Hi,

I'm working on a CVRPTW, where my y variable needs to show at which time my vehicle(s) arrives at a customer. My problem is that it is not working correctly for all of my customers. At the end it will subtract the distance from my upper bound time window.

Below my mod file:

int v=...; //number of customers

range V=1..v; //customers - where 1 is the depot

int d[V]=...; //demand for each customer

int k=...;//number of vehicles

range K=1..k; //number of vehicles

int C=...; //Capacity per vehicle

int c[V][V]=...; //cost between customers and depot

int e[V] = ...; // lower bound time window for each customer

int l[V] =...; // upper bound time window for each customer

int M=...; // big integer value

int S=...; // cost to use a vehicle

dvar boolean x[K][V][V]; // route driven between customer i and j

dvar int+ y[V]; // arriving time

dvar boolean g[K]; //if vehicle k is used

dexpr float m = max(a,b in V) (l[a] + c[a][b] - e[b]);

minimize sum(i,j in V, k in K) (c[i][j] * x[k][i][j]) + sum(k in K)(g[k] * S);

subject to {

 forall (j in V : j!=1)

 sum (k in K, i in V) x[k][i][j] == 1;

  

 forall(i in V : i!=1)

 sum (k in K, j in V) x[k][i][j] == 1;

  

 forall (i in V, k in K)

  x[k][i][i] == 0;

  

 sum (j in V) x[k][1][j] == 1;

  

 forall (h in V : h !=1, k in K)

  sum(i in V) x[k][i][h] - sum(j in V) x[k][h][j] == 0;

   

  forall (k in K)

  sum(j in V, i in V : i!=1) (d[i] * x[k][i][j]) <= C;

   

  forall (j in V)

   sum(k in K) x[k][1][j] <= k;

  

  forall (i in V)

   e[i] <= y[i] <= l[i];

   

  forall (k in K)

   g[k] <= sum(i, j in V) x[k][i][j];

   

  forall (k in K)

  sum(i, j in V) x[k][i][j] <= g[k] * M;

    

  forall (i,j in V : i!=j && j!=1, k in K)

   y[i] +c[i][j] <= y[j] + m* (1-x[k][i][j]);

   

   forall (k in K) 

   y[1] == 0;

 }

Right now I'm working with 45 customers (depot included, which is indicated by 1). 3 vehicles with a capacity of 500.

Thank you in advance for the help!

------------------------------
Arlien Kuijs
------------------------------

#DecisionOptimization

Can you please elaborate on what "it is not working" means? What exactly is the problem. It would also be nice to see your .dat (if possible) so that we can test your code here.

------------------------------
Daniel Junglas
------------------------------

Hi, 

Thank you for your quick response!

In the attachment I added my dat file.

the first visited customers y values are correct, since it is adding up. Some of the y variables have really high numbers, this my upper bound time window and a couple just a bit below my upper bound.




------------------------------
Arlien Kuijs
------------------------------

Original Message:
Sent: Wed June 10, 2020 04:01 AM
From: Daniel Junglas
Subject: y variable not working for CVRPTW

Can you please elaborate on what "it is not working" means? What exactly is the problem. It would also be nice to see your .dat (if possible) so that we can test your code here.

------------------------------
Daniel Junglas
------------------------------