2.1 H₂ Supply Chain for Captive Uses

H₂ production was 90 billion N m³ in Europe and 7 billion N m³ in France in 2007 (PHyRENEES, 2009). While most of it is produced onsite for captive uses, a significant portion of H₂ produced today is a by-product in the chemical industry (e.g., chloralkali electrolysis) and, since there is no specific purpose, this is sometimes wasted, flared, or treated by gas industries. According to the report presented by the Hydrogen Analysis Resource Centre (2015), four main hydrogen producers share the European market: Air Liquide, Air Products, Linde, and Praxair and in 2015, the total daily production capacity installed in Europe was around 3795 t of H₂ per day.

In current schemes, the logic of industrial distribution is generally as follows: hydrogen is produced in centralized plants or produced onsite for captive uses. It is obtained mainly from steam methane reforming (SMR) and coal gasification (CG) and also as a by-product from the chloralkali electrolysis plants. It is used either on-site or transported through short distances by pipelines. The option to be liquefied or compressed and then transported via tube trailers or tanker trucks also exists but is less frequent. Ammonia production plants, oil processing units, methanol, and metal industries are hydrogen customers with high demand volumes (see Fig. 4.1).

2.2 H₂ Supply Chain for Mobility

The HSC for H₂ as a fuel is defined as a system of activities from suppliers to customers. These activities involve an energy source and various stages such as production, storage, transportation, and dispensation of hydrogen to refueling stations. Unlike most other fuel infrastructures, hydrogen can be produced either centrally (similar to existing gasoline supply chains) or distributed (as small-scale units that can produce H₂ close to the use point in small quantities) modes at forecourt refueling stations (with reduction on the distribution cost). The stakeholders involve as economic, social, or institutional agents, business enterprises, and cultural groups of a specific country or region related to the hydrogen economy as displayed in Fig. 4.2. The projects dedicated to the deployment of hydrogen infrastructures are taken place worldwide. In Europe, the Hydrogen Mobility Europe 2 (H2ME) projects funded by the European Fuel Cells and Hydrogen Joint Undertaking will form the largest EU-funded project for hydrogen mobility and fuel cell electric vehicle (FCEV) deployment. The H2ME project is integrated by the Mobility Europe project (now call H2ME 1) with plans for 300 FCEVs and 29 hydrogen refueling stations (HRSs) to May 2020. H2ME 2 is a 6-year project which will run to the end of June 2022 and brings together 37 partners from 8 European countries. It will include the deployment and operation of 1230 FCEVs, the addition of 20 new HRSs, and will test the ability of electrolyzers to simultaneously feed hydrogen stations and help balance the electrical grid (“H2ME 2 launched in Europe to grow hydrogen fuelling infrastructure network and vehicle fleet,” 2016).

The selection of the “best” hydrogen pathway involves the comparison of various technological options in terms of multiple performance criteria, with the ultimate goal being to define a strategy whereby the infrastructure investment can be planned with confidence (Hugo et al., 2005). Several options are presented in Fig. 4.3. In that context, some questions arise:

HSC modeling requires the identification of the main characteristics of its activities. A general representation of the well-known options is displayed in Fig. 4.3. It must be emphasized that the degree of maturity is not the same among all the technologies.

3.1 Natural Gas

Natural gas is a fuel consisting mostly of methane produced through the decay of organic materials. Further, most natural gas is obtained from wells, although it can be produced from organic waste through the use of digesters (Bartels et al., 2010). Natural gas is typically found near areas with large oil or coal reserves. Worldwide, there are 187.8 trillion cube meters of proved reserves, which is expected to last about 56 years at the current consumption rate (BP Stat, 2013). The existing natural gas infrastructures can be used to power the hydrogen production units.

3.2 Coal

Coal reserves are abundant and relatively well distributed globally; it will continue to play a leading role until 2050 and beyond. It should nevertheless be noted that, as pointed out by the European Directive of 2009, a key level in the short term, to reduce GHG emissions, is to improve energy efficiency in end-use in different sectors: construction, transport, and industry. In this regard, using coal as an energy source is not without problem in that it releases carbon dioxide and other pollutants when combusted. CG and carbon sequestration (see also Section 4.5) may be able to minimize pollutants and greenhouse gases while changes in mining practices can reduce the environmental concerns.

