Leak detection and repair (LDAR) at liquids-rich shale gas sites: the case of Seven

Generations Energy

Daniel Roda-Stuart, Arvind RavikumarGroup Leader: Adam R. Brandt

Department of Energy Resources Engineering, Stanford University

NGI Annual Affiliates Meeting and Research Review, 2018October 18th 2018

Project: LCA of shale gas to China

• Partnership with Seven Generations Energy

• Analysis of proposed LNG to China pathway– Kakwa River project (NW Alberta)– Gas from liquids-rich Montney play– Gas to LNG unit at Sarita Bay on

Vancouver Island – LNG to Shenzen, Guangdong

China

2

What are environmental impacts of LNG from Alberta to China

compared with standard (coal-based) energy supply?

Direct access to leakage surveys

• Unique aspect of LCA study• Seven Generations provided complete access to

measured leakage data for LCA– Access to results from dozens of LDAR surveys– Pads, Super-Pads, Processing Plants, Compressors– >1000 leaks recorded over multiple years– Leaks and vents quantified and traced over time– Stanford researchers performed field work with leak

surveyors• First LCA performed with direct access leakage

data from the particular project being studied

3

Quantified leakage observations in database

4

52

Table 3-1. Detailed emissions distribution information for all leak and vent types found during initial surveys of all 7G facilities. All emission measurement units in kg/hr.

Figure 3-1. Emissions distribution for certain leak and vent types found during initial surveys at 7G super pads.

We also can plot a simple total distribution of all leaks and vents found and quantified at

7G sites, with site revisits excluded. Figure 3-2 shows the heavy-tailed distribution of leaks

and vents. All data included in this figure come directly from measurements, without any

supplemental data to account for detected leaks that were not quantified. As such, the

expected largest emitting sources, tanks, are not included. Regardless, we find that the top

5% of emitters account for 51.6% of quantified emissions.

Component Type CountTotal Emissions

[kg/hr] Mean 5% 25%Median

(50%) 75% 95%

Actuator 26 7.45 0.29 0.01 0.01 0.05 0.20 1.54Analyzer 14 1.34 0.10 0.01 0.04 0.09 0.13 0.22Catadyne Heater 248 19.50 0.08 0.01 0.05 0.05 0.14 0.19Chemical Injection Pump 54 8.39 0.16 0.01 0.01 0.02 0.06 0.37Compressor, Cylinder Head 12 1.30 0.11 0.01 0.01 0.06 0.20 0.29Controller 143 22.21 0.16 0.01 0.03 0.07 0.18 0.38Engine (Including Fuel/Start Gas) 28 0.88 0.03 0.01 0.01 0.02 0.05 0.08Flange Connection 15 0.45 0.03 0.01 0.01 0.01 0.03 0.09Meter / Instrumentation 20 2.11 0.11 0.01 0.02 0.03 0.17 0.37Misc 19 0.72 0.04 0.01 0.01 0.02 0.06 0.10Open Ended Line / Vent 115 51.96 0.45 0.01 0.03 0.15 0.38 2.08Other Tank 1 0.02 0.02 0.02 0.02 0.02 0.02 0.02Pneumatic Pump 4 0.16 0.04 0.01 0.01 0.04 0.07 0.07Pressure Safety Valve (PSV) 3 0.83 0.28 0.19 0.21 0.23 0.32 0.39Pump Packing 4 0.32 0.08 0.01 0.02 0.04 0.10 0.20Regulator 34 1.66 0.05 0.01 0.01 0.03 0.05 0.16Threaded Connection 313 13.67 0.04 0.01 0.01 0.01 0.04 0.15Valve 80 5.20 0.06 0.01 0.01 0.02 0.05 0.23

0.0

0.1

0.2

0.3

0.4

Analyzer Compressor, Cylinder Head Meter / Instrumentation Threaded Connection

Mea

sure

d Em

issi

ons

Rat

e [k

g/hr

]Total: 3316 kg CH4/d = $168,000 per y @ 3$/mcfMean: 2.9 kg CH4/d = $150 per y @ 3 $/mcfMax: 95.3 kg CH4/d = $4800 per y @ 3$/mcf

Example size distribution of leaks

5

52

Table 3-1. Detailed emissions distribution information for all leak and vent types found during initial surveys of all 7G facilities. All emission measurement units in kg/hr.

