---
publication_id: "OT-2026-000000037"
title: "Dynamic Slotting with Bounded Reslotting Rate and Affinity Measures"
published_at: "2026-09-15T19:01:39.237881Z"
canonical_url: "https://opentechnical.org/p/OT-2026-000000037"
categories:
  - "warehousing-storage-systems/warehouse-control/slotting"
keywords:
  - "warehouse slotting"
  - "inventory optimization"
  - "co-pick frequency"
  - "reslotting rate"
  - "affinity grouping"
  - "picking efficiency"
  - "storage management"
  - "seasonal inventory"
  - "cost-benefit analysis"
  - "warehouse control system"
---

# Dynamic Slotting with Bounded Reslotting Rate and Affinity Measures

## Abstract

This disclosure describes a system for optimizing inventory slotting within a storage facility by considering product affinity and co-pick frequency while adhering to a bounded reslotting rate. The arrangement determines optimal storage locations for items to minimize travel distance and pick time. It quantifies item affinity based on historical pick data, assigns a cost-benefit score to potential item movements, and prioritizes moves that offer the greatest operational improvement without exceeding a configurable limit on physical inventory relocations per period. The system also integrates mechanisms for managing seasonal inventory and new product introductions to maintain efficient storage utilization.

## 1. Technical Field

[0001] The present disclosure relates to the field of material handling and logistics, specifically to the management and optimization of inventory placement within storage facilities such as warehouses or distribution centers. It addresses methods for assigning storage locations to stock keeping units (SKUs) to enhance operational efficiency.

## 2. Technical Context

[0002] Traditional inventory slotting aims to place items in locations that minimize travel distance for picking operations or maximize storage density. However, frequently, an analytically derived optimal slotting plan can necessitate a large volume of inventory moves, potentially relocating a significant portion of the stored items. Implementing such a plan can be prohibitively disruptive and costly due to the labor, equipment, and downtime required for physical moves, leading to situations where optimal plans are generated but never fully executed. This results in suboptimal facility layouts persisting, undermining the potential gains of advanced slotting algorithms. There is a need for a slotting methodology that optimizes item placement while explicitly bounding the physical reslotting effort within practical operational limits.

## 3. System Overview

[0003] The disclosed system provides a framework for dynamic inventory slotting that optimizes item placement based on affinity measures, primarily co-pick frequency, while explicitly controlling the rate of physical inventory moves. The system processes historical pick data and current inventory information to calculate item affinities and identify potential improvements in pick path efficiency. It then evaluates the operational benefit of relocating items against the cost of performing such moves. A core component is a reslotting manager (18) that imposes a configurable limit on the number or volume of inventory moves permitted within a specified period, ensuring that slotting adjustments remain operationally feasible. The output is an actionable plan of proposed moves and new slot assignments that progressively steers the facility towards an optimized state without incurring excessive disruption. FIG. 1 illustrates the functional modules of the system.

## 4. Components

[0004] The system comprises several interacting components:

- **Data Ingestion Module (10)**: This module collects and pre-processes raw data from various sources. Inputs include historical pick transaction records, current inventory levels, item dimensions (length, width, height), weight, cube, and velocity data (e.g., picks per day). It also ingests facility layout data, such as slot coordinates, dimensions, capacity, and pick path definitions.
- **Affinity Calculation Engine (12)**: This engine analyzes historical pick transactions to determine the co-pick frequency between pairs of SKUs. For each pick list or order, it identifies which items are frequently picked together. An affinity score is computed for each SKU pair, representing the likelihood or frequency of their co-occurrence in pick tasks. This score can be a simple count, a normalized frequency, or a more complex metric considering order size or item contribution to travel time reduction.
- **Slotting Engine (14)**: This module identifies potential new slot assignments for SKUs. It considers factors such as item velocity, cube, weight, storage requirements (e.g., temperature control, hazardous materials), and the calculated affinity scores. Its objective function aims to minimize total travel distance, pick time, or maximize throughput, subject to physical slot constraints (e.g., size, capacity) and operational rules (e.g., ergonomic placement). It generates a set of candidate moves for evaluation.
- **Cost-Benefit Modeler (16)**: This component quantifies the operational benefit of a proposed move against its associated cost. The benefit is typically a reduction in predicted pick time or travel distance, often expressed in a monetary equivalent per period. The cost of a move includes labor time for relocation, equipment usage, potential disruption to ongoing operations, and administrative overhead. The model assigns a net score to each candidate move (Benefit - Cost).
- **Reslotting Manager (18)**: This critical component implements the bound on physical moves. It receives a ranked list of candidate moves from the Slotting Engine (14) and the Cost-Benefit Modeler (16). Based on a configurable maximum reslotting rate (e.g., number of SKUs moved per week, total cube moved per month), it selects the highest-scoring moves that collectively do not exceed this limit. This ensures that the generated slotting plan is operationally executable.
- **Output Interface (20)**: This module generates the final slotting plan, including specific instructions for inventory moves, new slot assignments, and any required label changes. The output can be integrated with a warehouse management system (WMS) or provided as a standalone task list for execution by warehouse personnel.

