By : Group 3 - Class C Akmal Fatah Fainusa 11/320145/TK/38979 Brithy Michelle Toar 11/312996/TK/37750 Ogissa Piertina Susilo 09/285235/TK/35694 Yunitha Ratnadilla 11/313560/TK/37951
DEPARTMENT OF MECHANICAL AND INDUSTRIAL ENGINEERING FACULTY OF ENGINEERING GADJAH MADA UNIVERSITY 2014
ii
LIST OF CONTENTS
Chapter I : Introduction ..........................................................................................................1 1.1 Background of Study ...........................................................................................1 1.2 Aims and Objectives ............................................................................................1 1.3 Assumptions and Contraints ................................................................................2 Chapter II : Institution Profile and Methods of Observation ................................................3 2.1 Brief Description of Institution ............................................................................3 2.2 Flow of System ....................................................................................................3 2.3 Activity Cycle Diagram .......................................................................................4 2.4 Servers, Resources, and Capacity ........................................................................5 2.5 Date and Time of Observation .............................................................................5 Chapter III : Simulation Design .............................................................................................6 3.1 Data Collection ....................................................................................................6 3.2 Goodness of Fit Test and Selecting Distribution .................................................6 3.2.1 Inter-arrival Time of Cars ...........................................................................6 3.2.2 Processing Time in Entry Ticket Counter ...................................................7 3.2.3 Processing Time in Exit Ticket Counter .....................................................9 3.2.4 Duration of Cars Being Parked ...................................................................10 3.3 Model and Output ................................................................................................12 Chapter IV : Results and Discussion ....................................................................................13 4.1 Real System Simulation .......................................................................................13 4.2 Verification of System .........................................................................................14 4.2.1 Visual Inspection ........................................................................................14 4.2.2 Flow Diagram .............................................................................................15 4.2.3 Animation of Simulation.............................................................................15 4.2.4 Mistake and Error Check ............................................................................15
ii
4.3 Validation of System............................................................................................15 4.3.1 Utility of Staff in Entry Ticket Counter ......................................................16 4.3.1.1 Normality Test ................................................................................16 4.3.1.2 Statistic Test ....................................................................................17 4.3.2 Average Waiting Time in Entry Ticket Counter .........................................18 4.3.2.1 Normality Test ................................................................................18 4.3.2.2 Statistic Test ....................................................................................20 4.4 Simulation Results ...............................................................................................21 4.5 Experiments with Model ......................................................................................22 4.5.1 Alternative 1: Vending Ticket Machine .....................................................22 4.5.2 Alternative 2: Additional Entry Ticket Counter .........................................24 4.6 Analysis and Discussion ......................................................................................27 Chapter V : Conclusion ..........................................................................................................28 References ..............................................................................................................................29 Appendices .............................................................................................................................30 1
CHAPTER I INTRODUCTION
1.1 Background of Study Generally, every system is expected to work in its most optimum productivity, so that it could generate more benefits for the firm or institution. Several tryouts should be carried out to the system with the purpose of finding out which alternative gives the highest contribution to the optimization of the system. Trials conducted to the system can be categorized into two types, which are experiments applied to the actual system and with a model of the real system. While the first category could be risky, impractical, and costly, building a mathematical model of an actual system turns out to be the most effective and efficient way of experimentation. Mathematical models can perform a varied range of systems. In this study, the system to be reproduced is the parking area system of a central hospital in Yogyakarta, known as RSUP Dr. Sardjito. With various health facilities located in the area, it is assumed that thousands of people come to this hospital every day for different needs. Such number of people will make use of the parking area, but considering the limited capacity of the parking lot, it is inevitable that some problems regarding the parking area system would arise. This study is conducted to build a model of the parking area system of RSUP Dr. Sardjito, and hopefully could give beneficial proposals of optimization for the problems regarding its parking area system.
