a. Internal cost: Located internally prior to leaving the factory Lost labor & materials, Lost sales (for capacity constraint) b. External cost: Located in the market, no,ceable by public Lost reputa,on, recall, warranty & liability costs Importance of conformance quality and defect preven,on: 1. Defects impact the ROIC = Profit ÷ Invested capital 2. Revenue = Volume x Price, Cost = Material + Labor cost 3. Volume = % of demand fulfilled x Demand 4. Invested Capital = Working Capital + Fixed Capital 5. Working Capital = Cost ÷ Inventory turns 6. Fixed Capital = Volume ÷ Capital Produc,vity (CP) Defects cause: Impact on the ROIC’s components 1. Variability: decrease: CP, Inventory turn, % demand fulfilled 2. Capacity loss: decreases: CP, % of demand fulfilled 3. Appraisal: decreases: CP, increases Material Cost 4. External failure: decreases Price, increases Material cost 5. Internal failure: increases Material cost 6. Decreases in demand Internal defect rates: Impact capacity requested by demand -­‐ Increases the amount of work flowing through a resource -­‐ Impacts u,liza,on of resource à may shi[ bo\leneck -­‐ Rework/ Iden,fy defects at source rather than at the end

-­‐ Total defects & cost reduced, capacity increase

-­‐ Easier to iden,fy the cause of the problem, be\er control PrevenCon beDer than inspecCon, which is non-­‐value adding -­‐ Inspec,on requires capital (Inspec,on costs) -­‐ S,ll requires scrap or rework (Internal costs) -­‐ Never perfect, defects s,ll pass through (External costs) -­‐ Removes responsibility of workers at each step -­‐ Inspec,on may become a bo\leneck PrevenCon: Developing Capable Business Processes -­‐ Costs less than sum of inspec,on, internal and external costs 1. Focus on problem: ROIC impact

a. Map the process flow

b. Iden,fy key variables and impact of defects on process 2. Collect data

a. Define specifica,ons: upper & lower specifica,on limits

b. Measure distribu,on of key variables

-­‐ Varia,ons =/= problem, unless outputs out of LSL/USL

c. Measure capabili,es:

Process Capability Measure, Cp = (USL-­‐LSL) ÷ 6σ

-­‐ Assumes the mean can be easily centered

-­‐ Defect rate harder to measure when defects are rare

-­‐ Required capability depends on:

1. Cost of defects, 2. Cost of improving the process.

Probability of defect:

NORMDIST(LSL,mean,stdev,1) + 1 - NORMDIST(USL,mean,stdev,1)
3. Iden,fy assignable causes:

a. Usage of Pareto Charts: Frequency of assignable causes

-­‐ Helps decide the assignable cause to eliminate first

-­‐ 80/20 rule: 80% of defects by 20% of defect types 4. Eliminate causes/ reduce variability

-­‐ Robust design, Crea,ng consistent process (ISO 9000)

Components of the Toyota ProducCon System: Lean opera,ons 1. Zero Muda (Non-­‐value added ac,vi,es) 2. Zero defects, breakdown, inventory, set-­‐up Cme 3. ProducCon flow synchronized with demand (JIT) à Requires low defect

-­‐ One-­‐unit-­‐at-­‐a-­‐,me flow: Mixed model produc,on, Piece-­‐by-­‐piece transfer

-­‐ Match produc,on demand with Takt ,me/ Cycle ,me

-­‐ Pull instead of push system: Supermarket, Respond to recent demand 4. Quality methods to reduce defects

-­‐ Fool-­‐Proofing (Poka-­‐yoke) and visual feedback i.e cut corner of SIM card

-­‐ Components are designed so there is only one single way of assembling

-­‐ Detect, Stop, Alert: Defects at machine (Jidoka) or assembly (Andon cord)

-­‐ Forces human interven,on, triggers process innova,on

-­‐ Built-­‐in-­‐quality inspec,on, inspec,on occurs at every step along the line

-­‐ Combined with reduced inventory (JIT), defects are exposed at the source 5. Flexibility – Catering to fluctua,ng demand

-­‐ Mul,-­‐task assignment: Workers skilled in handling mul,ple machines

-­‐ Adjustment of capacity to meet takt ,me/ cycle ,me by shi[ing workers 6. StandardizaCon of work: Crea,on of SOP is one way

