Need for a marginal methodology in assessing natural gas system methane emissions in response to incremental consumption

ABSTRACT

Accurate quantification of methane emissions from the natural gas system is important for establishing greenhouse gas inventories and understanding cause and effect for reducing emissions. Current carbon intensity methods generally assume methane emissions are proportional to gas throughput so that increases in gas consumption yield linear increases in emitted methane. However, emissions sources are diverse and many are not proportional to throughput. Insights into the causal drivers of system methane emissions, and how system-wide changes affect such drivers are required. The development of a novel cause-based methodology to assess marginal methane emissions per unit of fuel consumed is introduced.

Implications: The carbon intensities of technologies consuming natural gas are critical metrics currently used in policy decisions for reaching environmental goals. For example, the low-carbon fuel standard in California uses carbon intensity to determine incentives provided. Current methods generally assume methane emissions from the natural gas system are completely proportional to throughput. The proposed cause-based marginal emissions method will provide a better understanding of the actual drivers of emissions to support development of more effective mitigation measures. Additionally, increasing the accuracy of carbon intensity calculations supports the development of policies that can maximize the environmental benefits of alternative fuels, including reducing greenhouse gas emissions.

Introduction and background

Direct emissions of methane from the natural gas supply chain are a significant component of greenhouse gas (GHG) emission assessments for natural gas and natural-gas-based fuels including use in advanced technologies for power generation (Bauen and Hart 2000; Darrow and Tidball et al. 2015; De Bruijn 2005; Shaffer, Tarroja, and Samuelsen 2015) and transportation (Camuzeaux et al. 2015; Tong, Jaramillo et al. 2015a; Tong, Jaramillo et al. 2015b) contributing from 5 to 9% of attributed emissions in some cases (Dominguez-Faus 2016). Estimating the carbon intensities of natural gas technologies (the amount of CO₂ equivalent emissions resultant from one unit of fuel use) has a profound influence on policies and programs impacting clean energy technology deployment—most notably those seeking reductions in emissions of GHG to mitigate climate change, for example, California’s Low Carbon Fuel Standard pursuant to Assembly Bill 32 (Nunez 2006). Indeed, the importance of accurately estimating life-cycle GHG has been noted as being needed for robust policy development targeting low-carbon fuels (Mullins, Griffin, and Matthews 2010; Venkatesh et al. 2010), including natural gas (Venkatesh et al. 2011).

Quantifying the emissions of methane across the supply chain and the dependency of emissions on operational and other factors is no simple matter. The natural gas supply chain includes a large and complex infrastructure supporting a diverse range of operations including upstream (exploration and production activities associated with locating, extracting, and processing raw gas resources) (Alvarez et al. 2016; Brantley et al. 2014), midstream (gathering and processing, long-distance transmission and storage) (Marchese et al. 2015; Mitchell et al. 2015), and downstream (local distributions and end-use applications) (Alvarez et al. 2012; Lamb et al. 2015). Emissions of methane occur from a wide variety of discrete components, processes, and events, such as fugitive releases from valves, fittings, and compressors, purposeful venting from pneumatic devices or periodic venting from maintenance or upset, unburned methane in the exhaust of combustion devices, and pump-to-wheels activities for natural gas vehicles, including vehicle fueling (Allen et al. 2014a; Marchese et al. 2015; Simpson 2014; Von Fischer et al. 2017). Additionally, various techniques and methodologies have been used to characterize system methane emissions, including engineering analyses and quantification via both bottom-up and top-down approaches (Allen et al. 2014b; Karion et al. 2013; Miller et al. 2013). These methods show significant variation in total estimated emissions, resulting in significant uncertainty in available methane emissions inventories (Brandt et al. 2014).

Furthermore, past work has not viewed the measurement of emissions from the perspective of quantifying the marginal versus nonmarginal contributions, which is the central question in the proposed marginal emissions approach to carbon intensity quantification. No prior work has been found that systematically isolates the portion of modeled or measured emissions that result from incremental or decremental flow on the gas system as would appear in a marginal carbon-intensity calculation. There are some instances where the correlation of emissions to throughput has been reported (e.g., Mitchell et al. 2015). However, whether the relationship is causal and how throughput relates to marginal consumption have not been addressed. In the cited example, normalized (per unit flow) midstream emissions show a strong negative correlation to throughput, which may simply be a result of a substantial contribution from leaks whose rates do not change with throughput.

