SMU Evacuation Simulation
Senior Design: 2008
Conner Huckaby
Katelyn Pratt
Sarah Rives
Table of Contents
Management Summary 3
Background 4
Problem Situation 5
Analysis of the Situation 5
Technical Description of Pro-Model 5
Model 1: Actual Model Used by SMU 8
Model 2: Split Moody Exits 9
Model 3: Utilize Ownby Drive 11
Model 4: Force Exits 12
Conclusion and Critique 14
Appendix 1: Model 1 – Actual Model Used by SMU 16
Appendix 2: Model 2 – Split Moody Exits 23
Appendix 3: Model 3 – Utilization of Ownby Drive 32
Appendix 4: Forced Exits 41
Appendix 5: Supporting Documents from Traffic Access Study for SMU in University Park, TX. Jun 27, 2002 by DeShazo, Tang & Associates, Inc 48
Management Summary
Southern Methodist University, commonly referred to as SMU, will finish the construction of the Binkley parking garage in the summer of 2008. This new facility will open 850 new parking spaces on campus. However, the Binkley garage opening will cause new traffic troubles on campus because this parking structure is located kiddy-corner to Moody Garage, which has 830 parking spaces.
In the 2008 fall football season, SMU could see full parking facilities due to the combination of student, faculty, and visitor parking. Currently, SMU expects a car at the top of Moody garage to take 45 minutes to exit campus. But this 45 minute benchmark was calculated without taking into consideration the new Binkley garage. If both garages are full, how will traffic exit campus? Furthermore, how fast can both garages empty out? Is there a way to route traffic through campus in order for individual cars to exit campus quickly and with as little blockage and waiting time as possible?
This project simulates four different options SMU can take to route exiting traffic through campus with ProModel software. It begins by simulating the current evacuation plan in place. Pinpointing problems allowed us create new models to alleviate intersections, queues, and locations and ultimately decrease the time it takes to clear campus.
After analyzing all four models, we found that SMU should abolish its current traffic evacuation plan. SMU furthermore should implement an evacuation plan where traffic controllers will force traffic to exit in specific locations based on what garage they are in. The new plan will clear campus 13 minutes faster and decreases the percentage of full locations.
Background
In 1991, Allen Pitts and John Edleman senior design project formulated a traffic simulation for Southern Methodist University during fall football games. Their main goal was to assist SMU by simulating the flow of traffic around the general SMU area, make new suggestions as to where problems may occur, and give recommendations on how to alleviate such problems.
Seventeen years later, SMU still has traffic and parking problems around campus, especially during football games. In 2008, SMU constructed a new parking facility with 850 spaces along Binkley Ave to help alleviate campus parking problems. This parking garage is one block away from the already existing Moody Garage which offers 830 parking spaces. Additionally, SMU is acquiring the George W Bush Presidential Library which will add additional traffic to campus and additional parking.
Our senior design group was originally interested in recreating the 1991 traffic simulation created by Pitts and Edleman. We wanted to incorporate traffic with the new Binkley parking garage and the George W Bush Presidential Library.
However, our original project idea quickly changed. After meeting with Mark D Rhodes, Director of Parking and ID Card Services, we realized that the more immediate need lies with the evacuation of both the Moody and Binkley parking garages during football games in the fall of 2008. There was no need to take into account the presidential library, as it would not begin constructed until after the football season.
Problem Situation
SMU’s current evacuation plan after an event such as an SMU football game is inadequate because of the addition of the new Binkley garage. Rhodes hoped that we could create a new plan for routing cars exiting campus. We had to take into account the current organizational plan in place. This plan blocks off roads, blocks off parking spaces, and hires people to help direct traffic. We also decided to narrow our focus on only the traffic exiting from the two garages: Moody and Binkley.
Our main goal is to create a traffic evacuation simulation for SMU football games that will allow traffic parked in the main garages to exit campus as efficiently as possible, while taking in to account University parking needs.
