Process Analysis

Useful Links and Simulations: Process types

• Blocked and starved tasks:
https://s3.amazonaws.com/he-assets-prod/interactives/013_blocked
_and_starved/Launch.html
• Cycle times with multiple Workforce / workstations:
https://s3.amazonaws.com/he-assets-prod/interactives/022_cycle_ti
me_multiple_workers/Launch.html
• Batch and Single-piece Production:
https://s3.amazonaws.com/he-assets-prod/interactives/021_batch_a
nd_single_production_flow/Launch.html
• Machine-paced / Conveyor-paced assembly lines:
https://s3.amazonaws.com/he-assets-prod/interactives/032_conveyo
r_paced_assembly_line/Launch.html
Process Performance Metrics
Process Maps and Flow Diagrams
Process maps are planning and management tools that provide visual
representation of workflow in a process.

Purpose of Process Mapping:

• To establish a common understanding of the current processes and associated
operational requirements and deliverables.
• Clarify and resolve any ambiguity in flow of processes and enable focus.
• Simplify phenomenon when the existing processes are extremely complicated.
• Identify and eliminate waste and redundancies  improve efficiencies.
• Improve conformance quality: the ability to meet the specified customer stds.
Types of Process Flow maps
• Process Flowchart Map
• Swim lane Flowchart Maps / Process Blueprints
• Suppliers-Inputs-Process-Outputs-Customers (SIPOC) Tool
• Value Stream Map (VSM)
Process Analysis and Process Mapping
• Step-1: Identify the problem or process to map
• Question: Is there an inefficient process or a business problem that needs to be
improved?
• Violation of the customer promise is a major concern
• Lack of process standard, or variation among processors could be a concern here
• Process visualization could help: Process mapping
• Step-2: Clearly define the flow unit
• Step-3: List the activities involved
• Step-4: Identify and write down the sequence of steps:
• A precise identification of start and the stop is a must.
• Step-5: Map the process using standard process mapping symbols
• Step-6: Finalise and validate the process map
Process Mapping Symbols
Start / Begin OR Stop / End

Process Steps

Decision Nodes

Arrows / Direction of Process flow

OR Inventories
SIPOC Tool
The Six-Step Process:
1. Define / Name the process: Verb + Noun: e.g., Disburse Loan
2. Define Outputs: the tangible outputs that the process will deliver
3. Define Customers: the entities that receive outputs
• Each output must have a customer
4. Define Inputs: the things that trigger the process, e.g., Loan request
5. Define Suppliers: the entities that supply inputs
• Every input should have supplier
6. Define Sub-processes: the activities that convert inputs to outputs
• Form basis of extended process map
SIPOC Tool
Value Stream Mapping (VSM)
Process Flow Diagrams

Process Flow Diagram for a Café

 Process Flow Diagrams

Process Flow Diagram for making a chair: Parallel Processes

 Process Flow Diagrams

Process Flow Diagram for making a sequential process:

Process Flow Diagrams
Gantt Chart for the four-step sequential process:
 Process Flow Diagrams

Process Flow Diagram for making a sequential process: Combining tasks B and C
 Process Flow Diagrams

Process Flow Diagram for making a chair: Parallel Processes

Process Flowcharts
Process Maps: Swimlane Diagrams
Process Types
Types of Activities
• Value-Adding Activities (VA)
• Non-Value-Adding Activities (NVA)
• Value-Enabling Activities (VE)

Non-Value-Adding Activities: 7+1 Types of Waste (TIMWOOD ‘S’)

• Transport
• Inventory
• Motion
• Waiting
• Overproduction
• Overprocessing
• Defects
• Skill
 Product-Process Matrix

Adapted from Hayes and Wheelbright (1979)

 Process Flow analogy

What Goes into a Process Must Come Out of the Process.

Input Rate Must Be Less Than or Equal to the Output Rate; Otherwise, the System Will Overflow.
 R

The Penny Fab

• Characteristics:
• Four identical tools in series.
• Each takes 2 hours per piece (penny).
• No variability.
• Job releases to maintain constant WIP.