3.3 Biomass

Biomass consists of biological material that can be used for industrial purposes, such as plants, wood, or waste. Biomass has historically been used as a heating fuel through the use of combustion furnaces. There are four main categories of biomass resources that have the potential to be used as a feedstock: (1) energy crops which as the name implies are specifically grown for energy content (e.g., corn, soybeans, poplar trees, and algae); (2) agricultural waste, which includes crop and animal waste; (3) forestry waste from harvesting trees and clearing land; and (4) industrial and municipal waste (Bartels et al., 2010).

Many countries around the world have allocated the research and development toward the hydrogen production from biomass but some limitations exist. One issue with biomass is whether there is enough land to produce the amount of fuel and food needed in the world (Bartels et al., 2010). Another limitation is that the processes of hydrogen production from biomass are still in the development stage and require a strong effort in terms of R&D and demonstration activities (Balat and Balat, 2009). Also, the characteristics of biomass are very important since they can vary greatly from location to location, seasonally and yearly, so that the hydrogen production via biomass route may not be competitive with the hydrogen production with fossil fuels (Dagdougui, 2011a). However, the waste-to-energy application has received much attention due to its potential to become a major hydrogen source.

Biomass has been by far the largest source of RES consumed in the EU. Consumption of biomass was up to 13% in 2010, accounting for 69% of the total consumption of renewable sources. Most of this was used for power generation and in households (Eurostat–European Commission, 2012). Biomass plays a significant role in the current French energy mix for combustion and heat production (Direction générale de la compétitivité, de l’industrie et des services, 2011).

3.4 Electricity as an Energy Carrier for Hydrogen Production

Depending of the energy mix of each country, the use of electricity to produce hydrogen (via electrolysis) could determine the environmental impact of the H₂ fuel. For example, the French energy mix is dominated by nuclear power (78%) while the USA and China main source to produce electricity is coal, with 47% and 79%, respectively. Fig. 4.4 represented the global warming potential (GWP) per kW h of electricity for each particular production mix based on the Ecoinvent unit processes database with the method IMPACT 2002+ V2.10. France has the lowest GWP impact. The existing electricity infrastructures can then be used to power the hydrogen production units.

H₂ can be considered as a “clean” energy carrier (if produced solely from renewable energies) and is emission free at the point of final use (avoiding the transport-induced CO₂ emissions; Ball and Wietschel, 2008). Through the production of H₂, the losses resulted from the intermittence of solar, wind, and hydro sources can be reduced. Also, if electricity offers increases and electrolyzers are installed next to the renewable energy site, no need to invest in new electric grid would be another advantage. Development of renewable sources for electricity production in the next years is uncertain and will vary depending on the regional condition and political strategies.

4.1 Centralization Degree

To classify the degree of centralization of the HSC, two categories will be used either centralized or decentralized (on-site) units. A centralized production option would be analogous to current gasoline supply chains, where the economies of scale are capitalized upon within an industrial context and large quantities are produced at a central site and then distributed (Hugo et al., 2005; Murthy Konda et al., 2011). Centralized plants not only promise higher hydrogen production efficiency but also some difficulties are associated in high-volume hydrogen to be transported.

Decentralized production consists in small regional plants or even local filling stations that could generate hydrogen. While hydrogen generation efficiency for decentralized is lower than those for centralized plants, losses in hydrogen transport can make such a scheme more efficient (Kim et al., 2008; Haeseldonckx and D’haeseleer, 2011). There is a tendency in the literature to argue that decentralized production plants could overcome many of the infrastructural barriers facing a transition to hydrogen (Ball and Wietschel, 2008). Most studies consider the decentralized route as the key to by-passing the infrastructural problem (Haeseldonckx and D’haeseleer, 2011). A decentralized approach often results in higher costs as efficiencies are generally lower and because on-site production facilities are often dimensioned to cover peak demand (especially when no storage is foreseen or possible). However, a further increase of demand will require larger pipelines, which thus implies new investment costs (Haeseldonckx and D’haeseleer, 2011).

