Heuristic methods for job shop scheduling: Active schedule generation algorithm, non-delay schedule generation algorithm and heuristic schedule generation algorithm

PT. Multi Citra Busana is one of the knitting material convection companies that has a good local scale marketing area where production scheduling is usually done based on the order of arrival of orders or first come first serve (FCFS). As a result, there are still some schedules that are not precise, causing a large make-span. In observations made on 5 products, the company requires a total production time (make-span) of 11 hours 54 minutes in completing production activities. In this research, scheduling evaluation is conducted to minimize make-span using the Active Schedule Generation algorithm, Non-delay Schedule Generation and Heuristic Schedule Generation algorithm The approach used is to use a quantitative-comparative approach, namely research that compiles the calculation process to make comparisons between the results of the actual process with these methods. Through the Active Schedule Generation algorithm method, the value of make-span is 6 hours 26 minutes. Through the Non-delay Schedule Generation algorithm method, the value of make-span is 7 hours 1 minute. The Heuristic Schedule Generation algorithm method produces the smallest make-span value with a value of 6 hours 1 minute. Based on the comparison of the make-span values of the three methods used, it was concluded that the Heuristic Schedule Generation algorithm produced the smallest make-span valued at 20184.27 seconds or 6 hours 3 minutes shorter than the method applied by the company. Therefore, the Heuristic Schedule Generation algorithm method was chosen as the best method and can be applied in the company to minimize the make-span.


Introduction
The effectiveness and efficiency of energy is important in the production mechanism [1]. Scheduling can be defined as an effort to organize activities or work with the aim of achieving efficient use of facilities, time, and costs [2,3]. In the production process at PT. Multi Citra Clothing only serves orders (make by order) and involves many machines in the process. In fulfilling the customer's order, there are still machines that are unemployed while working on another job. The idle time of the machine that could have been used to do the jobs simultaneously was in vain so that there was often a delay in the completion of the job (make-span) time or exceeded the IOP Publishing doi:10.1088/1757-899X/1034/1/012114 2 maximum limit [4]. Make-span, which is the time required to complete all jobs in the shop, which consists of setup time between jobs and processing time per job [5]. Delay in fulfilling demand like this results in higher production costs because it takes longer to produce the product, the production schedule that should be able to run according to the plan finally changes and the loss is that consumer confidence in the company will decrease because it cannot fulfil orders in accordance with the time that has been set. Therefore, it is necessary to schedule each machine to minimize the amount and time of the delay. The reason why these methods were chosen is because this algorithm produces a fairly good scheduling sequence and is close to the optimal solution [6,7].

Method
Before the data is processed, the data that has been obtained is tested for the adequacy of the data first. To calculate the data adequacy test, use the equation 1 [8].
The value of N 'is the amount of data that must be available and can be seen in the data sufficiency table and the condition used to determine the adequacy of the data is if N' <N, then the data obtained is sufficient. But if N '> N data is not enough it is lacking and needs to be added to the data [8]. The data adequacy test formula is presented in equation 2 [8].
Where: k = Confidence level (99% = 3.95% = 2) s = Degree of accuracy N = Number of observational data N '= Amount of theoretical data ̅ = observational data Meanwhile, if the data obtained is not sufficient, data retrieval will be carried out. Activities like this continue to be carried out until the data are sufficient and appropriate for further processing. After the data obtained is considered sufficient then a uniformity of data is tested to separate data that have different characteristics due to various influences and to ascertain whether the data collected is from the same system. The formulas used in testing data uniformity are following equation 3 to 5 [8].
Where: UCL = upper control limit LCL = lower control limit ̅ = average data value σ = standard deviation k = level of confidence The data that has been tested is then calculated standard time or standard time. That is the actual time used by the operator to produce one unit of product type data (equation 6 and 7) [8]. standard time can be obtained by applying the formula: After the data to be processed passes the adequacy and uniformity test of the data and the default time has been obtained, the next step that must be taken is to process the data using the methods used. In measuring the average cycle time, there are measurement steps that must be taken [6]. The stages of the research carried out in accordance with Figure 1.

Conclusions
Based on the analysis and discussion carried out before, the following conclusions can be drawn. Based on the comparison of the make-span values of the three methods used, it was concluded that the heuristic schedule generation algorithm method produced the smallest makespsan worth 20184.27 seconds or 6 hours 1 minute seconds shorter than the method applied by the company. Thus, the heuristic schedule generation algorithm method was chosen as the best method and can be applied in the company to minimize the make-span.