Quantification of Hydrogen Emission Rates Using Downwind Plume Characterization Techniques

Abstract

Fugitive and operational hydrogen (H₂) emissions have been shown to offset the intended climate benefits of H₂ as a decarbonization tool because H₂ indirectly causes warming through atmospheric chemistry. However, little is known about the magnitude of value-chain H₂ emissions due to a lack of empirical measurements, the unavailability of precise, fast sensor technology, and the lack of established methods for emission quantification. In this study, a novel prototype H₂ sensor was deployed in a state-of-the-art mobile laboratory to demonstrate that H₂ emissions can be accurately quantified using downwind measurement techniques and analytics commonly used for other gases. A series of controlled H₂ release experiments were conducted with the corelease of multiple tracer gases. Tracer flux ratio and physics-based Bayesian plume model inversion methodologies were used to quantify the H₂ emission rates. We show that the H₂ sensor detects parts-per-billion or higher concentration enhancements downwind of the H₂ emission source within seconds, and the similar plume structures of H₂ and tracer gases confirm their codispersion in the atmosphere. This enables the quantification of H₂ emission rates at an accuracy comparable to that of other well-studied gases.

Keywords: H2 emissions; controlled-release experiments; mobile measurements; plume model inversion; tracer flux ratio.

Figure 1

Sample plume characterization: the left plot represents the 3D positioning of abovebackground H₂ enhancements for emission rate = 50 [SLPM] (8 plume passes), and the right plot shows the geographical demonstration of METEC in Fort Collins, CO.

Figure 2

Scatter plot of concentration ratios and the known released mass flow rate ratios during blended aboveground releases with C₂H₆ as the tracer gas. Individual concentration ratios (pink circles) are averaged for each release condition (black horizontal lines) with estimated uncertainties (gray error bars). A 1:1 line is shown (gray dashes) for comparison with a linear regression fit (solid blue line; fixed through the origin).

Figure 3

Histogram of Φ during blended aboveground releases with C₂H₆ as the tracer gas. A Gaussian fit (solid blue line) quantifies the mode of the distribution.

Figure 4

Evolution of posterior PDF (P(Q|C_H2^y)) derived from the recursive Bayesian inference plotted against a given range of candidate emission rates (Q) for each additional sensor pass for the H₂ test case of (a) aboveground 50 SLPM and 8 passes, (b) aboveground 20 SLPM and 10 passes, (c) aboveground 10 SLPM and 10 passes, (d) aboveground 5 SLPM and 7 passes, (e) underground 20 SLPM and 20 passes, and (f) underground 10 SLPM and 5 passes. The vertical red dashed line depicts Q^act, and the thick blue posterior line in each plot indicates the corresponding PDF of the last pass (after all data have been ingested by recursive inference). Six different test cases are depicted here (4 aboveground and 2 underground). Each panel contains the posterior PDFs derived from the Bayesian inference for the number of given transects (e.g., panel (a) has 8 PDFs corresponding to 8 passes). Following the PDFs on one sample panel, the PDFs become less spread and closer to the actual emission rate that is indicated by a vertical dashed line in each panel.

Figure 5

Standard deviation of σ_Q̂^ML in percentage for all aboveground experiments as additional passes are included.