1.2 Aims and Objectives In order to decide on the aims to be achieved, the problems occurring in the real system should be taken into consideration. Among all others, the main problem arising in the parking area system of RSUP Dr. Sardjito is about the waiting time (queuing time) of the vehicles (especially cars in this study) in the entry ticket counter. While the process time in the entry ticket counter varies from 2 seconds up to 25 seconds, the line can only provide for at most 10 cars. Furthermore, it is highly unlikely to expand the capacity of either the waiting line or the entry ticket counter since no more space is available in the area. Therefore, this study concerns in building a simulation model of the parking area of RSUP Dr. Sardjito which has the objective of minimizing the waiting 2
time of cars in the entry ticket counter. And based on the simulation, it is expected that some alternatives of optimization can be established to overcome the problem.
1.3 Assumptions and Constraints In building the simulation model, there are some assumptions and constraints set, which are: a. The system related to health facilities is excluded from the model. b. The observed entities are only the four-wheeled vehicles (cars in general). c. The existence of parking areas located outside RSUP Dr. Sardjito is neglected. d. The model in this study is limited to simulate the parking area system of RSUP Dr. Sardjito during visiting hours, in order to reduce the probability of visitors or patients staying overnight. e. Data collection for building the model is gathered only during visiting hours, which are categorized into morning visiting hours (10.00-12.00 WIB) and evening visiting hours (17.00-18.30 WIB).
3
CHAPTER II INSTITUTION PROFILE AND METHODS OF OBSERVATION
2.1 Brief Description of Institution RSUP Dr. Sardjito is the central hospital located in Kesehatan Street No. 1 Sekip, District of Sleman, Province of Daerah Istimewa Yogyakarta, and sited between the Faculty of Medicine and the Faculty of Engineering, Universitas Gadjah Mada. Built on 1982, this hospital has been cooperating with Faculty of Medicine in conducting the subjects for students learning to be doctors, specialists, and nurses. RSUP Dr. Sardjito owns various health facilities, ranging from cardiology, orthopedic, neural operations, medical rehabilitation, and many others. This hospital has a rather large parking area for four-wheeled vehicles which can provide up to 352 vehicles. The parking building is located in the west side of the hospital, comprised of 6 floors of parking area with different capacity for each floor.
2.2 Flow of System The flow of the parking area system of RSUP Dr. Sardjito is explained as follows: a. The system will be started when the car of patients or visitors enters the system through the gate and queues in the waiting line to get entry ticket. b. In the entry ticket counter, a staff will input data of the car, including the cars license plate and arrival time, and give a parking ticket. c. After getting the ticket, visitors or patients are allowed to enter the system to find any empty space and park their car. d. Afterwards, visitors or patients can go directly to any health facilities for their respective needs. The duration of visitors or patients stay inside the health facilities will determine the time of cars being parked. e. Once the visitors or patients are done, they will go back to the parking building to get into their car and move towards the exit ticket counter. f. In the exit ticket counter, visitors or patients will give the ticket back to the staff and pay the fare. g. The system will end once the car exits from the ticket counter. 4
2.3 Activity Cycle Diagram In the typical activity cycle diagram (ACD) based approach, a simulation model is viewed as a collection of interacting entities (Page, 1994). Entities are either idle - in notational or real queues - or active - engaged with other entities in time consuming activities. In the simulation model of parking area system of RSUP Dr. Sardjito, there are four different entities involved in the system, which are the cars, the staff on entry ticket counter, the staff on exit ticket counter, and the parking lot. List of entities with its assigned activities as well as the flow diagram of the system is described in Table 2.1 and Figure 2.1 as follows. Table 2.1 List of Entities and Activities Entity Activity Symbol Cars (temporary) Arriving at the entry ticket counter ARRIVAL Being processed by the staff on entry ticket counter IN Being parked in parking area PARK Being processed by the staff on exit ticket counter OUT Staff on entry ticket counter (permanent) Processing cars which are entering hospital area IN Staff on exit ticket counter (permanent) Processing cars which are leaving hospital area OUT Parking lot (permanent) Being used for parking cars PARK
Figure 2.1 Activity Cycle Diagram of Parking Area System of RSUP Dr. Sardjito
5
2.4 Servers, Resources, and Capacity Based on the observation of the real system, the numbers of servers and resources with its allocated capacity are listed below: a. Queue line for entering the system (capacity : 10 cars) b. Entry ticket counter (capacity : 1 car and resource : 1 staff) c. Parking lot (capacity : 352 cars) d. Queue line for leaving the system (capacity : 15 cars) e. Exit ticket counter (capacity : 1 car and resource : 1 staff)
2.5 Date and Time of Observation Data for building the model were collected on Tuesday, April 29 th 2014, and limited to the visiting hours of the hospital. Data were gathered from both the morning (09.30-11.00 WIB) and the evening (16.00-18.30 WIB) visiting hours in order to compare which data set could be more fitting for model input.