Considering what drives increases or decreases in emissions from the natural gas system is important in guiding policies and regulation of emissions and the potential deployment of natural gas-consuming technologies. For example, accurately accounting for direct methane emissions attributable to technology deployment activity requires understanding and quantifying emissions that will change in response to increased natural gas consumption, rather than total emissions from the system, that is, the marginal emissions associated with increased natural gas use.

Generally, carbon intensity (CI) calculation tools seeking to quantify the GHG impacts of different fuels in various applications such as transportation, power generation, and end use rely on averaging versus marginal analysis by adding system total emissions per unit of fuel to the carbon intensity, a pro rata allocation versus an incremental emissions approach. Examples of such tools include GREET (Argonne National Laboratory 2014) and CA-GREET (California Air Resources Board 2015), which use a simple, one-factor model for predicting attributable emissions at each stage of the production and delivery process by multiplying throughput (mass flow at a point in the system) by the estimated percentage of throughput lost through emissions. More sophisticated methods for estimating emissions have been used, including probabilistic modeling methods accounting for uncertainty associated with methane emissions in life-cycle GHG assessment of natural gas end uses (Venkatesh et al. 2011). A study by Tong, Jaramillo, and Azevedo establishes probabilistic ranges in place of point estimates for the carbon intensities of heavy duty natural gas trucks by assuming probabilistic distributions of methane emissions (Tong, Jaramillo et al. 2015a). Similarly, a Monte Carlo method in combination with a life-cycle analysis (LCA) was used by the same authors to assess the impacts of methane emissions on natural gas pathways for light-duty vehicles (Tong, Jaramillo et al. 2015b). However, Dominguez-Faus proposes that policymakers may favor the simplicity of a single estimate in making determinations regarding natural gas as a fuel and uses a scenario approach involving parameter variation to assess how methane emissions impact GHG emissions from natural gas trucks (Dominguez-Faus 2016). While these methods provide important and valuable contributions toward understanding the CI of natural gas use, they do not provide insight into what portion of methane emissions from the natural gas system is incremental due to consumption of natural gas and what portion is not. In the GREET models, for example, this leads to a systematic inaccuracy that increases the assessed climate impact of incremental use of natural gas. Thus, results from the proposed marginal methodology would help meet an important knowledge gap in regard to CI calculations.

Technically, a marginal analysis is the appropriate approach to use, but the issue of marginal versus average emissions methodology has not been addressed in the literature to any significant extent. This is most likely because very few fuels display a significant difference between marginal and average analysis. Because methane is itself a powerful greenhouse gas, system emissions play an important role in life-cycle analysis, but only marginal emissions should be properly attributed to specific end uses because the system itself and other end uses that already exist are responsible for the balance of emissions. The problem is analogous to estimating GHG emissions from the use of electricity as a vehicle fuel or other uses of electricity. A marginal approach is needed to estimate emission impacts of the electricity consumption, as measures that impact both supply and demand do not impact all generators, and emissions from generators, proportionately (Hawkes 2010; Siler-Evans, Azevedo, and Morgan 2012; Zivin, Kotchen, and Mansur 2014). Emission rates for generators vary extensively, and estimating how total emissions will change in response to changes in demand from electricity consumption (e.g., electric vehicle charging) requires an understanding of which generators respond and how they respond, that is, behavior of generators on the margin (Cullen 2010b; Kaffine and McBee et al. 2011; Razeghi, Brown, and Samuelsen 2011). Models have been developed for estimating the impacts of battery charging for electric vehicles using power system dispatch models to determine the marginal GHG emissions to more accurately estimate GHG attribution causes relative to using a system-average approach (Denholm and Short 2006; McCarthy and Yang 2010; Razeghi, Brown, and Samuelsen 2011). The analogous approach has not been used for methane emissions from the natural gas system. The authors note that the time scale of interest in a marginal analysis is case specific. Marginal dispatch on the electric grid is typically optimized over shorter periods, such as 15-minute intervals. When assessing the marginal emissions related to incremental use of natural gas, the relevant time frame ties to the time frame of deployment and use of the technology in question, and would be typically be months or years.

