ApEc 4821 Eidman

Scheduling Operations
Scheduling is an allocation decision. Scheduling decisions allocate resources over relatively short time periods: a few months, weeks, or hours. It uses the resources made available by facilities decisions and aggregate planning. The schedule indicates: What is to be done; when; by whom; and with what equipment. Scheduling decisions are concerned with tradeoffs among conflicting goals: high efficiency low inventories good customer service Scheduling decisions involves three distinct objectives: cost staying within budget schedule completing jobs by their due date (or hour) performance the performance of the product or service being provided Scheduling in Line Processes When several different products are made on the same line, scheduling problems stem from "changeover" time. The economic lot size is based on a trade-off between changeover cost and inventory-carrying cost. Once lot sizes have been determined, runout time calculations are used to schedule production: ri = Ii/di where, ri is runout time for the ith product, Ii is the current inventory for the ith product, di is the demand per period for the ith product. In line processes, a general scheduling rule is to schedule the product with the smallest runout time first. Products are scheduled by sequentially applying this rule, with sequential updating of inventory information. Loading Loading is a type of scheduling that loads or packs work into available work time. In general, the goals are to achieve a level load of work in all departments and to meet production deadlines. Well look at two basic ideas that come from the concepts of loading and that are useful in all scheduling decisions. Forward loading begins with the current date and loads jobs forward in time, taking into account the rough sequencing requirements and average waiting times. 1

The goals of forward loading are to estimate completion dates and capacity requirements. Backward loading begins with the due date for each job and loads requirements backwards, using average processing times. The goal of backward loading is to identify capacity requirements for each work center for each period. Gantt charting Gantt charts are a commonly used tool for determining and/or representing the exact sequence of operations at different work centers as well as waiting times and project completion times. Gantt charts are drawn with: Time across the top. Either scarce resources or jobs are listed down the side. Sequences of activities for individual jobs are marked on timelines for each resource. Gantt charts come in two forms: 1) by job or activity; 2) by machine. Gantt charts can be used to evaluate both job and machine performance. Thus, alternative sequencing options can be evaluated and the best (depending on the weighting of objectives) can be chosen. Machine performance can be evaluated by make span and by machine or work center utilization. Job performance can be evaluated by the sum of delivery times for all jobs scheduled. Make span is the time required to complete all work on all jobs. Machine or work center utilization is expressed at the percentage of time the machine is busy. To calculate this we start with calculating idle time: idle time % = {(# machines)*(make span) - work time} / (# machines)*(make span) utilization % = 100 - idle % The number of machines is constant, so is the work time (or processing time), so the amount of idle time is dependent on the make span which is different under alternative schedules (i.e., Gantt charts). So to minimize idle time, minimize make span. Sum of delivery times. Since work time is fixed, delivery times are inversely related to waiting time. So to minimize delivery times, minimize waiting time. Alternative plans can be evaluated with alternative Gantt charts.

Theory of Constraints A bottleneck work center is a work center whose capacity is less than the demand placed on it and less than the capacities of all other resources. The TOC argues that the bottleneck is the critical constraint that should be scheduled to achieve maximum output. Many steps can be taken to increase the capacity of the bottleneck resource. Some are : reduction of setup time use the bottleneck resource more hours per day. Dont stop it for breaks, lunch, or maintenance that can be deferred. add machines or labor to the bottleneck work center divert work that doesnt need to go through the bottleneck center to another work center, prevent work from reaching the bottleneck that has poor quality and will be scrapped later. Dispatching Dispatching decisions focus on which job to start next in a particular work center. These are complex decisions, because decisions about priorities in one work center may affect waiting times and utilization levels elsewhere. Common dispatching decision rules: MINPRT (minimum processing time) The next job to be done is chosen on the basis of which has the shortest processing time. The objective is to get the jobs done quickly so downstream processes do not have to wait for work. This rule is also called SPT or SOT for Shortest Processing/Operation Time MINPRT with truncation Use MINPRT except when a job has waited a specified period, then do that job. This modification eliminates long waits for long jobs but some efficiency is lost. MINSOP (minimum slack time per operation) Choose the next job on the basis of which job has the least slack time. Slack time is the difference between the amount of time until the due date and the amount of processing time left. To standardize or normalize, the slack time is divided by the number of operations. Critical Ratio The job with the minimum amount of time left until its due date relative to the remaining processing time (i.e., its CR) is chosen as the next job. CR= remaining time until due date remaining processing time FCFS (first come, first served) Choose the next job on the basis of which arrived first. 3

This is based on fairness criterion. MINSD (minimum planned start date) A previously prepared, master schedule is checked to see which job is scheduled to start first. The job scheduled to start first is started first. This is also called ESD (earliest start date). MINDD (minimum due date) The job with the earliest due date is chosen as the next job. This is also called EDD (earliest due date). RANDOM (random selection) Under this rule, a random process is used to choose the next job. As the book says, this rule is not used in real life, it is used as a benchmark to check for the performance of other rules. A study at Hughes aircraft (LeGrande) indicated that MINPRT does best in efficiency and flow rate measures; MINSOP does best in meeting due dates; FCFS and MINSD did worse than RANDOM for most criteria.