6
CHAPTER III SIMULATION DESIGN
3.1 Data Collection The data needed in building the model of parking area system of RSUP Dr. Sardjito are: a. Inter-arrival time of cars b. Processing time in the entry ticket counter c. Processing time in the exit ticket counter d. Duration of cars being parked
3.2 Goodness of Fit Test and Selecting Distribution 3.2.1 Inter-arrival Time of Cars From the observation, the number of cars entering the system during evening visiting hours was only 144 cars, while in the morning visiting hours it reached up to 194 cars. Considering that the latter total amount of cars could represent the system better, the set of data used in later tests are taken from the morning shift. Raw data of inter-arrival time during morning visiting hours is shown in Appendix 1. Before using the data for selecting distribution, it is preferred to check the uniformity and adequacy of the data set first. Test on the data uniformity reduces the total data from 194 to 192 cars, and test on the data adequacy gives the number of 2995 data needed in order to define more accurate distribution. Since it is not possible to gather so much data, the original data set will still be applied in selecting distribution. As shown in Figure 3.1, test using StatFit results in no theoretical distribution exactly fits the original data set. From this outcome, empirical distribution is applied for input in building the model. The graph of the empirical distribution is shown in Figure 3.2.
7
Figure 3.1 Auto-Fit Distribution Test of Inter-arrival Time Data
Class LB UB Freq. % 1 0 13 115 60% 2 14 27 14 7% 3 28 41 9 5% 4 42 55 15 8% 5 56 69 15 8% 6 70 83 9 5% 7 84 97 6 3% 8 98 111 2 1% 9 112 125 7 4% TOTAL 192 100% Figure 3.2 Empirical Distribution of Inter-arrival Time Data 3.2.2 Processing Time in Entry Ticket Counter The data set used in this section are picked out from the observation during morning visiting hours. Raw data of processing time in the entry ticket counter can be seen in Appendix 2. The test on data uniformity using control chart decreases the number of data to 187, while the test on data adequacy ensues in the total of 285 data needed in order to define more accurate distribution. Since it is not possible to gather more data, the original data set will still be applied in selecting distribution. 0 20 40 60 80 100 120 140 1 2 3 4 5 6 7 8 9 Inter-arrival Time 8
As shown in Figure 3.3, test using StatFit results in four theoretical distributions have passed the acceptance criteria. But before using any of four accepted distributions, it is preferred to check the outcomes of the goodness of fit tests first.
Figure 3.3 Auto-Fit Distribution Test of Processing Time in Entry Ticket Counter Data As can be seen in Figure 3.4, all four theoretical distributions which pass the auto-fit test have at least one reject result from the three type of goodness of fit test. Based on this result, it is preferred to use empirical distribution for input in building the model. The graph of the empirical distribution is shown in Figure 3.5.
Figure 3.4 Goodness of Fit Test for Processing Time in Entry Ticket Counter Data 9
Figure 3.4 Goodness of Fit Test for Processing Time in Entry Ticket Counter Data (cont.)
