TigerFlow: Service Based Flow Shop Simulator
The TigerFlow simulator is a service based platform for simulating flow-shops composed of machines and automated guided vehicle (AGVs). In a TigerFlow’ flow shop machines execute a collection of tasks in order to complete a job and AGVs move material from one machine to another for completing a task. In the current version of TigerFlow tasks are assigned to machines randomly. However, in the presence of machine failures, tasks are reallocated to other machines following a best effort approach.
The guide is structured as follows:
Configuring a flow shop
Scheduling a machine tasks
The TigerFlow platform is started by executing the following script in folder TIGERFLOW:
$ node bin/www.js
In TigerFlow entities communicate by exchanging messages over a network (cf. figure below).
When TigerFlow is initialized it starts the following services (cf. image below):
Network: responsible of broadcasting flow shop messages. In its current version, TigerFlow simulates a CAN bus network.
Cell: in charge of allocating tasks in machines.
Warehouse: responsible of dispatching AGVs to machines carrying material for task execution.
Configuring a flow shop
TigerFlow offers a Web UI for configuring a flow shop (localhost:3000). The interface offers buttons for adding n machines/AGVs to a flow shop (cf. figure below).
When a user adds an AGV to a flow shop,TigerFlow creates an AGV service instance and add it to the network.This can be seen in the Web UI in the log panel. In the same way, if a user adds a machine to a flow shop, TigerFlow creates a machine service instance with 10 randomly tasks (executed one at a time) and add it to the network.
The following images show the state of the flow shop after adding 2 machines and 1 AGVs to the flow shop.
Scheduling machine tasks
Machine tasks have each an ID, an estimated processing time, an estimated completion time and a due date. They can also be in one of three states at runtime: waiting, executing or executed state. You can follow the progress of a task in the Web UI.
When a new job comes, it can be submitted in the ‘add new job’ panel. If the checkbox is selected, the job will be a rush order. Otherwise the job will be a normal job with a very large due date and the due date setting will not work. Rush order means a job’s completion time should be earlier than due date or close to due date as much as possible. Obviously, normal jobs will be add to the end of task queue.
The following images show the state of the flow shop after submitting a rush order.
When a job’s due date is changed ,you can use ‘modify job’ panel to modify the job’s due date. If the job matches the delay condition that due date is one minute later than estimated completion time, the job will be rescheduled.
The following images show the state of the flow shop after modifying a job.
Overcoming machine failures
When one machine is out of action, all tasks in its task queue will be allocated to other machines. You can simulate machine failure with ‘close machine’ panel by closing a certain machine.