Figure 3-1. Emissions distribution for certain leak and vent types found during initial surveys at 7G super pads.

We also can plot a simple total distribution of all leaks and vents found and quantified at

7G sites, with site revisits excluded. Figure 3-2 shows the heavy-tailed distribution of leaks

and vents. All data included in this figure come directly from measurements, without any

supplemental data to account for detected leaks that were not quantified. As such, the

expected largest emitting sources, tanks, are not included. Regardless, we find that the top

5% of emitters account for 51.6% of quantified emissions.

Component Type CountTotal Emissions

[kg/hr] Mean 5% 25%Median (50%) 75% 95%

Actuator 26 7.45 0.29 0.01 0.01 0.05 0.20 1.54Analyzer 14 1.34 0.10 0.01 0.04 0.09 0.13 0.22Catadyne Heater 248 19.50 0.08 0.01 0.05 0.05 0.14 0.19Chemical Injection Pump 54 8.39 0.16 0.01 0.01 0.02 0.06 0.37Compressor, Cylinder Head 12 1.30 0.11 0.01 0.01 0.06 0.20 0.29Controller 143 22.21 0.16 0.01 0.03 0.07 0.18 0.38Engine (Including Fuel/Start Gas) 28 0.88 0.03 0.01 0.01 0.02 0.05 0.08Flange Connection 15 0.45 0.03 0.01 0.01 0.01 0.03 0.09Meter / Instrumentation 20 2.11 0.11 0.01 0.02 0.03 0.17 0.37Misc 19 0.72 0.04 0.01 0.01 0.02 0.06 0.10Open Ended Line / Vent 115 51.96 0.45 0.01 0.03 0.15 0.38 2.08Other Tank 1 0.02 0.02 0.02 0.02 0.02 0.02 0.02Pneumatic Pump 4 0.16 0.04 0.01 0.01 0.04 0.07 0.07Pressure Safety Valve (PSV) 3 0.83 0.28 0.19 0.21 0.23 0.32 0.39Pump Packing 4 0.32 0.08 0.01 0.02 0.04 0.10 0.20Regulator 34 1.66 0.05 0.01 0.01 0.03 0.05 0.16Threaded Connection 313 13.67 0.04 0.01 0.01 0.01 0.04 0.15Valve 80 5.20 0.06 0.01 0.01 0.02 0.05 0.23

0.0

0.1

0.2

0.3

0.4

Analyzer Compressor, Cylinder Head Meter / Instrumentation Threaded Connection

Mea

sure

d Em

issi

ons

Rat

e [k

g/hr

]

Many categories heavy tailed: - Mean much higher than median- Few observations much much larger

Classic heavy-tailed distribution

6

53

Figure 3-2. Distribution of quantified leak and vent emissions from initial visit 7G LDAR surveys.

3.1.2 Persistence Analysis

Four sites operated by 7G were visited two or more times in the datasets we had available.

For three of these sites, we present results both in terms of number of detected emissions

points and mass emissions rates. For these sites, we can classify leaks and vents as new or

old sources. New sources are those not found in previous surveys, and old sources refer to

a pre-existing emissions point that may or may not have been fixed before the resurvey.

For one processing plant we show data from three surveys, but do not have the level of

granularity to distinguish between new and old sources. Because 7G’s operations are

growing, many existing sites underwent equipment changes and expansions between

surveys. Any major site changes or expansions are listed in Table 3-2. Results are presented

in Figure 3-3 through Figure 3-6.

Table 3-2. Major site changes between initial and revisit surveys.

N = 1,133Total = 138.16 kg/hrMean = 0.12 kg/hrMin = 0.01 kg/hrMax = 3.97 kg/hr5% = 0.01 kg/hr25% = 0.01 kg/hrMedian = 0.04 kg/hr75% = 0.09 kg/hr95% = 0.35 kg/hr

Site Name Site Changes / Expansions

Satellite Pad 1Four new wells brought online, including four new separator modules and necessary piperack extensions.