Class LB UB Freq. % 1 2 3 17 9% 2 4 5 41 22% 3 6 7 42 22% 4 8 9 39 21% 5 10 11 24 13% 6 12 13 17 9% 7 14 15 5 3% 8 16 17 2 1% TOTAL 187 100% Figure 3.5 Empirical Distribution of Processing Time in Entry Ticket Counter Data 3.2.3 Processing Time in Exit Ticket Counter The data set used in this section are picked out from the observation during evening visiting hours, since more data were obtained during this period. Raw data of processing time in the exit ticket counter can be seen in Appendix 3. The test on data uniformity using control chart decreases the number of data from 259 to 250, while the test on data adequacy ensues in the total of 288 data needed in order to define more accurate distribution. Since the difference is not really wide, the original data set will still be applied in selecting distribution. 0 10 20 30 40 50 1 2 3 4 5 6 7 8 Processing Time in Entry Ticket Counter 10
As shown in Figure 3.6, test using StatFit results in no theoretical distribution exactly fits the original data set. From this outcome, applying empirical distribution for input in building the model is the last step. The calculation and graph of the empirical distribution is shown in Figure 3.7.
Figure 3.6 Auto-Fit Distribution Test of Processing Time in Exit Ticket Counter Data
Class LB UB Freq. % 1 3 4 40 16% 2 5 6 56 22% 3 7 8 65 26% 4 9 10 40 16% 5 11 12 24 10% 6 13 14 13 5% 7 15 16 7 3% 8 17 18 5 2% TOTAL 250 100% Figure 3.7 Empirical Distribution of Processing Time in Exit Ticket Counter Data
3.2.4 Duration of Cars Being Parked The data set used in this section are picked out from the observation during morning visiting hours with the sample size of 50. Raw data of duration of cars being parked can be seen in Appendix 4. 0 10 20 30 40 50 60 70 1 2 3 4 5 6 7 8 Processing Time in Exit Ticket Counter 11
The test on data uniformity using control chart indicates no data is being outlier, thus all data can be used for later steps. The test on data adequacy ensues in the total of 185 data needed in order to define more accurate distribution. Since it is not possible to gather more data, the original data set will still be applied in selecting distribution. As shown in Figure 3.8, test using StatFit results in nine theoretical distributions have passed the acceptance criteria. But before using any of them, it is preferred to check the outcomes of the goodness of fit tests first.
Figure 3.8 Auto-Fit Distribution Test of Duration of Cars Being Parked Data As can be seen in Figure 3.9, from the top-three accepted distributions, lognormal distribution is the only one having all do not reject results from the three type of goodness of fit test. Based on this outcome, lognormal distribution will be applied for the input in building the model.
Figure 3.9 Goodness of Fit Test for Processing Time in Entry Ticket Counter Data 12
On the other hand, since the model is created using Flexsim 6.0 the distributions are cross- checked using ExpertFit. The data of inter-arrival time, processing time in entry ticket counter, and processing time in exit ticket counter have the similar results, which mean empirical distributions are deemed as the fitting distribution. Meanwhile, the outcome of ExpertFit states the duration of cars being parked could be defined better using Johnson-bounded distribution (min = 1002.309394; max = 8570.981104; shape 1 = 0.154941; shape 2 = 1.075610), so this result is the one used for input in building the model.
3.3 Model and Output After selecting the fitting distribution for each data, the following step is building the model. The model of parking area system of RSUP Dr. Sardjito is created using Flexsim 6.0 and the system layout is displayed in Figure 3.10.
Figure 3.10 Layout of Parking Area System of RSUP Dr. Sardjito
13
CHAPTER IV RESULTS AND DISCUSSION
4.1 Real System Simulation After building the model, a simulation is required in order to check whether the model could work or not. Figure 4.1 displays the simulation of the parking area system of RSUP Dr. Sardjito. In this model, cars are symbolized with boxes. As can be seen from the Figure, cars need to wait in the queue line before being processed in the entry ticket counter. A staff is assigned in both entry and exit ticket counter as an operator for processing.
