term 234

All the actions, both value added and non-value added, currently required to bring a product through the main flows essential to every product: production flow from raw material to the customer, and design flow from concept to launch.

A pencil and paper tool that helps you see and understand the flow of material and information as a product makes its way through the value stream.

What the end customer actually wants, specified from the customer's perspective; the reference against which every action is judged value-adding or waste.

Waste: any activity that consumes resources without creating value for the customer.

Unevenness in production volume or mix, which forces capacity and inventory to be sized for peaks.

Overburden on people or equipment, typically caused by unlevelled demand.

Continuous improvement; incremental change that eliminates waste.

Improvement of material and information flow, requiring a high vantage point; management doing kaizen.

Elimination of waste at the shop floor team level, focused on people and process flow.

Getting one process to make only what the next process needs when it needs it, linking all processes from consumer back to raw material in a smooth flow.

The management philosophy of specifying value, identifying the value stream, creating flow, establishing pull, and pursuing perfection.

The production flow inside one plant, including shipment to the customer and delivery of supplied material — the level at which mapping should begin.

A group of products that pass through similar processing steps and over common equipment in your downstream processes.

The movement of material through the factory, drawn left to right on the bottom half of the map.

The flow that tells each process what to make or do next, drawn right to left on the top half of the map.

How often you should produce one part, based on the rate of sales: available working time per shift divided by customer demand rate per shift.

The shared awareness of pace that exists only when work is released and withdrawn in small consistent increments.

The time that elapses between one part coming off a process and the next part coming off — not the total time a part takes through all steps.

The time to switch a process from producing one product type to another.

The consistent increment of work released and withdrawn at the pacemaker, usually takt time multiplied by pack size.

The time it takes one part to make its way through the shop floor, from arrival as raw material to shipment to the customer.

The sum of the value-adding times for each process in the value stream.

A measure of production batch size expressed as the interval at which a process changes over to produce all its part variations.

Shift length minus breaks, meetings and cleanup — the numerator of takt time.

The fraction of time a machine actually runs when it is called upon to run.

The time a pull loop takes to put withdrawn parts back into the supermarket.

Producing one piece at a time, with each item passed immediately from one process step to the next without stagnation.

A process producing something regardless of the actual needs of the downstream customer process and pushing it ahead, typically to a schedule.

A method for controlling production between flows in which downstream withdrawal determines what the upstream process produces, when and how much.

A controlled inventory of parts used to schedule production at an upstream process; the customer withdraws what it needs and the supplier replenishes.

The specified location in a supermarket where a replenished container must be placed.

A capped, sequenced link between two decoupled processes; when it is full the supplying process must stop producing.

The supplying process produces a predetermined quantity directly to the customer process's order, without a supermarket.

A card or device that authorises production or conveyance of a specific quantity of a specific part.

The one-per-container kanban that tells a process how many of what may be produced and gives permission to do so.

A card or device that instructs the material handler to get and transfer parts, for example from a supermarket to the consuming process.

The one-per-batch kanban that signals when a reorder point is reached and another batch needs to be produced.

The place where kanban are collected and held for conveyance.

Levelling both the mix and the volume of production over a period of time, often using a box with a column per pitch and a row per product type.

Regularly releasing a small consistent amount of production instruction at the pacemaker while taking away an equal amount of finished goods.

The single process controlled by the outside customer's orders; how you control it sets the pace for all upstream processes.

Distributing the production of different products evenly over a time period rather than grouping them into long runs.

Releasing and withdrawing consistent increments of work so the production pace is predictable.

The pull created at the pacemaker by releasing and withdrawing small consistent increments of work.

A delivery route lining up several suppliers or customers so material can move frequently in small quantities.

Inventory held as a hedge against problems such as downtime; should be temporary and released only under strict rules.

Inventory held to protect against sudden fluctuations in customer orders.

A process operating on its own schedule, disconnected from any downstream customer, setting batch sizes that suit itself rather than the value stream.

Adjusting schedules based on physically checking inventory levels; an informal practice that is not a true pull.

