Simulering av industriella processer och logistiksystem MION40, HT 2012 Simulation Project Improving Operations at County Hospital County Hospital wishes to improve the service level of its regular X-ray operation which runs from 8 am to 8 pm. Patients have identified the total required time as the major service issue in this process. Management, on the other hand, is concerned with utilization of available resources. A process design team has been formed to study this problem. The process might be redesigned as a result of the team s recommendations. The team has defined the entry point to the X-ray process to be the instant a patient leaves the physician s office en-route to the X-ray lab. The exit point has been defined as the instant at which the patient and the completed X-ray enter the physician s office. There are broadly speaking two types of patients arriving to the X-ray process - emergency and non-emergency patients (priority level 1 and 2 respectively). The emergency patients arrive according to a Poisson process with a mean of 4 patients per hour. The non-emergency patients are registered as they enter and a sample of the arrival time data is provided in the Appendix. So far, no attempt has been made to further analyze this data, and there is no insight into what the arrival process looks like. The team has identified 12 activities in the current X-ray process (see Table 1), which is the same irrespective of the patient type. The only differences between patient categories are the activity times and their distributions, specified in Table 3. It should be noted that: The patient priority levels determine the service order at all the X-ray activities. Emergency patients (priority 1) come first at the expense of non-emergency patients (priority 2). However, ongoing service will never be interrupted at the benefit of high priority patients. 1
Table 1. Activities in the X-ray process Activity Description Type 1 Patient leaves physician s office with instructions. Start of the X-ray process 2 Patient is taken to the lab by an orderly on foot, in a Transportation wheel chair or lying in a bed. 3 The patient is left in the waiting area outside the X- Waiting ray lab in anticipation of an X-ray technician. 4 An X-ray technician fills out a standard form based on information supplied by the physician and the patient (done outside the X-ray lab). The technician then leaves the patient, which queues up in front of the X-ray labs. 5 The patient enters the X-ray lab, undresses and an X- ray technician takes the required X-rays (all done in the X-ray lab). 6 A dark room technician develops the X-rays. (Assume that the patient and the X-ray technician accompanies the X-rays) 7 The dark room technician and the X-ray technician check the X-rays for clarity. (Assume that the patient accompanies his/her X-rays) 8 If the X-rays are not clear, the patient needs to go back to the waiting room in anticipation of repeating steps 5, 6 and 7. Historically, the probability of rejecting X-rays has been 25%. If the X-rays are acceptable the patient proceeds to activity 9 while the X-rays are put in the outbox, where eventually the messenger service will pick them up. Business value added Value added Value added Inspection Decision 9 Patient waits for an orderly to take him/her back to Waiting the physician s office 10 Patient is taken back to the physician s office by an Transportation orderly. 11 A messenger service transfers the X-rays to the Transportation physician s office in batches of five jobs. 12 Patient and X-rays enter physician s office together. End 2
The resource data for the X-ray process are specified in Table 2 below. Table 2. Resource data for X-ray process Resource Activities No. of units available Orderlies 2 and 10 3 X-ray technician 4, 5, 6 and 7 3 X-ray lab 5 2 Dark room technician 6 and 7 2 Dark room 6 1 The orderlies will always try to take a patient back from the X-ray when they have dropped one off. Assume that the transportation time back from the X-ray lab is exactly the same as the transportation time to the X-ray lab. If no patient is ready to go back the orderly will wait for five minutes, if no patient becomes available during this time the orderly will return to the ward without a patient. The time for an orderly to walk back without a patient is always 5 minutes. The orderlies will never go and pick up a patient at the X-ray area without bringing another patient with them from the ward. Table 3. Activity times Activity Patient type Activity time Parameter values (minutes) distribution 1 All types Not Applicable Not Applicable 2 Emergency patients Non-emergency patients Max=9, Min=5 Max=12, Min=5 3 All types Not Applicable Not Applicable 4 All types Max=6, Min=4 5 Emergency patients Non-emergency patients Normal Normal =9, =4 =11, =4 6 All types Normal =12, =5 7 All types Constant Value=2 8 All types Constant Value=0 9 All types Not Applicable Not Applicable 10 Emergency patients Non-emergency patients Max=9, Min=5 Max=12, Min=5 11 All types Max=7, Min=3 12 All types Not Applicable Not Applicable 3