Figure 4.1 Running Simulation of Model The running time for this simulation is set to 5.400 seconds, in order to create the similar condition with the real system. On the other hand, setting the running time only up to an hour and a half affects the results as the outcome in the exit ticket counter cannot be taken into further analysis. Mainly caused by the operation time of a hospital which is 24 hours, the number of cars exiting the system during the observation is most likely be the accumulation of cars staying overnight. This explains the fact that there are entities (cars) which are already leaving the system in time 0. 14
On the contrary, when the simulation is started from time 0, the system is set to free from holding any entities (cars). Since it is assumed that the parking lot is empty in the beginning of the simulation, it is not possible that there will be any cars leaving the parking area in the first few minutes. Thus, it is hardly possible that the end report of the simulation in terms of the exit ticket counter can represent the real system condition. In addition, observation of the real system indicates that problems regarding queue line in the exit ticket counter is not as urgent as in the entry ticket counter. Other than having more space for waiting line, problems in the exit ticket counter incline more to the extraneous circumstances, such as the traffic happening in front of the hospital area, and less to the process. These are basically the reasons on why the objective and further analysis of the simulation model will be more focused in the entry ticket counter in the following sections. Based on the simulation, the overall model works well without any error hindering the process. Hence, the simulation model can be proceeded into the steps of verification and validation.
4.2 Verification of System Formally, verification is addressed to determine whether the model has satisfied the conditions imposed (Calegari and Szasz, 2013). In other words, verification is the process of proving that we are building the model in the right way. In doing the verification of the system, there are four methods which are chosen. Each method is explained as follows.
4.2.1 Visual Inspection Visual inspection has been done by a member of the group who did not partake in building the model. Figure 4.2 exhibits the visualization of the simulation. Based on the inspection, the model has successfully exemplified the real parking area system of RSUP Dr. Sardjito. The model layout also has enough resemblances with the real systems layout. The servers of the real system have been represented in the model as well, including the existence of entry gate, entry ticket counter, parking lot, exit ticket counter, and exit gate. 15
Figure 4.2 Visual Inspection of Model
4.2.2 Flow Diagram Verification with flow diagram uses activity cycle diagram (ACD) to check the logic for each action in each event type. The list of activities with the activity cycle diagram is shown in Table 2.1 and Figure 2.1.
4.2.3 Animation of Simulation From the simulation, it can be seen that the model performs the exactly same process with the real system. The system starts when there is an entity passing through the queue line towards the entry ticket counter, before being processed and searching for available parking spot. During the running, the simulation shows that there are entities queuing up in the waiting line.
4.2.4 Mistake and Error Check While doing the running, there is no error found from the model. The model progresses smoothly, which means that it can be said the model has been built right and worked well.
4.3 Validation of System Validation is generally considered as the process of determining whether the model has satisfied specific requirement to some acceptable level compared to the real system. In other words, validation is a process of proving that we are building the right model (Calegari and Szasz, 2013). In validating the system, there are two types of data that will be set as the original parameter in 16
this study, which are the utility of staff in the entry ticket counter and the average waiting time of cars to enter the entry ticket counter.
4.3.1 Utility of Staff in Entry Ticket Counter For validation, simulation of model is replicated for 30 times before being statistically tested and compared to the initial outcome from the real system. Replication data of utility of staff in the entry ticket counter can be seen in Appendix 5. Test on the data adequacy states that the minimum number of data needed for validation is 11. Since the replication number is already three times the number of the data needed, it can be directly proceeded to normality and statistic tests which both are performed using Minitab 15.
4.3.1.1 Normality Test Figure 4.3 up to 4.5 show the Anderson-Darling, Ryan-Joiner, and Kolmogorov-Smirnov normality tests applied to the replication data. With the initial setting of = 0.05, all three tests give the results of p-value > . Therefore, the data on utility of staff in the entry ticket counter can be dubbed as normal and able to be used in statistic test.
