybe of
different types + not only a load / unload time required by the worker but
also another task related to some unexpected event > fixing problems
include apart the fixing itself any reloading if required) > by means of
this example, we do not want to make a comparison b/w example #1 and #2,
instead we do want to remark more general issues.
First of all, it is worth remarking that all the issues above cited (as different /
varying processing times, frequencies of minor problems to fix, etc.) are all
operating conditions leading to deviations from a determined / ideal
scenario > in fact, we can consider that real operating scenarios are affected
by some stochastic variation of one or more variables (as a more real
assumption), e.g. stochastic occurrence of minor problems, stochastic
durations of some activities (also due to changes in the production mix, a part
in the operations itself on the machines).
(in the specific case of this example #2) the sequence of operations herein
resulting leads to the occurrence of some interference time (see the green
lines) > overall, lets estimate 4 units as the time losses over all machines:
roughly speaking, 1,5 (M1) + 2,5 (M2) +0,5 (M3) units > by distributing on
average these losses, we can conclude that 1,33 unit is lost on each machine
as interference time > there is then a capacity reduction due to these losses on
6
all machines.
Utilization can be now calculated > utilization rate of the worker is 19/29 = about 66
% > interference time is subsequent to this high utilization rate; the utilization rate of
the machines is also subsequent, also related to interference times (no matter now the
exact value). There is another general remark herein: high utilization worker > risk of
high interference time > reduction of production capacity (and utilization for real
processing) of machines.
6
State space method introduced, firstly, as a generic method for logical
analysis of the function(s) played by a system of interest/SoI, the method is
then applied to the machine interference problem.
7
estimate of the expected interference time: in fact, the method is an approximated
method, applicable by using simple spreadsheets but with some limiting assumption.
One limiting assumption is the fact that each machine is working independently of
the others, hence it will be requiring a worker or not, issueing the service request
independently of the others > this is not exactly true, being each machine / worker
scheduled by a unique master scheduling process, and the service requests arise
naturally as a consequence of this master schedule; hence, some how, we assume that it
is a rough approximation acceptable for a first indication. Nonetheless, some issues
/ phenomena are also naturally independent ones, hence the approximation is not that
bad in their regards: e.g., lets consider the random problems that may occur at
machines leading to minor stoppages, these are clearly leading to more independent
service requests (e.g. they depend on the specific material / item that is machined, and
random variations of materials flowing through each machine may lead to the
characteristics of independency remind but not to say: this is of course expect the
case that there is a clear trend / on average common to all materials, e.g. when there is
a change in the supply which leads to bad quality characteristics on average of the
supplied materials which affect all machines not independently).
7
State space method > state space table
(herein, it is applied as demo to a functional department in a job-shop system).
8
3), that were obtained by scanning all the possible combinations, ranging from the case
that all machines are issueing a service request (si,j elements of the matrix equal to
zero) to no machine is issuing a service request (si,j elements of the matrix equal to
one).
More precisely, in order to make this scanning and then fulfill the state space table, the
states of the system come out from all the combinations of states of the machines >
formula of combinations (w/o repetitions) used to calculate the number of
combinations of k machines out of n installed machines > this enables to enumerate the
number of states where k = 0, 1, 2, 3 service requests are held by the manufacturing
system to the worker (remind that the combinations w/o repetition can be written also
using factorials {n!}{k!(n-k)!}).
Generalization
As we will see, the state space method / table is a flexible modeling method, hence
applicable to different types of systems (not only job-shop/cells, also transfer lines).
8
State space method as a basis for deriving a probabilistic model, necessary
for a quantitative evaluation
The state space method prepares the ground for a quantitative evaluation. In
this regard, now the question is what is the probability that the service
demand/request is issued by one or more machines (this is a follow up of
the different combinations already expressed in the state space table)? This
means to compute, by means of a probabilistic model, first of all, the
probability that Machine Mj reaches state Sj = 0 (requiring a worker) at
time t, therefore the probability at manufacturing system level (probability
that 0, 1 n service requests are issued) > In this concern, overall, the
probability at system level results by simply scanning the state space table /
different combinations.
More precisely, we exploit simple rules of probabilistic calculus:
1. each machine is working independently of the others >
independence of the events > rule: P (A AND B) > P(A) x P(B) > in the
state space table, e.g., P(S1 = 1 AND S2 = 1 AND S3 =0) > P(S1 = 1) x
P(S2 = 1) x P(S3 =0) > so we can obtain the probability of a given
state of the system (in the matrix, along one row i);
2. states of the system are mutually exclusive, that is: the system may
reach one and only one of the 8 states shown in the state space table
(e.g. the state 111 is different from 011) > P (A XOR B) > P(A) +
P(B) > in the state space table, e.g., we can then calculate the sum of
9
all mutually exclusive states having one common function e.g. 1 service
request > P(1 service request) = P (((S1 = 1) AND P(S2 = 1) AND P(S3 =0))
XOR ) > sum +
Once the probability has been calculated, this can be used to estimate the expected
value for the interference time: in fact, after knowing the probability to be in a given
state i (number of service requests issued by the manufacturing system), the next step is
to model/estimate the impact correspondent to that given state of the system > we
then define a Yi: correspondent to each row, we have a performance Yi expressing, in
this case of machine interference problem, the impact as interference time
correspondent to a state i.
9
Summarizing the procedure:
1. The state space table is the first matrix, built in order to enumerate all the
combinations of machines states (to obtain the states of the system i) and,
subsequently, to define a system function (for each given state of the system
i);
2. correspondingly, a second matrix, table of probabilities, is derived: therein,
the elements pi,j are expressing the probabilities of the correspondent
elements in the state space table (pi,j of a machine j in a given state of the
system i, result of all the combinations);
3. within this second matrix, we derive the probability pi of the state i of the
system, exploiting the independence of working states / events of machines
(i.e. multiplication of pi,j); remarkable to say that the size of the matrix
depends on the number of machines installed in the manufacturing
system, that is: n machines > 2 raised to the power n are the number of states
reachable by the system / the rows of the matrix; therefore, of course,
summing up through all the states of the system / through all the rows leads to
a probability of 1 (all events/states may happen), which is subsequent to the
issue of mutual exclusiveness of different states of the system.
10
Completing the summary of the procedure:
4. Y is the performance of interest and depends on the state of the system i >
it changes depending on the system function/state i: if we have more service
requests issued at a given state i, we expect more interference time; a
simplified model will be used (and presented later in next highlight) in order
to estimate the expected value of the performance at state of the system i;
5. Each state of the system i has its own probability pi, so this is simply used as
probabilistic weight to calculate the overall expectation, related to all m
states in the matrix;
6. Calculation of the production capacity of the manufacturing
system will be a follow up, as E[Y] is an expected time loss,
hence inducing a capacity reduction;
7. Calculation of the utilization rate of workers and machines is also
under consideration, as we know E[Y], and this is a relevant time
component to be included in the time calendar where we will
calculate the utilization of the two resources ( not say but remind
example #2 / scheduling operating cycles of workers and machines).
11
Quick comment on the use of the operating times when the worker is required
or not required, as data in order to provide a rough estimate one could find a
machine requiring or not the worker attendance in a given time interval.
12
Quick comment on the use of the number of times when the worker is
required or not required, as data in order to provide a rough estimate one
could find a machine requiring or not the worker attendance in a given time
interval.
Still based on the Bernoulli distribution > in this case, the driver to estimate
the probability of success / failure is the number of times/occurrences: e.g.
number of materials causing blockage to a workstation (due to quality
characteristics out of tolerance for being machined) with respect to the total
number of materials flowing through the machine > i.e.: 1-quality yield factor
= P(Sj=0); in this case, this driver is more representative, as the occurrence of
blockages is clearly a minor number than the occurrence of normal operations
/ processing (remark: note that, in the case previously discussed,
load/unload/processing 1 load / unload = 1 normal operation / processing >
so we used ratio b/w their respective operating times).
13
Comment the hypothesis based on the scheduling the operating cycles of
workers and machines, and the possible interferences that may happen as a
consequence:
Many scenarios of interference can happen such as: i) worst scenario > totally
overlapping the operating times requiring the worker > total interference
time XOR best scenario > no overlapping > zero interference time) but it
seems reasonable to take also some average / practical scenario.
Reasonable: an uniform distribution that the time t when the request is raised
occurs when the worker is already busy > subsequently, applying the expected
value of the Unif (a,b) > (a+b)/2 > hence Delta t/2.
What happens in case of more requests? > depicts other possible scenario,
similar to this one shown (practical scenario with a cascade of workers
requests, when service requests are more than twos) > e.g. 3 machine
interference: one machine is promptly served, two machines are waiting;
amongst the twos waiting, if there is one single worker, the first served waits
exactly delta t required worker / 2; the second served waits delta t required
worker / 2 + delta t required worker (in fact, it has also to wait for the second
served to finish being served). The hypothesis behind this reasoning is that all
machines are requesting the worker at the same time: hence one machine is
served, and two machines are waiting due to the constraint of one single
worker.
14
Method for interference analysis / performance evaluation consideration
focused on the complexity of such analysis and needs for more accurate
modeling
15
1
3
Problem setting
3
Each configuration has different strengths and weaknesses:
job shop is the system characterized by low impact of breakdowns;
transfer line is the system characterized by high impact of breakdowns;
cell in between;
these differences are mostly due to different degree of flexibility -> routing
flexibility + the presence of buffers as decoupling point in the system > it is
a design choice, with different buffer capacity levels depending on the
system + the presence of over-capacity which creates some protection from
failures > this is another decision of system design choice relating to
capacity planning, not necessarily depending on the system configuration but
depending on the type of configuration one wants to achieve > featuring over
capacity and hence some more acquisition costs).
Job-shop case Cell (SCAMP) - transfer line case examples already discussed (go
back to the discussion, just as a remind ) with different buffer capacities, routing
flexibility, machine capacities.
Generalizing, we have to study the availability of the system starting from the
availabilities of each component (machine/station); it is needed to understand /
analyze how the machines contribute to the total production capacity of the
system, so also to analyze what happens at system level as a consequence of a
machine breakdown; to this end, there is the need to model / analyzethe system
configuration.
3
Problem setting occurrence of a failure and its impact at system level
> analyzing this problem can be considered as part of the design of a
manufacturing system
This problem is part of a design choice, i.e. considering both the buffer
sizing and the allocation of machines to stations within the system (more
than one machine in a given station to create some overcapacity with
respect to the demand to meet > parallelization of machines / resources to
carry out same operations on different products) is of significant importance
for many manufacturing systems > we want to evaluate how the number of
machines enables to meet the demand considering their effective /
actual production capacity (not only the theoretical capacity).
Technically speaking, in order to properly model the failure to eventually
study the impact at system level, it is needed to analyze the function of the
machines for the production capacity of the whole system (how much the
machine is contributing respect to the required capacity to meet?);
hence, we should start from knowing some coefficients of the machines
(already defined as availability A > the performance losses are represented
by unavailability 1-A), and use these coefficients to achieve an expression
of the system availability A system (as a function of the system of
components (A machines, A stations, A other devices as the material
handling system)).
4
Problem setting questions for the analysis and design of the manufacturing
system:
4
Method for availability analysis introduction to a simple method
5
Method for availability analysis example #2
6
Method for availability analysis example #3
All the production flows through machine M1 > the machine M1 is down >
the line is stopped immediately or (thanks to the presence of buffers) just
after the material present downward (WIP) in the line has been realized out
of the system (propagation of failures along the line). If the transfer was
sync, all should be stopped immediately.
Remark: in this lecture, we will not consider the presence of buffers, analyzing
the effect of machine failures on system capacity.
7
State space method introduced, firstly, as a generic method for logical
analysis of the function(s) played by a system of interest/SoI, the method is
then applied to the machine breakdown problem.
8
The final objective of this methods application is to providea rough estimate of
the expected availability / production capacity of the system: in fact, the method
is an approximated method, applicable by using simple spreadsheetsbut with
some limiting assumption. One limiting assumption is the fact that each machine is
working independently of the others. Hence, some how, we assume that it is a
rough approximation acceptablefor a first indication. Even tough, many
degradation phenomenaleading to the failure occurrence are naturally
independent ones, hencethe approximation is not that bad in their regards: e.g.,
lets consider that failure mechanisms can be interpreted as random problemsthat
may occur at machinesleading to the downtime, as independent issues.
8
State space method > state space table
(herein, it is applied as demo to a functional department in a job-shop system).
9
entire manufacturing system at the end of the interval (0;t] > therefore, the state
space table is simply representing/considering these wholestates set of the system
as rows (in the case, 8 = 2 raised to the power 3), that were obtained by scanning
all the possiblecombinations, ranging from the case that all machines are down
(si,j elements of the matrix equal to zero) to all machines are up (si,j elements of
the matrix equal to one) > subsequently, different system functions (due the
machines) are reached, ranging from the fact that contributions are fully achieved
(i.e. from the system perspective: no production capacity reduction at system
level (theoretical PC)) to no achievement at all (alias total capacity reduction).
More precisely, in order to make this scanning and then fulfill the state space table,
the states of the system come out from all the combinations of states of the
machines > formula of combinations (w/o repetitions) used to calculate the number
of combinations of k machines out of n installed machines > this enables to
enumerate the number of states where k = 0, 1, 2, 3 machines are down in the
manufacturing system (remind that the combinations w/o repetition can be written
also using factorials {n!}{k!(n-k)!}).
Generalization
As we will see, the state space method / table is a flexible modeling method, hence
applicable to different types of systems (not only job-shop/cells, also transfer lines).
9
State space method as a basis for deriving a probabilistic model,
necessary for a quantitative evaluation
The state space method prepares the ground for a quantitative evaluation.
In this regard, now the question is what is the probability that some
production capacity reduction happens (subsequent to the breakdowns
of one or more machines) (this is a follow up of the different combinations
already expressed in the state space table)? This means to compute, by
means of a probabilisticmodel, first of all, the probability that Machine
Mj reaches state Sj = 0 (fails) at time t, therefore the probability at
manufacturing systemlevel (probability that no or some capacity
reduction happens) > In this concern, overall, the probability at system
level results by simply scanning the state space table / different
combinations.
More precisely, we exploit simple rules of probabilistic calculus:
1. each machine is working / degrading / hence failing
independently of the others > independence of the events > rule: P
(A AND B) > P(A) x P(B) > in the state space table, e.g., P(S1 = 1
AND S2 = 1 AND S3 =0) > P(S1 = 1) x P(S2 = 1) x P(S3 =0) > so
we can obtain the probability of a given state of the system (in the
matrix, along onerow i);
2. states of the systemare mutually exclusive, that is: the system
10
may reach one and only one of the 8 states shown in the state space table
(e.g. the state 111 is different from 011) > P (A XOR B) > P(A) + P(B)
> in the state space table, e.g., we can then calculate the sum of all mutually
exclusive states having one common function e.g. 1 machine down >
P(1 machine down) = P (((S1 = 1) AND P(S2 = 1) AND P(S3 =0)) XOR
) > sum +
Once the probability has been calculated, this can be used to estimate the
expected valuefor the production capacity: in fact, after knowing the probability
to be in a given state i (number of breakdowns issued in the manufacturing system >
subsequent reduction in production capacity), the next step is to model/estimate
the impact correspondent to that given state of the system > we then define a Yi:
correspondent to each row, we have a performance Yi expressing, in this case of
machine breakdown problem, the impact of capacity reduction correspondent to a
state i > hence we obtain the valueof actual/effectiveproduction capacity reached
in that state i.
10
Summarizing the procedure:
1. The state space table is the first matrix, built in order to enumerate all
the combinations of machines states (to obtain the states of the system i)
and, subsequently, to define a system function (for each given state of the
system i);
2. correspondingly, a second matrix, table of probabilities, is derived:
therein, the elements pi,j are expressing the probabilities of the
correspondent elements in the state space table (pi,j of a machine j in a
given state of the system i, result of all the combinations);
3. within this second matrix, we derive the probability pi of the state i of the
system, exploiting the independence of working states / events of machines
(i.e. multiplication of pi,j); remarkable to say that the size of the matrix
depends on the number of machines installed in the manufacturing
system, that is: n machines > 2 raised to the power n are the number of
states reachable by the system / the rows of the matrix; therefore, of course,
summing up through all the states of the system / through all the rows leads
to a probability of 1 (all events/states may happen), which is subsequent to
the issue of mutual exclusiveness of different states of the system.
11
Completing the summary of the procedure:
12
Quick comment on the use of the operating times, as data in order to provide
a rough estimate one could find a machine functioning / not functioning (up /
down) in a given time interval.
13
Method for failure analysis / performance evaluation consideration
focused on the complexity of such analysis and needs for more accurate
modeling
14
Industrial Technologies Case Study
Federico Uslenghi
14th of October 2016
Federico Uslenghi
https://it.linkedin.com/in/federicouslenghi
2
Organizations New products - Productions
NEW PRODUCT
DEVELOPMENT
Market
Production factors Analysis
Operators/Employees
Product
Design
Equipment
Energy Product
Materials Industrialization
$ and
Data
PRODUCTION &
Purchasing Production Sales Distribution
LOGISTICS
4
Why this lesson today
Aims and Objectives of the Role and Responsibilities as an
Industrial Technology course Operations Manager
To develop competence Management of Plant
concerning the choice of the resources (people, assets,
configuration and the design etc) to ensure effective
of the production systems, fulfillment of end to end order
according to the required management process (from
performance targets order receivement to products
(flexibility, production capacity, shipment) to meet plant
lead time, WIP, etc.). performance targets
To analyze some industrial Management of people
technologies and related
Management of assets
managerial and
organizational issues.
From IT_M1_01
5
Examples
6
Case 1
Case 1
Increasing production volumes
7
Case 1 Moog Overview
8
Case 1 - Hardware definition & features
Ball screw
A ball screw is a rolling system
consisting of a threaded shaft (screw)
and a nut that internally threaded as
well, contains a determined number of
balls.
The return of the balls that run in the
nut during the rolling is permitted by
one or more mechanical elements called
liners
Key features:
High efficiency (both in driving and back
driving mode)
High reversibility
Low torque requirements
Field of application
precision positioning and precision
transport
9
Case 1 - Wortmann classification
ETO
CUSTOMERS
PTO
SUPPLIERS
MTO
ATO
MTS
11
Case 1 Problem statement
Output of flight simulation ball screw has been unstable and often not
enough to meet the customer demand
The future growth of business is putting manufacturing process and
capacity in trouble
Capacity has to be increased, output has to be made more stable
12
Case 1 problem statement
Production process and production system are already existing but as they are
set up they cannot cope with the increased demand
13
Case 1 Flight Simulation ball screw
Manufacturing cycle
14
Case 1 Inputs assessment
Demand: the provided forecast has to be
checked for consistency and accuracy
Other Data: BOMs, dwg are OK,
manufacturing cycles are OK for low
volumes but could be critical for higher
volumes
Raw Material: increased demand will
deplete raw material inventory much sooner
than expected. Immediate action required
Increase purchase order volume
Increase purchase order frequency
(remember the constrains! space)
Can the existing supply chain meet the
above requests?
15
Case 1 Resources assessment
The production program (work order) defines the sequence of
operations required to manufacture a product
Each step is detailed into its cycle time sheet
An increase in volume means a higher request of resources
Production program / Work order
Fase operations Machine Cycle time Set up time
10 Thread grinding Reishauer 1800 2,5 h 1h 24 pcs/wk = 72 h
20 Heat treatment outside operation 60 pcs/wk = 180 h
30 End shaft machining outside operation
40 OD grinding Tacchella 2500 3h 0,5 h
Delta = 108 h/wk
50 Assemblying 1h + set up times
= OD work orders
20 10Fase Thread
Heat operationsoutside
grinding
treatment Reishauer Machine
1800
operation y x Cycle
time Set up time
30 20 10
EndHeat
shaft machining
grinding outside
treatment
Thread operation
outside 1800z y x
operation
Reishauer
40 30 20
OD End
grinding machining Tacchella
shafttreatment
Heat outside2500
operation
outside x z y
operation
50 40 30
Assemblying
OD grinding
End shaft machiningTacchella x x z
2500operation
outside
50 40 Assemblying
OD grinding Tacchella 2500 x x
50 Assemblying x
17
Case 1 resources assessment
18
Case 1 Flight Simulation ball screw
Manufacturing cycle
19
Case 1 The long term solution
The long term volume will grow more and the process can be
improved on cost and service (lead time/on time delivery) to
ensure higher customer satisfaction
A different approach for the whole product or for some sub
assemblies can be developed to reduce cost and shorten the lead
time
Manufact
Engineering Purchasing Assembly Distribution
uring
ETO
CUSTOMERS
SUPPLIERS
PTO
MTO
ATO
MTS
Case 2
New product industrialization
21
Case 2 John Crane
22
Case 2 petrochemical plants
23
Case 2 John Crane Solaro
Environment and safety are key factors in designing new petrochemical
plant installations and upgrades to enhance plant reliability
Petrochemical plants make extensive use of rotating equipments
(pumps) to move hazardous (healh/environmet) fluids
John Crane design and manufacture high performance mechanical seals
for rotating equipments manufacturers
John Crane Solaro is specialized in seal support systems to provide the
best operating environment for mechanical seals in rotating equipment
24
Case 2 Context
25
Case 2 problem statement
Market
Analysis
Product
Design
Product
Industrialization
PRODUCTION &
Purchasing Production Sales Distribution
LOGISTICS
26
Case 2 - Wortmann classification
ETO
Prototypes &
CUSTOMERS
pilot run PTO
SUPPLIERS
First
MTO
production
lots
Final ATO
objective
MTS
28
Case 2 inputs assessment
Inputs
BOM: a preliminary BOM has been used for prototypes. The final
manufactruing BOM has to be defined with the complete selection
of every components/semi finished part/raw material
Drawings: preliminary drawings for testing prototypes have to be
revised to issue complete manufacturing drawings
Supply chain for material procurement has to be designed:
prototype parts have been sourced but the final dwg/BOM could
require a supply chain review
Manufacturing cycle
Technical BOM & preliminary work order have been used final
manufactruing work order /BOM/dwg to be defined
Resources
Temporary resources selected among those available were used for
prototypes stable, reliable, resources to be identified/developed
29
Case 2 problem variables
Production
BOM
program/work order
Drawings
Lay out/Space/logistic
Supply chain
Operations
Number of operators
Operators Decision
skills/certification/training Criteria:
Technology
selection Machines/fixtures Cost target
Volume
[pcs/year]
30
Case 2 technology options
31
Case 2 Technology selection Cost
assessment
First prototypes have been manufactured with traditional welding technology
The total manufacturing cost is too high and above the market targets
A different technology has to be found
32
Case 2 problem variables
Technology requires a
BOM Production higher degree of process
Drawings program/work order consistency: pipes must
have tight lenght and
Supply chain Space/logistic
parallelism tolerances,
Operations bends cannot be
produced by hands
Number of operators
New Suppliers able to
Orbital Operators meet new requirements
Welding skills/certification/training have to be found
Machines/fixtures Drawings/BOM to be
updated
Production program/Work order
Prototype 12 passes ORBITAL WELDING MACHINE
Fase Operations Machine Cycle time Set up time
10
20
30
Tube bending
Weld
Water pressure test
Ercolina
orbital machine
Hydrotest area
1h
3h
2h
0,2 h
1h
+
40 Pickling and passivation External operation
Remember the
33
selection criteria
Case 2 problem variables
Layout
BOM Production
A new weld bay has
Drawings program/work order to be created
Supply chain Layout moving other areas
The area dedicated
Space/logistic
to the old design will
Number of operators become available
Operators Operators
Orbital
skills/certification/training skills/training: new
Welding
technology requires
Machines/fixtures
proper training
Operators
certification: the
process and not the
operators must be coded
34
Case 3
Case 3
Health & Safety
35
Case 3 Health & Safety
36
Caso 3 John Crane
37
Case 3 Health & Safety
38
Thank you
Introduction to production systems
2
Production factors
Factors Products
Production
and
process
services
Operators
Equipment
Energy
Materials
$ and
Data
3
Production process
4
Production and Logistics
5
Production and others industrial processes
NEW PRODUCT
DEVELOPMENT
Market
Analysis
Product
Design
Product
Industrialization
PRODUCTION &
Purchasing Production Sales Distribution
LOGISTICS
6
How to define a production system
q Classifications
7
Production process
Control
Waiting
8
9
10
11
12
13
Technical drawings
14
List of parts
15
Bill of materials
16
Bill of materials
17
Bill of materials
18
Cycle time
19
Production program / Work order
20
Production program / Work order
21
Flow sheet of a production plant
22
Flow sheet of a production plant
Iron mill
23
Layout of a production plant
24
Layout of a production plant
25
Layout of a production plant
26
Layout of a production plant
27
Layout of a production plant
Owo
Taglio Piegatrici
PF
Stampa Finestr.
