MAJ Justin (JJ) Dwyer Advisor: Dr. Jeffery D. Weir Department of Operational Sciences (ENS) Air Force Institute of Technology (AFIT) Analysis of Military Entry Control Point Queueing Model 3: Additional Server Tandem INTRODUCTION Military Entry Control Facility (ECF) mission and purpose is to provide security to the instillation from unauthorized access and intercept contraband while maximizing traffic flow. Design of an ECF should maximize traffic flow without compromising security, safety or causing undue delays that may affect off- instillation public highway users or instillations operations. Overflow from the ECF queue can disrupt civilian traffic utilizing the surrounding roads of the instillation. PURPOSE Examine interactions that would minimize the interference of military ECF traffic with civilian population. Focus on arrival rates of vehicles, processing time of the guards, customer actions, and deploying additional server. Endstate: Provide insight to ECF operations that could lower interference with civilian traffic. METHODOLOGY All Models: - - Utilized discrete event simulations in SIMIO for models - Each model was replicated 30 times Model 1: Split to Individual Queue Model 2: Illogical Customers Model 3: Additional Server to Queue System Individual Lane Queue Overall ECP Queue Tandem Servers S S Parallel Servers S S 1 Service Distribution: Exp 3 3 µ − + Illogical Customers Blocking Access to Right Lane S S Single Server S Tandem Servers S S RESULTS & ANALYSIS Model 1: Split to Individual Queue Insight: Single lane queue of 20 is sustainable for utilization rates ρ 2 < 0.9. Model 2: Illogical Customers At each probability of choosing left lane we can find a point at which the overall queue length is no longer affected by the illogical customer choice. Insight: Fully open both lanes for queue not be affected by the illogical customer decisions. Model 3: Additional Server to Queue System Insight: If an ECF is capable of opening a parallel server, choose parallel server over tandem server. Parallel Utilization levels ρ < 1.6 open 2 nd sever at queue length < 17. Utilization levels ρ < 1.8 open 2 nd sever at queue length < 13. Utilization levels ρ < 2 open 2 nd sever at queue length < 5. Tandem Tandem model is only capable of sustaining a queue less than or equal to 20 for models with utilization levels p < 1.11 Insight: Tandem model servers act as batch server with batch service time equal to the max of the individual service times. CONCLUSION Model 1: Split and Model 2: Illogical Customers both emphasized the importance of fully opening both lanes of traffic to allow vehicles access to both queues From our analysis it was observed that tandem servers acted as a batch service with the batch service time equal to the max of the individual service times. If the focus of adding additional server is to reduce overall queue length, then the ECF should utilize parallel servers. Processing Time (1/μ) Inter-arrival Time (1/λ) Two Server ρ 2 = (λ/2μ) Length of Individual Queues Unrecoverable (Percent) Over Occurances (per hour) Recovery Time (minutes) 13.1 6.67 0.98 > 13 3.33% 0.183333333 4.47 ± 3.85 9.6 5.00 0.96 > 9 3.33% 0.1 4.81 ± 2.85 8.0 4.29 0.93 > 10 0.00% 0 - 11.07 6.00 0.92 > 4 0.00% 0 - 12 6.67 0.90 > 0 0.00% 0 - 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 2 4 6 8 10 12 14 16 18 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Over Occurances Per Hour Queue Recovery Time (Minutes) Queue Length at Which 2nd Server is Opened Processing Time (9 sec) Interarrival Time (8.57 sec*) ρ = 1.05 (1 Server) Recovery Time Ave Top / Bottom 95% Number of Occurances 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45 50 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Over Occurances Per Hour Length of Time 2nd Server Open (Minutes) Queue Length at Which 2nd Server is Opened Processing Time (9 sec) Interarrival Time (8.57 sec*) ρ = 1.05 (1 Server) Average Open Time Top / Bottom 95% Number of Occurances 0 5 10 15 20 25 30 35 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Average Ending Queue Length Queue Length at Which 2nd Server is Opened Processing Time (9 sec) Interarrival Time (8.57 sec*) ρ = 1.05 (1 Server) End Queue Length Top / Bottom 95% Scoring (from Fitness Function) - Three options were tied with a score of 4 (Queue = 3, 5, & 9) - Tie breaker goes to the longest queue length - “Best” option to open 2 nd server for this model is when queue length = 9 0 10 20 30 40 50 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Average Overall Queue Length Individual Lane Length Probability of Choosing Left Lane = 0.5 ρ = 0.96 Exp (9.6) ρ = 0.94 Exp (8) ρ = 0.90 Exp (12) ρ = 0.88 Exp (9.6) ρ = 0.87 Exp (8) ρ = 0.80 Exp (8) ρ = 0.80 Exp (9.6) ρ = 0.80 Exp (12) ρ = 0.72 Exp (9.6) ρ = 0.70 Exp (12) Overal queue length no longer affected by illogical customer decision when individual queue length equal this cutoff.