Small, in-line equipment scaled to the product family's actual volume so it can join a continuous flow.

A large, fast, shared machine that cannot practically be dedicated to one family or slowed to takt.

An arrangement placing process steps immediately adjacent so operators carry or pass parts from one step to the next.

A segment of the future-state map between pulls, used to break implementation into manageable pieces.

Reducing changeover time so smaller batches can be run more frequently.

A device that prevents an error, such as assembling the wrong component when all variants are kept at the line.

A documented, repeatable work sequence that makes production consistent and predictable to takt.

Maintenance practice aimed at eliminating the causes of unplanned downtime.

A bar chart of total cycle time per process or operator against the takt line.

The smallest increment of work that can be moved between operators when balancing a cell.

The sum of operator work across a cell; divided by takt it gives the required manning.

Focusing the organisation on a small number of specific targets, deployed through the levels.

A more dynamic evolution of policy deployment in which lower levels help formulate policy as well as carry it out.

Blending planning with trial and error: strive to accomplish the plan while accepting that deviations, once rigorously questioned, teach you something.

A periodic review, ideally conducted on the shop floor while walking the flow, that manages the exceptions rather than celebrating the wins.

A teacher whose learning curve you can borrow.

Treating all indirect operations as support for direct operations, with their work tied to the takt and pitch of value-adding operations.

A focused improvement activity; effective only when applied strategically inside a value stream design.

A constraint-focused improvement approach; in isolation it produces the same isolated victories over muda as a kaizen offensive.

A variation-reduction initiative; without a value stream design it fixes visible quality problems without improving the whole.

A computerised system that consolidates customer and shop-floor information and sends instructions to each process; its shop-floor scheduling elements should be removed in a lean future state.

Electronic transmission of orders and forecasts, drawn with a lightning-wiggle information arrow.

An order released to the floor that follows a job through the production process; releasing them early is a source of expediting.

A daily list issued to supervisors to sequence shop orders — a symptom of schedule churn.

Customer projection used for capacity planning; should not be used to schedule the shop floor day to day.

The near-term window in which a customer forecast is fixed.

The customer's next-day requirement, typically transmitted overnight.

The instruction sent to the shipping department stating what must ship and when.

Production triggered by a confirmed customer order, common where configurations vary greatly.

Production to replenish a finished-goods store from which the customer pulls.

Producing to the shipping schedule with no finished-goods supermarket; requires reliability or more safety stock.

Accumulated unfulfilled orders, which tempts firms to quote longer lead times rather than shorten them.

Manual intervention to accelerate specific orders, caused by releasing work too early and by late mix changes.

The icon indicating one area of material flow, ideally a continuous flow; it stops wherever processes are disconnected.

The box under a process recording cycle time, changeover, uptime, available time and other process data.

The icon capturing the location and amount of accumulated inventory, annotated with quantity and time.

The icon used for customers, suppliers and outside manufacturing processes.

The stepped line under the process boxes and triangles that compiles lead time and value-adding time.

The icon highlighting an improvement need critical to achieving the value stream vision.

The map of how the value stream operates today, drawn from information gathered by walking the flow.

The lean design to be implemented within a short period; the most important of the maps.

A one-page plan stating step by step what will be done by when, with measurable goals and named reviewers.

Process level, single plant door-to-door, multiple plants, and across companies — the scales at which a value stream can be mapped.

The number of parts one finished-goods container holds; commonly the basis for kanban size and pitch.

The unit in which parts are held and moved; should be sized for the operator, not the material handler.

A person viewed from above, drawn in the process box to show manning.

The supermarket level at which a signal kanban is sent to launch another batch.

How many times inventory is consumed and replaced in a year; the financial mirror of production lead time.

Processing time divided by production lead time — the ratio the map exists to expose.

The fraction of output that must be scrapped or reworked.

The fraction of parts that clear a process with no rework, measured at the process rather than at final inspection.

Available work time divided by cycle time, multiplied by uptime; a measure of current capacity if no changeovers are made.

The quantity produced between two changeovers.

