Queueing Model where inflow has to wait for outflow because of a shared channel

  • Context: Graduate 
  • Thread starter Thread starter spikkelvissie
  • Start date Start date
  • Tags Tags
    Channel Model
Click For Summary

Discussion Overview

The discussion revolves around the development of a mathematical queuing model to describe a scenario at a port where incoming boats must wait for outgoing boats due to a shared narrow channel. Participants explore the complexities of modeling this situation statically, considering factors such as arrival rates, service times, and the dynamics of boat movement within the port.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • One participant describes the scenario involving a steady arrival rate of boats and the constraints of a single queue in front of a narrow channel, emphasizing the need for a static queuing model.
  • Another participant suggests that the service time should include the total time a boat spends in the channel, which complicates the standard queuing model due to the interactions between incoming and outgoing boats.
  • Concerns are raised about the variability in service times at the berths and the impact of travel times within the port on the overall queuing model.
  • Some participants propose that a dynamic simulation might be more suitable for capturing the complexities of the scenario, while others seek a static model that can accommodate various input parameters.
  • Mathematical formulations are introduced, including the use of distributions like Poisson and gamma to describe arrival and service times, though the applicability of these models to the specific scenario remains uncertain.

Areas of Agreement / Disagreement

Participants express a range of views on the suitability of existing queuing models for the described scenario. There is no consensus on a specific model that adequately captures all variables involved, and the discussion reflects ongoing uncertainty and exploration of different approaches.

Contextual Notes

Participants note limitations in standard queuing models, particularly regarding the treatment of travel times and the interactions between incoming and outgoing boats. The discussion highlights the need for a model that can incorporate shared buffers and variable service times, which are not typically accounted for in traditional queuing theory.

spikkelvissie
Messages
3
Reaction score
0
Hi

I am look for a mathematical queuing model that can help with/solve the following scenario. I believe this scenario can me modeled in a dynamic simulation, but I am in need of a solution for a static model.

Scenario:
You are at a port. There is a steady, equally distributed arrival rate of incoming boats. A single queue of boats form in front of a channel. The channel is narrow and therefore the boats cannot pass each other in the channel. Inside the port are 2 berth with a constant service rate. The boats in the queue can only proceed to a berth if the channel is open and a berth is available. The boats that have been serviced at the berth has to return back through that same channel. If an incoming boat and outgoing boat wants to use the channel at the same time, the outgoing boat gets priority. There are no queues inside the port at the berths or at the channel going outward.

In summary: There is only 1 queue, outside the port at the channel, the queue time depends on the availability of the channel, the availability of the berth, the service time at the berth, etc.

Is there a queuing model that describes this scenario?

All of the network models I have looked at assumes the outflow uses a different channel than the inflow and that there are queues inside the port as well.

Please advise. Thank you.
 
Physics news on Phys.org
spikkelvissie said:
Hi

I am look for a mathematical queuing model that can help with/solve the following scenario. I believe this scenario can me modeled in a dynamic simulation, but I am in need of a solution for a static model.

Scenario:
You are at a port. There is a steady, equally distributed arrival rate of incoming boats. A single queue of boats form in front of a channel. The channel is narrow and therefore the boats cannot pass each other in the channel. Inside the port are 2 berth with a constant service rate. The boats in the queue can only proceed to a berth if the channel is open and a berth is available. The boats that have been serviced at the berth has to return back through that same channel. If an incoming boat and outgoing boat wants to use the channel at the same time, the outgoing boat gets priority. There are no queues inside the port at the berths or at the channel going outward.

As far as I can tell, you have a standard queue. The only difference is that the service time includes the entire time a ship is "in port" including from the time an incoming ship enters the channel until that same ship clears the channel on its way out. Even if there are two berths, the time a boat time spends in the channel is part of the service time for calculating mean waiting time. Having two berths potentially reduces mean service time at the berths, but total service time has to include the bottleneck the narrow channel represents.

EDIT: At the start of the service day you can bring two ships in, one behind the other. However, the variance of the service time at the two berths will cause staggering over time. You can use dynamic simulation to see how the variance of service time at the berths affects mean total service time. Note with the Poisson distribution, which usually describes arrival times, the variance is equal to the mean while the gamma distribution is often used for service times where the variance is k \theta^2.
 
