What Is Dispatching?

Dispatching is the real-time decision process that determines which job, task, or unit of work to release to a machine or resource at any given moment in a production or service system. It occupies the lowest and most time-sensitive layer of operations management, sitting beneath longer-horizon planning and scheduling activities. While a production schedule defines the sequence and timing of work over hours or days, dispatching executes those decisions moment by moment on the shop floor, responding to the actual state of machines, queues, and incoming jobs.

The field draws on operations research, control theory, and computer science. Its foundational concern is priority assignment: given a queue of jobs waiting for a resource, which should go first? The answer depends on objectives such as minimizing average cycle time, meeting due dates, balancing machine utilization, or reducing work-in-process inventory. Each objective leads to a different dispatching strategy, and no single rule dominates across all shop configurations.

Dispatching Rules

A dispatching rule is a local decision policy that assigns a priority score to each waiting job so the resource can select the highest-priority candidate. Classic rules include First-In First-Out (FIFO), Shortest Processing Time (SPT), and Earliest Due Date (EDD). SPT minimizes average flow time across all jobs but can starve long-duration tasks indefinitely. EDD minimizes the maximum tardiness but does not account for processing times. More sophisticated composite rules combine multiple signals, such as the Critical Ratio rule, which divides remaining slack time by remaining processing time to produce a single priority number. A survey of AI-enabled dynamic manufacturing scheduling published in ACM Computing Surveys documents the evolution from fixed heuristic rules toward machine-learning policies that learn dispatching behavior from simulation data.

Dynamic and Real-Time Dispatching

Static dispatching rules assume a stable environment, but real production systems experience machine breakdowns, rush orders, quality holds, and variable processing times. Dynamic dispatching adjusts priorities in response to changing shop conditions, using discrete-event simulation or state estimates to re-evaluate the queue at each decision point. Semiconductor wafer fabrication facilities, which can involve hundreds of operation steps and thousands of lots in progress simultaneously, require particularly sophisticated dynamic dispatching. Operations management research at Intel documented an integrated architecture in which near real-time scheduling and dispatching operate together, with the scheduling layer providing targets and the dispatching layer executing them under live shop conditions.

Simulation-based dispatching selects among a candidate set of rules by running short-horizon look-ahead simulations at each dispatch event and choosing the rule that produces the best projected outcome. This approach adds computational overhead but substantially outperforms fixed rules in environments with high variability.

Hierarchical Integration with Planning and Scheduling

Dispatching does not operate independently. In the standard three-level hierarchy of production control, planning sets aggregate capacity and material targets over weeks or months, scheduling converts those targets into specific job sequences over a shift or day, and dispatching releases individual jobs to resources in real time. Each level uses progressively more current information about the shop state. The boundaries between scheduling and dispatching can blur in highly automated systems where a scheduling algorithm refreshes its output every few minutes, effectively collapsing the two layers. A unified view of production system control in the Springer handbook on semiconductor manufacturing scheduling provides a formal treatment of how these levels relate.

Applications

Dispatching has applications in a range of fields, including:

• Semiconductor wafer fabrication for lot sequencing across hundreds of processing steps
• Materials handling and warehouse order fulfillment for pick-and-pack sequencing
• Hospital operating room scheduling for surgical case assignment
• Air traffic control for runway and gate sequencing
• Cloud computing for task assignment across distributed compute nodes