Operations Strategy Kanban Systems
Kanban Systems: overview; pull systems
The ‘Pull’ Principle
Pull means replacing stock after consumption has occurred, i.e. what we do today is to replace what was used yesterday.
This is an old established principle in production control (see previous discussion of 2 bin, reorder level systems) Pull usually associated with Kanban what makes Kanban different, i.e. a key part of ‘Just In Time’ (JIT) system, is the way the factory is operated.
2
Doug Love/Aston University
Kanban Systems: Overview
Kanban Material control system, plus Part of a related manufacturing philosophy Toyota Production System/Lean delivers Just-In-Time (JIT) performance.
There are different forms of Kanban Kanban square One card kanban Two card kanban Kanban-MRP
Doug Love/Aston University
3
Kanban Systems: kanban squares
Control using Kanban Squares
Parts
Assembly line 1
Parts
Assembly line 2
Parts
Assembly line 3
Parts Manufacture Raw Material Stocks
Finished Goods Stocks
Cycle: consume - replenish - consume Kanban Square
Operator working kanban square empty
4
Kanban Square
Operator stopped kanban square full Doug Love/Aston University
Kanban Systems; single stock point (1 card)
Kanban: single stock location or 1 card Kanban card & bin cycle round together Card specifies part number, qty, source, consumer Card authorises movement & production Number of ‘cards’ controls stock level. KK
Unit 2
K
M/c
M/c Make
Kanban Stock Position
KK KK KK
Unit 1
K
M/c M/c
M/c
Use KK KK KK KK
Doug Love/Aston University
5
Kanban Systems; calculation of number of kanbans
Number of kanbans in circuit The number of kanbans has to be calculated (at least initially) to ensure smooth operation Number of kanbans controls material in system Need enough material in loop to ensure stock is always available to customer. Disruptions (scrap, breakdowns etc.) are covered by the safety factor in the formula
The formula is: Number of kanbans = Demand Rate x (Replenishment Time) x (1+Safety Factor) Container capacity 1 Where Replenishment Time includes all the times required to: Transport to cell & between machines + pass the card to the 1st operation + Queue at machine(s) + Set machine(s) + Process the kanban quantity Safety Factor is estimated/revised by users (can be statistically calculated) 1NB
6
Container capacity can be simply the selected kanban card quantity
Doug Love/Aston University
Kanban Systems; dual stock (2 Card)
Kanban dual stock location or 2 card full
full
PP
Unit 1
M PP PP P Stock P
M/c Make PP PP PP
P
MM Stock
Use
M
M/c
SWOP CARDS
Unit 2
M
M
empty
Introduces the idea of ‘production’ and ‘move’ Kanbans. Customer empties container releases Move card + container Move card + bin taken to production cell Full container same part located and cards swapped (Move on full, Prod’n with empty). Move returns to customer on full bin Production joins queue at start of manufacturing process Production + full bin added to stock at supplier
Doug Love/Aston University
7
Kanban Systems; dual stock (2 Card) cont.
MM Stock
PP
Unit 1
M/c Make PP PP PP
PP PP P Stock P
M
Use
M
M/c Unit 2
M
MM
SWOP CARDS
P
M MM
MM Stock
M
Use
M/c Unit 3
MM
Can handle multiple ‘customers’ Allows stock to be located at producer & user units Kanbans in each circuit calculated separately (different loop delays & risks) Good where units are a long distance apart & improves accountability. 8
Doug Love/Aston University
Kanban: factory environment
Kanban - Environment The system is very disciplined but it is also simple, so low stock operation depends on changes to; plant (productive maintenance, capability, setups), working practices (flexibility, ownership), supplier relationships (partnership, delivery, quality), customer relationships (level MPS). Levelled scheduling means building a schedule by making: runners every day/period repeaters when required (but as often as possible) strangers fill up the capacity available
These changes eliminate disruption to produce smooth, predictable material flow
Doug Love/Aston University
9
Operations Strategy, Systems & Implementation; Kanban System Description
Kanban as part of Lean This old diagram shows some ‘building blocks’ to Just In Time operation J.I.T.
KANBAN
QUALITY ON LINE
RELIABLE MACHINES
SIMPLE CELL STRUCTURE
10
QUALITY CIRCLES
SUPPLIER INTEGRATIO N
CAPABLE PROCESSES
EDUCATION AND TRAINING
LEVEL SCHEDULING
SHORT CHANGEOVER TIMES
COMPLETE ACCURATE DATA
FLEXIBLE PEO PLE
Doug Love/Aston University
Kanban Systems; The improvement cycle
Driving the Improvement cycle
The elimination of disruption is aided by the improvement cycle: 1.
2. 3. 4.
remove a card (or few) wait for a stoppage to occur, note cause if no stoppage, goto 1 otherwise fix the cause of the stoppage, then goto 1
Note the direct link between the stock level and why it is needed, thus
Kanban is used to focus improvement effort where it is most needed
This process is a critical aspect of the success of the system
Doug Love/Aston University
11
Kanban Systems MPS
Kanban Limitations Repetition of demand is required Why?
Demand must be (relatively) smooth Especially in terms of overall volume Some change in mix OK if setups are quick
Direct Kanban links and smooth demand Collaborating customers (customer development)
Otherwise an assembly schedule is used the MPS/assembly schedule uses levelled scheduling to minimise variability may require customer-facing stock to damp fluctuation
12
Doug Love/Aston University
Kanban Systems MPS
Kanban – MRP These systems use both MRP and Kanban at the same time Each system is used for areas/cells/products that suit its characteristics In parallel systems some self-contained modules or product lines will be MRP controlled and others by Kanban MRP System
MRP for - ?
