Design of Experiments for Sensitivity Analysis of a Hydrogen Supply Chain Design Model

Abstract

Hydrogen is one of the most promising energy carriers in the quest for a more sustainable energy mix. In this paper, a model of the hydrogen supply chain (HSC) based on energy sources, production, storage, transportation, and market has been developed through a MILP formulation (Mixed Integer Linear Programming). Previous studies have shown that the start-up of the HSC deployment may be strongly penalized from an economic point of view. The objective of this work is to perform a sensitivity analysis to identify the major parameters (factors) and their interaction affecting an economic criterion, i.e., the total daily cost (TDC) (response), encompassing capital and operational expenditures. An adapted methodology for this SA is the design of experiments through the Factorial Design and Response Surface methods. Six key parameters are chosen (demand, capital change factor (CCF), storage and production capital costs (SCC, PCC), learning rate (LR), and unit production cost (UPC)). The demand is the factor that is by far the most significant parameter that strongly conditions the TDC optimization criterion, the second most significant parameter being the capital change factor. To a lesser extent, the other influencing factors are PCC and LR. The main interactions are found between demand, CCF, UPC, and SCC. The discussion has also shown that the calculation of UPC has to be improved taking into account the contribution of the fixed, electricity, and feedstock costs instead of being considered as a fixed parameter only depending on the size of the production unit. As any change that could occur relative to demand or CCF could strongly affect the response variable, more effort is also needed to find the more consistent way to model demand uncertainty in HSC design, especially since a long horizon time is considered for hydrogen deployment.

Supply chain network design

“…Under uncertainty, an MILP model was developed to find the optimal supply chain design and operation. Ochoa Robles et al (2018) used an MILP formulation to establish the hydrogen supply chain (HSC) network. In the HSC logistic model, a sensitivity analysis showed the most sensitive components and their interactions.…”

Calibration of a multi-mobile coil magnetic manipulation system utilizing a control-oriented magnetic model

“…However, as the number of parameters and size of the distribution space increases, so does the number of runs. Nevertheless, the same information can be obtained through fractional factorial design approaches such as Box-Behnken and Taguchi [29]. For this research, factorial design was used, it has the following advantages:…”

Optimization of a hydrogen supply chain network design under demand uncertainty by multi-objective genetic algorithms

“…In addition to hydrogen demand, one of the most significant parameters is feedstock cost ( Ochoa Robles et al, 2017), ( Ochoa Robles et al, 2017. In the original model ( De-León Almaraz et al, 2014 ), the unit production cost (UPC) of electricity remains fixed for all of the periods regardless of the technology, which was a severe simplification.…”

“…The second scientific barrier is to model the uncertainty related to different variables and parameters of the HSC, e.g., fuel price ( Sabio et al, 2010 ) or hydrogen demand, which have been identified as among the most significant parameters in the HSC ( Ochoa Robles et al, 2015, Ochoa Robles et al, 2017. Several methods have generally been mentioned to model demand uncertainty ( Chen and Lee, 2004, Jung et al, 2004, You and Grossmann, 2008 ): (i) the scenario-based approach; (ii) the distributionbased approach;(iii) the fuzzy-based approach; (iv) the deterministic planning and scheduling models, with the incorporation of safety stock levels; and (v) the spatially aggregated demand model.…”