Sliding shoe sorters are among the most mechanically sophisticated and throughput-critical machines in a modern distribution center. A shoe sorter moves cartons, polybags, and even small irregular items across a slat conveyor, using a series of individually actuated shoes to divert products off at predetermined destinations. Commissioning a new system — or accepting a major retrofit — is not a single event; it is an evidence-gathering process. This checklist-based guide describes the phases of commissioning and acceptance, what to look for, what can look right but still be wrong, and how to separate minor tuning excursions from fundamental design or installation faults. The content is intended as independent industrial education material. Site-specific procedures, lockout requirements, OEM documentation, and competent engineering judgment always take priority over anything presented here.
Operating Context and System Architecture #
Before checking sensors or measuring throughput, the commissioning team needs a clear mental model of the sorter as a complete control loop. A sliding shoe sorter does not merely move product; it establishes a contract of timing. The induction system presents items to the sorter at a controlled gap. The sorter tracks each item and its associated destination. The shoes move laterally, riding on guide tracks, to push the item off at the appropriate chute or spur. After the divert, a re-shoe section repositions the shoes to a neutral state for the next induction pass.
Each of these zones has independent failure modes, but their interactions are what create the most confusing symptoms. A mis-tuned photo-eye in the induction area can cause a missed gap that, five meters later, produces a two-for-one issue at a high-speed divert. A worn shoe link under one particular carton profile can create an intermittent jam that recirculation masks for hours. The acceptance process must therefore be organized so that each subsystem is verified in isolation first, then tested as an integrated system.
Three architecture families influence every subsequent test:
- Single-level vs. multi-level sorters: multi-level systems have separate slat conveyors stacked vertically. Acceptance must verify that cross-level transfer conveyors, merges, and safety interlock logic work independently.
- Continuous vs. indexed operation: most large systems are continuous, with induction running at a fixed belt or slat speed. Indexed systems are less common but require different validation for dwell times and shoe positioning.
- Modular slat construction: some designs use plastic slats, others stainless steel, others a combination. The coefficient of friction between the slat surface and the product footprint directly affects divert behavior, especially for thin or highly polished cartons.
The commissioning team should also agree, in writing, on who owns each finding. A “finding” in this context is a documented observation describing a behavior that deviates from the stated functional requirement. Findings can range from “cosmetic” (light guard paint scratch) to “blocking” (cannot pass final acceptance). Without a triage structure, the acceptance site visit becomes a series of diffuse discussions about vendor responsibility, and the subsequent handover documentation tends to understate genuine operational risk.
Pre-Power Checks and Mechanical Baseline #
Mechanical verification must happen with the system locked out and with no product on the conveyor. The sequence below assumes local safety rules, OEM lockout procedures, and competent supervision are in place. Never attempt to inspect or adjust moving components while the sorter is, or could become, energized.
Start with visual and tactile inspection of the entire conveying surface. Shoes should be present in every slot. Check for cracked slat edges, raised lugs, and any plastic flashing that could catch a polybag seam. Lift a sample of shoes by hand; they should translate across the slat width with moderate friction, without binding or free-play that would produce a clacking sound during high-speed operation. Mark any shoe that requires more than gentle force to move.
Pay particular attention to the guide tracks, especially at the entry and exit of each divert. The tracks are the sorter’s skeleton; any misalignment of 1 to 2 millimeters at a joint will manifest as a vibration that accelerates wear and causes shoes to intermittently skip. Verify that track joints are flush, not only visually but with a straightedge laid across the joint. If the OEM documentation gives a maximum track step, record the largest deviation found, even if it is within tolerance, because that data will be useful at the first maintenance interval.
Measure the chain or slat tension and inspect the sprocket engagement. Sliding shoe sorters can use a single continuous chain, multiple chain loops, or a gear-driven slat system. In each case, the drive should engage with the slat structure without a visible “pitch” or rhythmic lag. Turn the conveyor over manually, or by using the specified jog feature, for at least one full loop and listen for repeating noises. A rhythmic clicking, especially if it increases or decreases in frequency with speed, suggests a damaged slat, a bent shoe link, or a track irregularity.
