Constraints Management Intro to the Theory of Constraints

Constraints Management Intro to the Theory of Constraints (A lecture introducing a portion of the Physical side of the Theory of Constraints) James R. Holt, Ph. D. , PE Professor jholt@wsu. edu http: //www. engrmgt. wsu. edu/ Engineering & Technology © Washington State University 1 Management 2010

Welcome to the TOC! • TOC IS: § A set of Proven Solutions • DBR, CCPM, Replenishment, Sales/Marketing, Human Behavior, Measurements, Strategy § An Approach to Problems • Five Steps of Continuous Improvement § Tools for Discovery of New Solutions • What to Change, What to Change to, How to Cause the Change (The TP) © Washington State University-2010 2

Process Theory Input Process Output Larger Process Input Process Output Input Process © Washington State University-2010 Output 3

Systems Concepts • Organizations / Systems exist for a purpose • That purpose is better achieved by cooperation of multiple, independent elements linked together • Each Inter-linked event depends in some detail upon the other links. • The system owner determines purpose © Washington State University-2010 4

There is a “Weakest Link” • Different link capabilities, normal • • variation and changing workload make it impossible to balance everything. One element of the system is more limited than another. When the whole system is dependent upon the cooperation of all elements, the weakest link determines the strength of the chain. Load 100 © Washington State University-2010 5

Interconnections are Extensive • Every Systems have relatively few constraints and Many Interconnected relationships. § The generic problem with physical systems § The Five Focusing Steps § The Generic Physical Solution • Physical and Non-Physical Processes § § Flow systems (I, A, V, T structures / combinations) Distribution and Supply Chain Project Management control of these systems © Washington State University-2010 6

Interconnections are non -Trivial • • • A simple chain over simplifies reality Link 1 has a relationship with Link 5 has a different relationship with 1 1 2 3 4 5 6 7 8 9 © Washington State University-2010 7

Management of the Links Vs. Linkages • Even with the many links and linkages, there is only one constraint. Link 8 and 9 can combine to push on both Link 6 and Link 7 Link 1 and 2 can get together and lean on Link 3 or Link 8 1 2 3 4 5 6 7 8 9 © Washington State University-2010 There are 40, 000 first order effects and 1, 000+ second and higher order effects! 8

Traditional Approach: Divide and Conquer • Division of Labor breaks down linkages complex systems into manageable chunks. • Which is harder to manage? Left or Right? Left Right © Washington State University-2010 9

We Measure Operational Efficiency • Work flows from left to right through processes with capacity shown. Process A B C D E FG RM Capability Parts per Day Market Request 11 7 9 5 8 6 Too Much Overtime Chronic Complainer Excellent Efficiency--Near 100% © Washington State University-2010 10

Reward Based on Efficiency • Work flows from left to right. Process A B C D E FG RM Capability P/D 7 9 5 8 6 And soon, A & B slow Both found ways to look busy and appear to have a capacity of 5 parts/day. © Washington State University-2010 11

In reality. . . • Processes A and B won’t produce more than Process C for long. Process A B C D E FG RM Potential P/D 7 9 5 8 6 Reality 5 5 5 © Washington State University-2010 12

Then Variability Sets In • Processing times are just AVERAGE Estimates Process A B C D E FG RM Reality 5+/-2 © Washington State University-2010 5+/-2 13

What’s an Average? 50% • Half the time there are 5 or more per day at each process--Half the time less Process A B C D E FG RM Reality Probability 5+/-2 0. 5 Two at a time: Over all: 0. 25 5+/-2 0. 5 0. 25 3% Chance of 5 + per day © Washington State University-2010 14

Previous Solution: Inventory • Put a day of inventory at each process! WIP Process 5 5 A 5 B 5 C 5 D Total 25 E FG RM Variable 5+/-2 5+/-2 Process Inventory (WIP) acts as a buffer between processes. Inventory manager/expediter tries to smooth it out. But, flow though the line slows down. Costs go up. © Washington State University-2010 15