Stampa
UV
Fustelle
MP
Staccatura Piega
incolla
28
Types of production systems
29
Types of production technologies
30
Types of production technologies
31
Types of production technologies
32
Types of production technologies
33
Type of production technologies
34
Wortmann classification
ETO
PTO
CUSTOMERS
SUPPLIERS
MTO
ATO
MTS
36
Classification of production systems
37
Classification of production systems
Production
systems
Manufacturing
Process plants
plants
Continuos Parts
Assembly
flow production
Fixed
Job shops positions
Batch
Assembly
Cells shops
Transfer Assembly
lines lines
38
Example of production systems
40
Airbus
41
Illy Caff
42
Thought questions
q Seethis video:
https://www.youtube.com/watch?v=DTWnQDA
hp9k
Can you classify the different production processes
presented in?
43
1
3
Layout(*) and organisation of resources in the plant
The production system results from grouping machines based on their
homogeneity in terms of technological processes. It is also named as a
process layout, as expression of this grouping criterion (layout per process).
Equivalently, it is seen also as a set of different production departments
where each department is a group of machines that can provide the same
technological capability (depending on the technological process these
machines support).
The departments are indicated also as functional departments, remarking
the fact that there is a functional separation in different production
departments within the system (e.g., in a mechanical job-shop, one dept. is
providing a turning function, one dept. a milling function, etc.)
3
When a product is introduced in the product portfolio of the company, a
production process has to be developed; the technical office is then in chargeof
process development/process planning which means some activities.
The activities are (roughly speaking): i) defining the process plan of the product
i.e. the operations and sequence of operations to be performed inside the plant,
forming the process plan and correspondingly ii) looking for the different
technological processes involved in the job-shop, then identifying how the process
plan is released based on the technological capabilities offered by the system; iii)
this affects the material flows because the functional departments that should be
visited according to the process plan may be potentially different in terms of
process plans required by the different types of products (and in different order due
to the sequenceof operations) >>> this leads to the possibility of a different
(technological) routing through the plant as characteristic of each product.
Overall, respect to the production system, material flows are intertwined depending
on the product portfolio offered to the market. They are also variable through time
depending on the market demand.
3
Product routings and logistics in the factory
Product routings are supported by proper transport means: such transport
means are flexible in the routingsas they enable different material
movements. E.g., an hand cart forklift (manually guided, possibly
powered by electricmotor) can be used for handling materials from one
functional department to another, as required by product routings. The job-
shop is then characterised by an high material handling flexibility (i.e.
the ability to move different products efficiently for proper positioning and
processing through the manufacturing facility, as theoretical definition).
More precisely, the materials are moved according to transfer lot sizes(i.e.
number of work-pieces per lot) which are physically correspondent to /
limited by some supporting entities (e.g. pallets, bins). E.g.: pallets are
moved through hand cart fork lifts.
In general, these entities (e.g. pallets, bins) are used to load the transport
means and inter-operational buffers where the material is temporarily
stocked, waiting for the resource availability for next operations in the
process plan. As such, the buffers are stock holding points carrying out the
so called decoupling functionality (b/w departments).
In a job shop it is common to identify inter-operational buffers, either
within each department or, more commonly, as production system buffer
between the departments; the latter case guarantees a more economical
solution to achieve the same decoupling functionality (indeed, the buffers
4
are common space -> this results in less space required due to an aggregation of
space requirements from different departments i.e. advantages of centralization
logic of the decoupling function).
4
Human resources and organisation of resources in the plant
Grouping machines in functional departments (homogeneous for
technological capability) leads to a functional separation. It allows expertise
to be pooled (per function): it creates a favourable condition to support
knowledge transfer between workers operating in the same department.
The learning process can speed up, within each department, through
workers with highercompetences / experiences who transfer knowledge /
practical suggestions based on common problems (homogeneous for
technological capability) .
5
Strenghts
a fundamental s trength of the job-s hop is its high flexibility.
(Short/medium term) Mix flexibility can be defined as the ability to meet the markets requirements in terms of variety of products s upplied in a certain time; e.g. it can be meas ured as widenes s of the range of products /
product types . Volume flexibility can be defined as the ability to deal with variations in the aggregate demand, e.g. meas ured in relations hip to the variation of the production volume required by the market.
(Short/Medium/long term) Product flexibility can be defined as the ability to meet the demands of the market in terms of product s pecifications (in a certain time, i.e. als o s hort term) > product modification /
cus tomis ation; + the ability to vary in time the production mix (i.e. ability to deal with additions or s ubctrations over time) by launching new products > product innovation (medium/long term). Expans ion flexibility can
be defined as the the ability to eas ily add technological capacity and production capacity in the production s ys tem.
Mix flexibility
A job-s hop can produce a wide range of product types + can frequently change the mix over time in terms of relative production volumes (of thos e product types ), i.e. % res pect to the total production of a given mix in a
certain time (Mix Flexibility).
This is pos s ible becaus e the job-s hop is not cons trained to a unique product type, thanks to different characteris tics that can be found in a job-s hop, primarily: i) machine flexibility (the ability to proces s a variety/range of
different parts effectively, in other terms the ability to perform various types of operations w/o requiring a prohibtive effort in s witching from one operation to another; all in all, the machine is not dedicated to a s ingle product,
but flexible to product change overs , @ s ome cos t / time / impact (****)) + ii) material handling flexibility (the ability to move different products efficiently for proper pos itioning and proces s ing through the manufacturing
facility; material handling not rigid, not cons trained to the product routeing of s ingle products / s ingle group of products , but flexible to different product routeings (******)).
With s pecial concern to changes in relative production volumes (in the mix), routeing flexibility is particularly relevant, that is : the iii) routeing flexibility within each functional department (the ability to proces s a given s et
of parts / products on alternative machines ; not jus t one machine pos s ible for a required operation, but flexibility on the alternative machines that can be vis ited; worth a remark: not neces s arily all the machines can be vis ited,
s ome could be not vis ited due to s ome cons traint e.g. required vs . available workable volume) and iv) routeing flexibility through / betweeen functional departments (the ability to proces s a given s et of parts / products bas ed
on alternative technological cycles / proces s plans ). In fact, routeing flexibility is then a lever to cope with variations of the machine workload induced by variations in production volumes (rebalancing the workload )
(***********).
(***) not neces s arily a machine ins talled in a job-s hop is highly flexible; this depends on a des ign choice in regard to the type of machines s elected for the s ys tem; more the machine is a general purpos e type, more its range of products /
operations is wide; this is of cours e pos s ible due to the technological features of the machine.
(******) this is a clear characteris tic that can be naturally as s ociated to the job-s hop s ys tem: by definition there is no rigid interconnections , ultimately guaranteeing different product routings (adapted to the proces s ing requirements of
different products ).
(***********) it is worth remarking that the material handling flexibility enables / habilitates the routing flexibility as direct cons equence
Volume flexibility
A job-s hop can cope with variations in the aggregate demand (Volume flexibility) thanks to the s ame reas on: iii) routeing flexibility within each functional department and iv) routeing flexibility through / betweeen
functional departments .
Product flexibility
Product flexibility regards the new product launching after product innovation or cus tomization. A job-s hop can facilitate the introduction and engineering of new products / cus tomized products w/o affecting much the production
of product types already in the portfolio.
This is pos s ible thanks to different characteris tics that can be found in the job-s hop. New products / cus tomized products can be launched thanks to characteris tics previous ly dis cus s ed: i) machine flexibility (.); ii) material
handling flexibility ( + it is worth remarking the decoupling functionality of different s ys tem buffers & the fact that machines are decoupled as well within a dept., which facilitates the concurrent production of new products /
pre-s eries and products already exis tent in the portfolio). M oreover, it is worth remarking the benefit of routeing flexibility within or b/w functional departments , which can be a lever als o to reduce the impact of new product
launching (i.e. for s ome production period, s ome production capacity of one or more dept. s hould be dedicated to produce pre-s eries /trial production; hence, the production of product types already in the portfolio is continuing
with s ome reduction in production capacity due to pre-s eries ; routeing flexibility can be us ed to rebalance loads on the machine and thus to limit the impacts of pre-s eries /trials , if needed).
6
FURTHER STRENGHTS
low impact / vulnerability to breakdowns (short-term) and low obsolescence (long-term) of the job-shop can be seen as a
consequence of flexibility.
7
Relationship b/w Flexibility and Productivity
Flexibility causes penalties in time, effort, cost or performance (also quality) > overall, then Productivity (Y/X) is reduced. That is, e.g.: an high working time
nominally available (a high number of working hours) is required as input (X) in order to achieve the required output (Y), i.e. target production volume.
Some reasons can be discussed.
(r.1) Routeing flexibility within or b/w functional departments often corresponds to the fact that the group of machines that can provide the same
technological capability can provide it with different performances
E.g. 1 quality characteristics of the processed products can be different, because tolerance guaranteed by different machines could be not the same in the job-
shop > overall, less quality yield, high scrap rate or reworks > in general, qualitative characteristics of the products can vary for different work-pieces when
changing the machines where the product is processed, 2a vs. 2b; -> subsequent effects on quality + machine efficiency (non efficient use of machines) > more
hours needed to achieve the good production > reduced productivity.
E.g. 2 the cutting speed (in a metal cutting process) could be not the same in different machines, even if the technological capability is the same > possible
drawbacks in resulting performance (production rates) > reduced machine efficiency i.e., less efficient use of production hours (with lower speed) > more hours
in order to obtain the target production volume > reduced productivity
(r.2) Machine flexibility means the possibility to switch from one product/operation to another, but this corresponds to set-up time -> less hours used for
processing products/materials -> non efficient use, reduced efficiency of the machine working time (i.e. machines used to make setup) -> more hours
required to obtain the target production volume > reduced productivity
8
Weakness in the system design
This weakness is related to the design / planning of such a system: it is difficult to calculate / estimate the production
capacity available from the machines installed in the job-shop. The following factors motivate the difficulty.
Production mix: changes in the production mix (during time) changes in the machine workloads changes in
the utilization rates of machines different bottleneck machines > different production capacity available from
the system.
Technological characteristics of jobs / complexity of pieces to be manufactured: different process plans /
sequences of operations, different requirements in terms of working / processing times changes in the
machine workloads changes in the utilization rates of machines different bottleneck machines > different
production capacity available from the system
Alternative routings: changes in the process plans / product routings (b1) changes in the machine efficiency
(i.e. less productivity, more resources needed to achieve the target production volumes) > changes in effective /
actual workloads at some machines + (b2) possibility to rebalance workloads (i.e. some degree of freedom in
design choices) b/w machines, in order to reduce workloads at bottlenecks overall, (b1) + (b2), changes in the
utilization rates different bottleneck machines > different production capacity available from the system
Number of machines and their state: if some machine is not available, there is a reduction of capacity of a
functional department difficult to estimate the overall effect on the production capacity of the system
(combining, for ex., with the choice of alternative routings after the capacity reduction due to breakdown) >
different production capacity available from the system
Lot sizes: changes in the lot sizes (i.e. sizes of the production orders / jobs) changes in the number of set-ups
required (to change production orders / work-pieces in order to finally produce the target production volumes)
e.g. increasing lot sizes = savings number of set-ups and total set-up times, hence less effective / actual
workloads, higher utilization rate (for real processing) > different production capacity available from the system >
better utilized
Ability to schedule and control production orders / jobs: see previous discussion on scheduling complexity
dynamic bottlenecks, underutilization of some machines (in stand-by) lower utilization rate caused by
inadequate scheduling > different production capacity available from the system > worst utilization due to not
adequate schedule
Lot sizes + Ability to schedule and control production orders / jobs: (when set-up times are sequence
dependent) changes in the sequencing policies changes in the set-up times and hence utilization rate (for
processing) e.g. savings set-up times with adequate sequencing policy > different production capacity available
from the system > depending on more or less optimized sequences (on the minimization of set-up times or not)
The real dynamics of the system is a combination / superimposition of all these conditions > this leads to difficulty
of estimates of production capacity.
9
Example 1
Brief description of the processes (see below some descriptive notes in the
annex)
Brief description of the layout and possible material flows through the
layout (i.e. non-linear and crossing through different shops, intertwined) as
effect of different process plans / products
Expansion of the layout, probably progressive through years > for
example, one can guess that stand-alone, general purpose machines (i.e.
CNC machines) were added later in a separated part of the plant >
8
expansion of capacity and integration of some automation (i.e. technological
capability) in general purpose machines
More focus on the manual fabrication system whose operations are summarized in
the list of technological processes and applied to the layout considering the presence
of machines that can operate the technological processes; remark # 1: turning is not
separated neither according to processes internal / external turning nor to levels
of automatic control of machinery automatic lathe and manual lathe
Quality control has a separated room in the layout (collaudo separated from
technological processes)
Annex
Some short descriptions taken out (also partially) from http://www.britannica.com/
(e.g. http://www.britannica.com/EBchecked/topic/382939/milling-machine)
Milling machine
A milling machine is a device that rotates a circular tool that has a number of
cutting edges symmetrically arranged arout its axis.
The work-piece is commonly held in a vise or similar device clamped to a table
that can move in three perpendicular directions.
Disk- or barrel-shaped cutters are clamped through holes in their centres to arbors
(shafts) attached to the machine spindle; they have teeth on their peripheries only
or on both peripheries and faces.
Sawing machine
A sawing machine is a device for cutting up bars of material or for cutting out
shapes in plates of raw material.
The cutting tools of sawing machines may be thin metallic disks with teeth on
their edges, thin metal blades or flexible bands with teeth on one edge, or thin
grinding wheels.
The tools may use any of three actions in sawing: true cutting, grinding, or friction-
10
created melting.
10
purchase, operation and maintenance of specialized in-house deburring equipment
versus sending parts to a qualified contract deburring job shop.
Removing INSIDE and OUTSIDE burrs and sharp edges that are normally
created during cutting operations; it is possible to deburr round, oval, square,
rectangular tubes, irregular shaped profiles made of iron, stainless steel,
aluminium and plastic materials.
10
Example 2
8
Example2
Semiconductor manufacturing
A job-shop can be found not only in the mechanical sector. See for examplethe
semiconductor manufacturing, where many conditions typical of a job shop
can be identified, e.g.
8
13
14
(these are all questions that can be answered by using the system design model
presented in the remainder of this presentation) > suggest thinking to the
Mechoff case + consider also that some questions stimulate sensititivity
analysis (e.g. what happens if a machine break down, what happens with more
preventive maintenance, what happens if the production mix changes ) with
respect to a base scenario (how many machines / operator do we need to meet
demand)
15
Identification of product types suffers from uncertainty / uncertain data >
how to identify the product types?
Based on analogies with past experiences, similar plants / productions that a
corporate has been producing + Fostering future scenarios for potential
production mix
Besides, it is worth remarking that a sensitivity analysis can be useful in
order to cope with uncertainty: due to uncertainty, a set of credible
scenarios can be proven as a range to see how design decisions are
changing/robust (what happens if a scenario happens?)
16
17
18
It considers the time needed for the processing cycle on a work-piece; the
cycle also includes some auxiliary movements required to position the
work-piece during the processing cycle (e.g. FIDIA case / high milling
machines, focus on the processing cycle); if more operations are carried out
at different stepsof the process plan, T ij includesthe total time of these
operations (e.g. case of re-entrant material flows, e.g. semiconductor
manufacturing).
T ij does NOT include the set-up times: set-ups refer to times needed to
prepare the machine for the next processing cycle, e.g. (in case of machine
tools) make some machine regulation + change tools + fixing the work-
piece, positioning and clamping on the machine tables (needed work
holding device / system as clamping system), etc.
19
Coefficients (measuring time losses, cfr. Turco)
SCRAP RATE SR: percentage of materials out of tolerance (not achieved the target quality); the
material cannot be restored or repaired (reworked), hence it is discarded. More general, we can
consider also the possibility to rework the material (which causes time loss, as well)
TRIAL RATE TR: percentage of time, when the machine could theoretically be used, since the
technical conditions required for its use are fulfilled, but the time is dedicated to trial production
(subtracted as time loss). In fact, some external reasons (i.e. external to the machine) leads to non
utilization: in this case, the machine is not utilized because it is used for technical tests for trial
production / pre-series; therefore, this time is not planned for production, thus leading to time losses
for trials (i.e. subtraction of these time losses from the time that can be theoretically used)
AVAILABILITY: percentage of up time (intervals), when the machine is required for production
and actually available to work (w/o trials), with respect to the total time (up time + down time); down
time intervals regard the overall failure and maintenance downtime. In this case, the loss is caused by
some technical conditions & requirements (preventive maintenance planned in the production shift >
hence impacting on the required production time) + malfunctions / anomalies / failures (by definition,
impacting)
HUMAN COEFFICIENT: percentage of up time, when the machine is required and actually
available to work & the operator is available to carry out its required tasks. Sometimes, in fact, the
operator cannot be present because he/she is carrying out tasks in other machines; as auxiliary
resource to carry out tasks related to the operations of the machine (load/unload work-pieces, clamp
them etc.), sometimes he / she is busy when required (in other machines); this leads to the occurence
of some time losses on the machine under concern, which is then waiting for the worker; in other
words (alias), the machine is waiting in a stand-by time, caused by the organization of human task
allocations to machines leading to the occurrence of the problem of man-machine interference (this
theory will be considered later on during the course)
> Applying the formula, the yearly workload NH takes into account the need to load the machines (on
yearly basis). Thanks to the coefficients, NH also considers the presence of time losses within the
machine calendar time > thus leading to a gross / effective / actual workload required from the
machine
20
21
(see back the difficulty of production management) > scheduling complexity
> scheduling systemis not capable to fully utilize machines with production
orders, which finally corresponds to unefficient schedules > hence, in the
model, time losses caused by unefficient schedules are estimated by the SE
(correspondent to time losses due to scheduling efficiency)
22
23
Rounding down does not mean necessarily that the system is not able to
reach the target production volume. The reason is basically related to the
limits of the model with respect to the real operations: (i) coefficients are
only estimates (based on past averages, experiences, similar plants) + (ii)
coefficientscan be improved during operations (thanks, e.g., to better
preventive maintenance policies)
If the machine-typecost is high, it is wise to reduce (to reduce the financial
needs / investment costs). One strategy is to reduce the workload required
on the machine-type, by means of (i) outsourcing (i.e. moving workloads to
suppliers) or (ii) planning alternative routings (i.e. moving workloads to
other machine-types, less utilized or any how at least cost of the machine-
type)
24
(financia perspective) CAPEX versus OPEX > CAPEXs are splitted during
the facility / machine life time > splitted cost can be then comparable (so,
summed up) with costs representing OPEX. The mechanism to split is based
on financial issues not detailed in this course.
(as a result) two possible choices when evaluating the alternatives of system
design under the economical dimension (economicconvenience): (alt. 1)
same shifts / day for all the departments vs. (alt. 2) optimal shift for each
department. Alt. 2 minimizes the cost function expressed in the above
formula; but there are some costs still not considered in the formula: i) cost
of higher WIP & required space for stock holding, resulting from
decoupling functionality of departments operating in different shifts + ii)
other extra-costs induced bythe different shift operations, both
organizational extra-costs (e.g. workers with survellaince tasks through the
plant, workers with material handling tasks ) and due to some shared
plant structures whose operating costs are any how spent (e.g. heating
central system switched on even if some departments are closed / not
working during a shift)
25
26
27
28
29
1
3
Layout and organisation of resources in the plant
The production system results from grouping machines based on a basic
association between a set of parts / products to be produced and a set of
machines capable to support their production > ideally, parts / products are
assigned to families such that all the parts / products in the families are
processed on the same group of machines, and similarly machines are
grouped into cells if they process the same set of parts / products.
This approach (cellular manufacturing) leads to obtain cells made of different
machines, offering different technological capabilities in accordance to the
processing requirements of the parts / products assigned to the cell. In other
words, the process plan (i.e. processing requirements) i.e. the operations
and sequence of operations of each product / part in the families
associated to the cell finds the different technological processes directly
comprised within the machine groups forming the cell.
