This example shows how to set up and solve a mixed-integer linear programming problem. The problem is to find the optimal production and distribution levels among a set of factories, warehouses, and sales outlets. For the solver-based approach, see Factory, Warehouse, Sales Allocation Model: Solver-Based.
The example first generates random locations for factories, warehouses, and sales outlets. Feel free to modify the scaling parameter , which scales both the size of the grid in which the production and distribution facilities reside, but also scales the number of these facilities so that the density of facilities of each type per grid area is independent of .
For a given value of the scaling parameter , suppose that there are the following:
These facilities are on separate integer grid points between 1 and in the and directions. In order that the facilities have separate locations, you require that . In this example, take , , , and .
There are products made by the factories. Take .
The demand for each product in a sales outlet is . The demand is the quantity that can be sold in a time interval. One constraint on the model is that the demand is met, meaning the system produces and distributes exactly the quantities in the demand.
There are capacity constraints on each factory and each warehouse.
The production of product at factory is less than .
The capacity of warehouse is .
The amount of product that can be transported from warehouse to a sales outlet in the time interval is less than , where is the turnover rate of product .
Suppose that each sales outlet receives its supplies from just one warehouse. Part of the problem is to determine the cheapest mapping of sales outlets to warehouses.
The cost of transporting products from factory to warehouse, and from warehouse to sales outlet, depends on the distance between the facilities, and on the particular product. If is the distance between facilities and , then the cost of shipping a product between these facilities is the distance times the transportation cost :
The distance in this example is the grid distance, also known as the distance. It is the sum of the absolute difference in coordinates and coordinates.
The cost of making a unit of product in factory is .
Given a set of facility locations, and the demands and capacity constraints, find:
A production level of each product at each factory
A distribution schedule for products from factories to warehouses
A distribution schedule for products from warehouses to sales outlets
These quantities must ensure that demand is satisfied and total cost is minimized. Also, each sales outlet is required to receive all its products from exactly one warehouse.
The control variables, meaning the ones you can change in the optimization, are
= the amount of product that is transported from factory to warehouse
= a binary variable taking value 1 when sales outlet is associated with warehouse
The objective function to minimize is
The constraints are
(capacity of factory).
(demand is met).
(capacity of warehouse).
(each sales outlet associates to one warehouse).
The variables and appear in the objective and constraint functions linearly. Because is restricted to integer values, the problem is a mixed-integer linear program (MILP).
Set the values of the , , , and parameters, and generate the facility locations.
rng(1) % for reproducibility N = 20; % N from 10 to 30 seems to work. Choose large values with caution. N2 = N*N; f = 0.05; % density of factories w = 0.05; % density of warehouses s = 0.1; % density of sales outlets F = floor(f*N2); % number of factories W = floor(w*N2); % number of warehouses S = floor(s*N2); % number of sales outlets xyloc = randperm(N2,F+W+S); % unique locations of facilities [xloc,yloc] = ind2sub([N N],xyloc);
Of course, it is not realistic to take random locations for facilities. This example is intended to show solution techniques, not how to generate good facility locations.
Plot the facilities. Facilities 1 through F are factories, F+1 through F+W are warehouses, and F+W+1 through F+W+S are sales outlets.
h = figure; plot(xloc(1:F),yloc(1:F),'rs',xloc(F+1:F+W),yloc(F+1:F+W),'k*',... xloc(F+W+1:F+W+S),yloc(F+W+1:F+W+S),'bo'); lgnd = legend('Factory','Warehouse','Sales outlet','Location','EastOutside'); lgnd.AutoUpdate = 'off'; xlim([0 N+1]);ylim([0 N+1])
Generate random production costs, capacities, turnover rates, and demands.
P = 20; % 20 products % Production costs between 20 and 100 pcost = 80*rand(F,P) + 20; % Production capacity between 500 and 1500 for each product/factory pcap = 1000*rand(F,P) + 500; % Warehouse capacity between P*400 and P*800 for each product/warehouse wcap = P*400*rand(W,1) + P*400; % Product turnover rate between 1 and 3 for each product turn = 2*rand(1,P) + 1; % Product transport cost per distance between 5 and 10 for each product tcost = 5*rand(1,P) + 5; % Product demand by sales outlet between 200 and 500 for each % product/outlet d = 300*rand(S,P) + 200;
These random demands and capacities can lead to infeasible problems. In other words, sometimes the demand exceeds the production and warehouse capacity constraints. If you alter some parameters and get an infeasible problem, during solution you will get an exitflag of -2.
