Professional Documents
Culture Documents
Session9 Introduction To Queueing Systems
Session9 Introduction To Queueing Systems
Session9 Introduction To Queueing Systems
Operations
Jiankun Sun
Session 9
Introduction to Queueing Systems
Outline
Order queue
“buffer” size K
Incoming Answered
calls Staff calls
Calls processing
on hold calls
Case 1: Case 2:
Hour 1 2 3 4 Hour 1 2 3 4
Demand 5 5 5 5 Demand 3 3 7 7
Service 5 5 5 5 Service 3 3 6 6
Queue 0 Queue 1 2
- Unpredictable: variability
• Randomness in arrivals and service times, random availability of servers,
random routing in a process, etc.
• Queues build up even if utilization < 100%
- The Pine Valley Bank has a call centre that serves its customers.
- The management estimates that it takes a customer service
representative (CSR) 18 minutes on average to complete a call.
- There are on average 22.5 calls per hour.
- The call centre has 8 CSRs currently.
- Questions:
• Are the staff enough?
• What is the average waiting time of a customer?
• What is the probability that a customer waits for more than 1 minute?
• …
- Known facts
• Arrival rate = 22.5 calls/hr
• Service rate = 60/18 = 3.3 calls/hr/CSR
Nonlinear relationship:
Waiting time blows up
as arrival rate keeps
increasing!
}
Before GymFlow (Jan 17, 2013) 2 Weeks After 2 Months after
5:45 AM 6:45 AM 7:45 AM 8:45 AM 9:45 AM 10:45 AM 11:45 AM 12:45 PM 1:45 PM 2:45 PM 3:45 PM 4:45 PM 5:45 PM 6:45 PM 7:45 PM 8:45 PM 9:45 PM 10:45 PM 11:45 PM
𝜆
𝜇 𝜇
2𝜆
𝜆
𝜇 𝜇
System 1 System 2
- Performance comparison
• Utilization: 75% ( = 15/20) for both systems
• Average waiting time: 9 min in system 1 vs. 3.9 min in system 2
𝜆
𝜇 𝜇
2𝜆
𝜆
𝜇 𝜇
System 1 System 2
- Pooling benefits are lower when the systems that are pooled are
not truly independent.
• m = number of servers
• Utilization = flow rate/capacity = processing time/(m * inter-arrival time)
• CVa = standard deviation of inter-arrival time/average inter-arrival time
• CVp = standard deviation of processing time/average processing time