4.2 Steam Methane Reforming

Most of hydrogen (97%) stems from steam reforming of natural gas also known as SMR (Koroneos et al., 2004). SMR is used in the chemical and petrochemical industries; it is currently the cheapest production method and has the lowest CO₂ emissions of all fossil production routes (Ball and Wietschel, 2008).

The main steps during the production of hydrogen from natural gas are (Hajjaji, 2011) (a) production of the synthesis gas, (b) conversion of carbon monoxide to hydrogen (water shift gas), and (c) purification. The first stage (see Fig. 4.5) is a catalyzed endothermic reaction between methane (natural gas) with water vapor at high temperature (steam reforming) to produce synthetic gas, which mainly consists of carbon monoxide and hydrogen along with some water, carbon dioxide, and methane (Bartels et al., 2010). During steam reforming, hydrocarbons are catalytically split in the presence of steam at temperatures of 800–900°C (Landucci et al., 2007). Then, carbon monoxide is converted to carbon dioxide following the exothermic shift reaction. In the purification stage, pressure swing adsorption is the prevailing process (Koroneos et al., 2004) in which the reactive gas mixture, containing methane and hot steam, is fed to the tube side of a catalytic furnace reactor. Since the reaction is endothermic, the combustion of methane with air in the furnace side of the reactor provides the required reaction heat. The exhausted beds are regenerated via hydrogen washing, so even if a high purity product is obtained, about 25% of hydrogen is lost. The hydrogen-enriched purge gas is recycled to the furnace side of the reactor and burned together with the combustion methane (Landucci et al., 2007).

4.3 Biomass and Coal Gasification

Biomass gasification (BG) for hydrogen production, still at an early stage today, is expected to become the cheapest renewable hydrogen supply option in the coming decades although biomass has restricted potential and competes with other biofuels as well as heat and power generation (Ball and Wietschel, 2008). The BG process can be carried out (National Renewable Energy Laboratory, 2011) according to three main types of processes: fixed bed (downdraft and updraft), fluidized bed (bubbling fluidized bed, circulating fluidized bed), and entrained flow gasifiers.

Each type can use one or a combination of gasification agents, including steam, air, and oxygen, to promote conversion. Gasification is an endothermic process and requires a heat source to promote reaction (Fig. 4.6). Indirect gasification typically uses steam whereas direct gasification uses high-pressure air or oxygen as agents. Indirect gasifier temperatures are in the range of 750–900 °C and a mix of syngas, char, and tars is produced. One disadvantage of this approach is that a char combustor, a steam reformer, and an extra compressor are needed to boost the syngas pressure before the acid gas is cleaned up. During direct gasification, biomass under pressure in the presence of oxygen and steam produce medium thermal energy syngas and heat via an exothermic process. The heat is captured in the gasifier and combined with oxygen to maintain temperatures of 850–1100 °C. This process also needs an expensive air separation unit (ASU) for oxygen supply.

4.4 Water Electrolysis

Water electrolyzers can be divided into two categories, alkaline and proton exchange membrane (PEM) electrolyzers. According to Ball and Wietschel (2008), electrolysis processes are more expensive than SMR and only applied if high-purity hydrogen is required since the work of Bartels et al. (2010) reports that H₂ production from electrolysis may become economically competitive because fossil fuel feedstock costs also increase, and technological advancements decrease the cost of alternative energy types (Murthy Konda et al., 2011). Significant cost reductions are also expected for many materials, and catalysts and cell components used in PEM electrolyzers could benefit from large-scale production of PEM fuel cell of similar concept and design (Grigoriev et al., 2009).

The governing reaction of electrolysis is shown in Fig. 4.7. PEM technology provides an example of “zero-gap” configuration, in which electrodes are in direct contact with the surface of the ion exchange membrane. This cell concept offers some significant advantages compared to more conventional “gap cells”: (i) no circulating electrolyte is required, pure water being the only reactant supplied to the anodes; (ii) gaseous products H₂ and O₂ are produced at the backside of the interpolar field, offering the possibility of reducing ohmic drops and increasing current densities. As a result, low-energy consumption (4.0–4.2 kW/N m³ H₂) and high hydrogen purity (>99.99%) are obtained (Grigoriev et al., 2009; Dagdougui, 2011b).

