WH-AS4 - Order-Picking Improvement

Aims

Develop and evaluate picking strategies for the SPR warehouse. Determine what picking changes can achieve in the current warehouse and how picking should operate with your group’s proposed storage, layout, and material-handling decisions.

Learning objectives

After completing this assignment, you will be able to:

select picking decisions suited to the observed workload;
develop a picking process that fits the warehouse’s storage, layout, and equipment;
compare strategies using quantitative models and feasibility checks; and
justify a recommendation using performance, resource requirements, and limitations.

Study case

Reuse your cleaned data, activity profiles, location coordinates, and previous results for the SPR warehouse, and extend your analysis to the picking process.

Tasks

Before comparing strategies, make the current warehouse picking process explicit. Aim to cover the information pertinent to the picking decisions you will change, so your baseline scenario is clear, traceable, and reproducible. For example, you may need to:

define the product population, warehouse area, and demand period, and justify that scope;
describe the baseline process, from order release and work assignment to completed orders; and
identify the company facts, assumptions, and missing inputs used in your model.

The following two tasks are independent, but you may reuse the same model and code for both.

1. Improve picking in the current warehouse

Change picking only. Keep the baseline storage assignments, physical layout, and material-handling equipment fixed.

Choose at least one picking strategy different from the historical (or simulated) baseline.
Implement a routing heuristic or optimization algorithm that generates feasible routes for the orders or batches in your study. Specify the routing rules, any other changed picking decisions, resource requirements, and assumptions.
Quantitatively compare the proposed strategy with baseline picking and explain the result.

Task 1 requires route generation for the selected orders or batches. An average-distance formula alone does not meet this requirement.

2. Adapt picking to the retained designs

Account for your earlier design decisions. Consider each storage, layout, and equipment combination your group still retains for the final recommendation.

List the retained combinations and the constraints they impose on picking.
Define at least one compatible picking strategy for each combination.
Quantitatively evaluate each combination with its picking strategy and compare the results with the baseline.

For example, a new layout may need different routes or zone handoffs, while different equipment may change load capacity or access. One strategy can serve several combinations if you check its compatibility and performance in each. Discarded alternatives are outside this task.

Task 2 allows either route generation or a travel model. You may implement a routing heuristic or optimization algorithm, or use analytical formulas or approximations to estimate travel for the proposed picking strategy. For simplicity, generating individual routes is optional in this task. In either case, define the picking rules and justify how the model represents the retained design and workload. Check the model’s aisle geometry, depot position, storage pattern, and demand assumptions against your proposal. Use your case data to estimate its inputs, explain any approximation or adaptation, and reflect on the limitations.

Optional travel models

The following travel models are optional examples:

Hall (1993): distance approximations for traversal (S-shape), midpoint, and largest-gap routing in a single-block warehouse, assuming independently and uniformly distributed pick locations.
Le-Duc and de Koster (2005): average travel-distance approximations for return routing in a two-block warehouse with a central cross-aisle and class-based storage. Pick locations are uniform within each class zone; the model assumes picking by order and excludes batching effects.
Dijkstra and Roodbergen (2017): exact expected route-length formulas for return, S-shape, midpoint, and largest-gap routing for a specified storage assignment in a single-block warehouse. The formulas assume independent item demand and sufficient carrying capacity; exactness concerns expected distance under the chosen routing policy, rather than an optimal route.

Examples of picking decisions

Table 24.1 gives examples of decisions you can change. Select a routing method for Task 1 and any other picking decisions relevant to your case. Explain why the selected methods and their assumptions fit the warehouse.

Table 24.1: Examples of optional picking decisions for the current warehouse or retained design proposals.

Decision	Examples
Routing	S-shape (traversal), return, largest-gap, or combined routes, where the aisle structure permits them.
Batching	Pick orders separately or combine them into batches based on shared locations or priorities, within carrying capacity.
Staffing and work assignment	Change picker numbers, assign and sequence orders or batches, or balance workload across existing zones.
Zone picking	Pass orders sequentially between zones or pick their parts in parallel and consolidate them.
Sorting	Sort items by customer order during picking (sort-while-pick) or after picking (pick-and-sort).

