Housekeeping

Judge
MST Business
Michaelmas 2009 School • Part I (lecture 1-4): Factory-level planning
• Part II (lecture 5-8): Supply chain management

Operations p
• Four supervisions ((see Camtools for g
groups
p and dates))
Not always right answer, so no cribs!
Management • Past exams / “The Goal” Goldratt (session 5)

• Contact details:
Centre for Process Excellence and Innovation
M i k Stevens
Merieke St 14 Trumpington Street
Tel: 01223 7 48827
E il [email protected]
Email: 700@ k
Page 2

Session Overview Learning Objectives

Part I Part II
1. Process Fundamentals and types of manufacturing 5. Factory-level Scheduling and MRP Systems
• Understand the basic decisions in Operations Management,
ƒ Information Flow, Material Flow ƒ g
Order Fulfilment strategies and their implications
p on firm p
performance
ƒ Productivity, Improvement, Capacity, Utilisation ƒ Push versus Pull scheduling
ƒ Objectives and Tradeoffs ƒ
ƒ
MRP systems
Just-in-Time and Kanban systems
• Be able to manage inventory, schedule processes, develop
2. Inventory Management
ƒ Reordering Policies y
6. Toyota Production System
y and Lean Thinking
g basic forecasts
ƒ ABC Analysis, Pareto Chart ƒ 7 Wastes, 5 Lean Principles
ƒ Little’s Law ƒ Lean Toolbox, Lean Services • Understand the Lean, Six Sigma and TOC improvement
ƒ EOQ, EPQ, POQ models and their limitations ƒ Implementing Lean
3 Forecasting and Assembly Line Balancing
3. 7 Quality Management
7. Management, Six Sigma
Sigma, Project Man
Man. concepts
p
ƒ Moving Average ƒ Service industry
ƒ Exponential Smoothing ƒ Demand and revenue management • Be able to manage a project (Critical Path Method)
ƒ Decomposition and Fourier Analysis ƒ Sigma Level of processes
ƒ Forecast errors ƒ Project Management and NPD,
NPD differences to
ƒ Assembly Line Balancing manufacturing operations. Critical path method. • Understand the need to manage the wider supply chain
4. Machine-level Scheduling 8. Supply Chain Management
ƒ Single machine: Minimising average completion/flow ƒ SCM as competitive advantage
ti
time, maximum
i llateness,
t number
b off late
l t jobs,
j b average ƒ Diff
Difference OM andd SCM
tardiness ƒ Managing Supplier Relations
ƒ Two machines in-line: minimising makespan ƒ Supply Chain Collaboration
ƒ Course Review

Page 4
What is Operations Management? Why does OM matter?
– Operations Management (OM) is the activity of managing the
resources which are devoted to the production and delivery of products
• One of the three core activities of the firm
and services to end customers. ƒ Marketing and sales
ƒ Product/service development
– OM uses p process thinking g to meet and exceed customer demand
ƒ Operations
while using all resources efficiently in order to maximise the value of the
organisation. • Operational decisions have strong financial
– Anything which repeats in the operation can be seen as a process. i li ti
implications...
The repetitive nature of processes allows for improvement.
• ... and often forgotten: also determine the day-to-
day to
– Project management : low volume, high variety operations with defined day service level at the customer end!
beginning and end.

– Distinction between operations which manufacture goods and those

which provide services (but is this distinction really meaningful?)

Page 5 Page 6

Objectives for Today

Judge Business School • Representing processes diagramatically
• Quantitative and qualitative measures we use to describe
processes
Session 1 • Categorise different types of manufacturing processes
Process Fundamentals • Id
Identify
tif th
the objectives
bj ti iinvolved
l d iin operations
ti
management, the costs, and the trade-offs
OM objectives and trade-offs

Page 8
Input - Output Diagram Productivity, Efficiency and Improvement
Basic definitions in Operations Management all relate to the
Input-Output process model:
I
Inputs
t O t t
Outputs
– Productivity = Ratio of Σ Outputs to Σ Inputs
Subsystem
ƒ E.g. Vehicles produced per labour hour
or ƒ “Benchmarking” as comparative analysis across firms
ƒ What are the main mechanisms for increasing productivity?
Transformation
process
– Efficiency
ƒ Aims at usingg the least amount of resources to p
produce a g
given g
good or
service at the lowest possible unit cost
Flows of orders, information, data,
energy, material, people, etc.
– Improvement = [ Productivity(t) / Productivity(t
Productivity(t-1)
1) -1]*100
1]*100
Input-output diagrams are a way of defining a system, ƒ Considers output per unit resource over time
e g a manufacturing System
e.g. ƒ Tracks productivity development over time

Page 9 Page 10

Example: A Bakery I Example: A Bakery II

• Processes are shown as boxes

boxes, inventories as triangles
triangles, • What are the implications of making this change to the
arrows depict flows flow?

Page 11 Page 12
 Example: A Bakery III A Simplified Manufacturing System
External
influences;
• Product variety adds complexity System boundary Taxation etc
Taxation, etc.

R
Raw material
t i l Planning &
suppliers Purchasing Scheduling

Goods In/ Component Final Customers

Receiving
g Manufacture Assembly
y FGI

Component WIP WIP

Raw
suppliers Materials
M t i l Manufacturing
system

Elements outside El
Elements
t in
i the
th system
t
the system

Page 13 Page 14

Capacity and Utilisation

• Cycle time: average time between units of output emerging from
External process
Raw material influences;
Taxation, etc.
• Capacity: maximum level of value-added activity a process is capable
suppliers System boundary
of over a period of time (units/time)
• Utilisation: Ratio of the actual output from a process to its design
Goods In/ Component Final
capacity (potential output if all capacity was used)
Customers
FGI
Receiving Manufacture Assembly
• Throughput:
Th h t the
th ti
time ffor a unitit to
t move through
th h a process
WIP WIP
Component Raw
suppliers Materials Manufacturing
system What determines the throughput of a system?
Capacity utilisation is a critical success factor :
- What happens if you don’t have sufficient capacity?
Elements in the Elements in the system
environment
- Can you think of examples of costs incurred by unused resources?

Page 15 Page 16
Exercise capacity utilisation Overall Equipment Effectiveness (OEE)

Work Work Work Available time

Cycle Time Capacity Capacity
station A station B station C (min/unit) (Output / Utilisation Downtime
hour) (%) (labour, e.g. Tot. operating time
Availabilit breaks), Availability a=
Breakdowns, Available time
Total operating
p g time y losses
1 worker
worker, 1 worker
worker, 1 worker
worker, Set ups /
Set-ups
3 mins 5 mins 2 mins change-overs,
maintenance
2
1 worker, 1 worker,
workers
3 mins 2 mins
2.5 mins Equipment ‘idle’
Speed Net operating time
1.2 plus slow Performance p=
2 workers, 0.8 workers Net operating time losse Tot. operating time
workers running
2.5 mins 2.5 mins s equipment
2 5 mins
2.5

“Load balancing”
Value- Quality Defects / scrap
p Value-added time
Quality q =
added time losses Net operating time

ÆEfficiency is generally driven by capacity utilisation and

labour productivity... OEE = a * p * q
Page 17 Page 18

Exercise OEE
Maximum time available:
Management decides machine works 150 hrs: available time
Availability losses: 10 hrs (machine set-up) + 5 hrs (breakdowns)
Total operating time: available time – availability losses:
p
Speed losses: 5 hrs ((idling)
g) + 10% ((when the machine runs):
)

Net operating time: total operating time – speed losses: Types

yp of Manufacturing
g and
3% defects, valuable operating time: net operating time – quality losses Service Operations
a: p: q:
OEE (a * p * q) =

Page 19
Plane and car manufacture Manufacturing Process Types
• How do these two operations differ?

High
Project

Jobbing

Variety
Batch

Mass

Conti-

Low
nuous

Low Volume High

Page 21 Page 22

Projects - Millau Viaduct Job Shop - Flow Chart

Annealing Stamping Painting

Assembly

Washing Welding Brazing

• Labour and equipment is often brought to location of assembly
• Physical size and degree of customisation key factors Process-driven split
p into centres,, complex
p
routing and scheduling.
Page 23 Page 24
Job Shop: Aero Engines & Machine Tools Batch: Textile Production

• Volume key factor in justifying automation

• Volume does not justify dedicated lines or machinery • Short life
life-cycle
cycle means that machines need to be flexible for re
re-
• Parts often travel between work-shops, thus ‘job shop’ use with next batch/product
• Work centres are grouped by type of process: welding, drilling, painting • Changeovers between products

Page 25 Page 26

Ford Highland Park Model T Assembly Line

Moving Assembly Line in 1913

Running
Commutators Front Axles
Boards

Assembly

Radiators Gas tanks Rear Axles

250,000
, Vehicles Per Year,, One Model

Page 27 Page 28
Mass/Line production: Automobiles Continuous processing: Oil refinery

• Volume does justify dedicated lines

• Cycle time is set to pace entire factory
• Multi-model lines • Flow processes, often driven by chemical/physical needs
• Limited flexibility regarding volume and new models • Individual product is often not an entity (e.g. petrol)

Page 29 Page 30

Process Characteristics Manufacturing Process Types

Process Process
The single most important feature of a process in a business tasks flow

High
operation is the trade-off in its design between production Diverse / Project
Intermittent
complex
volume and p product variety y
Jobbing
ƒ Defines types of job design required
ƒ Defines necessary tools and technology

Variety
Batch
ƒ Defines cost structure
ƒ Defines relationship with suppliers Mass

ƒ Establishes customer expectations Conti-

Low
Repeated Continuous nuous
/ simple
Low Volume High

Page 31 Page 32
 What are we trying to achieve in OM?

Internal objectives: “shareholder

shareholder value”
value
ƒ 100% effective use of resources
ƒ Minimal operating expenses: Zero defects,
defects zero stock

External objectives:
ƒ
ƒ
Objectives
j and Trade-offs ƒ
ƒ Additional metrics: Flexibility, Safety and Service

“The right product, at the right time, at the right quality,

at the right price”

Page 34

Cost Implications of OM Decisions All OM Decisions are Trade-offs

Cost of inventory Customer service vs. operational
FORTUNE May 22,
ƒ Cost of capital & warehousing cost 2006
ƒ Cost of handling, quality implications, obsolescence ƒ Response time, order fulfilment Interview with Bill
Amelio, CEO of
Cost of production Setup cost vs. inventory Lenovo
ƒ Cost of inventory is a function of production batch sizes ƒ What are the optimal batch sizes in
production? Q: Will you manufacture
ƒ Cost of machines, labour
everything in China?
ƒ Opportunity
O t it costt off set-ups
t Inventory cost vs. ordering cost
Cost of logistics & distribution ƒ What are optimal order quantities? A: For notebooks it's definitely
in China, but for desktops
ƒ Cost of transportation, depending on frequency we'll do final assembly close
…Where to find the balance? to the customer. You don't
Cost of sales
want to stick them on a boat
boat,
ƒ Opportunity cost of lost sales because a PC's value drops
Cost of making a sale (interface to marketing...) each week anywhere from
half a percent to a percent
half-a-percent
and a half.
Page 35 Page 36
 Objectives Today
Judge Business School • What is inventory?
• Inventory cycles
• Arguments in favour and against inventory
Session 2 • Little’s Law
Inventory Management • Parts classification: Pareto and ABC Analysis
• Batch sizing decisions

Page 38

Inventory – Definitions
‘An accumulation of a commodity that will be used to satisfy
future demand.’
ƒ Johnson and Montgomery (Operations Research)

‘The stocks or items used to support production (raw

materials and work-in-process items), supporting activities
( i t
(maintenance, repair,
i andd operating
ti supplies),
li ) and
d customer
t
service (finished goods and spare parts).’
ƒ APICS (Association for Operations Management) Dictionary

‘Dead material.’
ƒ Taiichi
T ii hi Oh
Ohno (F
(Father
th off th
the T
Toyota
t Production
P d ti S System)
t )

‘A substitute for information.’

Wh is
What i needed,
d d what
h isi waste?
? ƒ Michael Hammer (Process Reengineering Guru)

Page 39 Page 40
Types of Inventory The Inventory Cycle (‘Sawtooth’)
Raw Materials:
ƒ Materials to which the manufacturer has not yet added value
Work-in-progress or Work-in-process (WIP): Order Quantity (Q)
ƒ Materials to which the manufacturer has added some value but still has

evel
more to add

Inventory Le
Finished Goods
ƒ Goods ready for shipment to the customers, with no more value to be
added
ƒ Also consider service parts...
Safety and Cycle Stock Reorder point
ƒ Safety stock: non-active component to protect against fluctuations of
demand, production and supply
ƒ Cycle stock: active component that depletes over time, and is 0 Lead Time
replenished cyclically
Order time Order
Terminology: Stock-keeping Unit (SKU): an item at particular location Placed (LT) Received

Page 41 Page 42

Reorder Point with Safety Stock Inventory – Arguments in favour

Little’s Law (see later) implies:
ƒ There is a minimum inventory needed to run the factory
y level

Buffer against uncertainty

ƒ Market demand (seasonality,
(seasonality promotions
promotions, etc
etc.))
Inventory

ƒ Production throughput (quality, machine breakdown, etc.)

ƒ Supply of components

Reorder point Exploitation of price fluctuations

ƒ Raw materials: cocoa, coffee, etc

Safety stock Smoothing or levelling of production

ƒ Small variation can be buffered through final goods inventory

0 Enables the achievement of economies of scale

LT LT
e
Time

Page 43 Page 44
Inventory – Arguments against Hidden Costs of Inventory
Cost involved:
• Longer lead times
ƒ Cost of capital: value
value*ii, i=interest rate per unit time
ƒ Opportunity cost: How much would the capital earn otherwise? • Reduced responsiveness
ƒ Depreciation of goods
ƒ Stock obsolescence and deterioration • Underlying problems are hidden rather than being
ƒ Quality defects due to handling exposed and solved
ƒ Labour and handling
• Quality
Q lit problems
bl are nott id
identified
tifi d iimmediately
di t l
ƒ Warehousing, rent and energy
ƒ Insurance and overhead to admin labour, space, etc. • No incentive for improvement of the process
Overall costs:
ƒ Typical estimate is 20-30%, but often excludes quality,
depreciation, and opportunity cost
ƒ Key issue: estimates almost always too conservative!