To begin specifying the problem, generate the distance arrays
distfw = zeros(F,W); % Allocate matrix for factory-warehouse distances for ii = 1:F for jj = 1:W distfw(ii,jj) = abs(xloc(ii) - xloc(F + jj)) + abs(yloc(ii) ... - yloc(F + jj)); end end distsw = zeros(S,W); % Allocate matrix for sales outlet-warehouse distances for ii = 1:S for jj = 1:W distsw(ii,jj) = abs(xloc(F + W + ii) - xloc(F + jj)) ... + abs(yloc(F + W + ii) - yloc(F + jj)); end end
Create variables for the optimization problem.
x represents the production, a continuous variable, with dimension
y represents the binary allocation of sales outlet to warehouse, an
x = optimvar('x',P,F,W,'LowerBound',0); y = optimvar('y',S,W,'Type','integer','LowerBound',0,'UpperBound',1);
Now create the constraints. The first constraint is a capacity constraint on production.
capconstr = sum(x,3) <= pcap';
The next constraint is that the demand is met at each sales outlet.
demconstr = squeeze(sum(x,2)) == d'*y;
There is a capacity constraint at each warehouse.
warecap = sum(diag(1./turn)*(d'*y),1) <= wcap';
Finally, there is a requirement that each sales outlet connects to exactly one warehouse.
salesware = sum(y,2) == ones(S,1);
Create an optimization problem.
factoryprob = optimproblem;
The objective function has three parts. The first part is the sum of the production costs.
objfun1 = sum(sum(sum(x,3).*(pcost'),2),1);
The second part is the sum of the transportation costs from factories to warehouses.
objfun2 = 0; for p = 1:P objfun2 = objfun2 + tcost(p)*sum(sum(squeeze(x(p,:,:)).*distfw)); end
The third part is the sum of the transportation costs from warehouses to sales outlets.
r = sum(distsw.*y,2); % r is a length s vector v = d*(tcost(:)); objfun3 = sum(v.*r);
The objective function to minimize is the sum of the three parts.
factoryprob.Objective = objfun1 + objfun2 + objfun3;
Include the constraints in the problem.
factoryprob.Constraints.capconstr = capconstr; factoryprob.Constraints.demconstr = demconstr; factoryprob.Constraints.warecap = warecap; factoryprob.Constraints.salesware = salesware;
Turn off iterative display so that you don't get hundreds of lines of output. Include a plot function to monitor the solution progress.
opts = optimoptions('intlinprog','Display','off','PlotFcn',@optimplotmilp);
Call the solver to find the solution.
[sol,fval,exitflag,output] = solve(factoryprob,'options',opts);
if isempty(sol) % If the problem is infeasible or you stopped early with no solution disp('The solver did not return a solution.') return % Stop the script because there is nothing to examine end
Examine the exit flag and the infeasibility of the solution.
exitflag = OptimalSolution
infeas1 = max(max(infeasibility(capconstr,sol)))
infeas1 = 9.0949e-13
infeas2 = max(max(infeasibility(demconstr,sol)))
infeas2 = 8.0718e-12
infeas3 = max(infeasibility(warecap,sol))
infeas3 = 0
infeas4 = max(infeasibility(salesware,sol))
infeas4 = 2.4425e-15
y portion of the solution to be exactly integer-valued. To understand why these variables might not be exactly integers, see Some “Integer” Solutions Are Not Integers.
sol.y = round(sol.y); % get integer solutions
How many sales outlets are associated with each warehouse? Notice that, in this case, some warehouses have 0 associated outlets, meaning the warehouses are not in use in the optimal solution.
outlets = sum(sol.y,1)
outlets = 1×20 2 0 3 2 2 2 3 2 3 2 1 0 0 3 4 3 2 3 2 1
Plot the connection between each sales outlet and its warehouse.
figure(h); hold on for ii = 1:S jj = find(sol.y(ii,:)); % Index of warehouse associated with ii xsales = xloc(F+W+ii); ysales = yloc(F+W+ii); xwarehouse = xloc(F+jj); ywarehouse = yloc(F+jj); if rand(1) < .5 % Draw y direction first half the time plot([xsales,xsales,xwarehouse],[ysales,ywarehouse,ywarehouse],'g--') else % Draw x direction first the rest of the time plot([xsales,xwarehouse,xwarehouse],[ysales,ysales,ywarehouse],'g--') end end hold off title('Mapping of sales outlets to warehouses')
The black * with no green lines represent the unused warehouses.