An ideal factor-based model of methane emissions should correlate emissions with actual drivers to provide an accurate prediction of changes due to incremental shifts in natural gas consumption. Therefore, the development of a cause-based method of allocating direct methane emissions from the natural gas system to incremental changes in natural gas consumption—the methane marginal emissions quantification model—is proposed. The method seeks to identify the marginal methane emitters—the sources of emissions that change in response to incremental natural gas demand—and to better understand and quantify methane emissions by accounting for the range of factors that drive emissions.

The prior body of work discussed in the preceding provides a starting point for developing a marginal emissions approach. A significant body of data on emissions exists for all elements of the natural gas production and delivery chain. In many instances, it may be possible to reassess existing measurements from the perspective of cause and effect. In other instances, component modeling and/or new measurements will be needed. That work, once completed, can be implemented into CI calculators such as GREET in a straightforward way by replacing the system-average emissions factor for the production and delivery chain with the appropriate marginal emissions factor.

It must be noted that the scope of this work does not include resolving discrepancies in reported overall direct methane emissions from the natural gas system, including (1) accuracies of official inventories and discrepancies with top-down measurements (Brandt et al. 2014; Zavala-Araiza et al. 2015b), (2) the presence and identification of “super emitters” (Lavoie et al. 2015; Zavala-Araiza et al. 2015a), (3) climate implications of methane emissions from unconventional gas reserves (Newell and Raimi 2014; Omara et al. 2016; Weber and Clavin 2012), and others. Rather, the objective of this work is to contribute to the knowledge base regarding causes and influences of methane emissions from the natural gas system. Benefits of a marginal method include greater clarity into the climate implications of increased or decreased natural gas consumption in energy end-use sectors and insights regarding opportunities for mitigation; for example, quantifying changes in emitted methane across the supply chain from increased natural gas use for vehicle fueling (i.e., a marginal perspective) would more accurately estimate carbon intensities for natural gas vehicles and provide a more complete understanding of low-carbon transportation pathways.

Marginal emissions quantification methodology

Understanding the incremental rate of direct methane emissions from the various stages of the natural gas supply chain is necessary to assess the climate impact of applications that utilize natural gas as a fuel or feedstock. At a micro-level, the mechanisms causing methane emissions from the natural gas system are complex. However, for purposes of assessing the incremental emissions related to alternative scenarios for natural gas end use, a macro approach can be used focusing on emissions driven by incremental natural gas consumption. For this purpose, we propose to categorize emissions sources into three primary types: time-based, random-event-based, and throughput-based, as shown in Table 1. As an example, the reason pneumatic controllers are considered random here is that the control variable triggering actuation of the device is not correlated to throughput. Thus, in this instance the term “random” is being used to imply a lack of correlation. It should be noted that some sources (likely even the majority of sources) may exhibit a combination of these three macro causal factors. In addition, care must be taken to identify sources that depend on throughput indirectly, such as control activation events that appear random but where the event frequency increases with throughput (instantaneous or cumulative). Additionally, any hypothesized dependence in the table may or may not be confirmed through further analysis. Furthermore, consideration of changes to the natural gas system itself must be made to ensure appropriate marginal assessment. For example, increased use of natural gas leading to system expansion would result in additional equipment on the system that would increase emissions, and would be considered marginal in a scenario where system expansion was relevant, for example, a long-term scenario for natural gas use.

Table 1. Drivers of methane emissions from sources in the natural gas system. scf: standard cubic feet, CH₄: methane.

Category	Description	Potential equations	NG system examples
Time	Continuous emissions via leakage or venting. Events occurring at a known frequency	Emissions (scf CH4) = Time interval (time) * Magnitude (scf CH4/time)	Continuous pneumatic controllers; Fugitive leaks
Random event	Emissions related to discrete events or occurrences with no systematic or causal dependence on system flow	Emissions (scf CH4) = Event probability (event/time) * Magnitude (scf CH4/event)	System damage; Intermittent pneumatic controllers (IPC)
Throughput	Event or occurrence-based emissions whose frequency and magnitude depends upon system flow rate or cumulative flow over a time period	Emissions (scf CH4) = Volume throughput (scf gas) * Magnitude (scf CH4/scf gas)	Compressor operations emissions; Manual liquid unloadings