Page 45 Page 46

Rock – Boat analogy Inventory reduction only is fatal!

Reduced
Rock = Problems Inventory Rock = Problems Inventory Level

Defects Defects
Supply Shortages Supply
Damage Issues
Equipment
Damage Issues Shortages
Breakdowns Equipment
Breakdowns

Page 47 Page 48
Little’s Law Exercise Little’s Law
John D.C. Little’s Theorem (or Little's Law) gives a simple relation • A company assembles computers. The process has
between inventory and lead-time. Applies to all types of systems!
three stages – assembly,
assembly testing and packing – which
take 975 minutes in total
I = R *T • A work day has 7
7.5
5 hours
ƒ I is the number of items or inventory in a system [units] • Average daily demand is 1,600 units
ƒ R is the production rate at which items arrive/leave [units/day]
ƒ T is the lead-time (here, the time a job spends in the system) [days] • Current WIP levels (for all three presses combined) are
ƒ g , steady-state
all based on average, y values. 4,800 units. Consultants hired by the CEO think this is
I determines the minimum pipeline stock needed! too much, and suggest to reduce stock by 50%
py
• Your reply?

Page 49 Page 50

Measuring Inventory Performance New Vehicle Inventories in the US

Quantity of Inventory [units]
DOI
Days of Inventory (DOI) = Average
g Demand [units/day]
[ / y] 200
Chrysler

180
Ford
Stock Turns is the number of times an organisation replaces 160
GM
its stocks during a period (usually measured annually) Honda
140

Cost of Goods Sold in Period [£] 120

Mazda

Stock Turns = Mitsubishi

Average Inventory Valuation [£] 100
Nissan

80

ƒ Typical stock turns: 5 to 20

20, world
world-class
class lean manufacturers achieve Toyota
y

>40.
60
Volkswagen

ƒ Company A had a starting WIP of £1.75m, and a closing WIP of 40 BMW

£1 2
£1.25m at the
h end d off the
h year. T
Totall sales
l iin the
h year were £36m.
£36 20
Hyundai

Calculate stock turns. Average

0
6

Ja 6
M 7

Se 7

Ja 7
M 8

Se 8

Ja 8
M 9

Se 9

Ja 9
M 0

Se 0

Ja 0
M 1

Se 1

Ja 1
M 2

Se 2

Ja 2
M 3

Se 3

Ja 3
M 4

Se 4

Ja 4
M 5

Se 5
05
-9

-9

9
-9

9
-9

9
-9

0
-0

0
-0

0
-0

0
-0

0
-0

0
-0
p-

n-

p-

n-

p-

n-

p-

n-

p-

n-

p-

n-

p-

n-

p-

n-

p-

n-

p-
ay

ay

ay

ay

ay

ay

ay

ay

ay

ay
n
Ja

Se
M

Page 51 Page 52
Part Classification Pareto, ABC & the ’80-20 Rule’
All inventory control models are part-specific % of Impact
(Value x Volume)
ƒ Attention given to part depends on cost impact
100%
ABC Classification, H Ford Dickie, 1951 95%

ƒ Number of parts versus Value x Volume 80%

ƒ A: 20% of parts = 80% of cost (unit cost x quantity)
ƒ B:
B 30% off parts
t = 15% off costt
ƒ C: 50% of parts = 5% of cost C
(exact percentages differ from one author to another) B

Pareto’s Law or Analysis, the ‘80-20 Rule’ A

ƒ Vilfredo Pareto (1848-1923), study of income in Italy in 1897

c.20%
0 c.50% % of SKU
SKU’ss

Page 53 Page 54

ABC Classification: Impact Runners, Repeaters, Strangers

Try to ensure regularity in operations
• A-parts: watch closely, minimise stock, aim for flow Classify parts by order frequency:

ƒ Runners: high demand, aim to make continuously, use JIT.

• B-parts: review ordering policy from time to time, observe
ƒ Repeaters:
R t repetitive
titi ddemand,
d b
butt d
does nott jjustify
tif continuous
ti
production. Try to produce regularly, even if quantity varies.
• C-parts: automate replenishment,
replenishment use reorder point as a
ƒ Strangers: spare parts etc., rarely ordered. Make to order, as
trigger demand does not justify stock holding.

Phil Crosby: “ballet, not hockey”

Page 55 Page 56
 Basic Approaches to Ordering
• 1. Fixed Order Quantity Models
ƒ Economic Order Quantity (EOQ)
ƒ Re-order Point
• 2.
2 Fixed Period Requirements
ƒ Fixed Period Ordering
ƒ Lot-for-Lot ordering
g
ƒ Period Order Quantity
Ordering and • 3. Variable order q
quantity
y and ordering
g interval
B t h Sizing
Batch Si i ƒ Least Unit Cost
Decisions
ec s o s ƒ Least Total Cost
ƒ Part-Period Balancing
• 4. Material Requirements Planning (MRP)
ƒ Calculates time-phased requirements
Page 58

Fixed Order Quantity Systems (s,Q) Fixed Period Ordering Systems (R,S)
Fixed Order point
nventorry Levell

S
Order

evel
Quantity
=Q

ntory Le
Q
In

s ROP
O Inven
Time 0 R R R
0

• Order Quantity is variable, ordering time interval R is fixed

• (s,Q): Order time interval is variable, order quantity Q is fixed • Order up to level S
• Every R time periods
periods, system orders to replenish to S
but only if stock currently below s
Page 59 Page 60
‘Lot-for-Lot’ Ordering Batch Sizing: Determination of Q when ordering

• Also called ‘pass-on-orders’, or ‘order-up-to’ model (OUT)

• Using
g a large
g order size ((i.e. ordering
g infrequently):
q y)
• Simply passes on customer orders to the supplier as they we suffer a large inventory holding cost.
come in, without interference
• Using a small order size (i
(i.e.
e ordering frequently):
• One only orders from the supplier what is demanded by we suffer a large fixed cost of ordering
the customer ƒ Clerical / labour cost of p
processing
g an order
ƒ Any fixed costs imposed by supplier
• No fixed order quantity, but fixed time intervals (each
ƒ Inspection and return of poor quality products
period)
ƒ T
Transport t costs
t
• Optimal solution for inventory cost! ƒ Handling costs
ƒ Labour cost of organising transportation
• ...but ordering cost an issue?
• Where to find the balance?

Page 61 Page 62

Economic Order Quantity Total Cost Formula

The order quantity that minimizes the total cost per period There are two parts to the total cost per period
ƒ The holding cost depends on average stock: Q/2
ƒ Derived by F.W. Harris, manufacturing engineer with ƒ Ordering cost depends on number of orders per period: D/Q
Westinghouse Corp
Corp., in 1913
This gives the total cost per period formula, as a function of
ƒ Rediscovered and applied by management consultant the batch size ordered (Q)
R.H. Wilson in 1934,, thus often called Wilson-Harris
lot size formula Q D
T (Q) = C H + CO
2 Q
D = annual demand
Q = batch (lot) size
CO = (fixed) cost of placing one order
CH = cost of holding one item in store for one period

Page 63 Page 64
 EOQ basic trade-off model Economic Order Quantity
Cost per
period Ordering
O d i
Slope = 0
Total CO Cost
EOQ = 2 D
Minimum
CH Holding Cost
Holding cost,
cost CH
Annual Demand

Ordering cost, CO (Fixed) Cost per order placed = CO [£]

Cost per unit to hold one item for one period = CH [£]
Optimal order Order quantity Q Demand rate per period = D [units/time]
Order quantity = Q [units]
Length of Order Cycle = (Q/D) [time]

Page 65 Page 66

Why didn’t we include the Example - EOQ

variable (unit) cost? A retailer expects to sell about 200 units of a product p.a. The storage
space taken up in his premises by one unit of this product is valued at
Shouldn’t the formula for total cost per period be: £20 per year. Interest rates are expected to remain close to 10% per
year and one unit is bought
y g at £100.
Q D
T (Q) = C H + CO + DCV 1. If the cost associated with ordering is £35 per order, what is the
2 Q economic order quantity?
q y
2. For administrative convenience, we can only order in minimum order
where CV = variable cost (cost per item)? quantities of multiples of 10. What is the total cost in case of ordering 20,
and 30?
Yes, though here it makes no difference to the optimal Q, 3. How does the EOQ change if we assume a more realistic 20%, or even
so we often ignore the term DCV 40% iinventory
t h
holding
ldi cost,
t iin addition
dditi tto storage?
t ?

Page 67 Page 68
EOQ considering volume discounts Period Order Quantity (POQ)
EOQ logic, modified so that we order to cover demand for a
whole number of periods,
periods while still minimising cost
Example:
ƒ D = 200 units per year and EOQ = 58 units, with “period” equal to
one month
ƒ EOQ/D = 58 units / 200 units = 0
0.29
29 years between orders
ƒ 0.29 years = 3.48 months, so order every 3 months to cover
p
expected demand in the next 3 months

Same logic as EOQ, except that ordering interval is

computed not ordering quantity
computed, quantity.
Also known as Economic Time Cycle

Page 69 Page 70

Batch Sizing: Determining Q in Production Economic Production Quantity (EPQ)

• What if we are producing the batch ourselves, rather than Assume a constant production rate of R>D for each batch
ordering it in from an external supplier? Inventory

• Most of the issues are the same as they were when we Q(1-D/R)
Slope= -D
were ordering the batch in. The differences are: Slope=R-D

((1)) The cost of ordering g becomes the cost of setup p

ƒ Clerical / labour cost of setting up a machine t
Q/D
ƒ Loss of production while set-up takes place
ƒ Return of poor quality products after start-up Analysis is as before
before, except CH is replaced by CH(1 – D/R)

(2) The batch does not now arrive instantaneously 2 D CS

EPQ =
• The optimum batch size is known as the Economic C H (1 − D / R )
Production Quantity (EPQ)

Page 71 Page 72
Problems with EOQ and EPQ Variable Order Quantity and Ordering Interval
Rigid Assumptions! Methods that allow lot size & ordering interval to vary from period to period
1. Demand is constant and steady, and continues indefinitely ƒ We still assume that demand is known, even if it is not constant
2
2. EOQ assumes whole h l replenishment
l i h t llott arrives
i att same titime
ƒ Seek to cover demand for a whole number of periods (why?)
3. Replenishment lead-time is known
4. Order size is not constrained by supplier, no min/max restrictions ƒ As in EOQ, objective of minimising the sum of setup and inventory cost
5. Holding cost per item per period is a constant
6. Cost of ordering/setup varies linearly with replenishment size LUC - Least Unit Cost
7. Item is independent of others; benefits from joint reviews are ignored ƒ See next slide
8. Doesn’t encourage us to decrease fixed ordering/setup costs
LTC - Least Total Cost
Problem Cost Accuracy:
ƒ Consider seeking
g to cover demand for next 1,2,3...
, , periods
p (as
( LUC))
ƒ How much does a set-up or placing an order cost? ƒ Choose n to most closely balance set-up and inventory cost for this batch
ƒ Holding costs: often calculated at interest level (cost of capital)
PPB - Part-Period Balancing
Conclusion
ƒ Basic version as LTC, but advanced versions include ‘look-ahead / look-back’
ƒ Simple model, often reasonable, though with limitations
facility to see if simple modifications to schedule reduce total costs.

Page 73 Page 74

Least Unit Cost Exercise LUC

• Heuristic (Greek: ‘find’): ‘quick and dirty’, or ‘sub-optimal’
method Period 1 2 3 4 5 6 7 8

• Basic idea: Consider seeking to cover demand for next Requirements 25 30 0 50 0 65 35 35

1,2,3... p
periods. Find cost/unit for each case. Stopp jjust
before this starts to rise. Restart calculation from there. End of Period Stock

1. Cover demand Batch Cost Cost /Unit

• Assume we suffer holding cost on only items held over for 1
from one period to the next for 1,2
for 1,2,3
for 1,2,3,4
• Example 2. Cover demand Batch Cost Cost /Unit
ƒ Set-up Cost £100, inventory holding cost = £1 / period / item for 4
for 4,5
for 4
4,5,6
56
for 4,5,6,7
Period 1 2 3 4 5 6 7 8 3. Cover demand Batch Cost Cost /Unit
for 7
for 7,8
Requirements 25 30 0 50 0 65 35 35
Page 75 Page 76
 Objectives for Today
Judge Business School • Why forecasting is needed
• Different types of demand fluctuations
• Moving averages (MA)
Session 3 • Exponential smoothing (ES)
Forecasting
• Advantages and disadvantages of MA & ES
Assembly Line Balancing
• Other forecasting methods
• Assembly line balancing

Page 78

Why do we need to forecast? Why do we need to forecast?

Example I Example II
• You are the plant manager at Manufacturing Excellence • You are the store manager at one of AllFoodYouNeed
Ltd., a local producer of metal parts for the automotive Plc.’s stores. Recently, you have noticed that you lose
industry You are currently running at over 100% of your
industry. money on selling freshly squeezed orange juice due to a
capacity, causing a significant increase in unit costs. high percentage of unsold juice that cannot be kept
However, you can still not satisfy all of your customers’ overnight.
orders.
• You have made the following observations over the past
p
• Expandingg yyour p
production capacity
p y would take 6 months two weeks:
and cost about £2 million. Week 1 2
Day Mo Tu We Th Fr Sa Su Mo Tu We Th Fr Sa Su
• Should you make the investment? Juice produced (litres) 100 100 100 100 100 100 100 100 100 100 100 100 100 100
Juice unsold (litres) 19 27 38 27 18 7 5 17 28 36 29 17 6 4
Demand (litres) 81 73 62 73 82 93 95 83 72 64 71 83 94 96

• What improvements would you suggest?

Page 79 Page 80
Two Forms of Demand Forecasting Methods - Overview
1. Qualitative (long-term, 2-10 years)
• Independent demand
ƒ Market Surveys
y
ƒ finished products ƒ Delphi Study: ask the experts...
ƒ based on market demand ƒ Problems: bias, ignorance
ƒ requires forecasting
2. Quantitative (short to medium term, 0-2 years)
• Dependent demand A. Extrinsic (based on external patterns beyond firm level)
ƒ parts that go into the finished products ƒ Regression,
R i correlation
l ti andd econometrics,
ti medium
di tterm (1
(1-2
2 years))
ƒ Problem: will miss unusual events and short term issues
ƒ dependent demand is a known function of independent demand
ƒ no forecasting is required B IIntrinsic
B. t i i (based
(b d on patterns
tt off actual
t l data
d t att firm
fi level)
l l)
ƒ Short term (up to 12 months)
• Can products have partially dependent and independent ƒ Moving average, exponential smoothing (extrapolation methods)
demand at the same time? ƒ Time
Ti series,
i Decomposition/Fourier
D iti /F i A Analysis
l i
ƒ Problem: almost exclusively based on historical data!

