In order to simulate subsurface hydrogen storage processes under various cushion gas conditions, scientists at Los Alamos National Laboratory are creating strong machine learning models, which are an application of artificial intelligence. In the future low-carbon economy, this will be essential.
Deep saline aquifers, or depleted hydrocarbon reserves, are among the most practicable places to store hydrogen, according to team leader scientist Mohamed Mehana. “But in order to accomplish this, cushion gasses must first be injected into the reservoir, replacing the preexisting fluids and offering the pressure support necessary for hydrogen recovery.”
The effects of cushion gasses—most commonly methane, carbon dioxide, or nitrogen—on these subterranean hydrogen storage systems have been investigated by scientists. Nevertheless, the impact of cushion gasses on the efficiency of underground hydrogen storage activities has never been completely recognized.
The Los Alamos team effectively examined a variety of cushion gas situations in a recent publication that was published in the International Journal of Hydrogen Energy. This investigation yielded important new information about the impacts of different cushion gasses on underground hydrogen storage performance.
Part of the country’s decarbonization endeavor includes scaling the hydrogen economy. Hydrogen gas will also need to be created and stored locally, just like gasoline, in order to power clean-energy semi-trucks, directly produce electricity, and keep solar power plants operational throughout the winter.
To achieve this level of development, the country will have to tap into a vast array of subterranean reserves. One particular set of geological and operational conditions was the subject of all previous research. However, the Los Alamos team’s model took into account a variety of geological circumstances, the presence of water, and the operational impact of various cushion gasses in order to replicate real-world events.
Shaowen Mao, a postdoctoral research associate with the Los Alamos team, explained that the peculiar features of hydrogen and its intricate operating requirements make underground hydrogen storage a complex process. During the withdrawal stages, we must minimize the risks associated with water production while optimizing hydrogen recoverability and purity. It is crucial to comprehend these and other elements in order to make subsurface hydrogen storage financially feasible.
In order to replicate the variety of real-world circumstances, the Los Alamos researchers employed a deep neural network machine learning model, which examined combinations of geological and operational data. The team included important discoveries in the study, a few of which were as follows:
- Because of its enhanced cycle performance over time, underground hydrogen storage in porous rocks holds great technical promise.
- the benefits and drawbacks of storing hydrogen underground in depleted hydrocarbon reserves and saline aquifers, and
- the effect of different cushion gas scenarios on well injectivity in porous rocks, water production risk, hydrogen recoverability, and purity.
Years of research
This work expands upon years of research on hydrogen storage at Los Alamos, one of the first universities to investigate this subject from several perspectives.
In order to better understand how cushion gas affects the performance of underground hydrogen storage, Los Alamos scientists have studied the flow and transport behavior of hydrogen in the underground environment.
This research has also included an ongoing investigation of possible hydrogen storage sites in the Intermountain West region, which mixes machine learning-powered simulations with the physics of underground geological formations.
Additionally, a different area of research has focused on creating instruments that can evaluate the performance, risk, and dependability of hydrogen storage under various scenarios. The result of this later effort is OPERATE-H2, which is the first software accessible to the industry that integrates cutting-edge machine learning to optimize hydrogen storage.