As noted, the physics of actual emissions are dependent upon a complex set of underlying variables that are many more than the three modeling factors proposed. For example, time-dependent emissions from a pinhole leak will have a functional dependence on pressure and temperature. Dependencies are also a function of the infrastructure materials and equipment actually used. The primary driver of a pinhole leak is pressure, but emissions are better described for current purposes in terms of emissions per unit of time (time as the causal factor for modeling) because changes in pressure and temperature will generally fluctuate around a system average at a given location and the instantaneous leak rate is not relevant to the analysis. Simply quantifying average leak rate per unit of time over the relevant time period provides an accurate picture of marginal emissions (in this case, the emissions do not depend on gas flow rate on the system so the marginal emissions from this source would be zero to first approximation). Thus, a degree of averaging will be used within the marginal approach while maintaining the distinction between marginal and nonmarginal elements. Equation 1 represents the approach in simplified mathematical form. For end-use carbon intensity calculations, only the throughput term would contribute to marginal emissions:

(1)

where i = [1,N] represents individual emissions sources,

N represents total number of emissions sources,

a, b, c represent emissions distribution coefficients of time, event, and throughput, respectively,

E_T, E_E, E_TP represent emissions rates driven by time, event, and throughput during a time-frame interval (scf of methane/time interval), respectively,

E_∑ represents overall emissions for individual emissions sources (scf of methane/time interval), and

∑E represents total emissions for all emissions sources (scf of methane/time interval).

At the source level, equations can be used to approximate emissions in relation to causal drivers. Equation 2 presents a volumetric flow rate expression for an orifice showing that flow is governed by system pressure, fluid properties, and leak geometry, rather than throughput. Although it is not expected that this equation accurately predicts emissions from fugitive leaks, such equations can provide an accurate determination of the parametric dependence of emissions on system variables:

(2)

where Q is volume flow rate through the orifice (leak proxy),

C is orifice plate coefficient of discharge,

β is expansion factor,

d is internal orifice diameter,

ΔP is pressure difference across orifice, and

ρ is fluid density.

Additionally, the model should include predictable changes in system design and management practices in the natural gas supply chain depending upon the time horizon of the analysis. For example, proposed EPA regulations on system methane emissions targeting a 40% reduction by 2030 would impact the baseline case in a 2030 analysis. Therefore, a time horizon must be specified in any specific marginal analysis.

Development of a complete assessment of marginal emissions across the natural gas delivery chain requires assessment of a large number of potential emissions categories, for example, the U.S. EPA-characterized emission sources (including the 51 discrete types shown in Table 2) along with potential emissions of methane (U.S. EPA 2016). Table 2 also provides a preliminary assessment by the authors identifying the sources for which a portion, or all, of attributed emissions are independent of system throughput (i.e., nonmarginal emissions). Based upon this preliminary assessment, which indicates that well over 90% of emissions sources may show a difference between marginal and average emissions, the implications of a using a marginal versus average approach could be significant. If, for example, sources with potential independence of throughput are weighted 25% independent (as a conservative estimate for illustration purposes), then by using a marginal approach, attributed methane emissions could be 22% lower than current methods may estimate (i.e., assuming all sources in Table 2 are proportional to throughput). Higher proportions of non-throughput-driven emissions will have a proportionately higher impact. This illustration and many other scenarios and hypotheses could be validated or modified in the proposed development of the marginal emissions methodology.

Table 2. Potential methane emission sources from the natural gas system. From U.S. EPA (U.S. EPA 2016).

