Chapter 8

Pull Production Systems

USA Snapshots
v Strongest union states
Nationally, 13.9% of nonagricultural workers are union members. States
with the largest number of workers who belong to unions.

Hawaii 26.5%
New York 25.4%
New Jersey 22.0%
Michigan 21.6%
Washington 21.2%
Alaska 20.4%

2
Source: Bureau of Labor Statistics data for 1998, most recent available By Suzy Parker, USA Today
 Feats of Engineering
v Engineers picked the following as the 20th century’s greatest engineering
achievements

Electrification 100%
Automobile 95%
Airplane 88%
Safe water/supply/treat 86%
Electronics 80%
Radio and TV 73%
Agricultural mechanization 70%
Computers 63%
Telephone 62%
Air conditioning/refrigeration 54%

Source: National Academy of Engineering By Anne R. Carey and Quin Tian, USA Today 3

USA Snapshots
v The percentage of workers belonging to unions stayed at 13.9% in 1999 (16.5
M), ending a decade of decline.

1979 24.1%
1989 16.4%
1990
1999 13.9%

Source: Bureau of Labor Statistics By Anne R. Carey and Sam Ward, USA Today

4
 Lowest Union Membership
v States with the lowest percent of workers who belong to unions:

North Carolina 4.2%

South Carolina 4.5%
Mississippi 5.6%
Texas 5.9%
Arkansas 6.2%
South Dakota 6.4%
Source: Bureau of Labor Statistics data for 1998, most recent available By Gary Visgaitis, USA Today

Sources of Waste

v Overproduction
v Inventories
v Transportation
v Delays
v Defective Products
v Processing
v Motion

6
 Causes of Waste

v Departmental layouts v Long setup times

v Processes not capable v Unaligned performance
v No standard practices measures
v No adherence to standard v Lack of workplace
practices organization
v Poorly performing suppliers v Poor forecasting and
v Ineffective maintenance production
planning/scheduling
v Lack of training

Objective
Products are to be made:
v In the required quantities
v At the right time
v With the highest quantity
v At the lowest cost

Results from:
v The least non-value added activity
v Production problems easily identified and remedied

The greater the demand variation and the larger the number of product
configurations, the more difficult this task becomes

8
 Push Production System
v Schedule is based on when an order is to arrive at an operation (or
due date priority of items in WIP)
v Production is generally performed in batches (MRP lot sizing rules)
v Routing sheets, schedules, and work instructions follow the jobs

Or work instructions could be maintained at a given operation

v A schedule is generated for each operation

v Processing time is typically less than 10% of total lead time
v Priorities change at each operation
v Process and transfer batches could change the batch size
v Control process is costly and wasteful
v Expediting causes problems
- As some jobs move up in priority, others must move down
9

Push Production System (cont.)

v Schedules for every job at every workstation could be centrally prepared
(possibly by MRP)
v Schedule at each operation is dependant on:
– Lead time (could be estimated LT from this point to finish)
– Critical Ratio (time to due date / expected flow time to finish)
– # of operations
– Capacity at workstation

Supplier 1 A 2 B 3 C Customer
(Raw Material) (Finished
Product)

10
 Pull Production Systems
v Detailed production schedules at each operation are eliminated

v Schedule is placed at most downstream operation that can achieve

continuous flow (or close to continuous flow) - e.g., an assembly line

v Quantities and timing of work should migrate to workers instead of being

retained within a central scheduling group

v Signals by downstream operation requests upstream producer to

replenish goods (consumption signals replenishment)

v Downstream operation keeps enough in stock to satisfy demand during

replenishment lead time

v Uses small buffers for imbalance of production and demand

11

Flow of material and signals in a pull

system

Supplier 1 A 2 B 3 C Customer
(Raw Material) (Finished
Product)

Material Flow

Order Signal Information

1 In-Process Buffer A Operation

12
 Buffer
v Pull production is considered stockless
v Impossible to produce in a lean environment with no inventory

Zero inventory equals zero output

v Must hold inventory in the process to meet lead times desired

– For example, if I can meet the customer lead time and still flow through
operations B and C, then inventory is held in front of B, but not
between B and C.

Supplier 1 A 2 C Customer
(Raw Material) B FIFO (Finished
Product)
13

Buffer (cont.)
v Need a buffer for each kind of part in kanban pull systems

v Difference from push – less inventory in each buffer

v Size of buffers dependant on demand rate and replenishment times (time

from generating signal to replenishment of the inventory).

v Replenishment times are often a function of setup times.

