1. Push and Pull Systems: Lecture 12
MRP and push systems
Kanban and CONWIP
Push and pull

2. MRP and Push Systems Demands for chairs are given as follows
Week
chair
How do we schedule the production, given its BOM?
chair
frame
4 legs
seat

3. Information Part Chair Leg Seat Frame On-hand 340 900 300 500
Leadtime
Lot size L4L
Safety stock

4. MRP Mechanism week 1 2 3 4 5 6 Chair (340) 200 250 220 300 500
250
220
300
500
Net demand
110
Leg (900)
440
880
1200
2000
order
800
1020
Seat (300)
400
480
Frame (400)
Netting demands
Offsetting leadtimes

5. MRP Record for the Leg week 1 2 3 4 5 6 7 8 MRP record for the Leg 440
Gross Requirements
440
880
1200
2000
800
1880
360
Scheduled. Receipt
Project balance
(900)
460
380
200
Planned order release
1020
LS = 800, LT = 1 week, SS = 200

6. MRP Record for the Seat week 1 2 3 4 5 6 7 8 110 220 300 500 200 470
Gross Requirements
110
220
300
500
200
470 90
Scheduled. Receipt
Project balance
(300)
190
370 70 50
250
350
280
Planned order release
400
480
LS = 400, LT = 1 weeks, SS = 50

7. MRP Record for the Frame
week 1 2 3 4 5 6 7 8
Gross Requirements
110
220
300
500
200
470 90
Scheduled. Receipt
400
Project balance
(100)
390
170
270
370
310
Planned order release
LS = 400, LT = 2 weeks, SS = 100

8. MRP Terms
Gross requirement is the anticipated future usage during a period
Scheduled receipt is order placed before and will arrive at the beginning of a period
Projected available balance is the inventory status at the end  of a period
Planned order release is order to be released at the beginning of a period
Net demand of the end product is the amount that should be available at the beginning of the period

9. Connecting Records
Now that each frame needs 4 rivets to put together. For rivets, leadtime= 1 wk, safety stock = 400, lot size = Current on-hand is 412
chair
frame
4 legs
seat
rivet (4)

10. frame/week 1 2 3 4 5 6 7 8
Gross Requirements
110
220
300
500
200
470 90
Scheduled. Receipt
400
Project balance
(100)
390
170
270
370
310
Planned order release
MRP record for rivet
Gross Requirements
Scheduled. Receipt
1600
Balance (412)
Planned order release
LS=1600, LT=1, SS=400

11. Rolling-Horizon: Week 12 3 4 5 6 7 8 9 10 11 25 36 27 26 28 30

12. At the Beginning of Week 21 2 3 4 5 6 7 8 9 10 11 25 36 27 26 28 30 23

13. At the Beginning of Week 31 2 3 4 5 6 7 8 9 10 11 25 36 27 26 28 30 23 31

14. At the Beginning of Week 41 2 3 4 5 6 7 8 9 10 11 25 36 27 26 28 30 23 31

15. Exercise 1 Fill the following MRP table week . Part A 1 2 3 4 5 6 .
Gross req
Scheduled rec
Project available
balance (21)
Planned order release
Q=20; LT=1; SS=0
During week 1, the following occur: find 12 additional units in inventory;
actual requirement is 15 units; wk 3 and wk 4 requirements reduced to 16
and increased to 14, respectively; wk 7 requirement is 15

16. Exercise 1 Continues Fill the following MRP table week .
Part A
Gross req
Scheduled rec
Project available
balance ( )
Planned order release
Q=20; LT=1; SS=0
During week 1, the following occur: find 12 additional units in inventory;
actual requirement is 15 units; wk 3 and wk 4 requirements reduced to 16
and increased to 14, respectively; wk 7 requirement is 15

17. MRP Based Control Forecast demand generate time phased plans
Planned orders FG
demand
machine
buffer
Parts

18. Push Production Systems
Forecast or anticipate demand
Translate forecasts into production plan
Convert production plan into schedules of lower level components in a top-down fashion
Executing schedules in a bottom-up fashion

19. Comments on MRP
MRP started a new way for factory management and is one of the most important manufacturing management paradigms
MRP process can be easily computerized, this led to the earliest MRP system by an IBM team led by Orlicky
MRP was the first major industry application of computer and has been the core (engine) of the information backbone systems
MRP logic is flawed. The fundamental weakness is the constant leadtime assumption, which amounts to the assumption of infinite processing capacity and no uncertainty in the system

20. MRPII: A Push Planning Framework
Figure 1
Low level planning
BOM.
Routing
Inventory
Status
Execution

21. Kanban and CONWIP Toyota production system
Continuous improvement
Smooth production
Eliminate wastes
How do we create a smooth (stable) production flow so that production matches demand and wastes in production process are minimized?