PEM electrolysis is a viable alternative for generating hydrogen from RES. As long as electricity comes from a clean source, electrolysis is a clean process. But producing hydrogen via electrolysis and then using hydrogen to produce electricity again is associated with considerable losses (Hake et al., 2006). In addition, a PEM electrolyzer can deliver hydrogen at high pressure, which will in turn be attractive for the application where hydrogen needs to be transported or stored (injected to a hydrogen tank or a pipeline network or be transferred to the liquefaction process). A power consumption of 52.5 kW h/kg is used by the PEM (which is equivalent to about 75% in efficiency) (Dagdougui, 2011b).

4.5 Carbon Capture and Storage

H₂ supply chains offer the possibility of capturing most of the CO₂ emissions (~90%) and sequestering them (Murthy Konda et al., 2011). Carbon capture and storage (CCS) is an energy intensive and costly process involving several steps: CO₂ capture, pressurization, transportation, and final disposal into geological formations or in aquifers (Haeseldonckx and D’haeseleer, 2011; Direction générale de la compétitivité, de l’industrie et des services, 2011). There are three main technology options for CO₂ capture and storage: precombustion capture, postcombustion capture, and oxyfuel combustion.

5.1 Liquefaction

Liquefaction can greatly increase the volumetric density of hydrogen, as compared to compressed gas, but the liquefaction process itself is very capital and energy intensive (Johnson et al., 2008). Hydrogen liquefaction and use of liquid hydrogen are usually practiced only when high storage density is required, e.g., in aerospace applications (Dagdougui, 2011b). But H₂ can also be liquefied for being transported in larger volumes by cryogenic trucks (800 times denser than the gaseous state). However, a certain level of cryogenic technology is needed to be implemented, either to liquefy or to keep in the liquid state (Patay, 2008). The extensive isolation technology and fueling process, as well as the loss of hydrogen by evaporation effects during storage periods, are further disadvantages of liquid hydrogen (LH₂) storage system (Hake et al., 2006). A temperature of 20.3 K or −253°C is required. It is not obtained only by supplying cold gas but by the combined effect of cooling and clean the adiabatic expansion of the gas after it has been previously compressed (AFH2, 2011). The electricity use in a liquefier (kW h/kg H₂) is estimated as a function of size with a minimum electricity use of around 9 kW h/kg H₂ for the most efficient, large liquefiers (Johnson et al., 2008) compared to 4.1 kW h/kg H₂ needed for compression (Grol et al., 2005). There are currently around 30 liquefaction units worldwide. Their daily productions range from a few tonnes to 54 t for the largest (Sacramento, USA, Union Carbide). World production of LH₂ would reach 355 t. The majority (84%) of this production lies in the USA: its main use is the petrochemical industry (33.5%) followed by aerospace applications (18.6%) (e.g., the first factory was built for the Apollo program in 1960) (AFH2, 2011).

The liquefaction system can turn the hydrogen gas to medium pressure liquid hydrogen. A condenser can be located on the same site a production unit (referring as “liquefier attached”). If not attached, a condenser can be located anywhere in the country. Each of liquefier can supply several filling stations (Patay, 2008).

5.2 Liquid H₂ Storage

Like all cryogenic liquids, liquid hydrogen is stored in cryostats containers double insulation whose purpose is to limit the inevitable heat outside contributions. The former is a thermal insulation barrier against the flow of heat by direct conduction, provided by the vacuum maintained in the double wall of the cryostat. The latter, against the flow of heat by radiation, is provided by a multireflecting sheet metal disposed in the space between the walls. A cryostat is not sealed so that a continuous evolution of hydrogen gas (precisely vapor hydrogen) is observed. This constant evaporation, which corresponds to a weight loss of 0.5 to 1% per day depending on the technology used, is one of the major disadvantages of storing hydrogen in its liquid form (AFH2, 2011). Liquid hydrogen plays an important role during the transition phase (until 2030) and in connecting outlying areas, such as along motorways or in rural areas (Ball and Wietschel, 2008). The works of Almansoori and Shah (2006, 2009) concluded that from an economic viewpoint, designing a liquid hydrogen based supply network is extensively cheaper than designing a gaseous-based network.