Main sector	Emission source				Potential methane emissions, kt	Percentage (%)	Potential for nonmarginal emissions*
Production	Gas wells (nonassociated gas wells and gas wells with hydraulic fracturing)				51.4	1.09	✓
	Well pad equipment (heaters, separators, dehydrators, meters/piping, compressors)				291.2	6.18	✓
	Gathering and boosting (stations, pipeline leaks)				2134.2	45.31	✓
	Drilling, well completion, and well workers				28.11	0.60	✓
	Produced water from coal methane (Powder River and Black Warrior)				60.9	1.29	✓
	Pneumatic device vents				1065.2	22.62	✓
	Dehydrator vents				21.7	0.46	✓
	Pumps (chemical injection, Kimray)				196.5	4.17	✓
	Condensate tank vents				33.7	0.72	✓
	Compressor exhaust vented (gas engines)				248.6	5.28
	Liquid unloading (vent using plungers and vent without using plungers)				218.4	4.64
	Blowdowns (vessel and compressor blowdown, compressor starts, etc.)				207.1	4.40	✓
	Upsets (pressure relief valves and mishaps)				2.1	0.04	✓
	Offshore (shallow and deep water gas platforms; Gulf of Mexico and Pacific)				150.6	3.20	✓
Total production					4709.7	100.00
Processing	Plant fugitives, reciprocating compressors, centrifugal compressors, flares, dehydrators				156.3	35.13	✓
	Compressor Exhaust (gas engines, gas turbines, AGR vents, pneumatic devices)				253.1	56.89	✓
	Blowdowns/venting				35.5	7.98	✓
Total processing					444.9	100.00
Transmission & Storage	Stations venting	Transmission and storage			183.3	10.86	✓
	Stations fugitives	Transmission and storage			142.2	8.43	✓
	Dehydrator vents	Transmission and storage			6.5	0.39	✓
	Pneumatic venting	Transmission and storage			52.1	3.09	✓
	Wells	Storage			92.6	5.49	✓
	M&R (Trans. Co. Interconnect, farm taps, direct sales)				92.5	5.48	✓
	Compressors	Compressors fugitive	Reciprocating	Transmission	339.4	20.11	✓
				Storage	104.2	6.17	✓
			Centrifugal	Transmission	115.7	6.86	✓
		Compressors exhaust	Engines, turbines, generators		288.3	17.08
	Pipeline venting				183.6	10.88	✓
	Pipeline leaks/fugitive				3.3	0.20	✓
	LNG operation (storage, compressor exhaust, import terminal)				83.92	4.97	✓
Total transmission and storage					1687.6	100.00
Local Distribution	Underground pipelines	Mains	Cast iron		32.1	6.69	✓
			Unprotected steels		48.1	10.02	✓
			Protected steels		46.9	9.77	✓
			Plastic		20.4	4.25	✓
		Services	Unprotected steels		47.8	9.96	✓
			Protected steels		18.6	3.87	✓
			Plastic		12.5	2.60	✓
			Copper		4.4	0.92	✓
	Metering and regulating (M&R) facilities	M&R	M&R (>300 psi)		8.6	1.79	✓
			M&R (100–300 psi)		14.6	3.04	✓
			M&R (<100 psi)		5.7	1.19	✓
		Regulators	Regulators (>300 psi)		3.8	0.79	✓
			Regulators (100–300 psi)		1.9	0.40	✓
			Regulators (40–100 psi)		6.5	1.35	✓
			Regulators (<40 psi)		0.4	0.08	✓
		Regulator-vaults	Regulator-vaults (>300 psi)		0.2	0.04	✓
			Regulator-vaults (100–300 psi)		0.6	0.12	✓
			Regulator-vaults (40–100 psi)		0.4	0.08	✓
	Upsets (mishaps/dig-ins)				67.1	13.98	✓
	Routine maintenance (pressure relief valve releases and pipeline blowdown)				5.5	1.15	✓
	Customer meters (residential and commercial/industrial)				134	27.91	✓
Total local distribution					480.1	100.00

Notes. *Preliminary assessment by authors based on survey of the literature cited. Identifies categories of emissions in which some portion may arise from sources independent of throughput. It should be noted that the majority likely exhibit some degree of both independence and dependence, i.e., partial dependence.

Downstream example—drivers of emissions from meter set assemblies (MSA)

In order to further illustrate the implications of the marginal methodology, this section presents preliminary results from experimental work seeking to quantify the relationship between emissions and drivers for residential meter set assemblies (MSA) as an example. The results for MSAs are not generalizable to other parts of the system but they illustrate the understanding required for all components of the natural gas system to elucidate the relationship between emissions and causal drivers to accurately estimate methane marginal emissions from changes in natural gas consumption. More accurate assessment and quantification of causal factors will be developed in the next phase of this work for important emission sources throughout the system.

Customer meters account for 25% of total methane emissions from the distribution sector, emitting an estimated 112 million metric tons of methane to the atmosphere annually in the United States from residential, commercial, and industrial meters. Among these sources, residential MSA are the dominant contributor to total customer meter gas emissions (95%) (Lamb et al. 2015). Hence, this discussion of emissions characterization focuses upon residential meters. Residential MSA measure and regulate the natural gas supply to residential households, and each consists of a riser connecting the gas supply pipeline to a regulator (Figure 1). The regulator provides a set gas pressure to the meter independent of fluctuations in flow rate or supply pressure. The regulator is connected to the inlet of a customer meter where gas consumption is measured and recorded. The outlet of the meter is joined to the residential gas line.