System Variability Takes Over-Chaos An Average of 5 means sometimes 3 and some times 7 Process WIP A 3 B 0 C 10 D 8 E 4 Total 25 FG RM Variable 5+/-2 5+/-2 Process And, the work-in-process can shift around creating large queues at some locations. © Washington State University-2010 16

System Variability Takes Over-Chaos Process WIP A 3 B 0 C 10 D 8 E 4 Total 25 FG RM Variable Process 5+/-2 5+/-2 Other workstations can be starved for work. The work they could be doing is delayed because it is not there. They can’t take advantage of their extra capability. So. . . © Washington State University-2010 17

System Variability Takes Over-Chaos Process WIP A 3 B 5 C 10 D 8 E 4 Total X 25 30 FG RM Variable 5+/-2 5+/-2 Process So… Management Helps! Management puts in more work(Inventory) to give everyone something to do! Result: More and more delay happens! It takes longer and longer from time of release to final shipping. © Washington State University-2010 18

Attempts to Control WIP • “Put a lid on it”-Use Kanban Cards-JIT WIP 5 Process A 5 B 5 C 5 D 5 Total 25 E FG RM Variable 5+/-2 5+/-2 Process Just-In-Time uses Kanban Cards to limit the queues building in the system. No more than 5 parts are allowed at any station. Looks good, but is it? © Washington State University-2010 19

Effects of Inventory Limits on Production • What does a Kanban card of 5 Mean? WIP 5 Process A 5 B 5 5 C 5 D Total 25 E FG RM Variable Process 5+/-2 After Kanban Before Kanban 5+/-2 Average = 5 5+/-2 Can’t exceed 5 5+/-2 Average = 3. 5 © Washington State University-2010 20

Operation’s Dilemma Produce a lot Increase work-inprocess Manage production effectively Costs & delivery in control Decrease work-inprocess Assumption: We can’t both increase WIP and decrease WIP at the same time. Injection: Put a large inventory where it’s needed and low everywhere else! © Washington State University-2010 21

TOC Steps to Continuous Improvement Step 1. Identify the system’s constraint. Step 2. Decide how to Exploit the system’s constraint. Step 3. Subordinate everything else to the above decision. Step 4. Elevate the system’s constraint. Step 5. Warning!!! If a constraint is broken, go back to Step 1. But don’t allow Inertia to become a constraint. © Washington State University-2010 22

Five Steps Applied to Flow Operations 12 WIP A B Total C D 12 E RM FG 7 9 X 5. 5 5 XXX 7 8 6 Five Focusing Steps Step 1. Identify the Constraint (The Drum) Step 2. Exploit the Constraint (Buffer the Drum) Step 3. Subordinate Everything Else (Rope) Step 4. Elevate the Constraint ($? ) Step 5. If the Constraint Moves, Start Over © Washington State University-2010 23

Understanding Buffers WIP Total 12/5=2. 5 Days A B C D E FG RM 7 9 5 8 6 • The “Buffer” is Time! • In general, the buffer is the total time from work release until the work arrives at the constraint. • Contents of the buffer ebb and flow within the buffer • If different items spend different time at the constraint, then number of items in the buffer changes • but Time in the buffer remains constant. © Washington State University-2010 24

We need more than one Buffer Raw Material Buffer A B C D E Finished Goods Buffer FG RM 7 9 5 8 6 There is variability in the Constraint. To protect our delivery to our customer we need a finished goods buffer. There is variability in our suppliers. We need to protect ourselves from unreliable delivery. © Washington State University-2010 25

Buffer Time is Constant. Predictable Raw Material Buffer A B C D E Finished Goods Buffer FG RM Raw Material Buffer 2 Days 7 9 Constraint Buffer 2. 5 Days 5 8 6 Finished Goods Buffer 1 Day Processing Lead Time is Constant © Washington State University-2010 26