22. Kanban Control Scheme
Kanban control ensures that parts are not made except in response to a demand. FG
machine
buffer
Parts
Kanban

23. Dealing with Interruptions
Don’t produce defect
Don’t pass defect
Don’t accept defect FG
machine
buffer
Parts
Kanbans

24. Kanban Calculus (1) Work Station
throughput rate
Work Station
A single server queue with a base stock control policy

25. CT-100 Assembly (service rate )
Example
CT-100 Kanban System. Orders arrive 250 per month on average, and assembly capacity is averaged 300 per month. Suppose that a Kanban system is used to produce CT Determine total number of Kanbans allowed in the system, given a Kanban safety factor of 0.2 and container size of 5.
Estimated flow time = 0.2 months
CT-100 Assembly (service rate )
throughput
demand

26. CONWIP Constant work in process
Keep a fixed level of WIP (CONWIP). One finished part triggers a new part released to the first machine. FG
machine
buffer
Parts
Cards

27. Card Count in CONWIP Stable line speed and WIP level
Characterize the line with
- Practical production rate, RP
- Minimum practical leadtime, TP
By Little’s law, the CONWIP level IC is
IC= RPTP (2)
TP includes all (reasonable) detractors include setups, failures, etc.

28. Kanban Versus CONWIP CONWIP Kanban
MPS generates a backlog in front of a line
backlog can be sequenced
a card releases the first job to the line
requires a stable volume but can tolerate more of the product mix changes-- shifting bottleneck
Kanban
card for each station and part number
requires a stable product mix and volume

29. Variations of CONWIP
Tandem CONWIP: a long line separated to two or more sections
CONWIP with shared resource: two CONWIP controlled lines share a common machine
Pull from bottleneck: for a line with a stable bottleneck, control only the upstream section from the bottleneck

30. Push and Pull
Push
demands
Pull
Push: schedules the release of works based on (forecasted) external demands
Pull: authorizes the release of work based on system status

31. CONWIP and MRP: Observability
Control throughput and observe WIP
demand --- throughput
fluctuation --- congestion
Control WIP and observe throughput
Observability
- WIP is easy to observe and easy to control
- TH is tied with capacity, when effective capacity changes, it is hard to control

32. CONWIP and MRP: Efficiency
For a given level of throughput, a push system will have more WIP than an equivalent CONWIP
A CONWIP system can be modeled as a closed queueing network in which the number of customers is a constant
A MRP control system can be approximately seen as an open queueing network in which each machine has fixed workload

33. Efficiency Example
A line with 5 identical exponential single-machine stations with µ=1 job/hour
For CONWIP control with a fixed w (WIP):
Throughput rate TH = w µ /(5/µ+ w – 1)=w/(4+w)
For MRP, no control after release. The line behaves like 5 independent stations
ρ = TH/ µ for every station
The total

34. Hierarchical Planning in a Pull System
Marketing
Parameters
FORECASTING
Product/Process
Parameters
CAPACITY/FACILITY
PLANNING
WORKFORCE
PLANNING
Labor
Policies
Capacity
Plan
Personnel
Plan
AGGREGATE
PLANNING
Aggregate
Plan
Strategy
WIP/QUOTA
SETTING
Customer
Demands
Master
Production
Schedule
DEMAND
MANAGEMENT
WIP
Position
SEQUENCING &
SCHEDULING
Tactics
REAL-TIME
SIMULATION
Work
Schedule
Work
Forecast
SHOP FLOOR
CONTROL
Control
PRODUCTION
TRACKING

35. Push and Pull in Supply Chains
Extend the idea of pull beyond production control to supply chain: produce or make product available only when demand is received
Make-to-order instead of make-to-stock
The overall leadtime is often too long under the current competitive environment
Assemble-to-order make-to-stock for some upstream supply chain and make-to-order for the lower stream supply chain

36. A Hybrid Push-Pull System
supermarket
suppliers
Item 1
Inbound
Logistics
(pull/push)
Assembly
(pull/push)
delivery
Item n
Mixed Layout in Supermarket Manufacturing

37. Today’s Takeaways
MRP is an important manufacturing paradigm and provides the basis for push systems
Kanban and CONWIP are pull control with a theoretical basis in queueing theory, i.e., equations (1) and (2)
Push schedule job release while pull control the workload, and is more efficient and robust
MRP is easily computerized