8
in reality, even if the technological processes were the same, there were different
capabilities in terms, e.g., of installed electric power (different kWs) > different
machines could process materials at different hardness > basically the material flows of
different products / parts were then directed to specific machines (depending on their
processing requirements and specific limits of the machines);
Suggestion: to take advantage by re-arranging the machines according to cellular
manufacturing, resulting in three independent cells, grouping together machines
according to the type of material as criterion > overall: no inter-cell move, cell
independence + linear material flows (in/out of the cell) + closeness / adjacency of
machines within the cell > potentials for time saving in material handling for
parts / products assigned to the cell (based on a proper re-layout within the cell > this
is required to decide how the different machines are installed in the space dedicated to
each single cell)
5
Example 2
8
pallets carrying work-pieces to be manufactured / after their manufacturing; the
conveyor tracks are made of different segments (along the loops) computer-controlled
carry-and-free, e.g. in order to enable stopping in front of the required workstations;
last but not least, manned operations is required at load/unload stations.
6
Example 3
8
8
As summary of the previous examples, it is worth remarking the following
issues:
example 1 was a re-arrangement of existent equipment on the factory floor
(i.e., from job-shop to cells);
example 2 and 3 were considering the incorporation of flexible automation
(FMS, FMC);
both they were subsequent to the so called cell formation, leading to the
identification of machines and associated parts / part families grouped to form
a cellular manufacturing configuration.
Manufacturing cells provide different advantages with respect to the job-shop.
10
As a consequence
Batch size / lot size reduction > possible thanks to the set-up time reduction >
subsequently (lot-sizing policy, economic perspective) economic order quantity would
be reduced + (production logistics perspective) even if with minor batch sizes, proper
machine utilization can be guaranteed w/o relevant set-up time loss (i.e. even if with
more set-ups due to minor batch sizes, reduced set-up time loss on the machine) > less
time loss for setup means then more production capacity that is effectively utilized for
working / processing work-pieces.
10
possible exactly because of the low distances / high material handling productivity;
accordingly the transfer batch can already start new operations at next workstations w/o
waiting finished operations of the remaining parts of the production batch at the
previous workstation (> overlapping policy)
More reliable estimates of delivery lead times > if queuing times are low due to low
WIP, variab ility of queuing times is also low > low variability of manufactu ring LTs
> more reliable estimates of delivery lead times > possibility to guarantee high
delivery reliability (i.e. meeting the due dates) and better use of management system
which assumes knowing the lead time to plan in respect to some dates (i.e. MRP)
10
Initial Remark
Before introducing these further strenghts, it is worth remarking that the
traditional approach (and the majority of approaches) used to form & design
cells consider cells in terms of their respective parts and machines, and regards
the machines capacities as the factors that limit production (Russell et al.
1991). Besides, traditionally, methods only considered humans and their
assignment to cells in terms of their labour capacity (Russell et al. 1991), or
rate at which they can produce a part, while they do NOT in terms of the skills
they possess, which is more a recent issue. But, as real problem (and lever to
gain opportunities), the complete implementation of cellular manufacturing
into an organization would require a major modification of the production
system, resulting in very significant changes in worker roles (Hedge et al.
1994).
An important requirement for cellular manufacturing is in fact an increased
level of technical skills and flexibility for workers, along with the ability to
work in teams. Johnson and Manoochehri (1990) summarize the human
resource requirements for success in cellular manufacturing: i) dedicated
workers who have multiple skills (technical as well as communication and
interpersonal skills); ii) who have the discipline to follow strict methods and
procedures; iii) who are willing to make decisions and accept responsibility;
and iii) who are committed to effic ient and effective production.
11
first of all when the cell is manned (i.e. low level of automation), the workers have
major responsibilities for setup, processing, material handling and also inspections (i.e.
quality inspections). In particular, due to closeness / adjacency of machines within
the cell, the workers have responsibilities to / can undertake a range of task along the
process plans (operated through different machines within the cell) of parts within the
family, e.g. processing tasks along the process plans > job enlargement; moreover,
quality inspections are assigned to operators, which could lead to potentials of quality
problem solving and any related decisions (even if it is worth remarking that product /
process engineering is usually centralized and out of the cell, hence problem solving is
not totally delegated) > job enrichment (involving the vertical integration of tasks and
the integration of responsibility within the cell, thus assigning some autonomy to the
cell itself) Another e.g. of job enrichment is the assignment of control and
production advancement tasks, with some responsibility (and autonomy) on disturbance
handling of scheduled activities within the whole cell
it is worth underlining that more operators working within a cell are usually assigned to
work regularly within the same cell > this is favourable for team working, with
common responsibilities grounded on the job enlargement / job enrichment > leading
the team to be responsible for the whole process plans of parts within the families
associated to the cell, comprising quality characteristics
11
Cellular manufacturing is characterized by the following weaknesses:
Difficulties with work load balancing between cells (and between their
machines) > a cell introduces some rigidity (it is less flexible ) with
respect to the jobshop > this is a direct consequence of many matters: i)
families cell association; ii) the intentional design choice to avoid as much
as possible inter-cell moves (i.e. the best choice should be no move, cell
independence) iii) the autonomy of each cell (assigned with the integration of
responsibility within the cell) > as a consequence of the rigidity, it is
possible that the work load generated by different families (in different
cells) is not balanced across the cells (in some planning periods, some cells
over loaded vs. some cells under loaded, and workloads cannot be trasferred
easily, or cannot be transferred at all, from one cell/its machines to the other
cell/its machines, as alternative routing); within a job-shop, there is more
flexibility: it is easy to assign a production lot / production batch to the
underloaded machines, thanks to the routeing flexibility.
(Subsequently) Problems related to production mix variability > when
production volumes between families are changing (as production mix is
variable in different planning periods), workloads change leading to work
load unbalance between cells (in some planning periods, some cells/their
machines over loaded vs. some cells/their machines under loaded); within a
job shop, changes in production volumes between parts/products are
compensated and absorbed thanks to flexibility.
12
Difficulties with the application to the whole stages of the production chain > the
configuration of manufacturing cells is usually not applicable to the whole production
chain, i.e. to all its stages; hence, the advantages obtained with the cells are limited
and mitigated due to the influent characteristics of down and upward stages.
In some cases, necessity of more machines than in a job shop > (reasoning on a long
term perspective) this is again due to the difficulties of work load balancing between
cells/machines (cells/machines are over vs. under-loaded, in the long term) > it could
happen that ceteris paribus the total production volumes the number of required
machines is higher than in the case of the job-shop; in the job-shop, machines are
loaded by all products, hence it happens that some overloading parts/products are
compensated with other under-loading parts/products, and the number of machines
might be lower exactly for this reason.
Difficulties to manage technological operations outside cells > even if the objective
is to process all operations of parts of identified families within the associated cell,
sometimes this is not possible > some cell dependence / inter-cell move is required or,
at least, it is required to finish some operations out of the cells, either in a common
resource in-house or in the premises of suppliers specialized in some technological
processes. E.g. the heat treatments are typical technological processes / operations
which, for the costs of the required machines / plants, cannot be duplicated in more
cells; hence, it is preferable to keep them as common resources, either within only one
selected cell or (often) outside all the cells > the disadvantage is that some material
flows are now intertwined due to families pertaining to different cells , with the
subsequent effect of reduced gains in terms of expected performance (e.g. queuing time
/ manufacturing LT variability in front of the machine outside cells could be higher,
etc.).
Problems related to breakdowns > in a cell, the number of machines of the same
type is lower than in the case of the job shop (at worst case, it is one machine) + the
cell is built based on similarities of parts within families, hence an high share of
parts may require to visit the few machines of the same type, requiring their capacity
> when a breakdown fails, the impact is high, at worst stopping the cell, at best
resulting in high capacity reduction impacting on high share of parts processed by the
cell.
12
Group Technology (GT) is an approach to manufacturing and engineering
management that helps managing diversity by capitalizing on underlying
similarities of parts/products and required activities/operations.
In general, by grouping items/parts which share common traits/similarities,
GT facilitates the rationalization of activities/operations in a wide variety
of functional areas, including purchasing, design, and manufacturing.
Concerning manufacturing, cellular manufacturing, discussed so far, is an
application of GT: thanks to GT, manufacturing cells are created and
operated > rationalization is achieved at shop floor level.
1) First step (data collection) aims at collecting data in regard to the product
types, already manufactured and expected in the future, estimating the
required production volumes (yearly demand) for each product type +
defining their technological routings.
2) During the second step (codification and classification of products), all
product types / part types should be assigned a part code (remark not to say
but have in mind: in case PCA analysis is used); afterwards, they should be
classified according to their characteristics, i.e. those relevant for further
steps of GT application > characteristics such as, e.g., shapes, dimensions,
13
materials, required tolerances (remark not to say but have in mind: both in case of eye-
balling and PCA analysis is used) (we have already seen in some examples that these
characteristics may motivate / lead to manufacturing cell formation).
3) Third step (rationalization and standardization of products) deals with an in-depth
analysis of product / part types aimed at avoiding the product variety which is not
necessary. Indeed, product variety may be induced by many reasons, such as: i) the
introduction of new products in the product portfolio, ii) the existent product
modifications and improvements, iii) the customization requests from clients. If we
avoid product features / specs which are not really needed, product (re)design will
have a great impact on the shop floor and on the definition of the proper configuration
of the production system.
4) Fourth step (rationalization and standardization of technological routings of
products) aims at reducing the variety of the technological routings of the product
types (hence, not only product variety reduction as in the previous step). Ideally, in
fact, if all product types had the same technological routings, production
management could be simplified, material flows intertwining could be avoided,
with subsequent advantages on better workload balancing in respect to the
production capacity, achieving, overall, better performances. Two types of
rationalization / standardization, reducing technological routings variety, may be
achieved at two levels: i) (macro-level) operations sequences + ii) (micro-level) for
each operations, standardization as much as possible of tools and fixtures needed to
produce work-pieces. A good practice to this end is simply to identify existing tools
and standardize fixtures according to product shapes, during the product design and
process planning/engineering steps.
5) Fifth step (identification of part families) aims at identifying the part families based
on similarities of different parts / products in the product portfolio > parts / products
are grouped into families based on similar characteristics: e.g. shapes, dimensions,
materials, required tolerances, operations sequences (see back, linking to step 2,
required to prepare the needed information)
6) Sixth step (identification of machine groups forming the cells) consists of grouping
machines according to the part families identified so far at previous step; this is done
sequentially, that is: after the identification of part families, we identify the needed
machines, according to their operations sequences;
7) Fifth + six steps simultaneously > identification of part families / machine grouping
forming cells can be also achieved simultaneously > cell formation is concurrent to
part family identification (see back, linking to step 4, which is related to the needed
information, i.e. technological routings, with operations sequences and machines to be
visited).
Step 5 and 6 can be supported by different types of methods, presented in the remainder.
13
Methods for Part-Family Identification (PFI), based on the classification of
products (step 2 - 5)
14
>>> Part coding analysis (PCA) methods rely on a coding system. This is used i) to
assign (alpha-)numerical weights / digits to the part / product characteristics and,
afterwards, ii) to identify families by using some familiarization scheme based on the
numerical weights so assigned. More precisely > PFI can be automatically implemented
by means of an algorithm that i) applies a filter to the whole parts set, basing on the
part codes to look for similarities, and ii) forms part families using the filtered parts.
PCA-based systems are traditionally product design oriented or shape-based, therefore
they are ideal also for component (such as e.g. fixtures) variety reduction; other PCA-
based systems may integrate different characteristics, spread from the geometric
features (shapes and dimensions) to the technological features (material, processing
requirements, tolerances ).
Plus: more formalized method, which leads to repeatability and objective outcomes.
Minus: the coding system that are normally available in a company are not necessarily
ready for GT > e.g. they were built more for commercial purposes > this is the reason why
step 2 of GT is required (prepare the coding system ..) > implementing a coding system
may not be straightforward.
14
(step 5 of GT) Part / product families are identified (PFI) based on
geometrical features of products > i.e., shapes and dimensions of different
products.
(Commenting the picture) >>> dimensions (@ similar shapes) of workpieces /
products are initially different, when looking at the whole portfolio;
dimensions are then used in order to lead to the identification of 3 part
families, grouping the parts with similar dimensions > hence, there is a
similarity for what concern the required workable volume > this is a useful
information to select (at the next step of GT) proper machines.
This method is not only useful to identify part families but, as previously
discussed, it helps also to standardize production components such as
fixtures (i.e. similar shapes & dimensions > similar fixtures with minor
adjustments in order to adapt to shapes & dimensions of different products).
15
(step 5 of GT) Part / product families are identified (PFI) based on
technological features of products > i.e., materials, processing
requirements to obtain the required prod specs / features, tollerances of
different products.
Commenting the picture > even if dimensions and shapes of workpieces /
products are now quite different, there is another kind of similarity,
concerning the processing requirements > turning operations are required by
all workpieces > hence, similar processing requirements are used to identify
part families > these processing requirements might then lead (at the next
step of GT) to the selection of the same machine types.
16
Part coding analysis (PCA) methods rely on a coding system. The part code
results after assigning a set of numerical weights / digits correspondent to the
different part / product characteristics established within the coding system
(see the table for an overview of such characteristics). Indeed, each
numerical weight will have a meaning according to the coding system so
defined. E.g. part code 31000 means metallic plain part while 000
could be used to expand the code with other information, in order to consider
further characteristics.
Commenting this example >>> when classifying the parts, parts are
recognized in terms of the following characteristics (geometric /
technological features): i) material (metallic, non metallic, ); ii) shapes; iii)
processing requirements (remark: some processing requirements are tacitly
expressed beyond the shapes shown in the product drawings > e.g. bent
requires bending processes; besides, some other processing requirements are
expressed tacitly beyond the type of raw material > cast, stamped, forged
requires to eliminate burrs, i.e. requires a deburring process.
Part Family Identification / PFI will be subsequent to having all the part
codes available, then applying the relevant filters to identify the required
similarities.
17
Part coding analysis (PCA) methods > this is the Opitz coding system > it
comprises 9 digits representing a wide set of characteristics > it is based on
geometrical and technological features, therefore different characteristics
pertaining to these features.
Commenting this example >>> look at the 1st digit > part class, amongst
rotational parts > there is clear specs in regard to the ratio of dimensions
(length L / diameter D of the rotational shape) > well formalized method,
helpful for the expert classifying based on the coding system.
Indeed, this coding system is even more formalized than the coding system of
example 1, and more complete with more information based on the 9 digits.
More in general, PCA, as said, guarantee more repeatable / objective PFI,
thanks to their enhanced formalization.
18
Methods for simultaneous Part-Family Identification and Machine
Grouping (PFI/MG) (step 5 + 6 of GT, simultaneous)
Different methods are available amongst production based methods > we focus
on cluster analysis methods and, in particular, two well known techniques
within this type of methods.
19
Cluster analysis is composed of many diverse techniques for recognizing
structures in complex data sets. Generally speaking, the main objective of
this cell formation tool is to group either objects or entities or attributes into
clusters such that individual elements within a cluster have a high degree of
natural association among themselves and very little natural association
between clusters.
Introdution to ROC
ROC is a procedure / algorithm that can be classified as an array-based
clustering. In array-based clustering, the processing requirements of parts on
machines can be represented by the machin e/part matrix formulation. The
machine/part matrix has zero and one entries (aij) > A 1 entry in row i and
column j of the matrix indicates that part j has an operation on machine i,
whereas a 0 entry indicates that it does not.
The array-based clustering techniques try to allocate machines to groups and
parts to associated families by appropriately rearranging the order of rows
and columns of the matrix to find a block diagonal form of the aij = 1 entries
in the machine-part matrix > (cells are identified along the block diagonal
form).
Overall, ROC procedure/algorithm uses as input the machine/part matrix and
it aims at rearranging the rows and columns after converting the aij = 1 / 0
entries to binary numbers.
20
ROC Procedure / algorithm
Step 1 > associate a b inary number to each row of the matrix > remember that in
mathematics and digital electronics, a binary number is a number expressed in the
binary numeral system, or base-2 numeral system, which represents numeric values
using two different symbols: typically 0 (zero) and 1 (one) > hence to express the
binary number of each row you have to sum extended from j = 1 to j = J (j-th index of
the columns) of [aij multiplied by 2 raised to the power J-j];
Step 2 > order rows according to a des cendent order of the binary numbers so
calculated (descending from top to bottom-hand side of the matrix);
Step 3 > associate a binary number to each column of the matrix > hence to express
the binary number of each column you have to sum extended from i = 1 to i = I (i-th
index of the rows) of [aij multiplied by 2 raised to the power I-i]
Step 4 > order columns according to a descendent order of the binary numbers so
calculated (descending from left to right-hand side of the matrix)
Step 5 > if on steps 2 and 4 no reordering happened go to step 6, otherwise go to step 1
Step 6: stop
20
Comment the example
The machine/part matrix formulation is a matrix where the rows represent
the machine types (rows i) and the columns the product types (columns j).
The machine/part matrix formulation provides the information of which
machines are utilized / required / visited by different types of products /
parts (aij= 1 or 0 entries).
No additional information is available in regard to the operations sequences,
workloads generated by different product types on different machine types.
Example of computation of binary number of row #1 (sum extended to
columns) (step 1).
21
Comment the example
descending order of rows (step 2)
example of computation of binary numbers of column #2 (step 3)
22
Comment the example
descending order of columns (step 4)
no further re-ordering can happen (step 5) > end the procedure (step 6)
23
on by sub-suppliers (i.e. limited to operations outside the cell > problem of
operations outside the cell not solved or the entire technological routing of the
product, depends on costs).
23
Another technique of cluster analysis is based on the adoption of so called
similarity coefficients > w e now introduce Similarity coefficient (SC) based
methods; thes e methods rely on the computation of similarity coefficients in
conjunction with the use of some clustering algorithm / procedure, in order
to finally form the manufacturing cells.
24
machines (parts) / machine groups, accordingly w ith the similarity coefficients. Once
available the tree, the machine groups forming the cell can be identified considering
existent constraints and objectives, e.g. a minimum threshold for the similarity
coefficients could be fixed and machines are grouped into cells considering their
similarity coefficients in the tree (i.e. higher than the threshold).
24
Comment the example
25
Comment the example
26
Comment the example
27
Comment the example
28
Comment the example
29
Comment the example
30
Comment the example
31
Comment the example
32
Once Part-Family Identification and Machine Grouping (PFI/MG) (step 5 +
6 of GT, simultaneous) is achieved, no verification has been done respect to the
capacity limits.
Cell formation has identified only the machine types to be grouped within a
cell, not their number.
We are now ready - knowing from previous steps of GT the production mix,
technological routings, part families and machines forming the cell - to
calculate the number of machines for each machine type necessary for each
cell
The economic convenience of operating in different shifts / day should be also
considered (as in the case of the jobshop).
33
In Virtual cellular manufacturing, grouping of resources (i.e. machines) is not
reflected in the physical structure > no re-arrangement in terms of layout;
grouping is reflected in the PP&C system, which then enables to associate
part families to group of machines > it is clear that this association is not
rigid and may change in different planning periods.
More precisely:
Depending on the production mix at a given time horizon, machines across
various functional departments may be identified to form logical (virtual)
groups, instead of physically repositioning machines one adjacent to the other.
In order to achieve virtualization, it is then relevant to design logically how
activities can be organized / controlled by re-assigning part families to
groups of machines during the normal planning & control cycle.
One specific weakness of traditional cellular manufacturing that is addressed
by VMC is related to the problems related to production mix variability >
when production volumes between families are changing (as production
mix is variable in different planning periods), PP&C will aim at work load
re-balance between virtual cells (hence, smoothing, in some planning
periods, the differences b/w some cells/their machines over loaded vs. some
cells/their machines under loaded); that is: within a job shop / functional
layout, the PPC builds up different virtual cells, gaining the advantages
expected by cellular manufacturing (except those related to closeness /
adjacency of machines) without loosing the plus of flexibility - typical of
34
job-shops which can be used to compensate and absorbe the variations of production
volumes in the production mix from time to time.
34
Comment the figure:
Given the jobshop as physical structure, with material flows intertwining, the
PPC system identifies group (Part family identification / Machine grouping
forming cells) for a given planning horizon, so creating during this planning
horizon the virtual cells, with subsequent advantages due to cell
autonomy and rationalization of material flows (intertwining). The fact
that this grouping is virtual is a prerequisite for the possibility of (virtual) re-
arrangement in the next planning periods (with production mix variations).
35
Design of Manufacturing Systems Transfer
Lines
General features
Examples
Strengths and weaknesses
System design
2
Transfer lines general features
Each transfer line consists of a series of machines where a
single product type (or a limited number of product types
within the same family) flows, resulting in a routing through
the machines.
3
Example 1 (1/3)
Perspective: from the load station
Tube loading
First machine
Second machine
4
Example 1 (2/3)
Perspective: from the load station
Second machine
Third machine
Fourth machine
5
Example 1 (3/3)
Perspective: from the unload station
Fourth machine
Tube unloading
6
Example 2
7
Example 3
MANNED
Product types associated OPERATIONS
to Rotary Transfer Line AT LOAD /
UNLOAD
STATIONS
8
Some examples
https://www.youtube.com/watch?v=Zx2GWXs
wP2I
https://www.youtube.com/watch?v=UO5dVrzw
e0Y
https://www.youtube.com/watch?v=PxJqQnek
JV4
9
Some examples
https://www.youtube.com/watch?v=uGyfm5Pi
wQM
https://www.youtube.com/watch?v=LxKrDX2b
4ZU
10
Transfer Lines general features
3 3 3 3
N=4
A, B, C D, E, F G, H, I J, K, L CT=3
PC=20 p/h
4 4 4
N=3
A, B, C, D E, F, G, H I, J, K, L CT=4
PC=15 p/h
Overall:
Low WIP
Low Lead time (also considering variability)
13
Transfer lines Strenghts
14
Transfer Lines Weaknesses
Low flexibility
High investment needed
Long time required to start new productions
High risk of obsolescence
Significant impact of failures
15
Rough design of a transfer line (single-model)
where
CT = cycle time of the line [seconds/piece]
16
Rough design of a transfer line (single-model)
where
A = line availability (0 < A 1)
SR = scrap rate (0 SR < 1)
Assumptions
Pieces are manufactured in batches (batch A, batch
B, batch C and so on); changing production from one
batch to another requires a setup
Setup times do not depend on the production (batch)
sequence
18
Rough design of a transfer line (multi-model)
Approximating:
20
Rough design of a transfer line (multi-model)
22
Design of Assembly Systems Fixed position
assembly
q General features
q Examples
q Assembly shop
q Assembly line
3
Fixed position assembly general features
In a fixed position assembly, the product is assembled in a
single site, rather than being moved through a set of
assembly stations. Materials (i.e. components), equipment,
tools are brought to the site.
