Simulation-Based Sequencing

Simulation-based sequencing provides an attractive alternative to algorithmic sequencing by providing a simple, yet very flexible method for constructing a schedule. In general, a simulation-based sequencer can also produce any schedule that is produced by an algorithmic sequencer. However, a simulation-based sequencer can also consider many schedules that cannot be considered by an algorithmic sequencer. Hence, the simulation-based sequencing approach, sometimes referred to as a Resource Based Method or Parallel Loading, can create better schedules than an algorithmic sequencer, but this would depend very much on what you consider to be a 'better' schedule.

In contrast to the algorithmic sequencer, the simulation-based sequencer selects and loads an operation at a time. By loading individual operations rather than entire orders, the simulation-based sequencer has finer control over the way the operations are loaded onto the planning board. This operation-at-a-time loading is the key to the added flexibility in schedule generation using the simulation-based sequencer.

A second fundamental difference between the algorithmic and simulation-based sequencer is that the simulation-based sequencer constructs the schedule in a single time pass by moving forward from one event time to the next. The simulation-based sequencer begins at the current time and loads all operations that can start now. Note that these operations don't come from a single order as in the case of the order-at-a-time sequencer, but can be taken from the entire set of orders. Once all operations that can start at this event time have been loaded, the simulation-based sequencer advances time to the next event e.g., the first completion time for any operation on the planning board. In this case, a busy resource has just changed to idle; hence, the simulation-based sequencer attempts to load additional operations at this new event time. The simulation-based sequencer continues in this fashion, advancing time forward to the next event time and loading additional operations as resources become idle until all operations have been loaded.

The following set of figures illustrates steps in the simulation-based construction process. In each of the figures, the vertical line denotes the current value of simulated time. The first step in the simulation-based sequencer is to load all operations that can begin at the current time.

In this example:

Step 1: Load Operation A-10 and B-10
Step 1: Operation 10 for Order A (A-10) can be loaded on Resource 1 and Operation 10 for Order B (B-10) can be loaded on Resource 2.

Step 2: Load Operation C-10
Step 2: The simulated time is advanced to the ending time of Operation A-10, and Operation C-10 is then loaded on Resource 1.

Step 3: Load Operation A-20
Step 3: Simulated time is advanced to the end of Operation B-10, and Operation A-20 is loaded on Resource 2.

Step 4: Load Operation B-20
Step 4: Simulated time is advanced to the end of Operation C-10, and Operation B-20 is loaded on Resource 1.

Step 5: Load Operation C-20
Step 5: Simulated time is advanced to the end of Operation B-20. Since no further operations remain to be loaded on Resource 1, simulated time is advanced to the end of Operation A-20 where the final operation, C-20, is loaded on Resource 2.

In contrast to the algorithmic sequencer, the simulation-based sequencer only moves time forward. It only advances time forward once all operations that are to be loaded at that time have been loaded. Hence, the simulation-based sequencer temporarily stops time and examines the entire order set, advancing to the next event time once all operations have been loaded. The algorithmic sequencer, on the other hand, is constantly moving back and forth in time as it loads all operations for each order. It starts at the current time and goes forward in time loading all operations for this first order. It then goes back again to the current time and again moves forward in time loading all operations for the second order. It continues in this fashion, moving forward and backward in time, until all orders are fully loaded on the planning board. Hence, the algorithmic sequencer fixes an order and works across time, advancing to the next order once all operations have been loaded.

In the Preactor 400 APS simulation-based sequencer, we control the schedule using dispatching rules that select the next operation to load.

Typical rules select the next operation

• with the shortest setup time
• with the shortest processing time
• with the earliest due date
• with the lowest value of critical ratio
• in a preferred sequence
• etc...

Although the loading concepts employed by the Simulation Based Sequencer provide great flexibility in schedule generation, the effective use of this flexibility is influenced by the dispatching rules chosen to select the next operation for the resource from the queue of work waiting to be processed. Preactor 400 APS employs both standard and customized dispatching rules and this is discussed further in the section dedicated to Preactor 400 APS.

There are disadvantages in using a simulation-based sequencer too. An example would be use a dispatching rule which minimizes lateness, such as 'critical ratio', where selection of operations would be based on the remaining operation time compared to the delivery time. In this situation, if capacity is a problem, a simulation-based sequencer would tend to minimize total lateness of all the orders but you would have a higher number of late orders (i.e. a large number of slightly late orders). An algorithmic sequencer using the due date priority option would produce fewer late orders (i.e. a small number of very late orders). In this situation the user could focus attention more easily on those orders that are predicted to be late.

An algorithmic sequencer can also control some process parameters more effectively than a simulation based sequencer. For example in many food and process industries it is often a requirement to control the amount of time that work waits between process steps (food can spoil). Because an algorithmic sequencer loads a complete order at a time it can easily check if the intervals between the operations are correct, and adjust them if required.

The simulation-based sequencer has less control because when it loads the first operation it does not know when the next operation will be loaded. Later, when the next operation is loaded, it may not be possible to move the first operation because other work has been loaded in between.

In general, a simulation based sequencer using First In First Out (FIFO) as the dispatching rule, will generate a schedule with higher resource utilization and a shorter overall schedule span than an algorithmic sequencer, but at the expense of higher work in progress.