Satellite Pad 2 No major changes.

Compressor Station No major changes.

Processing Plant Increased condensate tank size from 750 bbl to 1000 bbl (between 2nd and 3rd visit).

Single largest observation100 x larger than median33 x larger than mean

Database ofquantified leaks

Does not includeestimated leaks

Total volume = 173 mcf/d

Un-quantified leaks

• Leaks are quantified by contractor with Hi-Flow sampler– Entrains entire gas flow into sensor port

• A number of leaks were found and reported but not quantified– Inaccessible, unsafe, or out of measurement range (e.g.,

tank hatch)• Missing volumes estimated

– Tanks from average of previously reported tank-associated emissions (6 studies in US)

– Other sources apply average of quantified leaks from that leak type and site type (e.g., pressure safety valve at super pad)

7

Results of 7G leakage surveys

8

37

Figure 2-10. PLR decreases linearly with increasing natural gas production on a log-log basis.

2.2.3.4 Land Use Emissions

Land use impacts depend on the amount of land impacted and the carbon emissions per

unit of land impacted. We discuss these elements separately below, for both upstream and

midstream portions of our analysis.

Land impacted

Land-use data from 7G for existing production levels are scaled to 1 bcf/d production using

linear extrapolation. Current upstream land use is taken as the summed reclamation area of

all production sites, processing plants and compressor stations, access roads, housing

structures, and storage areas. This area is multiplied by the scale-up factor to approximate

future upstream land use area. Midstream land use refers to area occupied by a pipeline

from Grande Prairie to coastal B.C., assumed to be 1100 km in length. A right-of-way of

70 meters for the pipeline is adopted based on the on TransCanada environmental impact

analysis documents [45].

R2 = 0.83

0.01

0.10

1.00

1 2 5 10 20 50Natural Gas Production [mmscf/d]

Prop

ortio

nal L

oss

Rat

e [%

] Satellite PadSuper Pad

Larger sites have proportionallyless loss of gas

Site-level loss rate ranges from 1% to 0.02%

Overall linearly decreasing function in log-log space

which integrate an average emission factor with activity datareflecting the total number of well sites.Influence of High-Emitting Sites on Total Methane

Emissions. Within each basin (Figure 1) and production bin(Figure 3, notched box plots), the mean CH4 emission rate ishigher than the median because of the disproportionateinfluence of low frequency, high emitting sites. The high CH4

emitters are commonly referred to as “super emitters”; sincewe lack information on the site-level CH4 sources, we denotethem simply as “high emitters” and identify them as the top 5%of sites based on the cumulative fraction of CH4 emissions.Figure 4a shows, empirically, that the top 5% of high-emitting

sites account for 57% (CI: 40−70%) of cumulative CH4

emissions, with each of these sites having site-level CH4

emissions >13 kg/h/site. Furthermore, their cumulative CH4

emissions are equivalent to 1.6% (CI: 1.1−2.2%) of their totalCH4 production. This result is consistent with the observationby Brandt et al.44 that the largest 5% of leaks from NG systemstypically contribute over 50% of total leakage volume. Overall,our results suggest that CH4 emission models (or CH4

emission factors) that do not adequately capture thedisproportionate contribution of high emitters may signifi-cantly underestimate total emissions.

Figure 3. Relationship between site-level CH4 emissions and NG production. (a) absolute and (b) production-normalized CH4 emissions. Solidcyan lines show quadratic robust weighted least-squares regressions with bisquare weighting (see Materials and Methods) performed on the entiredata set. Measured site-level CH4 emissions were also binned by deciles of their site-level NG production, which are numbered sequentially on thetop x axis. The notched box plots (outliers not shown) visually depict the data spread in each production decile. The black horizontal line in eachnotched box shows the median. The triangular purple symbols show the mean CH4 emission rate in each production decile and the solid red linesshow the polynomial fit through the mean CH4 emission rate in each decile. These regression equations are (a) log10[CH4 (kg/h)] = 0.30 ± 0.14 ×log10[Prod (Mcfd)] − 0.23 ± 0.38; radj2 = 0.72) and (b) log10[%CH4(kg/h/kg/h)] = −0.71 ± 0.15 × log10[Prod(Mcfd)] + 2.0 ± 0.41; radj2 = 0.93) forthe absolute and production-normalized CH4 emissions, respectively.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b03535Environ. Sci. Technol. XXXX, XXX, XXX−XXX

F

Similar relationship found for ~1000 wells across 8 US basins

by Omara et al. 2018

What happens over time?