Page 81 Page 82

Patterns of Demand Forecast via Moving Average

• Most basic approach; we assume mean demand constant
35 A step change Trend
• To forecast demand in next period we average demands

25 Constant in the recent past periods

uantity

mean
demand • A 4-period moving average is the average of the last 4
15
Qu

ti
time periods
i d
5
• The general formula for moving average is:
-5
-10 0 10 20 30 40 St = ( x t + x t -1 + .. + x t -n +1 ) / n
Time Random noise
NB: by convention, St is an average based on data up to time
t, but used as a forecast for time t+1

Page 83 Page 84
Exercise 4-Period Moving Average Forecasting Exercise 4-Period Moving Average Forecasting
Time Demand Forecast Time Demand Forecast
-3 504 -3 504
-2 484 -2 484
-1 493 -1 493
0 423 0 423
1 458 1 458
2 440 2 440
3 485 3 485
4 395 4 395
5 368 5 368
6 344 6 344

Page 85 Page 86

Age of Data and Weights Exponential Smoothing

The more data periods are used, the older the data gets: • Like a moving average, but weight given to each period is
ƒ 1 period MA: average age of data is 1 period a fixed
fi d proportion
ti off weight
i ht given
i tto succeeding
di period
i d
ƒ 2 period MA: average age is (1+2)/2 = 1.5 periods
ƒ
ƒ
3 period MA: average age is (1+2+3)/3 = 2 periods
...
{ }
S t = k ⋅ xt + (1 − α ) ⋅ xt −1 + (1 − α ) 2 ⋅ xt − 2 + (1 − α ) 3 ⋅ xt −3 + ...
ƒ np period MA: average
g age
g is ((1+2+...+n)/n=(n+1)/2
) ( ) pperiods
• 0 < α < 1,
1 so weights get smaller as we recede into the past

• In a MA forecast, each period has a weight of 1/n k k(1-α)

( ) )2
k(1-α)
( )3 ...
k(1-α)
(
• Perhaps the more recent past should have greater
weight? • The constant k is chosen so that the weights sum to 1:

k + k(1-α) + k(1-α)2 + k(1-α)3 + ... = 1

and this (it can be shown) requires k=α

Page 87 Page 88
Exponential Smoothing Forecast Exponential Smoothing: Updating Formula
• Exponential smoothing forecasts contain information on all previous
demands, each demand is given a weight that is decreasing • Forecast for period (t+1) can be quickly calculated from
exponentially back in time. the forecast for period t:
• g constant: 0 < α < 1
Smoothing
St = α xt + α (1 − α ) xt-1 + α (1 − α ) 2 xt-2 + α (1 − α )3 xt-3 …
• The general formula for exponential smoothing is:
S8 = α x8 + α (1 − α ) x7 + α (1 − α ) 2 x6 + α (1 − α )3 x5 …

St = α ⋅ xt + α ⋅ (1 − α ) xt −1 + α ⋅ (1 − α ) 2 xt − 2 + α ⋅ (1 − α ) 3 xt −3 + ... S9 = α x9 + α (1 − α ) x8 + α (1 − α ) 2 x7 + α (1 − α ) 3 x6 …
= α x9 + (1 − α ) [α x8 + α (1 − α ) x7 + α (1 − α ) 2 x6 + … ]
St is based on all (available) data up to period t to forecast xt+1
= α x9 + (1 − α ) S8

St = α xt + (1 − α ) St −1
Alternative notation:
St = St −1 +αε t ε t = xt − St −1
Page 89 Page 90

Exercise Exponential Smoothing Forecasting Exercise Exponential Smoothing Forecasting

Time Demand Forecast α=0.1 Forecast α=0.9 Time Demand Forecast α=0.1 Forecast α=0.9

0 504 0 504
1 484 1 484
2 493 2 493
3 423 3 423
4 458 4 458
5 440 5 440
6 485 6 485
7 395 7 395
8 368 8 368
9 344 9 344
Page 91 Page 92
Exponential Smoothing: α = ? ES: Step change in demand
Choosing α: depends on how rapidly we want the smoothed
value to respond to changes in demand The forecast “locks
“locks-on”
on” to the demand (that’s good)
ƒ Usually α=0.1 to 0.3
ƒ The larger α, the more responsive (nervous) is the forecast
12
1.2
ƒ The smaller α, the less reactive and more stable the forecast
1

S t = α x t + α (1 − α ) x t-1 + α (1 − α ) x t-2 + α (1 − α ) x 3 ..
2 3
08
0.8

Quantity
y
0.6
Demand (Input)
Steady
• α=0.1 10% 9% 8.1% 7.3% 0.4 state
Weights 0.2
Simple or constant (Type 1) ES

• α=0.9 90% 9% 0.9% 0.09%

0
-10 0 10 20 30

It can be shown that the average age of data = 1/α Time

Page 93 Page 94

ES: “ramp” (linear trend) demand Summary of Simple ES

ƒ The method we have seen so far is known as simple exponential
The forecast does not “lock-on” (that’s not so good) smoothing

35 ƒ It copes OK with step changes in demand

Demand (Input)
30 ƒ It does not cope well with linear trends
25 Simple or constant (Type 1) ES
ƒ An adaptation of simple exponential smoothing can cope with
Quantiity

20
linear trends: double exponential smoothing (sometimes known as
15
Type 2 exponential smoothing)
10
5
0
-10 0 10 20 30

Time

Page 95 Page 96
Definition of Double ES Approaches to Forecasting
Input, xt (Demand) • One approach is to extrapolate past data (MA, ES)
((S1)t ((S2)t
ES1 ES2 ƒ This works fine for minor variability
ƒ Problem: response to step changes not immediate, forecast may
2 + be late in reacting to trends
-1
1
ALPHA = 0.3
t Demand S1 S2 2*S1 - S2 • Another approach is to learn about the underlying
0
1
0
0
0.00
0 00
0.00
0.00
0 00
0.00
0.00
0 00
0.00
properties of the demand time series
2 1 0.00 0.00 0.00
3 2 0.30 0.00 0.60 Output, yt ƒ E.g. demand for turkeys goes up every Christmas
4 3 0.81 0.09 1.53
5
6
4
5
1.47
2.23
0.31
0 3
0.65
2.63
63
3.80
(Forecast) • How to analyse:
7 6 3.06 1.13 4.99
8 7 3.94 1.71 6.18 ƒ Regression analysis: what factors matter? E.g. weather and
9 8 4.86 2.38 7.34 sporting
p g events to p
predict beer consumption
p
10 9 5 80
5.80 3 12
3.12 8 48
8.48
11 10 6.76 3.93 9.60 ƒ Decomposition/Fourier analysis: the idea is that one complex
12 11 7.73 4.78 10.69
13 12 8.71 5.66 11.76 demand time series can be decomposed into a set of simpler
14 13 9.70 6.58 12.82 series
15 14 10.69 7.51 13.86
16 15 11.68 8.47 14.90

Page 97 Page 98

Decomposition Models Fourier Analysis Example:

New Vehicle Orders
Demand
d

Trend with Fourier Analysis for Demand Pattern of Model A

seasonal pattern 600
D

Original Data
500
Fourier Transform

400 Weekly Mov. Avg.

Time

nits per day

300
• Additive Model:
Demand = (Trend) + (Seasonal) + (Cyclic) + (Randomness) Un 200

• Multiplicative Model: 100

Demand = (Trend) * (Seasonal) * (Cyclic) * (Randomness)

0
• Cyclic component is like the seasonal component, but with a longer cycle

115
118
100
103
106
109
112
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
97
1
4
7

period: simple models ignore it

-100
Days

Page 99 Page 100

Forecast Accuracy Forecast Error
ƒ The forecast is always wrong! • We define et = xt - St-1 as the forecast error in period t

ƒ The longer the forecast horizon, the worse the forecast • Let e1, e2, …, en be the forecast errors observed over
n periods
ƒ The less aggregated,
aggregated the worse the forecast
• A measure of forecast error over n periods is the
Mean Absolute Deviation (MAD):
n
MAD = 1 ∑ | e |
n
i =1 i
• Another measure of forecast error is the Mean
q
Squared Error (MSE):
( )
n
MSE = 1 ∑ e2
n i =1
=1 i
Page 101 Page 102

Traditional Assembly Line

Assembly
Line Balancing
g

Page 104
 Assembly Line
• An assembly line consists of a sequence of operations
• Some operations modify a single part or item
• Other operations assemble two or more parts together

Page 105 Page 106

Operations and Stations Remember our solution from week 1:

• Crucial question: how many people do we need to do
Work Work Work
the work? Cycle Time Capacity Capacity
station A station B station C (min/unit) (Output / Utilisation
hour) (%)
• Naïve answer: n operations need n people
1 worker
worker, 1 worker
worker, 1 worker
worker,
5 12 67%
• Problem: 3 mins 5 mins 2 mins

ƒ Some operations may take less time than others, so some 2

1 worker, 1 worker,
workers 3 20 83%
operators will be idle part of their time 3 mins
2.5 mins
2 mins

ƒ We might not need to produce items at such a high rate 1.2

2 workers, 0.8 workers
workers 2.5 24 100%
2.5 mins 2.5 mins
• Thus,
Thus one person can cover more than one operation 2 5 mins
2.5

Station 1 Station 2 Station 3 “Load balancing”

Operation 1 Operation 3 Operation 5

Operation 2 Operation 4
ÆEfficiency is generally driven by capacity utilisation and
• Hence we group operations into stations labour productivity...
Page 107 Page 108
Balancing the Assembly Line Balancing the Assembly Line
ƒ Ideally, the time required for stations to carry out their ƒ We ask each station to complete its work within a fixed
operations should not vary much from one station to time, known as the cycle time
another, otherwise: ƒ Then we allocate operations to stations, ensuring that:
ƒ some stations will be partly idle ƒ each station can complete its operations with the cycle time
ƒ and/or
d/ some stations
t ti will
ill b
be overloaded
l d d ƒ there is a similar amount of idle time at each station
ƒ NB: the cycle time is also the time between successive
ƒ This is called “balancing”
balancing the assembly line it
items reaching
hi th the end
d off th
the assembly
bl line
li
ƒ Hence by reducing the cycle time, we increase the
production rate
ƒ “Takt time” = adjusting the cycle time needed to meet
customer demand

Page 109 Page 110

Precedence Constraints Example

t=0.6
• Some operations cannot be done before others t=0.3
#1 #3
• Arrow means that one operation has to be completed t=0.5
before another can start #6
#4
#8
Length of time needed for operation t=1.0 t=0.7
t=0.6 #2
t=0.3 #5 #7 #9
#
#3 t=0 7
t=0.7
#1 t=0.4 t=0.6 t=0.8
t=0.5
#6 • Suppose we want to produce 40 items/hour
Opn number #4
#8
inside circle • 60 min per hour / 40 items per hour = 1.5 min cycle time
t=1.0 t=0.7
• Work content = sum of all operation (processing) times
#2 #5 #7 #9 = 5.6 minutes
t=0.7
t=0.4 tt=0
0.6
6 tt=0.8
0.8 • Minimum number of stations 5.6min / 1.5min = 3.73 (i.e. 4 stations)

Page 111 Page 112

A possible allocation of operations Evaluating our allocation
t=0.6 t=0.6
t=0.3 t=0.3
#1 #3 #1 #3
t=0.5 t=0.5
#4 #6 #4 #6
#8 #8
t=1.0 t=0.7 t=1.0 t=0.7
#2 #5 #7 #9 #2 #5 #7 #9
t=0 7
t=0.7 t=0 7
t=0.7
t=0.4 t=0.6 t=0.8 t=0.4 t=0.6 t=0.8

• Station 1: #1,#2,#3 = 1.3 mins • We define:

Total processing time per item
• Station 2: #4,#6
, = 1.5 mins (is
( this OK?)) Balancing
g Loss = 1 -
Number of stations * Cycle time
• Station 3: #5,#7 = 1.4 mins
• Station 4: #8,#9 = 1.4 mins • Here, balancing loss =

Page 113 Page 114

Heuristic Methods for Assembly Line Balancing Longest Sequential Chain of Followers
• Operation: #1 (4 followers), #2,#3,#4 (3 followers),
• Small problems can be solved “by eye”; #5,#6 (2 followers), #7 (1 follower), #8,#9 (0 followers)
l
larger ones need
dhheuristic
i ti methods
th d • Allocate jobs with more followers first
• Break ties byy taking
g jjobs with longest
g p
processing
g times
• In allocating operations to stations
stations, allocate ‘most
most first, thus Op# 4 (1.0) before #2 (0.4) before #3 (0.3)…
deserving’ operation first, followed by second ‘most
Order of allocation:
#1 (0.6) #4 (1.0) #5 (0.6) #8 (0.7)
deserving’, etc. #1 to Station 1
#4 to Station 2
#2 (0.4)
(0 4) #6 (0.5)
(0 5) #7 (0.8)
(0 8) #9 (0.7)
(0 7) (no room on Stn.1)
• Examples of heuristics: #2 to Station 1
#3 to Station 1
1 Longest sequential chain of followers
1. #3 (0.3)
(0 3) #5 to Station 3
(no room on Stn. 2)
2. Total number of followers #6 to Station 2
1.3 1.5 1.4 1.4 etc
3 Ranked
3. R k d positional
iti l weights
i ht (not
( t partt off syllabus
ll b now))

Page 115 Page 116

Total Number of Followers The good, the bad, the ugly...
• Operation: #1 (7 followers), #2,#3,#4, (4 followers), #5,#6 Material flow
(3 followers), #7 (2 followers), #8, #9 (0 followers) Material flow

Better: Operators can trade elements

Bad: Operators caged. No chance
• Take longest first to trade elements of work
of work. Can add and subtract
operators. Trained workers will
between them. self balance
self-balance.