For the electrical portion of pre-power checks, verify that every limit switch, photo-eye, and proximity sensor is securely mounted and aimed. Record the alignment values for each photocell where the sensor provides a discrete output or an analog signal strength reading. Sensor alignment that is merely functional at low speed may degrade when vibration increases at rated speed. Confirm that cable chains are routed away from the shoes and slats, and that internal wiring has enough service loop to accommodate track motion without strain.
Controls and Sensor Verification #
With power safely applied and all guarding correctly in place, the next phase is to validate the control system’s logical response. This is the layer where communication faults, wrong addressing, and diagnostic gaps make themselves visible.
Each physical destination on the sorter should have a unique logical node. If the controls database refers to destinations by an alphanumeric string (for example, “P3-112”), the physical signage at the chute and the display on the human-machine interface (HMI) must match exactly. A discrepancy of one character can cause a sortation error that is very difficult to troubleshoot after acceptance, because the operator sees a product at the wrong location and the system’s trace report claims the intended destination was correct.
Verify the fieldbus intercommunication between the sorter controller, the warehouse control system (WCS), the warehouse execution system (WES), and the programmable logic controller (PLC). The typical failure mode is not a complete loss of communication but a repeated transient dropout. To detect this, use the diagnostics page of the PLC program to observe the communication status or health counter for at least 30 minutes of idle operation. Non-zero or incrementing error counters are not always problematic — many industrial protocols have a defined mechanism for retries — but they must be stable and within the negotiated tolerance. An error counter that increments every 10 minutes during idle operation will likely worsen under load due to EMI from motors and drives.
Test the emergency stop chain and all safety interlock gates, but only with designated safety personnel and in full compliance with site procedures. The goal is not to see whether the sorter stops; it is to see how it stops. A sorter that is structurally stopped but leaves the shoe logic in an inconsistent state — for example, shoes mid-divert for a carton that is 20 centimeters past a chute — is more dangerous after a restart than faulting entirely. Record whether the recovery sequence requires manual reset, and whether the WCS transaction for in-flight items is cleared or flagged.
Divert Logic and Destination Accuracy #
Destination accuracy is the primary key performance indicator for a shoe sorter. The most important point to understand about acceptance testing is that a test carton with a common box size is not a valid substitute for the actual product mix. If operations expect to handle a high percentage of low-friction polybags and thin flat-packs, then the acceptance test set must heavily weight those profiles.
Create a test matrix that maps product anatomy to expected behavior. Independent variables include length (along direction of travel), width, height, weight, and the friction coefficient between the carton and the slat, as well as the friction between the carton and the extruded aluminum or plastic divert rail at the chute. A practical test matrix for a standard installation might look like the following
- Small rigid carton (200–300 mm long) with a flat, high-friction surface
- Long thin polybag (400–600 mm along travel) with flexible contents
- Wide but flat corrugated box (width greater than shoe travel length)
- Heavy, minimally flexed carton (higher mass per square centimeter)
- Mixed product with minimal gap between items
For each test batch, document the induction speed, the product’s entry position on the slat, the intended destination, the actual destination, and the pitch angle at the divert moment. A carton that arrives at the chute nose skewed by more than 5 degrees is a symptom of a timing or track issue, not a cosmetic problem, because skew increases the probability of a jam at the chute diverter and creates an unpredictable footprint in the recirculation lane.
A key acceptance criterion is the “no misroute” window. In a correctly functioning shoe sorter, a carton that begins its divert motion too late will continue past the chute and recirculate. Observing recirculation is not necessarily an error. However, a carton that begins its divert motion too early may end up in a chute physically adjacent to the intended one. That is a misroute, and it implies a timing error logic failure. During acceptance, count every misroute, not just the ones that resulted in a visible wrong item. The WCS exceptions log should be cross-referenced with visual observation to verify that recirculated items are correctly logged as such.
Also check the release logic at the divert plane. Shoes should move smoothly from the neutral position onto the guide track, with no oscillating or hunting before they contact the product. A shoe that vibrates as it changes track is subject to premature wear and can impart a lateral jolt to a lightweight product.
