The goalThe fundamental for a manager is to specify the goal pursued by the organization, and to induce everyone to think how at their level they are contributing to it. Productivity and measurement make no sense if one has not specified what 'The Goal' is. Strategy, infrastructure, measurements, rewards all should be aimed toward this goal. That is the most important, even more important than making money, it is know what the ultimate goal is. make money in pursuit of the goal. On financial measures The first measure for a plant manager is to make money (in the pursuit of the goal). Traditional measures of making money are: NET PROFIT (an absolute measure), ROI (a relative measure), and CASH (a lifeline/oxygen). These need to be translated into operational measures that can be used to manage a plant. Three operational measures are mentioned:- throughput: rate at which money is generated through sales- inventory: investments into plant's products that will become throughput in the future- operating expense: all the money needed to convert operating expense into throughput The relation between these are as follows: increasing throughput and reducing operating expense all contribute to increasing net profit, ROI, and cash flow. Net profit = throughput - operating expense. The relation for inventory is different: inventory is paid for by operating expense. Determine the amount of inventory which maximizes net profit. Note that reducing inventory might actually reduce future throughput, hence this should be handled with care. The agenda is set: maximize throughput and reduce operating expense. Determine the level of inventory which maximizes net profit. Goal : increase throughput while at the same time decreasing inventory and operating expense On bottlenecks An hour lost on the bottleneck is an hour lost on the entire system. System rate equals bottleneck rate. The bottleneck determines throughput. A system typically has very few bottlenecks. An hour lost on a non-bottleneck resource is just a mirage. Bottlenecks determine throughput and inventory in the system. Identify your bottlenecks. Exploit your bottlenecks to the fullest. Subordinate everything else to the decisions taken to better exploit your constraints. Elevate your constraints. Until at some point, another constraint arises. Start again. The utilization of a non-bottleneck resource is determined by the bottleneck, and not by its own capacity. The bottleneck controls the flow of all materials. It sets the pace (drum), it pulls materials to its buffer, protects itself with buffers. Variability is bad (see Queuing theory above) One should balance flows (variability), not capacities, ideal factory is not balanced To optimize a bottleneck: Prevent it from starving (buffers, priorities); ensure that it does only useful work (ex: quality controls before it); exploit it to the fullest; organize production so that the bottleneck pulls the production On the notion of a balanced factory An ideal factory is not balanced. Statistical fluctuations and dependent events will make a balanced system perform poorly (see XCELL simulations). They only incease inventories (if allowed), operating expense, and decrease throughput. One should balance flows, not capacities. Look at the bottleneck, and adjust flow to demand. A process of on-going improvement: 1. Identify the systems constraints. 2. Decide how to exploit the system’s constraints (usar ate ao limite, mas ter em atencao a perda de proditividade por a capacidade de utilizacao se aproximar de 100% 3. Subordinate everything else to the above decisions 4. Elevate the system’s constraints (increased capacity, an old machine that is recovered, new machine, overtime) 5. Warning: If in the previous steps a constraint has been broken (something else is now a constraint), go back to step 1, bit do not allow Inertia to cause a system’s constraint Manager abilities: 1. What to change? 2. What to change to? (consider negative effects) 3. How to cause the change? Inventory economics The designations are the same as in Inventory Economics: M = maximum inventory level (units) R = re-order inventory level (units) Q = quantity ordered (units/order) NR = expected number of backlogged units per re-order cycle (units/re-order cycle) NE = expected number of excess inventory units per re-order cycle (units/re-order cycle) Q/D = average length of the re-order period (year) Pi = probability that demand is greater than or equal to inventory level I Demand D = average annual demand (units/year) d = average demand (units/time) d = standard deviation of demand (units/time) Lead time L = average lead time (time) L = standard deviation of lead time (time) DL = average demand during lead time (units/lead time) DL = standard deviation of demand during lead time (units/lead time) Replenish Rt = average replenish time (time) Rt = standard deviation of replenish time (time) DRt = average demand during replenish time (units/replenish time) DRt = standard deviation of demand during replenish time (units/replenish time) Costs TC = total annual inventory cost ($/year) C = annual inventory cost excluding the acquisition cost ($/year) S = fix cost of placing an order ($/order) c = variable cost per unit when placing an order = unit acquisition cost ($/unit) r = opportunity cost of capital (%/year) h = physical holding cost of one unit of