32 30 28 26 24 22 99 95 90 80 70 60 50 40 30 20 10 5 1 util P e r c e n t Mean 26.67 StDev 2.189 N 30 AD 0.578 P-Value 0.121 Utilitas Petugas Loket Masuk Normal
Figure 4.3 Anderson-Darling Normality Test of Utility of Staff in Entry Ticket Counter 17
32 30 28 26 24 22 99 95 90 80 70 60 50 40 30 20 10 5 1 util P e r c e n t Mean 26.67 StDev 2.189 N 30 RJ 0.975 P-Value >0.100 Utilitas Petugas Loket Masuk Normal
Figure 4.4 Ryan-Joiner Normality Test of Utility of Staff in Entry Ticket Counter
32 30 28 26 24 22 99 95 90 80 70 60 50 40 30 20 10 5 1 util P e r c e n t Mean 26.67 StDev 2.189 N 30 KS 0.125 P-Value >0.150 Utilitas Petugas Loket Masuk Normal
Figure 4.5 Kolmogorov-Smirnov Normality Test of Utility of Staff in Entry Ticket Counter
4.3.1.2 Statistic Test The statistic test for system validation is performed using t-test. The value of parameter from the observation is calculated from the total processing time per total observation time (an hour and half). With the initial parameter of 26.02%, Figure 4.6 shows the one sample t-test done to the simulation result. (H0 : = 26.02 ; H1 : 26.02 ; = 0.05) 18
Figure 4.6 One Sample T-Test for Utility of Staff in Entry Ticket Counter As can be seen from the result, the p-value is stated to be 0.114, which is much higher than the initial of 0.05. From this outcome, we fail to reject the H0 with the statistical confidence level of 95% and it can be said that the replication data have the similar mean compared to the original data. Therefore, it can be concluded that outcomes in terms of utility of staff in the entry ticket counter have been statistically validated.
4.3.2 Average Waiting Time in Entry Ticket Counter Similar with the validation in the previous section, simulation of model is replicated for 30 times before being statistically tested and compared to the initial outcome from the real system. Replication data of average waiting time in the entry ticket counter can be seen in Appendix 5. Test on the data adequacy states that the minimum number of data needed for validation is 7. Since the replication number is already four times the number of the data needed, it can be directly proceeded to normality and statistic tests which both are performed using Minitab 15.
4.3.2.1 Normality Test Figure 4.7 up to 4.9 show the Anderson-Darling, Ryan-Joiner, and Kolmogorov-Smirnov normality tests applied to the replication data. With the initial setting of = 0.05, all three tests give the results of p-value > . Therefore, the data on average waiting time in the entry ticket counter can be dubbed as normal and able to be used in statistic test.
19
21 20 19 18 17 16 15 99 95 90 80 70 60 50 40 30 20 10 5 1 Validasi Antrian P e r c e n t Mean 18,26 StDev 1,172 N 30 AD 0,435 P-Value 0,281 Probability Plot of Validasi Antrian Normal
Figure 4.7 Anderson-Darling Normality Test of Average Waiting Time in Entry Ticket Counter
21 20 19 18 17 16 15 99 95 90 80 70 60 50 40 30 20 10 5 1 Validasi Antrian P e r c e n t Mean 18,26 StDev 1,172 N 30 RJ 0,978 P-Value >0,100 Probability Plot of Validasi Antrian Normal
Figure 4.8 Ryan-Joiner Normality Test of Average Waiting Time in Entry Ticket Counter
20
21 20 19 18 17 16 15 99 95 90 80 70 60 50 40 30 20 10 5 1 Validasi Antrian P e r c e n t Mean 18,26 StDev 1,172 N 30 KS 0,108 P-Value >0,150 Probability Plot of Validasi Antrian Normal
Figure 4.9 Kolmogorov-Smirnov Normality Test of Average Waiting Time in Entry Ticket Counter
4.3.1.2 Statistic Test The statistic test for system validation is performed using t-test. The value of parameter from the observation is calculated from the total waiting time for all cars per the number of cars who have to queue up in line. With the initial mean of 18.39 seconds, Figure 4.10 shows the one sample t-test done to the simulation result. (H0 : = 18.39 ; H1 : 18.39 ; = 0.05)
Figure 4.10 One Sample T-Test for Average Waiting Time in Entry Ticket Counter As can be seen from the result, the p-value is stated to be 0.534, which is much higher than the initial of 0.05. From this outcome, we fail to reject the H0 with the statistical confidence level of 95% and it can be said that the replication data have the similar mean compared to the original data. Therefore, it can be concluded that outcomes in terms of average waiting time in the entry ticket counter have been statistically validated.