Last edited:
Thank you. I guess the bigger problem is that there are many variables a standard queuing model doesn't account for. For example the travel time inside the port, more than one ship can use the channel behind each other, the time it takes to get through the channel vs. the time it takes to travel from the berth to the channel. The dynamic model would probably be best, but my colleague asked me to look into a way to solve this statically. I am no expert in queuing theory, just did it in University. Making the service time include the time it takes for the boat to leave the port would make this a much simpler problem, but I believe there will be instances where a ship can enter the port while the boat from the berth is still on it's way to the channel, it is this variability that makes it complicated.

It would be ideal if there existed a queuing model where instead of just having arrival and service rate as an input, one can input the shared buffer, the rate of the buffer and the travel time between the buffer and the service point. I believe this travel time has to be separate from service time, since another boat my enter the port while the previous boat is traveling in the port. Unfortunately I don't have any solid data at the moment, I was looking for a generic model which can be used given different types of ports with the same setup.

Thank you for your help. Regards
 
spikkelvissie said:
Thank you. I guess the bigger problem is that there are many variables a standard queuing model doesn't account for. For example the travel time inside the port, more than one ship can use the channel behind each other, the time it takes to get through the channel vs. the time it takes to travel from the berth to the channel. The dynamic model would probably be best, but my colleague asked me to look into a way to solve this statically. I am no expert in queuing theory, just did it in University. Making the service time include the time it takes for the boat to leave the port would make this a much simpler problem, but I believe there will be instances where a ship can enter the port while the boat from the berth is still on it's way to the channel, it is this variability that makes it complicated.

It would be ideal if there existed a queuing model where instead of just having arrival and service rate as an input, one can input the shared buffer, the rate of the buffer and the travel time between the buffer and the service point. I believe this travel time has to be separate from service time, since another boat my enter the port while the previous boat is traveling in the port. Unfortunately I don't have any solid data at the moment, I was looking for a generic model which can be used given different types of ports with the same setup.

Thank you for your help. Regards

You're welcome, but I think your main interest is the mean queue outside of the channel. Queuing theory, like any probabilistic model, is for calculating the expectation (mean) and the variability around the mean. The theory is that over extended time, the actual mean queue approaches the calculated mean queue. If you treat the entire port operation as a "black box" your only interest is the time between ships entering the channel vs the time between new arrivals to the queue. Mathematically the black box might be modeled as:

S= F(t, k, \theta) + C_t where t is time, (k=1) is a random service encounter using a test variable where \theta ^ {-1} is the mean joint rate of ships served by the two berths. The test variable has an exponential distribution. C is mean transit time within the port (in and out). You most likely would be using the Erlang distribution for some k once you've established a mean service time.

Even with a simulation to evaluate all kinds of special situations, you need information on mean values and measures of variation (variance). With this model, the variance is calculated from the parameters.
 
Last edited:
SW VandeCarr said:
You're welcome, but I think your main interest is the mean queue outside of the channel. Queuing theory, like any probabilistic model, is for calculating the expectation (mean) and the variability around the mean. The theory is that over extended time, the actual mean queue approaches the calculated mean queue. If you treat the entire port operation as a "black box" your only interest is the time between ships entering the channel vs the time between new arrivals to the queue. Mathematically the black box might be modeled as:

S= F(t, k, \theta) + C_t where t is time, (k=1) is a random service encounter using a test variable where \theta ^ {-1} is the mean joint rate of ships served by the two berths. The test variable has an exponential distribution. C is mean transit time within the port (in and out). You most likely would be using the Erlang distribution for some k once you've established a mean service time.

Even with a simulation to evaluate all kinds of special situations, you need information on mean values and measures of variation (variance). With this model, the variance is calculated from the parameters.

Thank you, this helps a lot. I really appreciate it. Regards
 

Similar threads

Graduate Calculating Waiting Time with Weighted Servers in Queueing Theory
  • · Replies 3 ·
Replies
3
Views
2K
Undergrad Real life queue - how to model?
  • · Replies 2 ·
Replies
2
Views
2K