Recirculation and Throughput Stability #
Recirculation is the sorter’s way of saying “I did not handle this item at the first opportunity.” A well-designed system uses recirculation as a controlled buffer for merge coordination, missed inductions, and out-of-sequence items. But recirculation also creates the risk of self-inflicted failure: a loose or mis-labelled carton in the recirculation lane can disrupt later induction, and a damaged shoe can gradually damage every product it contacts over several laps.
During the throughput stability test, run the sorter at rated speed, but use a combination of induction profiles: 80% direct induction, 20% recirculated product. Track how many items re-enter the sorter on each lap. If a single item recirculates more than twice, investigate whether the divert decision is being repeated without resolution. This can occur when the WCS loses track of an item’s true physical location and reassigns it to the same destination over and over. The sorter chute may already be full, or the divert might be mechanically failing, but the control system “thinks” the item is still in queue. This is one of the most difficult field issues to diagnose because the visible symptom (repeated recirculation) is the same as for a mechanical jam, while the actual cause is a plan/track inconsistency.
Throughput stability is not measured as a single average. Measure throughput in 5-minute windows, and record the coefficient of variation across those windows. A system that averages 6,000 items per hour but varies between 2,800 and 8,400 in successive 5-minute windows is unstable, regardless of the theoretical capacity. The variance will create downstream starvation or overflow at the shipping lanes and are likely to cause operational complaints within the first weeks of production.
Note the relationship between gap and speed. Many shoe sorter installations exhibit a characteristic “speed cliff,” a point at which a small increase in line speed causes a disproportionately large increase in miscounts or diverts failed. This cliff is typically determined by the sensor reaction time in the induction, the sensor-to-actuator latency, and the shoe’s mechanical travel time. Learn where this cliff is and document it as an operating boundary, not just a failure point. If the cliff appears at a speed lower than specified, investigate the induction sensor response time and the controller’s scan cycle rather than immediately blaming mechanically slow shoes.
Diagnostic Table: Common Symptoms and Evidence #
| Symptom | Likely Contributing System | Diagnostic Evidence to Collect | Interpretation Note |
|---|---|---|---|
| Intermittent misroute at one chute | Divert track alignment, shoe actuation, or WCS mapping | Pitch angle of product at chute entry; shoe position sensor state; time stamp of WCS decision | If the error only appears at one chute, focus on mechanical height and timing, not global timing. |
| Same item recirculates repeatedly | WCS tracking, destination status, or physical sensor flag | Full tracing record for that item; control and physical reads of the chute occupancy indicator | A “ghost” occupied chute is often the cause; verify the chute is physically empty. |
| Continuous misfeed at induction | Induction sensor alignment or merge logic | Sensor signal strengths; footage of items entering at non-standard gaps | Mis-timed merge logic can look like a sensor issue; capture gap spacing from high-speed video. |
| High shoe wear in a single longitudinal zone | Track joint step or slat warpage | Straightedge reading at every joint across the zone; thermal reading of adjacent track | Wear that is localized to 300 mm of the track is a mechanical defect, not a lubricant anomaly. |
| Motor torque spikes without apparent jam | Slat pitch, sprocket indexing, or a bent shoe link | Drive current data; rotational speed oscillation; audibly locate the tick | Do not assume a logic issue. Cycle the sorter without product and watch for the spike. |
| Occasional product split out of cell | Pitch angle too large or product friction too low | Footage of the product leaving the slat; recording of shoe contact point | This is often a controlled tuning issue, but repeated splits at one destination indicate a downstream chute conflict. |
Interpretation Errors and Decision Boundaries #
Commissioning engineers and site teams often interpret the same observation differently. The most common interpretation error is attributing a control logic failure to a mechanical fault, or the reverse. A shoe that rides up over a track joint destructively will sound like a mechanical problem, but if it only occurs when the sorter is carrying a heavy carton, the trigger may be a slightly oversize track pocket that allows slat expansion under load. The shoe is failing mechanically, but the root cause is a design tolerance error, not a component quality error.
Another interpretation error is the “destination accuracy tail” — measuring only the overall accuracy percentage without segmenting by product type. A system that achieves 99.95% overall accuracy can easily have a 100% error rate on the hardest 2% of the product mix. For an e-commerce operation selling small, flexible items, a sorter that handles flat-rigid cartons perfectly but fails on polybags will generate a disproportionate load on the manual recirculation station. Always compare accuracy segmented by class of product or by source induction lane.