Upstream
Order Operation Container emptied
Queue 2
1 I
Downstream
Operation
Consumption Signal
14
 Pull vs. Push Production
Push Pull
1) Schedules order releases based on 1) Order releases occur when the
lead times and the master schedule. downstream buffer reaches a given
level.
2) Lot sizes are based on lot sizing 2) Lot sizes are determined at the
rules from the master schedule. shop-floor level and are based on
demand and replenishment
requirements of downstream
inventory buffers.
3) Priorites based on rules (ex: 3) Priorities are determined by
earliest due date, SPT, FIFO) but are operators using a sequence board.
often changed on the floor according
to the work schedule.
4) Expedited orders will be prioritized 4) Expedited orders will be prioritized
according to capacity and urgency. according to capacity and urgency.

15

Containers and Cards

v Standard-sized containers are used to control inventory buffers

v Kanban cards are used to differentiate containers

v Kanbans include:
– Kind of material
– Quantity of material
– Origin or producer of the material (upstream source)
– Consumer of material (downstream customer)
– Could be bar-coded to ease the pain of inventory transactions

16
 Containers and Cards (cont.)
v Rules for kanbans:
– No material may be placed in a container without a card
– No container moved without a card
– The quantity of material produced will never exceed the amount on the
card
– No material may be processed without a card

17

Rules for Pull Production

v Downstream operations withdraw only the quantity of items they need from
upstream operations. The quantity is controlled by the number of cards.

v Each operation produces items in the quantity and sequence indicated by the
cards.

v A card must always be attached to a container. No withdrawal or production is

permitted without a kanban.

v Only non-defective items are sent downstream. Defective items are withheld and
the operation is stopped until the source of defect is remedied.

v The production process is smoothed to achieve level production. Small demand

variations are accommodated in the system by adjusting the number of cards.

v The number of cards (for a given part, subassembly, etc.) is gradually reduced to
decrease WIP and expose areas that are wasteful and in need of improvement.

18
 Continuous Improvement
v Goal: To continually reduce inventory buffers to expose sources of waste
and eliminate them.

Initial Visible Productivity Savings ($)

PRODUCTION MATERIAL
CONTROL
PRODUCT QUALITY =

LOYAL CUSTOMERS SUPPLIER BASE

FLEXIBILITY SAVINGS

=
COMPETITIVE
ADVANTAGE DISTRIBUTION SAVINGS
= ACTIVITY-
INVENTORY $ + TIME ENGINEERING
MARKET DESIGN SAVINGS BASED
SHARE + TIME + $ COSTING
NEW MARKETS &
STREAMLINED RESPONSIVE PROCESS
INFORMATION SYSTEMS COSTING
BUSINESS PROCESSES

19
SURFACE POTENTIAL GAINS ACROSS THE VALUE STREAM

Necessary Conditions for Pull

v Shop supervisors and teams of workers must have more planning and control responsibility.

v Must produce to only meet demand; no overproduction(major component of waste).

v Company goal must be to reduce excess buffers.

v Preventive maintenance must be used to decrease process variation and increase capacity.

v Prevent defects from occurring by using quality assurance methods.

v Reduce setup times for quick changeovers.

v Link all operations in the process and level the capacity.

v Production plans and schedules must be somewhat uniform.

v Develop cooperative work attitudes and teamwork!

20
 How to Achieve Pull Production
v Issues to address:
– How do we signal production?
– How do we signal movement of materials?
– How large should our lot sizes be?
– How do the workers know what to do next?

21

Reorder Point System

v Uses fixed quantities
v Reorder point – when the buffer falls below a critical level

ROP = D * LT + SS
D = Average demand rate (units / time)
LT = average lead time between order and replenishment
LT = P + C
P = Production time
C = Conveyance time
P and C are also both averages

22
 Reorder Point System
v Example:
Demand is 1500 per month. There are twenty work days in a month. A
safety stock of 50 units is needed because of high demand variation.
Setup takes 1/10 of a day, production requires half the day, and the lot
waits 4/10 of the day. When an order is generated, it is received by the
supplier in 1/4 of the day and then requires one day to receive the
shipment. What is the ROP?

23

Number of Containers
v Number of containers depends on ROP and capacity of containers

= D * (P + C) + SS
K
Q
Q = capacity of the container (units per container)
K = max number of full containers
(K may represent the number of kanbans)
**Note: Too many containers may cause complications**

Using the previous example, how many containers do we need if each container
holds 50 units?