• Parameters: 0.5 pennies/hour

rb = 8 hours
T0 =

p. 4
 R

The Penny Fab

p. 4
The Penny Fab (WIP=1)

Time = 0 hours
The Penny Fab (WIP=1)

Time = 2 hours
The Penny Fab (WIP=1)

Time = 4 hours
The Penny Fab (WIP=1)

Time = 6 hours
The Penny Fab (WIP=1)

Time = 8 hours
The Penny Fab (WIP=1)

Time = 10 hours
The Penny Fab (WIP=1)

Time = 12 hours
The Penny Fab (WIP=1)

Time = 14 hours
The Penny Fab (WIP=1)

Time = 16 hours
Penny Fab Performance

WIP TH FT THFT
1 0.125 8 1
2
3
4
5
6
See corresponding point on graphs on p.4 and in departure time table
The Penny Fab (WIP=2)

Time = 0 hours
The Penny Fab (WIP=2)

Time = 2 hours
The Penny Fab (WIP=2)

Time = 4 hours
The Penny Fab (WIP=2)

Time = 6 hours
The Penny Fab (WIP=2)

Time = 8 hours
The Penny Fab (WIP=2)

Time = 10 hours
The Penny Fab (WIP=2)

Time = 12 hours
The Penny Fab (WIP=2)

Time = 14 hours
The Penny Fab (WIP=2)

Time = 16 hours
The Penny Fab (WIP=2)

Time = 18 hours
Penny Fab Performance

WIP TH FT THFT
1 0.125 8 1
2 0.250 8 2
3
4
5
6
See corresponding point on graphs on p.4 and in departure time table
The Penny Fab (WIP=4)

Time = 0 hours
The Penny Fab (WIP=4)

Time = 2 hours
The Penny Fab (WIP=4)

Time = 4 hours
The Penny Fab (WIP=4)

Time = 6 hours
The Penny Fab (WIP=4)

Time = 8 hours
The Penny Fab (WIP=4)

Time = 10 hours
The Penny Fab (WIP=4)

Time = 12 hours
The Penny Fab (WIP=4)

Time = 14 hours
Penny Fab Performance

WIP TH FT THFT
1 0.125 8 1
2 0.250 8 2
3 0.375 8 3
4 0.500 8 4
5
6
See corresponding point on graphs on p.4 and in departure time table
The Penny Fab (WIP=5)

Time = 0 hours
The Penny Fab (WIP=5)

Time = 2 hours
The Penny Fab (WIP=5)

Time = 4 hours
The Penny Fab (WIP=5)

Time = 6 hours
The Penny Fab (WIP=5)

Time = 8 hours
The Penny Fab (WIP=5)

Time = 10 hours
The Penny Fab (WIP=5)

Time = 12 hours
Penny Fab Performance

WIP TH FT THFT
1 0.125 8 1
2 0.250 8 2
3 0.375 8 3
4 0.500 8 4
5 0.500 10 5
6 0.500 12 6

See corresponding point on graphs on p.4 and in departure time table

 H
Penny Fab One Animation
Departure Times

• w=1 w=2 w=3 w=4 w=5 w=6:

• 8 8 8 8 8 8
• 16 10 10 10 10 10
• 24 16 12 12 12 12
• 32 18 16 14 14 14
• 40 24 18 16 16 16
• 48 26 20 18 18 18
• . .
• . .
• THAvg 0.125 0.25 0.375 0.5 0.5 0.5
 R

TH vs. WIP: Best Case

0.6
rb 0.5
0.4
TH
0.3
1/T0
0.2
0.1
0
0 1 2 3 4 5 6 7 8 9 10 11 12
W0
WIP
p. 4
 R

FT vs. WIP: Best Case

26
24
22
20
18
16 1/rb
14
FT
12
10
T0 8
6
4
2
0
0 1 2 3 4 5 6 7 8 9 10 11 12
W0 WIP
p. 4
 R

A Manufacturing Law p. 4

• Little's Law: The fundamental relation between WIP,

FT, and TH over the long-term is:
W IP  TH  FT
parts
parts   hr
hr

• Insights:
• Fundamental relationship
• Simple units transformation
• Definition of Flow time (FT = WIP/TH)
Little’s Law (WIP = TH*FT) check in HAL case: 1400 X 34 = 47,600
Little’s Law Analogy

Arrivals WIP: Items in Queue Departures

AND Items in service

WIP = TH x FT
(Number = Rate X Time)

Analogous to:
Distance = Speed X Time Interval
 H
Relating D = S x T to Little’s Law
Time = T (FT)

Speed (S)
• Distance (D)

Arrival Rate (TH)

• Number of elements
waiting to be serviced
(WIP)

Little’s Law : WIP = TH x FT

 H

Best Case Performance ..

