Blog

2026.11.10

Two-Bin Kanban Line-Side Replenishment System: Implementation Guide for Manufacturers

Two-Bin Kanban Line-Side Replenishment System: Implementation Guide for Manufacturers

A two-bin kanban line-side replenishment system does more than detect an empty container. The signal must reach the supply supermarket, generate the right pick, travel with the right material, and close when the line receives it. This guide is for manufacturing, logistics, and IT teams in Thailand planning that complete loop. It covers item selection, cards and scans, ERP/MES/WMS inventory boundaries, exceptions, RFP requirements, and FAT/SAT acceptance.

Define the process: line-side two-bin replenishment

In a two-bin setup, two containers of the same part sit close to the point of use. When one is emptied, its container or card becomes a replenishment signal while production continues from the other. A material handler picks a standard quantity from the supply supermarket and delivers it to the assigned line location. The cycle closes when the full container is received and the empty container or card is collected. Oracle’s flow-manufacturing guide describes this empty-bin signal and use of the second bin.

This process has a different boundary from replenishing a warehouse pick face from reserve storage: that pick face serves warehouse order picking, while this article’s destination is a production line. It is also different from returning unused surplus material from a line to the warehouse after quality disposition. That is a reverse flow. An RFP should state the forward boundary explicitly: empty bin at the line, release from the supply supermarket, and receipt at the line. Mixing a return transaction into this flow confuses demand, physical stock, and consumption.

Kanban is an authorization and instruction signal in a pull system, not merely a part label. A card can identify the part, pack quantity, source, storage address, and consuming process. Lean Enterprise Institute distinguishes withdrawal kanban, which authorizes movement to the downstream process, from production kanban, which authorizes making more at the supplying process. Define which signal your line-side system creates before designing a screen.

Select items that can actually run on two bins

Do not force every part into a two-bin process. Start with repeated consumption, a stable pack quantity, enough line-side space, and a dependable supply point. Highly variable special parts, oversized materials, or parts subject to strict batch selection may need another method. Oracle’s kanban overview describes relatively stable demand and medium-to-high production volume as typical application conditions.

For each candidate, record part and revision, pack quantity, hourly use variation, replenishment lead time, supermarket availability, route, bin dimensions, rack depth, inspection status, and substitution rules. An average daily use rate alone misses the rush before a break, changeovers, and a late milk run. A single worst-case number applied everywhere creates excess line stock. Observe demand and replenishment by time of day and separate normal service from exceptions before choosing an initial bin quantity. The observation period is a project decision, not a universal standard.

As a design estimate, the remaining bin must cover expected use during the whole replenishment cycle plus a margin for variation. Check the estimate against real pack sizes, container capacity, route schedules, and quality-held stock. If a part cannot wait until the first container is fully empty, move the signal point to the start of use or consider more than two containers. Record the measured lead time and consumption behind that decision, then adjust during a physical trial.

Two-Bin Kanban Line-Side Replenishment System: Implementation Guide for Manufacturers - figure 1

Make the empty bin, card, and scan one state transition

The most damaging gap is an empty container that never becomes a visible request. Pair the physical action with the system action: place the bin in a marked spot, deposit a card, or scan a code. If scanning creates the official request, work instructions must say that placing the bin alone is insufficient. If collecting the card is the authoritative trigger, the scan can record time or confirm identity, with a defined manual process during scanner failure. Turning both actions into independent requests creates duplicates.

The minimum useful journey is full at line, in use, empty/requested, allocated at the supermarket, picked, in transit, received at line, and empty bin/card collected. Not every transition needs another scan. Choose evidence points according to error risk and labor cost. Each request should have a unique ID, part and revision, line and rack location, bin or card ID, quantity, event time, owner, and status. A network retry must retain the same request ID so it cannot create a second pick.

Decide what a scanned code identifies. A part number alone cannot distinguish two physical bins; a container ID alone may not reveal a wrong part loaded inside. Options include a card ID linked to a controlled part-and-location master, or a unique bin ID linked to the part and lot. GS1’s barcode overview explains identification of items, locations, lots, and other attributes. Choose GS1 or internal encoding according to partners and existing systems, and test reading distance and label durability at the actual line.