Inventory quantity divided by the daily customer requirement.

The distance material moves through the plant; a proxy for transport waste.

The fraction of shipments delivered when the customer required them.

The person with lead responsibility for understanding and improving a product family's value stream, reporting to the top person on site.

A coach who helps value stream managers see waste and introduce the practices that remove its sources.

The person who withdraws kanban at the pitch increment and moves containers between supermarket, cell and staging.

The department that receives customer orders, levels them, and sends instruction to the single scheduling point.

The person who drives the lean transformation forward, ideally the one asking the question.

An organisation focused on a product family rather than a department.

A supplier pack unit that cannot be broken; the supplier can deliver any number of full coils but not a partial one.

An operation performed off site, such as painting, which adds transport and lead time in the middle of the stream.

The plant boundary where supplied material arrives and where an inbound supermarket can be held.

Where finished goods are assembled according to the shipping schedule before loading.

Completed product held before shipment, either in a supermarket the customer pulls from or staged directly for shipping.

Partly finished material between process steps.

Supplied material before the first process, drawn at the far left of the map.

Customer-specified packaging that circulates between plants and constrains pack size.

Assigning work elements to operators so the operator times are as even as possible and no element starts before its predecessors finish.

A left-to-right chart of which work elements must be completed before others can begin; the blueprint a balance has to obey.

A work element that must be finished before another can start, constraining which operator can be given the later element.

The sum of the standard element times, being the total work required to build one unit; written Σtₑ.

Total working time in one shift after lunch and breaks are taken out; the T in the manning and cycle time formulas.

The number of finished units the shift is meant to complete; the N that fixes both the cycle time and the number of operators.

Unit time divided by the number of operators: the work each operator would carry if the work divided perfectly evenly.

The largest of the operator times, and so the real interval between finished units, however fast the schedule asked the line to run.

The share of paid operator time spent idle because work was assigned unevenly: the gap between the effective cycle time and the average operator time, as a fraction of the former.

Average operator time divided by effective cycle time; the complement of balance delay, so the two always sum to one.

Two or more identical stations working alternate units, used when one element takes longer than takt and cannot be divided.

A set of operators whose work elements are mutually exclusive from every other set, such as a paint booth or a feeder line, each with its own schedule and cycle time.

A line that builds a component fed into the main line, synchronised to it so the two flows stay in step.

One line building two or more models at the same time, balanced on the work each element takes over the whole shift rather than on one unit.

Whether a given work element is used by a given model; what makes mixed-model balancing differ from single-model balancing.

Assembly where each unit is a customer order with its own combination of features and options, so the element times at a station vary unit by unit.

The share of work content a set of models holds in common; a numeric test of whether they belong on one line and one map.

The parts and quantities needed for one unit, which combined with the balance gives the parts each station must be stocked with for a shift.

The first work element to use a given part, which is what decides the station that part must be delivered to.

Leading at the point of value creation rather than from an office: managers observe the process where it runs and support it, resting on four interlocking elements — visual management, a communication structure, a problem-solving process, and standard-based process observation.

Making the state of a process visible where it runs, so a deviation from the standard is seen by the people who can answer it rather than reported upwards later.

A number carried on the shop floor board to show whether a process is meeting its standard; useful only while it is current, since short-cycle management cannot act on a figure that describes last month.

Managing a process in fixed short intervals — the shift, the hour, the pitch — so a deviation is caught within the interval it happened in instead of at the end of the month.

The gap between what a process is doing and the standard that describes it; the only trigger shop floor management acts on, and the reason a standard must exist before an improvement can be claimed.

Watching a running process against its written standard, which is what turns a standard from a document into a check that the work is still being done the way it was agreed.

The agreed route by which an unanswered deviation reaches the next management level, pulled by the process the way a kanban is pulled by consumption rather than pushed by a report.

A defined sequence for taking a deviation to its cause and a countermeasure, so problems are closed by evidence rather than by opinion.

Going to the place where the work happens to see the process for oneself; the habit that makes every other shop floor practice possible.