• Example: For Penny Fab, rb = 0.5 and T0 = 8, so W0 = 0.5  8 = 4,
8, if w  4
FTbest 
2 w, otherwise.

w / 8, if w  4
TH best 
0.5, otherwise.

which are as per the curves we plotted.

 R

p. 4
Best Case Performance
• Best Case Law: The minimum Flow time (FTbest) for a given WIP level,
w, is given by
T0 , if w  W0
FTbest 
 w / rb , otherwise.

The maximum throughput (THbest) for a given WIP level, w is given by,
 w / T0 , if w  W0
TH best 
 rb , otherwise.
 Penny Fab Two:
Parallel Stages with Non-identical Times
 R

Penny Fab Two

2 hr
5 hr 3 hr

10 hr
p. 5
 R

Penny Fab Two

Station Number of Process Station

Number Machines Time Rate
1 1 2 hr 0.5 j/hr
2 2 5 hr 0.4 j/hr
3 6 10 hr 0.6 j/hr
4 2 3 hr 0.67 j/hr

rb = ____________ T0 = ____________ W0 = ____________

p. 5
 R

Penny Fab Two

Station Number of Process Station

Number Machines Time Rate
1 1 2 hr 0.5 j/hr
2 2 5 hr 0.4 j/hr
3 6 10 hr 0.6 j/hr
4 2 3 hr 0.67 j/hr

0.4 p/hr
rb = ____________ 20 hr
T0 = ____________ 8 pennies
W0 = ____________

p. 5
Penny Fab Two Simulation (Time=0)

2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=2)

7
4

2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=4)

7
6
9
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=6)

7
8
9
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=7)
17

12
8
9
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=8)
17

12
10
9
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=9)
17

19

12
10
14
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=10)
17

19

12
12
14
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=12)
17

19

17 22
14
14
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=14)
17

19

17 22
16
19 24
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=16)
17

19

17 22

19 24
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=17)
27

19

22 22 20

19 24
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=19)
27

29

22 22 20

24 24 22
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=20)
27
Note: job will arrive at
bottleneck just in time
to prevent starvation. 29

22 22
22
24 24 22
2 hr
5 hr 3 hr

10 hr
Penny Fab Two Simulation (Time=22)
27

29

27 32 25
24
24 24
2 hr
5 hr 3 hr

Note: job will arrive at

bottleneck just in time
to prevent starvation. 10 hr
Penny Fab Two Simulation (Time=24)
27

29

27 32 25

29 34 27
2 hr
5 hr 3 hr

And so on….
Bottleneck will stay busy;
all other stations
10 hr will starve periodically
 H

Penny Fab Two Animation

• Departure times:

• 20, 22, 25, 27, 30, 32, 35, 37 ..

• THAvg = two units every 5 hours i.e. 2/5 = 0.4 per hour
0.4 p/hr 20 hr 8 pennies
rb = ____________ T0 = ____________ W0 = ____________
Effects of Variability
 R

Successive Units on Time Line

How variable is this system ?

p. 5
 R

Understanding Variability

• Variability: Non-uniformity of a class of entities

• (departure from regular, predictable behavior)

• Associated with, but not identical to, Randomness

• (variability without explicitly analyzed cause)

p. 5
 H
Variability: Where does it come from?
Tasks:
• Inherent variation
• Lack of SOPs
• Quality (scrap / rework)

Processing
Buffer
Input:
• Unpredicted
Volume swings
• Random arrivals Resources:
(randomness is the rule, • Breakdowns / Maintenance
not the exception) • Operator absence
• Incoming quality • Set-up times
• Product Mix
Routes:
Especially relevant in service operations • Variable routing
(i.e. not just factories): • Dedicated machines
• emergency room
• air-line check in
• call center
• check-outs at cashier
Where Does Variability Come From ?
• Sources of Variability:
• setups and batching • work-pace variation
• machine failures • differential skill levels
• materials shortages • engineering change orders
• yield loss • customer orders
• rework • product differentiation
• operator unavailability • material handling
Quantifying Variability
 R

Some Notations ..