4
Fixed position assembly general features
5
Examples
6
Examples
Workbench
7
Examples
8
Examples
9
Fixed position assembly Strengths
q High flexibility
q Low investment
q Job enlargement, enrichment and rotation for the
employee
10
Fixed position assembly Weaknesses
11
Rough design of a fixed position assembly
12
1
3
Assembly systems are systems which join together components (coming from
systems making the parts production or subassemblies) in order to obtain finished
products.
Manual assembly systems are composed of several stations in which one or
more workers executes assembly activities. Stations are linked through a
handling system (or transfer system).
CONFIGURATION:
Fixed position assembly: the product does not move while being assembled,
the required components are brought to the working station. All the assembly
activities are executed in only one workstation. For this reason this solution is
typically applied to heavy and bulky products (difficult to handle), e.g.
machine tools, ships, airplanes.
CRITICAL POINT: material handling systems are different and for different reasons ->
should be taken carefully into account when designing an assembly system:
i) Material handling systems for material feeding (moving parts/components to be
supplied; note the challenge: trend to have more and more part numbers for
producing diversity of product variants in a product family -> trend to variety);
ii) Material handling systems for moving the assembly (product) on progress (again
there is some effect of variety but more on the line balancing and sequencing
problem).
3
(as opposite to flexibility )
Clearly in this scheme the fixed position assembly is located in the upper corner, to
the right: depending just on the workforce skills, this system can guarantee:
high flexibility (the ability to adapt is gained by training the operators which are -by
definition- the most versatile existing resource) -> operators in this system are highly
skilled / multiskilled;
high variety and low repetitiveness -> substantially, the operator has to execute all the
assembly activities required by the product; or at least a big portion of activities along the
assembly process;
low volume as the high variety (low repetitiveness) of the executed activities brings to
lower efficiencies for different reasons.
3
Description of this type of system
Every workstation (i.e. the manual station) could be modeled as a unique block,
physically correspondent to a single site (as space in the shop floor) for which:
the input are the various necessary components to be assembled (all those
required by the BOM);
inside the system, such components are assembled together; all the tools and
equipment required to assemble the components must be there (in the site);
the output is the unique finished product.
Every workstation is involved in the assembly of a different types of product (in
the example 2 workstations assembly the product A, the other twos assembly the
product B and the product C, respectively -> mix flexibility is reached thanks to
the parallel stations prepared for different product types, i.e. equipped with
proper materials/components, tools, , and trained workers).
4
Physical structure and organization of the single site:
components are stocked around the site (challenge: need for room/space to
place components);
assembly equipment and tools are placed around the working position where the
product to be assembled is positioned (need for room/space to place );
the product is placed in the central area of the workstation (working position).
in the case of heavy and bulky products, they are located on the ground /
floor in a delimited zone;
conversely, small products are placed on a worktable / assembly work
bench.
the operator works on the product while standing and moving in the surrounding
area.
NOTE: the space required depends on the kind of product to be assembled, the
number of parts to be assembled and their characteristics (amount and size), the
number of tools/equipment .
8
efficient and to be safe;
provide the equipment onto adjustable bases to adapt their height to workers
characteristics;
etc.
5
CASE OF SMALL PRODUCTS -> repair shop (shop replaceable units of radars
in jet fighters)
(Introduction)
The bill of material (BOM) for maintaining the radars embedded in military
airplanes is usually multi level.
The FRU (Field Replaceable Unit) are functional units that can be
replaced by ground operators: when a FRU fails, ground operators
substitute it with a new FRU he/she has available in the warehouse.
The damaged FRU is hence sent to the Maintenance Center.
New FRU means the FRU should be considered as it is brand-new (i.e. as
good as new in reliability theory): it can be that a FRU has been already
submitted to a previous revision/reparation, where components have been
already replaced because broken or because they reached their end of life
stage. It means FRU is not functioning, due to a failure; failure is generally
defined as the situation when an entity (i.e. FRU or other entities in the BOM)
cant execute a required function (UNI EN 13306).
A single FRU is made of many SRU (Shop Replaceable Unit), which in turn
are made of several assembled electro-mechanical components. There are
two ways to repair a FRU, either the damaged SRU is replaced with a
new one, or some of the electro-mechanical components are directly
replaced in the original SRU.
8
(Comment on the fixed position assembly of this case)
Both reparations (SRU replacement or component replacement) can only be executed
by the specialized workers from the Maintenance Center.
The workers are working in a workbench on such small objects/products and fragile
+ note that the workers are multi-skilled and specialized.
Note the ergonomic solutions in the workbench . (e.g pneumatic screwdriver).
6
CASE OF SMALL PRODUCTS
8
CASE OF (RATHER) SMALL AND FRAGILE PRODUCTS
Hi-fi systems, high quality production (take care of non quality generation) +
high level end item (in the market), moderate to low volumes;
Linn Products developed the single-station build concept where one operator
assembles, tests and packs the product;
each operator works at a fully equipped position to which material for
a day's build is delivered / fed by Automated Guided
Vehicle(AGV);
he then builds, packages, tests and signs the product ready for
despatch to maximise performance and quality;
Whilst a typical hi-fi industry product may be manufactured in minutes as
cycle times, assembly cycle times at Linn can be hours for the most
sophisticated and innovative designs (which is aligned to the low volumes ->
low requirements for the production capacity).
8
Aero engines: it is clear that specialized workers are required, to which different
assembly tasks are assigned; it is basically another solution where the site is built by
handling the product thanks to an equipment fixed on the roof.
8
9
High flexibility
Mix flexibility: (short period) it is possible to assemble different types of
product in the same time, meeting specific and periodically changing demand
requirements (i.e. high variety in the same time, different variety at different
times).
This is due to the independence and the decoupling of the
workstations (in a given period, if you need to assemble more products
of type X, it will be sufficient to dedicate / prepare more workstations
to that type of product) -> of course workers should be trained to
change.
Product flexibility: to introduce a new product, operators just need to be
trained to acquire the new required skills for new types of product, while the
physical structure can remain substantially the same (new product being close to
characteristics of existent products -> normally few simple tools / equipment are
required even with changed products -> it is quick to start with new production
of new product types / prepare the workstation to this end).
Expansion flexibility: (long period) it is not difficult to add new workstations,
thanks to their independence/decoupling, the only constraint is room/space.
Low investment
it is, of course, due to the system simplicity -> the building of new
workstations doesnt require particular structures; of course expansion
10
flexibility may be possible (if there is room/space for this):
the investment value may be increasing if some necessary equipment are
expensive;
in this case, it is advisable to buy a limited number of equipment and, through
a correct scheduling, make them available to the different workstations
whenever they are required (obviously, this solution would make the logistics
management more complex and lead to some interference time)
10
Potentials for intertwining of material flows
Every necessary material/component has to be brought to every workstation +
every finished product has to be handled from every workstation. Such issues
creates problems relating to internal, intertwined flows. The problem will be
more relevant with the increase of workstations number.
Conversely in assembly lines, the material flow is extremely rationale -> every
single workstation along the line is fed with just its pertinent components; then
the finished product is taken from just one specific point (the end point of the
line), instead of every single workstation.
For this reason, in fixed position assembly, a logistics management of flows is
challenged by many issues, and inefficiencies are existent, due to such a
complexity on the floor, having different I/O points, the site of the
materials, possibly with different requirements, i.e. multi/mixed model
assembly).
High WIP
Cycle times are high at each workstation -> subsequently, one or more pallets
of components (as input) are kept for a long time as WIP; besides, also the
finished products (as output) are normally not delivered immediately after
the end of the operations (still a WIP then) waiting for some time before the
material handling (i.e. complexity in logistics) come to serve. Then, in buffers
by the station, many components and finished products may be stocked.
11
On the contrary, in assembly line there is a natural and continuous flow of semi-
finished assembly and finished products, which leads to limited WIP along the line and
finished products being regularly delivered out of the line to the further logistics
activities
11
The rough design of a fixed position assembly simply consists of determining
the necessary number of single sites/manual stations. Such number referring
to a piece of product type j has to be calculated starting from the knowledge of:
the average (total) time required in order to complete the assembly process
on a piece of product type j;
the required production capacity for product type j.
The value of the requested production capacity has to take into account the
value of the waste which could be produced during the assembly process;
generally the waste in manual assembly acquires the meaning of
rework/repair, typically for products of high value. It means that PCj consists
of the target demand and the losses of time (consider the scrap rate => rework
rate due to quality control,..).
12
be correct to build up a fixed position assembly: actually if the number of stations
increases, for ex., the intertwining of material flows issue becomes excessively
relevant, and in general, the strategy should look for other solutions (i.e. lines) more
oriented to the efficiency (eventually with some flexibility).
12
Design of Assembly Systems Assembly
lines (part 1)
q General features
q Strengths and weaknesses
q Types of assembly lines
q Examples
q2
Manual assembly systems general features
q Assembly shop
q Assembly line
3
Assembly line general features
Each assembly line consists of a series of stations where
the product is progressively assembled.
4
Assembly line general features
5
Examples
In the 1910s , Walter Flanders designed the first assembly line at Henry Fords car factory.
6
Examples
7
Examples
Todays assembly line in a factory manufacturing wire harness solutions and electrical assemblies
8
Examples
9
Assembly Lines Strengths
10
Assembly Lines Weaknesses
q Low flexibility
q Long time required to start new productions
q Repetitive work
q Line balancing might be difficult
11
Types of assembly lines
12
Paced lines
13
Machine-paced lines
q Strengths
Cycle time and production capacity are perfectly
controlled
q Weaknesses
Probability of no completion (at the line stations)
and problems of unfinished pieces
14
Operator-paced lines
q Strengths
No problem of unfinished pieces
q Weaknesses
Cycle time is variable and it is determined by the
slowest operator
15
Unpaced lines
16
Unpaced lines
q Strengths
No problem of unfinished pieces
Cycle time can be exceeded, but only occasionally
(i.e. problems of blocking and starvation)
q Weaknesses
Cycle time and production capacity are not
perfectly controlled
17
Continuous flow lines
18
Continuous flow lines
19
Continuous flow lines
20
Examples
21
Examples
22
1
3
Assembly line In such configuration:
assembly operations of the assembly process are distributed amongst different
stations (allocated / distributed properly to different workstations /
machines along the line so to balance the workloads amongst them);
operators are stationary in the workstations and a material handling system is
used; such material handling system is a transfer system that may consist of a
roller conveyor, conveyor belt, etc. and it is used to move the semi-finished
assembly through the workstations, where the parts are progressively added
in sequence until the finished assembly is realized by the line.
The assembly line can be of three types, depending on the way the assembly is
moved:
intermittent and synchronous (the movement is intermittent -> intermittently,
the product under assembly moves i.e. every regular / fixed time period,
this movement is sync with the end of operations of all work-pieces under
assembly at all stations -> this corresponds to a line featuring the absence of
buffers between workstations);
intermittent and asynchronous (as buffers are now present b/w between
workstations, movement is still intermittent, at every regular on average
time but the work-piece is moved away from each station independently,
i.e. some independence);
continuous (operators move together with the assembly and, at the same
time, they work on it; or, similarly, the assembly is moving through the stations at a
low speed, and the operators are concurrently moving around, carrying out the
assigned operations at their workstations).
Positioning this CONFIGURATION into the scheme, the assembly line is located in the
lower corner, to the left. Such system can guarantee:
low flexibility: this is primarily due to the fact that stations are rigidly connected to
each other through the material handling system; therefore, the sequence of visited
workstations is fixed in time; this brings a structural constraint when assigning
operations to the line (i.e. when making the line balancing); in other words, the sequence
of the assembly operations which has to be performed to obtain a given finished product
will be then correspondingly allocated into a sequence of workstations (i.e. according to
the serial production flow, characteristic of this system);
low variety and high repetitiveness: at every station each operator has to execute few
assembly operations related to the product (normally few product variants); this
corresponds to high repetitiveness; high repetitiveness means high efficiency because
each worker becomes due to making many times the same, few operations very skilled
but on a small range of operations, thus increasing his/her efficiency on single tasks ->
(i.e. division of labour, Adam Smith). (minor issue: the fact that workers do not have to
switch tasks during the day further saves time and money);
high volume -> the configuration of the line is thought to meet the assembly
requirements of the specific product types allowing its quick assembly in order to sustain
high volumes along the year / high requested production capacity -> thanks to the
division of labour, which means high repetitiveness and low variety, means also low times
for the operations at each station and, so, low cycle time.
3
System description
Every line can be modeled as a series of workstations linked by the rigid
transfer system.
The handling system is integrated and could be installed on the roof or on
the floor on in between.
Thanks to the material handling system, the workpiece visits all the
stations in a serial and rigid way.
Every line (which is composed of more workstations) could be involved in the
assembly of different types of product (in the example, first line assembles
the product , the second assembles the product and the third assembles the
product ), and could be managed as:
single-model line (only one product type);
multi-model line;
mixed-model line (more flexible solution, even if it is always a line).
The entire assembly process is decomposed into elementary assembly tasks,
then these tasks are allocated to the stations; the logic of allocation is dictated
by certain constraints and/or the optimization of certain objective
functions.
Types of product
It is a solution chosen for products for which demand is high.
Typical examples of assembly lines can be found in the final assembly of
automobiles, of electric/electronic components
When some product variety is required, even at high volume, a mixed-model line may
be operated as a solution.
4
Physical structure and organization of the line:
Each station is dedicated to a few tasks, it means that, unlike the fixed
position assembly, for assembly lines each station is just equipped with its
specific tools, without the need for sharing specific tools/equipment.
Unlike the fixed position assembly, in this case necessary components are
brought to their corresponding stations, the product being assembled is
moved along the line and the finished product is taken from just one specific
point (the end point of the line). Therefore, there are just more rational flows,
from these two types, of the product being assembled and of the
component/part feeding.
8
The first historical example of assembly line was designed at Henry Fords
factory in the 1910s. It represents the first industrial application of Taylor theory.
Such theory aimed at splitting the overall complex workload in elementary
and simple activities to be allocated to many people .
8
7
This second example is very different, and developed based on principles of lean
manufacturing. It is good example, I will not underline much of these principles
in theory. In a general sense, it is a good example of an intermittent
asynchronous line (buffer size = 1 pallet), with short number of stations
(subsequent to the lower complexity of the product in terms of BOM),
integration in the line of operations of assembly and quality test and other
clever solutions which can be explained based on the lean manufacturing theory.
https://www.youtube.com/watch?v=ovMz2gTjvFA
8
9
Line is a solution for high volume, featuring efficient and limited cost of
operations.
Low WIP
Unlike the fixed position assembly, in assembly line components are stocked
just in correspondence of the afferent stations;
Moreover, the product being assemble remains at the station just for the
duration of a short cycle time (less physical quantities, then less working
capital than the fixed position assembly); and when it exits at the end of the
line it is soon delivered (to the next logistics activities).
10
system.
10
Line is a solution with the disadvantage, amongst the others, of low flexibility.
Long time required to start new productions (connected to the low flexibility,
it is equivalent to say a low product flexibility, in terms of product flexibility)
the introduction of a new product type, assuming that it could be assembled in
the existing physical line, requires at least the rebalancing of the stations, a
complex problem in general -> see next the difficulties of line balancing in
assembly lines (while, in the case of fixed position assembly, for a product
11
introduction it is simpler, as main challenge required is just to train an already versatile
workforce).
Repetitive work
Operators execute a few tasks which have really brief durations (in terms of minutes)
-> it may lead to dissatisfaction (see the case of lines wherein assembly processes are
distributed in very small portion to each station -> Ford T).
11
Assembly lines can be:
paced, if the transfer is intermittent/synchronous (with absence of buffers
between workstations) -> machine/operator-paced;
un-paced, if the transfer is intermittent/asynchronous (with the presence of
buffers between workstations);
continuous flow, if the transfer through stations is continuous while the
operators are making the assigned operations within the station ().
12
Paced lines (machine-paced lines)
Workstations are not separated by buffers.
The material handling/transfer system is generally a conveyor which moves the pallet/base where the parts are placed
and assembled.
The material handling/transfer system generates an intermittent movement:
during the transfer phase, the conveyor moves simultaneously every base/pallet from a station to the
following one;
then it stops for a period which lasts the cycle time of the line (this is giving the pace of the line -> the
time allowed for an operator to work on the job is limited;
during the stop phase operators can execute their assigned tasks (and should complete their operations ).
Criticality:
At each station/operations cycle, there is the risk/possibility of failing to complete the assigned tasks within the
cycle time. This problem is due to the time variability of manual activities (i.e. occasionally the operator could need
more time to finish his/her tasks, even if the average time of assigned tasks to the station is lower than the CT, in this
scenario the assembly jobs move anyway to the following stations every CT, but the following operators can make
only the operations that have precedent operations already done, precedence constraints -> this is limiting their
activities in relation to the state of the product being assembled -> it results in an unfinished piece at the end of the
line).
To avoid the problem of no completion:
a first solution could be the increase of the cycle time, but this implies a reduction of the line production
capacity (thus not meeting the target demand);
alternatively, it could be increased the number of stations (to allocate fewer operations / fewer times at each
station), this implies the disadvantage of additional costs (i.e. additional installation costs, manpower costs due
to more stations, large is the line -> more stations/more WIP/more WIP costs/more space costs, ).
13
Strengths: cycle time and production capacity are perfectly controlled
Since the handling system moves automatically the pieces, as it starts every
regular time given by a timer, there is perfect control of the cycle time, thus
there is an accurate control on the production capacity.
14
Strengths: no problem of unfinished pieces
The possibility for the operators to give their approval before the product
moves, eliminates the problem of unfinished pieces.
The risk of giving approval too late compared to the theoretical cycle time:
increases with the operator utilization level (i.e. when the sum of durations of
the assembly tasks allocated to a station is close to the theoretical cycle time,
risk is higher);
increases with the rise of the number of stations (because having more
stations increases the probability of one station leads to a delay).
15
The ideal solution would be that operators could help each other to minimize the delay
generated with respect to the CT (help chain (?)). However this solution requires some
additional conditions:
workers should be multi tasking/multi skilled, since they need to be capable to execute
other tasks different from their owns, assigned to their stations;
stations should be close, because operators need to reach other stations in a short time;
cycle time should be sufficiently long, because operators need a certain time to
provide the help intervention to other stations;
assembled pieces should not be too small, otherwise two operators couldnt work
together on it.
15
Unpaced lines
Workstations are separated by buffers (where workpieces can be stocked waiting for
next operations at downstream workstations). Such buffers should be opportunely
sized, they often may represent portions of the transfer system itself (e.g. conveyor
belt).
Unlike paced lines, in this case operators are not forced to finish their tasks within
the cycle time: occasionally, they can also spend more time to finish them because,
while they are still busy in finishing their activities, buffers avoid line stoppages
because they enable to feed downstream stations with the already stocked pieces
and provide room to stock pieces produced from upstream stations.
REMARK: clearly operators cant systematically exceed the cycle time, otherwise
downstream buffers would result empty and/or upstream buffers would result full. Which
means stopping the production of respective stations (up/downwards) and, by
propagating to the bottleneck station, a reduced production capacity of the line.
More precisely, since buffers have finite capacity, two kinds of problems could happen:
problem of blocking: when a buffer is full the upstream station cant drop the piece
(having finished its assigned task on the current piece);
problem of starvation: when a buffer is empty the downstream station cant take a
new piece (having finished its assigned task on the current piece).
If these events are frequent, there is a risk for the reduction of the line production
capacity. The frequency (of such events) depends on:
the sizing of buffers for this reason it is important during the design of the line- to
size buffers correctly (to reduce probability of blocking/starvation);
the balancing of the line (the presence of stations overloaded with work increases the
frequency of these events directly down and upstream).
16
Strengths: no problem of unfinished pieces
thanks to the presence of buffers, which decouples operations at different
stations and thus allows to conclude the tasks at each station.
Strengths: cycle time can be exceeded, but only occasionally (i.e. problems of
blocking and starvation)
thanks to the presence of buffers, operators are not forced to finish their tasks
always within the cycle time: occasionally they can also spend more time to
finish them because, while they are still busy in finishing their activities,
buffers avoid line stoppages because they feed downstream stations with the
stocked pieces.. etc.
clearly they cant systematically exceed the cycle time, otherwise downstream
buffers would result empty and/or upstream buffers would result full.
Weaknesses: cycle time and production capacity are not perfectly controlled
the cycle time depends (on average) on the bottlenecks operator/station,
which then limits the production capacity;
the propagation of blocking and starvation may affect the cycle time of the
line, i.e. affecting the bottleneck; but this depends on the sizing of buffers, that
can reduce (or not) the probability of blocking or starvation; this adds to the
natural variability of the bottleneck itself.
17
How does it work?
The material handling system moves at a constant and very low speed + Every station is coincident with a (physical) portion (part)
of the line (of a length L).
Every time a workpiece fixed at the handling system enters a station:
the operator (of the station) goes up the platform where the product is fixed, moves with it within the station while
making the assigned assembly tasks, when station ends he / she goes down on the floor and back to the begin of the
station;
The operator may stand and move along the station making the assigned assembly tasks, while the assembly is moving; in
case of small products, he / she may be sit and move slight movements forewords on the line.
Given the conveyor velocity vc, the distance D between two consecutive pieces (blocked on their respective bases / on the platform)
determines the cycle time: CT = D/v c.
Note that the length of a station is a quantity necessarily larger (than) or equal to the distance D:
it can never be smaller than D -> it means giving to operators a time which is at least equal to the cycle time, in order to complete
the assigned assembly tasks;
indeed, when L > D, then L > CT * v c it means the conveyor acts as a buffer the operator has some extra time to finish his /
her activities.
In continuous flow lines, the increase of buffers size (i.e. increase of the stations size / length) implies:
an increase of the actual production capacity due to a decrease of the unfinished pieces phenomenon (the theoretical
cycle/production capacity is fixed as it is defined by D and vc -> so it is reduced the loss due to unfinished pieces;
also the occupation of more space, longer flow time to traverse the line, and higher WIP level, as there are both pieces within
stations and buffers).
