HyNOON

HyNOON is a computer model that searches for the economically optimal long-term strategy for production, transmission, and distribution of hydrogen across large areas of the United States. We created HyNOON as a research tool for the National Renewable Energy Laboratory, to help them analyze how a long-term transition to a hydrogen economy might occur, and what it might cost.

The HyNOON user enters:

• the costs and efficiencies of hydrogen production technologies like steam reformation and electrolysis
• the cost of the feedstock on which each production technology depends
• the cost of transporting hydrogen via pipelines and trucks
• the hydrogen demand of each city under consideration

All of these inputs can vary over the period of time that the analysis covers, which is typically fifty years. HyNOON constructs a long-term strategy to supply the hydrogen demand over that period in a way that minimizes the overall life-cycle cost. It may connect several cities together in a distribution network (via pipelines or hydrogen trucks) that evolves over time.

The image below shows the results of an analysis of several cities in the northeastern US. HyNOON has decided to serve most of the cities via two pipeline networks, each of which radiates outward from a central production facility in progressively smaller-diameter pipelines. A few of the smaller cities are not served by the pipeline network, but instead receive their own dedicated hydrogen production facilities.

(Click for full-size image)

HyNOON creates a "blueprint" for the construction of hydrogen production facilities, pipelines, and trucking links that supply the area"s hydrogen demand at the lowest total life cycle cost. The four maps and the table below show the output of a sample HyNOON analysis of twelve cities in Southern California. This sample analysis covers the 50-year period from the 2010 to 2060. The hydrogen load in each city varies over the 50-year period according to its population and an assumption about how quickly and how completely that city"s transportation sector will convert to hydrogen. The size of the circles reflect the relative sizes of the hydrogen loads in each city, and the width of the solid lines reflect the relative diameter of the pipelines. Dotted lines indicate liquid truck links.

(Click for full-size image)

In 2010 the hydrogen load is zero in each city. In 2014 HyNOON builds pipelines between the cities with the largest future hydrogen loads. These pipelines are greatly oversized for 2014, but optimally sized considering the future hydrogen demands. A small steam methane reformation (SMR) plant supplies the cities in the pipeline network, while isolated electrolysis plants supply the smaller cities. In 2028, near the end of the useful life of the electrolysis plants built in 2014, HyNOON replaces the electrolysis plants with liquid hydrogen truck links from various points in the pipeline network. A larger SMR plant at Los Angeles now supplies hydrogen to all twelve cities. No changes occur after 2028 except for the construction in 2032 of an even larger SMR plant at Los Angeles. The total life cycle cost of this blueprint is $8.76 billion.

The image below shows the results of another HyNOON analysis, this time of 54 cities in Florida. The map shows the optimal final build-out strategy. In this case HyNOON has chosen to construct three separate liquid truck networks, one emanating from Miami, one from Orlando, and one from Pensacola.

(Click for full-size image)

Optimization Algorithm

The goal of HyNOON’s optimization algorithm is to find the blueprint that minimizes the total life cycle cost. HyNOON uses an optimization technique called simulated annealing to accomplish this goal. The simulated annealing process starts with a feasible solution, makes a small random change to the solution, then decides whether or not to accept that change. If it accepts the change, the modified solution serves as the new starting point and the algorithm continues by considering a further change to that solution. If it rejects the change, the algorithm reverses that change and considers some other change. The algorithm continues until it can no longer improve the solution.

The key to the simulated annealing algorithm is the criterion it uses to accept or reject changes. It accepts any "downhill move", which is a change that improves the solution, for example by lowering its cost. But it does not reject outright any "uphill move". Rather, it treats uphill moves probabilistically, accepting some and rejecting others by random probability. As the search proceeds, this probability of accepting uphill moves steadily diminishes. This probabilistic acceptance of uphill moves allows the simulated annealing algorithm to avoid getting stuck in a "local optimum", which is a solution that is the best in its neighbourhood, but not the best overall. The ability to avoid local optima is crucial in solving certain optimization problems, particularly the class of combinatorial optimization problems, of which the hydrogen production and transmission problem is an example.

In HyNOON, the "solution" that the simulated annealing algorithm works to improve is the blueprint. The starting point is a simple blueprint wherein every city gets its own SMR plant. The changes that the algorithm can consider include changing a production facility from one technology to another (e.g SMR to electrolysis, or vice versa), changing the retirement date of a production facility, replacing a production facility with a transmission link from another city, changing a transmission link from one type to another (e.g. pipeline to gas truck), or moving one end of a transmission link from one city to another. Most applications of HyNOON require tens of thousands of such modifications to converge on an optimal or near-optimal blueprint.