#1 (0.6)
(0 6) #4 (1.0)
(1 0) #5 (0.6)
(0 6) #8 (0.7)
(0 7) Better: Operators can help
Material each other. Might
increase output with
#2 (0.4)
(0 4) #6 (0.5)
(0 5) #7 (0.8)
(0 8) #9 (0.7)
(0 7) third operator.
operator
Bad: Operators birdcaged. No
chance to increase input
with a third operator.
#3 (0.3)
(0 3) Best: Effective operator access.

Bad: Straight line difficult to balance.

1.3 1.5 1.4 1.4

Page 117 Page 118

Manufacturing Cell
Judge Business School

Session 4
Machine level Scheduling
Machine-level

Page 119
Objectives for Today Scheduling
• What is scheduling? • Deciding in what order and when particular operations
should be carried out
• Scheduling machines in a job shop environment
• For projects:
ƒ Minimising
g average
g completion/flow
p time
ƒ Scheduling is essentially project management
ƒ Minimising maximum lateness ƒ Critical Path Analysis (CPA)
ƒ Minimise number of late jobs
ƒ Mi i i average ttardiness
Minimise di • For batch and flow production:
ƒ Minimise makespan in a 2-machine flowshop ƒ Standard routing, but some variety. Computer-planned schedules
(MRP / MRPII / ERP), or visual tools (Just-in-Time / Lean)
ƒ Cyclic scheduling, capacity planning key (bottlenecks)
• For job shops (our topic today)
ƒ No standard routing, no standard lead-times
ƒ Multiple machines, identical or different, parallel or in-line, etc...

Page 121 Page 122

Job Shop - Flow Chart Types of Problems in Job Shop

• In a job shop situation, different jobs have different
A
Annealing
li St
Stamping
i P i ti
Painting routings
ti and
d diff
differentt processing
i titimes
• Analysis of complex cases is often based on concepts
from simple cases such as just one machine, or two
Assembly sequential machines, or two parallel machines

M/C 1
Washing Welding Brazing
M/C 1 M/C 1 M/C 2
M/C 2
Process split into independent work
centres complex routing and scheduling
centres, scheduling.
Page 123 Page 124
Scheduling Objectives Minimising Average Completion Time (1 m/c)
The objectives we might face in scheduling are: Job A B C D
M/C 1
ƒ Minimise average
g completion/flow
p time of a set of jjobs Processing Times 11 3 4 2
ƒ Minimise the maximum lateness
ƒ Minimise the number of late jobs • If we do jobs in alphabetical order, and start at t=0, the schedule is:

ƒ Minimise average lateness/tardiness A11B14C18D20 NB useful notation: JobCompletion time

ƒ Minimise “makespan” (time between first job starting and last job • The overall completion time (20) is does not depend on the schedule, but we
finishing)
g) might want to minimise average completion time.
• For our schedule, average completion time = (11+14+18+20)/4 = 15.75
For each objective, an algorithm or heuristic exists to help
j
us achieve our objective • We can do better: put jobs with Shortest Processing Times earlier:
D2B5C9A20
ƒ Algorithm: gives us the optimal solution
• For new schedule, average completion time = (2+5+9+20)/4 = 9
ƒ Heuristic: gives us a reasonable, but not necessarily optimal
solution
l ti • SPT Rule: minimises average completion/flow time on one machine
• Flow time = time that a job spends in the system

Page 125 Page 126

Minimising Maximum Lateness (1 m/c) Minimising the number of late jobs (1 m/c)
Job A B C D Job A B C D E F G H I
M/C 1 M/C 1
Processing Time 3 3 4 5 Processing Time 2 7 3 8 4 4 6 8 5
Due Date 9 8 16 9 Due Date 5 10 15 22 23 24 25 30 33

• If we do jobs in alphabetical order

order, the schedule is: • F this
For thi problem,
bl it iis impossible
i ibl to
t avoid
id some jjobs
b bbeing
i llate
t
A3B6C10D15 • We might want to minimise the number of late jobs (egg to minimise
• For this schedule,
schedule maximum lateness (job D) = (15-9) = 6 the cost of subcontracting them)
• For this problem, it is impossible to avoid some job being late • Moore’s Algorithm:
1.
1 Schedule jobs by EDD
• W might
We i ht wantt to
t minimise
i i i th
the maximum
i llateness:
t tto d
do thi
this, putt th
the
jobs in order of Earliest Due Date: 2. If no job is late, go to step 6
3. Find the first late job (call it the kth job)
B3A6D11C15
4
4. F
From amongstt jobs
j b 1 tto k,
k remove the th job
j b with
ith th
the llongestt processing
i
• For new schedule, maximum lateness (job D) = (11-9) = 2 time
5. Return to step 1 with one fewer job to consider
• EDD Rule: minimises maximum lateness on one machine
6 The schedule is the EDD schedule, plus removed jobs (in
6. ( any order))

Page 127 Page 128

Exercise Moore’s Algorithm I Exercise Moore’s Algorithm II
Job A B C D E F G H I Job A B C D E F G H I
M/C 1 M/C 1
Processing Time 2 7 3 8 4 4 6 8 5 Processing Time 2 7 3 8 4 4 6 8 5
Due Date 5 10 15 22 23 24 25 30 33 Due Date 5 10 15 22 23 24 25 30 33

• Jobs already listed in EDD order

order, so start with this • New
N EDD order
d iis:
schedule
J b
Job J b
Job
Processing Time Processing Time
Completion
l Time Completion
l Time

Due Date Due Date

Page 129 Page 130

Exercise Moore’s Algorithm III Exercise Moore’s Algorithm IV

Job A B C D E F G H I Job A B C D E F G H I
M/C 1 M/C 1
Processing Time 2 7 3 8 4 4 6 8 5 Processing Time 2 7 3 8 4 4 6 8 5
Due Date 5 10 15 22 23 24 25 30 33 Due Date 5 10 15 22 23 24 25 30 33

• New
N EDD order
d iis: • Final schedule is:

J b
Job Job

Processing Time Processing Time

Completion
l Time Completion Time

Due Date Due Date

• Moore’s Algorithm: minimises the number of late jobs on

one machine

Page 131 Page 132

Minimising Average Tardiness I Minimising Average Tardiness II
M/C 1 M/C 1

• Observation:
Ob ti minimising
i i i i average llateness
t iis a d
daft
ft id
idea!! • W
We could
ld ttry EDD
EDD; if th
there iis only
l one llate
t jjob,
b EDD
minimises the average tardiness
• Lateness = (completion date) – (due date)
hence it could be negative • But for more than one late job, there is no algorithm for
this problem
• But in real life, there is often little/no benefit from finishing
a job early • However, there is a heuristic: the Modified Due Date
(MDD) Rule: put jobs in order of increasing MDD
• Instead,, define Tardiness of a job
j = maximum of 0 and
the Lateness of the job (we write: max(0,Lateness)) • At time t, MDD = the maximum of the due date and the
earliest time at which we can complete a job:
• Minimising average tardiness is a meaningful objective
i.e. MDDi = max(di,t+pi)
• How do we achieve it? where di = due date, pi = processing time for job i

Page 133 Page 134

Exercise MDD I Exercise MDD II M/C 1

M/C 1
• At t=0, Modified Due Date = Due Date for each job j:

Job A B C D E F G H I Job A B C D E F G H I

Processing Times 2 7 3 8 4 4 6 8 5 Processing Times 2 7 3 8 4 4 6 8 5

Due Dates 5 10 15 22 23 24 25 30 33 Due Dates 5 10 15 22 23 24 25 30 33

Modified Due Dates

Here, ... has the smallest MDD (...), and completes at t= ...

Page 135 Page 136

Exercise MDD III Exercise MDD IV
M/C 1 M/C 1
• t= • t=

Job A B C D E F G H I Job A B C D E F G H I

Processing Times 2 7 3 8 4 4 6 8 5 Processing Times 2 7 3 8 4 4 6 8 5

Due Dates 5 10 15 22 23 24 25 30 33 Due Dates 5 10 15 22 23 24 25 30 33

Modified Due Dates Modified Due Dates

Here, ... has the smallest MDD (...), and completes at t= ... Here, ... has the smallest MDD (...), and completes at t= ...

Page 137 Page 138

Exercise MDD V Exercise MDD VI

M/C 1 M/C 1
• t= • t=

Job A B C D E F G H I Job A B C D E F G H I

Processing Times 2 7 3 8 4 4 6 8 5 Processing Times 2 7 3 8 4 4 6 8 5

Due Dates 5 10 15 22 23 24 25 30 33 Due Dates 5 10 15 22 23 24 25 30 33

Modified Due Dates Modified Due Dates

Here, ... has the smallest MDD (...), and completes at t= ... Here, ... has the smallest MDD (...), and completes at t= ...

Page 139 Page 140

Exercise MDD VII Exercise MDD VIII
M/C 1 M/C 1
• t= • t=

Job A B C D E F G H I Job A B C D E F G H I

Processing Times 2 7 3 8 4 4 6 8 5 Processing Times 2 7 3 8 4 4 6 8 5

Due Dates 5 10 15 22 23 24 25 30 33 Due Dates 5 10 15 22 23 24 25 30 33

Modified Due Dates Modified Due Dates

Here, ... has the smallest MDD (...), and completes at t= ... Here, ... has the smallest MDD (...), and completes at t= ...

Page 141 Page 142

Exercise MDD IX Exercise MDD X

M/C 1 • t= M/C 1
• t=

Job A B C D E F G H I
Job A B C D E F G H I
Processing Times 2 7 3 8 4 4 6 8 5
Processing Times 2 7 3 8 4 4 6 8 5
Due Dates 5 10 15 22 23 24 25 30 33
Due Dates 5 10 15 22 23 24 25 30 33
Modified Due Dates
Modified Due Dates
Final job is ..., which completes at t=...
Here, ... has the smallest MDD (...), and completes at t= ... How to measure the quality of my schedule?
Final Schedule:
Original Due Dates:
Total tardiness is , the average tardiness is

Page 143 Page 144

Job shop scheduling with more than Try ABCDE….
one machine
hi M/C 1 M/C 2 Jobs A B C D E M/C 1 M/C 2

Processing Time - m/c 1 6 7 4 6 2

• Forget due dates
Processing Time - m/c 2 5 6 6 3 4
• Makespan: time between first job starting and last job finishing
• Flowshop: every job visits the same machines in the same order
M/C 1 A B C D E
• p in a 2-machine flowshop
Problem: Minimise makespan p 0 6 13 17 23 25

M/C 2 A B C D E
Jobs A B C D E 6 11 13 19 25 28 32

Processing Time - m/c 1 6 7 4 6 2

Processing Time - m/c 2 5 6 6 3 4
Î Makespan = 32. Can we do better?

Page 145 Page 146

Minimising Makespan Johnson’s Rule

Observations M/C 1 M/C 2 M/C 1 M/C 2

• The makespan on m/c 1 is fixed

1 From
1. F the
th jobs
j b nott yett assigned,
i d fifind
d th
the jjob
b with
ith shortest
h t t
• Thus, to minimise overall makespan, we need to minimise makespan processing time on either m/c.
on m/c 2
• Two ways to do that
2. Which machine does that processing time occur on?
ƒ Minimise amount of time m/c 2 continues to p
process beyond
y m/c 1
ƒ If that
th t shortest
h t t processingi time
ti occurs on m/c
/ 1,
1 assign
i the
th job
j b tto
ƒ Minimise amount of time m/c 2 needs to wait before starting to process
the next free slot in the schedule
jobs
• C
Conclusion:
l i
ƒ If that shortest processing time occurs on m/c 2, assign the job to
ƒ Last jobs in schedule should have short processing times on m/c 2 the last free slot in the schedule.
ƒ First jobs in the schedule should have short processing times on m/c 1

3. Go back to Step 1 until all the jobs are assigned.

Page 147 Page 148

Exercise Johnson’s Rule I Exercise Johnson’s Rule II
Jobs A B C D E M/C 1 M/C 2 Jobs A B C D E M/C 1 M/C 2

Processing Time - m/c 1 6 7 4 6 2 Processing Time - m/c 1 6 7 4 6 2

Processing Time - m/c 2 5 6 6 3 4 Processing Time - m/c 2 5 6 6 3 4

ƒ Job on m/c M/C 1

ƒ Job on m/c
ƒ Job on m/c
ƒ Job on m/c
ƒ Job → between jobs already allocated M/C 2
This yields the schedule:
Makespan =

Johnson’s Rule: minimises makespan in a two machine

fl
flowshop
h
Page 149 Page 150

Summary of Scheduling Rules

• SPT Rule: minimises average completion/flow time on one machine
Judge Business School
• EDD Rule: minimises maximum lateness on one machine
• Moore
Moore’s
s Algorithm: minimises the number of late jobs on one machine
• MDD Rule: (heuristic) minimises average tardiness on one machine Session 5
• Johnson
Johnson’s
s Rule: minimises makespan in a two machine flowshop Factory level Scheduling
Factory-level

• Flow time = time that a job spends in the system

• Lateness = (completion date) – (due date)
• Tardiness = max(0, lateness)
• Makespan = time between first job starting and last job finishing

Page 151
 Marketing, Design, Manufacture

Order Fulfilment
Strategies

Page 153

Conflicting Objectives How to deal with Customer Orders?