(by comparing to what it is happening in un-paced lines, there was an increase of the actual production capacity due to a decrease of
effects of blocking and starvation thanks to better / higher buffer sizes -> i.e. for other reasons, not unfinished pieces, but probability of
starvation/blocking -> the actual production capacity is increased).
PROBLEM: In continuous flow lines, since the handling system moves constantly the parts, there is perfect control of the cycle time, thus
there is an accurate control on the production capacity. However (as for paced lines), as a consequence of having perfectly controlled
cycle times, there is no guarantee of the completion at the stations, thus there are problems of unfinished pieces.
As for paced lines, this problem can be solved by introducing the possibility for operators to stop the line, then it is possible to distinguish
two cases of continuous flow lines:
Case 1: operators cant stop the line;
Case 2: operators can stop the line.
18
Strengths: cycle time and production capacity are perfectly controlled
Since the handling system moves constantly and automatically the parts, there
is perfect control of the cycle time, thus there is an accurate control on the
production capacity.
19
no problem of unfinished
Strengths:
pieces
If any problems, the possibility for the operators to stop the line eliminates the
issue of unfinished pieces.
20
PC assembly
intermittent, asynchronous
conveyor through which the operator, sitting in his / her work bench, pushed
the assembled piece downard
material feeding done on the other side of the workbench (the one not adjacent
to the line.
21
Different solutions in the automotive sector of continuous flow line:
Cabin truck (the cabin or cab of a truck is an enclosed space in a truck where
the driver is seated) is moved on a platform and the operator is seated in the
cab to make the assembly task
Car assembly car body has already painted (in a paint shop) is now subject
to a final assembly; the operator standing by the line is making the assigned
operations / car door assembly similar solutions but as a preassembly
In both cases, components are stocked on the back along the
line/station
In both cases, the transporter is an overhead conveyor for car bodies
http://www.bmwgroup.com/e/0_0_www_bmwgroup_com/produktion/f
ahrzeugfertigung/automobilfertigung/erlebnis_produktion/prod_prozes
se.shtml
22
Design of Assembly Systems Assembly
lines (part 2)
2
Design of a manual assembly line
q Assembly process
Assembly Graph
(showing precedence
relationships amongst
operations)
3
Balancing constraints
q Cycle time
q Precedence relationships among operations
q Incompatibility between operations that cannot
be assigned to the same station
q Opportunity or necessity to assign some
operations to the same station
q Constraints related to space
q Constraints related to workers
q Constraints related to the material feeding
4
Balancing objectives
q Technical objectives
Minimizing the number of stations, given the cycle time
Minimizing the cycle time, given the number of stations
Minimizing the total idle time
N
IT = n CT i =1 ti
where:
n = number of stations
CT = cycle time
N = number of assembly operations
ti= time to perform operation i (i.e. unit working time)
5
Balancing objectives
q Technical objectives
Minimizing the probability of no completion
in a machine-paced line, or
in a continuous flow line, in case the operator cant stop the line
Minimizing the probability that the times of operations in
one or more stations exceeds CT
in an operator-paced line, or
in a continuous flow line, in case the operator can stop the line
6
Balancing objectives
q Economical objectives
Minimizing the total expected cost (TEC)
TEC = LC + E_CUT
LC = line cost (equipment cost + operators cost)
E_CUT = expected cost of unfinished operations (i.e. tasks)
7
Performance indicators
q Balance Efficiency
E = T / (K CT)
where:
K is the number of stations
q Balance Delay
D = (K CT T) / (K CT) = 1 - E
8
Line balancing A simple method
t
iS
i
UR =
CT
where:
ti = mean time of task i
S = set of tasks assigned to the operator
CT = cycle time
9
Line balancing A simple method
Steps
1. Draw the precedence graph (assembly graph)
2. Calculate the total tasks time T (sum of all tasks times)
It can be defined as total assembly work content or,
equivalently, total assembly time
3. Calculate the cycle time CT
CT = available time / demand
4. Calculate the minimum number of stations K*:
K* = T/(CT )
10
Line balancing A simple method
11
Line balancing Positional Weight Heuristic
Prioritization
A task is prioritized by the cumulative assembly time
associated with the task itself and its successors
PW(i) = ti + w tw ; w in S(i)
where:
S(i) successor tasks to task i
12
Line balancing Positional Weight Heuristic
Prioritization steps
1. Task ordering
For all tasks i, compute the positional weight PW(i)
Rank tasks by non-increasing PW
2. Task assignment
For ranked tasks, assign task i to the first feasible
station (obey the precedence relationships; do not
exceed cycle time; obey other constraints)
13
Positional Weight Heuristic Example
CT = 70
14
Positional Weight Heuristic Example
15
Positional Weight Heuristic Example
16
Line balancing with parallel stations Example 1
17
Line balancing with parallel stations Example 1
18
Line balancing with parallel stations Example 2
19
Line balancing with parallel stations Example 2
20
Line balancing Probability of no-completion
Pk P*
where:
Pk = probability of no-completion of task k
P* = maximum probability of no-completion
21
Line balancing Probability of no-completion
Steps
1. Calculate the remaining time related to task k
RTk = CT ti
iS
where:
CT = cycle time
ti = mean time of task i (time required to perform
task i)
S = set of tasks assigned to the operator (task k
included)
22
Line balancing Probability of no-completion
23
Line balancing Probability of no-completion
24
Line balancing Probability of no-completion
25
Line balancing Probability of no-completion
26
Probability of no-completion - Example
Station x
P* = 0.1 CT = 100 PI?
Task (ti ; i) RTI = 100 (56+15+12+12) = 5
D (56; 5) (52+12+0.82+0.82) = 5.22
E (15; 1) ZI = 0.96
H (12; 0.8) (ZI) = 0.8315
I (12; 0.8) PI = 1 - (ZI) = 0.1685 > P*
The constraint (PI P*) is not satisfied
27
Line balancing A linear programming method
i = index of task (i = 1, ..., N)
k = index of station (k = 1, ..., K)
CT = cycle time
ti = mean time of task i
cik = cost coefficient such that Ncik ci k+1 (k = 1, ..., (K-1))
IP = set of task pairs (u,v) with a precedence relationship
ZD = set of task pairs that cannot be assigned to the same station
ZS = set of task pairs that must be assigned to the same station
Objective function
N K
c
i =1 k =1
ik xik = min
Subject to
N
t x
i =1
i ik CT k = 1, , K
K
x ik =1 i = 1, , N
k =1
h
xvh xuj h = 1, , K (u, v) IP
j =1
29
Line balancing A linear programming method
30
1
Design of a manual assembly line
For this type of systems, the design also consists in determining the number of stations which constitute the line.
The first step is to identify the production mix: quantities and types of products to realize during the expected useful life of the line. For the sake of simplicity, we now refer to the single-
model line. We need to know the target demand.
INPUT 2 Afterwards, it is necessary to define/evaluate the time of each assembly operation/task. Since durations are not deterministic, they are difficult to be determined (unlike parts
production lines). There are 2 different methods to evaluate the time of each assembly operation:
using the MTM (Motion Time Measurement) approach, which using appropriate tables, allows to calculate durations of assembly operations. This method decomposes the manual
operation into the basic motions or human movements required to perform it, and assigns to each motion a predetermined standard time which is determined by the nature of the
motion and the conditions under which it is made; when all such times are added up, it provides standard time for the assembly operation/task; standard times can be finally determined
by adding suitable allowances (which take into account fatigue and working conditions of the operator). This approach is:
time consuming and expensive;
accurate.
observing and measuring by means of a chronometer the execution of an operation for a certain number of times, then calculating the mean value and standard deviation. This
approach is:
time consuming as the number of times must be sufficiently high to guarantee the statistical significance of the summary statistics;
allows to have reliable estimations (since it is calculated based on selected operators).
Independently from the evaluation method, this step allows to associate to every operation its mean execution time and standard deviation.
INPUT 3 Knowing the requested production capacity (P) due to the target demand, it is then possible to calculate the requested cycle time: CT = 1/P -> in ALB type 1
CT is a constraint/input
DESIGN PROBLEM Afterwards it is necessary to balance the line (assembly line balancing ALB): to allocate the operations to the different stations, respecting constraints and
considering specific objectives. It is then known the required number of stations as well as the operations assigned to each stations.
When solving ALB, it is possible to size the workforce: the number of operators doesnt necessarily coincide with the number of stations.
it could be decided to have more operators in certain stations (possible in case of bulky products) -> parallelization of resources;
It could be decided to use a pool of jolly operators (to replace workers of the line when they have a break);
it should be decided a pool of workers considering absenteeism, illness.
This is a generic model. In particular the assembly line balancing is an activity equally important for all the types of assembly lines. Obviously, there are instead specific aspects which
should be considered when dealing with particular types, for instance:
in un-paced lines buffers must be correctly sized, to guarantee an adequate decoupling of operations b/w stations;
in continuous flow lines the length of every station (i.e. buffers) and conveyor velocity must be defined.
2
Example of definition of the balancing constraints:
the precedence relationships among operations (which are synthesized
in the table) in this example are showed by the Assembly Graph;
the assembly process table is an useful and immediate tool which is
helpful during the design phase, for consultation of predetermined data .
Respect the cycle time
The workload at each station (due to the allocated operations) can not exceed the cycle time, otherwise production capacity
is affected.
E.g. 1 machine-paced lines -> high number of unfinished pieces / E.g. 2 un-paced lined: the average time of the bottleneck >
CT so not respect of requested PC
4
In reality, ALB problem can be solved, subject to constraints, and with reference to
different objectives:
technical objectives
economical objectives (unlike the technical ones, these explicitly consider cost
data)
TECHNICAL OBJECTIVES
Minimizing the number of stations, given the cycle time as constraint (1st type
ALB Assembly Line Balancing problem) This is the most frequent case in practice:
given the cycle time (determined by the requested production capacity), it is
minimized the total number of stations (therefore it is minimized the total
number of operators, which is a cost item).
Minimizing the cycle time, given the number of stations (2nd type ALB
Assembly Line Balancing problem) This case is not at all frequent in reality: given
the number of stations, it is minimized the cycle time, therefore it is maximized the
production capacity.
Minimizing the total idle time The total idle time IT is a residual/remaining time
(explain the meaning, e.g. case of paced line, N stations each of them has the time
allowed for an operator to work while the sum is the total time required to make the
assembly) -> it represents the time not used in order to execute assembly
operations. Minimizing the total idle time means solving one of the two problems: 1st
type ALB or 2nd type ALB , precisely:
Given the cycle time, minimizing the total idle time means minimizing the
total number of stations (since the sum of the times needed to perform the
required operations, depending on the technology used for the assembly, is
fixed);
Given the number of stations, minimizing the total idle time means
minimizing the cycle time.
5
TECHNICAL OBJECTIVES These objectives are two sides of the same coin:
Minimizing the probability of no completion
in a machine-paced line, or
in a continuous flow line, in case the operator
cant stop the line
Minimizing the probability that the times
of operations in one or more stations
exceeds CT
in an operator-paced line, or
in a continuous flow line, in case the operator can
stop the line.
Minimizing this type of objective function can lead to unsatisfactory results, particularly:
in situations where the cycle time is given (ALB 1st type), it is likely to generate a
number of stations excessively high (deciding for having a lower number of
operations to avoid unfinished operations allocated to each station, therefore a
lower probability of no completion / of stopping the line is obtained -> means an
higher number of stations);
in situations where the number of stations is given (ALB 2 nd type), it is likely to
generate an excessively high cycle time (and consequently an insufficient production
capacity) as we would like to give higher CT with respect to the time of the assigned
operations).
6
ECONOMICAL OBJECTIVES
Minimizing the total expected cost (TEC) The following assumptions may be introduced:
the unfinished subassembly continues to proceed along the line, then workers perform only the operations which are not affected by non-completion / unfinished operations
other cases may be possible to manage the unfinished operations;
at each station only one operator works
other cases may happen with more operators in the same station (parallelization).
TEC = LC + E_CUT
Since we are talking about unfinished operations, this objective applies only for machine-paced lines or in continuous flow lines.
LC = line cost (equipment cost + operators cost):
operator cost depends on the number of shifts;
equipment cost depends on the installed facilities.
LC is proportional to the number of stations (in realty this relationship is not linear because in one station could work more than one operator or because of jolly operators
etc.).
There is a trade-off between the line cost (LC) and the expected cost of unfinished operations (E_CUT):
It is possible to reduce the probability of no completion (then the related costs, E_CUT), by reducing operators utilization - which means to increase the number of stations
and/or the cycle time (ALB 1 st or 2 nd type). Consequently there is an increase of the line cost (LC).
It is possible to reduce the line cost (LC), increasing operators utilization which means to reduce the number of stations and/or the cycle time (ALB 1 st or 2 nd type).
Consequently there is an increase of the probability of unfinished pieces (then E_CUT increases).
The objective of minimizing the total expected cost allows to overcome the limitations related to previously described technical objectives: the optimization of the 1 s t
type ALB problem or the 2 nd type ALB problem or the minimization of the total idle time produce solutions leading to a solution on the left side of drawn diagram: they find solutions
which correspond to the choice of lower numbers of stations i.e. (higher saturation/utilization) such as to have lower line costs (LC). But, looking at the graph it is evident that, from the
total expected cost point of view, none of the solutions placed on the left side is the best. TEC is minimized at the intermediate values of number of stations (hence intermediate utilization).
Regarding the unfinished subassemblies management policy, it is possible to identify different cases, each of them determines different values of the cost of unfinished operations:
the unfinished subassembly continues to proceed along the line, then in the downstream stations, only the operations not affected by no-completion are executed (in the assembly
graph operations, that are NOT directly or indirectly connected to the unfinished operation); therefore the piece needs to be finished after its exit from the line, in a hospital station;
the unfinished subassembly is immediately removed from the line and finished in a hospital station; this solution -as the previous one- implies the operating costs corresponding
to the hospital station;
if in a s tat ion the cycl e tim e is exc eeded, the l ine is stopped unti l the complet ion of the operat ions; in this c ase costs are generat ed by the opportunity cost linked to the decre ase of
the production capa city. This policy is typic al for Just in T ime systems , be cause i t allows great er quali ty control (unfinished item s are not moved) and advantages in terms of tim e
(no extra work to finish operations). Moreover, in JIT syste ms, tim e vari abil ity is reduced by the presenc e of mul ti tasking operators s ince they help ea ch other (an operator who has
already finished his operations can help another who is nearby);
there are jolly operators, who can intervene when some other operator needs help; in this case the cost is generated by the presence of the jolly operators; there are multi-
skilled operators which can help adjacent operators (i.e. if they have finished their assigned tasks).
7
BALANCE EFFICIENCY Balance Efficiency is the time required for
completing the assembly process with the respect to the time allocated
(allowed) in order to complete it (along the line)
T = total time needed to perform the required operations, it depends on the
technology used for the assembly, then it is fixed; T=ti
K = actual number of stations resulting at the end
of the design
0 < E 1 The balance is efficient if the time
allowance is slightly longer than the time needed
(it means that E tends to 1); the balance is
inefficient if the time allowance is much longer
than the time needed (E tends to 0).
8
ASSEMBLY LINE BALANCING methods There are a few methods /
algorithms in order to formalize and solve the ALB problem of a manual
assembly line, following the previously described constraints and objectives.
The presented models refer to the single-model line.
9
to the same station; then another operation is
considered and it is verified the possibility to
allocate it to the same station (again, verifying the
precedence constraints and the UR constraint);
if UR is not , it is created another station and
the operation is allocated to the new station, etc.
The procedure ends when all the operations have been
assigned to stations.
9
Method I: maximum value of utilization rate of operators imposed as a design
criterion - Description of the steps of the method:
10
STEP 5. Assign tasks to stations, respecting the constraints (CT, precedence
relationships, a, etc.)
If there is more than one task available to be assigned, use a rule to prioritise tasks.
11
Method II: Prioritization
Unlike the method of the maximum value of utilization rate of operators imposed
as a design criterion, this method allows to find an unique solution; anyway, since
procedure is heuristic, this solution can not be the optimal one.
The basic logic of this heuristic methodology is to give higher priority to tasks
followed by higher numbers of tasks. Such attribution allows to:
- respect precedence constraints;
- set free a greater number of branches of the assembly graph (i.e. executing
first tasks which take precedence over the other activities, in the following stations it
will be granted greater freedom in choosing between the remaining tasks).
12
STEP 1. Task ordering
1) For all tasks i, compute the positional weight PW(i)
PW(i) is the positional weight of the task i. It is calculated following
the assembly graph.
The times required to perform the operations/tasks and
precedence constraints are taken from the assembly graph.
2) Rank tasks by non-increasing PW (thus ranking the PW(i) in non
ascendant order.)
13
Example
14
STEP 1 Task ordering
PW(a)= ta + wTw = 20 + 6 + 5 + 5 + 15 + 10 + 15 + 46 +
16 = 138
(..)
Task a rank 1
15
STEP 2 Task assignment
When at station 1 it is not assigned any task a the time remaining is the CT =
70
When at station 1 it is assigned task a the time remaining is the 70 20
(duration of the task a) = 50
.
The assignment ends when the CT is totally exploited (no more tasks can be
assigned to the corresponding station).
Note about utilization UR lower than 90%, max time allowed to make the
work is 63 (so -7), so we have to move downward (to 2 nd station), task g.
16
The minimum number of stations is T/CT = 15/4 = 4
But (by considering precedence constraints and the CT constraint) it is
clear that in reality the minimum number is 5 stations: AC _ B _ D _ EF_
G (B and D should be always alone as there are no possibility to add short
time operations with them).
We cannot reach the optimal number of stations unless parallelizing
stations: if we parallelize we can obtain 4 stations (i.e. each workload of
parallelized working position could be CT*2 -> more time window /
allowed to make the assembly; in general a working position within a
station (of K parallel working positions) has a time allowed for assembly
equal to CT*K
By parallelizing it is possible to make a better line balancing, as we are
reducing constraint on the time allowed at each station.
Where to parallelize? Heuristics: when there are operations in the
sequence which required loosened constraints (i.e. relax /
loosen constraints) -> those that have higher time requirements
(i.e. and cannot be divided in further subtasks)-> B _ D (higher
times) to make
17
18
As the time required for the operation/task B is 5, you should have to accept
a new CT=5 ( > CTinput=4), thus implying a reduction of the PC (alias not
feasible solution to respect a given CT);
otherwise you can parallelize the station in which the operation/task
B is done, thus not reducing the PC -> we avoid a PC reduction.
This solution may become necessary whenever PC is very high -> CT would
be low, incompatible with the time requirements of each single operation
that cannot be divided anymore (thus we rely on the effect on an higher time
window for the operations within the working position of the stations,
K*CT when we parallelize to make the solution feasible). Of course we
can also parallelize the whole line; CT of the line is CT = m/PC -> so effect
on an higher time window for the operations within the working position
of the stations
19
Pros/Benefits deriving from parallelization:
(Higher availability with reduced prod capacity -> see what said for
transfer lines)
Better balancing (see example 1) -> reduced number or resources
Feasibility to respect constraints on the PC even with operations
with larger time durations (see example 2) -> guarancee to keep
high PC
Job enlargement since workers are required to do more
operations, ; -> job satisfaction.
20
Method III: Managing the probability of no-completion This method has
some analogies with the method of the maximum value of utilization rate of
operators imposed as a design criterion. For each task this constraint has to
be satisfied: Pk P*
Starting point: the line doesnt exist. Then, it is created the first station:
considering an operation/task which has not precedence constraints, it is
calculated its probability of no-completion (Pk) and, if it is lower than P*
(if this condition is not verified for the first
assigned operation, the problem doesnt have a
feasible solution), this operation is allocated to
the station.
21
Afterwards, checking that precedence constraints are
fulfilled, another operation is considered:
if its Pk summed to the previous Pk is still P*, the
new operation is allocated to the same station. Then
another operation is considered and it is verified the
possibility to allocate it to the same station (again,
verifying the precedence constraints and the Pk P*
constraint) .
if its Pk summed to the previous Pk is not P*, it is
created another station and the operation is allocated
to the new station.
The procedure ends when all the operations of the
assembly graph have been assigned to stations.
21
RTk = remaining time it is the mean remaining time after the operation
k is assigned
22
23
For normal distribution:
P(is Ti CT) probability of completion
1 P(is Ti CT) probability of no-completion
24
Values of (Zk)
25
Demonstration of the relation Pk = 1 (Zk) :
Then:
is (Ti is ti )/is i2 ~ N(0,1)
26
If it is overcame the threshold of P*:
it is possible to accept it implies the choice of an higher expected cost of
unfinished operations (E_CUT) but this may be not unacceptable (it may
be acceptable when it induces low increase of Pk, especially with
additional operations with low std. deviations)
it is possible to parallelize stations it doesnt just solve the problem but
it also allows to exploit other opportunities (see previous examples) -> that is:
relaxing constraints may mean also giving more possibilities in order to
better assign/optimize assignment in a station, thus reducing /
accommodating probabilities of no completion.
27
Method IV: linear programming method
ti= mean time of task i (is a deterministic value because of the assumption A)
cik = cost coefficient such that Ncik ci k+1 (k = 1, ..., K-1)): push to fill in first the
previous station. It is a fictitious cost for pushing to utilize the previous station.
IP= set of task pairs (u,v) with a precedence relationship (It is possible to use a
matrix to enumerate the precedence relationships in a quick way i.e. Hoffman
Matrix)
28
The model requires to predefine the number of stations K. Such value should be
sufficiently high to enable a feasible solution; to minimize operators idle time, the
model will force as much as possible operations in the first stations, leaving the
others empty. The empty stations will be then eliminated. Thus, cost coefficients cik
have precisely that role to force the utilization of the first stations before using the
others.
Indicating with K* the number of stations identified by the model, the resulting
solution can be evaluated using the balance delay.