5.3 Compression

The production of hydrogen and conversion to gaseous form is less expensive than producing it and converting to liquid form. However, this difference in production costs does not come close to bridge the gap of higher transportation cost of gaseous hydrogen compared to liquid hydrogen (Ingason et al., 2008).

The conditioning system can turn the medium pressure hydrogen gas to a high pressure hydrogen gas or compressed hydrogen (CH₂). Currently, 200 or 350 bar pressure bottles are distributed in the industry and developments now concern tanks can withstand pressures of 700 bar (CEA, 2013). A conditioning center can be located on the same site a production unit (referred as “attached conditioning center”). If not attached, a conditioning center can be located anywhere in the country. Each conditioning center can supply several groups of filling stations (Patay, 2008).

5.4 Gaseous H₂ Storage

The required space of gaseous storage could be an important issue. Given the fact that large storage facilities can only maintain pressures up to 12–16 bar, this means that up to 1 million m³ of space is required for storage capacity for 10 million N m³ H₂ (Haeseldonckx and D’haeseleer, 2011). In addition, storing such large quantities of hydrogen also brings along serious safety issues. Large hydrogen storage facilities are unlikely to be located near densely populated areas.

The risk of leakage of hydrogen must also be considered in light of flammable and explosive nature of the gas under certain conditions. However, due to the small size of the molecule, hydrogen is able to penetrate many materials, including certain metals. It weakens, moreover, some making brittle. The study of high-pressure storage is therefore essential to test the resistance of materials to hydrogen pressure. These materials must be resistant but relatively light (mobility requires). The metal tanks currently used still prove expensive and heavy in relation to the amount of gas they can take. Nonmetallic tanks but most of polymeric materials (nonload-bearing nonmetal liner axial and hoop wrapped with resin-impregnated continuous filament¹) are being developed to address these constraints (CEA, 2013).

5.5 Metal Hydrides

An alternative to the traditional storage methods (liquid and gas) is proposed through the use of advanced solid materials. Certain materials absorb hydrogen under moderate pressure at low temperatures, forming reversible hydrogen compounds called hydrides. This type of hydrogen storage is often called “solid” hydrogen storage since hydrogen becomes part of the solid material through some physicochemical bonding (Dagdougui, 2011b). This kind of storage works like a sponge; it absorbs H2 and stocks it. It is therefore recovered at the outlet of H₂ gas, high purity, low pressure. For use in cars, it will be compressed to 700 bar (350 bar or depending on the model selected) and put in the filling station. This technology is well adapted for on-site applications. More information about this kind of storage can be found in “McPhy – Solid hydrogen storage”.

6.1 Pipeline

Pipelines have been used to transport hydrogen for more than 50 years (Ball and Wietschel, 2008). The longest hydrogen pipeline in the world to supply chemical and petrochemical industries (about 1050 km in France, Germany, and the Benelux countries) is operated by Air Liquide (CEA, 2013). The United States has more than 720 km of hydrogen pipelines concentrated along the Gulf Coast and Great Lakes, the estimation of the capital cost of hydrogen transmission pipelines range from 200,000 to 1000,000 US$/km (Dagdougui, 2011b).

From a production unit, the gas is transported through a transmission line medium pressure (100 bar) (Patay, 2008; Haeseldonckx and D’haeseleer, 2011). This means that less space is required to store the same amount of hydrogen. In addition, as hydrogen is stored throughout the entire pipeline network, there are no large concentrations of hydrogen at the same location, improving the overall safety. The exact amount of hydrogen that can be stored depends on the maximum and minimum pressure, the hydrogen flow, and the length and diameter of the pipeline (Haeseldonckx and D’haeseleer, 2011).

