Same nodes but different solution time (VRP Time Windows)

回答済み

Rihot Gusron

2022年07月20日 08:09 編集済み

• 
• 
• 

Hi gurobi support,

i just modeled the VRP with Time Windows from this book.

The things is i did tried using dummy datas first, with 10 nodes and all the parameter needed for the model to run. The running time was short and can get the solution. But when i tried with the same nodes (10 nodes) but with real datas, the solver still running for like 8 hours. Is anyone can give me some advice about this?

My datas look like this:

Earliest time(at each nodes) : All zero

Latest time (at each nodes) : All 480 (in minutes)

Demand(at each nodes) : [0 50 50 60 50 90 50  50 48  50 ]

Vehicle Capacity (at each vehicle) : [2000 1000 500]

 

1

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2025年11月16日 23:32

  This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum, or try Gurobot, our chatbot interface offering instant, expert-level support.
• 

  David Torres Sanchez

  • Gurobi Staff

  2022年07月20日 11:13 編集済み

  Hi Rihot,
  
  Sounds like there is an error in your formulation.
  You can try writing it to a LP file and make sure that the constraints match the formulation in the book: \(\texttt{model.write("vrptw.lp")}\).
  
  Cheers, 
  David

  1
• 

  Rihot Gusron

  2022年07月20日 16:47

  • 
  • 
  • 

  Hi Mr. Sanchez,

  Thanks for the response. I did double check the model and seems can't find the error. Actually the time windows constraint seems suspicious, as when i comment that out, the solver was found the solution in a matter of second. I just extracted the lp model and hosted on google drive. Do you mind to have a look and maybe give any insight/advice about this?

  The link to .lp : click here

  Best Regard.

  0
• 

  David Torres Sanchez

  • Gurobi Staff

  2022年07月21日 10:06 編集済み

  Hi Rihot,

  You are right, as far as I can tell the time window constraints are fine. Looks like these:

  $$u_{ik} +s_i + t_{ij} - u_{jk} \leq M(1-x_{ijk})~~\forall (i,j)$$

  With \(M=1000\).
  
  Maybe try constraining the time windows a little bit more, e.g. provide a lower window and maybe a different upper window for each node.
  If you are struggling with data creation, you can try using some real instances available and just grab fewer nodes: http://web.cba.neu.edu/~msolomon/c101.htm
  
  Cheers, 
  David

  0

投稿コメントは受け付けていません。