A problem, its analysis, its countermeasures and its follow-up set out on one sheet, so the reasoning is visible and can be argued with.

Plan, do, check, act: the loop that makes an improvement an experiment with a result rather than a change with an opinion.

A repeated routine for improving towards a target condition in small steps, each step a PDCA cycle, so the capability to improve is practised rather than assumed.

The state a process is to reach next, described precisely enough that reaching it is a fact rather than a judgement; the object of the improvement kata.

The second pillar of the Toyota Production System: a process detects its own abnormality and stops, decoupling the operator from the machine so a defect is contained where it was made instead of being passed on.

Japanese for lantern: the signal a process raises when it cannot meet its standard, and — as a board combining several signals in one place — the means by which support knows where to go without being told.

Good parts against what the planned working time could have produced, as the product of the availability, performance and quality factors; it deliberately excludes utilisation losses, because time nobody scheduled is not a loss the process caused.

Production time over possible operating time: the share of scheduled time the plant was actually able to run, lost to breakdowns, adjustment and set-up.

Parts produced against the target output for the time the plant ran: what short stops, idling and reduced speed cost, once availability has already been accounted for.

Good parts over produced parts, counting rework as a loss; the third factor of OEE.

Possible operating time over the twenty-four hours of a day. It is the one loss OEE leaves out, and reporting it in place of effectiveness is what makes a plant chase running time instead of good parts.

The document listing every operation a part passes through with its times and resources; the source a current-state map is checked against once the flow has been walked.

The batch size at which set-up cost and holding cost balance. It is an answer to a question lean changes: reduce the set-up and the economical batch falls, which is why quick changeover comes before batch-size arithmetic.

Operating profit against the capital tied up to earn it; the measure through which inventory reduction and floor space released show up as a business result rather than a shop floor one.

The time one operator spends on a unit, including walking, loading and any work done while the machine runs. It is the figure a cell is balanced on, not the machine's automatic time.

The time the machine runs unattended after it has been loaded. In a cell this is in-cycle only when the operator has other work to do while it runs; otherwise the operator waits on the machine.

Operator work that can be done while the machine is running. Moving work into the automatic cycle is how a cell absorbs walking and loading without adding stations.

Operator work that can only be done while the machine is stopped — load, unload, inspect. It sets a floor under the operator cycle that no amount of walking reduction will beat.

The exact number of pieces a cell is allowed to hold so one-piece flow stays one-piece. More than the standard is overproduction; less starves the next operator.

The sheet that plots each operator's manual work, walking and machine auto time against takt, so a cell can be manned from evidence rather than from habit.

A line that stops only at a station boundary when a problem is pulled, so a defect is contained without crashing mid-cycle work.

One operator running several machines in a cell, possible only when walking is short and machine auto time covers the walk to the next load.

A cell laid out so the operator starts and finishes at the same end, cutting walk and making manning flexible: one person can run the whole U, or several can split it, without moving machines.

Moving one unit at a time to the next step as soon as the work is done. Batching inside a cell is still a batch, however small the cell looks on the map.

Operators who can run more than one station in the cell, which is what lets manning follow takt instead of following job titles.

Setup work that can only be done while the machine is stopped. SMED's first move is to stop treating external work as internal.

Setup work that can be done while the machine is still running the previous job — staging the next die, preheating, fetching tools.

Shingo's method for taking a changeover under ten minutes: separate internal from external, convert internal to external, then streamline what must still stop the machine.

A changeover reduced to a single motion, the end-state SMED aims at once internal work has been designed out.

Checking the conditions that cause a defect before the defect can be made, rather than inspecting the part after. Shingo's name for the only inspection that can reach zero defects.

Sorting good from bad after the fact. It finds defects; it does not prevent them, and it does not belong in a future-state data box as a quality strategy.

Sampling after the fact to feed information back. Faster than judgment inspection at reducing defects, still too late to reach zero.

The operator inspects the unit they just made, immediately. Faster feedback than a downstream inspector, still a check after the error.

The next process inspects the last process's work before starting its own. Catches what self-check missed; still not source inspection.

