DWSRP-TW-SC Datasets

Published: 26 May 2026| Version 1 | DOI: 10.17632/dbsf73xpg5.1
Contributors:
Onur Demiray, Eda Yucel, Doruk Tolga

Description

This dataset provides benchmark instances for the Dynamic Workforce Scheduling and Routing Problem with Time Windows and Skill Constraints (DWSRP-TW-SC). Each instance represents a workforce routing scenario in which a heterogeneous, multi-skilled team serves geographically distributed service tasks that arrive dynamically throughout a 540-minute (9-hour) working day. Tasks are characterized by an arrival time, earliest and latest start times, processing time, priority weight, location, and required skills. Teams are characterized by their skill capabilities. Tasks whose skill requirements cannot be covered by a single team require synchronized service by multiple teams. Instance parameters: Each super-instance is denoted I(N, K, δ), where N is the total number of tasks, K is the number of teams, and δ ∈ {0.2, 0.4, 0.6, 0.8} is the target degree of dynamism. The combinations (N, K) ∈ {(30, 2), (40, 3), (50, 4), (60, 5), (75, 5)} yield 20 super-instances per set. Task locations are drawn uniformly within a 25 km × 25 km region; rectilinear distances and a team travel speed of 30 km/h are assumed. Processing times follow Uniform(5, 25), priority weights Uniform(1, 5), and the skill set has |Q| = 5. Skill requirements and team capabilities are generated by a Bernoulli process with threshold 0.5, with every task requiring at least one skill. Folder structure: TightInstances — Tight time windows: lᵢ = min{aᵢ + Uniform(10, 50), τᵐᵃˣ}. Synchronization is implicit and handled at the optimization level. TightInstances_D — Duplicated variants of the tight instances. Each synchronization-requiring task is split into the minimum number of single-skill-coverable sub-tasks sharing the same arrival time, time window, processing time, weight, and location, and must be executed simultaneously. LooseInstances — Identical to TightInstances except lᵢ = τᵐᵃˣ = 540 for all tasks (relaxed time windows). LooseInstances_D — Duplicated variants of the loose instances. Each synchronization-requiring task is split into the minimum number of single-skill-coverable sub-tasks sharing the same arrival time, time window, processing time, weight, and location, and must be executed simultaneously. TightInstances_A_(ω, υ) — Anticipated-task variants of the tight instances, with (ω, υ) ∈ {(40, 35), (50, 20), (60, 15), (70, 10), (80, 5)}. LooseInstances_A_(ω, υ) — Anticipated-task variants of the loose instances, with (ω, υ) ∈ {(40, 35), (50, 20), (60, 15), (70, 10), (80, 5)}. Small_Medium_DWSRP-TW-SC_Instances — Small to medium size instances with all tasks known at t = 0, used to benchmark the ALNS against exact and constructive baselines.

Files

Categories

Vehicle Routing Problem, Dynamic Routing, Workforce Planning, Dynamic Optimization

Licence