This objective function does not allow to allocate equally the total idle time between
stations. It would be better to allocate equally the total idle time between stations:
to ensure more flexibility in production volumes;
to ensure greater job satisfaction;
to better accomodate the variability of the time to perform operations/tasks (it is
better acceptable a growth of P* along the line, not a full utilization in first stations)
29
4. The fourth constraints class ensures that if u and v belong to ZD, they are
not assigned to the same station (negative zoning)
5. The fifth constraints class ensures that if u and v belong to ZS, they are
assigned to the same station (positive zoning)
30
Design of Assembly Systems Assembly
lines (part 3)
2
Effect of buffer in a line Example 1
ST = 30 (40 %)
ST = 72 (50 %)
ST = 60 ST = 120 (10 %)
Station 1 Station 2
3
Design of unpaced lines buffer size
4
Design of unpaced lines buffer size
5
Design of continuous flow lines station length
D = CT * V = V / PC
where:
CT = cycle time
V = conveyor speed (constant)
PC = production capacity
6
Design of continuous flow lines station length
FT CT L/VD/V LD
where:
FT = flow time
CT = cycle time
V = conveyor speed (constant)
7
Design of continuous flow lines station length
8
Design of continuous flow lines station length
Open stations
9
1
Design of un-paced lines
Decoupling function
Due to the decoupling function, an operator in a station can occasionally exceed (go beyond) the CT (i.e.
the requested CT in order to reach the target PC), as the station downward and upward can continue their
operations, respectively thanks to the material available on progress (i.e. downward can be fed by a new
piece) and the remaining space / capacity of the buffer (i.e. upward can drop the piece).
REMEMBER THAT: Since buffers have finite capacity, two kinds of problems could happen (when a
station up/downward finishes its operations):
problem of blocking: when a buffer is full the upstream station cant drop the piece;
problem of starvation: when a buffer is empty the downstream station cant take the piece.
If these events are frequent, there is the reduction of the line production capacity (make example). The
frequency (of such events) depends on:
the sizing of buffers for this reason it is important during the design of the line to size buffers
correctly.
Here the focus is on the sizing of buffers! -> go to example 1 and then back here
In un-paced lines buffers have to be sized. Two methodologies allow to do it, as they enable to model
variability:
analytical approach, based on queuing theory, which requires the formulation of hypotheses (for instance
it has to be assumed that the service time follows an exponential distribution) that rarely fit to reality;
simulation, which allows to use various probability distributions and provides more realistic results.
2
Effect of blocking and starvation -> loss of time due to these phenomena at
each station + loss of PC at the end of the line (over a given period).
Thinking on the line dynamics, due to variability and limited buffer size,
blocking and starvation may happen and this would lead to losses.
Lets make a scenario analysis (which of course is one example, out of many, that
can happen with some probability)
Scenario when blocking happens -> sequence of 120 sec at station 2
Start at WIP = 0 pieces + Limit of Buffer size = 2 pieces; WIP is
growing up to reaching the limit;
When WIP reaches the limit, blocking happens: at the end of 5
operation at station 1 (5 cycle), blocking happens -> lost time at
station 1;
In the mean time, 3 pieces were exiting the line (every 120 sec).
Scenario when starvation happens -> sequence of 30 sec at station 2
Start at WIP = 2 pieces (@ Buffer size limit = 2 pieces); WIP is
decreasing up to reaching the limit (zero);
3
When it reaches the limit, starvation happens: at the end of 5 operation at
station 2 (5 cycle), starvation happens -> lost time at station 2
In the mean time, 5 pieces were exiting the line (every 30 sec), instead of 6 as
potentials in the period.
Overall, there is a lost capacity (1 piece lost if compared to average expected for the
period): how many pieces would happen w/o considering the Buffer size limit
-> balanced, make the average over the whole mix
-> balanced but considering the variability and buffer size limit, the two
scenarios show that in a given period there is a loss of 1 piece ; this happens
as combined effect of blocking and starvation: due to buffer size station 1
was blocked, so we lost opportunity to stock more materials when upward
station was quicker usable to feed stations in future periods when the
upward was slower . So we experienced at the end of the period 1 lost piece
out of the line (with respect to the line pace) -> in the short term period
experienced, we would expect 9 pieces exiting. 8 instead exited.
In the long term period these losses may be avoided (or reduceed) if we
increase the buffer size (think e.g. of a BS equal to 3, the same scenario would
not experience neither blocking nor starvation, so no lost PC).
3
It is possible to identify a typical behavior of the production capacity of a line as a
function of:
the buffer size;
the coefficient of variation (CV = standard deviation of the times required for the
assembly tasks / mean of the times required for the assembly tasks).
This can be proven by simulation / other methods (even the scenario analysis done was
showing that, with increased buffer size, given the variability).
To synthesize, considering all these relevant aspects, it is crucial to find the optimal
buffer size.
4
First two points are a summary of the previous slide, where the figure showed
why buffer size has to be increased to protect from assembly time variability.
5
Design of continuous flow lines
The conveyor moves at constant and really low speed.
When a piece gets in a station (which corresponds to a segment / portion of the line), the operator climbs on the transfer (material
handling) platform and, while moving, he / she performs the assembly operations. When he has finished, he steps down the platform
and he walks back to the starting point.
Other solutions can be: i) the operator may stand and move along the station making the assigned assembly tasks, while the assembly
is moving; ii) in case of small products, he / she may be sit and move slight movements forewords on the line.
Given the conveyor velocity vc, the distance D between two consecutive pieces / assemblies (blocked on their respective bases / on the
transfer platform) determines the cycle time: CT = D/v c. To make the distance D (thus being in the same position of precedent piece), the
successive piece requires CT, given the conveyor velocity V -> the distance D between two consecutive pieces determines the cycle time:
CT = D/V. Having the distance exactly equal to CT*V means that a piece comes out the line each CT.
6
(Reading from left to right) Given the flow time, it is possible to determine the
station length. Station length should be, at least, equal to the distance between
consecutive assemblies.
(Reading from right to left) The longer is a station, the more an operator has a
time buffer to complete his / her tasks in the eventuality he may exceed
(occasionally) the cycle time. The more is FT, as a time frame (time buffer)
within it is possible to conclude.
7
The longer is a station, the more an operator has a time buffer to complete his tasks in the
eventuality he exceeds the cycle time (as it is shown in the figure).
This condition pushes to adopt very high station lengths. However it determines the
disadvantage of having high investment costs and space occupancy costs.
Furthermore, since increasing stations length, as it is fixed the distance between consecutive
assemblies, there are many products being assembled along the line, and so a high WIP
(increases the working capital).
To synthesize:
the distance between two consecutive assemblies is related to the cycle time and the
production capacity; to modify such a distance means to modify the cycle time, thus the
production capacity;
the station length is not related to the production capacity, but it affects the probability of no
completion, the overall length of the line (and, so, the investment costs) and the quantity of
products WIP which are being assembled along the line; therefore the optimal solution of buffer
sizes would be based on cost, considering expected cost related to the unfinished pieces + line
cost (including WIP).
Techniques to make the buffer sizing are similar as in the case of unpaced lines.
analytical approach, based on queuing theory, which requires the formulation of hypotheses
(for instance it has to be assumed that the service time follows an exponential distribution) that
rarely fit to reality;
simulation, which allows to use various probability distributions and provides more realistic
results.
The behaviour observed by means of a simulation starts from the Lmin, and measures the prob of
no completion (# unfinished pieces/tot at the station(s)) which is reduced as lenght increases (as
it could be expected) -> reducing the ECUT but increasing the LC.
Remark in case of a Continuous flow line when the operator can stop the line: the number of
times this happens is low if L is high; this loss of PC is then low. There is still a trade-off.
8
In continuous flow lines stations can be all open or closed (but in the same line
there could be both closed and open stations in the same line).
Open stations stations dont have defined boundaries that separate each
other, then the operator can easily cross over them, in order to anticipate begin
or to postpone end of his / her assigned operations. In this case, along the line
there are areas which are shared by operators who belong to adjacent stations:
an upstream zone, which the operator of the station k shares with the operator
of the station k1;
a downstream zone, which the operator of the station k shares with the
operator of the station k+1.
This does not imply, necessarily, that the operator can work concurrently in
the shared area/segment as there are still the constraints as precedence
relationships amongst operations.
9
Closed station operators cant cross over stations because it is unwanted or sometime
it is impossible. For instance, a closed station is strictly required:
if in the station particular environmental conditions (isolated because of painting
operations using spray, heating chambers, -> example with robots, automated);
Physical interference problems b/w operators of adjacent stations are undesirable.
9
1
3
In recent times, the need to offer diversified products (product differentiation
as business strategy, products with different variants / optionals offered to
the market) has strongly affected assembly systems. There are 3 possible
configurations (involving different kinds of production mix management):
Single-model (more lines, one for each product);
Multi-model;
Mixed-model.
Single model It is the most immediate solution: there is one dedicated assembly
line for each product model to be assembled. It is adequate when the production
volumes are high and stable. Main characteristics of this solution are:
low flexibility: each line is dedicated to a single product type; any kind of
change, both in product features and in product quantities, can lead to heavy
changes in the line design (rebalancing/reconfiguration);
easy management: since each line assembles a single product type/model,
there is no complexity in managing flows of materials (i.e. managing flows of
the different components required in accordance to the product model) and
information (operators need NOT to be informed about tasks to be performed,
no new instructions as the model is always the same).
Multi-model The line can assemble more than one product type (more than one
model / variant). Anyway, products are managed in batches since their setups are
relevant in cost and time (i.e. setups consist primarily of replacing components). As a
consequence, along the line, it is not possible that two different products are assembled at
the same time (except in transients when changing the product types with no line
stoppage). Main characteristics of this solution:
batch sizing and sequencing: batches need to be properly sized and sequenced; batch sizing =
f(inventory holding costs, setup costs); batch sequencing may be sequence dependent, when
changing from one product model to the other;
cycle time and number of stations dependent on the product type to be assembled -> you
should find a compromise in solutions; it can happen that there may be high differences in the
number of stations required by each product model (problem setting: given their requested CT,
determine the number of stations) as the time requirements are different-> problem in number of
operators / their excess;
high stock of finished product as, due to high set-up, it is not possible to chase the product rate;
inventory are higher as higher batch size are required.
Mixed-model The line can assemble more than one product type and there is no need to
manage products in separate batches owing to the strong reduction in setup times. It is
possible to realize sequences of product types totally different from each other (even
achieving the unitary batch size, that is now possible).
Typically, this solution is associated to continuous flow lines or un-paced lines. Instead,
it is not associable to paced lines because the cycle time should be, in this case, fixed to
the longest assembly time / cycle time sought among all the assembled product types (to
be able to conclude in this time window all the operations assigned); then, it would lead to
a loss in production capacity because such cycle time would be too long for the majority
of the product types. Main characteristics of this solution (to be discussed later in detals):
need to reduce setup times to systematically and precisely follow the demand;
stock of finished products can be kept low, thanks to limited batch sizes (for this
reason mixed-model lines are highly used in just in time contexts);
need for an appropriate sequencing of the models -> it is a fundamental problem
influent for the performance of the line (i.e. influent for PC losses in un-paced lines or for
the probability of no completion in continuous flow lines);
difficult components flow management: complexity is growing with the rise of
product models; operations and components required are changing and this builds
complexity in the component logistics; complex flow in managing the input of
components to each station of the line (i.e. complex logistics management in the factory to
feed the components to each station, which is due to the product model sequencing in the
line).
difficult management of product sequences due to parallel stations (due to variability
of operations which may change the sequence that was in input).
(Further explanations about multi-model and mixed-model lines are in the following
slides)
3
Multi-model The line can assemble more than one product type. Anyway
products are managed in batches since their setup are long. As a consequence,
along the line, it is impossible that two different products are assembled at the
same time.
4
primarily due to time needed to replace components as stocks of parts in the stations /
equipment in the stations, when the line is stopped to change as well as to provide new
instructions to operators:
time to change components low if product models in sequence are very similar ->
leading to similar components required by the station; high if product models are very
different > high opportunity cost due to the line stoppage;
sequencing is important for setup time reduction, as it may reduce the opportunity cost
due to the line stoppage (no big changes are required, the time is short with a product
changeover).
E.g. un-paced lines https://www.youtube.com/watch?v=ovMz2gTjvFA for example
different lenghts of wire harness solutions and electrical assemblies may change some
raw components, the equipment used to join the wires etc. adjusting to the length ->
we have to stop, not to interfere with line operators.
consider also the effects of transient subsequent to set-up, when you stop or not the line (in
order to empty the line with the old product model / to fill it with the new product model):
E.g. un-paced line from A to B; if done while the line is stopped (not in shadow time)
transient is 30 minutes x WIP to empty the line + 25 minutes x Number of stations
(when line is stopped); if stopped, 30 minutes x WIP to empty the line + 25 minutes x
Last station (when line is not stopped); it should be possible to change components
during the line run (changing progressively); we save time for the transient; any way,
there is a production lost due to bottleneck effect -> e.g. in an un-paced line, 25
minutes, thus quicker than 30 -> leads to WIP building -> approaching blocking -> risk
of stopping operations of a station due to blocking in the future -> this will be paid
later, as we have seen, in a balanced line, it corresponds to a lost capacity
E.g. again un-paced lines with common components / equipment, change in shadow
time progressively from the first station on while change over is being applied (from A
to B), e.g. the change of components can be made along the side of the line that is free
from installation (see the PC asm line)
Some other examples of changes may regard reconfiguration of the line as:
Rearranging number of stations (when it is possible as the installation can be
reconfiguration, e.g. of simple tables work benches);
Changing parameters, e.g. distance b/w pieces in a continuous flow line, timer of the
machine-paced line;
4
The balancing of multi-model lines can be seen as the balancing of single-
model line (monoproduct) since the line assembles a single product type at a
time (a batch of product type is produced). Except for the transient it is exactly
producing a single model.
STEP 1) (described above) The relation (above) has validity if at each station
there is only one operator.
The numerator is the total time required to produce the whole range of
products (since it is the sum -considering the needed quantities of each
product- of all the times required to assemble each product model).
The previously described relation K* = T/CT * is equivalent to this one,
since it still represents the ratio: required workload / available time, but in
5
this case it is calculated along the year (the previous one was per product). The K* is
just the minimum value from which to start.
Note that CT = H / Qj, so it is the requested CT, hence it comes out the equivalence in
case of single product type.
K* (the minimum number of stations) is calculated for all models because it is
important to keep a sufficiently high utilization of operators (which decreases when
this number increases).
5
(Other comments to the steps):
STEP 3) Since the line balancing is based on the cycle times TCj calculated in
the previous step (2), for each model j the balancing is performed using any
algorithm among those seen for mono-product lines.
app of the algorithm is done to minimize the number of stations/expected
cost of the line (considering also probability of no completion), given cycle
time as constraint (1st type ALB Assembly Line Balancing problem);
the number of stations may be different for each product model j, which
means the need of an adjustment, when reconfiguration of number of
stations is not possible / not desirable (i.e. not desirable -> effect on workers
-> if it is decided a lower number of stations, a surplus of workers for products
with low time requirements happens -> should be reallocated).
STEP 4) Adjust the line balancing if needed, through some corrective measures
on the results obtained in step 3 (e.g. keeping the same number of stations for all
models).
E.g.1 For a machine paced line -> lets say that the result is K1** = 10 e K2** =
15. We keep 15, so we redistribute operations of 1 amongst 15. We can now have
for sure a low UR, we could reduce the CT (by changing the timer) and make
operations quickly at line level (obtain the Qj in lower planned time); this avoids
some too low UR of operators, w/o increasing much the prob of no completion.
E.g.2 for an un-paced line -> reduced times assigned for each station with higher
6
number of stations than required -> low CT, high PC, low time to reach the Qj
6
STEP 5) The necessary work capacity has to be equal to the available work
capacity, considering all set-ups.
7
Mixed-model
The line can assemble more than one product type and there is no need to
manage products in batches thanks to the strong reduction in setup times. It
is possible to realize sequences of product types totally different from each
other.
Typically this solution is associated to continuous flow lines or un-paced lines (it
is not associable to paced lines because the cycle time should be the longest
assembly time sought among all the assembled products, then it would lead to
a loss in production capacity because such cycle time would be too long for the
majority of the products).
8
and suppliers) and the downstream systems (distribution system);
difficult components flow management. As the line can assemble more than
one product type, each station has to be fed with all possible components which
could be required for each specific product model (this complexity increases with
the increase of the product types). Solutions to this issue could be:
use of small station warehouses, periodically fed (this
solution is suitable for small components);
use of assembly kits (that are prepared upstream the
line), which contain components associated to specific
products: they go ahead along the line together with
the corresponding product which is being assembled.
use of overhead conveyors (in particular for large
components).
difficult parallel stations management: the presence of parallel stations could
modify the sequence of production because of the variability associated to
operation times (this complexity increases when the synchronization of the feeding
of the stations depends on the sequence imposed at the beginning of the line). A
solution to this problem could be preparing a buffer upstream parallel stations to
restore the initial sequence.
8
For production sequencing, the main objectives are two:
9
The bad sequencing leads to a no-completion. This is due to the choice to
assemble at first the product model (A) with shortest assembly time and after the
other one (B)
CASE 2:
The sequencing is better than CASE 1, even if at the fourth unit there is the danger
/ risk of no completion (remember that assembly times are aleatory).
CASE 3:
The sequencing is optimal. The operator finishes always his tasks with a safety
(time) buffer, then the probability of no completion is minimized. This solution
contemplates the alternation of models: in this way the operator has always the
possibility to recover the time spent to assemble the model which requires more
time (B) at the following cycle (assembling the model which requires less time).
9
The balancing problem in mixed-model lines has some specific issues which
make it more difficult than the balancing problem in multi-model lines. Anyway
the main objectives are the same (already seen in multi-model lines):
Minimizing the number of stations, given the cycle time (line balancing):
once defined the production mix (required models and their demands), the
objective is to minimize the number of stations, then to minimize setup costs,
labor costs, etc.
Minimizing the probability of no completion (station balancing): in this case
the objective is to minimize the no-completion phenomena in the stations.
Generally, particular objective functions are used for allowing to indirectly
minimize such phenomena.
10
the second unit B has already got in the station and he finishes to work on it in
correspondence with the downstream boundary of the station. Clearly, the third unit B
incurs the no-completion phenomenon.
The causes which have led to the no-completion are two:
the great difference between the assembly times of the two models;
the production sequence (see the previous example).
In the second case (Case 2) the production sequence is the same of Case 1. Even CT,
station length and the overall workload of the station are unchanged. However, in this
case, the assembly times are changed (better uniformity): the assembly time of the
model A is 1,5 min. and the assembly time of the model B is 2,5 min. This change allows
to avoid the no-completion phenomenon, thanks to the better station balancing.
10
To balance stations, it has to be minimized this objective function (above).
twk*w is the mean weighted time assigned to
station k, weights are given by the required
quantities of each model;
therefore tjk - twk*w is the variation between
the time required for the model j at the station k and the
mean weighted time which the station has to
process. This term is squared to avoid
compensations.
the denominator twk*w normalizes the result.
11
12
To balance the line, it has to be minimized this objective function (above).
Line balancing tasks have to be assigned to each station in a way that balances
the workload between stations.
TMk is the mean workload of the station k;
ST is the number of stations
TMk - (TMy / ST) is the variation between the workload of the station k
and an ideal workload of the stations (obtained if you uniformly
distribute the workload between stations).
13
Design of Assembly Systems Assembly
shop
q General features
q Examples
q2
Assembly systems general features
q Assembly line
3
Assembly shop general features
An assembly shop consists of a series of stations and each
station (generally, more than one station) is assigned a phase
of the assembly process of a product type.
4
Examples
5
Examples
6
Examples
7
Examples
q https://www.youtube.com/watch?v=zxCQzzDS
cHU
8
Assembly shop Strengths & Weaknesses
q Strengths
The stations (phases) are decoupled by buffers
The cycle time is not a constraint
The flexibility is high
Mix, product and expansion flexibility
q Weaknesses
Investment depends on the level of automation of the
system
It might be difficult to manage the flows of products
and components
The complexity of production planning and control can
cause bottlenecks and idle-times
9
1
3
Assembly systems are systems which join together components (coming from
manufacturing or subassembly systems) in order to obtain finished products.
Manual assembly systems are composed of several stations in which one or
more workers executes assembly activities; stations are linked through a
handling system (or transfer system).
Automated assembly systems are composed of several stations in which
assembly activities are automated. They can be rigid or flexible (robot). In
rigid systems each assembly machine can execute one or a few assembly
activities; in flexible systems robots can execute more assembly activities.
Assembly shop
the product has to be moved through different manual/automated assembly
workstations;
there is no rigid transport system, every product could have its specific flow
through the workstations (flexibility in routings due to flex material
handling);
the transfer could be manual, automated (using AGV, Automated Guided
Vehicles) or semi-automated;
to each workstation (generally, more than one station in parallel) is assigned a phase
of the assembly process of the product;
this system is useful and relies on the decoupling of stations to launch different
product types (some variations in the mix -> even with mixed-model
production)
In this scheme the assembly shop is located in the central area: it is an intermediate
solution between the other two solutions (fixed position assembly and assembly line).
Thanks to the routing flexibility (owing to the the absence of rigidity of the handling
system), the assembly shop system ensures a good degree of flexibility, enabling
especially mix and product flexibility.
Mix flexibility: (short period) it is possible to assemble different types of
product in the same time, meeting specific and periodically changing demand
requirements (high variety in the same time); (besides the flexibility in material
handling, during the move phase) this is also linked to the independence and the
decoupling of the workstations;
Product flexibility: to introduce a new product, operators just need to be trained
to acquire the new required skills, while the physical structure can remain the
same.
It is also ensured a good degree of variety and repetitiveness: each assembly phase
(which is assigned to each workstation) represents a good set of assembly tasks, which
ensures the low repetitiveness of the executed activities.
3
The presence of different product types having a wide range of different assembly cycles/processes or different
required volumes makes the assembly shop a suitable alternative to the line solution (one or more lines). In
particular, this requires a strong decoupling of the stations, so to enable managing this mix.
The assembly cycle/process is decomposed in a certain number of phases, each of such phases is assigned to a
certain number of parallel stations, as in figure. Generally each single station (to which a phase is assigned) is able
to perform the entire phase.
Stations belonging to consecutive phases are not connected by a rigid transfer system, but a flexible one. Then, there is
not only a single flow for all products, but flows are interwoven (intertwined).
Each station has an upstream (or input) buffer (where products are dropped, waiting to be assembled) and a
downstream (or output) buffer (where the assembled products are dropped, waiting to be handled to the subsequent
station); buffers decouple phases: if they are properly sized the system doesnt depend on the handling system.
The handling system provides to stations also the needed components (inserted in suitable kits).
The handling system can be traditional (carts guided by operators) or automated (AGV). In case of AGV, there are many
possible configurations, for instance:
taxi: the handling system transfers the product from one phase to the subsequent one:
the product being assembled is handled by the AGV and it is dropped in the upstream buffer of a station, then
the AGV moves away to perform another transfer. When the operator finishes to perform the assembly he drops
the product in the downstream buffer and he calls back the AGV (not necessarily the same AGV which dropped
the product at his station). Meanwhile, another AGV has dropped another product in the upstream buffer of the
station.