Decision boundaries define whether a discovered issue blocks acceptance or can be logged as a deferred maintenance item. A blocked decision boundary should include at least one of the following criteria:
- Safety-related: any interlock, guard, or emergency stop that does not behave consistently according to the site safety procedures.
- Misroute or item damage rate that exceeds the value specified in the approval criteria, when measured over the full acceptance run, not a single batch.
- A throughput shortfall that is not attributable entirely to site-provided induction starvation. This means the sorter is given sufficient input volume and still cannot meet the target.
- Any failure that causes the sorter to fault more than a specified number of times in a defined window. A single fault during the speed test should be noted but not necessarily block acceptance, unless it is a repeated fault with no verified root cause.
- The inability of the control system to recover from a fault without operator intervention at a frequency too high for normal operation.
Conversely, items that should not block acceptance are: minor cosmetic paint damage, non-repeatable single shoe binding that disappears after lubrication, or a chute light alignment that requires re-adjustment after first contact with product. These should be documented as punch-list items with an agreed completion date.
The team must also avoid the common bias of “last test bias”: if the final 30 minutes of a three-hour test go smoothly, the entire three hours are remembered as smooth. Instead, keep a running sheet with a one-line comment every 10 minutes. The comment sheet is what establishes whether the system’s performance was steady or front-loaded.
Documentation, Sign-Off and Transition to Operations #
The sign-off document is more than a spreadsheet of pass/fail ticks. It should contain the threshold criteria before the test, field observations during the test, defining evidence for each failure. It should also include the exact test configuration that was used. That means the software version of the PLC, WCS, and any vision system; the hardware revision of major components such as the drive motor and shoe guide; and, importantly, the set of parameters controlling shoe timing, divert delay, and induction gap.
For a robust handover, provide a parameter change log. If the vendor or site team adjusts the divert delay during the acceptance window, record the original value, the new value, and the reason. This log is invaluable six months later when a maintenance technician changes a shoe beam and asks, “Why is this parameter set to 250 ms?” Without the log, the technician will either reset it incorrectly, reverting to a worse state, or will not touch it, preserving a setting that was only correct for one product profile.
During the first 72 hours of production — sometimes called “ramp-up” — the commissioning team should not disappear immediately. Run a structured shadowing period in which the maintenance engineer, the controls engineer, and the WCS administrator observe the sorter under real order profiles. The acceptance test always uses a finite set of product samples; production reveals anomalies that never appeared in testing. Any observed deviation during ramp-up should be compared with the baseline acceptance thresholds. A subtle recirculation pattern that appears only at 3 a.m., when the induction merge logic is starved and then suddenly flooded, is typical.
Finally, confirm that spare parts are physically available, not just ordered, for the components most likely to wear: shoes, shoe pins, track joints, and drive chain. Spare part inventory is sometimes handled by the procurement department and not the engineering team, but a failure to have specific spare components in stock during the first month of operation is one of the most common root causes of an otherwise avoidable catastrophic downtime.
Key Takeaways #
- Commissioning is an evidence-based process: every acceptance decision should be traceable to a timed observation, a sensor reading, or a mechanical measurement, never to an informal impression.
- Mechanical baseline is the foundation: track alignment, shoe friction, and slat continuity must be established before the control system can be verified, or electrical tuning will mask physical defects.
- Accuracy acceptance must be split by product profile and induction lane; a single aggregate percentage value hides the failure modes that operational teams care about most.
- Recirculation repeats are a diagnostic treasure and a source of risk: trace a repeated recirculation to the WCS decision and the physical chute status before finalizing any acceptance decision.
- Throughput stability is measured in windows, not as an average; the variance across five-minute intervals is a stronger indicator of production readiness than the theoretical maximum rate.
- Document the parameter settings, the test configuration, and the reason for every change made during the acceptance window; this preserves knowledge long after the commissioning team leaves.
- Establish clear decision boundaries before testing begins so that a single noteworthy observation does not derail (or unfairly validate) the system in a way that site management could later contest.
- Safety, lockout, and OEM documentation always take precedence: the acceptance checklist is a tool, not a substitute for competent engineering judgment and site-specific procedural compliance.