21
4.4 Simulation Results After being verified and validated, the model is again simulated in order to derive some alternatives to answer the problem in the objective function. With the running time of 5.400 seconds, general reports from the simulation can be seen in Figure 4.11 and Figure 4.12 below.
Figure 4.11 Summary Report (Initial Condition)
Figure 4.12 State Report (Initial Condition) With the objective function of minimizing the waiting time of cars in the entry ticket counter, some variables are chosen to be the parameters of comparison in the following alternatives of optimizations. These variables comprise of average content, maximum stay time, average stay time, %blocked, and %empty of waiting line in entry ticket counter. From the simulation, the waiting line has the average content of 0.79 entities (cars) per second. Cars need to wait up to 65.41 seconds before being able to enter the entry ticket counter, with the average time of waiting is 19.05 seconds. The percentage of blocked entities is fairly high, 22
which is stated in 15.35%. Throughout the running for one and a half hour, the waiting line has the percentage of 65.85% being in the state of empty. These parameters will be the value to be appraised in deciding the optimal solution of optimization.
4.5 Experiments with Model Going by the objective of minimizing the waiting time before entering the entry ticket counter, there are two major options which could be considered for optimizations, those are: a. Replacing the entry ticket counter to vending ticket machine b. Adding one more entry ticket counter and staff alongside the existing one
4.5.1 Alternative 1: Vending Ticket Machine The idea of using vending ticket machine is derived from the real system condition in which processing is still performed manually by an assigned staff and the time could vary from 2 to 25 seconds. The moderately high variability of processing time is mainly influenced by the lacking of signs within the first few meters after the entry gate. Visitors or patients who have never visited the hospital would be confused on how to reach their particular destination of health facility, thus decide to ask the staff while getting their parking ticket. Figure 4.13 displays the layout of the model after implementing this solution.
Figure 4.13 Layout of Alternative 1: Vending Ticket Machine 23
Assuming that a vending ticket machine could perform a standardized processing time of 3 seconds, this alternative is expected to reduce the duration of waiting time of the cars before being able to enter the hospital area. With the running time of one and a half hour, Figure 4.14 and 4.15 exhibits the summary and state report of the first offered solution.
Figure 4.15 State Report (Vending Ticket Machine) Based on the reports, the average content of the waiting line is now 0.53 entities (cars) per second. With a homogenous processing time, the maximum stay time of a car has significantly decreased to 26.73 seconds and the average duration of being in the waiting line is 12.36 seconds. Using a vending ticket machine also gives a low percentage of entities being blocked, which is 5.05% and during the simulation the waiting line has the percentage of 73.60% being in the state of empty. Table 4.1 lists the comparison of the results between the initial system condition and the first alternative of optimization using vending ticket machine.
24
Table 4.1 Comparison of Results - Initial Condition and Alternative 1 Parameter Initial Condition Vending Ticket Machine Average content 0.79 entities per second 0.53 entities per second Maximum stay time 65.41 seconds 26.73 seconds Average stay time 19.05 seconds 12.36 seconds %Blocked 15.35% 5.05% %Empty 65.85% 73.60% Seeing from the data above, it can be seen that basically implementing this alternative gives positive result towards the effort to fulfil the objective function. Having a standardized processing time for all vehicles significantly reduces the duration of cars queuing up in the waiting line and subsequently decrease the percentage of entities being blocked, which also indicate that there is less queue occurring in the waiting line. On further notes, implementing this alternative will cost around US $500-700 to the management. Cost will solely depend on the type of vending ticket machine the management would like to buy. On the other hand, it is suggested to invest some more in the signs for facility locations to lessen the confusion of first-time visitors and patients since there will be no more staff around the entry gate to inform them.
4.5.2 Alternative 2: Additional Entry Ticket Counter Another option that could be implemented in order to minimize the waiting time is adding more ticket counter in the entry gate. But since it has been stated that there is not much available space in the entry area, the idea of adding more ticket counters is limited to only one counter. Adding more than one counter is most likely not possible, as the ticket counter would already invade the area of the health facilities of the hospital. Figure 4.16 displays the layout of the model after implementing this solution.