## 5. Operation

[0006] The system operates in a cyclical manner, typically on a predefined schedule (e.g., daily, weekly, monthly).

[0007] **Start-up Sequence:** Upon initial activation or after a major configuration change, the Data Ingestion Module (10) performs a full load of historical pick data, current inventory, and facility layout information. The Affinity Calculation Engine (12) then processes this comprehensive dataset to establish baseline affinity scores for all active SKU pairs. The Slotting Engine (14) performs an initial optimization run, generating a broad set of potential slot reassignments based on these affinities and other item characteristics. This initial set of moves is then passed to the Cost-Benefit Modeler (16).

[0008] **Steady Running:** During normal operation, the system follows a continuous optimization loop:

- **Data Refresh**: The Data Ingestion Module (10) periodically updates its dataset with new pick transactions, inventory changes, and any facility layout modifications. This ensures the system operates with current information.
- **Affinity Re-calculation**: The Affinity Calculation Engine (12) incrementally updates or periodically re-calculates affinity scores based on the most recent pick data. This allows affinities to evolve with changing demand patterns.
- **Candidate Move Generation**: The Slotting Engine (14) identifies potential improvements by evaluating current item placements against updated affinities, velocities, and available slots. It proposes candidate moves for individual SKUs or groups of highly affiliated SKUs. For example, if items A and B are frequently co-picked, and they are currently in disparate locations, the Slotting Engine (14) might propose moving B closer to A (or vice-versa, or both to a new joint location).
- **Cost-Benefit Analysis**: The Cost-Benefit Modeler (16) evaluates each candidate move. The benefit is calculated as the projected reduction in pick path distance or time over a defined future period (e.g., 30, 60, or 90 days), multiplied by the item's anticipated pick frequency. The cost is a predetermined value per unit moved, adjusted for factors like distance of move, item weight, or special handling requirements.
- **Move Prioritization and Filtering**: The Reslotting Manager (18) receives the ranked list of candidate moves, sorted by their net cost-benefit score (benefit minus cost). It then applies the predefined bound on physical moves. For example, if the bound is 50 moves per week, it selects the top 50 moves from the ranked list. If a move of SKU A to slot X creates a conflict for SKU B already in slot X, the Reslotting Manager (18) resolves this by either rejecting the lower-priority move or finding an alternative slot for SKU B if its move is also high priority and fits within the bound.
- **Plan Generation**: The selected moves are compiled into a slotting plan by the Output Interface (20), which specifies the item, its current location, and its new target location. This plan is then released for execution. The cycle repeats, with subsequent cycles building upon the executed moves, continuously refining the slotting arrangement.

[0010] **Fault or Interruption Handling:** In the event of a system interruption, the last successfully generated slotting plan remains active. Upon restoration, the system can either resume from its last completed step or perform a full re-initialization, depending on the nature of the interruption and the age of the last data refresh. Data integrity checks are performed during the Data Ingestion (10) phase to ensure that processing proceeds with valid information. If a proposed move cannot be executed physically (e.g., slot already occupied, item not found), this is reported back to the system, and the Slotting Engine (14) will consider this slot unavailable for the current item in subsequent cycles until the discrepancy is resolved.

## 6. Example Embodiments

[0011] **Embodiment 1: Dynamic Slotting for High-Velocity Items with Co-Pick Affinity** In this embodiment, the system prioritizes high-velocity SKUs that are frequently picked together. The Affinity Calculation Engine (12) places a higher weighting on co-picks within the same customer order for items with an average daily pick count exceeding a threshold of 10. The Slotting Engine (14) then attempts to co-locate these high-affinity, high-velocity pairs or groups within a defined "fast-pick" zone, such as depicted by item (30) in FIG. 2, which is typically nearest to dispatch. The Cost-Benefit Modeler (16) calculates the benefit of such moves as a direct reduction in operator travel time, assuming a fixed time value per unit of distance. The Reslotting Manager (18) is configured with a bound of 100 item moves per day, focusing on items that offer at least a 15% reduction in predicted pick path distance if moved.