24
 Container Size
v The smaller the container:
– The shorter the reach for parts
– The less space taken up
– The closer the container may be placed to the workstation
– The shorter the assembly line (POU WIP)
– The easier to tilt the parts toward the worker (ergonomics)

v Try to keep the size of the containers similar

v Rule of thumb – containers should hold about 10% of daily demand

v May need material handlers to move containers (in contrast to operators

doing their own material handling)

25

Outbound and Inbound Buffers

Outbound Inbound
Buffers Buffers
M M 2 Buffers
1 S S Movement of parts

3 Movement of parts
V V and containers

v Due to distance between operations, buffers may be needed after the

upstream station and before the downstream station.

v Need a signal to move material from the outbound buffer to the inbound
buffer (withdrawal kanban).

v Need a signal to produce at the upstream station (production kanban).

26
 KANBAN ~means Visual Signal

There are two basic types of Kanban Signals:

1. In-Process Kanban ~ a visual
signal to pace the movement of
products in a flow manufacturing
process

RED
KANBAN CARD ID: 0 Production PRINT DATE:
2. Material Kanban ~ a visual *0* Unit of Lot 3/16/95

signal to replenish materials LOTSIZE NUMBER:

1 of 4
PART NO: 12-00327-09
12-00327-09
REF. DRAWING NO.:
dw1234

consumed in a lean production PART NAME: KANBAN QUANTITY: 3

*3* REVISION NO.:
dr987
MODULE 69NT CONTAINER TYPE: BOX
process Supply Point: Supply B
*Supply B*
USAGE POINT: USAGE C Notes:
*USAGE C* NOTE 1

27

Types of Material Kanbans

v Production kanban – enables processing of parts

– No production is allowed without it

v Withdrawal (conveyance or move) kanban – enables transportation of

parts
– No container may be moved downstream from an outbound buffer
without it

v Supplier kanban – enables procurement of parts

– links the facility with its suppliers
– Another form of withdrawal kanban

28
 How Three Types of Kanban
Work Together
Blue = Withdrawal Kanban (WK)
Red = Production Kanban (PK) IB = Initial Buffer
FB = Final Buffer
Green = Supplier Kanban (SK)
supplier

planner IB FB IB FB

(fax) SK PK WK PK
Process attach Process
WK
#1 #2
FIFO

Kanban Kanban

Order Post Receiving Post

29

How Three Types of Kanban

Work Together (Cont.)
v When container in the initial buffer is opened or is empty, the withdrawal kanban is
removed from the container (or container and card positioned in signaling
location) and used to requisition material from final buffer of previous operation.

v The withdrawal kanban is placed on the container pulled from the outbound
buffer (production kanban taken off container to trigger production, placed on
receiving post) of upstream operation and taken to initial buffer of next operation.

v The production kanban is placed in the receiving post (FIFO) and acts to schedule
production.

v The supplier kanban acts just like a withdrawal kanban, except that the upstream
operation is an external supplier.

30
 Kanban Calculations

D x (1 + SF) x LT Signal =
Number of Kanbans = Full Container D = Average Demand
Kanban Size
Opened SF = Safety Factor
D x (1 + SF) x LT Signal = LT = Kanban Cycle Time (time for
material to be replenished once
= +1 Container a signal has been received)
Kanban Size
Empty

Signal could be: empty container

container opened
full container taken away

Number of kanbans = number of cards (not necessarily full containers)

required for system to work
D x (1 + SF) x LT
If the number of kanbans = 2, then: Kanban Size =
2

31

Interaction Between Production Kanban

and Withdrawal Kanban

Operation #1 Operation #2

Withdrawal Production Withdrawal Production

Kanban Kanban Kanban Kanban

Production Rate = 12 units/minute Demand Rate = 6 units/minute

= 0.2 units / second = 0.1 units / second
Kanban Size = 6 units Kanban Size = 6 units
Let Safety Factor = 10%
Lead Time = 30 seconds

32
 Interaction Between Production Kanban
and Withdrawal Kanban

Withdrawal kanban - signal when container is opened

D (1+SF) LT
Number of Kanbans =
Kanban Size
=

Withdrawal kanban - signal when container is empty

Number of Kanbans =

Withdrawal kanban - signal when container is empty & LT = 120 seconds

Number of Kanbans =

33

Interaction Between Production Kanban

and Withdrawal Kanban

Production Kanban

Production Rate = 12 units/minute

Demand Rate = 6 units/minute
(a) Waiting time to work into schedule
Time (sec.) % Total
60 40
70 30
80 30

Wait Time = ?