Shingo's system: source inspection plus poka-yoke, aimed at zero defects rather than at an acceptable quality level.

A physical contact, counting or motion-step device that either stops the process when a condition is wrong or prevents the wrong action from being possible.

Shingo's distinction: a process is the path a product takes; an operation is what a person or machine does at one place. Improving operations inside a wasteful process leaves the waste.

Producing without the warehouses that batch-and-queue treats as normal. The aim of TPS as Shingo wrote it, not a side effect of faster machines.

Using a single sheet to make the problem, the reasoning and the proposed action visible so a manager coaches judgment rather than issuing answers. The sheet is the residue; the conversation is the work.

Preparing the roots before the tree is moved: talking the proposal through with everyone it will touch, so the decision when it comes is already understood.

Passing a proposal up and down the organisation so each level adds what only it can see, rather than cascading a finished plan.

An A3 whose left side states the background, current condition, goal and root cause, and whose right side states countermeasures, a plan and follow-up.

An A3 written to recommend a course of action, still built on a current-condition and a gap, so the recommendation can be argued with.

The eight-step process that turns a map into a managed programme: commit, choose the family, learn lean, map current, map future, kaizen plans, implement, and keep the gains.

The dated, named actions that take a future-state map off the wall and onto the floor, loop by loop, with a reviewer for each step.

The way a small change in customer demand becomes a large swing upstream when each process adds its own batch, forecast and safety. The map shows it as inventory triangles growing toward the supplier.

The share of work a process receives that it can act on without going back for missing or wrong information. Rolled across the stream it is the information-flow twin of first-time-through.

The time someone is actually working on the item. Martin and Osterling keep it separate from lead time so a box that is busy for minutes and waits for days cannot hide behind one number.

The gap the timeline exists to show: how long the item waits versus how long it is worked. A practical map states both, or it is a process list.

Work the customer would pay for, as distinct from necessary non-value work (compliance, changeover) and pure waste. A data box that only has cycle time cannot make that cut.

Leaving wasteful work elements out while the cell is still on paper. Rother and Harris insist on it before any machine moves, because waste designed out costs nothing and waste built in has to be removed twice.

The ladder from a machine that only cuts to one that loads, cycles, unloads and ejects unaided. The great divide is auto-eject: below it an operator must stand and wait for the part, above it the operator can walk to the next machine, which is what makes multi-process handling possible at all.

The machine returns the finished part on its own. Rother and Harris call for it wherever the operator needs both hands on the part, because without it the part is double-handled and the cell pays for the same motion twice.

Loading every operator but one to nearly a full takt and leaving the remaining slack on a single operator, rather than spreading the idle time thinly across the cell. Spread evenly, the waste is invisible and permanent; concentrated, it is the next kaizen and can be removed.

One worker taking a part through several *different* processes, laid out in flow order. Shingo's worked example puts two workers through three processes and gains 27% on output per worker without buying a machine, because the part stops waiting between operations.

One worker minding several machines doing the *same* operation. It raises machine utilisation and leaves the part waiting between processes, which is why Shingo treats it as the weaker of the two arrangements and multi-process handling as the aim.

Carrying out dissimilar operations in one smooth movement rather than as separate scheduled steps. Shingo puts it alongside levelling as the way a plant absorbs mix without holding stock between operations.

Separating worker from machine and giving the machine the ability to stop itself, without buying full automation. Shingo's point is that almost all the benefit of automation comes from this step, and almost all the cost comes after it.

Load is the work that has to be done; capacity is what machine and operator can do. Toyota calls holding the two in balance "levelling", and Shingo's instruction for reaching it is to round down the peaks and fill in the valleys at the final process.

The wait a finished piece spends beside the rest of its lot before the lot moves. It is invisible in cycle time and it is most of the lead time in batch production, which is why Shingo separates it from process delay.

The wait a whole lot spends before the next process can take it. Shingo pairs it with lot delay so a map can say which of the two a countermeasure would actually remove.