[0012] **Embodiment 2: Seasonal Item Management with Pre-emptive Slotting** This embodiment focuses on managing seasonal inventory fluctuations. The Data Ingestion Module (10) receives a forecast of seasonal item demand peaks and troughs, often several weeks in advance. For seasonal items, the Affinity Calculation Engine (12) may use historical pick data from previous seasonal cycles rather than just recent data to predict co-pick patterns. The Slotting Engine (14) identifies optimal locations for these seasonal items (e.g., moving them from bulk storage (34) to medium pick slots (32) prior to their peak season or vice-versa post-peak). The Cost-Benefit Modeler (16) evaluates these pre-emptive moves by factoring in the expected increased pick volume during the peak and the potential for reduced put-away time post-peak. The Reslotting Manager (18) allows for a higher move bound (e.g., 200 moves per week) during specific seasonal changeover periods, enabling a more aggressive re-arrangement of inventory to prepare for or recover from a seasonal demand surge.

[0013] **Embodiment 3: New Product Introduction with Buffer Slotting** This embodiment addresses the integration of new SKUs into the facility. When a new product is introduced, it initially lacks pick history, so affinity data is unavailable. The Data Ingestion Module (10) flags these items as "new." The Slotting Engine (14) initially places new SKUs in designated buffer or "quarantine" slots, often similar to medium pick slots (32), strategically chosen for their accessibility but not necessarily optimized for affinity. As pick data accumulates for these new items, the Affinity Calculation Engine (12) begins to compute affinities. After a predetermined period (e.g., 4 weeks) or a minimum number of picks (e.g., 50 picks), the Slotting Engine (14) then proposes moves for these new items to more optimized locations based on their newly established velocities and affinities. The Cost-Benefit Modeler (16) prioritizes these moves to integrate the new items into efficient picking paths. The Reslotting Manager (18) ensures that these integration moves for new products are balanced with ongoing optimization moves for existing inventory within the overall move bound.

## 7. Alternative Configurations

[0014] The affinity measure can be configured to include other metrics beyond simple co-pick frequency. For instance, the system could incorporate a weighting based on the cubic volume of items picked together, prioritizing groups of large items for co-location to reduce travel for forklift operations. Another alternative is to use a monetary value of co-picked items, emphasizing higher-value groupings. The bound on reslotting rate can be dynamic, varying based on current warehouse workload or available labor resources, rather than a fixed number. For example, if the WMS reports low picking activity, the Reslotting Manager (18) might temporarily increase the allowed moves. Furthermore, the cost component of a move in the Cost-Benefit Modeler (16) can be differentiated based on the type of equipment required (e.g., manual, forklift, automated guided vehicle) or the specific zone of the warehouse.

## 8. Parameter Variations

| Parameter | Range | Effect of Moving Towards Lower End | Effect of Moving Towards Higher End |
| --- | --- | --- | --- |
| **Co-pick analysis window** | 7 to 180 days | Reacts faster to recent demand shifts, more volatile affinities. | Smoother affinities, less sensitive to short-term fluctuations. |
| **Reslotting move bound** | 10 to 500 moves/period | Minimizes disruption, slower convergence to optimal slotting. | Accelerates optimization, higher short-term operational cost. |
| **Move cost per item** | 0.5 to 5.0 monetary units | Prioritizes more moves, greater emphasis on potential benefits. | Restricts moves to only high-benefit candidates, less frequent changes. |
| **Benefit projection period** | 14 to 90 days | Rewards short-term gains, suitable for volatile demand. | Considers longer-term efficiency, less reactive to immediate trends. |
| **Affinity weighting factor** | 0.1 to 1.0 (relative to velocity) | Slotting dominated by item velocity, less by co-pick patterns. | Slotting strongly influenced by co-pick relationships, potentially grouping slower items. |
| **Seasonal item buffer time** | 7 to 60 days before peak | Less time for pre-emptive moves, potential for rush operations. | Ample time for smooth transitions, higher holding costs in buffer. |

## 9. Additional Implementations

[0016] The system can be extended to integrate with other warehouse automation technologies. For example, if an automated storage and retrieval system (AS/RS) is present, the Slotting Engine (14) can incorporate AS/RS specific constraints like bin dimensions and retrieval times, and the Reslotting Manager (18) can leverage the AS/RS's automated movement capabilities, potentially allowing for a significantly higher move bound due to reduced labor costs. Another implementation could involve real-time feedback from pickers via handheld devices; if a proposed move causes unforeseen operational issues (e.g., access difficulty), this feedback could be captured by the Data Ingestion Module (10) and used by the Cost-Benefit Modeler (16) to adjust future move costs or by the Slotting Engine (14) to refine slot suitability. Furthermore, the system could incorporate a multi-echelon optimization, considering not only slotting within a single facility but also the flow of inventory between different facilities in a network, using affinity data to optimize cross-docking or merge-in-transit operations.

## 10. Figures

![FIG. 1](figures/fig-001.svg) A block diagram illustrating the main functional modules of the slotting optimization system.

![FIG. 2](figures/fig-002.svg) An elevation view of a storage bay with distinct slot types for different inventory characteristics.