Conflicting objectives:
Order Fulfilment Process
ƒ 100% utilisation of resources Supply ƒ From ‘order
order to delivery
delivery’ (OTD)
ƒ Zero stock ƒ Key process, day-to-day challenge!
ƒ Any volume and product mix
ƒ No lead time Demand ƒ Determines customer service and majority of cost

This requires co-ordination within and between firms aiming at: Order fulfilment strategy
ƒ Balanced flow of work - all tasks take the same time ƒ How far back does the order go into the supply chain?
ƒ Steady demand for products - and hence for inputs
ƒ Which is appropriate under which circumstances?
Ideal demand is smooth and predictable:
ƒ Total demand = maximum output capacity of resources Two basic factors are important
ƒ Any changes are perfectly forecast in sufficient time to allow capacity
change ƒ Production Lead-time P: How long does it take to make the
But…. product?
ƒ Real demand is usually not predictable ƒ Demand Lead-time D: How long is the customer willing to wait for
ƒ Demand has peaks - lunchtime/Saturday/summer etc the product?
g p
ƒ Demand varies through product life cycle
y & competition
p

Page 155
The P:D Ratio P:D Ratio
Incoming
g P
Customer D Engineer-to-Order
Orders
Customer Lead-time D Design
P
D Make-to-Order
Components
Start Production Process Lead-time P Finish P
D Assemble-to-Order
Components
P D
Make-to-Stock

FGI
Hal Mather, 1988, ‘Competitive Manufacturing’, Prentice Hall Time
Supply Chain Factory Marketplace
p

P:D ratio MTS, ATO, MTO, ETO

Make-to-Stock (MTS/BTS) / Make-to-Forecast (MTF/BTF)
ƒ Goods made to be placed in stock prior to receiving an order. Typical of
commodities and continuous processing.
processing
Engineer-to-Order
ƒ Efficient production, but risk of obsolescence / high stock cost
PULL
Assemble-to-Order
sse b e to O de (ATO)
( O)
ƒ Producers hold components stock to assemble an order as required by the
customer. Typical of line/ batch production.
Make-to-Order
ƒ Responsive
R i ffor customised
t i d products,
d t b butt still
till costt off stock
t k

Make-to-Order / Build-to-Order (MTO/BTO)

ƒ Material ordered and product or service made only after the buyers order is
PUSH Assemble-to-Order
received. Line and job shop production.
ƒ No FGI cost, but potentially less efficient production.

Engineer to order (ETO)

Make-to-Stock ƒ Product designed & built to customer order. Typical of projects.

Page 159
 Order Fulfilment in Electronics Sector - Mass Customization
E
Example:
l DELL
..an oxymoron?
Sales Order
Component Assembly 1-2hrs The need to ‘customize’ mass-produced goods to
Warehouse 25min
5 customer needs
An umbrella concept
Suppliers
pp Logistics
Customer
Centre ƒ Build-to-Order
Long ƒ Assemble-to-Order
Transportation 3-Day Order Fulfilment
within UK;
ƒ L
Late C
Configuration
fi i
Lead Time
“Merge-in Transit” ƒ Customisation at point of use
distribution ƒ C t i ti th
Customisation through
h service
i

Production Planning and Control

Basic problem: how to convert customer orders into:
ƒ A production schedule for the multiple process stages on the shop-
floor (work orders)
ƒ A supplier schedule for the required materials and components
(purchase orders)
ƒ Ad
delivery
li schedule
h d l at the
h customer iinterface
f (d
(delivery
li promise)
i )

Production Planning
g
T
Two basic
b i approaches:
h PUSH and
d PULL scheduling
h d li
and Control

Page 163 Page 164

 A Simplified Manufacturing System
External
influences;
System boundary Taxation etc
Taxation, etc.

R
Raw material
t i l Planning &
suppliers Purchasing Scheduling

Material Requirements
q Planning
g
Goods In/ Component Final Customers
Receiving
g Manufacture Assembly
y FGI

Component WIP WIP

(MRP)
Raw
suppliers Materials
M t i l Manufacturing
system

Elements outside El
Elements
t in
i the
th system
t
the system

Page 165 Page 166

“Push” Scheduling MRP – Material Requirements Planning

Central Planning MRP system were invented in 1960s to cope with computational
and
d Scheduling
S h d li complexity of scheduling
ƒ Need to plan at top level-item, for all items made in the plant
ƒ What happens if components are used in several final products?
Order Order Order
– Computerised inventory control & production planning system

Process Process – Schedules component items and processes when needed - no earlier
Suppliers and no later
Raw #1 #2
Materials WIP Definitions:
‘MRP
MRP is a set of priority planning techniques for planning component
– In a PUSH system, the orders are planned and issued centrally items below the product or end item level’
– Upon completion, the order is moved forward, until the next process is ‘A set of techniques which uses the bill of materials (BOMs), inventory
iissued
d with
ith th
the order
d tto start
t t processing
i it data and the master production schedule to calculate future
requirements for materials. It essentially makes recommendations to
– Hence, the LAST process sees the new order FIRST release material to the production system.’
– This is called BACKWARD scheduling

Page 167 Page 168

New Names, but mostly same Logic... MRP: 3 Inputs, 3 Outputs
MRP / MRP I - Material Requirements Planning
ƒ Typical
yp systems:
y BPICS
ƒ OUTDATED! No capacity feedback loop Master Production Schedule

Æ ‘Closed loop’ MRP

Rough
R h
MRP II - Manufacturing Resource Planning Material Cut
Product Inventory Capacity
ƒ Typical systems: Manugistics, i2, Peoplesoft Requirements
Structure Master Planning
Pl
Planning
i
DRP - Distribution Resource Planning Tree File (RCCP)
ERP- Enterprise Resource Planning
ƒ updates MRP II with relational DBMS, GUI & client/server architecture Planned Order Releases
ƒ Typical systems: SAP R/3, mySAP, Baan
APS – Advanced Planning and Scheduling
Work Orders Purchase Orders Rescheduling Notices
ƒ Often add-ons to support ERP, short-term planning and scheduling
ƒ Typical systems: i2 Rhythm

Page 169 Page 170

1. Master Production Schedule 1. Master Production Schedule

Drives MRP process with a schedule of finished products
Outlines production schedule for top-level items only MPS Period
Item 1 2 3 4 5 6 7 8
Quantities represent production
production, not demand!
Clipboard 86 93 119 100 100 100 100 100
Quantities may consist of a combination of customer orders Lapboard 0 50 0 50 0 50 0 50
& demand
d d fforecasts
t Lapdesk 75 120 47 20 17 10 0 0
Quantities represent what needs to be produced, not what Pencil Case 125 125 125 125 125 125 125 125
can be produced
Should be checked byy Rough-Cut-Capacity
g p y Planning
g
(RCCP) routine

Page 171 Page 172

 2. Product Structure Tree 2. Bill of Materials
or: Bill-of-Materials (BOM) LEVEL ITEM Unit of Measure Quantity
0---- Clipboard Ea 1
Clipboard Level 0 -1--- Clip Assembly Ea 1
--2-- Top Clip Ea 1
---3- Sheet Metal In2 8
Clip Assembly Rivet Board
(1) (2) (1) Level 1 --2-- Bottom Clip Ea 1
---3- Sheet Metal In2 8
--2-- Pivot Ea 1
Top Clip Bottom Clip Pivot Spring Pressboard Finish ---3- Iron Rod In 3
(1) (1) (1) (1) (1) (2oz )
(2oz.) Level 2
--2-- Spring Ea 1
---3- Spring Steel In 10
Sheet Sheet Spring Iron
-1--- Rivet Ea 2
Metal Metal Steel Rod
(8 in2) (8 in2) (10 in.) (3 in.) -1--- Board Ea 1
Level 3
--2-- Press Board Ea 1
--2-- Finish Oz 2
Page 173
Page 174

3. Inventory Master File MRP Order Action Logic

Description Inventory Policy Determine net requirements by
It
Item Board
B d L d ti
Lead time 2
Item no. 7341 On-hand stock 53
Item type Manuf
Manuf. Holding cost 1 Adding allocations
Product/sales class Ass’y Ordering/setup cost 50
Value class B y stock
Safety 25
to gross requirements
Buyer/planner RSR Reorder point 39 minus scheduled receipts
Vendor/drawing 07142 EOQ 316
Phantom code N Minimum order qty 100
minus on hand balance
on-hand
Unit price/cost 1.25 Maximum order qty 500 equals net requirements
Pegging Y Multiple order qty 100
LLC 1 Annual demand 5,000

Page 175 Page 176

The MRP Gross to Net Logic Problems with MRP Systems
• MRP Systems tend to hold more inventory than
Projected necessary
Future Available Inventory
Demands Resources Status • MRP Systems
y are designed
g for batch & q
queue, not for
flow
Gross On hand • MRP Systems lengthen lead times unnecessarily
Req’ments balance

PLUS MINUS PLUS EQUALS Net • MRP Systems inherently distort demand patterns in the
Req’ments
R ’ t supply chain
Scheduled
Allocations
Receipts

Page 178

MRP is...
A good database
A good transaction processor
Essential source of information and almost impossible to replace in many
p
companies!

MRP is not..
A good scheduler
Why? Just-in-Time
– Assumes infinite capacity & Lean Production
– Works on fixed batches
– Works on fixed lead times
– Schedules backwards,, therefore cannot synchronise
y

Page 179 Page 180

 “Pull” Scheduling The Automatic Loom of Sakichi Toyoda

Central Planning Order

and
d Scheduling
S h d li

Raw Process Process

Suppliers WIP
Materials #1 #2

– In a PULL system, processes are triggered by a replenishment signal

– Upon withdrawal of material from inventory, the preceding process is
authorised to start processing
processing, and ONLY THEN!
– Hence, the FIRST process sees the new order FIRST
– This is called FORWARD scheduling
– Simplest form: two-bin approach
Page 181 Page 182

Toyota Production System (TPS) Just-in-Time Scheduling

Taiichi Ohno starts developing TPS in engine shop in 1948
The idea of JIT merges:
ƒ Pillars of TPS are ‘autonomation’
autonomation (jidouka),
(jidouka) based on Sakichi
Sakichi’s s loom,
loom and
ƒ The supermarket concept of “on-the-shelf inventory”
Just-in-Time (JIT), which came from Kiichiro Toyoda.
ƒ The two-bin replenishment system
ƒ Toyoda Kiichiro: “..in a comprehensive industry such as automobile
manufacturing,
f t i th best
the b t way to t work k would
ld be
b tot have
h allll the
th parts
t for
f Production orders are based on replenishment
assembly at the side of the line just in time for their user”. Ohno 1997.
The demand signal is conveyed via kanban cards
Principles of TPS:
ƒ Type 1: Production kanban
ƒ Only make what is required, when it is required, just in time ƒ Type 2: Withdrawal kanban
ƒ Use small batches The system is tightly controlled,
controlled as it is “fragile”
fragile , i.e.
i e unable
ƒ Reduce seven wastes (overproduction, inventory, transportation, to cope with large swings in volume or product mix
motion, inappropriate processing, defects, waiting/delay)
Need to keep schedule variability within 5-10%
Toyota overcame obstacles of economically producing a high variety of
products in small batches Just-in-Time is central part of the Toyota Production System
(TPS) also referred to as “Lean
(TPS), Lean Production
Production”
Supported by tightly synchronised supplier network at close proximity

Page 184 Page 185

A two kanban system A Typical Kanban Sequence
#6: Th
#6 The number
b off Triangle
Ti l
Kanbans accumulates and
eventually reaches the
target batch size.

#7 : When the target batch line

is reached, the kanbans for the #5: The PRODUCTION #3: The PRODUCTION card
Customer Kanban Manufacturing Product are placed in the press Card (Triangle Kanban) is detached, and MOVE card
Process Supermarket Process queue (file).
(file) is sent to the is attached to the container
BATCH BOARD.
Pull Pull #2: A MOVE card is
sent to the p
post press
p
Batch Board supermarket

#1: The cell

Withdrawn New #9: The Raw Material
uses a container
t i
of product
Product Production Kanban is detached Press
and moved to the
Kanban Post.
Post

#10: The triangle

#8: When the file reaches (PRODUCTION kanban) is
the beginning of the queue
queue, placed in the finished #4: The Container with
the product is made. goods container, which the MOVE card is returned
is moved to the super- to the cell.
Page 186
market. The cycle is complete.

Page 188
Traditional vs.. JIT Inventory TPS Elements - Taiichi Ohno
All processes driven to be in control and capable
ƒ Standardised work practices
p
Traditional ƒ “Simplify, highlight deviations, mistake-proof”
Inventory Level
Problems are natural and are opportunities to learn, not to blame!
ƒ Most problems arise from not following standards
ƒ Every problem has root cause and counter-measures
Every activity must add value
Average
Inventory ƒ Eliminate waste
JIT Make what customers want when they want it, just-in-time
ƒ Smooth production “pulse”
Select and invest in people
0 Time
ƒ Managers chosen as best teachers/problem-solvers
ƒ Empowerment & multi
multi-skilling
skilling

Page 190 Page 191

TOC - Key Concepts

– The main concept is FLOW - the importance of
constraints and bottlenecks!
– What is a bottleneck?
– Idea: bottlenecks control the performance of a system:
“An hour lost at a bottleneck is an hour lost to the system”
(Goldratt)
– Only three measures matter: Throughput, Operating
Expense, Inventory
Theory
y of Constraints – Focus improvement on the bottlenecks,
bottlenecks allow the
bottleneck to control the activity of the system
– A busy ope
operation
at o iss NOT
O necessarily
ecessa y a p
productive
oduct e oone
e

Æ Balance flow, not capacity!

Page 192 Page 193

Drum, Buffer + Rope Scheduling
How to schedule operations to maximise throughput? Judge Business School
ƒ Protect throughput at bottleneck with buffer
ƒ Bottleneck becomes pacemaker for entire process

Session 6
Process
#1 Toyota Production
Bottle- Process

I
Neck #3 System and Lean Thinking
R
Resource
Process
#2 ROPE BUFFER DRUM

Page 194

Two Types of Improvement

1. ‘Kaikaku’ or ‘Kaizen-Blitz’
ƒ Step change
ƒ Short-term execution
ƒ 2-3 day workshops - implement changes on the spot!
ƒ Limited / narrow scope (e
(e.g.
g one machine or area on shop-floor)
2. ‘Kaizen’ or Continuous Improvement (CI)
ƒ Literally “Changing
Changing something for the better”
better by eliminating waste
ƒ Long-term activities

Improvement
p ƒ
ƒ
Gradual / incremental improvements
N lilimit
No it iin ffocus, often
ft extends
t d tto suppliers
li and
d di
distribution
t ib ti
Kaizen Office: dedicated team leading improvement activities
Æ Kaizen and Kaikaku only work in conjunction!