• t: Average of random variable (time - arrival, processing,

departure etc.)
• σ2 : Variance of random variable
• c2 : SCV, Squared Coefficient of Variation, = σ2/t2
• mf : Mean Time Between Failures (MTBF)
• mr : Mean Time To Repair (MTTR)
• A: Availability of machine = mf / (mf + mr)
• ts : Setup time
• Ns : Number of units produced between setups p. 3
 R

Variability Classes in Factory Physics

Low variability Moderate variability High variability

(LV) (MV) (HV)
ce
0 0.75 1.33

• Effective Process Times:

• HV, LV, and MV are all possible in effective process times
• LV: Actual Process times without outages
• MV: Effective Process times with short adjustments (e.g.
Setups)
• HV: Effective Process times with long outages(e.g.
Failures)
p. 5
 R

Variability Formulae

p. 5
 R

Variability Formulae

 t0 / A

mr
 c  (1  c ) A(1  A)
2
0
2
r
t0

p. 5
Situation II: Tortoise and Hare - Availability

•Hare X19: •Tortoise:

t0 = 15 min t0 = 15 min
0 = 3.35 min 0 = 3.35 min
c02 = 02 /t02 = 3.352/152 = 0.05 c02 = 02 /t02 = 3.352/152 = 0.05
mf = 12.4 hrs (744 min) mf = 1.9 hrs (114 min)
mr = 4.133 hrs (248 min) mr = 0.633 hrs (38 min)
cr = 1.0 cr = 1.0

•Availability:
m

f 744
 0 .75
mf

114
 0 .75
m f  mr 744  248 m f  mr 114  38
A= A=

No difference between machines in terms of availability.

Tortoise and Hare - Variability
•Hare X19: •Tortoise 2000
t0 15 t0 15
  20 min   20 min
• te = A 0 .75 • te = A 0 .75
m mr
c  (1  c ) A(1  A) r 
2
0
2
r
c  (1  c ) A(1  A)
2
0
2
r 
t0 t0
•ce2 = 248 •ce2 =0.05  (1  1)0.75(1  0.75) 38 
0.05  (1  1)0.75(1  0.75) 
15 15
6.25 high variabilit y 1.0 moderate variabilit y

Hare X19 is much more variable than Tortoise 2000!

Variability Effects

• Effect of Breakdowns: Hare or Tortoise?

Not to
scale
 The Student Syndrome

Degree
of
Effort Uniform
effort line

Deadline

Time on Project
Not to
scale
Variability Effects

• Example of Setups …
Variability Effects: Setup Example ...
• Data:
• Fast, inflexible machine – 2 hr setup every 10 jobs
t0  1 hr
N s  10 jobs/setup
t s  2 hrs
te  t0  t s / N s  1  2 / 10  1.2 hrs
re  1 / te  1 /(1  2 / 10)  0.8333 jobs/hr

• Slower, flexible machine – no tsetups

 1.2 hrs
0

re  1 / t0  1 / 1.2  0.833 jobs/hr

• Traditional Analysis?

No difference!
 H

... Setup Example …

• Revised Approach: Compare mean and variance
• Fast, inflexible machine – 2 hr setup every 10 jobs
t0  1 hr
c02  0.0625
N s  10 jobs/setup
t s  2 hrs
cs2  0.0625
te  t0  t s / N s  1  2 / 10  1.2 hrs
re  1 / te  1 /(1  2 / 10)  0.8333 jobs/hr
 cs2 N s  1 
σ    t 
2
e
2
0
2
s 2 
  0.4475
 Ns Ns 
ce2  0.31
 H

… Setup Example …
• Slower, flexible machine – no setups
t0  1.2 hrs
c02  0.25

re  1 / t0  1 / 1.2  0.833 jobs/hr

ce2  c02  0.25

• Conclusion: Flexibility can reduce variability.

 H

… Setup Example
• New Machine: Consider a third machine same as previous machine with setups, but with shorter,
more frequent setups
N s  5 jobs/setup
t s  1 hr

• Analysis: re  1 / te  1 /(1  1 / 5)  0.833 jobs/hr

 cs2 N s  1 
σ    t 
2
e
2
0
2
s 2 
  0.2350
 Ns Ns 
ce2  0.16

Shorter, more frequent setups induce less variability.

