Analytical network process
An ANP–SWOT approach for interdependency analysis and prioritizing the Iran's steel
scrap industry strategies
abstract
Supplying raw materials plays a central role in the Iran's crude steel plants. Iran's scrap recycling industry, as a major supplier of steel scraps, requires identifying and performing the efficient strategies to utilize the substantial amount of steel scrap. In this research, by designing a model and implementing the efficient strategic factors (Strengths, Weaknesses, Opportunities and Threats), via the SWOT analysis, the appropriated strategies, SO, ST, WO and WT are determined. This model consists of four factors, 14 sub-factors and eight strategies. Since the SWOT method is unable to rank the factors and strategies, the MADM technique is applied to prioritize them. The dependency between the factors affects the priority of strategies. Therefore, AHP and ANP methods are applied discretely to examine whether the dependency among the factors changes the priorities of strategies. Comparison of the outcomes from the AHP and ANP methods presents that the dependency among the factors in ANP method causes major changes in ranking the alternatives particularly for the principal alternatives.
Analytical network process
ANP is a developed method of AHP that presented by Saaty. Unlike AHP method, ANP has network and cluster structure. The hierarchy formation in AHP is a linear top to down structure while network is a non-linear structures that extent in all directions (Sevkli et al., 2012). This enables ANP to model the complex problems in the real world. This method is able to consider the mutual and interdependent relationships among factors and subfactors and rank alternatives, criteria and sub-criteria with considering mentioned relationships (Saaty, 1994b, 1996, 2004). Generally, the ANP technique comprises two methods for ranking the alternatives. The first technique is creating super matrix. Wij, ith row and jth column of the matrix, is the principal eigenvector, that represents the influence of the elements compared in the ith cluster on jth cluster. When Wij¼0, it means the ith cluster has no influence on jth one. The weighted super matrix is derived by conveying all column sums to unity value exactly. To create super matrix, the weight of options is normalized and their priorities are determined. The weighted super matrix is raised to powers until the global matrix of priorities is obtained. Finally the alternatives, criteria and sub-criteria are sorted (Gwo-Hshiung et al., 2011).