inventory per year ($/unit/year) H = r c + h = marginal holding cost of one unit of inventory ($/unit/year) b = backlog cost per unit of unsatisfied demand ($/unit) e = cost of excess inventory ($/unit) 2 Re-order policies in continuos review Uncertain demand and lead time This model is based on the following assumptions: A single product. Continuously review inventory, meaning that the inventory level is known at every point in time. Demand for the product is normally distributed. Lead time is normally distributed. Inventory is replenished by placing orders, which arrive all at once. Inventory procurement costs consists of a fixed cost to place an order, and a variable cost for each unit ordered. All unsatisfied demand is backlogged at a variable cost per unit. The difference between this model and the model with constant lead time will be just the determination of the optimal re-order level. Nevertheless the total procedure and equations to solve a problem of this type is presented bellow. The determination of the optimal reorder quantity (Q) and optimal reorder inventory level (R) is done by the following procedure. Step 0 - Determination of DL and DL (if not known) DL d L DL L d2 d L2 and Step 1 - Determination of an estimation for Q (Q1) Q1 This is done by using EOQ order size: 2SD H Repeat steps 2 through 4 until both differences between (Ri+1 - Ri) and (Qi+1 - Qi) are insignificant: Step 2 - Determination of an estimation for R (Ri) H Qi P D L R P Z Z R i) determination of ZR bD R i D L DL Z R ii) determination of Ri Step 3 - Determination of an estimation for NR (NR i) N Ri DL e 2 Note: N Ri 1 DL 2 Z 2R 2 DL Z R P Z Z R e x DL where P Z Z R H Qi bD 2 2 2DL x R dx and R Step 4 - Determination of an estimation for Q (Q i) This is done by using EOQ order size: Qi ZR R DL DL 2 S b N R D H 3 The total expected annual inventory cost will have the following five components: D 1. Ordering cos ts S ($/year) Q Q 2. Cycle stock cos ts H ($/year) 2 3. Safety stock cos ts R D L H ($/year) D 4. Stockout cos ts b N R ($/year) Q 5. Aquisition cos ts c D ($/year) So, the annual inventory costs will be: D Q D C S H R D L H b N R Q 2 Q and D Q D TC S H R D L H b N R c D Q 2 Q The maximum inventory level will follow a normal distribution with the following properties: Expected max imum inventory level Q R D L S tan dard deviation DL 4 Session #1: Operations management in a service context (Shouldice Hospital) Concept: Process Management: the “Focused” Factory (small differences between customers); managerial issues in a line flow: high utilization, limit variability, dedicated equipment, minimal planning (scheduling arrivals of patients and of those operating the process). Very limited variability in the process enables Shouldice to overcome the trade-off between customer service quality (almost no congestion) and utilization reduced variability on the system, focus on 1 task (=standardized); pre-selection of homogeneous easy cases of ingenual hernias; 1 proven technique. reduced variability in arrivals: arrival time in the morning, 35 people reduced variability in inputs: quality control (self-diagnosed, pre-selected patients) flexibility of doctors through cross-training, to avoid bottlenecks. flow regulation: additional work requirements are absorbed by the workforce incremental improvement due to specialization: high volume of similar cases means steep learning curve, so better intervention & quality of treatment. better ability to manage variability through diagnosis. process risk: radical technological change. what, if anything, makes it so hard to replicate the system? it is more than a technique (culture = customer involvement with exercise and reassuring newcomers; strategy = tight quality control (pre-selection) reduces variability, use of same techniques; infrastructure) customers are part of the operational system (characteristic of services) A perfect process for a very standardized product (DJC, etc.) Session 1: Applying operations management in a service context (Shouldice Hospital). mapping a process: the flowchart, experiencing the process as the customer does, what is the process i (as employee) am contributing to?. an example of a line process in a service context. the chararacteristic of services: customers form an integral part of the operational system (this is particularly well illustrated here). managerial issues in a line flow: high utilization, limit variability, dedicated equipment, minimal planning (scheduling arrivals of patients and of those operating the process), standard relations with suppliers . in sum: a world of constancy, repetition, and incremental improvement. advantages of the focused 'factory': the system focuses on 1 task (system = task), minimize tradeoffs, clarity in vision, and in execution, no confusion, almost no - the world champions of ingenual hernia operations. analysis of system throughput times, and of potential bottlenecks. flow regulation: additional work requirements are absorbed by the workforce . process risk: radical technological change. what, if anything, makes it so hard to replicate this system? 