• Conclusion:
 Propagation of Variability
From Single Station to Serial Stations (Tandem Queues) ..
Propagation of Variability

re (i) re (i+1)
ra (i) rd (i) = ra (i+1) rd (i+1)
i i+1
cd2 (i) = ca2 (i+1)
ca (i)
2
cd2 (i+1)
ce2 (i) ce2 (i+1)

Departure variability depends on arrival variability and

process variability along with the station utilization.
Propagation of Variability –
High Utilization Station
LV HV HV

HV HV HV

LV LV
LV

HV LV LV

Conclusion: flow variability out of a high utilization station (u → 1)

is determined mostly by process variability at that station (cd2 → ce2).
Propagation of Variability –
Low Utilization Station
LV LV
HV

HV HV HV

LV LV
LV

HV LV HV

Conclusion: flow variability out of a low utilization station (u → 0) is

determined mostly by process variability into that station (cd2 → ca2).
Variability Linking Relationship
A useful interpolation incorporating this intuition:
• Single Machine Station:
cd2  u 2 ce2  (1  u 2 )ca2

where u is the station utilization given by u = rate

• Multi-Machine2 Station:2 u2 2
cd  1  (1  u )(c  1) 
2
a (ce  1)
m
ra te
u
m
where m is the number of (identical) machines and
Flow Time is Affected by:

• Utilization

• Variability

• Both magnitude and position matter

 Variability Placement Example ..

• Two machines in series, each processing in large batches

due to long setups and resulting in large processing time SCV.

ra=0.9/hr te=1 hr. te=1 hr.

ca2=1.0 ce2=9.0 ce2=9.0

 .. Variability Placement Example ..

• From queueing equations, FTq(1) = 45 hrs.

• From linking equation, ca2(2) = cd2(1) = u2 ce2 + (1 - u2)
ca2 = 7.48

• Applying same equations at station 2,

• FTq(2) = 74.16 hrs.
• and
• Total Flow Time = Queue Time + Process Time = 45 + 1
+ 74.16 + 1 = 121.16 hrs.
 .. Variability Placement Example

• Suppose one flexible machine with lower setup time (and

hence lower batch size and therefore lower SCV) is available to
replace one of the two machines, which one should it replace ?
Assume te=1 hr. and ce2=0.25 for the new machine.

• Replace Machine 2:
• FTq(2) reduces to 34.79, so Total Flow Time = 81.79

• Replace Machine 1:
• FTq(1) reduces to 5.63, cd2(1) = 0.39, and as a result FTq(2)
reduces to 42.27
• Total Flow Time = 49.90 (significantly less)
Propagation Laws

• Variability Placement:
• Variability early in a routing has a larger
impact on WIP and flow times than equivalent
variability later in the routing.

• Pay Me Now or Pay Me Later:

• If you cannot pay for variability reduction, you
will pay in one or more of the following ways:
• 1. Long flow times and high WIP levels.
• 2. Wasted capacity (low utilization of resources).
• 3. Lost throughput.
“Pay Me Now or Pay Me Later” Scenario (Fig.
9.1)

Each “Job” consists of 50 pieces.

Each station’s effective time is 20 minutes.
Plant works 1440 minutes per day.
“Pay Me Now or Pay Me Later” Results
(Simulation Based)
Some explanatory calculations for Table 9.3

• Utilization values are Actual Throughput divided by Ideal

Throughput (50*1440/20 = 3600, i.e. 2.5 pieces per min.)

• T0 values are 20+20 = 40’ for Cases 1, 2 and 4; and 20+10

= 30’ for case 3

• W0 values are calculated from rb X T0 , which works out to

2.5 X 40 = 100 for Cases 1, 2 and 4; and 2.5 X 30 = 75 for
Case 3

• Check the degree to which Little’s Law is satisfied in

different cases
Understanding & Modelling Queues
Waiting & Service Time

Inventory
waiting
Outflow
Inflow

Entry to system Begin Service Departure

Time in queue Service Time

Flow Time
Waiting & Service with Multiple, Parallel Resources
Inventory
in service

Inventory
waiting
Inflow
Outflow

Entry to system Begin Service Departure

Time in queue Service Time

Flow Time
.. Notations ..