25
Figure 4.16 Layout of Alternative 2: Additional Entry Ticket Counter Even with the varying processing time, adding one more entry ticket counter is expected to somehow reduce the duration of waiting time of the cars before being able to enter the hospital area. With the running time of one and a half hour, Figure 4.17 and 4.18 exhibits the summary and state report of the second offered solution.
Figure 4.18 State Report (Additional Entry Ticket Counter) After adding one more counter, the average content is seen to be decreasing to 0.58 entities (cars) per second. The maximum stay time has reduced to 30.02 seconds and the average waiting time is 12.41 seconds. With the existence of another server, the percentage of entities being blocked is lowered to 5.23% and the waiting line has the percentage of 72.89% being empty. Table 4.2 shows the comparison of the results between the initial system condition and the second alternative of optimization with adding one more entry ticket counter. Table 4.2 Comparison of Results - Initial Condition and Alternative 2 Parameter Initial Condition Additional Entry Ticket Counter Average content 0.79 entities per second 0.58 entities per second Maximum stay time 65.41 seconds 30.02 seconds Average stay time 19.05 seconds 12.341seconds %Blocked 15.35% 5.23% %Empty 65.85% 72.89% Based on the table above, adding one more entry ticket counter also gives positive result towards the effort to fulfil the objective function. More server means that more entities could be processed at the same time, so cars need less time to queue up and also less entities being blocked from entering the system. In terms of investment, this solution is estimated to cost the management around US $1.000-1.200, which would be spent in building the counter (estimated cost is US $500), buying both hardware and software for integrated ticketing system (estimated cost is US $500), and also hiring another employee to work in the counter (local minimum wage level is estimated to be US 27
$120 per month). In addition, the management should also do bi-monthly maintenance for checking up the hardware.
4.6 Analysis and Discussion From the two alternatives of optimization, the results for each proposed solution could be appraised one to another as figured in Table 4.3 below. Table 4.3 Comparisons between Alternatives Parameter Vending Ticket Machine Additional Entry Ticket Counter Decrease of average content 0.26 0.21 Decrease of maximum stay time 59.13% 54.10% Decrease of average stay time 35.12% 34.86% Decrease of %blocked 10.30% 10.12% Increase of %empty 7.75% 7.04%
From the results above, it can be concluded that the using of entry ticket counter with vending ticket machine contributes to more significant outcomes in the objective of minimizing the waiting time. Substituting manual operator (staff) to automated one (vending machine) will affect the occupancy rate of the parking lot, which has risen from the amount of 167 cars to 183 cars for the running time of one and a half hour. But considering that this change can be considerably accepted, the proposed solution of investing in a vending ticket machine to replace the existing entry ticket counter is still advisable.
28
CHAPTER V CONCLUSION
The most crucial problem occurring in the parking area system of RSUP Dr. Sardjito happens in the waiting line to enter the hospital area. With such confined space that can only provide at most 10 cars in total, the simulation model has the objective of minimizing the waiting time of cars in the queue line. Verification and validation processes have been applied to the model in order to get as accurate result as possible to the real system. The model has been verified using some methods as well as validated with the variables of utility of staff and average waiting time both I the entry ticket counter. Taking the issues of available space and waiting time into consideration, there are two proposed alternatives of system optimization. After comparing the results, the idea of replacing the entry ticket counter to vending ticket machine gives more significant result in minimizing the waiting time in the queue line. Thus, this option is more advisable to be applied in the real system.
29
REFERENCES
Calegari, D., and Szasz, N., 2013, Verification of Model Transformations: A Survey of the State- of-the-Art, Electronic Notes in Theoretical Computer Science, vol. 292, pp. 5-25. Page, E.H., 1994, Simulation Modeling Methodology: Principles and Etiology of Decision Support, Virginia Polytechnic Institute and State University, Virginia.
30
APPENDICES Appendix 1. Observation Data of Inter-arrival Time during Morning Visiting Hours