Page 196 Page 197

 Radical vs. Incremental Improvement
Standardise and
“Continuous”
nce
maintain
Perrforman

i
improvementt
Improvement

Actual
improvement

“Breakthrough
Improvements” Actual
Lean Thinking
g
improvement

Time

Page 198 Page 199

Where does the term ‘LEAN’ come from? The NUMMI Experience
Joint Venture GM/Toyota plant, closed 1982, reopened in 1984
ƒ MIT International Motor Vehicle Program
Dramatic improvements compared to previous GM plant
ƒ Started in 1979, ongoing to date! ƒ Assembly hours/car 36 to 19

ƒ 5 Year - 5 Million Dollar Program ƒ Assembly defects/car 1.5 to 0.5

ƒ Worker absenteeism 15% to 1.5%
ƒ Coordinated by MIT – Researchers from around the World
Changes:
ƒ Research
R h iinto
t allll aspects
t off the
th Toyota
T t Production
P d ti System
S t ƒ Toyota
T management – Lean
L P
Production
d i principles
i i l
ƒ Work organization – teams, few job classifications, focus on quality,
ƒ Developed assembly plant methodology ƒ Still unionized workforce

ƒ Comparative research – the International Assembly Plant Study Transfer to other GM plants took many years
ƒ GM management lacked commitment – embarrassment for Roger Smith
ƒ Policy forums for senior industry
industry, government and union officials
ƒ NUMMI plant visits for management were for brief time periods only
ƒ ‘Lean Production’ coined by John Krafcik (1989) ƒ Visiting teams below critical size
Æ b t NUMMI showed
Æ..but: h d that
th t llean was nott culturally
lt ll bound!
b d!

Page 200 Page 201

The “DNA” of TPS Waste – the central Focus of Lean!
Should Toyota be afraid of copycats?
Waste is anything which does not add value to a
ƒ It iis possible
ibl tto copy shop-floor
h fl techniques
t h i product or service in any office or manufacturing
ƒ Some improvement in productivity and quality will result activity!
ƒ Lack of flexibility in adjusting to change will persist
ƒ System will not able to learn autonomously ..essentially everything the customer is not
ƒ Continuous improvement needs to be driven by prepared to pay for!
workforce, cannot be dictated by management
Frequently used term: “muda” (=waste)
Dynamic learning capability is key advantage: ƒ Muda (waste)
Lean is a mindset
mindset, a system -- not a ‘toolbox’!
toolbox ! ƒ Muri (excessive strain)
ƒ Mura (unevenness or irregularity)

Page 202 Page 203

Ohno’s 7 Wastes Think about a Petrol Station…

Raw
Materials
£
Overproduction
Transporting
£ £
£ £
£ £ Unnecessary
£ £ £ £
£ Motion
£ £
Defects/ £ £ Waiting
Rework £ £ ££
£ ££ £ £ £ £
£ £
£ £ £ £ £
£ £
£ £ £ £
Inappropriate £ £ £
£ £ £££
Processing
g £ £ ££
£ £
£ £
Despatch
£
£ Unnecessary
£ £
£ I
Inventory
t
Page 204 Page 205
Does this add more Value? Create no value but are Cost Elements
required by product
development, order filling
Formula 1: or production system

100 litres of fuel As perceived

by the customer
Value
12 litres/second
lit / d Type 1 Added NNVA
NNVA Work
Value
4 new tyres Added
Type
yp 2 NVA Work
NVA
Clear ‘windscreen’
…in
i 77-10
10 seconds!
d ! Those actions that
don’t add value -
eliminate right away
Total cost of operation
- Condition under continuous
Total cost of operation improvement -
- Actual condition -
Page 206 Page 207

Five Lean Principles Features of a Lean Organisation

1. Specify what creates value from the So what does it look like if these principles are put
customers perspective into practice?
2. Identify all steps across the whole Features to look for:
value stream
ƒ The learning organisation: mistakes are opportunities!
3 M
3. Make
k th
those actions
ti that
th t create
t value
l ƒ Visual management: targets and achievements
flow ƒ Continuous Improvement activities
4. Only make what is pulled by the ƒ Standard operating procedures (SOPs)
customer just-in-time ƒ Empowerment
ƒ Multi-skilling
5. Strive for perfection by continually
removing successive layers of waste ƒ Teamwork

Page 208 Page 209

The “House of Lean”

Policy Deployment
Demand Smoothing
Lean
Value
Stream
Pro
Pro-
JIT Jidouka
motion
Organi-
sation
Cases:
Office
Production Smoothing Lean Retailing
5S / Shop Floor Teams / 7 Q Tools / PDCA
Lean Services

Page 210 Page 211

Exercise 7 Wastes Case I: Lean Retailing at Tesco

Identify
y examples
p of the 7 Wastes in a p
posh Tesco: Lean Road Map
restaurant, from a customer’s point of view! ƒ How to make products flow right to the shelf?
ƒ Making products flow requires involvement of partners
in the value chain

Are there additional wastes? Aim:

ƒ To ensure that the pproduct flowed from the supplier
pp
through the distribution centre to store with minimal
delays,
y so giving
g g true “one touch replenishment”.

Page 212 Page 213

Before Merchandisable Units After Merchandisable Units

Page 214 Page 215

Case III: Fujitsu The Seven Service Wastes

1. Delay on the part of customers waiting for service, for delivery in queues, for
Fujitsu service operations (call centre) response, in queues, not arriving as promised. The customer’s time is not
ƒ Efficiency
Efficiency-driven,
driven, standard times and op/procedures free to the provider!
ƒ Main measures: calls per man day, av. call handling time 2. Duplication. Having to re-enter data, repeat details on forms, copy
information across, answer queries from several sources within the same
Æ Is this “lean”? g
organisation
Main idea: 3. Unnecessary movement. Queuing several times, lack of “one-stop
shopping”, poor ergonomics in the service encounter
ƒ Use custo
customere feedback
eedbac to e eradicate
ad cate root
oot cause p
problem!
ob e
ƒ Prevent errors from recurring- “failure demand” 4. Unclear communication, and the wastes of seeking clarification, confusion
over product or service use, wasting time finding a location that may result in
ƒ 40-90% of call volume was preventable
misuse or duplication
Outcomes: 5. Incorrect inventory. Being out-of-stock, unable to get exactly what was
ƒ Demand down by 60%, cost reduction of 64%, productivity increased required to the customer
by 45%
6. Opportunity lost to retain or win customers, failure to establish rapport,
ƒ Customer satisfaction up 28%, employee satisfaction up 40% ignoring customers, unfriendliness, rudeness
ƒ £200m new business generated
7. Errors in the service transaction,, product
p defects in the product-service
p
bundle, lost or damaged goods
Source: Bicheno (2004) “The Lean Toolbox”
Page 217 Page 218
 Sid’s Heroes – Case Study
Sid’s Heroes was a BBC series on manufacturingg
improvement based on lean techniques in the late 1980s

Six companies were visited by Sid, three-day

three day workshops were held in
each company

Most were very successful…

successful

1. What operational improvements been made, and will these

improvements be sustainable?
Implementation
2. What are the root causes for the failure,, and who is to blame?
of Improvement programs
3. Could this failure have been prevented, and if so, how?

Page 219 Page 220

Successful Implementation
Judge Business School
1 Senior management buy
1. buy-in
in and support is a critical
enabler
ƒ Rewards and responsibilities often need to be changed
2. Long-term support & consistent objectives are key Session 7
ƒ Beware of “Fad
Fad of the Month
Month” Problem Service
3. Ownership of improvements needs to reside at process Quality and Six Sigma
level
ƒ Beware of how consulting resources are used Project Management
4 Measures need to be aligned to top
4. top-level
level goal
ƒ “What you get is what you measure”!
ƒ The “Bottom-up”
Bottom up myth

Page 221
 Service Process Matrix
Low
Degree of customisation High

Service Factory Service Shop

Costs associated with ‘Customised mass

Low
facility & equipment services’.

ct
er contac
e.g.. Airlines, Hotels, e.g.. Car repair,
Fast food ‘Routine’ hospitals

Custome
Service Operations
p
Mass Service Professional
Customers treated as Service

C
similar Costs principally for

High
e.g.. Retailing, ‘elite’ labour.
banking e.g.. Lawyers,
accountants

Source: Service Operations Management, Roger Schmenner, Prentice Hall

Page 223

Demand and Revenue Management

What are the key differences between

Manufacturing g and Service Operations?
p

What are the implications for

Operations Management?

Page 226
2
Page 225
Demand & Revenue Management Revenue Management: How does it work?
Problem: finite capacity, uncertain demand
Price
1. Segmentation
500 ƒ Some customers value time
ƒ Some customers value money
400
2 Differentiation
2.
P)

BA: LHR
LHR-HAM
HAM
Airfare (in US$ or GBP

300 ƒ Change demand by adjusting price 350

Ryan: STN-
LBC
ƒ Different price for each customer-time combo
200
3. Adaptation 200
JetBlue: JFK-
OAK ƒ Simplification needed 150
100 125
ƒ S
Stochastic problem Æ deterministic LP 100
AA: JFK-OAK
0 ƒ Continuous re-calculation

y (am)
y (pm)
5 days
3 days

Departure Date
weeks

weeks

weeks

weeks

weeks

weeks

weeks

weeks

weeks
week
Demand

1w
9w

8w

7w

6w

5w

4w

3w

2w

2w

1 day
1 day
ÆIt might make sense to sell below cost at times,
if price paid > (variable cost for empty seat+
g
marginal cost for full seat))
ÆMax profit might not occur at full load!
Page 227 Page 228

Distinguish Order Winners from Order Qualifiers The Kano Model I - Product Features
Order Qualifiers: hygiene factor, needed to be considered by customer
Must be (Basics): characteristics or features taken for
Order Winners: distinguishing factors that drive customer choice
granted. (Hotel: clean sheets & hot water)
Order
winning More is better (Performance): we are disappointed if a
criteria need is poorly met, but have increasing satisfaction the
better it is met.
met (Hotel : response time for room service).
service)
Number of orders

Delighter: features that surprise and delight in a

Order positive way (Hotel: wine and flowers upon arrival)
qualifying
criteria
Reversal (Tontini): features that annoy (TV in a smart
restaurant)
P d t specification
Product ifi ti - features
f t

Page 229 Page 230

 The Kano Model Apply the Kano Model..
..to a one-week skiing holiday package!
D li h
Delight
What are the Must Be factors?

More is
What are the Performance factors?
Delighters
Customer better
satisfaction Neutral What are the Delighters?
Must be
What are the Reversal factors?
Reversal
(T ti i)
(Tontini)
Dissatisfaction
Absent Fulfilled
Presence
of characteristic

Page 231 Page 232

Quality Failures in the news

The Challenger Space Shuttle disaster (1986)
ƒ Ignoring “O ring” problems

The Japan-BNFL scandal

ƒ £100m compensation for faked quality data (size of pellets)

The Paris Concorde disaster (2000)

Quality ƒ Poor maintenance on main gear, overload, poor wing tank protection

The Bridgestone - Firestone / Ford Defender debacle (2000)

ƒ Tires decomposing at high speeds, killing 121 people in US
ƒ Recall alone cost £7.5bn
“It
It costs five to seven times as M
Mercedes
d A A-Class
Cl “El
“Elchtest”
ht t” (1997)
much to get a new customer as it
ƒ Retrofitting of ESP systems!
does to satisfy and keep one.”
(TARP research; Struebing, 1996)

Page 233 Page 234

History of Quality Philosophies Deming

1. Inspect & Reject

(SPC to reduce cost)

2. Quality control: aim at Zero “In God We Trust …

Defects: SPC in effect makes a
certain number of defects inevitable
- all others must bring data”
3. Quality
y assurance: Use q
quality
y
standards to ensure documentation
of a ‘quality process’

4. Total Quality Management: standards

are self limiting, unresponsive and resented
so make quality a “moral” issue for everyone
W Edwards Deming

Page 235 Page 236

Deming – Father of the Quality Movement PDCA: The Deming Cycle

Taught Statistical Process Control (SPC)
Insisted that top management attended his courses (what to do
Plan p )
for improvement)
Thesis: natural variation is inherent in all processes
processes. It is the (Implement
(I l t
task of management to understand and control the causes of more widely;
undue variation Standardise)

Management is responsible for about 85% of quality Act

problems, shop floor employees only control 15% Do (..it, on a trial
basis;
Emphasis on continuous improvement “Hold the gains” experiment)
Check
Proposed PDCA Wheel (if it works; the risks; the
variation, LEARN!)

Page 237 Page 238

Joseph Juran’s Trilogy Cost of Quality (after Juran)
Appraisal Costs
Juran’s
Juran s Trilogy ƒ To assure outgoing & incoming quality

ƒ Quality Planning ƒ Appraisal activities to detect nonconforming items

ƒ Quality
Q lit C Control
t l ƒ Acceptance sampling
sampling, inspection
inspection, final testing

ƒ Quality Improvement Prevention Costs

ƒ Prevent rework
rework, scrap,
scrap and other failures
ƒ Activities include process control, preventive maintenance, most
Quality “Hold the gains” Quality ISO 9000 activities, training
Control Planning
Failure Costs
ƒ Internal failure: scrap, rework, rectification, retest, and including
opportunity costs
Quality
“Pareto: ƒ External failure: warranty, returns, customer dissatisfaction,
“Breakthrough” Improvement focus on the vital few”
customer defection

Page 239 Page 240

Six Sigma: Normal distribution

Six Sigma
g Source:
www stat yale edu
www.stat.yale.edu

Normal distribution
µ: mean
σ: standard deviation

Page 241 Page 242

Where are those six sigma’s? History of Six Sigma
Source: www.six-
6σ Goal declared by Motorola in 1987
sigma training org
sigma-training-org ƒ At the time it was operating at 3.8σ = defects 10,000ppm

Key Steps
ƒ Statistical tools used to analyse processes
ƒ Reached quality plateau of 5
5.2σ
2σ (108 ppm) in 1 year and reduced
throughput time by 90% in many processes
ƒ Dedicated ‘Black
Black Belt’
Belt resources in 1991
ƒ A formal discipline and structured approach developed

Outcome: achieved 5.83σ (7 ppm) in three years

Page 243 Page 244

Six Sigma background Sources of variability

1. Supply uncertainty
Core Principles:
ƒ Materials and component quality
– Everything
E thi iis a process ƒ Supplier delivery reliability
– Every process has ƒ Raw material prices
variation ƒ Shifts in technology
2. Throughput uncertainty
– Every process can be
measured ƒ M hi b
Machine breakdown
kd
ƒ Defects and rework
– Every process can be
improved and variation 3 Demand uncertainty
3.
reduced ƒ Variability in product use & product purchase
ƒ Seasonality and trends in demand
– The target is 3.4 defects ƒ Fashion & product life cycle effects
per million opportunities in ƒ Competitor action
customer output
Source: adapted from Davis 1993
Page 245 Page 246
The Six Sigma Approach The Inspection Exercise I
• Six Sigma aims to define the causes of defects, measure
th
those defects,
d f t and d analyze
l th
them so th
thatt they
th can be
b
reduced.
• During the Measure Phase, the overall performance of the
core business process is measured.
• A Six Sigma defect is defined as anything outside of
customer specifications
• This is called an “opportunity” (to do better, essentially…)
• A Six Sigma opportunity is the total quantity of chances for
a defect

Page 247 Page 248

The Inspection Exercise II How to calculate a sigma level?