For this kind of system, the presence of a control and transfer optimization system is really important: such
system must minimize the number of transfers (w/o moving pieces, empty) and the handling distances.
The taxi system can be used when durations of assembly activities -at least- last a few minutes (typically
cycle time so the time assigned to the station should be at least 5-6 minutes).
working table: the AGV -instead of dropping products in buffers- are provided with adjustable working
tables which allow operators to perform assembly activities while moving.
This solution is recommended when the cycle time is short, since it is not convenient to drop a product at a
station if it will have to leave the station after a very short time.
Viceversa, if the time assigned to the station is high, this may require an oversizing of AGVs, so it would be
not convenient.
4
Two cases of a car manufacturer and semiconductor manufacturer
comments on mid variety / mix & volume
The handling system can be manual, automated (AGV) or semi-automated.
AGV (automated) control of the travelling mission computer already
knows where the vehicle has to go (and controls the execution of the move
phase); besides, a supervisor is present as a computer that assigns the
travelling missions to the AGVs (so scheduling the routing of workpieces
through the plant);
Hand cart (semi automated) the operators have to move pieces from
stations to stations and have to make the L/U; at each station there are lights
which help to understand the state of the stations; because of the closeness
between stations, it is not so important to optimize travels, and we can rely on
the operators flexibility to make the L/U of small sized materials, moved by an
hand cart (some wastes in move can be experienced); the operator would know
when to move when he sees the control lights at green state (to update the
operator in regard to the actual dynamics); it is probably using a card system
that can be directly placed in the hand cart (the card is saying what is the
operations and type of stations) -> so he knows where to move
The stations are equipped with automated machine.
(in the case of the car manufacturing) robot making the welding operations
etc.
5
AGV
Zone where the AGVs are charged their electric battery
Stations with I/O buffers, many paths to reach stations
6
Flexibility implemented in a specific solutions adopted -> conveyor with an
inner loop and many loops leading to the stations
The central conveyor tracks are used as inner loop for recirculating
pallets with the product under assembly; this serves as a buffer storage;
The outer conveyor tracks have gates to the inner loop + the outer
tracks provide the function of move towards the station where the
phases are assigned
There could be manual or automated stations
https://www.google.it/search?q=gripper&biw=1920&bih=971&source=lnms&
tbm=isch&sa=X&ved=0ahUKEwjfm5byu7DKAhXB1RoKHdmHDqgQ_AUI
BigB#tbm=isch&q=welding+robot+comau&imgrc=WtMizS8H WU0PVM%3
(ex. Of robots in this case not in an assembly shop)
7
8
Strengths
The stations (phases) are decoupled by buffers
The cycle time is not a constraint, operators are not forced to finish
systematically the assembly activities in the given cycle time.
The flexibility is high
Particularly, it is useful when productive mix varies over time in terms
of product types and (relative) quantities of product types.
As stations are not coupled, it is possible to plan different routings, so
assigning different phases to different (parallel) stations from time to time
Mix, product and expansion flexibility.
Mix flexibility means variation in models and volumes related to models
(also in automated system => welding tong of robot), Expansion
flexibility means expanding the already existing capacity.
Weaknesses
Investment depends on the level of automation of the system and generally
the space occupied is higher than the other assembly systems.
It might be difficult to manage the flows of products and components
(solution: assembly kits).
The complexity of production planning and control can cause bottlenecks and
idle-times (comes as a consequence of different product types in the mix and
9
mixed model production to answer to changing demands)
9
Design of Assembly Systems Automated
assembly line
q General features
q Examples
q System design
q2
Assembly systems general features
q Assembly shop
3
Assembly line general features
Each assembly line consists of a series of stations where
the product is progressively assembled.
4
Assembly line Example
5
Assembly line Example
Example of vibratory feeders (suiting different applications
and component shapes and sizes).
6
Assembly line Example
7
Assembly line Example
8
Assembly line Example
Bottling line
9
1
3
In the automated assembly lines operations are performed by machines rather
than by operators.
All different configuration may be found: paced & continuous flow line (both
lines and rotary tables) as well as unpaced lines; the transfer is moved onto a
support
Continuous flow line: support and device, making the task, are moving at the
same speed, along the time span of the operation; no relative speed (circular or
alternating move)
Pace line (circular or linear move -> support should be back)
Unpaced line: good for protection from PC reduction vs. cost of WIP, space, buffer size
3
System description
Every line can be modeled as a series of workstations linked by the rigid
transfer system.
The handling system is integrated and could be on the roof or on the floor
(). It visits all the stations in a serial and rigid way.
Every line (which is composed of more workstations) could be involved in the
assembly of a different type of product (in the example, the first line
assembles the product , the second assembles the product and the third
assembles the product ).
The entire assembly process is decomposed into elementary tasks, then these
tasks are attributed to the stations. The logic of attribution is dictated by certain
constraints and/or the optimization of certain objective functions.
Products
It is a solution chosen for products for which demand is high.
Unlike manual assembly lines, automated assembly system have the following
characteristics, which has an impact on the way they are designed:
determinism of time: since machines repeat the same operations with extreme
uniformity and since they require really constant conditions, also the execution times
of the operations are almost constant; therefore, this absence of uncertainty avoids
the no completion phenomenon (which, instead, is typical of manual assembly lines);
machines stoppages (due to blockages caused by sensors detecting problems): each
automatic machine is equipped with a sensor system, which detects any improper
execution of the assembly operation; a typical example is the force sensor: if, during
the execution of the assembly operation, the sensor detects a force greater than a
predetermined threshold (this is a symptom of a problem), the machine is blocked
and then it requires the operator intervention;
stoppages have different operational impacts, different from breakdowns since they are
more frequent and have short durations; moreover, unlike breakdowns, stoppages
are wanted/desired stops, since they address the presence of imperfections (i.e.
material imperfections / quality issue).
4
Components (refrigerators) -> thermocouple
high volume (part common to many finished product types);
high standardization;
few components to be assembled.
Feeding system for automated assembly lines -> vibratory bowl feeders
Vibratory bowl feeders are the most common device used to feed
components in Industrial Assembly Automation Applications:
Vibratory feeders are self-contained systems, comprising of a
specially tooled bowl feeder that orients the components, a
vibrating drive unit upon which the bowl feeder is mounted, and a
variable-amplitude control box.
Configuration of this case may be: Paced line or Un-paced line
5
Example of components: plastic bottle caps, nails, razor blade handle, plugs
for non reusable syringes
The bowl feeder includes an out feed accumulation track to convey parts
along and discharge into the assembly machine.
The track is designed to sort and orient the parts in consistent,
repeatable positions.
The drive unit, available in both electromagnetic and pneumatic drives,
vibrates the bowl feeder, forcing the parts to move up a circular, inclined
track.
The track length, width, and depth are carefully chosen to suit the application
and component shape and size. Special track coatings are applied according
to shape size and material of the parts which aids traction, damage to the
product and lower acoustic levels.
A variable speed control box is used for controlling the vibration of the bowl
feeder and stops the flow of parts to the out feed track via sensors.
Vibratory feeders offer a cost-effective alternative to manual labour, saving
manufacturers time and labour costs. Several factors must be considered
when selecting a parts feeder, including the industry, application, material
properties, and product volume.
6
Small Products with variants (some mix flexibility) -> valves
Multi-model line
Rotary table (always limited in space constraints)
Vibratory Feed bowlers are requiring space (small components) +
small assembly machines (instead of bigger machines in parts
production)
7
Product BOM built into the layout
Electric switches (some how more complex, so complexity is broken
down in many stations, making subassemblies )
Unpaced line (lenght of the handling system queuing in front of the
assembly machines)
The lenght is enough to manage small / frequent problems (blockages)
Layout occupation (due to the lenght) is motivated by reduction of PC
losses
8
Automated bottling line (a number of operations: filling bottles with the liquid
+ packaging etc.)
Continuous flow line
when they are filled with liquid, bottles need a lot of time; at the same
time the requested PC is high, so the CT is very low;
how to cope with this challenge? (focus on filling operatoon)
Parallelizzation -> while filling is working at the same time different
bottles are plugged in different working positions and moved-> the
station has a lenght / has a rotational speed / enough to enable the long
time for the filling operating cycle -> but every filling cycle top k
bottles exit -> top/k is what is needed (respecting the low CT)
Note that the bottles are fixed and so we can avoid problems of liquid
leakage, etc.
Integrated concepts of lines to the industry. This is a competitive alternative to
traditional lines that adds value to the complete packaging line.
Suitable for all liquids packaged in PET, our Combi systems combine
blow moulding, filling and capping in a single integrated system;
This eliminates conveying, empty bottle handling, accumulation and
storage, line layout with a smaller footprint, -> integrated system
(fabricating the PET)
9
Design of process plants
q General features
q Examples
q System design
q2
Process plants general features
A process plant is formed by a series of production
equipment used to make non reversible chemical-physical
transformation of materials through a fixed technological
routing.
3
Process plants general features
q The production flow is serial, analytical or synthetic.
RM FP
s.1 s.2 s.3 s.4
s.3
RM s.4.2 A2
RM = raw materials s.1 s.2
a, b, c = raw materials
FP = finished products
A1, A2, B1 = finished products Continuous transformation
(with different products) s.5 B1
s = stages of the production process
BF = buffers / storages
RM s.1 BF s.3
a a a
FP
s.4
RM s.1 BF s.2
b b b
Discontinuous
transformation
RM
c
4
Process plants general features
q Process plants are highly automated:
relevance of technological parameters of the production
process (temperatures, pressures, );
significant investment in sensors, equipment control, etc.;
control often automated with supervisory intervention.
Signal processing
Raw materials data Process variables Equipment data/events Product Quality Environmental conditions
PRODUCTION PROCESS
Materials Products
Energy
5
Example 1
Oxygen 200 kg/hr
CO2 800 kg/hr
Ethylene
4.000 kg/hr
Oxygen
20.000 kg/hr
VAPOR-
Ethylene
REACTOR LIQUID
40.000 kg/hr SEPARATOR
Ethylene Oxide
DISTILLATION 58.673 kg/hr
Reactions:
C2H4 + O2 C2H4O (desired) DISTILLATION
C2H4 + 3 O2 2 CO2 + 2 H2O (undesired)
C2H4O + 2 O2 2 CO2 + 2 H2O (undesired)
.
H 2O
327 kg/hr
6
Example 2
C
A
>
s.1 s.2
B s.6
>
> > s.3 s.4 s.5
> FP
Flow sheet for the production process of Nylon 6,6 (continuous process)
7
Example 3
BF
A s.1
s.3 s.4 s.5 > s.6
> s.2
BF
B s.1
s.3 s.4 s.5 > s.6
> s.2
BF
s.1
s.3 s.4 s.5 > s.6
s.2
FP
A, B = silos of raw materials
FP = silos of finished products
BF = buffer / storage
s = stages of the production process
Flow sheet for the production process of Nylon 6,6 (batch process)
8
Process plants general features
9
Process plants general features
q Low flexibility
q High investment needed
q High risk of obsolescence
q Significant impact of failures
q Importance of variations in process conditions
10
Rough design of a process plant (continuous flow)
1. Define the production flows according to the
technological routing required for the product
2. Identify all the production equipment types that
are needed and the bottleneck
3. Define the theoretical production capacity
TPC [ton/hour]
where
TPC = theoretical production capacity in [ton/hour],
or other similar units, i.e. [kg/hours]
11
Rough design of a process plant (continuous flow)
4. Calculate the actual production capacity
where
A = line availability (0 < A 1)
SR = scrap rate (0 SR < 1) (varying according
to the process conditions kept in the system)
q Assumptions
Production in batches (batch A, batch B, batch C and
so on);
Setup times are required and do not depend on the
production (batch) sequence
13
Rough design of a process plant (batch)
1. Identify the production mix
2. Define the production flows according to the
technological routing required for the products (in
the production mix)
3. Identify all the production equipment types that
are needed
4. Calculate yearly workload and number of hours
available for each type of production equipment i
5. Calculate the number of production equipment of
type i necessary to produce the production mix
Rough design of a process plant (batch)
N Qj 1
NH i = j =1 + STTij NC j
TPC (1 SR ) A
ij ij i
where
15
Rough design of a process plant (batch)
16
Rough design of a process plant (batch)
AHi(s) = WHi(s) SE
where
WHi(s) =yearly working time available (depending on the
number of shifts per day)
SE = scheduling efficiency (0 < SE 1)
17
Rough design of a process plant (batch)
NH i
NM i ( s ) =
AH i ( s)
18
Rough design of a process plant (cont. flow/batch)
19
1
3
Products, processes and product routings
Petrochemical /
reversible transformation. These sectors are:
3
impurities / wastes which are reused as much as possible or removed to reduce
environmental impact (e.g. air pollution ).
3
Different characteristics of production flow are
observable in process plants.
4
order to obtain homogeneous characteristics for next phases
> e.g. after the grinding process takes place in a raw mill to
reduce the particle size of the raw material, the output of the
grinding process called raw meal is transferred to a
homogenization silo before the clinker manufacturing
process; before the grinding process, there is a pre-blending
storage where materials loaded at different times in the
storage are blended before entering the grinding process) >
(discontinuous transformation > batch process as in
cement, nylon production processes).
4
(To achieve the transformations) keeping the optimal / standard conditions
of certain technological parameters is a critical issue of these types of
plants; in other words, variations in process conditions are critical and
should be mitigated in order to maintain as stable as possible quality and
efficiency of the production, as well as safe production.
Variations in many variables (materials, process itself, environment) are
naturally occurring, leading to challenges in the process control problem,
because these are occurring also in real time. In this regard, it is worth
mentioning that there are control architectures developed in process plants
normally having different layers in a hierarchy from sensors / actuators
(lower level) to digital control on field (ex. pump, valve control, equipment
control in general) to supervisory control both automated or semi-
automated at upper level of the hierarchy.
An example of supervisory control is tangible in control rooms, see the case of
the refinery
http://pascagoula.chevron.com/home/abouttherefinery/whatwedo/processingan
drefining.aspx.
Comment shortly the scheme, as logical scheme which can then be
implemented in the control architecture.
5
Example 1
(Short intro) Ethylene oxide is a product used industrially for making many
consumer products as well as non-consumer chemicals and intermediates; it is
important or critical to the production of detergents, thickeners, solvents,
plastics, and various organic chemicals. ; it is industrially produced by
direct oxidation of ethylene (in the presence of a catalyst).
The process is schemed out as a Block Flow Diagram. A BFD is a simplified
representation of the main phases of a production process through the
equipment used to carry out the correspondent operations (only critical
equipment for the process are drawn in a BFD, not other components which
will be included in more detailed schemes, such as e.g. storage tanks, valves,
heat exchangers, all functional components for the operations); subsequent
the flows are represented by directed arrows, normally having the flows from
left to right, expect for the recirculating flows >>> on the whole, this results in
drawing a fixed technological routing through the sequence of visited
equipment realizing the different phases of the production process.
The BFD is an initial representation based on the principle of material flow
conservation (mass balances), and it is depicting the production flows that
the production system should be capable to produce, both primary flows,
any intermediate flows and scraps / re-entrants or reworks (said in general
terms). This is, at least, the overall macro-perspective of what is actually
occurring within each operation along the production process.
Indeed, from a micro-perspective of the plant/process, some chemical reaction
correspondingly happens within the reactor, both desired/main and undesired/side
reactions; this leads to the production flows within the BFD.
More precisely, note that the relationship 2:1 b/w ethylene and oxygen flows (40000:
20000 kg/hr) is correspondent to what formulated in the main chemical reaction: if this
reaction just happens (one molecule of ethylene and a half oxygen are needed in order
to obtain ethylene oxide, C2H40, which is the desired outcome), we would have just
ethylene and oxygen flows (40000: 20000 kg/hr) as inputs and outputs. But undesired
reactions occur, resulting in CO2 and H2O; all in all, we have an expected outcome at
the end of the reactions, considering both undesired reactions and also the fact that
the conversion of the desired reaction is not 100%, i.e. not all the ethylene and
oxygen are reacting.
For this process characteristics, some further operations are then required within the
technological routing: after the vapor-liquid separation and the subsequent distillation
in two sub-phases (subsequent equipment, distillation tower), we obtain the final
product (C2H4O) + the elimination of all side products (CO2, O2, H2O) + the
recirculation of raw material ethylene (previously not reacted) which is sent back to be
reprocessed.
1st concluding remark: all the info shown in the BFD are resulting from knowing more
detailed models in regard to different chemical reactions happening within the reactor,
plus other transformations later on required; e.g. separator is just resulting in a quantity
of more volatile materials released to the outside environment (oxygen and CO2 carbon
dioxide) etc. (alias it is like saying: from knowledge/models in chemical
engineering to knowledge/models in industrial engineering domain).
2nd concluding remark: on the whole, we can understand that the model at higher
abstract modeling level (BFD showing production flows) is constrained by what is
happening in terms of physical / chemical transformations along the production
process > we can synthesize that we are obtaining an initial analysis in terms of
production flows which will then enable to define T/APC (theoretical/actual
production capacity) considering that some material is also leaving the process
(Oxygen, CO2, H2O) > 58673 kg/hr (> necessarily, some unavoidable material weight
loss, due to the inherent process characteristics, may happen); any other variation in the
process conditions may of course lead to further material loss (e.g. reactions should
happen at given conditions such as temperature etc, if standard conditions are kept,
otherwise ) .
6
Example 2
1st concluding remark: this is a clear example where the process plant is designed to
operate as a continuous flow production process: all phases of the production process
are regulated in strictly dependency and control of standard conditions should be
kept at a production regime;
2nd concluding remark: in this concern, it is worth pointing out that a warm-up period
is required to achieve the production regime; the process conditions during the warm-
up are not the optimal / standard one, therefore scrap is over-generated. Also for this
reason, the type of plant is planned to produce on a continuous production cycle (24
H) for a production campaign of many months (e.g. 3 months) (to avoid high scrap
rate every time the campaign is started).
3rd concluding remark: if, in the plant, standard flow rate Q cannot be achieved, due to
variations in target production volume / rhythms, there is need that the phases of the
7
production process are regulated under different conditions, leading to different
performances (with respect to standard conditions); the same considerations should be
applicable if we change the product type; in general, high sophistications are required
with new regulations / set-ups, leading to some costs (more scrap rate) even at regime,
if we are not capable to achieve optimal / standard conditions > then regulate but at a
cost which could not be neglected > low flexibility > that is: the range of production
volumes / products is very limited, and the changes would be any how at relevant
costs.
7
Example 3
1st concluding remark: batch process is another solution when our strategy is based on
flexibility (more product variety, market driven, ) and we can accept higher costs.
Of course flexibility is any how limited by many constraints (see for ex. some fixed
routing b/w s.3, s.4, s.5.
8
For what concern logistics within the plant (production logistics), this is a
natural consequence of the fixed technological routing, resulting in the plant
structure.
pipelines (including pipes which means also a set of instrumentations along
the pipes, required to control the transport of the materials), operated/used
both in continuous and discontinuous flow transformation; of course, not all is
made by means of pipelines, also some transport system are used, more based
on serial transporters, as trucks, dumpers, trains etc. (see the case of a copper
mine and transportation of extracted materials > the extraction of raw material
can be both underground i.e. sinking a vertical shaft into the Earth to an
appropriate depth and driving horizontal tunnels into the ore, or open pit i.e.
90% of ore is mined by this method; ores near the surface can be quarried after
removal of the surface layers);
in general, the production logistics can be considered simple, first of all
because it is majorly designed/built into the plant structure, with few
management decisions still left on the routing side; furthermore, products are
few product variants, and may require few materials within the product
recipe (taken from some silos concentrating those materials); overall,
movements can be done in relevant batches (as dumpers) or by continuous
transport activities b/w stages or stages and storages (along pipelines, so
back to routing built in the plant design). All in all, not as many discrete
components moved, as loaded units to be moved through the factory, as
happens in manufacturing.
9
First of all, for what concern production management of the plant, this is limited to
some decisions.
Management decisions are limited to two main problems: batch sizing + batch
sequencing within a production campaign. The decision is the number of production
campaigns / number of batches (hence, batch sizing) to be produced for each single
product (i.e. short digression to better understand this management decision > this is
related to the typical set-up/cost of inventory holding costs optimization > when the
number of production campaigns / batches is decided (e.g. by using Magee Boodmans
model), the batch size is also defined, being equal to Dj / No (i.e. demand of product j /
number of production campaigns where, for each campaign, only a single product is
produced)); batch sequencing is needed when different products are realized by the
plant (also relevant to optimize if a sequence dependency exist).
Regarding quality issues, as discussed this depends on the stability of process conditions:
Keep stable conditions during the production phase;
Limit the transients as warm-up which are usually associated with scraps.
9
Low flexibility is clearly a characteristic of this kind of plants concerning
different flexibility dimensions and for many reasons. Even in the most flexible
solution (batch process production) we have:
limited set of products, batch wise production (due to batch sizing and
sequencing, first one product type for a campaign, then another product type,
etc.) > limited mix flexibility > it is any how still today challenging to be
market-driven;
a plant is not easily (re)configured / (re)converted because of the fixed plant
structures (i.e. piping ) + the high cost of the plant shutdown (remember
the need for continuous production to reach economies of scale) > limited
expansion flexibility > this is done any how; a plant is revamped in some
parts for its aging or, some times, it is subject to reconvertion in order to
keep in pace with new requirements and technological developments that
enables the achievement of those requirements (e.g., environmental objectives
on impacts + improvement of some performances due to technological
upgrades with new technological issues >>> i) cite the BAT, even if BAT
should be normally considered at the begin of life, while upgrades are along a
very long life of the equipment > often more decades; ii) again the case
http://pascagoula.chevron.com/home/abouttherefinery/whatwedo/processingan
drefining.aspx, it mentions some upgrades;
due the high required investment, this type of plant is usually built to work on
more production shifts / high number of production shifts per week
(24H/7D)> hence, a small demand variation cannot be absorbed at all, if not
10
changing some speed parameters (@ risk because we may cause scraps and any how
not optimal process conditions); remaining shifts or overtime are normally few or not
any at all > limited volume flexibility;
Other characterisics:
High investment needed > as said, in order to achieve higher economies of scale, a
trend was observed towards increasing plant sizes and capacities > inducing very, very
high investment costs; on the other hand, there are some new trends which are actually
observable on small-sizes of this kind of plants, which means of course orienteering to
other objectives, relating to product diversification and more market driven approach.
