Supporting Demand Supply Network Optimization with Petri Nets - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 376

Algorithm 1. Main Analysis Routine
RG_main (DSNnet_system - cf. Definition 3) returns all DSN setups with cost {
RG[] = array of new Reachability Graphs;
total_RG = new Reachability Graph;
for each initial customer node do
RG[i] = RG_1_customer(customer_node[i]);
end
for i = 0. number of customer nodes -1 do
total_RG = aggregate_RG( total_RG, RG[i]);
end
for each path in total_RG do
investment_cost = compute_investments(path);
add investment_cost to the path's cost;
end
return total_RG;
}
complexity. For, assume that we have C customers and P is the maximum of the number of DSN setup
options for a customer. Then the size of the state space (and computation time) grows exponentially as
O(P C ) . Parallelizing this part of the algorithm is included in the topics for future research.
r eachability Analysis r outine for a Single Customer node
The algorithm that computes the reachability graph for one customer is presented in Algorithm 2. It
follows the traditional reachability analysis algorithm with the addition of AND and OR nodes.
This part of the algorithm computes investment costs that depend on production volume in a stepwise
manner. For example, if a manufacturing line capacity for a particular phone is 7500 phones per day,
this algorithm computes the number of needed manufacturing lines based on total volume throughput,
and calculates the needed investments.
Algorithm Validation
The algorithm was validated in two steps. First, a series of tests with different product structures and
customers was carried out to ensure that the paths were computed correctly. This step concentrated on
the correct functioning of the AND and OR nodes which were our additions to the traditional reachabil-
ity analysis algorithm. In the second phase, the cost figures obtained for a single setup were compared
with those obtained from the previous spreadsheet-based tool. The second validation phase uncovered
some errors in cost computation, which were corrected and revalidated.