Buffer Management Constraint Buffer WIP A Total 12/5=2. 5 Days B C D E RM FG 7 WO 21 WO 20 WO 19 WO 18 9 WO 17 WO 16 WO 15 WO 14 5 8 WO 13 WO 12 WO 11 WO 10 2. 5 Days 0 6 • The Constraint is scheduled very carefully • Buffer Managed by location • Individual activities in the buffer are not scheduled Time until Scheduled at Constraint © Washington State University-2010 27

Problem Identification RM A B C D E RM WO 19 Delayed Parts FG 7 WO 21 WO 20 WO 19 WO 18 2. 5 Days 9 WO 17 WO 16 WO 15 WO 14 5 8 WO 13 WO 12 WO 11 WO 10 6 Constraint schedule is in jeopardy! (WO 11 is in the Red Zone! A Red Zone Hole) Watch WO 14 (Yellow) 0 WO 19 OK (Green) Green Time until Scheduled at Constraint © Washington State University-2010 28

Additional Buffers • Constraint Buffer (as we discussed) § Protects the Constraint from running out of work • Finished Goods Buffer § Protects customer delivery from Constraint variation • Raw Material Buffer § Protects the Release of material from suppliers • Assembly Buffer § Facilitates speedy flow of products © Washington State University-2010 29

Additional Buffers Buffer Types: Constraint FG RM Assembly Ropes Constraint Finished goods A B C D E RM FG 7 9 5 F G H 8 7 6 8 6 RM Assembly Raw Material Note: The Assembly Buffer is not used in S-DBR anymore. The concept is included here to help you understand Feeder Buffers in Projects (which you will see later). © Washington State University-2010 30

DBR and Human Behavior Issues • Drum-Buffer-Rope overcomes the problems of system structure § DBR de-couples interdependency § DBR allows variability to work in our favor § Protects the throughput capacity of the system • Measurements should be in place to encourage the right behaviors § Maintain Buffers, Increase throughput, Reduce variability, … © Washington State University-2010 31

Distribution System • Unlike a Factory, there is no single person managing all the movements. • Retail Systems include time delay between demand cycles. • Production occurs to forecast. • Delivery Systems focus on efficiency--Transfer in large batches (long time between shipments). • Errors in forecast are magnified ten fold. • Too much of the wrong inventory, too little of the right. • Magnitude of Missed Sales is not Known. © Washington State University-2010 32

Forecast Accuracy Point where the world changes 100% Accuracy of Forecast Effective Response Zone Now ---> Death Response Zone Future © Washington State University-2010 33

Pushing Inventory to the Retail Store BEFORE Manufacturing Warehouse Distribution © Washington State University-2010 Stores 34

Locate Inventory Where it Provides Best Protection After-Fast Production-Fast Delivery Aggregated Variability Manufacturing Warehouse Distribution © Washington State University-2010 Stores 35

Supply Chain Processes • Supply Chain is made up of many independent links (Businesses or Business Units) • Individual links do not provide a completed product • There is significant interface problems § Timing, Quality, Price, Value • Links are in competition with each other / Leverage each other © Washington State University-2010 36

Typical Supply Chain Raw Materials Refine / Prepare Distribute Produce Retail © Washington State University-2010 Transport Customer 37

Long and Short Duration Supply Chains Farmer Cannery Wholesale Retail Customer Dairy Cows Creamery Deliver Retail Customer © Washington State University-2010 38

Complex Combinations Brakes Car Lot Tires Bumpers Upholstery Manuf. Car Engine Transmission © Washington State University-2010 Car Lot 39

Dedicated Chains Mine Smelter Independent Business Unit Rolling Mill Product Independent Business Unit Steel Sales Independent Business Unit Single Firm - Totally Owned Industry - Sole Source Transfer Prices Fixed by Policy © Washington State University-2010 40

