NREL is a national laboratory of the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy operated by the Alliance for Sustainable Energy, LLC Optimal Regional Layout of Least-Cost Hydrogen Infrastructure Brian W. Bush Olga Sozinova Marc W. Melaina National Renewable Energy Laboratory 4 May 2010 NHA Hydrogen Conference & Expo

Outline Overview of the Scenario Evaluation and Regionalization Analysis (SERA) model Disaggregation of the 2008 National Academy of Sciences (NAS) demand scenarios for hydrogen fuel-cell vehicle (FCV) roll-out Application to California infrastructure optimization Application to competition between on-site steam methane reforming (SMA) and combined heat, hydrogen and power (CHHP) from stationary fuel cell systems National Renewable Energy Laboratory Innovation for Our Energy Future 2

Scenario Evaluation and Regionalization Analysis (SERA) Goals Determine optimal regional infrastructure development patterns for hydrogen, given resource availability and technology cost. Geospatially and temporally resolve the expansion of production, transmission, and distribution infrastructure components. Key analysis questions Which pathways will provide least-cost hydrogen for a specified demand? What network economies can be achieved by linking production facilities to multiple demand centers? How will particular technologies compete with one another? (e.g., central vs. onsite) National Renewable Energy Laboratory Innovation for Our Energy Future 3 SERA is a tool for studying regional build-outs of renewable energy infrastructures over time by optimizing on the delivered cost of hydrogen

National Renewable Energy Laboratory Innovation for Our Energy Future 4 Key Functionality of SERA Regional or national scope –Optimizes infrastructure at and between urban areas Detailed spatial resolution –Any number of feedstock pricing regions Annual temporal resolution –Infrastructure built when needed, according to time-varying demands in urban areas Integrated hydrogen networks –Production, transmission, and delivery components City-specific cash flows and delivered costs Interoperable with other hydrogen analysis tools and databases –Leverages existing capabilities –Avoids duplication of existing functionality

National Renewable Energy Laboratory Innovation for Our Energy Future 5 How SERA Works SERA searches for optimal infrastructure architectures to meet time- varying demands in multiple urban areas within a specified region.

Objective function –Total discounted cash flow for the whole H 2 infrastructure  This could also be done at five-year increments or in other time frames. –Other objective functions could be used. Constraints –Demands must be fully met: i.e., no shortfalls allowed. –Capacity constraints on technologies must be satisfied. –Lifetimes of infrastructure components must be respected. –Transmission networks must be “tree-like”: i.e., no loops or cycles are allowed currently. Exogenous inputs –Annual H 2 demands at cities –Regional feedstock prices –Infrastructure characteristics  Production technologies  Transmission technologies  Distribution costs Optimization Technique Used in SERA 6 National Renewable Energy Laboratory Innovation for Our Energy Future Pipeline Liquid Truck Route Urban Demand Central Production

Analysis Process National Renewable Energy Laboratory Innovation for Our Energy Future 7

Interoperability with HyDRA National Renewable Energy Laboratory Innovation for Our Energy Future 8 8 SERA provides... Hydrogen infrastructure, by year Hydrogen delivered cost, by year HyDRA provides... Hydrogen demand scenarios Energy/feedstock price forecasts Existing infrastructure Background information

Spatiotemporally detailed nationwide FCV rollout scenario 1.Start with the city-specific schedule of FCV introductions in the NAS study. 2.Divide FCVs among NAS-defined cities in proportion to their populations. 3.For the “nationwide” category, assign FCV introduction dates to census-defined urban according to log-linear regression result. 4.Apportion the “nationwide” FCVs among the cities present in each year, in proportion to population. 5.Use the FCV ageing, VMT, and other NAS FCV stock assumptions to compute FCV stock, VMT, and hydrogen demand on a city-by-city basis. 6.Delay FCV introductions to cities where the demand is less than 1500 kg/day. The resulting scenario matches the NAS scenario in terms of FCV introduction, FCV stock, VMT, and hydrogen demand on a year-by-year basis in addition to matching the city-specific schedule in the study. National Renewable Energy Laboratory Innovation for Our Energy Future 9 New FCVs, by city and year Hydrogen Demand, by city and year

California Hydrogen Demand Scenarios Three NAS scenarios were disaggregated and specialized to California: –1 “Hydrogen Success” –1a “Hydrogen Accelerated” –1b “Hydrogen Partial Success” The national scenario was partitioned to the state level based on VMT, according to FWHA statistics Within the state, hydrogen demand was scaled according to urban population. 210 urban areas 210 urban demands

Infrastructure Differences for California Demand Scenarios 1b (“Partial”) 1a (“Accelerated”) 2045

Patterns in Optimal Infrastructure Choice National Renewable Energy Laboratory Innovation for Our Energy Future 12 Note: Multiple plants at the same location are shown as single points in the above plot.

National Renewable Energy Laboratory Innovation for Our Energy Future 13 Conclusions from California Scenario Studies Observations and semi-insights: 1.Pipeline infrastructure and (to a lesser extent) other transmission infrastructure is non-optimally costly for the levels of demand considered here—it is only when feedstock costs to onsite production technologies are raised substantially (or the deployment of those technologies forbidden) that transmission infrastructure comes into play significantly. 2.Some of the potential technologies (e.g., central grid electrolysis) rarely come into use because they are generally more costly than others (e.g. central biomass gasification) in the cost inputs. 3.Hydrogen cost may vary widely (an order of magnitude) with locality and time. 4.The construction of production plants that are not fully utilized in the early years of their lifetime substantially increases delivered hydrogen cost in those years. Detailed analysis of the interplay between technology costs will be required to verify and defend insights gained in optimization studies. Average Delivered Cost [$/kg] < > 7

Station Placement Algorithm We implemented a spatiotemporal station placement technique that produces semi-realistic spatial, temporal, capacity distributions for hydrogen refueling stations: 1.Time-dependent hydrogen demand for the urban area is used to estimate the number of stations that would be built in each year. 2.The stations are sized stochastically according to an empirically determined capacity distribution. 3.The stations are located stochastically within the urban area of interest, according to a uniform distribution. This station network expansion algorithm is based upon empirical data for existing gasoline stations, simulations of how those networks have evolved over time, and the resulting station size distributions. (cf. Melaina and Bremson, 2008) National Renewable Energy Laboratory Innovation for Our Energy Future 14 Refueling Stations in Los Angeles

Low-Cost Mixes of CHHP & On-Site SMR Goal: To gain insights regarding the cost- competition between two prominent intra- urban hydrogen-production technologies relevant for the first decades of fuel-cell vehicle (FCV) use. Hypothesis: In cases where there is a low demand for hydrogen in the early years of FCV scenarios, CHHP may have cost advantages over on-site SMR production for certain refueling stations. National Renewable Energy Laboratory Innovation for Our Energy Future 15 Potential CHHP Locations in Los Angeles Delivered Cost for Three Scenarios Technology Mix for Three Scenarios

Conclusions and Next Steps Capabilities of the SERA model –Integrated, cross-cutting model –Scenario-oriented analysis compatible with H2A models –Searches for optimal combinations of hydrogen production and transmission infrastructure to meet time-varying demand in urban areas over a region. Ongoing work –Application of SERA to more elaborate scenarios –Elaboration of models (especially cost models) for specific studies –Further incorporation of CHHP and biogas technologies Future work –Examining price points between competing technologies –Coupling SERA to a discrete-choice model for hydrogen demand National Renewable Energy Laboratory Innovation for Our Energy Future 16