5 Session #2: Ordering a product - Scheduling(EOQ/EPQ) The simple economics of ordering a product or scheduling of a machine. the basic ordering model: EOQ. tradeoff between order cost and inventory costs. economies of scale in distribution, cost savings due to setup cost reduction. EOQ with space constraints (ex 2), illustrating organization by normative regulation (EOQ) and by pricing (space)Optional (not for final exam):. setup time reduction. multi-product EOQ: coordination of production on a single machine. the simple economics of a production/distibution system Concept: Inventory systems EOQ/EPQ Basic ordering model: EOQ: Minimize total cost (Inv holding cost + ordering cost) Order/Production quantity EOQ = Q = sqrt(2SD/H) (where: S: order cost; H: annual holding cost, D: annual demand) Cycle time T = Q/D; Number of reorders = D/Q HowTo: For problems more complicated than the basic EOQ, just express the new constraint in terms of costs (e.g. space, stockout penalty) and mimic the EOQ formula. Several inventory models: Continuous review/ Periodic review EOQ curve is flat around EOQ optimum => no great precision needed. 6 Session #3: Protecting against demand variability (American Hospital) Structure of a continuous review inventory management policy: reorder point (R), order quantity (Q). the economic approach to setting the policy parameters or controls: minimize the expected cost function (inventory cost, order cost, and stockout costs). relation with the EOQ and the newsboy models: cycle stock and safety stock. measuring service quality: by stockout (or backorder) frequency, by the expected number of stockouts (or backorders) per inventory cycle, or by the expected number of stockouts (or backorders) per year. measuring economic performance: computing the various cost components of the total cost: inventory holding cost, setup or order cost, and stockout (or backorder) costOptional (not for exam):. periodic review: similarity of arguments, different controls Concept: Continuous review model; managing uncertain demand ------------------------------------------------------------------HowTo: Minimize total cost (TC=ordering cost + cycle stock holding cost + safety stock holding cost + stockout penalty). Adapt the EOQ formula. Find cost minimizing R: Find optimum probability of stockout (analysis per cycle) marginal benefit = marginal cost, P(DL>R) * b = H (Q/D); take E(units short per cycle) from table, stockout penalty = E(uspc)*D/Q*b What you control in the inventory policy: Q (order quantity) and R (reorder point). What you don’t control: demand (follows a normal distribution) Reorder point R = L + ZR * L (average demand during lead time + safety stock), with L = sqrt(L)*1 (portfolio theory); Safety stock = R - L Service quality: stockout frequency (i.e., ns/Q: the fraction of back-ordered demand), expected number short per cycle or expected number of stockout per year 7 Session #4: Queuing Theory (XCELL) Session 4/5: Factory physics: flows and congestion XCELL:.simulations: examining product flows in serial production environments. effect of variability on throughput, utilization, and cycle times. value of buffer inventory. imbalanced lines: examining the effect of unequal speeds QUEUEING THEORY: . explaining congestion in environments with sufficient average capacity, or the problem of serving clients when service is variable and arrival times are as well. a simple memoryless service, memoryless arrival model. explaining the 'multiplier effect' that appears when utilization approaches 100%: the multiplier that needs to be applied to the service (or value-adding) time becomes infinitely large and explains why even short jobs have to wait a lot (on average). the tradeoff between server utilization (or efficiency) and service quality (or waiting time) Concept: Explaining congestion when there is enough capacity. Two extremes: Memoryless (exponential) and constant service/arrival rate. Average length of the queue: Lq = UTIL2 / (1- UTIL) (expo., 1 server, 1 queue), where UTIL = AR / (n* SR), where AR = lambda = arrival rate = 1 / average inter-arrival time, SR = mu = service rate = 1 / average service time, n = # of servers Average number in system: Ls = Lq + UTIL; Average waiting time in the queue: Wq = Lq / AR; Little’s Law (time in the system): Ws = Ls / AR The multiplier: ratio of time in the system (Ws) to value-adding time (1/SR) (= 1 / [ 1 - UTIL ], therefore UTIL -> 100% then multipilier -> infinity). Trade-off between server utilization (efficiency) and service quality (waiting time). To cope with the complexity, use a computer simulation to model the problem. Variability is bad. It makes your stations starving and blocking Buffers have value only if they build up and down. If the system is not balanced, you only build up inventory (see The Goal). Buffers split/decouple the process The less utilized a station is, the more variability it can absorb. Try to break-up your system into balanced subsystems. Pool stations to absorb variability 8 Session #6: The Newsboy problem (Night & Day) Introduction to capacity choices: base versus peak, fixed versus variable. economic analysis of the decision problem: expected value calculation of various capacity choices, going beyond the expected value: examination of the entire profile of risk, risk aversion: avoiding the large deviations from the expected value. a short-cut: marginal analysis: calculation of the break-even probability of use of the marginal unit (= investment cost/potential value), or of calculating the economic stockout probability. the notion of safety stock (excess of order