• m: Number of servers for a multiple server station

• u: Utilization of resource = ra/re or te/ta (or rate/m for a
multiple server station)
• r: Rate per unit time of a process-centre

• Subscripts Used:
• 'a' - arrivals 'd' - departures 'e' - effective process
time
• 's' - setup time 'b' - bottleneck 'q' - in queue
Queuing Based Relationships

• From M/M/1 to G/G/m …

• Derivation of formulae based on Queuing Theory
• (page 6 of handout)
Queuing Relationships

G/G/1: Kingman’s Approximation - VUT

22
+
c ce
a u
FTq (G / G / 1) = ( )( )te
2 1- u

G/G/m: Generalized VUT

22 2(m+1) -1
+
c ce u
a
FT q (G/G/m) = ( ) te
2 m(1 - u)
Summary Results of Queuing Analysis

• VUT Equation
• FTq = V X U X T
• Where:
• Variability Factor V = (ca2 + ce2)/2
• Utilization Factor U = u/(1-u) or u{[√2(m+1)] – 1} / m(1-u)
• Time Factor T = te (i.e. effective process time after adjusting all detractors)

• Useful model of single and multi-machine workstations

• Separate terms for variability, utilization, process time.
• FTq (and other measures) increase with ca2 and ce2
• Arrival variability, process variability, or both can combine to inflate queue time
• General, fast and accurate – and easily implemented in a spreadsheet (VUT) or
packages like MPX.
Flow Time vs. Utilization
Utilization & Variability Laws

• In steady state, increasing variability always increases average

flow times and WIP levels.

• If a system increases utilization without making any other

changes, average flow times will increase in a highly non-linear
fashion.

• In steady state, all plants will release work at an average rate

that is strictly less than the average capacity.
 The Cost of Queues

• - Customer
• - Provider
• - Society
• ..
 Capacity versus Flow Time Tradeoff
The Total-Cost Curve is roughly U-Shaped
The wall shows Bottleneck capacity
Operation is over-utilized
TH rate here is too low
Cost per Unit Time

Total Cost

Capacity Cost

Flow Time Cost

0 2% 25%
Slack % = fraction of time capacity not active = 1 – u
 Limitation of Analysis

• VUT Equation:
• characterizes stations with infinite space for queueing
• useful for seeing what will happen to WIP, CT without restrictions

• But real world systems often constrain WIP:

• physical constraints (e.g., space or spoilage)
• logical constraints (e.g., kanbans)

• Blocking Models:
• estimate WIP and TH for given set of rates, buffer sizes
• much more complex than non-blocking (open) models, often
require simulation to evaluate realistic systems

• Capturing Variability: c2 may miss some patterns of variability

Queuing Relationships: M/M/1 ..

u
WIP(M/M/1) =
1- u

WIP(M/M/1) u te
FT(M/M/1) = = =
ra r a (1 - u) 1 - u

Using substitutions in combination: ra = 1/ ta; re = 1/te

.. Queuing Relationships: M/M/1

u
FT q (M/M/1) = FT(M/M/1) - t e = t e
1- u

2
u
WIP q (M/M/1) = r a FT q =
1- u

Using substitutions in combination: ra = 1/ ta; re = 1/te

Queuing Relationships: M/M/m

2(m+1) -1
te u
Ft q (M/M/m) =
m(1 - u)

Queuing Relationships: G/G/1

(Kingman’s Approximation - VUT)
2 2 2 2
c a + ce u c a + ce
FT q (G/G/1) = ( )( )te = ( ) FT q (M/M/1)
2 1- u 2
Queuing Relationships: G/G/m

Analogously from previous result:

2 2
+
c ce a
FT q (G/G/m) = ( ) FT q (M/M/m)
2

Final relationship (generalized VUT):

2 2 2(m+1) -1
+
c ce ua
FT q (G/G/m) = ( ) te
2 m(1 - u)
 Economics of Waiting Lines

Waiting

Cost
Capacity

None A Lot
Facility Capacity
Flow Time vs. Throughput Capacity
Analysis of Processes with Batching
Process Batching
• Types of Process Batching:
1. Serial Batching:
• processing is one-at-a-time, but with setups between items
(e.g. fabrication, assembly)
• “batch size” is number of consecutive jobs between setups
• batching used to reduce loss of capacity due to setups