MRP Cells
Kanban for - ? Kanban Cells
Doug Love/Aston University
13
Kanban Systems: Kanban-MRP Sandwich
Parts
Assembly line 1
Parts
Assembly line 2
Parts
Assembly line 3
Parts Manufacture Raw Material Stocks
MRP Control
Kanban Pull Control
Finished Goods Stocks
MRP Control
MRP Push linking Purchases with MPS
Kanban-MRP “Sandwich” MRP controls MPS and purchases Kanban controls all parts/assemblies made inside the factory MPS drives assembly/product manufacture 14
Doug Love/Aston University
Kanban Systems: Kanban-MRP Sandwich Parts
Parts
Parts
Kanban-MRP “Sandwich” MRP system changed: BOM: MRP does not control/stock any kanban controlled part Kanban parts entered as phantoms to implement this active BOM much simplified
Lead times include allowance for kanban controlled cycles Critical to synchronisation of supply
Inventory No tracking of kanban parts & no related transactions Backflushing widely used to avoid recording issues of purchased items
Levelled schedules required for viable operation Used where kanban links to customers & suppliers not feasible or may be an intermediate step in implementation
Doug Love/Aston University
15
Kanban Tutorials: Review of principles
Kanban Tutorial – Review of Principles* 1. Is kanban a manufacturing philosophy or a control mechanism? 2. What directly & indirectly determines the level of work in progress in a kanban-controlled manufacturing system? 3. 2 bin systems are a special case of which kanban mechanism? 4. In comparison with 1 card systems, 2 card kanban can be said to provide a better fit to the causes of delay in the system(s). Why? 5. How can the number of kanbans in the conveyance circuit be said to depend on motorway traffic jams? 6. How can the number of kanbans in the production circuit depend on the handling system used? 7. How can the number of kanbans in the production circuit depend on the TPM? 8. What do you have to do to make a manufacturing system (that uses kanban) run with less work in progress and/or material stocks? 9. When designing a kanban system there is little point in trying to calculate the number of kanbans in the circuit very accurately – why? 10. Why can 2 card kanban be said to provide better accountability than 1 card? *based on material in the lecture, Control Systems text notes and other sources in the bibliography 16
Doug Love/Aston University
Kanban Calc Tutorial 1
Calculation Tutorial 1 – 2 card system design Demand Rate ( per 8hr day) =
1200
=
2.5
Production Circuit The production circuit includes both operations in the supplier cell.
/min
Container Size
Operation Data
Weight Each Size (LxWxD)
1 Kg 0.2x0.1x0.05m
Handling Method Manual Hand Truck Power Truck
Capacity Kg 40 500 2000
Container Tote Pan Bin Stillage
Volume m3 0.054 0.211 1.000
0.001
Machine (mins/each) Operator Perf variability +/- %
m3
Other Data Element
Operation (all times in Minutes) 10 20 Average Max. Average Max. 120 180 120 200 30 40 20 25 20 45 20 45
Queue wait machining Setting Transport to next stage
The above times include the effect of miscellaneous losses, breakdowns etc. The time to pass the kanban back to op 10 is estimated at = 20 mins
Move/Conveyance Circuit The supplier and customer cells are located on the opposite sides of a large industrial site. Transport Data Transport Element Time (Mins) Average Max Wait Truck 15 30 Customer->Supplier Cell 15 18 Swop Cards 2 2 Supplier -> Customer Cell 20 24
NB the formula is, No kanbans=
Operation 10 20 2 1 20 30
Demand Rate x(Sum of Times in circuit) x (1 + Safety Factor) container capacity
Doug Love/Aston University
17
Kanban Calc Tutorial 2
Calculation Tutorial 2 – 1 or 2 card system? Cycle of activities Description of activity Empty => supplier Call truck Wait truck Transport to loading bay Wait loading/lorry or van available Load on lorry Transport to supplier Wait unloading Unload Wait truck Transport to manufacturing cell
Time (mins) 2 30 3 240 60 480 10 20 20 3
Swop cards
2
Production wait free operator transport card/bin to 1st op machine queue until parts for minimum run available set machine process run quantity move to next 2nd operation queue waiting machine free set machine process move to cell storage area
10 2 calculate 30 2 2 240 30 3 5
Wait truck Transport to loading bay
NB data continues on next page
18
20 3
Comment put up signal next time on circuit
wait while other items loaded time depends on route
forklift drop off & place cards in rack
done by forklift driver
minimum run quantity =
2000
NB std time is 2 mins each 1st operator does this
NB std time is 3 mins each 2nd operator does this forklift
Doug Love/Aston University
Kanban Calc Tutorial 2
Calculation Tutorial 2 – 1 or 2 card system? cont. Supplier => return full Wait loading/lorry or van available Load on lorry Transport to customer Wait unloading Unload Wait truck Transport to assembly cell
Notes Variabilty of the times has been estimated at: Production Elements Transport Elements
30% 50%
Possible container capacities are (in units): Large Medium Small
2000 400 50
Demand Estimated demand is (units/hr)
19
240 60 480 10 20 30 5
wait while other items loaded time depends on route
drop off by track
25
Doug Love/Aston University