Assignments Analyzing the current process design Phase I a) Draw a flow chart over the X-ray process. The flowchart should describe the process in detail. A careful documentation and a good understanding of the process will make it easier to build the simulation model in phase 2. The flowchart should be drawn on the computer. The most important objective is to convey an unambiguous and detailed description of the process, using a particular syntax or set of symbols is of less importance. It is recommended not to proceed with the following questions before you are sure that the solution to a) describes the problem correctly. Phase II b) Develop a simulation model of the process. Modeling hint: For batching and un-batching, use the blocks Batch and Unbatch from the Discrete Event Library with the option Preserve uniqueness checked. Modeling hint: Build the model incrementally based on your flowchart, do not try to put everything together at once and then test if it works. The keyword is Baby steps As a first check that everything works as supposed to it is often useful to run a shorter simulation with animation. Use different symbols for different types of items, for example different types of labor and different types of jobs. c) As a first cut analysis, run a 1 day simulation with Random seed=100, using the correct activity time distributions. Look at the average cycle time, the throughput, the resource, the queue and the activity statistics. What seems to be the problems in this process? 4
Phase III d) Analyze the input data regarding non-emergency patients inter-arrival times. Use the statistical tools discussed in class to determine a distribution that adequately describes the data. The analysis should be done without the help of the statfit software available in Extend. e) Simulate 30 days of operation and compute the cycle time and daily throughput (average and standard deviation). Also compute the activity and resource utilization statistics and queue statistics with 95% confidence intervals (use the statistics blocks in Extend). Assume that any patients remaining in the system at the end of the day will be taken care of by the night shift. Every morning the system is considered empty. Are there any surprises compared to the results in c) above? f) Assess the performance of the process using the values calculated in e). Where is the bottleneck? What seems to be the problem in terms of cycle time and throughput? Suggest and evaluate a new process design g) Based on your insights about current operations suggest at least three plausible design changes with the purpose of reducing the average cycle time and increasing the throughput. h) Investigate the performance of the new process design in terms of the cycle times, and the daily throughput. Also, look at the resource, activity utilization and queue statistics with 95% confidence intervals as before. What is your conclusion? Is the new design significantly better than the old process in terms of cycle times and daily throughput (the conclusion must be statistically supported)? Are there any obvious drawbacks? Instructions for the final report Each group should produce a well-written and well-structured typed report (OBS! One report per group!). The report should explain the problem, your approach to solve it, your findings, results and conclusions. The questions a)-h) above obviously need to be treated fully and model printout and data on which you base your conclusions needs to be made available in appendices. The underlying Extend models should be provided in electronic format via e-mail to Fredrik Olsson. The report, excluding the electronic Extend models, shall be submitted in paper format to Fredrik Olsson. Each report must clearly state the name of all the group members. 5
Appendix Arrival data for non-emergency patients Patient # Time of arrival (in minutes from time zero) Patient # Time of arrival (in minutes from time zero) 1 1.49 31 183.77 2 3.01 32 183.80 3 6.29 33 185.46 4 7.76 34 187.54 5 10.87 35 199.79 6 13.54 36 213.09 7 45.13 37 227.22 8 47.03 38 235.60 9 47.97 39 259.43 10 52.02 40 272.97 11 53.81 41 275.60 12 55.07 42 286.18 13 57.25 43 287.90 14 59.56 44 293.97 15 73.87 45 297.86 16 92.60 46 298.56 17 111.60 47 306.59 18 122.03 48 321.28 19 128.93 49 325.51 20 133.67 50 328.73 21 135.82 51 329.21 22 138.01 52 329.50 23 138.53 53 339.11 24 144.39 54 343.71 25 149.49 55 356.54 26 158.34 56 356.68 27 170.66 57 366.32 28 173.11 58 371.05 29 176.59 59 374.01 30 177.85 60 374.86 This material is also available in electronic format on the WEB. 6