Pipelines for medium and large fueling stations may become relevant once a significant market penetration of hydrogen vehicles has been achieved, but these are mostly used for local distribution in highly populated areas and for large-scale interregional energy transport according to the HyWays roadmap (European Commission, 2008). In Ball and Wietschel (2008), a pipeline network could be constructed after 2030 when the potential demand can justify the high investment.

6.2 Tube Trailer

From a conditioning center, compressed hydrogen can be transported at around 200–250 bar by tube trailers. With the appearance of decentralized, regional production, tube trailers use is a solution for the transition phase toward the use of pipelines (European Commission, 2008). Commercial tube trailers are well established. Generally, transporting CH₂ over the road in high-pressure tube trailers is expensive and used primarily for short distances; it becomes cost prohibitive when transporting farther than about 321 km from the point of production in the study of Dagdougui (2011b). Compressed gas truck delivery is not considered as a long-term delivery solution because their low hydrogen capacity would necessitate too many deliveries (Yang and Ogden, 2013).

6.3 Tanker Truck

From the liquefaction unit, LH₂ can be transported by tanker trucks (cryogenic liquid hydrogen tankers). This transportation mode is the most economical pathway for medium market penetration (Dagdougui, 2011b). They could transport relatively large amounts of hydrogen and reach markets located throughout large geographic areas. Forty ton trucks can carry 3500 kg of LH₂ so that the transport of liquid hydrogen is limited by volume, not by weight (Bossel, 2006).

8.1 Mathematical Modeling

In the mathematical formulation, mixed integer linear programing (MILP) approaches have been widely used. The MILP method consists of maximizing or minimizing an objective function as a function of parameters, variables, and several constraints on these variables (Haeseldonckx and D’haeseleer, 2011). The use of integer variables in general and in particular binary dramatically broadens the capabilities of linear programing modeling, enabling the disjunction of constraints, the logical implication, and general restrictions to the model incorporating certain nonlinear behaviors of reality. Many practical optimization problems lead to consider an extremely large number of feasible solutions so that the problem can be viewed as a combinatorial one. Several solvers can be used to solve MILP problems. In particular, GAMS environment (Brooke et al., 1988), offers a wide variety of optimization procedures that includes well-known algorithms for the solution of MILP (Geletu, 2008): Branch & Bound, Benders Decomposition, Cutting Plane (Gomory) algorithm, and Branch & Cut. Usually, these algorithms are used in combination with the simplex algorithm and/or the interior-point method. For instance, some of the solvers that can solve MILP problems are BARON, BDMLP, LINDO GLOBAL, MOSEK, OSL, XPRESS, and CPLEX.

Almansoori and Shah (2006) have clearly introduced a general model that determines the optimal design of a network (production, transportation, and storage) for vehicle use where the network is demand driven. The model was applied to a Great Britain case study. The same authors extended the model in 2009 (Almansoori and Shah, 2009), to consider the availability of energy sources and their logistics, as well as the variation of hydrogen demand over a long-term planning horizon leading to phased infrastructure development as well as the possibility of selecting different scales of production and storage technologies. Other works (Almansoori and Shah, 2012a) take into account demand uncertainty arising from long-term variation in hydrogen demand using a scenario-based approach: the model adds another echelon including refueling stations and local distribution of hydrogen minimizing the total daily cost.

Hugo et al. (2005) developed an optimization-based formulation that investigates different hydrogen pathways in Germany. The model identifies the optimal infrastructure in terms of both investment and environmental criteria for many alternatives of H₂ configurations. This model has been extended and considered as a basis for other works such as Li et al. (2008) for the case study in China. At the same time in Iran, a model for investigation of optimal hydrogen pathway and evaluation of environmental impacts of hydrogen supply system was examined by Qadrdan et al. (2008). Another study also considered hydrogen from water, using electricity from hydro and geothermal power in Iceland for exportation (Ingason et al., 2008).

Additional items have been integrated in Kim et al. (2008) models as deterministic versus stochastic approach to consider demand uncertainty in the new model. The model they proposed determines a configuration that is the best for a given set of demand scenarios with known probabilities. The stochastic programing technique used is based on a two-stage stochastic linear programing approach with fixed recourse, also known as scenario analysis. A strategic design of hydrogen infrastructure was later developed to consider cost and safety using multiobjective optimization where the relative risk of hydrogen activities is determined by risk ratings calculated based on a risk index method (Kim and Moon, 2008).