Separate human-readable card fields from machine-readable data. Consider part, description, standard bin quantity, source, line rack, card number, route, revision, and effective date. When a rack moves, an old card can send a perfectly valid request to the wrong location. A reprint must invalidate the previous card; two live cards must not represent one container. Test mounting and legibility where oil, water, and repeated cleaning are present. When retiring a card, also check racks, collection boxes, transport carts, and the supermarket for old copies. Quarantine any full bin carrying a retired card until its part, lot, quantity, and destination are checked. A scan of a retired card must not create a replenishment request; log its number and discovery location and alert the owner. Share the retirement list across shifts so an old container cannot re-enter the line the next day.

Connect the request to supermarket picking

An empty-bin request must be checked against physical stock at the supply supermarket. Picking every request immediately can multiply walking and split deliveries to the same line. Waiting to batch every request may empty the second bin. Show the request time, next planned route, estimated remaining-bin endurance, and part criticality so a dispatcher can prioritize work. If software proposes a priority, supervisors need a reasoned override with an audit record.

At pick time, verify the supermarket slot, part and revision, lot or expiry where relevant, quantity, and bin condition. Decide whether one request maps to one picked bin or several requests are consolidated before being sorted by destination. In the latter case, allocated at source and staged for a specific line are distinct states. A picker must not silently substitute another part or lot; engineering, quality, and planning approval rules govern substitutions.

A supermarket is an upstream supply point that replenishes what downstream users withdraw. Lean Enterprise Institute’s supermarket definition describes a card or empty-bin signal returning upstream after withdrawal. Line-side software cannot manufacture stock that the supermarket lacks. Replenishing the supermarket itself belongs to another production kanban, purchase process, or reorder-point process. For that upstream decision, see our reorder-point system implementation guide.

Confirm handoff and receipt instead of assuming delivery

Picked is not delivered. A bin can be on a cart for the wrong line, left near the point of use without being placed on the correct rack, or marked received before physical arrival. Specify where release from the supermarket, carrier handoff, and line receipt are recorded, including time and owner. Decide whether each bin requires a scan or a route-level receipt is acceptable, based on misdelivery risk and labor. A colored container alone is weak identification in mixed-model or night-shift operation.

At receipt, compare rack location and container tag, then display a concise part, destination, and image on the handheld. If the network is unavailable and material still moves, issue a numbered exception record. On recovery, record both the actual physical receipt time and the later data-entry time. Do not present an after-the-fact entry as if it were a real-time scan. An event ID helps detect duplicate entry and premature receipt.

The line should check destination, part, revision, lot, quantity, and visible condition. A bin under quality hold cannot become available line stock just because it arrived. Plan when an empty container and card are collected; otherwise an old card can trigger another request on the next route. Either close collection in the same workflow or monitor it as an exception.

Two-Bin Kanban Line-Side Replenishment System: Implementation Guide for Manufacturers - figure 2

Set ERP, MES, and WMS inventory boundaries first

The hard question is when stock changes state. One design reduces supermarket stock and transfers it to line stock at issue or receipt. Another records actual component use in MES and then posts or aggregates it to ERP. A backflush site may post theoretical consumption against completed production later. Empty and full kanban scans do not have one universal accounting meaning: postings vary by replenishment strategy and configuration. Define the authorized transaction point so ERP, MES, and WMS do not each subtract the same physical movement.

ERP commonly owns purchasing and financial stock, WMS owns supermarket bins and logistics tasks, and MES owns line use and production reporting. Treat this as a discovery hypothesis, then document the actual system of record, master-data owner, and allowed updates. The OPC Foundation’s ISA-95 overview provides a useful model for the information boundary between enterprise resource and manufacturing operations management, including material information.

At minimum distinguish warehouse reserve stock, available supermarket stock, picked stock in transit, full bins at the line, a bin in use, quality-held stock, and empty containers. An empty bin is not always identical to zero accounting inventory; physical remnants and standard pack quantity can differ. Record variance reasons and decide how the ERP and costing process will handle them. The final financial treatment must follow the company’s configuration and accounting policy.