Analysis of the Situation
Our general approach to the SMU traffic evacuation problem was to create an exiting traffic simulation. We felt that ProModel would be useful in analyzing the situation because ProModel is predictive technology that provides simulation-based, decision making tools and techniques to improve performance.
We first simulated the current evacuation plan SMU has in place. We then examined the results and pinpointed key problems. Upon investigation we realized that the key problems were high intersection utilization, queues that are too full, flow restrictions, blocked locations, and full locations. Pinpointing these problems allowed us to utilize additional model options to alleviate intersections, queues, and locations. We simulated each model separately and compared the time to clear in order to choose the best alternative exiting plan.
For each model, we followed the current SMU plan in place and kept the Boulevard and the areas surrounding the Meadows Museum off limits to through traffic. This left any exit in the southwest area of campus unavailable. Additionally, we assumed that Binkley Ave could be used if necessary. Furthermore, we assumed any exits to the north through the Highland Park neighborhood are extraneous and could be combined to exit at Airline and Hillcrest and any additional exits to the 75 access road through the Highland Park neighborhood could be combined to exit at SMU and 75.
Technical Description of Pro-Model
ProModel is a discrete event simulation tool with many applications to operations research. It is primarily used for the planning and design of warehouses and logistics. ProModel is a combination of a simulation tool and animation creator, allowing for graphical input of modeling constraints. Each model is created by independent timing and capacity restrictions.
The creation of the timing constraints of each model consists of defining the speed of the entity (individual cars in this situation) and the lengths of the various paths. The speed of the entity used was chosen as an average speed of cars during similar loads. Car direction was also defined based on the observations by DeShazo, Tang & Associates, Inc shown in Appendix 5. In addition to the constants involved with simulating the model, defined factors were manually coded. The time factors coded include traffic lights and intersection behavior. The traffic lights were created by sampling the simulation clock and dividing by cycle time, resulting in the remainder of time per entity in the cycle. The entities are then forced to wait until the next cycle begins before they proceed. The intersection behavior was created by modeling the average time from a sample of hesitation times during different times throughout the day.
Separate from timing, the capacity of the system is determined by entity size and queue length. Each queue models one direction of a street. The lengths of the queues were determined by standard maps of the campus. By defining the individual queue capacity, the system determines the wait time that will incur at the previous intersection.
The simulation is started by defining arrivals at the three garages. The arrivals are the entrance of the entity into the system being simulated. The total number of arrivals was determined by the capacity of the two garages. Arrival time was determined by an exponential distribution and average speed out of the garage from a sample of tests in the current garages. An exponential distribution was chosen because it was the distribution that modeled the nature of our simulation the best, a high volume at the start of the simulation decreasing over time. By simulating an exponential variation in the sample garage flows, we are able to accurately simulate the real situation.
The output of ProModel is a statistical analysis of the time and constraint utilization during the simulation. By observing stressed points of the model we are able to pinpoint the potential areas for change to decrease the overall run time of the model.
Model 1: Actual Model Used by SMU
Overview
Model 1 is the actual evacuation plan SMU has in place. It utilizes only one exit out of each garage. This is mainly because Binkley Ave is blocked between Airline and Dublin and because the Boulevard is blocked during game days. The major intersections at play here are between Binkley/ Airline and SMU/ Airline.
Flowchart
Analysis
The current plan takes cars 80.64 minutes to clear campus. The intersection at SMU/Airline was utilized the most at 94.6%. The busiest queue was the queue at SMU/ Airline and was full 87.2 % of the time. Given the limited amount of options generated by this plan is not surprising that the intersections at Binkley/Airline and at SMU/Dublin were also greatly utilized: 92.32% and 92.38%. It is also important to note that the intersection at Binkley/Airline was blocked 39.6 % of the time. If this plan stays in place, the average car will take 9.32 minutes to exit and will be blocked for 5.69 minutes. Therefore, a car can expect to be blocked 60.5% of the time. In order to decrease clearing time we must look at decreasing the amount of traffic at these three intersections and alleviating the queue at SMU/ Airline.