Step 1: Calculate the DPMO
First we calculate Defects Per Million Opportunities (DPMO) and based on that a Sigma is
decided from a predefined table (see handout):

Number of defects observed

DPMO = -------------------------------------------------- x 1,000,000
Number of Units
Where:
• Number for defects is total number of defects found;
• Number of units is the number of units produced ;

Step
p 2: Covert DPMO into a Sigma
g Level
• Use the conversion table

Cave: A 1.5
1 5 Sigma process walk is not considered here,
here but is commonly factored
into the calculation!
Page 250
 Yield % Sigma Defects Per Million
Opportunities

1.5σ Process walk

99.9997 6.00 3.4
99.9995
99.9992
99.9990
99 9980
99.9980
5.92
5.81
5.76
5 61
5.61
5
8
10
20
Yield to
99.9970
99.9960
99.9930
99.9900
5.51
5.44
5.31
5.22
30
40
70
100
Sigma
99.9850
99.9770
99.9670
99.9520
5.12
5.00
4.91
4.80
150
230
330
480
Conversion
99.9320
99.9040
99.8650
99.8140
4.70
4.60
4.50
4.40
680
960
1,350
1,860
Table
99 7450
99.7450 4 30
4.30 2 550
2,550
99.6540 4.20 3,460
99.5340 4.10 4,660
99.3790 4.00 6,210
99.1810 3.90 8,190
98.9300 3.80 10,700
98.6100 3.70 13,900
98 2200
98.2200 3 60
3.60 17 800
17,800
97.7300 3.50 22,700
97.1300 3.40 28,700
96.4100 3.30 35,900
95.5400 3.20 44,600
94.5200 3.10 54,800
93.3200 3.00 66,800
91.9200 2.90 80,800
90.3200 2.80 96,800
88.5000 2.70 115,000
86.5000 2.60 135,000
84.2000 2.50 158,000
81.6000 2.40 184,000
78 8000
78.8000
75.8000
2 30
2.30
2.20
212 000
212,000
242,000
LSL: Lower Specification Limit
72.6000
69.2000
2.10
2.00
274,000
308,000 USL: Upper Specification Limit
65.6000 1.90 344,000
61.8000 1.80 382,000
58.0000 1.70 420,000
54 0000
54.0000
50.0000
1 60
1.60
1.50
460 000
460,000
500,000
Source: Integrated Enterprise Excellence Volume III
46.0000 1.40 540,000
43.0000 1.32 570,000
39.0000 1.22 610,000
Page 251 35.0000 1.11 650,000 Page 252
31.0000 1.00 690,000

The Magnitude of Difference... How to use Six Sigma

1. A statistical measure of products, processes, and service
Sigmas % Good Spelling Time Distance PPM excellence.
31% 170 spelling 31.75 years per From here to ƒ Using “sigma”-levels allows for the comparison across
1s 632,120
errors per page century the moon industries, processes and services
25 spelling 4.5 years per 1.5 laps around 2
2. A business goal for achieving operational excellence:
2s 69.1% errors per page century the world
308,537
ƒ Improve process performance
One trip from
3s 93 3%
93.3%
1.5 spelling errors
per page
3.5 months per
North to South 66 803
66,803 ƒ Reduce cycle times
century
Brazil
ƒ Reduce defects
1 spelling error in 2.5 days per 45-minute drive
4s 99.4% each 30 pages century on a highway
6,210 3. A disciplined, data-driven problem solving technique:
1 spelling error A short drive to ƒ Customer-focused
30 minutes per
5s 99.98% in one the closest gas 233
encyclopedia
century
station ƒ Team-based
1 spelling error in
6 seconds per 4 steps in any ƒ Results-oriented
6s 99.9997% all books in one
century direction
3.4
small library

Process Six Sigma

Sigma, Design for Six Sigma
Sigma, New Six Sigma

Page 253 Page 254

 Six Sigma – DMAIC Technique
Problem Definition Focus
A Identify Project CTBs Practical
Define B Develop Team Charter
D fi
Define C Define Process Map Problem
What is the frequency of defects?
1 Select CTB Characteristic Y
Statistical
aracterize

Measure 2 Define Performance Standards Y

3 Validate Measurement System Y Problem
4 Establish Process Capability Y

When, where and why do defects occur?

Statistical
Cha

Analyze 5 Define Performance Objective X’s

6 Identify Sources of Variation X’s Solution
How can we improve the process?
7 Screen Potential Causes Vital few X’s
Xs
Practical
P ti l
Lean,, Six Sigma,
g , TOC:
Improve
When to use which approach?
8 Discover Variable Relationships Vital few X’s
9 Establish Operating Tolerances Vital few X’s Solution
Optimise

How can we maintain the improvement?

10 Validate Measurement System Vital few X’s Guaranteed
G d
Control 11 Determine Process Capability Vital few X’s
Life
12 Implement Process Controls Vital few X’s

Page 255 Page 256

Lean, Strategic vs. Operational Perspective

6 Si
Sigma,
TOC Lean
Thinking
Strategic Level:
5 Principles Understand Value

Quality L
Lean Variability
TQM, TQC Production 6σ, SPC
Operational
Responsiveness Level Availability Level (Tools):
Agile, Postponement Scheduling, TPM Eliminate Waste
Kanban,
Capacity Takt Time,
Prod. Control
Prod
Drum-Buffer-Rope, etc.
etc
MRPI+II, ERP, APS
Source: Dave Nave, TOC
Quality Progress 2002
Source: Hines, Holweg, Rich (2004) “Learning to Evolve,
A Review of Contemporary Lean Thinking”, IJOPM
Page 257 Page 258
 Project Management &
New Product Development (NPD)
• The
Th kkey distinction
di ti ti tto manufacturing
f t i and d service
i iis n=1
1!
• Project Management
ƒ There can be a fixed deadline (2012 Olympics)
ƒ There can be a fixed budget (public procurement)
ƒ There
Th generally
ll is
i a sett off constraints
t i t that
th t need
d to
t be
b balanced
b l d
• New Product Development
Project
j Management
g
ƒ The process from idea, concept, design engineering, prototyping,
testing, to product launch (“Job zero”)
ƒ New product development is a critical capability, no more or no
less important than manufacturing, or marketing & sales
• From an operational point of view, NPD is a special case
of project management
Page 259 Page 260

Critical Path Method: “Time is Money” The Steps in CPM

Step 1. A Network diagram is used to show precedence relationships among activities that
• Key question in project management: how to schedule activities in the
comprise project: Activities appear on arcs (“Activity on arc”); arcs define precedence.
project to meet the given objectives of time, budget, or both?
Step 2
2. Four pieces of information required for each activity:
• The method that does this is called the “Critical Path Method”, ƒ Normal time: time to complete activity under normal conditions
developed by Lockyer in 1978. ƒ Normal cost: cost of activity if completed in normal time
• I simple
In i l tterms, the
th critical
iti l path
th is
i the
th least
l t amountt off time
ti ƒ Crash time: minimum time in which activity can be completed
needed for the whole project to complete. ƒ Crash cost: cost of activity if completed in minimum time
Step 3. Project network is assumed to use normal times and costs for all activities.
• However, most activities can be expedited at extra cost (called
However
“crash” time, and “crash” cost) through overtime, outsourcing, etc. Step 4. If resulting project completion time is satisfactory, all activities are scheduled at
normal completion times.
• CPM helps p determine whether ƒ If completion
p time is too long,
g, p
project
j can be completed
p in less time at g
greater cost by
y
‘crashing’ (reducing the activity time of) some activities:
or not this is beneficial!
Crash cost – normal cost
Crash cost /unit time =
Normal time – crash time

Note: In general, do not crash an activity as far as possible, only as far as necessary.
Step 5
5. Cost of project is given by total normal cost of all activities plus the cost
incurred by crashing

Page 261
Exercise Addenbrooke’s Hospital I Exercise Addenbrooke’s Hospital II
• The Head of Department at Haematology wants to The Department Head knows how long each activity takes, its cost, and
establish a new diagnostic service for Leukaemia patients. the cost and extent that expediting
p g is p
possible:
(a) Select operators normally takes 3 weeks at a cost of £4000, but could
• The project budget is £40,000, it needs to be in place be reduced to as little as 1 week, if he is willing to expend £8000.
within six weeks, and the procedure requires a new piece
of equipment to conduct the test. (b) Procure equipment takes 2 weeks at a cost of £5000. He cannot
obtain the equipment more quickly, at any price.
• The project has the initial steps: (c) Complete EU paperwork takes 6 weeks at a cost of £10,000, which
ƒ Select operators for the new equipment can be expedited by hiring more clerks, who can get this down to 4
ƒ Procure the new piece of equipment weeks,, if he is willing
g to pay
p y £14,000.
,
ƒ Complete auditing paperwork on Medical Equipment Purchases (d) Train operators normally takes 4 weeks, at a cost of £8000, but can
be done in 2 weeks, but then the cost is £13,000.
• Further steps include
ƒ Train the operators (e) Test equipment normally takes 3 weeks and costs £6000, but can be
done in 1 week, if he will pay £14,000 instead.
q p
ƒ Test the equipment for safety
y

Page 263 Page 264

Exercise Addenbrooke’s Hospital III Exercise Addenbrooke’s Hospital IV

• The normal project time is the longest path through the network, from Start to
End, called the critical path. Here, the critical path is ......, with length .....
weeks.
• The normal project cost is the sum of the normal costs for all activities. Here,
the normal activity costs sum to £..........
• The project completion time can be reduced only by crashing an activity on
the critical path, i.e., reducing the time of the activity by an appropriate
amount. Here, we can reduce the completion time to .... weeks by crashing
either acti
activity
it .... or acti
activity
it ....
• It costs £........./week to crash activity “....” [ ]
• It costs £........./week
£ /week to crash activity ““....”
”[ ]
• Thus, we choose to crash .... by ..... week, which costs £.......... Although it is
possible to crash “.....” by as much as ..... weeks, we only need to crash
it ..... week in order to meet the ...... week deadline.
• Outcome:

Page 265 Page 266

PERT, Critical Chain and Gantt Chart
Program (or Project) Evaluation and Review Technique
(PERT) Judge Business School
ƒ Charts events in numbers (10s, 20s, 30s,..)
ƒ Is able to assign probabilities to tasks
Critical Chain (adoption of TOC in NPD by Goldratt)
Session 8
ƒ Focus on resources and their dependencies in order to understand
critical
i i l path
h Supply Chain Management
ƒ Do not search for optimal solution due to uncertainty. 80/20 rule.
ƒ Add buffers (time, resources)
ƒ Monitor progress via consumption of buffers, not by task completion
Gantt chart
ƒ Graphical representation of critical path

Page 267

Why do we talk about it? Supply Chain Management (SCM)

as Competitive Advantage
Traditional thinking: competition is driven by the 4P’s
ƒ Supply chain capabilities significant determinant of competitiveness
ƒ Wal-Mart versus K-Mart
ƒ Co
Compaq/HP
paq/ versus
e sus Dell
e How do companies
p use SCM as competitive
p
A final product is not the sole achievement of the OEM instrument?
ƒ Customer experience
p is determined by
y supply
pp y chain: q
quality,
y cost,
delivery Examples?
ƒ Significant proportion of value sourced from suppliers!
Supply chains are connected systems:
ƒ Competitiveness of one tier is a function of the supply and distribution
functions, i.e. surrounding tiers.
“Value Chains compete, not individual companies!” (M Christopher)

Page 269 Page 270

Operations vs. SCM Inventory Profile of the
Operations Management Supply Chain Management UK AAutomotive
i S Supplyl Ch
Chain
i
ƒ Main concern is to maximise the ƒ Main concern is to manage inter-firm
efficiency of the internal processes networks to maximise the value to 100
involved the ultimate customer Max
ƒ Focus: focal firm only (“user/chooser”)

ory
Average

Days of Invento
ƒ “Optimises” intra-firm processes with ƒ Focus: dynamic interplay between
Min
respect to the impact on the focal firms
50
firm, and immediate customer
ƒ Sub-optimisation
Sub optimisation of the operation at
the local firm in order to support
holistic strategy

0
Local versus Global Optimisation:

mbly WIP
Bought--in Parts

se Parts

mbly WIP

mbly WIP
ed Parts

On-siite Parts

d Transit

Disttribution

ustomer
Inbound Transit
Material

Dispatch
SCM is the differential decision making when

Raw M

In-hous

Cu
Finishe
Pre-Assem

Outbound
Assem

Assem
considering the firm within its value chain network!

Source: Holweg and Pil 2004

Page 271 Page 272

Inventory Profile of the Aligning Supply Chain Incentives

UK AAutomotive
i S Supplyl Ch
Chain
i Supply chain management involves managing “.. upstream and
downstream relationships with suppliers and customers in order to
create enhanced value in the final market place at less cost to the
100
supply chain as a whole” (M Christopher)
Max
ory

Average What mechanisms are at hand to implement this “sub

sub-
of Invento

Raw Materials Min Assembly optimisation”?

Distribution
and
50 components Plant 1. Power balance: dictate strategy
6% 73%
Days o

21% 2. Shared rewards: long-term lock-in, shared gains

How to go about it (Narayanan and Raman, HBR 2004)?

0
1. Acknowledge that an incentive misalignment exists
Pre-Assembly WIP
Bought-iin Parts

ed Parts
bly WIP

bly WIP
se Parts

On-sitte Parts

Outbound Transit

Distrribution

ustomer
Transit
Material

spatch

2. Diagnose
g the cause for the misalignment
g
Assemb

Assemb
Inbound T

Dis
Raw M

In-hous

Finishe

Cu

3. Change incentives (contracts, performance measures) to reward

partners for acting in the supply chain’s best interests
P

4. Review periodically, and educate managers across tiers.

Source: Holweg and Pil 2004
Page 273 Page 274
Power in the Supply Chain
– Everyone in the supply chain seeks to appropriate value for themselves
from participation!

