Project report subway - Arena (simulation)
- 1. SIMULATION FOR UC MAIN CAMPUS PROJECT REPORT POORVI DESHPANDE M12388313 DEC 3, 2017
- 2. ACKNOWLEDGEMENT I would like to thank Professor David Kelton for guidance provided during the course of project. I would like to express my gratitude to employees of Subway, UC Main Street, UC, who provided valuable information and granted me the permission to collect data.
- 3. CONTENTS 1. Introduction 1.1 Objective ………………………………….…………………4 1.2 Current Process.…………………………….………………..4 1.3 Present Model ….……………………………………………5 2. Model Assumptions …………………………………………………...6 3. Data Collection ………………………………………………………..6 2.1 Inputs for Model …………………………………………....6 2.2 Data Distribution …………………………………………...6 4. Base Model ……………………………………………………………10 5. Model Results …………………………………………………………12 6. Alternate Model ……………………………………………………….14 7. Results ………………………………………………………………...15 8. Conclusion …………………………………………………………….17
- 4. 1. INTRODUCTION Subway is an American fast food restaurant which has approximately 45,000 stores located in more than 100 countries. It is one of the fastest growing franchises in the world. This project aims to simulate one of its outlet at the UC Main Campus which is frequented by many students of University of Cincinnati day in and day out. The software used for simulation is Arena. Arena facilitates us to learn important statistics of a model like total time average, waiting time of customers in the system, utilization of various resources etc. Subway opens at 9 AM and closes at 8 PM. It typically witnesses rush around lunch hours 12 PM to 3 PM. According to the discussion with a subway employee, I gathered some information about the number of people usually in a queue during rush hours. During the rush hours, the wait time could be anywhere between 5 to 10 minutes with approximately 10 people in the queue. This could lead to customers leaving the store deciding against buying a subway sandwich and probably head in search for another luncheon. This results in loss of customers for the store. 1.1 OBJECTIVE To simulate current system at the subway outlet at UC main campus and uncover ways to effectively improve the efficiency of resources and reduce the waiting time of customers for the whole service i.e. from walk in to exit. 1.2 CURRENT PROCESS As customers enter the restaurant, they either go straight to the service counter or wait in the queue area for their turn. At the counter they select the type of bread following which they choose whether they want meat in their sandwich or not. The customer also choose if he/she wishes to toast their bread. Next, they can choose condiments for their sandwich and then finally proceed to pay the bill. The drinks section is separate and therefore does not get counted in the services at the counter. In any food chain, it is it is quite expected that there would be a queue formed during the peak hours or when many customers arrive at the same time. The queue length is often one of the determining factors to choose restaurants and joints. Therefore, our goal is to come up with a model that reduces the wait time for the customer.
- 5. 1.3 PRESENT MODEL From the point customer enters the subway joint, he goes through many stages before he has his order in his hands. These processes are : • Bread Selection • Meat Selection • Toasting the bread • Adding vegetables , sauces and condiments • Adding drink to the order • Bill Payment Subway outlet at UC is particularly busy during the lunch hours as well as when the classes break. They open up another service counter during this time (12:00 pm to 3:00 pm) to mitigate the long queue that is expected to form in these hours. The current staff schedule is as follows: Table 1.1 Scheduling of staff
- 6. 2. MODEL ASSUMPTIONS The following assumptions were made to model the subway restaurant: 4. The demand remains same on all days of the week. 5. The schedule of the staff does not change throughout the week. 6. Every staff is equally skilled in all processes involved in making a sub. 7. The drinks section is not included in the service counter process time. 3. DATA COLLECTION 3.1 Inputs for the model The time intervals of the following processes were noted to learn the estimate of the distribution of data. • Inter-arrival time of customers • Bread Selection Time • Toasting Time • Process time for choosing vegetables/condiments • Billing time 3.2 Data Distribution 1. Inter-arrival Time of Customers: The frequency of customers is less during the initial hours. Slowly on the rise, it reaches maximum at the peak lunch hours. A rough schedule of customer arrival is as follows: Figure 3.1 Schedule of customer arrival per hour 0 20 40 60 80 100 120 140 # CUSTOMERS
- 7. 2. Bread Selection Time: Figure 3.2 Distribution of time for bread selection 3. Toasting time : It was found that the toasting time was a uniform distribution between 30 sec and 40 sec depending on the meat and length of bread.
- 8. 4. Process time for choosing vegetables/condiments: The distribution best fit for vegetable selection was found to be 36.5 + 24* BETA (1.18, 0.856) Figure 3.3 Distribution of time for vegetable selection
- 9. 5. Billing Time: Figure 3.4 Distribution of time for billing
- 10. 4. BASE MODEL Following is the model that is currently being implemented in subway restaurant. Figure 4.1 Arena model for subway simulation There are essentially 2 service counters, one of which opens up only during the rush hours i.e 12 PM till 3 PM. Main counter is open for all working hours. The various types of modules used in the model are: • The create module for the arrival of customers. • The decide module to decide which queue should a customer take when both queues are open. The logic behind the choice is that the customer would always go for the queue which has shorter queue length. Another decide module designates a 50% by chance split for toasting the bread assuming that 50 % people go for a toasted bread.
- 11. • The process modules that operate different functions in the making of the sandwich. Below are the snapshots of the model: Resources: Figure 5.1 Resources for subway simulation Queues: Figure 5.2 Queue for subway simulation Entities: Figure 5.3 Entities for subway simulation
- 12. 5. MODEL RESULTS The model was run for 50 replications, the results of which are as follows: Table 5.1 Output performance matrix of Subway simulation 1. Total number of customers completely served and exited from the system is 640. Figure 5.1 Number of customer served in Subway simulation
- 13. 2. Time of customers in queues : Figure 5.2 Queue times Subway simulation 3. Scheduled Utilization of resources: Figure 5.3 Scheduled Utilization of resources in Subway simulation
- 14. 6. ALTERNATE MODEL An improved model could be anything that reduces the wait time of the customers by either increasing the resources or modifying their order. One way to achieve that would be to increase the staff during the rush hours which might improve upon the waiting time in the queues. However, it will also mean the cost increment for the outlet as the new staff would have to be paid accordingly for the given time interval. Thus, I have proposed a shift in the staff in this new model. It essentially removes the second billing counter and employs the billing staff at the second vegetable selection counter as it is the most exhaustive process of all. The two billing counters are combined into one. This could be done as I noticed that the billing counter witnesses no queues in the base model. Figure 6.1 Alternate model for Subway simulation We now have our staff distributed as : Table 6.1 Schedule of staff in improved model of Subway simulation
- 15. The total number of staff remains the same. The improvement due to the proposed model is that the waiting per customer on an average is reduced by 15.68%. 7. RESULTS The results from the base and the best model was compared. Table 7.1 Output Performance matrix for improved model of Subway simulation 1. Number of customers completely served is 645. Figure 7.1 Number of customers served in improved Subway simulation
- 16. 2. Queues: Figure 7.2 Queue times in improved Subway simulation 3. Scheduled Utilization of resources: Figure 7.3 Scheduled Utilization in improved Subway simulation
- 17. 4. Comparison from Process Analyzer Figure 7.4 Chart in process Analyzer for best case 8. CONCLUSION After simulating different scenarios and analyzing them in the process analyzer, we can conclude that adding another resource at the vegetable and sauce counter can increase the outlet efficiency leading to serve more number of customers and reducing their total time in the system by 6 seconds. We can also improve the system performance by cross skilling the employees and utilizing the toast time for the bread selection of the next customer.