Model 2: Split Moody Exits
Overview
Model 2 attempts to alleviate the three intersections highly utilized in Model 1 by using both exits out of Moody Garage: the south and the north. This model also opens up Binkley Ave for through traffic in order to alleviate traffic at the SMU/Airline queue. Model 2 also continues to only route the traffic coming out of the one exit of Binkley Garage to the east in order to avoid the Boulevard on game days.
Flowchart
Analysis
In model 2, it takes cars 70.57 minutes to clear campus. This model did accomplish our goal of freeing up the SMU/Airline intersection which is now utilized only 26.4%. Model 2 also decreased the utilization of the queue at SMU/Airline to 6.1% and this queue is never full. However, this model increased how much the intersection at Binkley/Airline was used to 94%. But, the percentage of cars blocked at the Binkley/Airline intersection decreased to 33.8%. It is also important to note that between Model 1 and Model 2 the average time per entry at Binkley Garage increased. The queue at the Binkley Garage exit is full 90.4 % of the time. On the downside, this model overall increases the amount of time a car is blocked to 6.53 minutes, meaning a car will be blocked 64.4% of the time it is trying to exit campus. In order to further decreasing clearing time, we hope to decrease the fullness at the queue outside of Binkley Garage and decrease the amount the intersection at Binkley/Airline is used. We also hope to decrease the percentage of time a car is blocked while exiting campus.
Model 3: Utilize Ownby Drive
Overview
Model 3 attempts to alleviate traffic at the Binkley/ Airline intersection and continue to relieve traffic at the SMU/ Airline intersection by routing north bound traffic out of Binkley Garage through Ownby Drive. This model does not take into account pedestrian traffic, as pedestrians are more likely since Ownby is closer to the tailgating festivities. Model 3 also continues to route traffic out of both the northern and southern exits of Moody Garage.
Flowchart
Analysis
Cars in Model 3 take 222.2 minutes to clear campus: clearly our worst model by far. Model 3 did manage reach our goal to decrease the percentage utilization at the intersections at Binkley/Airline and Binkley/ Dublin 80.2% and 85.75%. However, the average contents at Binkley/ Dublin queue increased to 24.3. Also, the percent the south exit queue was used increased by an additional 10% to 28.6%. Also detrimental to this model is the fact that we did not solve the percentage of cars blocked at the Binkley/Airline or the Binkley/Dublin intersection. In fact, we made it worse. Binkley/Airline is now blocked 80.1% of the time and Binkley/Dublin is blocked 57.9% of the time. Model 3 did manage to decrease the percentage that the queue outside of Binkley Garage was full to 79.46 percent. However, the queue at Binkley/Dublin and the queue at the South exit increased to being full 80.4% and 85.47% of the time. Furthermore, cars are blocked on average 50.52 minutes in model 3. This means that a car will be blocked 71.94% it is trying to exit campus. This is the worst exit time, blocked time, and percentages full yet. Based on the data, Model 3 should never be implemented.
Model 4: Force Exits
Overview
Model 4 takes a completely different approach to exiting traffic from campus. Instead of cars making a choice of which direction to exit, we have traffic control staff make the decision for them. In Model 4 we direct all traffic exiting Moody Garage east to Interstate 75. We also direct all traffic leaving Binkley Garage only to the north and south exits. We split number of cars going to the north and south, but still reflected that less traffic will choose north rather than south. It is also important to mention that Moody Garage will once again only have one exit location to the north. Binkley Ave would still be utilized for through traffic, just not as an exit out of Moody Garage. We attempted a forced approach because this traffic simulation is commonly found after large events, such as after a Dallas Cowboys football game. Furthermore, this model hopefully will continue to alleviate traffic at the SMU/ Airline and Airline/ Binkley intersections.