– Certain players recognise that they have limited power to appropriate

value, but would seek to leverage more value if they could

– Understanding power structures is important for explaining inter-

organisational dynamics

– Toyota model: based on transforming power through creation of

hierarchies of structural dominance
Managing
g g Supplier
pp Relations
– Toyota is dominant player in the system, able to control key resources that
appropriate value

– Creates dependents (suppliers) who provide no threat to the flow of value

appropriation

Source: Cox 1999

Page 275 Page 276

The Supplier Relationship I The Supplier Relationship II

Strategic Coalitions (Porter, 1985) Adversarial versus collaborative sourcing
ƒ Idea of settingg up
p strategic
g coalitions within the value system
y to broaden up
p ƒ Arms-length
Arms length & price driven
the effective scope of the company's chain. versus
ƒ Importance and impact of coalitions as means of gaining the cost or ƒ Collaborative, trust-driven supplier relationships
differentiation advantages by vertically linking companies without actually
integrating them. Research shows that
ƒ Long-term agreements among independent firms that go beyond normal ƒ plants with fewer suppliers have better quality
market transactions.
ƒ trusting relationships show better performance
“Co-makership”, (Merli 1991)
Linked to debate on Lean Production & JIT
ƒ Supplier-client relations as a main business asset.
ƒ long-term
long term and stable relationships,
relationships a limited number of suppliers,
suppliers a global ƒ Japanese
p model showed superiority
p y also in the supply
pp y chain
certification system and supplier rating based on cost, rather than on price. ƒ See: Womack, Jones and Roos “The Machine”, 1990
ƒ The cost rating considers: quality cost, delivery related cost (for reserve Key studies
inventory, production interruptions and even lost sales), response time cost,
supply lot cost, costs linked to lack of improvement, technological ƒ Macbeth & Ferguson 1994: “Partnership Sourcing”
obsolescence cost. ƒ Lamming 1993: “Beyond Lean Supply”
g World-Class Suppliers”
ƒ Hines 1994: “Creating pp

Page 277 Page 278

 Why are the Japanese so much Secrets of Japanese Success
better…? 1. Long-term collaborative relationship
ƒ Trust and commitment, respect of the right
g of mutual existence
Performance Indicator Suppliers
to US auto
Suppliers to
Japanese
Suppliers
of… 2. Dual sourcing
plants transplants Chrysler Ford GM Honda Nissan Toyota ƒ Component volume is adjusted according to performance
Inventory turns 25.4 38.3 28.3 24.4 25.5 38.4 49.2 52.4
ƒ Constant positive pressure
Percentage change in + 0.65% - 0.85% + 0.69% + 0.58% + 0.74% - 0.9% - 0.7% - 1.3% 3. Improvement
manufacturing costs
compared with
previous year
ƒ C ll b ti with
Collaboration ith suppliers
li on operational
ti l iimprovement;
t example:l
Percentage of late 2.96% 1.38% 4.45% 1.70% 3.04% 2.11% 1.08% 0.44%
Toyota’s Supplier Support Center (TSSC) in Kentucky
deliveries ƒ Annual cost reductions are realised in collaboration, not isolation
Emergency shipment 714 371 1,235 446 616 423 379 204 4. Operations and logistics
costs in US$m
ƒ Level production schedules to avoid spikes in the supply chain
ƒ Milk-round
Milk round delivery systems that can handle mixed
mixed-load,
load small
small-lot
lot
Source: Liker and Wu 2000 deliveries
ƒ Disciplined system of JIT delivery windows at the plant; suppliers deliver
only what is needed
needed, even if this compromises load efficiency in transport

Page 279 Page 280

From ‘Exit’ and ‘Voice’ to ‘Hybrid Supplier Relations: Critical Choices

Collaborative’ Mode of Exchange
• Long-term collaboration or short term flexibility?
EXIT VOICE HYBRID COLLABORATIVE

Arms-length & transactional Long-term & relational Long-term & relational • Problem solving: exit, voice or hybrid strategy?
Open for new suppliers to bid Set of potential suppliers mostly Open to new suppliers
suppliers, after a vetting
closed period
• Single, dual or multiple sourcing?
Competitive selection by low bid – Selection based on capabilities -- Competitive assessment -- intermediate
frequent and speedy exit exit rare and slow frequency and speed of exit

Design simplified by customer to enlarge Design controlled by customer, Larger design role for supplier, attention
pool of suppliers supplier involved via resident
engineer
to supplier design capabilities
Æ What strategic
g value does the supplier
pp
No equity stake Often an equity stake Equity stake depends on criticality of
contribute?
technology

Contracts for governance Norms /dialogue for governance Norms + process management routines
for governance
Æ Quality
Quality, Cost,
Cost Delivery,
Delivery Service..
Service but also
Codified procedures Tacit procedures Process management routines make
Innovation, Technology, Responsiveness, etc.
procedures explicit
p p

Source: Helper 2005

Page 282
Building Deep Supplier Relations
Conduct joint improvement activities. Supervise your suppliers.
ƒ Exchange best practices with suppliers
suppliers. ƒ Send monthly report cards to core suppliers
suppliers.
ƒ Initiate kaizen projects at suppliers' facilities. ƒ Provide immediate and constant feedback.
ƒ Set up supplier study groups. ƒ Get senior managers involved in solving
problems.
Share information intensively
y but
ƒ Turn
T supplier
li rivalry
i l iinto
t opportunity.
t it
selectively.
ƒ Source each component from two or three
ƒ Set specific times, places, and agendas for vendors.
meetings.
ƒ Use
U rigid
i id fformats
t ffor sharing
h i iinformation.
f ti Create compatible production
ƒ Insist on accurate data collection.
philosophies and systems.
ƒ Share information in a structured fashion.
Develop suppliers
suppliers' technical capabilities
capabilities.
ƒ Set up joint ventures with existing suppliers
to transfer knowledge and maintain control.
ƒ Understand how your suppliers work.
The Renault-Nissan Case
ƒ Build suppliers' problem-solving skills. ƒ Learn about suppliers' businesses.
ƒ Develop a common lexicon. ƒ Go see how suppliers work.
ƒ Hone core suppliers'
suppliers innovation capabilities. ƒ Respect suppliers'
suppliers capabilities.
capabilities
ƒ Commit to co-prosperity.

Source: Liker and Choi 2004.

Page 283 Page 284

Crisis at Nissan Problems identified by Alliance team

FY’99 FY’06
ƒ Lack of p
profit orientation
Global sales 2.53 million units 3.48 million units
ƒ Insufficient focus on customers
Global market share 4.9% 9% (combined with
Renault) ƒ Lack of cross-functional, cross-border, intra-
Domestic market share 19.7% 13.2%
hierarchical lines within the company

Consolidated operating 1.4%

1 4% 9 2% (2007: 7
9.2% 7.3%)
3%) ƒ Lack of a “sense of urgency”
g y
profit margin
ƒ No shared vision or common long-term plan
Net income -684 billion JPY 461billion JPY

Debt 2 trillion JPY 2001: zero

Page 285 Page 286

Solution: Reduction.. Purchasing: Organizational change I
RNPO (2001)
50% in the number of suppliers
44%
20% in overall purchasing costs RENAULT
15%
NISSAN

50% in lead times 50% RENAULT-

NISSAN
50%

BV Attempt to have
50% in number of platforms best of both worlds?
100%

30% in capacity
p y
RNPO
20% in administration and sales cost
20% in number of distribution subsidiaries
Common global purchasing turnover Nissan and Renault
10% iin number
b off retail
t il outlets
tl t 2001: 30% Æ 2008: 85%
Page 287 Page 288

Purchasing: Organizational change II Did Nissan go too far?

MC-AFL (2003)

MC-AFL: Administration for Affiliated Companies, set up 2004: industry-wide steel shortage
g
to improve relationship with suppliers after rigorous
Nissan stopped production for 5 days in November
November-
supply chain restructuring
December 2004, and again for 2 days in March
New Group Enforcement: part of the Nissan Production 2005
Way
Æ 40,000
40 000 cum
cum. units loss of production
E.g. 2005: Nissan acquired 42% of previous supplier
Calsonic-Kansei Æ 16 billion JPY (£120 million) loss in profit

Page 289 Page 290

Implications
1. Inflicting a “blanket change” on an entire productive system as
Nissan tried to achieve, can cause unwanted results.

2 Nissan and Renault attempted to get the “best

2. best from both worlds”
worlds , but
did not fully consider the consequences of “mixing and matching”
parts of systems that had been developed as intrinsic wholes
wholes.

3. The lead-time for adjusting the various features of supplier Supply

pp y Chain Collaboration
relationships differs considerably, first and foremost for trust!

4 With hindsight it can be argued that a more balanced evaluation of

4.
the supply chain, i.e. not only financial but also operational, would
have led to a more balanced approach to reform
reform.

Page 291 Page 292

The traditional Supply Chain I The traditional Supply Chain II

Source: Holweg et al 2005 Source: Holweg et al 2005

Page 293 Page 294
Type 1: Information Sharing Type 2: Vendor Managed Inventory
(EPOS and/or Forecast)

Source: Holweg et al 2005 Source: Holweg et al 2005

Page 295 Page 296

Type 3: Collaborative Forecasting Supply Chain Collaboration

and Replenishment (CPAR/CPFR)
Type 1 Type 3

Forecast Collaboration
Collaborative Collaborative

Yes
Forecasting/ Forecasting and
EPOS Exchange Replenishment

Type 0 T
Type 2
Traditional Vendor

No
Supply Chain Managed
Replenishment

F
No Yes

Inventory Collaboration

Source: Holweg et al 2005 Source: Holweg et al 2005

Page 297 Page 298
 What is Outsourcing?
Traditional economic theory suggests that a firm should not make in
house what a supplier can provide for less
But: what about long-term competitiveness?
Key question: what is core competence, what is not?

1. Manufacturing outsourcing (e.g. Flextronics, BenQ, Magna Steyr)

ƒ Contract manufacturing
2. Design outsourcing (e.g. Pininfarina, Bertone)
Managing
g g the Global Supply
pp y Chain ƒ Specialist design houses

Outsourcing 3. Business process outsourcing (e.g. Xchanging, Capita)

ƒ HR services, call centres, indirect purchasing
ƒ Transaction-based versus customised services
ƒ The role of tacit knowledge

Page 299

Where does the money go? Take a £15 ‘Made in China’ Key Driver of outsourcing:
plastic ‘Professor
Professor Dumbledore’
Dumbledore figure … Labour Cost Differential (domestic)
£15
Employement and W ages in the US Auto Industry 1980 - 2000

600 OEM 50

550 Supplier
OEM wage
500 40

o yees
Supplier wage
450

$ per hour
Price

'0 00s E m plo

in the 400 30
shop
350

300 20

250

200 10
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000
Adapted from Slack et al. 2007 Source: Holweg and Pil (2004)
Key Driver of outsourcing: But not only labour costs: Assessing
Labour Cost Differential (internat.)
(internat ) Market Potential
Measure of potential: Vehicles in operation:
(Persons per vehicle) passenger cars (CV),
(CV) ratio of passenger cars/CVs

Germany $29.91 Korea $10.28 ƒ Argentina: 5.5 ƒ Argentina: 5.7m (1.5m) 3.8
ƒ Brazil: 77
7.7 ƒ Brazil: 19 4m
19.4m (4 8m)
(4.8m) 40
4.0
ƒ Mexico: 4.6 ƒ Mexico: 15.6m (7.9m) 2.0
US $21.97 Czech Rep. $4.71
ƒ China: 37.5 ƒ China: 11.0m ((24m)) 0.46
UK $20 37
$20.37 Brazil $2 67
$2.67 (2002: 87.6) (2002: 0.30)
ƒ India: 85.8 ƒ India 8.1m (4.9m) 1.7
Japan $20.09 Mexico $2.48
ƒ Czech Rep: 2.2 ƒ Czech Rep 4.1m (562k) 7.3
Spain $14.96 China $1.3 ƒ Russia: 4.3 ƒ Russian Fed: 26.8m (5.9m) 4.5
ƒ Poland: 2.5 ƒ Poland: 13.4m (2.3m) 5.8

Source: US Department of Labor Statistics, Wards, ONS. ƒ USA: 1.6 ƒ USA: 135.1m (108.9m) 1.2
ƒ Germany: 1.7 ƒ Germany: 46.6m (3.2m) 14.6
ƒ Japan: 17
1.7 ƒ Japan: 57 5m
57.5m (16 7m) 3.4
(16.7m) 34

Dialectics of Outsourcing Outsourcing: A Word on China…

Pros Cons China: threat and opportunity
ƒ Domestic sales increase
ƒ Focus on core ƒ Loss of control over
competences process
Problems:
ƒ Harness lower labour cost ƒ Limited ability to improve ƒ Abundant low-cost labour resources..
at supplier processes ƒ ..but
b t mostt high-value
hi h l partst still
till iimported
t d
ƒ Access to technology ƒ Risk of opportunistic ƒ Education & knowledge base
ƒ IPR: the ‘copycat
copycat cars’
cars
ƒ Stable and predictable b h i
behaviour off supplier
li
ƒ GM Chevy Spark - Chery QQ
financial planning in fee-for- ƒ Loss of human capital and ƒ Honda motorcycles, VW parts…
transaction services tacit knowledge
ƒ Less investment risk

Page 306
Outsourcing: A word on Nike... The 3 Supply Chain ‘Enemies’
1. Inventory & delays
ƒ Time worsens ‘swing’
swing of amplification
ƒ Decision delays require stock
ƒ Safety stock decisions send false signals

2. Unreliability or uncertainty
ƒ Any kind of uncertainty needs to be covered with inventory
ƒ Unreliable processes cause unreliable delivery

3. Hand-offs or decision points

ƒ Every hand-off or tier in the system bears danger of distortion!

3T’s of Effective Supply Chains Review of Learning Objectives

• Understand the basic decisions in Operations Management,
and their implications
p on firm p
performance
• Be able to manage inventory, schedule processes, develop
Trust basic forecasts
• Understand the Lean, Six Sigma and TOC improvement
concepts
p
Transparency
Time • Be able to manage a project (Critical Path Method)
• Understand the need to manage the wider supply chain

R Wilding: “The 3T’s of highly effective supply chains”

Page 309
 Thank you &

Good luck!

Page 311