Competitive Chains Oil Well Refinery Chemical Plant Cloth Mill Dress Factory Customer Transfer Prices at Market Prices © Washington State University-2010 41

Simple Measures Drive Behavior • These two measures drive the behavior we want in the production line: Throughput Dollar Days (TDD) Says, “Don’t miss a delivery (avoid failure). And, if you do, fix it fast!” Inventory Dollar Days (IDD) Says, “Don’t let Inventory sit around idle in places where it does no good. Quickly move it to where it protects TDD and then reduce it both in quantity and in time held. ” • TDD and IDD become a Drill Sargent Mentality: MOVE IT! © Washington State University-2010 42

TDD and IDD Help Track Down Problems If A caused the problem, it shows in A’s measures Process A B C D RM E FG Capability 7 9 5 8 6 parts/day If B caused the problem, it shows in B’s measures. But, what if A and B were grouped? And they were measured at a Team, would there be even better performance? © Washington State University-2010 43

And What About D and E? Then, what is the logical measure for D and E? Process A B C D RM E FG Capability 7 9 5 8 6 parts/day D and E team to deliver to the customer. A missed customer delivery is TDDs. The IDD (Inventory held * time held) tells how effective the D and E Team is! © Washington State University-2010 44

What if the Customer is the Constraint? We still use strategic placement of protective inventory internally Process A B We protect our distribution with Finished Goods C D RM Market wants 4 parts per day E FG Capability 7 9 5 8 6 parts/day We really could treat (measure) the effectiveness of the whole line as one big team. TDD=Effectiveness in Delivery IDD =Effective use of resources (and tracking improvements) © Washington State University-2010 45

Projects Are Handled the Same Way A B C D Jobs Process Flow Type I A->B->C->D Type II C->A->B->D Type III A->B->B->D Type IV C->B->A->B Each Job type has four Resources A, B, C, D days of processing for each receive work as it the four resources. flows in different Hum? patterns Release one job per day and every body stays busy. Right? © Washington State University-2010 46

Projects Are Handled the Same A B C D Jobs Process Flow Type I A->B->C->D Type II C->A->B->D Type III A->B->B->D Type IV C->B->A->B Internally, B is the There are 16 processes constraint. We can treat on the 4 job times but 6 of them go through it just like the product line B. TDD=Effectiveness in Delivery C and D only have 3 processes. IDD =Effective use of resources (and tracking improvements) Hum? © Washington State University-2010 47

Projects are Balancing Acts! 3 Goals: Quality and Scope Timing and Schedule © Washington State University-2010 Budgeted Costs 48

Then, Things Combine … Quality and Scope Timing and Schedule Budgeted Costs Slippery Slope of Structure / Behavior Precedence Structure Statistical Variation © Washington State University-2010 Human Behavior 49

And, Reality Appears … Quality and Scope Timing and Schedule Budgeted Costs In an Every Changing Environment nce e d e c e Pr re Structu Bumpy Road of Reality Statis tic Variat al ion © Washington State University-2010 an m u H ior v a Beh 50

The Project Dilemma • It Looks Like a Lose-Lose Situation. Meet needs of Commitment in Danger Compensate for Early Mis-Estimates Not Jeopardize Other Original Commitments Not Compensate for Early Mis-estimates Meet Original Commitments © Washington State University-2010 51

Resolving Project Problem Options Add more time & money and decrease scope Meet any Commitment in Danger Compensate for Early Mis. Estimates Not Jeopardize Other Original Commitments Not Compensate for Early Misestimates Meet Original Commitments Use our Safety Buffer Correctly © Washington State University-2010 52

Consider the Aspects of Projects Good Statistics Central Limit Theorem (add enough things together and everything looks normal) © Washington State University-2010 53

Typical Activity Duration Normal Duration Time Standard Deviation Project Task Duration Time Mean Median 50% Probable 85 -90% Probable © Washington State University-2010 54