quantity over average demand). application to inventory theory: look at inventory as an investment which might generate potential value (with some probability), verify the rapidly decreasing marginal value of inventory, invest into inventory until you break-even on the last unit. pricing out information: expected value of (perfect) information. the basic problem to be solved: the lack of fit between a relatively rigid technology (day crews) and a variable demand Concept: Capacity choice (base/peak, fixed/variable, safety stock), economic and marginal analyses Economic analysis: Minimize Expected cost (E(C) = pi * Ci); use the payoff matrix (line: actual demand; column: what you can provide) Marginal analysis: invest in an extra unit while expected benefits - expected cost > 0 (-c + V*p >0), where p is the cumulative probability Value of perfect information: Expected value of the best payoffs (i.e, with the perfect information E(C) = pi * min(Ci) - Expected cost without the info. Concept : introduction to capacity choices: base / peak, fixed / variable, problem of stockout. Use economic analysis of the decision problem: minimize expected value of costs. Beyond expected value: examination of the entire profile of risk, risk aversion: avoiding the large deviations from the expected value. Invest in inventory until marginal value = 0 (or just < 0), expressed by : Expected value of investment = -C (investment in one more unit of resource) + p (probability that demand will be above number of units purchased here day crews) * Value (cost) of stock out (here the fact that you have to use night crews). Notion of safety stock (excess of order quantity over average demand). Pricing out information: expected value of (perfect) information. Basic problem to be solved: the lack of fit between a relatively rigid technology (day crews) and a variable demand. 9 Session #7: Analysing Complex Tasks (Chambord) Concept: Managing complex projects (Gantt, PERT; CPM) project process: interrelated set of activities, uniqueness (contrast with Shouldice) trying to even out use of resources, avoid unfeasibility (resource need>capacity) project analysis: task diagram, precedence, tasks on arcs or tasks on nodes critical path: sequence of activities that takes longest time to complete (no slack) ES, EF (ES+t), LF, LS (LF-t); slack = LS-ES or LF-EF cost/time (penalty costs) tradeoffs: crashing generally doesn’t compensate defend from uncertainty: regular monitoring and revision, try to absorb uncertainty off critical path (vulnerability of dense set of critical paths to uncertainty) need to revise regularly as you have more information (tasks, precedences, etc) common problems in project management: insufficient; excessive use of crashing; biased estimates determinants of project success: 1) context: market pull (importance to the market); organizational push (importance to the organization); culture (are we good at it?). 2) definition: 80% of cost determined at the design stage; complete specifications; understanding of objectives and characteristics of process; challenge & motivation. 3) evaluation: outside evaluation; experience; ability to learn. 4) resources. 5) methodology (analysis, milestones; evaluations; learning by doing. why do projects often come in so late (maximize utilization; the problem of common resources) and so much over budget (the Chunnel example)? Session 7: Analysing complex tasks (Chambord). illustrating a project process: specifics, contrast with Shouldice. project analysis: task diagram, precedence, tasks on arcs and tasks on nodes (the latter is more natural). critical path, computation, ES, LS, EF, LF, slack. crashing and cost/time tradeoffs. discussion of uncertainty: regular project schedule revision, try to absorb uncertainty off critical path, vulnerability of dense set of critical paths to uncertainty. the problem: you need to revise regularly as you have new information on task times, precedences, etc. common problems in project management: insufficient monitoring, excessive use of crashing, biased estimates. why do projects often come in so late (adler et al. article: the problem of common resources) and so much over budget (the channel tunnel example)?. importance of project definition and corporate alignment: what does this project do for the corporation, who pulls it out of the organization from the outside? 10 Session #8: Process Flows & Identifying Bottlenecks (National Cranberry) Concept: an example of a line process in an industrial context Flowcharting and identification of bottleneck(s) and potential bottleneck(s): separator and dryer (if %wet is high), possibly bagging (multiplicity of shipping choices should minimize the risk) Differential roles of bins in peak (excess supply storage) and off-peak (storing of work so you can run the facility at reasonable efficiencies) Bottlenecks are a function of the input mix Identification of key issues: - peak sizing: which fraction of the peak are we going to build our hardware for (we have more flexibility with staffing)? - the problem of truck waiting time and of overtime Trade-off between queues/waiting time and utilization/idle capacity model 1) consider berries only, take average of 20 peak days, doesn’t explain queues model 2) separate wet and dry berries, still does not explain truck waiting model 3) disaggregate the peak period: the