Patay (2008) has developed a strategic and tactical model for the French territory and Benelux region in cooperation with Air Liquide Company. In this work, the demand and energy costs are exogenous parameters given the uncertainty on these parameters for a long-term horizon. This work studies the transportation and distribution nodes taking into account an economic criterion. The multiperiod model is to scale the infrastructure in each time period of a time horizon (horizon year 2050). The originality of Patay (2008) problem results from the time horizon considered and the dynamic dimension, as well as the size of the supply chain considered.

Guillén Gosálbez et al. (2010) proposed a bi-criterion formulation that considers simultaneously the total cost and life cycle impact of the hydrogen infrastructure and to develop an efficient solution method that overcomes the numerical difficulties associated with the resulting large scale MILP. Sabio et al. (2010) also developed an approach, which allows controlling the variation of the economic performance of the hydrogen network in the space of uncertain parameters examined the case study of Spain. Later, Sabio et al. (2011) take into account eight environmental indicators in a two-step method based on a combination of MILP multiobjective optimization with a postoptimal analysis by principal component analysis to detect and omit redundant environmental indicators.

Murthy Konda et al. (2011) considered the technological diversity of the H₂ supply pathways together with the spatial-temporal characteristics to optimize a large-scale HSC. They calculate the transportation costs based in Almansoori and Shah (2006, 2009) approaches. The original models are modified (e.g., inclusion of existing plants, capacity expansion, and pipeline features) and analysis is extended to incorporate the computation of delivered cost of H₂, well-to-tank emission, and energy efficiency analyzes. In Haeseldonckx and D’haeseleer (2011), the objective is not only to find the optimal set of activated hydrogen production plants but also to implement a hydrogen infrastructure optimization algorithm that has to decide which hydrogen-production plants will be invested in and which plants will not.

The thesis work of Dagdougui (2011b) describes the risk hazards (delimitation and explanation of potential risks in some parts of the hydrogen infrastructure: pipeline and storage tank) to demonstrate the consequence of hydrogen accident in case of a future infrastructure operation. The risk is integrated into the HSC to minimize the global risk to population and environment. The model is applied to regional case studies to the region of Liguria (North of Italy) and Morocco. A GIS-based methodology was coupled based on the clean feedstock for hydrogen production.

A sensitivity analysis was implemented to investigate the uncertainties of some parameters on the design of the future hydrogen infrastructure in Liu et al. (2012) on the analysis of hydrogen demand from hydrogen FCEVs in Ontario, Canada. Though Yang and Ogden (2013) describes the development and use of a hydrogen infrastructure optimization model using the H2TIMES modeling framework to analyze hydrogen development in California to 2050, H2TIMES is a quasi-spatial model that develops the infrastructure to supply hydrogen fuel in order to meet demand in eight separate California regions in a least cost manner subject to various resource, technology, and policy constraints.

It can be highlighted that several mono-objective optimization approaches have been developed or extended as in Qadrdan et al. (2008), Almansoori and Shah (2006, 2009, 2012a), Ball et al. (2006), Hugo et al. (2005), Kamarudin et al. (2009), Kim et al. (2008), Patay (2008), Liu et al. (2012), and Yang and Ogden (2013). In these studies, the cost is the objective to be minimized. Multiobjective optimization studies are relatively scarce and criteria to be analyzed are based on economic and environmental performances; some examples are presented in Guillén Gosálbez et al. (2010), Hugo et al. (2005), Li et al. (2008), and Sabio et al. (2011): minimizing the expected total discounted cost and the associated financial risk (Sabio et al., 2010) and minimizing the total cost of the network and the total relative risk of the network (Kim and Moon, 2008; Dagdougui, 2011b).

