"Look-ahead” Sequencing Rule ( Part I )

OVERVIEW

Sequencing rules are one of the areas where performance improvement is possible in terms of production output, resource usage. Better the usage and utilization of resources is, better would be the end result w. r. t. standard manufacturing norms. This is not only creates a better work flow but also reduces the global operating costs.

Sequencing rule is a rule that prioritizes the jobs waiting for processing on a particular machine. The prioritization scheme normally takes into account the attributes of the job, machine and current time as well.

It is important to note the different between databases oriented applications and real intelligent applications. In general, database oriented applications always talks about the past – what had happened, whereas intelligent applications talks about the future – what will happen. This result is derived from the past and current state of the environment.

Primarily, this is nothing but generating a solution by “Looking – Ahead”. It would be interesting here to apply the good old concept of ‘Look ahead’ can also be applied to sequencing rule to generate new sequencing rule(s), with predefined objective function(s).

This is essentially a global sequencing, which considers the current resource and looks ahead to successor operations and respective resources to find out an efficient assignment for current resource.

In another knowledge paper we talked about standard sequencing rules, which are applied in general. Here we will try to illustrate sequencing rule, with look-ahead approach to derive heuristic sequencing rule. It would be appropriate to mention here that such heuristic sequencing rules can give better performance result in specific scenarios or environments.


(I) PROBLEM DEFINITION

The good old concept of ‘Look ahead’ can also be applied to sequencing rule to generate new sequencing rule(s), with predefined objective function. This is essentially a global sequencing, which considers the current resource and looks ahead to successor operations and respective resources to find out an efficient assignment for current resource. To illustrate the problem better, let us try to create a sample problem definition:


EXPRESSION OF PROBLEM:

1] Operations J1/10,….,J4/10 are to be scheduled on Wc1. Find out a sequence which shall satisfy a defined objective function, by looking ahead.

2] Next operations of all current operations [J1/10,….,J4/10 ] are operations [ J1/20,…, J4/20 ].

3] Above illustration shows feasible tuple / resources where next operation’s can be carried out. It provides respective processing times and queue lengths on each feasible resource. It is assumed that unit of queue length and processing time is same.

OBJECTIVE:

Find out operation sequence to be processed on current resource in order to achieve desired result. For example, the objective is to move the operation as fast as possible. In other words, try to minimize the WIP, in order to generate better throughput.

INFORMATION AVAILABLE [CALCULATED]:

I] List of the operations to be scheduled on current resource. [ curr_op_list ]

II] List of all immediate successor operations list of curr_op_list. [succ_op_list ]

III] List of all feasible tuples for succ_op_list. [ f_tuple]

IV] Merged calendar for all f_tuple. [cal_tuple]

V] Waiting queue at f-tuple [units same as processing time unit for operations] [ f_t_queue]

VI] Earliest start date for succ_op_list on each f_tuple. [ esd_succ_op ]

VII] Minimum esd_succ_op for succ_op_list. [one value per operation in curr_op_list] min_esd_op

VIII] Precedence constraint among succ_op_list pree_succ_op