second half of the peak period is wetter, and you are then running out of holding bins - trucks become holders of inventory Anticipate 1971: wet % increases 20% - model 3 forecasts high truck waiting times Possible solutions: flexible staffing, start earlier in peak (-queues =overtime); buy bin (-queues =overtime); buy dryer (-queues -overtime) Session 8: Understanding process flows and identifying bottlenecks (Cranberry). an example of a line process in an industrial context. flowcharting and identification of bottleneck(s) and potential bottleneck(s): separator and dryer (if %wet is high), possibly bagging (except for saying that the multiplicity of shipping choices should minimize the risk of this one becoming a bottleneck -- something to be verified in a second round of analysis). identification of key issues:- peak sizing: which fraction of the peak are we going to build our hardware for (we have more flexibility with staffing)? - the problem of truck waiting time and of overtime. diagnosis of truck waiting times in 1970: try simple answers first, complexify if needed- model 1: consider berries only, take average day of 20 peak days -- you can explain why the line takes so long to process the berries during peak- model 2: separate wet and dry berries, still does not explain truck waiting- model 3: you have to disaggregate the peak period (eg. take first 10 days, and then a second period consisting of last 10 days): the second half of the peak period is wetter, and you are then running out of holding bins -- trucks become holders of inventory. anticipating 1971: wet% increases 20% -- model 3 forecasts high truck waiting times possible solutions: flexible staffing: in peak you can start right away -- this takes care of most of the problem differential roles of bins in peak (excess supply storage) and off-peak (storing of work so you can run the facility at reasonable efficiencies). repetitive nature of analysis: do not be afraid to make simplifying assumptions, see where that leads too, make things more complex only when and if needed 11 Session #9: Resource Planning/Allocation (Sheridan Motors; ABC Company) Concept: Linear programming, aggregate planning Marginal costing needed for decision making, don’t consider fixed costs Once we determine we can cover fixed costs, LP helps determine optimal throughput and product mix given constraints (the optimal throughput is independent of the fixed costs). LP gives the solution, not the explanation. LP provides information for: 1) aggregate planning - what should we be planning for 2) resource decisions - alternative strategies: - chase: adjust production and workforce/overtime to match demand, no inventory - level: constant production or workforce, build inventory 3) sensitivity analysis - test different scenarios and analyse impact optimal solution: corner solution or pt. on object. function tangent to feasible area. LP provides production options, not just optimal solution. Manager must choose option aligned with corporate goal. Sheridan: Shadow price/dual price = economic implication of bottleneck (binding constraint); opportunity cost of one unit of capacity; how much objective function will change with a unit increase of capacity - if shadow price = 0, then there is slack or surplus - if shadow price >0, then slack = 0, constraint is binding (bottleneck) ABC: aggregate planning 1) Objective function: cost = 800W + 300O + 500H + 500F + 2,25I 2) Flow constraints (people): W0 + H1 - F1 = W1; (product): I1 + P1 = D1 + I1 3) Capacity constraints (people): O1 <= W1; 50W1 + 12,5 O1>= P1 4) Capacity constraints (machines): P1 <= 20000 5) Accounting constraints: W = W1 + W2 … + W12; O, H, F, I (same thing) 6) Additional constraints: I12 = I 0; W12 = W0 (restart period 1 after period 12) Session 9: Fundamentals of resource planning and allocation (Sheridan Motors and ABC Company)(This session is summarized in the note summarizing the review session, and containing Sample Exam Questions) 12 Session #10: Planning and Controlling Operations (American Semi-Conduct.) Concept: Planning and Controlling Operations; match processes and products What is 'The Goal'? To what extent is on-schedule delivery important? Identification of key business decisions: extent of standardization in the standard business, and growing importance of customized business Operations in the respective business processes: - standard business: customer pull from the die bank, replenish the die bank using a reorder point policy (customer values speed) - customized business: you are selling engineering, fab time, as well as final assembly and testing (customer can deal with uncertainty, lead time estimates) Importance of planning if both business processes co-exist: determine average customer leadtimes for the fab, then limit congestion in the fab by allowing new orders to be processed only when others are completed (CONWIP or constant WIP policy) Simplify planning and increase customer focus: let marketing decide which order to process next, let production manage throughput once the orders are released Two manufacturing tasks: maximize today's throughput and tomorrow's throughput (learning and yield improvement) Problem of OASIS: conservative (focus on OSD), but market values speed! Benefits of TEMPO: linking OASIS (order entry) with WIPSYS (production and inventory status) for improved