2. Parallel Batching:
• processing itself is done in batches (e.g., plating, baking,
transport, education)
• “batch size” is number of jobs run concurrently
• batching used to increase effective rate of process

Results in a mismatch between demand and supply rates for some time.
Batching Situations

process
buffer process buffer
process
BATCHING EFFECTS:
Inventory vs. Capacity Perspective
 Batch Size Choice Example
• Example based on *:

• Steer Support:
• Setup time: 1 hour
• Processing time: 1 minute

• Ribs (2 per assembly):

• Setup time: 1 hour
• Processing time: 0.5 minute per rib = 1 minute per set

• Assembly:
• No setup
• Processing time 3 minutes per set

* Cachon, G. & Terwiesche, C., “Matching Supply with Demand”, Tata McGraw Hill, 2012
Batch size and the location of the
bottleneck ..

Milling Machine Assembly process

Set-up time 120 minutes -

Processing time 2 minutes/unit 3 minutes/unit

Batch Size
Capacity given Batch Size =
Set-up time + Batch-size*Time per unit

• Note:
• Times are aggregated per unit end product
• Capacity is in end product output rates
 .. Batch size and the location of the bottleneck

Milling Machine Assembly process

Set-up time 120 minutes -

Processing time 2 minutes/unit 3 minutes/unit
Capacity (B=12) 0.0833 units/minute 0.333 units/minute
Capacity (B=60) 0.25 units/minute 0.333 units/minute

Capacity (B=120) 0.333 units/minute 0.333 units/minute

Capacity (B=200) 0.3846 units/minute 0.333 units/minute

Capacity (B=300) 0.4166 units/minute 0.333 units/minute
 Batch Operations Save Capacity
Inventory
[in units of xootrs]

133
Steer support
inventory

Rib
inventory

200 260 460 520 600 800 860 1060 1120 1200 1400 1460 Time [minutes]

Production cycle

Produce ribs Produce steer supports Idle time

Set-up from Ribs to Steer support Set-up from steer support to ribs

For a batch size of 200 units of end product

 But Batch Operations Carry Inventory
Production with large batches Production with small batches

Cycle
Cycle
Inventory
Inventory
Produce Sedan

Produce Station wagon

Beginning of End of Beginning of End of

Month Month Month Month
Optimal Batch Size

Assembly 360 Time

60 120 60 120

Setup ribs Produce ribs Setup supports Produce supports

Neither Machinists nor Assemblers are idle, but ………

The example is optimized assuming that the Setups are fixed and
cannot be shortened. Attacking setup times (e.g. through SMED) would
enable smaller batches, resulting in lower waiting time, smoother flow,
and reduced inventory.
The example also assumes that there is no variability at all.
BATCHING EFFECTS:
Flow Time Perspective

-
 PROCESS BATCHING H

EFFECTS
• Notation: Setup time = ts, Processing Time = t0, Process
batch quantity = Ns
• If ts = 4 hrs., t0 = 1 min., then:
• Ns=1000 Ns=4000

• Total Time 1240’ 4240’

• Setup Time % 240/1240 240/4240

= 19.3% = 5.6%
• Avg. Processing
Time/ Unit 1.24’ 1.06’
Serial Batching Schematic ..
Process Batching Effects
(Contd.)
• If ts reduced to 1 hour:
Ns=1000 Ns=250

• Total Time 1060’ 310’

• Setup Time % 60/1060 60/310

= 5.6% = 19.3%
• Avg. Processing
Time/ Unit 1.06’ 1.24’
 Process Batching Effects
(Contd.)
• Average processing time per unit = t0 + ts/Ns
(“allocation” or “utilization” effect)

• Total wait + process time = ts + Ns*t0 for each unit in

the batch (“cousins” effect)

• Large batches (Ns’s) also cause other batches to wait

in similar fashion (“rivals” effect)

• First effect pushes towards larger batches, while

latter two effects push towards smaller batches
.. Serial Batching Schematic
 Flow Time vs. Batch Size
(with 5 hr. setup time)

Flow Times (hrs)

Split lot: moving part of the batch to next process before full batch is complete .
 Flow Time vs. Batch Size
(2.5 hr. setup time)

