On Network Optimization

One way of describing a network is a set of points (supply chain term: nodes) connected by routes (supply chain term: arcs).  Network optimization is a very common subject to study in transportation, logistic, manufacturing, and operation management -- and for a good reason: an optimal network usually leads to a lean operation.  Example of the commonly used network optimization are the minimum cost network and the minimum time network.


The most 'ancient' but still surprisingly well algorithm for the network optimization is the greedy algorithm: connecting the nodes by the shortest arc.  In a simple case, greedy is usually right.  The more 'advanced,' that is, more time and computing power saving method, is the dynamic algorithm.  But since greedy is much simpler and usually right, it is the most common method used by non-programmer.  Nowadays, however, due to the increase in the computing power of our chips, it is now possible to find the most optimal solution using the more brutal force -- aka brute force method.  That is, to collect data for every possible route and then sort for the most optimum.  It is not the best way, but since today's average computer can probably calculate more than 2^30 computation per second, it is possible.


And then came the power of the spreadsheet.  In case you have not realized it yet, Microsoft Excel is extremely powerful.  Excel also came in with the "Solver" add-in, which let you use the possibly simplest method of network optimization without the needs of knowing programming algorithm.  This is called Linear Programming method.


Microsoft's Guide to Optimization: Excel Solver Tutorial
The built-in Excel Solver has some limitation.  If your variables are large and complicated enough that  you encountered error message such as "the problem is too large for solver to handle," then you might consider the upgraded, premium Solver, or even Oracle Crystal Ball.  The later is very expensive, but can easily do monte carlo simulation and more.


In this blog you will find some examples of linear programmings that are tailored for transportation and logistic purposes.


Hope this helps.