The pathway toward a hydrogen economy as an energy vector must encompass a broad range of items concerning the three pillars of sustainability based on economic, environmental, and social impacts. In this sense, multiobjective optimization has been tackled in the work of De-León Almaraz et al. (2013) demonstrating the importance of the use of multicriteria decision-making for the HSC network design by minimizing the cost, the environmental impact, and the safety risk simultaneously. The sensitivity of geographical scale was also analyzed in De-León Almaraz et al. (2014) to solve a real problem of the HSC in the Midi-Pyrénées region in France for a time horizon 2010–2050 (Fig. 4.10). In order to analyze the economies of scale and the real geographical implications, a comparison between a regional and a national case for France has been discussed in De-León Almaraz et al. (2015).

Recent models are focusing on the integration of carbon capture and storage technologies and in the utilization of pipelines resulting in centralized HSCs using fossil fuels instead of renewable energies if CCS technology is available. One example is the SHIPMod presented in Agnolucci et al. (2013) for an optimization-based framework for the design of HSC and CCS pipeline networks over a long planning horizon. A conclusion of this work is that varying the level and the spatial pattern of demand has significant impacts on both the optimal supply system and on the overall costs of delivered hydrogen. Later, in the work of Moreno-Benito et al. (2016), the SHIPMod has included additional options such as hydrogen imports in the UK for a multiperiod problem until 2070 by optimizing the present value. Almansoori and Betancourt-Torcat (2016) present an extension of previous models improving the definition of the objective function, in this case the total annualized cost to plan the HSC under emission constraints. A case study for Germany in the year 2030 is treated considering also CCS decisions. Besides, some works are evaluating only technologies that use RESs in their models, e.g., renewable electricity (Kim and Kim, 2016) and biomass (Woo et al., 2016).

Recently, a detailed sensitivity analysis using design of experiments methodology is presented in Ochoa Robles et al. (2015). The main conclusions are related to the importance to find better ways to represent the demand uncertainty since the demand is an exogenous parameter in the mathematical models. Demand uncertainty has been a common topic in recent years in this subject, e.g., Dayhim et al. (2014) minimize the total daily social cost (with categories such as economy, emissions, energy consumption, and risk) of the HSC with uncertain demand applying scenario-based approach to solve the proposed multiperiod two-stage stochastic model. Nunes et al. (2015) proposed to represent the logistic infrastructure, considering uncertainty in the demand forecast evaluating different investment alternatives using the sample average approximation method to reduce the number of scenarios, only the cost is minimized based on previous studies of Almansoori and Shah (2012b) and Dayhim et al. (2014).

Finally, the use of a metaheuristic tool to design and solve the HSC has been presented by Ochoa Robles et al. (2016). This work uses genetic algorithms instead of exact methods for multiobjective optimization minimizing the cost and the CO₂ emission decreasing the computation effort to generate the Pareto fronts.

8.2 Geographical Simulation

Literature review reveals that few researchers have used the spatial dimension to construct the infrastructure for hydrogen. Some examples of geographic approaches include the study of Ball et al. (2006) who developed the Model for Optimization of Regional Hydrogen Supply approach of the energy system with the integration of geographic aspects in the analysis by the GIS-based method for Germany. This model identifies the cost-optimal way for constructing and implementing an initial hydrogen supply infrastructure as well as possible trade-offs between hydrogen production and electricity generation within a country-specific context (high degree of regionalization) (Ball et al., 2006). Johnson et al. (2008) used also GIS for modeling regional hydrogen infrastructure deployment using detailed spatial data and applied the methodology to a case study of a potential coal-based hydrogen transportation system in Ohio with CCS. The objective in this work was to optimize hydrogen infrastructure design for the entire state. The MARKAL model has been applied to the UK- and GIS-based spatial model has been developed to represent the layout of HSC (Yang and Ogden, 2013). A study presented in Baufumé et al. (2013) assumes a high penetration of hydrogen-fueled vehicles for Germany in 2050 and investigates the structure of a potential pipeline network for hydrogen transmission and distribution under different scenarios for H₂ production and demand. More recently, a detailed economics model of hydrogen infrastructure in California has been developed and applied to assess several potential fuel cell vehicle deployment rate and hydrogen station technology scenarios with GIS tools (Brown et al., 2013).