Flow Times (hrs)

Split lot: moving part of the batch to next process before full batch is complete .
Superimposed Batch Size

Note: Cycle times to be read as Flow times

Move Batching Effects
 MOVE BATCHING EFFECTS ..
• Smaller Move Batches help to reduce
Flow time by better overlapping of
work

• Larger Move Batches reduce Material

Handling frequency and cost

• Larger Move Batches make flow

timings “lumpier” (proportionate to
b: see illustration)
Move Batching Illustration
Op.1 Op.2

Lead Time

Op.1A Op.1B Actual schedule

depends on which
operation is faster
Op.2A Op.2B

Lead Time
 Move Batching Effect Example
• A Fork Lift Truck brings material in lots of 16 at deterministic intervals of 8
hours.

• Inter-arrival times: 8 hours for first piece off the truck, and 0’s for remaining
15 pieces.

• Average inter-arrival time:

• ta = Σt / n = [8 + 0 + … + 0] / 16 = 0.5

• Variance of inter-arrival time:

• σa2 = Σt2 /n - ta2 = [82 + 0 + … + 0] /16 - 0.52 = 3.75

• SCV of inter-arrival time:

• ca2 = σa2/ ta2 = 3.75/ 0.52 = 15 i.e. (b-1)
 MOVE BATCHING EFFECTS
• Smaller Move Batches help to reduce Larger Move batches (b>1) cause
Flow time by better overlapping of wait-for-batch time
work (proportionate to b:
{b-1}{1/ra}/2) in addition to
• Larger Move Batches reduce Material usual process queue time
Handling frequency and cost
Larger Move batches also
• Larger Move Batches make flow somewhat dampen flow
timings “lumpier” (proportionate to timing variability (by a factor
b: see illustration) of 1/b: c2batch = c2individual/b)
 Batching Laws

• Process Batching Law: In stations with batch operations or significant

changeover times,
1. The minimum process batch size that yields a stable system may be greater
than one.
2. As process batch size becomes large, flow time grows proportionally with
batch size.
3. Flow time at the station will be minimized for some process batch size, which
may be greater than one.

• Move Batching Law:

Flow times over a segment of a routing are roughly proportional to the
transfer batch sizes used over that segment (provided there is no waiting for
the conveyance device).
The Power of Aggregation: Push vs Pull:
Example
A textile company produces and offers variants of shirts, comprising four
colours. Demand for each variant is N(100,10). The company wants to meet
95% of the demand successfully.

The company is considering two process designs:

1. The dyeing is done earlier in the process.
2. The dyeing is done later in the process.

Represent both processes using process maps.

Find how much FG inventory needs to be stocked in each case.

Useful Links and Simulations: Queueing Systems
• Queueing systems: Cumulative arrivals and departures:
https://s3.amazonaws.com/he-assets-prod/interactives/010_cumulati
ve_arrivals/Launch.html
• Average time in Queue:
https://s3.amazonaws.com/he-assets-prod/interactives/040_avg_tim
e_queue_single/Launch.html
• Average time in Queue_Multiple Servers:
https://s3.amazonaws.com/he-assets-prod/interactives/041_avg_tim
e_queue_multiple/Launch.html
Useful Links and Simulations: Process Control and
Quality
• Visualising Defect rates:
https://s3.amazonaws.com/he-assets-prod/interactives/001_defect_r
ate/launch.html
• Central Limit Theorem:
https://s3.amazonaws.com/he-assets-prod/interactives/051_central_l
imit_theorem/Launch.html
• Process Capability Index:
https://s3.amazonaws.com/he-assets-prod/interactives/002_process_
capabilities/launch.html
• Correlation Coefficient:
https://s3.amazonaws.com/he-assets-prod/interactives/043_correlati
on_coefficient/Launch.html
• Six sigma compliance and Process Capability:
Useful Links and Simulations: Managing Quality
• Failure costs in Processing Chip: Cost of Damage Control – The sooner
the better:
https://s3.amazonaws.com/he-assets-prod/interactives/053_fc_proce
ssing/Launch.html
• Failure costs in Agriculture: Cost of Damage – The earlier in the supply
chain, the better:
https://s3.amazonaws.com/he-assets-prod/interactives/052_fc_agric
ulture/Launch.html