production control (leadtime quotation, status update for customers) Limits of TEMPO: fluid flow models ignore congestion due to variability - makes deterministic assumptions with regard to flow Variability of re-entrant flow lines and effect of re-entry on congestion (the 'effective' processing rate of lots is a function of how often they need to re-enter the system). Session 10: Planning and controlling operations (American Semiconductor) What is 'The Goal'? To what extent is OSD important? Identification of key business decisions: extent of standardization in the standard business, and growing importance of customized business. Operations in the respective business processes: standard business:customer pull from the die bank, replenish the die bank using a reorder point policycustomized business: you are selling engineering, fab time, as well as final assembly and testing. Importance of planning if both business process co-exist: determine average customer leadtimes for the fab, then limit congestion in the fab by allowing new orders to be processed only when others are completed (constant WIP policy). Simplify planning and increase customer focus: let marketing decide which order to process next (customer priority), let production manage throughput once the orders are released. Two manufacturing tasks: maximize today's throughput and tomorrow's throughput (learning and yield reduction). Benefits of TEMPO: linking OASIS (order entry) with WIPSYS (production and inventory status) for improved production control (leadtime quotation, status update for customers). Limits of TEMPO: fluid flow models ignore congestion due to variability -- makes deterministic assumptions with regard to flow. Variability of re-entrant flow lines and effect of re-entry on congestion (the 'effective' processing rate of lots is a function of how often they need to re-enter the system) 13 Session #11: Just-in-Time Operations (Bose Corporation) Concept: Just-In-Time (JIT): seamless flow of raw materials to customers, through entire value-chain; improved co-ordination Supplier must do the same thing, work within the same high quality levels. Benefits of bringing suppliers ‘in-house’: 1) Reduce costs (less staff, less inventory, setup cost reduction) 2) Volume, economies of scale when ordering (price, logistics) 3) Alignment, commitment from suppliers (common goal) 4) Suppliers get involved in Engineering, eventually supplier does it all 5) focus on core competency (Shouldice lesson) 6) Supplier is better in R&D in its business, experience with other customers Conditions for success (JIT II): 1)Trust, partnership (supplier/producer) - confidentiality and common goals 2) Location of supplier (close) 3) Mutually beneficial to both supplier (volume) and producer (cost), win-win Kanban: Limit the WIP in the system (Little’s Law) Session 11:Just-in-time operations (Bose Corporation: JIT II) Procurement: the 'virtual' or the 'administrative' factory (it is the part of the factory that handles the items that are not made internally (growing importance due to the organizations refocusing on 'core competences') Benefits and liabilities of the make/buy decision on plastics JIT: concept, goal, implication Benefits and risks of JIT II for Bose A new way of working with suppliers: partnerships for (final) value: joining forces to conquer market share (in finished goods), and their requirements in terms of trust, discipline, Illustration of moving to a service notion of supply and of business simplification in supply relationships 14 Session #12: Identifying and controlling for quality (Dynamic Seal) Concept : Quality and process control; Statistic Process Control (SPC) Process in stat. control: If subject only to random causes (stable, free of assignable causes) Gaining control : investigating => identify assignable cause => eliminating cause from process => reduced variability Control limits : establish UCL, LCL and hereafter sample to catch problems early on in process => reduce rework/scrap and minimize defects => higher throughput, minimize oper. expense, leadtime, inventories and inspection costs. Control charts : R-charts : UCL = D4* R and LCL=D3*R (R = average of several past R values, D3 and D4 = constants providing 3 limits. Xbar-charts : UCL = +A2* R; LCL=-A2*R (R = average of several R values, A2 and A2 = constant providing 3 limits for process mean). Xbar-charts : UCL = +z/sqrt(n); LCL=z/sqrt(n); = sqrt(n)*A2*R(avg)/3 Capability analysis : establish , , distribution through continous examination of output (+statistics), before starting operating process (ideal conditions). Process Capability Ratio (Cp) = (upper specification - lower specification) / 6 Cp < 1 => process not capable; Cp>1 => capable; Cp >>1 capable and robust Process Capability Index (Cpk) = minimum of (( - lower specification) / 3) or ((upper specification - ) / 3) Improve capability : systematic changes (often identified through elimination of assignable causes) Conformance analysis : establish that process still under control through sampling (+statistics) at regular intervals (increasingly spaced as you gain control). SPC build quality into the process and fosters atmosphere of continous improvement. Based on empowerment and on-line monitoring. All staff participating in the production process is responsible for quality control, not only inspectors, creating