So, what is the Behavior? • Engineering Pessimism: Estimate a safe value (85%) Level of Effort • Engineering Optimism: I’m good, I can beat 50%. • Student Syndrome: “Why start now? It isn’t due until Friday? ” (There is more urgent homework or parties. ) • Parkinson's Law: Work Expands to full the time available (Just keep tweaking! More is better!) Assigned Date Time--> • Empirical evidence shows most tasks complete on or after the due date © Washington State University-2010 55

Engineering Perpetual Motion (overtime) Actual Work Load Level of Effort Assigned Date Normal Work Load Time--> © Washington State University-2010 56

The result is Bad Multi- Tasking Your Work A 1 Others Awaiting You A 3 A 2 Your Work Project Manager A Ten Days Each Task B 1 B 3 B 2 Project Manager B © Washington State University-2010 57

Politically Correct Schedule 30 Days Flow A 1 A 2 A 3 B 1 10 B 2 20 30 © Washington State University-2010 B 3 40 50 58

More Like Actual Schedule 40 Days Flow A 1 A 2 A 3 B 1 B 2 B 3 10 20 30 © Washington State University-2010 40 50 59

Elements of the Project Management Solution • Prioritize • Don’t Schedule Conflicts • Avoid Bad Multi-Tasking • Don’t Release Too Early/Too Late • Buffer Critical Chain § Buffers: Project / Feeding / Resources • Schedule 50% Estimate Completion • Communicate “Time Remaining” • Negotiate Capability Not Dates • No Milestones © Washington State University-2010 60

TOC Flow Time 20 Days Flow A 1 A 2 A 3 B 1 10 20 B 2 30 © Washington State University-2010 B 3 40 50 61

Don’t Schedule Conflict Before After TOC Leveling © Washington State University-2010 62

Buffer the Project and NOT Individual Activities Before with 85% Estimates The original Estimats with 50% Individual Buffers Task Buffer TOC Aggregated Schedule with a Project Buffer Task Buffer © Washington State University-2010 63

Protect the Critical Chain Project Buffer Feeding Buffer © Washington State University-2010 64

Communication Buffer for Resources on the Critical Chain Lt. Green be ready Buffer Pink get ready Blue be ready Buffer Green be ready Cyan Resource be ready Project Buffer Feeding Buffer © Washington State University-2010 65

The Simple Line Diagram Was Too Simplistic RM FG FG Aircraft assembly is more of an “A” Plant RM RM © Washington State University-2010 RM 66

The “A” Plant Has Some Long Duration Processes FG RM RM RM © Washington State University-2010 67

Pull Tight the Longest Path (and Shake) Fastest Possible Flow Time RM FG Critical Assembly Joins RM RM RM © Washington State University-2010 RM RM 68

How Could we Fairly Measure Feeder Chains? RM FG RM RM RM TDD-On Missed Delivery to main line RM RM IDD-On Effective Use of Resources RM (and monitoring improvements) Could this also apply to suppliers? © Washington State University-2010 69

Washington State University’s Engineering Management Program Teaches All This and More Finance TOC Capital Projects Thinking Uncertainty Processes Investment Physical Measures Systems Behavior Projects People Full Theory Organizations Scheduling Performance Manage Measurement Quality Assignments Design for Quality Experiments © Washington State University-2010 Operations Optimization Simulation Decisions Reliability Supply Chain Strategy Corporate Departmental Subordination Focus 70

WSU-Engineering & Technology Management “Engineering Business” • Lectures in Evenings over the Internet (Education in your Kitchen!) • Mature, experienced Faculty • Mature, experienced Students • Compliments Undergraduate and Graduate Programs • Provide VALUE to the Student and Corporation (80+ projects averaged $70, 000 each) (11 Projects over $1, 000 each) • http: //www. engrmgt. wsu. edu/ • jholt@wsu. edu © Washington State University-2010 71