Design separate events for request, allocation, issue, receipt, consumption, cancellation, and variance correction. Carry unique event IDs, part, quantity, location, time, source event, and result so a retry is recognized as the same transaction. GS1’s EPCIS explanation describes visibility events through what, when, where, and why. A factory need not implement EPCIS to benefit from disciplined event design.

Moving stock between warehouse locations and releasing stock from a supermarket to a production line have different business meanings. Our warehouse location movement guide addresses the former. This guide starts at the line-use replenishment request and continues through receipt. Draw the WMS-to-MES handoff and any ERP posting in the RFP.

Design exceptions before they stop the line

The priority exception list includes an uncollected card, duplicate scan, lost bin, delayed request, empty supermarket location, wrong item pick, late carrier, rejected receipt, quality-held lot, and changed rack address. For each, define who detects it, who decides, who corrects data, and what restores normal operation. “Contact an administrator” is insufficient for a night shift with no administrator on site.

A duplicate scan should show the existing request rather than create another. Cancellation rules differ before picking, after picking, during transit, and after line receipt. Deleting a received request can move book stock back while the material remains physically at the line. Keep the original event and create a correction after checking the physical destination and quantity. The GS1 EPCIS standard similarly describes correction through additional events rather than erasing history.

Stockout is not the same state as request creation. Alert the line supervisor with the estimated endurance of the remaining bin, then follow approved rules for substitution, urgent production, or transfers from another line. Software should execute only an approved substitute, with the reason recorded. For a disconnected handheld, specify which work may proceed, how a paper or offline record receives a unique ID, and how competing offline scans are reconciled. Reconcile bin, card, physical location, and final receiver, not merely the order in which devices reconnect. Rehearse this in SAT.

Measure replenishment quality, not request volume

Useful measures include request-to-receipt lead time, percentage delivered before the second bin empties, line downtime caused by missing material, misdeliveries, duplicate requests, stock differences, lost cards, and emergency-route share. More requests do not prove better usage: bins may simply be too small. Fewer requests may mean excess line stock. Read these measures alongside bin size and production volume.

Split lead time into request-to-pick, pick-to-dispatch, and dispatch-to-line receipt. A total alone cannot show whether delay came from stockout, picking, or transport. Compare shifts, weekends, and changeover days, then confirm top causes at the worksite. Set targets from measured distributions and tolerated line risk, not an unsupported universal number. During reviews, compare digital status with physical racks: a “received” bin may be absent, a “full” bin may carry a quality-hold tag, or yesterday’s card may still sit in a collection box. Digitization exposes discrepancies; it does not replace physical checks.

Put functions, data, and acceptance criteria in the RFP

State the target lines, item count, shifts, supply points, current ERP/MES/WMS, devices, network, and measurable aim. Include peak request rate, mixed-model production, lot control, quality hold, offline work, label maintenance, and API constraints. That lets bidders distinguish configuration from development and price the right scope.

Group functions into empty-bin capture, duplicate prevention, supermarket allocation and picking, transport handoff, line receipt, exception handling, supervisor view, and audit. For each, provide input, owner, resulting state, failure behavior, permissions, and timestamp. Card printing, replacement, and invalidation belong in scope. Ask bidders to walk one real part and container through a complete cycle, not only demonstrate attractive screens.

Nonfunctional requirements include operating hours, offline tolerance, retry behavior, concurrent users, device durability, barcode readability, access controls, backup, and log retention. Test peak shift changes rather than average activity. For every integration specify direction, frequency, fields, IDs, retry rules, error alerts, reconciliation, and operator ownership. An ERP stock transaction must only fire at the approved business point.

Compare bids across software, scanners, printers, labels, wireless coverage, integrations, master-data cleanup, training, support, and change rates. Use your site’s actual material shortages and logistics labor to estimate benefits. Require assumptions and a measurement method beside each promised benefit, rather than copying an industry ROI figure into your business case.