(b) Run Time = ?

D (1+SF) LT
Number of Kanbans =
Kanban Size
=
34
 How else can I make the

D = average demand
SF = safety factor
LT = kanban replenishment
time

pull system work?

Interchange the production kanban and withdrawal kanban.
 Production KCT Withdrawal KCT
Withdrawal
Do not interchange the production kanban and withdrawal kanban. kanbans for
materials
pull from
warehouse

Production Kanban Flow

Likely greater variability in KCT!

35

Flow of Materials and

Production Kanbans
Initial Buffer Final Buffer withdrawal
Process 6 kanban

4 attached
5 production

kanban
removed
3 2 1
Kanban Kanban
Order Post Receiving Post

Material Flow Number of = D x (1+SF) x LT

Production Kanban Flow Kanbans Kanban Size

Kanban Cycle Time =

11 kanban waiting time in the receiving post +
2 kanban transfer time to ordering post +
3 kanban waiting time in ordering post +

4 lot processing time (internal setup time + run time + in-process waiting time) +
5 container transfer time to final buffer +
6 container waiting time in final buffer 36
 Flow of Materials and
Production Kanbans
Example: Do not need to consider whether
To add one or not with production
D = 200 units/hr
Kanban - signal will always occur
SF = 10% When container is full.
LT = 3 hours

D x (1+SF) x LT = 200 x (1.10) x 3 = 6.6

Number of Kanbans =
Kanban Size 100
Round up to 7!
KCT = 100 x 7 = 3.5 hours
200

Container will be received in outbound buffer 30 minutes before

needed
37

Flow of Materials and

Withdrawal Kanbans
Upstream Process Downstream Process
Final Buffer 3 Initial Buffer
4
2 1

Ordering

Post
Material Flow D x (1+SF) x LT

Number of Kanbans =
Material Kanban Flow Kanban Size

Kanban Cycle Time =

1 kanban waiting time in receiving post +
2 kanban conveyance time to upstream buffer +
3 container conveyance time to downstream buffer +
4 container waiting time in downstream buffer

Note: Formula assumes the LT begins when container is opened at the downstream buffer. If the
discipline assumes it begins when container is empty, must add one. 38
 Supplier Kanbans with
Constant Order Cycle
v Permit withdrawal from suppliers finished goods inventory and follow the
product until they are processed at one of the companies operations.

v Product ordering and withdrawal occur at fixed times.

v Quantity ordered may vary.

v Withdrawal times and frequencies of deliveries are negotiated between

the supplier and the company.

39

Supplier Kanban Cycle Time

(An Example)
Negotiated number of deliveries = 1 daily
Negotiated delay = 24 hrs. (from day one 1 PM to day two 1 PM)
Kanban conveyance time to supplier = 5 hrs. (from 8 AM to 1 PM)
Truck waiting time at supplier plant = 1 hr. (from 1 PM to 2 PM)
Material conveyance time from supplier to company = 5 hrs.
(from 2 PM to 7 PM)

Kanban minimum lead time = 24 + 5 + 1 + 5 = 35 hours

Lead time must be expressed in days 2 days

Number of kanbans delivered depends on demand generated during the last kanban
cycle.

40
 Kanban Calculations
Example

Number kanbans for part Z.

Product Z used for models A, B and C as follows:
Model Number Part Z(Q)
A 1
B 3
C 2
Daily rate at capacity, Dc:
Model Dc
A 21
B 16
C 32
LT = Replenishment Time (fraction of days) = 0.25
SF = Safety Factor = 1.10
K = Kanban Size = 15
Number of
= [(21 x 1) + (16 x 3) + (32 x 2)] x 0.25 x 1.10
Kanbans 15
= 36.58 = 2.44 3
15 41

Signal Kanban
v Special type of production kanban.

v SP-kanban is for ordering batches or quantities in excess of one container.

v Batch size can be any multiple number of containers.

v Using SP-kanban is like collecting several production kanbans

before production occurs (also called unit of lot by some).

v SP-kanban works well in cases where containers are stacked and stored.

v SM-kanban is used in conjunction with an SP-kanban to order materials

required to produce the product.