Evaluation requirements

Apply the following requirements to both tasks:

Model: use the route generation required in Task 1 and a suitable route-generation or travel model in Task 2. Calculation, optimization, and simulation may support the evaluation. State the model’s inputs, relationships, assumptions, and constraints. For optimization, define the decision variables and objective, and claim optimality only when supported.
Fair comparison: use the same demand population, period, units, evaluation assumptions, and external conditions. Identify the design and resource settings that differ. For random models, use matched seeds, justify the replications, and report variability. Retain failed runs and unfinished orders, and state the denominators used. Distinguish measured results from estimates and use a consistent evaluation method within each comparison.
Performance: show the baseline, picking-only proposal, and every retained combination with its picking strategy. Define the measures and units, such as travel distance, labor hours, completion time, throughput, or service. Report resource changes alongside performance changes. A travel-distance model measures travel only. To estimate total labor or completion time, account for relevant travel speeds, picking and handling times, and waiting or coordination effects.
Feasibility: check load capacity, access, labor and equipment availability, replenishment, handoffs, consolidation, and order completion where relevant. State unverified conditions as limitations.

Distinguish the picking-only effect from the combined effect of earlier design changes and picking. Where feasible, compare the baseline or another common picking strategy with the tailored strategy within a retained design. If that comparison is infeasible, explain why and report the result as a combined design effect.

Submission requirements

Management report

Include the analysis in your group’s shared management report. Aim to cover:

The baseline process, scope, assumptions, and strategies for both tasks.
A scenario comparison table showing design choices, picking rules, resources, performance, and change from the baseline, supported by feasibility and sensitivity results.
A recommendation for the current warehouse and retained proposals, explaining benefits, resource trade-offs, limitations, and implementation conditions.
Assumptions or parameters (sensitivity analysis) that can influence the recommendation. For example, assess how a change in picker numbers affects completion time, travel distance, and labor requirements.

Choose figures that support the decision and follow best practices for captions and visualizations. Keep full calculations in the code package.

Supporting code

Extend the existing documented entry point to read the original files in data/raw/ and generate the results.
Save scenario definitions, inputs, assumptions, settings, seeds where applicable, and the tables, KPIs, and figures used in the report.
When using a solver, retain its status and any available bound or gap.

Use any suitable tools or programming language and follow the existing final project file requirements. The report and code remain one submission package.

Assessment criteria

Table 24.2 shows the rubric used to assess the report and supporting code. Each criterion is worth 2 points, split equally between the two tasks. Task 1 and Task 2 each contribute up to 5 points, for a total of 10 points. The Task 2 score covers all retained design combinations together.

1 point per task: complete, correct, and decision-relevant evidence.
0.5 points per task: partially convincing evidence with a material omission, weakness, or minor error.
0 points per task: missing evidence or evidence that is materially incorrect, infeasible, or unsupported.

Shared baseline definitions, assumptions, code, and sensitivity results may support both tasks without being repeated. One sensitivity analysis is required for AS4 overall; use its results where relevant to the recommendations.

Table 24.2: Evidence for full credit in WH-AS4, with 1 point per task for each of five criteria.

Criterion (2 points each)	Task 1: current warehouse (1 point)	Task 2: retained designs (1 point)
Process definition	Baseline and proposed rules, roles, scope, and assumptions are clear and internally consistent.	Picking rules, roles, scope, and assumptions are clear and internally consistent for each retained combination.
Design logic and integration	Routing and other picking choices suit the workload and fixed warehouse, with justified methods and assumptions.	Picking choices suit the workload and retained storage, layout, and equipment combinations, with justified methods and assumptions.
Quantitative evaluation	Implemented feasible routes support correct, fair comparisons using appropriate measures of the picking-only effect.	Suitable routing or travel models support correct, fair comparisons for each combination, with picking and combined design effects correctly attributed.
Operational feasibility	Capacity, access, labor, equipment, replenishment, and order completion are supported for the current warehouse; limitations are explicit.	Capacity, access, resources, replenishment interfaces, handoffs, and order completion are supported for each retained design; limitations are explicit.
Decision justification	The recommendation explains picking-only benefits, resource trade-offs, limitations, and implementation conditions, using sensitivity results where relevant.	The recommendation explains combined design benefits, resource trade-offs, limitations, and implementation conditions, using sensitivity results where relevant.