-­‐ Reduce variability, process won’t need buffer or suffer loss in throughput

-­‐ Quar,le analysis: Comparing top and bo\om quar,les performance

-­‐ 7. Worker involvement: Reducing waste of human intellect

-­‐ Quality circles: Workers meet to jointly solve produc,on problems

-­‐ Con,nuous improvement of process (Kaizen)

-­‐ Fishbone diagram (Ishikawa): Cause-­‐effect

-­‐ Diagonal lines represent main causes, the main line = outcome variable

-­‐ Skill-­‐development, cross training (Skill based salary, life,me employment)

-­‐ Workers are empowered to make process improvements Resource yield: 1 – (Flow rate of defects at resource ÷ Flow rate) OR: Flow rate of successful unit processed at resource ÷ Flow rate Where defects are scrapped, to get X goods, produce = X ÷ Process yield -­‐ Process yield = Product of the yields of each resource in the process Yield determines value of a product: For demand-­‐constrained: Value is independent of defects, sales not affected At beginning: Value = Input cost (Raw materials) At n: Value = (1 ÷ Resource yield of n) x Input value at n + Variable cost at n For capacity-­‐constrained: Products scrapped a[er bo\leneck affect sales

At bo\leneck: Computa,on switches from cost based to selling price based At the end: Value = Marginal extra revenue of each product

PrevenCon: Developing Capable Business Processes (CONTINUED)

-­‐ Costs less than sum of inspec,on, internal and external costs 5. Evaluate results: Same tools used in collec,ng data 6. Monitor conformance:

a. Control chart: Tracks process outputs over ,me

-­‐ USL/LSL usually set at 3σ = 99.7% of process output within limits

-­‐ R-­‐Chart: USL/LSL = Factor for Upper/Lower limit x Average Range

-­‐ X-­‐bar Chart: USL/LSL = Ave. X-­‐bar +/-­‐ Factor for X-­‐bar x Ave. Range

-­‐ X-­‐bar = Mean, R = Range = max – min

-­‐ A\ribute control chart: Uses p = % of defec,ve goods, stdev =

-­‐ Use a\ribute charts: When there are many poten,al defect causes

b. Common cause & assignable cause varia,ons:

-­‐ Common cause à Pure randomness, unpredictable outcomes

-­‐ Assignable cause à Changes parameter of underlying distribu,on

-­‐ Try to assign causes to common cause varia,ons and control them

-­‐ Whether it is Common/ assigned: Based on knowledge of observer

Control charts determine when the process varia,on is common cause -­‐ If distribu,on parameters shi[, managers can inves,gate, assign a cause Control charts evaluate if certain improvements achieve the desired effect Unable to reduce variables (conformance quality) à Focus on design quality -­‐ Create robust design à Allow for input varia,ons w/o varia,ons in output

QM: ConCnuous improvement toolbox: Pull/Kanban, Quick changeover, Quality at source, Plant layout, Batch reduc,on, TPM, 5S , Standardize work, Visual systems, POUS, Quality Circle Employee Involvement (Kaizen) 1.

Sort through and sort out, remove unnecessary item

4.

Standardize: Make it obvious where things belong 2.

Straighten: Organize remaining items

-­‐ Use lines, labels, signs. Make 1,2,3 habitual.

-­‐ Make sure they’re easily and immediately retrievable

5.

Sustain: Create rules, guidelines, cleaning charts 3.

Shine: Keep the workplace neat and clean

-­‐ Give recogni,on for good habits to sustain successes 7 Source of waste (Muda: Non-­‐value adding acCviCes) 1.

Overproduc,on: Producing too much, too soon

4.

Over-­‐processing: Excessive ,me spent on the step 2.

Wai,ng: Idle resources or flow units wait for resource

5.

Inventory: Accumula,on à overproduc,on, failed JIT

-­‐ Low u,liza,on à resource idling, wai,ng for flow unit

6.

Rework: Increases variability and consumes capacity 3.

Transport: Movement of half-­‐finished goods (internal)

7.