ownership of the quality control and improvement. Traditional systems react to errors, SPC anticipates, is active. Management role : empower workers to control own processes and decide when to redesign. Worker role: eliminate assignable causes and maintain process in control (produce, control and take corrective action). Cost of quality / non-quality : prevention, appraisal (inspection), internal failure (scrap, rework etc.) and external failure (lost customers etc.) Advantages of introducing gradually (ex: limited part of factory): demonstration value, easier investigations, choose demanding client, learn before extending, get worker’s input and commitment in return Session 12:Identifying and controlling for quality (Dynamic Seal). SPC view of the world (Deming). QC issues in the United Aircraft cell. Preliminary estimation of quality costs (or better costs of non-quality). Quality assurance through inspection (separating good output from bad one). Illustrating capability analysis (housing) . An example of conformance verification through control charting (ring). How do you introduce SPC at Dynamic Seal: the benefits of starting in small part of the plant (demonstration value, investigations are easier, you have a demanding client with clear customer specifications, you can learn from this effort before extending it, you give the workers the chance for their input and you can obtain their commitment in return 15 Session #13: Reengineering (Manzana Insurance) Concept: Business Process Reengineering, redesign the process(clean sheet app.) Move away from functional to process-based operations; streamline/simplify processes; remove obstacles to customer responsiveness (boundary less); eliminate non-value-added time; introduce IT to enable process exploit synergies. Problems: business processes are different and interfering with each other; the incentive system is driving people to pay insufficient attention to policy renewal (too much attention to new business) - less renewals and more new policies mean increased risk and greater claims losses; backlogs of all policies, primarily renewals; underwriting process represents bottleneck; incentive system reduces throughput by favoring slowest policies to process; geographic orientation of underwriters leads to over-utilization of UW1 (98%) and unavoidable bottleneck Solutions: Short-run: process flow organization: pool the UW's to allow much quicker response time.; change incentives to ensure timely (earlier) handling of the renewal business, can still be done second priority if they are started on time. BPR solution: introduce IT for file exchange and sharing with agents - more effective information transmission; DC’s become scanners or disappear; integration of underwriting, rating and policy writing; retain two processes: underwriting and administrative (agent interaction, data-integrity checking, writing, identification of special cases); let UW’s manage exceptions, standardize other activities. Session 13: Reengineering (Manzana Insurance). BPR: what are some of the key ideas . What is the goal? Renew policies or grow the portfolio? If both, realize that the business processes are different, and that currently they are interfering with each other (too much attention to new business). the infrastructure (incentive system) is driving people to pay insufficient attention to policy renewal (which is key to long term performance). short-run process flow organization: pooling of the UW's allow much greater service quality (response time) -- verification through utilization and queueing formulas (of Chapter 9). A key requirement: the effective (and early) handling of the renewal business -- can be done second priority if it is started on time. the BPR solution: introduce IT for customer exchange and file sharing, for more effective information transmission, pooling of operators, integrating of functions, difficulty of integrating the UW's with the administrators, managing exceptions, standardizing others 16 Session #14: Operations Strategy (American Connectors Company) Concept :Operations Strategy; Material Requirements Planning (MRP) Conclusions: link between products and process characteristics; pay attention to business dynamics; mass customization has its limits. There are different operations that should not be mixed: Product-oriented (DJC) and batch-oriented (ACC).If you mix them, none works. Product-oriented: focus on product (simplify it); line flows; standardized/repetitive work; high volume; dedicated/optimized and non-flexible machinery; high utilization / low variability; simple planning (pull); low number of suppliers (close relationship); low customization; low number of references; increase process efficiency; stabilize demand; do not accept all orders; makes money in the late phases of the product life cycle. Batch oriented: focus on the process; layout by functions; jumbled flows; customized/variable work; medium/low volume, performing/flexible machinery (need investment); low utilization / high variability; complex planning (MRP) high number of suppliers (lose relationship); high customization; high number of references; accept all new orders; makes money in the early phases of the product life cycle (higher margins). DJC showed how powerful the continuous learning can be. DJC improves the process, not the product. ACC does the