42
 Signal Kanbans and Their Locations
Part No. D-411

Part Housing

Name
Lot 18 Reorder 9

Size Point
SP-kanban
Material J5 Storage No. 18 Lot Size Reorder Point
Area
SM 18 6
SM
kanban Part No.
D-411

SP Go get material Process

signal Mold
#3
Signals production

43

Positioning the Signals

K = Number of full containers in the stack where the card is located.
SP

K = D (LT) + S S
SP ,
Q
where D = average demand
LT = Replenishment time
Q = kanban size
K = location of the SM- kanban
SM

'
K D ( C - P)
SM = + K SP
Q
C = Total time between when the materials are ordered and when they arrive for usage.
P' = Time between when production is first ordered and when setup begins.
I f ( C - 'P) p 0 then round to the lowest negative integer (e.g., - 0 . 3 = -1)

44
 Example 6 - Calculation of SP - Kanbans
D = 30 units per day
Q = 3 units
Batch size = 18 units
P = 75 minutes (includes time spent at ordering post, setup, run product and back to buffer.
Getting into schedule =16 minutes
Setup = 16 minutes
Processing time = 2 (18) = 36 minutes
In-process wait and time to move to buffer = 2 + 5 =7 minutes; Total = 75 minutes
P = 7 5 minute s = 75/48 0 = . 1562 5 days

K = 30 (.15625) = 1 . 5625
SP
3

Must be an integer, round up to 2. Safety stock is 28% (.4375/1.5625 x 100%)

When buffer reaches two full containers, a production order for a new batch of six containers will be placed in the kanban mailbox.

45

Example 7 - Calculations for SM- Kanbans

D=30units/day,Q=3
Time to obtain raw materials for the run, produce and move to inventory buffer includes:

Move to order post, travel to location where materials are stored, wait for order accumulation, time to operation, setup, and production.

C = 42 minutes
P' = Time for SP- kanban to sit in mailbox,move to order post, and wait in order post before setup occurs.
P' = 16 minutes
( C - P') = 26 minutes = 26/480 = .0542 days

K = 30 (.0542) + 2 = 2.542, Round up puts K at three containers.

SM
3 SM

46
 CONstant Work In Process (CONWIP)
v Process:
– Product leaves operation D and the card is removed
– Card is sent to operation A
– A product on the schedule is paired up with the card
– The product and card are pushed through the operations in FIFO order

Schedule
1 2 3
A B C D

In-Process Buffer Operation

47

CONWIP (cont.)
v Card authorizes production
v Advantages:
– Card Count
• Card is not product or part specific
• Product, accompanying the card, is specified by the schedule
• Reduces the overall number of cards
• Good for products with lower demand (job shop: hours out = hours in)
• Easy to remove cards
– Floating Bottlenecks
• Tolerates changes in product mix and volume easier than Kanban
• Bottleneck may move for various products
Automatically buffers the bottleneck
v Products sharing operations, but moving through different routing sequences, must
use traditional kanban
-Conwip could be used for a segment of the routing
48
 Other Scheduling Techniques
v Wheeled carts - empty cart or space on the floor signals replenishment
– Used to move large objects or large batches
v Kanban squares - replenishment occurs when signal mark appears
– ex: stack of staples is replenished on a shelf if inventory falls below a
certain point
v Golf balls - colored golf ball signals which product to replenish (good when
there are only a few products)

49

Other Scheduling Techniques

v Electronic kanban – uses keyboards and monitors to convey what should be
replenished

v Clothespin Clips – Each clothespin clip is associated with a product and has its
SKU number on it
– When a product is used, the clip is removed and placed on a wire designated
for the SKU
– The product with the most clips on one wire has priority

v Milkrun – material handler makes periodic rounds to drop off empty containers and
pick up full containers

v Kanban Sequence Board – cards are posted on a board so the next product in the
sequence may be found easily

50
 Kanban Sequence Board
R14 R19 R20 S5 S6 S9

Green

Yellow

Red

• When cards arrive they are sorted and hung on the board by item
• Cards are hung from the top of the board
• As items move from green to yellow and on to red they grow in priority

51

To Pull or Not to Pull

v Pull systems need fairly stable, continuous demand

v Most pull systems need every kind of material to be held in a buffer

v Pull works better for repetitive, standardized products

v Pull is easier if products are very similar

v Pull needs to operate in an environment with few defects

52
 Lean Enterprise: Hybrid Execution
System

High PULL

Volume Third
Hybrid Control Systems: Dimension:
Constraint
MRP Plan/Pull Execution Management

Low MRP Plan Traditional MRP

Pull Execution PUSH
Low Product Variety High

53