FAT and SAT: test the complete loop and its exceptions

At FAT, create one empty-bin request with test data and follow allocation, picking, transport, receipt, and card collection. Test repeated scans, disconnected devices, source stockout, wrong-item pick, rejected receipt, cancellation, retries, and held lots. Do not accept a screen-only demonstration. Verify the same request ID and reconciled quantities in all relevant systems. After retry, one request must still produce one intended pick. After cancellation, physical and book stock must have an explained state.

SAT uses the real rack, bins, tags, handhelds, wireless network, routes, and shift teams. Empty the first bin and measure whether the replacement reaches the correct location before the second one is depleted. Check scanning through oil, reflections, distance, and protective gloves; check route access and space. Record by part and shift, then adjust bin quantity or delivery frequency. The FAT pass cannot replace a physical-site acceptance test.

Two-Bin Kanban Line-Side Replenishment System: Implementation Guide for Manufacturers - figure 3

Before go-live, invalidate old cards, reconcile opening stock, check location masters, decide how open requests transfer, and define how to revert to controlled paper work. During the first shifts compare digital state with the physical containers on every route. Name who can approve a correction. Training must include where to place empties, when to stop an exception, and how to reject an incorrect delivery. Manufacturing, logistics, quality, and IT should accept the same test record.

Roll out from one line and learn the operating conditions

Choose a pilot line with repeatable parts but enough route, shift, and exception variation to teach you something. An oversimplified trial hides mixed-model and night-shift problems; plant-wide rollout on day one replicates bad cards and rack masters at scale. During the pilot declare whether the old paper method or new digital method is the transaction of record. Avoid two stock deductions caused by dual entry.

After the cycle is stable, define entry criteria for additional parts: bin quantity, number of cards, rack, source, lead time, quality class, and ERP/MES/WMS posting point. When removing a part, retire cards and dispose of remaining bins. A rack move updates the physical sign and printed cards as well as the device master. Assign approval and ownership for these changes as part of the system.

Toyota’s history of its production system explains how kanban strengthened supermarket-style production by moving what was needed, when needed, in the needed quantity. Digitizing a card alone does not create that pull behavior. Demand must arise from actual consumption; withdrawal and upstream replenishment must connect; and exceptions must not lead to unrecorded advance deliveries.

Frequently asked questions

Does every component need two-bin kanban replenishment?

No. It suits repeat consumption, standard containers, adequate space, and a repeatable source. Compare other methods for large, rare, custom, or tightly batch-controlled parts. Validate demand rate, lead time, rack space, and lot constraints using actual material.

Does an empty-bin scan reduce ERP inventory?

Not necessarily. The scan can create a replenishment request. Issue from the supermarket, receipt at the line, and actual consumption are separate events. Configure financial posting for the approved business point and avoid double deduction across ERP, MES, and WMS.

Should we use physical cards or barcodes?

Cards provide visible work instructions; barcodes make identity and time capture easier. When combined, define one authoritative trigger. Test lost cards, damaged print, and offline devices, with controlled paper evidence and later reconciliation.

How should bin quantity be set?

Check whether the remaining bin covers time-of-day demand through the replenishment lead time. Account for pack quantity, route frequency, quality hold, and variation. Adjust with real material at SAT. If the second bin cannot last, move the signal point or change bin count and route design.

What is the difference between FAT and SAT?

FAT verifies state transitions, integrations, retries, duplicate prevention, and stock reconciliation in a test setting. SAT verifies the same flow with real racks, containers, devices, wireless coverage, routes, shifts, and parts. Both must exercise exceptions.

Conclusion

Line-side two-bin replenishment turns actual consumption into an authorized withdrawal and gets the correct container to the correct point of use before production runs out. Select parts and bin sizes from measurements, choose one official card/scan trigger, and track allocation, handoff, receipt, and empty-bin return. Define ERP/MES/WMS stock boundaries and exception rules before testing. Prove one line’s cycle through FAT and SAT, tune bins and routes, and then scale.

If you are assessing whether two-bin supply fits your parts or how to separate the ERP/MES/WMS transactions, you can contact TOMAS TECH with your candidate items and current delivery route. We can start by mapping one complete cycle and its acceptance criteria.

Sources