Replenishing when stock falls to a calculated point rather than to a schedule. Shingo treats it as the ancestor of kanban and is explicit about the limit: it regulates quantity well and does nothing about overproduction between the points.

A poka-yoke that stops the machine or the line when the condition is wrong. Shingo's preferred regulatory function, because it does not depend on anyone noticing.

A poka-yoke that raises a buzzer or a light and lets production continue. Shingo rates it a weaker regulatory function precisely because defects keep coming if the signal is missed.

Detecting the wrong condition by physical contact — a pin, a limit switch, a shape that will not seat. The most robust of the three setting methods because it needs no counting and no attention.

Detecting an abnormality by checking that a fixed number of motions or parts happened — ten strips of tape, six screws. Shingo's answer to the operation that is right only when it is complete.

Detecting an abnormality by checking that the steps happened in the required order. It catches the omission that leaves a good-looking part with a step missing.

Checking every piece rather than a sample. Shingo's argument is that this is affordable only when the check is built into the process as a poka-yoke; done by people it is expensive, tiring and still misses defects.

Preliminary: internal and external setup are not distinguished. Stage 1: separate them. Stage 2: convert internal to external. Stage 3: streamline every remaining element. The order is the method; skipping to stage 3 buys minutes where stages 1 and 2 buy hours.

A fastening that holds with one motion — pear-shaped hole, U-slot, cam — instead of a bolt that has to be wound out along its whole thread. Shingo's point about bolts is that only the last turn tightens and only the first loosens; the rest is waste.

Holding the die or tool without threaded fasteners at all — wedges, springs, magnets, taper pins. The end of the road that functional clamping starts down.

A standard plate the next tool is set up on off-line, so the machine only ever receives one shape. It converts centring and squaring — classic internal work — into external work.

Making every die the same working height so shut-height adjustment disappears. Shingo is emphatic that eliminating an adjustment beats automating it.

Two people working the changeover at once, front and back of the press, so a twelve-minute setup becomes six rather than twelve minutes of one person walking around the machine.

Replacing "set it and try it" with fixed settings, gauges and stops. Adjustment is the part of a changeover that has no fixed duration, which is why it dominates the times nobody can predict.

Changing what is made and how it is designed to be made, rather than working faster at what already exists. Shingo pairs it with passive improvement and is clear which one has the bigger ceiling.

Accepting the object and the process as given and improving the motions inside them. Necessary, bounded, and the trap a plant falls into when it never asks whether the operation should exist.

Making many variants in small quantities to order. Shingo's treatment is the answer to the objection that levelling only works for high volume: it is changeover, not volume, that decides whether mixed models can be run.

Resources held ready to absorb variation in what the customer asks for, so takt can still be met when the volume is not exactly what was planned. They have to be available immediately — a buffer that needs a budget conversation is not a buffer.

Resources held against the plant's own failures — an absent person, a system down — as opposed to buffer resources, which cover variation in demand. Keeping the two separate stops one shortage being explained away by the other.

A board that releases work by pitch increment and shows, within the increment, whether the pace is being held. Its virtue is that a problem is visible in one pitch rather than at the end of the day.

The person who moves work and kanban on a fixed route at the pitch. Tapping and Shuker's image is a post office: the heijunka box holds the mail and the runner delivers it on a timetable.

A controlled store of work between two office processes, picked from rather than pushed into, so the downstream process takes what it can do and the upstream process replenishes what was taken.

The few measures a value stream is actually steered by, chosen so each one can be broken into components that name a specific waste. A measure that cannot be decomposed cannot direct an improvement.

Authority earned by taking responsibility at the level where the work happens, rather than issued with a job title. Shook separates responsibility from authority deliberately: the A3 owner has the first and has to pull the second.

The normal condition of an A3 owner, and the point of the method rather than a flaw in it: agreement has to be won from the people who hold the resources, which is what makes the reasoning on the sheet matter.

Total process time divided by total lead time, printed on the map header beside rolled %C&A. On the book's own example map it is 3.9%, which is the number that ends the argument about whether the stream is busy or waiting.