NDD WindowWindow

Overview

A window defines a feasible interval on a dimension. Windows are most commonly used to represent time windows (e.g. a delivery must occur between 08:00 and 12:00) but can be applied to any dimension. Windows can be specified on location attributes, task attributes, and vehicle shift tasks.

Applicable models

• NDD ndd-cmibu6krtqja

Window

Schema definition

message Window {
  required float start = 1;
  required float end = 2;
  optional float slackCostCoef = 3 [default = 0];
  optional float tardyCostCoef = 4 [default = 0];
  optional float slackMax = 5 [default = -1];
  optional float tardyMax = 6 [default = -1];
}

Fields

start

The start of the feasible interval on the dimension. In the context of time, this is the earliest acceptable arrival time at a location or task.

end

The end of the feasible interval on the dimension. In the context of time, this is the latest acceptable arrival time.

slackCostCoef

The cost coefficient applied per unit of slack on this window. Slack represents the amount of dimensional quantity added to meet the window start (e.g. waiting time). A higher coefficient penalises earlyness more heavily.

tardyCostCoef

The cost coefficient applied per unit of tardiness on this window. Tardiness represents exceeding the window end. A higher coefficient penalises lateness more heavily.

slackMax

An override for the maximum slack permitted on this specific window. A value of -1 (the default) defers to the dimension-level slackMax setting.

tardyMax

An override for the maximum tardiness permitted on this specific window. A value of -1 (the default) defers to the dimension-level tardyMax setting.

Examples

A morning delivery window from 08:00 to 12:00 (in minutes from midnight) with no slack or tardy cost:

start: 480
end: 720
slackCostCoef: 0
tardyCostCoef: 0

A window with a soft end, permitting up to 30 minutes of lateness at an additional cost:

start: 480
end: 720
slackCostCoef: 0
tardyCostCoef: 0.5
tardyMax: 30

Notes

• Multiple windows may be specified on a single task attribute or location attribute. The solver will select the most cost-effective window to satisfy.
• When slackMax or tardyMax is set to -1, the dimension-level defaults are used. Set these to 0 to create a hard window that permits no earlyness or lateness respectively.