opposite. Session 14: The concept of an operations strategy. Included with returned case write-up of American Connector Company 17 Session #15: Logistics and Distribution (Polaroid Corporation) Concept: whether to implement centralized distribution in Europe have centralized warehousing in the U.S.; currently 12 warehouses in Europe significant country differences (some have strict service requests, different distribution systems, different packaging requests) recent custom simplification and deregulation of shipping presents opportunities for more efficient transportation segmentation by product better than by region (different service requirements)? Options: 1) Third party logistics - different core competence, own process; we won’t be able to deal with the changes - too important, very close to the customer, key to deliver service, not just transport (logistic process: shipment of physical goods, info exchange), lose flexibility 2) Regional warehouses: reduce the # of warehouses by consolidating by region - incremental approach, intermediate low-investment solution, centralize later 3) Direct shipping (gain economies of scale by pooling inventories) - reduced inventories & probability of stockouts: If you combine the stock in one warehouse the std dev = sq root of 12 x std dev of 1 warehouse (assuming cov = 0). Also, assuming you keep the same level of total safety stock, the mean of the big warehouse will be 12 times the mean of the individual warehouse, so the probability of running out of stock is practically 0% (better leadtime/service). This also reduces the costs of transferring products between locations to combat shortages. - lower inventory carrying costs, lower inventories (cycle stock economies - EOQ). - potential cost-savings from staff reductions in warehousing, shipping, & traffic management. - savings from eliminating warehouse rental/ownership costs. - strategic importance of controlling distribution channels. - alternatively, maintain same level of service but drastically reduce stock. - resistance on the part of national warehouses since the warehouses/buffers allow them to provide higher quality service; they don’t see the cost-savings benefits because they’re not being charged for inventories. 4) Keep as is but implement integrated IT - current system not working efficiently, can be improved - start charging for inventory to create incentives for economy - facilitate cross-sharing of safety stock and internal sales by creating transfer prices - common languages on packaging to improve sharing of safety stock - develop integrated IT system Session 15: Logistics and distribution (Polaroid European Distribution System). Economics of scale in distribution Examination of alternatives Short-run improvements in the system that will facilitate change (eg. charge countries for inventory, credit them for internal sales). Logistics: the 'real' factory and the 'administrative' factory (administration). Implementation issues: which country goes first? Do you move to a new system at once?. Why is logistics not paid attention to? . A process view of logistics . Limits of third party solutions. History: what happened?. Bottlenecks in organizations: make sure no function (incl. logisitics) is a bottleneck for the business. 18 Session #16: Capacity and Facilities Management (Eli Lilly and Company) Concept: Operations can be a competitive weapon; Anticipate future environment: higher costs (R&D, marketing, manufacturing); lower prices (Government, generics, price sensitivity, competition); future competitive advantage (brand name, access to the market, manufacturing capabilities vs R&D today). reducing clinical trial time (7 years) implies great savings (uncertainty); to reduce lead time you have to reduce also process development - start before clinical trials 1) Dedicated facility - capacity leveled for peak demand, effective capacity very low (ex: 2 first years) - risk of even lower util, because demand/dosage did not materialise 2) Flexible plant - much higher util, even during kick-off phase - if the product is a blockbuster, then build dedicated facility; more time to get market info, reduce risk, increase util, reduce manufacturing cost - if you go for specialized you can use flexibles as a safety stock of capacity; build plants with lower capacity, use flexible to produce the rest if needed - cost 6,5x higher per kg of capacity, but util of flexible can be very low or zero - process engineering no longer a bottleneck, you get earlier to the market, extra sales - I can have a set of flexibles with different flexibility levels (portfolio effects) Session 16: Capacity and facilities management (Eli Lilly and Company: the flexible facility decision). Fundamental change in the pharma business. Benefits of manufacturing improvements in process development leadtime and in manufacturing cost reduction. Qualitative discussion of the benefits of the introduction of flexible facilities (improved risk management of the specialized facility sizing). Quantitative comparison of manufacturing cost of flexible and specialized facilities. Conservative nature of early cost estimates on flexible facility (ignores learning by doing benefits). Additional dimension: once you are considering a set of flexible plants, they all do not have to be equally flexible (which allows you to further reduce their costs) 19