Mo,on: Only one “right way” of doing a task

-­‐ Plant layout follows process flow, minimize movement

-­‐ Analyze and op,mize task using ergonomics

Outputs of MRP: Times for related opera,ons, ,metable for orders, changes to orders and opera,ons, excep,on reports and problems, performance reports, planning reports for longer-­‐term decisions, inventory transac,on records Benefits: Material supply is directly linked to known demand: Lower stock levels, higher stock turn over, be\er customer service, higher u,liza,on, less ,me spent on emergency orders, expose hidden problems Benefits: Links tacCcal master planning with operaConal scheduling: Can be used for short-­‐term planning, assigns priori,es for jobs supplying materials and provides early warning Problems: Reduced flexibility due to plan based orders, needs a lot of data manipula,on, systems can be very complex, assumes constant lead ,me independent of quan,,es ordered, ignores the order in which materials are readily made, assumes order of usage specified in the bill of materials is right, lot sizes suggested by MRP can be inefficient (not EOQ), may not recognize capacity and constraints, can be expensive and ,me consuming to implement

Design quality à Design specifica,on (Product or service) -­‐ Target value & tolerance level (customer acceptance level) Conformance quality à Produc,on and delivery process

Just-­‐in-­‐Cme (JIT) principles of the TPS 1. Mixed model produc,on: e.g. ABC, 3:2:1 propor,on

a. Produce AAABBCAAABBC, not AAAAAABBBBCC

-­‐ Creates steady workflow, steady demand

-­‐ Requires aggressive set-­‐up ,me reduc,ons 2. Piece-­‐by-­‐piece transfer: Batch -­‐ 1 unit, Limit transport 3. Pull system: Only replacing WIP inventory consumed

-­‐ Push system: Resources based on forecasts

a. Downstream relays demand to upstream produc,on

b. Shop floor control: Automa,c & Visible

c. Problems: Exposed, crea,ng urgency

d. Reac,on to changes and problems:

Immediate, on-­‐line and visible

e. Opera,ons are physically linked 4. Producing at demand rate:

-­‐ Requires a forecasts of resource capacity requested

-­‐ Capacity must be planned Pull system and Defect reducCon 1. Avoids overproduc,on, inventory buildup 2. Reduces excessive flow ,mes 3. With variability, pull system à Lost throughput

-­‐ Due to blocking or starva,on

a. But this brings a\en,on to produc,on problems:

-­‐ Set-­‐up ,mes, Resource break-­‐downs and defects

Which is usually buffered or hidden by high inventory

Which can be reduced to reduce waste 4. A pull system is supposed to react to actual demand

-­‐ In reality, most systems use forecasted demand first LimitaCons of independent demand methods: 1. Slow movers -­‐ such as spare parts

-­‐ Forecasts = close to 0, result in items never in stock

2. Batch produc,on – Forecasts either too high or low 3. Costs rely on accoun,ng conven,ons 4. Forecasts upon historical, no other sources Assump,ons: Independent demand, demand can be forecasted, stock is always available, costs are fixed distribu,on of variables is correct, cannot control variables Adjusted, Independent o[en works for dependent demand

Components of qualitaCve forecasCng:

a.

Based on observa,ons taken at regular ,me intervals

-­‐ Constant series: Constant with slight varia,ons

-­‐ Trend – Presents opportuni,es for growth

-­‐ Seasonality – Can be both trend and seasonality b.

Causal: Rela,onship that can be used to forecast

-­‐ Principle tool: Regression analysis, find high adjusted R2 c.

ProjecCve: Predict based on previous observed value

-­‐ Detects pa\ers, usually for short term forecasts

Measures of forecast error: a. Mean forecast error (MFE):

-­‐ Measures bias, not accuracy b. Mean absolute devia,on (MAD):

-­‐ Reflects varia,on of average from actual c. Mean squared error (MSE): Analyze the paDerns of demand using moving average -­‐ When demand has no observable trend or seasonality -­‐ Systema,c component of demand = Constant level -­‐ Level in period t = ave. demand over last N period -­‐ Larger N = stability, Smaller N = responsiveness

Analyze paDern of demand with exponenCal smoothing -­‐ When demand has no observable trend or seasonality -­‐ Systema,c component of demand = level

-­‐ Forecast = Weighted ave. of last demand and forecast

= α x Latest demand + (1 – α )(Previous forecast) -­‐ Larger α = Responsiveness, smaller α = stability -­‐ α works as a smoothing constant (usually = 0.1 to 0.2 Both moving average and exponen,al smoothing lag behind trend and only have one parameter Poisson distribuCon: Defined only by its mean -­‐ Good for modeling demands with low means (