Figure 1. A representative meter set assembly (MSA) with major components highlighted.

The primary sources of emissions from MSA are fugitive emissions (leaks) from threaded connections and regulator venting events. Regulator venting events only occur when system pressure exceeds a set point and are rare and rather random, so threaded connection leaks are the dominant source of leaks (Innovative Environmental Solutions 2009). Leaks from threaded connections depend predominantly on system pressure and are weakly related to system flow (throughput), if at all. Therefore, a marginal emissions approach would likely model MSA emissions as independent of throughput and would attribute no incremental emissions increase or decrease in response to changes in gas flowing through the meter. In fact, the emissions may be pressure dependent and increased use would most likely reduce the pressure and, therefore, the leak rate. The implication is that a marginal analysis approach would attribute minimal incremental carbon intensity to end uses of natural gas related to the metering element of the delivery chain. To the extent that reducing this source of emissions is a policy goal, the key implication is that curtailing end use is not an effective tool to accomplish this goal.

To confirm the characterization of MSA emissions as independent of throughput, a number of tests were conducted with actual residential meters sets at the Southern California Gas Company Engineering Analysis Center. The following figures show the relationship between inlet pressure and volumetric flow rate versus methane emissions in the simulated external leak MSA. Figure 2 shows a positive linear correlation with pressure for all three systems. Both curve fits for MSA#2 and MSA#3 have a reasonable R² value, while that of MSA#1 is quite high, with all MSA showing higher leak rate with higher pressure. Figure 3 shows that all three MSA tested exhibit a decreasing trend of the emissions with increasing throughput. MSA#3 has a high R² value, while those of MSA#2 and especially MSA#1 are low. Nonetheless, the results of Figure 3 demonstrate that emitted gases tend to decrease with volume throughput.

Figure 2. Methane emissions from MSA with simulated external leak systems in response to inlet pressure variation.

Figure 3. Methane emissions from MSA with simulated external leak systems in response to flow rate variation.

Conclusion

The prior body of work on methane emissions from the natural gas system is extensive and includes emission-factor measurements for all elements of the production and delivery system, including unintended emissions such as simple leaks and equipment malfunctions and emissions related to system operation such as pneumatic controls. This provides a robust component taxonomy and inventory, and an upper bound on marginal emissions. The gap to be addressed in future work is that prior measurements do not characterize emissions in a way that separately quantifies marginal and non-marginal elements. The authors propose an engineering-based characterization of the direct emissions of methane from the natural gas system to determine relationship to factors upon which increases and decreases in emissions depend. The analysis will also include consideration of potential changes over time due to management practices throughout the natural gas supply chain. The results will be used to develop a more accurate method for estimating and attributing methane emissions from the natural gas system to assess impacts of incremental changes in natural gas consumption. The method will be based on a marginal approach accounting for the physical and operational parameters of the gas system and verified via field analysis and computational simulation where appropriate. The methodology will provide technical insights into the causes of methane emissions, better inform carbon intensity assessments, and a better tool for assessing the potential climate impacts of alternative natural gas technology deployment scenarios.

Highlights

• Methane emission from the natural gas system influence carbon intensity calculations.

• Various causal factors govern emissions sources across the natural gas life cycle.

• Current methods generally predict emissions as proportional to system throughput.

• A marginal method is proposed to better estimate incremental emissions per unit fuel consumed.

• Results support alignment of policy and regulation based on causes of emissions.

Additional information

Notes on contributors

Michael Mac Kinnon

Michael Mac Kinnon is a senior scientist in the Advanced Power and Energy Program at the University of California, Irvine.

Zahra Heydarzadeh

Zahra Heydarzadeh is a mechanical engineering Ph.D. student in the Advanced Power and Energy Program at the University of California, Irvine.

Quy Doan

Quy Doan is a mechanical engineering M.S. student at the University of Southern California and an intern at Southern California Gas Company.

Cuong Ngo

Cuong Ngo is a chemical engineering student at California Polytechnical University, Pomona, and a student intern at Southern California Gas Company.

Jeff Reed

Jeff Reed is Chief Scientist for Renewable Fuels and Energy Storage in the Advanced Power and Energy Program at the University of California, Irvine.

Jacob Brouwer

Jacob Brouwer is a professor of mechanical and aerospace engineering and Associate Director of the Advanced Power and Energy Program at the University of California, Irvine.