High risk of obsolescence > there is a clear high risk of obsolescence with the limited
set of product types (or one type) > facility lifetime is strictly related to products
lifetime; this is a general concern, which exist in these plants, and it is evan amplified
(respect to manufacturing);
Significant impact of failures > primarily, there is a series logic in this type of plants,
with some degree of parallelization / reduncancy, but not much + presence of relevant
buffers (with high materials / WIP stocked therein) just in the case of batch processes,
which are of course useful in order to protect of failures (while occupying high
spaces/high WIP, but this can be considered a minor concern respect to the problem of
failures > also because the value of WIP could be not high, being the product/intermediate
a commodity); in the case of continuous flow production of course an unexpected
shutdown is really really critical (also for safe issues);
Importance of variations in process conditions already discussed and a clear intrinsic
challenge of process plants.
10
Steps of system design - case of one product
11
Steps of system design - case of one product
Step 4 remember the fact that the SR may be dependent on the process
constraints -> if we do not operate at standard process conditions (i.e. either
warm-up or different conditions from process conditions at the nominal /
standard regime), the SR may be a different value, respect to the best value;
Step 5 few options may be available, for ex. duplicating some equipment but
this should be clearly verified with knowledgeable engineers on the process.
12
Modeling assumptions of system design - case of batch processes
The equipment are used according to the batch wise processing approach:
batch A of a product, then batch B of another product, etc. >>> to change from
one batch to another, it is required a setup for the new campaign (with the new
product).
Set-up times are considered sequence independent: this could be either
because in reality there is no sequence dependence (negligible), or it is the
result of an optimization procedure that enabled to identify best batch
sequencing (i.e. first batch A, then batch B, etc.) to reduce set-up times > in
this last case, set-up times are given by assuming the scenario of optimal
sequence.
13
Steps of system design - case of batch production process
14
Coefficients (measuring time losses, cfr. Turco)
Observe that only relevant, typical coefficients should be considered wisely because
we have to apply to the problem of the case > e.g. the HC is not considered > there is
no clear machine interference problem as the manufacturing case; on the other hand, the
intervention on the production equipment is clearly possible (even if not desirable) for a
shutdown > maintenance.
SCRAP RATE SR: percentage of materials out of tolerance (not achieved the target
quality)
AVAILABILITY: percentage of up time (intervals), when the machine is required
for production and actually available to work (w/o trials), with respect to the total time
(up time + down time); down time intervals regard the overall failure and maintenance
downtime.
Observe also that here we are using the TPCij (not Tij as in the job formula) > this is the
production capacity for a given equipment i, supporting the chemical-physical
transformation of product j at a given step in its routing > here we are not talking
about a single workpiece manufactured as in the case of the job shop
Applying the formula, the yearly workload / required capacity NH takes into
account the need to load the machines (on yearly basis). Thanks to the coefficients,
NH also considers the presence of time losses within the machine calendar time >
thus leading to a gross / effective workload required from the machine.
15
Few quick notes:
STT is long hours or day (using hours is a better, practical measure);
one campaign is normally coincident with one product batch;
number of campaigns NC may be also different for products from equipment
to equipment, one different case can be then also expressed as NCij.
16
(see back the difficulty of production management) > scheduling complexity is
not high + it would be mandatory to keep high utilization > scheduling system
should perfectly utilize all plants > so SE can be considered high; it is not high
because of some complexity induced by the different product routing and some
constraints along the process.
17
18
(financia perspective) CAPEX versus OPEX > CAPEXs are splitted during the
facility / machine life time > splitted cost can be then comparable (so, summed
up) with costs representing OPEX. The mechanism to split is based on
financial issues not detailed in this course.
Comments on the typical costs.
19
INDUSTRIAL TECHNOLOGIES
IT 03/10/2016
Production (process) = set of activities (process) required to produce goods or services delivered
to the market by a company all the activities that deal with the transformational input-output.
A process is a list/group of activities.
Production system = a subsystem of the company. It uses resources as inputs raw materials,
semi-finished goods, energy, information, knowledge, etc. to provide/produce products and
services in order to satisfy the customer needs and the objectives established by the companys
strategy. System is a concept, generic, we are not yet referring to a physical plant/factory but we
are still dealing with the idea, the design, the concept of the factory (we deal with the system and
not the factory).
Production plant = physical plant where the production system is established (i.e., the physical
building). There are further steps for designing and installing the plant that we are not dealing
with in this course.
Our work in this course involves how to put together all the elements (e.g., operators, ) to create
a production process.
There are different kinds of activities:
- Activities concerning the need to get the material, information and resources.
- Activities dealing with the physical transformation of the raw material.
- Activities concerning the distribution of the production results.
Logistics means finding the raw materials to do production and distributing the final products
(outbound logistics involve procurement and distribution). It also involves how we handle the
material. There are inbound logistic (production), which deal with what happens inside the plant.
We are not dealing with the full lifecycle of the industrial company. Production is one of the main
activities among the main activities, relevant for bringing the product to the customer.
An industrial company is the result of two main processes: operations and NPD, and these two
processes meet in production. In the intersection of these two, there are several design activities
happening. Therefore, we need to understand how implementing the production process is
normally done.
Consulting companies follow these steps to launch design activities of production processes:
- Production process ASME Diagram. There is an international standard created for this by
the American Society of Manufacturing Engineers. They said that production is a process,
made most of the time by the same kind of activities (which can be described in a standard
way). Production is where the transformation of physical stuff is happening (e.g., a
machine molding, a milling machine working on a metal sheet, an operator doing
1
INDUSTRIAL TECHNOLOGIES
something by hand). Waiting activities: WIP. Checking that what was happening was
done well: control activities. Then there is the flow of materials through the system. With
these 4 elements we can describe everything, this kind of language is known everywhere
as ASME. In this example we are dealing with the cements production process, which is
dealing with raw materials coming from 2 sources, 2 flows that join later on. In the end, the
final result is divided in 2 main flows. The circles mark several types of sub-processes. We
have, thanks to this diagram, an idea of how this happens.
Next to it we have a typical paper used by production engineers for checking the
production processes inside the plant.
We see in the following image (a) a
combination of two raw materials or the
assembly of a physical product. The
divergent process in image (b) marks
instead something starting from a natural
product that may have a bi-side result
(e.g., oil to create diesel/gasoline, cheese,
or a physical artifact that is mounted and
then has two different outcomes that can
are separated). The (c) may show two
different products having a common
platform.
Assembly line (picture on the right) = list of sub-processes used to
make sub-assemblies eventually grouped together to make an
assembly. We can have on one hand some assemblies and on the
other hand other processes.
2
INDUSTRIAL TECHNOLOGIES
One of the main steps for an industrial engineer is trying to understand a companys
process (it will be in the exam for sure).
- Bill of materials. We are dealing with physical artifacts, and we are dealing with the design
of production systems which we use to produce these products. Organizing a list of
components means dealing a bill of materials. Physical procedures for describing how a
component is done. It is normally a list of components (no pictures) in a proper list.
All of the exercises will start with this.
- Cycle times (Production time / Processing time) / Production programs.
Cycle time: In the following graphics we see a CT sheet where we can see the timings
forecasted for doing a certain activity in a certain machine (attached normally to a machine
or to the operators location). How many sec/min/h we need for producing a certain kind
of product. Our job is not to design the CT, but to understand how many
machines/lines/operators we need to buy given the CT. The CT is the typical result of the
work of production or process engineers (experts of the specific production technology
who are aware of the time that is needed for performing an activity). We are just designing
the system according to the technology used, to produce a certain amount of products.
Work order / Production program / Order of production: minimum information given to a
production machine to produce. Which activities we need to do and for how long. In our
exercises these information will be given, we will start from these tables for designing the
3
INDUSTRIAL TECHNOLOGIES
4
INDUSTRIAL TECHNOLOGIES
We will not get into detail about technologies, but we will deal with many different
technologies. In the central picture on the slides there is a cooking tower for the cement
industry. We will see many different types of production with different types of
technologies. We are just stressing the fact that we will deal with different technologies.
- Classifications. We need the element to classify the system. All companies, at a certain
point, work in ways that are very similar to each other. We should keep in mind these
commonalities. There are 2 classifications: Wortmann classification and Free Axis
classification.
o Wortmann classification: codified in the 70s by prof. Wortmann (from the
Netherlands). Each company can be described according to whats happening inside
putting it into a kind of chain that goes from supplier to customer. A company can
be described into its minimum activities: engineering (design of the product),
purchasing (buying), manufacturing (production, making the transformation),
assembly (putting elements together), distribution. The customer is buying the
product from the company, and this happens through an order. The Wortmann
classification classifies companies and their behavior according to how the
company is reacting to the customers orders. Here there is a difference between
the different behaviors of the company.
ETO and PTO indicate a company that engineers or purchases the product once
they got the order. Fincantieri is a typical ETO company, designing a cruise-ship
according to the order of the new cruise-ship order.
Italy is the first producer in the world for production CNC machines. These
machines are not produced to be put in stock (they cost about half a million, so it
would be a really expensive stock). But also waiting for an order to purchase the
material and make the machine from scratch. It would take way too long (the
normal timings in Italy, Germany, is 3 months the time-to-market, which was 1
year until some time ago). You can find a solution in the middle, like ATO (having
pieces in stock and assembling the machine to order), or MTO where some pieces
are produced from raw materials once the order is received.
o 3 axes (Abbel) classification.
Demand axis = how the company responds to the demand.
5
INDUSTRIAL TECHNOLOGIES
Process axis = Italy is the second country in the world for manufacturing GDP. If we
look around, we can easily recognize two types of production: process production /
process industries, industries based in big plants, many pipes/towers, consumption
of energy, normally producing semi-finished goods and raw materials. Their name
comes from the fact that they perform chemical processes and physical
transformations, and they are producing primary goods. They are big plants, fully
automatized, requiring a lot of investments.
There are other kinds of industries, called manufacturing industries, the industrial
production apart from process industries. These can be normally described into two
main types: machines that are making physical transformations in the artifacts (e.g.,
drilling, C machines), and they are called parts production. Then there are others
that are making assembly activities. We have to divide these two: first of all,
because their design is quite different, and second of all, because we have
companies that have focused on parts production and others as assembly
providers. F.ex. in the automotive industry, the oldest industry in the world, most of
the car makers (except from some niche situations), the car company itself is just
assembling the cars through components made by suppliers, which are producing
the same component for different car makers. The reality can be much more
complex, where in the same company there are both, mixed. e
Volume axis = how much at the same time. A company which needs to produce a
lot of products may better have a plant in continuous production. Most of the
companies, most of the time work on batch production.
o Taxonomy of the classification. In this course we will go through all of these boxes
to understand how to design a
production system starting from a
job shop and getting to the
process plants.
6
INDUSTRIAL TECHNOLOGIES
7
INDUSTRIAL TECHNOLOGIES
main elements are normally in trade-off and are the main elements we need to consider.
Matrice volume/variet per I diversi tipi di processo The kind of production process we choose
depends on the volumes and variety we have in production. When designing we will always start
from an analysis like this.
Basic elements
- Time will always be our main design variable.
- Cost
- Performance
Main operational performances:
Lead time (throughput time, flowing time): duration of time from start of production activity to
finish [hours, minutes, seconds] (def.) time necessary to perform all the activities inside the
factory from the customer order arrival to the moment in which the material is ready to be
delivered.
In our course we are interested in the
production lead time (while the total LT is
often referred to as lead time for order),
the time that is needed to prepare the
machine, loading and unloading,
performing the first quality check,
throughput time
We will work on production systems, and our decision will have an impact on production lead time
(which has a direct impact on when we will be able to deliver). Non chiede i nomi delle parti del LT.
Setup time: the time it takes to set-up a production resource for processing a new work order
[hours, minutes, seconds]
WIP (Work In Progress): the amount of inventory that is being processed or is waiting on the
shop floor [quantity of work orders, but also time needed to finish the quantity of work orders]
-> we can always translate quantity in terms of time (our design variable)
Production rate (throughput rate, rhythm, production capacity): number of products that a
production system is able to produce in a time unit [pieces / h, products / day, etc.]
Cycle time is the time period elapsing b/w the exit time of the precedent work-piece and exit
time of the successive work-piece from a system, it is the inverse of production rate [time / piece]
Other operational performances (non-linear, that companies use more often):
Utilization (saturation): ratio of effective production hours vs. available hours (production hours
could include setup, available hours could include maintenance); how saturated resources are.
Flexibility: the ability to easily handle variations in demand mix (different types of products
produced
Availability: It measures the impact of breakdowns and stoppages on the time when a machine
is theoretically available to be used. We just need to know that a certain type of layout or system
may be more (or less) available than another type of system.
Dynamics of a production system
Bottleneck = slowest machine in the system.
What is the minimum amount of time that the piece will require to be produced? The sum of all
the production times needed to produce = LTmin.
Although some machines will be waiting for material coming from the bottleneck.
We can buy another machine of the same kind of the bottleneck, not to reduce LT (it is always
16min), but to increase the throughput rate.
8
INDUSTRIAL TECHNOLOGIES
Different types of production systems have different results in terms of performance (by
definition).
An important issue is that there is a direct correlation between LT and WIP part of the dynamism
of production processes. The lower the WIP, the lower the LT. Having a lower LT means most of
the time to have less uncertainty, an easier control of production and be able to plan better in
advance and be able to deliver on time. We have to be able to deliver the production to when it
has been planned; it is a matter of market performance. There are some variables that are only a
task of production system design, but there are other issues we are concerned with LT and WIP.
In production companies there is often a lot of WIP that is not properly managed, causing
problems in LT and consequently in delivery time (cites Littles law, we only need to know the first
two columns are always correlated).
What we have introduced until now are normally called performances (performances = the result
(measured against a multi-dimensional, numerical scale) of the logistic and productive process),
but according to the situation we are in, there are some elements that are up to us (relevant
performances for us), while others will not be up to us (operative conditions production design).
It may happen that the company owner (often happens in family-run companies) doesnt want to
buy new machines, and you can only use old machines, it is not up to you to solve that issue. This
is an operative condition that you, as a manager, cannot change. Instead, it is up to you as a
production manager to work on quality (produce something properly done), or to work on the
delivery time of your order with the machines you have. There are some situations that are
fixed/blocked, e.g., the budget. There are constraints that will be given to our design that should
be considered.
Costs of Production Systems
The main costs to be taken into account when evaluating a production system are the following:
Costs of installation (most of the time in CAPEX capital asset expenditure) *
Operating costs (in OPEX)
Relevant costs to be considered in design decisions for a production system could be:
Real costs (installation and operating costs); there is real money behind.
Opportunistic / figurative costs (e.g. depreciation, inefficiency costs, etc.): sometimes we have
to take decisions based on these costs, which are not moving real money but represent an
economic value of a decision.
Installation costs can be defined as all the expenses that the company has to invest in a plant to
enable production. They must be estimated in the design phase for evaluating the affordability
9
INDUSTRIAL TECHNOLOGIES
and return of the related investment. To start an industrial activity, it is necessary to have
adequate financial resources (i.e. the capital) to be invested in all production factors that are
necessary for creating the production capacity
Installation costs in CAPEX include several cost elements:
Feasibility study (preliminary economic analysis of the project)
Development of the project
Acquisition of the ground;
Building & infrastructure construction
Installation of plant services
Acquisition and installation of machinery and equipment and human resource hiring
Intangible cost of knowledge assets (know-how, patent acquisition, payment of royalties)
Interest payable on any mortgages or loans for investment
These costs are initial costs that happen at the same time moment, some of them are capital
assets and others are operative expenditures.
Installation costs in OPEX, they refer to the set of non-durable production assets and financial
payable to start production, like inventories of finished products and raw materials, account
receivable (deferred payment, typically 30-60 days), any cash for running start-up.
Operating Costs are all costs to be sustained in a given period of time (typically one year), for the
operation of the plant:
Variable operating costs: which include all costs of operation that depend on the volume of
production, like raw materials, components, energy (typical variable operating cost), commissions
due to the sellers, transportation, etc.
Fixed operating costs are independent of the production volume, and shall include all those
expenses that remain the same independently from the production level, like: overhead
(insurance, communication, building conditioning, etc), expenditure related to technical and
administrative staff, rents.
Semi-variable operating costs are related not only to the volume of production, but also have a
fixed part, independent of the volume of production
These costs are real money that is moving.
Sometimes opportunistic / figurative costs are though much more important than real money.
These include inefficiency costs (opportunity cost, losses, ). Inefficiency costs do not correspond
to a real flow of money, but rather represent a loss of opportunity (e.g. in terms of reduction in
production volume and thus in income loss); defined as the loss of income resulting from
inefficiency (of a machine or plant) with respect to a predefined standard.
Some of these costs include:
- Overtime work costs.
- Subcontractors costs.
- Stockout cost: the customer is asking me a product but I am not able to provide it at the
time it is needed: I am losign the opportunity to sell this product, and it is a typical cost of
inefficiency (I wasnt able to satisfy my customer) that is normally measured in terms of
margin that was lost, or penalties written in the contract.
- Stockholding cost: the fact that I am keeping a great amount of material in stock ->
blocking my money. I am losing the opportunity to use my money better. This implies the
dilemma: is it better to design a system that is very fast of production (limiting flexibility),
or to have a more flexible system that involves stocks?
- Setup cost.
10
INDUSTRIAL TECHNOLOGIES
When we are making decisions, we should decide what costs are relevant. For the design of a
production system a cost should be:
future (past is already sunk)
avoidable (if we do not follow a certain design / plan we do not sustain that cost)
differential (with respect to other alternative designs / plans)
We are dealing with systems production systems that will have a long life.
In this type of situation, it is important to consider another economic effect depreciation.
Depreciation cost is more up to accountants, but we need to take it into account.
Normally, when we have to evaluate whether it is convenient to invest in a new plant/machine:
- Cost analysis
- Revenue / savings analysis (non lha nominata)
- Cash Flow analysis
- Discounted Cash Flow Analysis (NPV)
- IRR (not so used) / Payback time
Illy case https://youtu.be/r0czaqVP_P0 https://youtu.be/PoFysyHHMUQ
https://youtu.be/cvkPDc1oXy4
Notes: Coffee beans come in from the harbor (4 miles away). Not yet roasted. They only use
Arabica.
A machine separates the beans from 72 slaloms for the green beans through a network of
tubes. To keep the whole operation running smoothly it takes an operator and a sophisticated
software.
They have a university of coffee to spread the knowledge about high quality coffee.
There are different tasting levels to ensure the quality of the batches.
- 2nd phase: Roasting process -> 15000 compounds make the taste. In this process the beans
take their taste.
- 3rd phase: Cooling process. A compressed air system breaks the beans, then air brings them
to the maturation process for 20 days.
- 4th phase: for 3-4 weeks the coffee is put in pressurized containers to augment its aromatic
intensity. This allows oils to expand.
- 5th phase: packaging machine.
- Containers are produced internally to preserve the quality. Machines work on continuous
cycles of 3 shifts. Bending and pressing.
- Sensory lab to map defects.
- Production of coffee machines and capsules as well.
The main content is a process industry, working with big plants. We also have seen other
departments that are more similar to manufacturing.
Always starting from the point that companies exist for producing products to be sold. This is a
fact: companies exist to satisfy demand, and all the design activities will always start from an
understanding of the demand. In reality, companies produce several products.
There is a relevant part about the treatment that should be done on the coffee. They are putting a
lot of time and analysis to analyze the raw material. At the end of the day, the steps i.e., drying,
cooking, cooling all have something in common: they are highly automated (big machines),
while operators are controlling, verifying, monitoring through flow sheets, supervising. This is
quite typical in all process industries, no matter the sector they belong to (e.g., chemical
11
INDUSTRIAL TECHNOLOGIES
12
INDUSTRIAL TECHNOLOGIES
13
INDUSTRIAL TECHNOLOGIES
14
INDUSTRIAL TECHNOLOGIES
availability can be evaluated by analyzing the configuration of the system based on the availability
of each single machine. A plant, a production facility, is a system which according to how it has
been designed could have more or less impact if there are more or less failures inside. We can
design a production system that is more available than another.
What is the impact of unavailability on the production capacity?
Our main interest is to have an estimation of this impact, and to do so we need to evaluate the
configuration of the system to understand how the production flow is affected.
The impact is different if the blockage happens in a machine in a station where several machines
work in parallel, or if it happens in one which has no parallelized machine.
If the failure happens in M1, the availability of the whole system depends on its availability.
We have the same assumptions as in machine interference:
the state of the system depends on the state reached by its machines;
o the state of the system as function of the state of its machines;
each machine is working / degrading independently of the others;
o independent events/states for each machine (a machine is failing independently of
the others);
We are interested in calculating the system availability to get a rough estimation of its impact on
production capacity.
15
INDUSTRIAL TECHNOLOGIES
If I have three machines working in parallel, and one of them is not working, I will have a reduction
in PC. If I have two not working, the reduction of PC that I am expecting will be higher.
1. Model the state space table of the system (si,j)
2. Model the correspondent table of probabilities (pi,j)
3. Evaluate the probability of each state of the system: i=1m j=1 pi,j
o We are interested in PC instead of time here; evaluate PC = Yi
o i=1m j=1n E[Yi] x pi,j = 1
4. Evaluate the expected production capacity due to all states: E[Y] = i=1m (Yi x pi)
5. Calculate the expected system availability due to all states: E[Y] / TPC
o TPC = capacity that in theory our system should be able to produce
Availability of the system can depend on many factors.
(*) Overavailability
16
Filename: Notes.docx
Folder: /Users/silviasanasi/Documents/UNI/Master/Industrial Technologies/IT Terzi
Template: /Users/silviasanasi/Library/Group Containers/UBF8T346G9.Office/User
Content.localized/Templates.localized/Normal.dotm
Title:
Subject:
Author: Silvia Sanasi
Keywords:
Comments:
Creation Date: 10/3/16 8:24:00 AM
Change Number: 465
Last Saved On: 2/20/17 11:01:00 AM
Last Saved By: Silvia Sanasi
Total Editing Time: 261 Minutes
Last Printed On: 2/26/17 12:03:00 PM
As of Last Complete Printing
Number of Pages: 16
Number of Words: 5.442 (approx.)
Number of Characters: 31.020 (approx.)