9 54

Figure 3-3. Persistence of leaks and vents for 7G satellite pad 1, in terms of number of detected emission points (top) and methane emissions rate (bottom).

55

Figure 3-4. Persistence of leaks and vents for 7G satellite pad 2, in terms of number of detected emission points (top) and methane emissions rate (bottom).

54

Figure 3-3. Persistence of leaks and vents for 7G satellite pad 1, in terms of number of detected emission points (top) and methane emissions rate (bottom).

55

Figure 3-4. Persistence of leaks and vents for 7G satellite pad 2, in terms of number of detected emission points (top) and methane emissions rate (bottom).

Larger sites – Compressors and gas plant

10 56

Figure 3-5. Persistence of leaks and vents for 7G compressor station, in terms of number of detected emission points (top) and methane emissions rate (bottom).

56

Figure 3-5. Persistence of leaks and vents for 7G compressor station, in terms of number of detected emission points (top) and methane emissions rate (bottom).

57

Figure 3-6. Persistence of leaks and vents for a gas processing plant over three surveys.

We find that in all cases, both the number of emissions points and the site-wide emissions

rate decreased between repeat surveys. We also observed that leaks that were fixed tended

to remain fixed and did not reappear. Likewise, leaks that were not fixed tended to persist

and did not go away on their own, indicating that leak persistence is very high.

Consequently, LDAR surveys are resulting in emissions reductions, so long as detected

leaks are fixed. Repeat LDAR surveys regularly found new emission sources, even without

significant site changes occurring, supporting the idea that LDAR surveys must be done

regularly to find and fix new emission sources. Because emissions sources can arise from

equipment failure or breakdown, repeated surveys are useful even in well-managed sites.

The decrease in vents found in repeat surveys was not solely due to intermittent venting or

site changes. Catadyne heaters are seasonal pieces of equipment that are used to maintain

a certain temperature near temperature-sensitive equipment. They are only utilized during

cold weather events and are known for emitting large amounts of gas through poor

combustion. 7G classifies these emission points as vents. In our datasets, initial visits

tended to coincide with cold weather events, with many catadyne heaters operating.

Revisits did not occur during such cold weather events, and as such, the number of catadyne

heaters in use were far less. Thus, the large decrease in vents is not without explanation.

We also find that LDAR surveys performed at site commissioning are valuable, as

equipment is either not properly installed, not fully tightened, or becomes strained under

Large variation between contractors

• LDAR effectiveness found to vary between contractors

• Contractor B used most commonly

• In two cases, Contractor A visited same site within 2 months of Contractor B

• Uneven performance in reporting emissions

• Only surveys from Contractor B used

11

13

Figure 2-3. Differences in number of emissions sources found by two different LDAR survey contractors at the same sites. Satellite pad visited two months apart (left), and super pad visited three and a half months apart (right).

2.1.4 Supplementing Incomplete Data

Not all detected leaks and vents were quantified, as the Hi-Flow sampler requires that the

emission point be within reach of the surveyor and safe to measure. For instance, tank

emission points have never been quantified in any 7G survey. Other emissions, such as

from overhead pipe racks, were also inaccessible. Rather than assigning these emission

points a value of zero, we chose to supplement the empirical LDAR dataset and assign an

estimated flow rate to these unquantified points. For tank-related emission sources,

emission factors (EFs) from a database of measured tank emissions were applied. This

database was compiled by Stanford University researcher Jacob Englander, and includes

613 unique tank-specific emission measurements, broken down by site-type, from a

number of studies [11, 34, 35, 36, 37, 38, 39]. For non-tank-related emission sources, that

were unquantified, we applied the average flow rate from leaks that were quantified. We

use only initial surveys for determining these average flow rates, to better represent native

leak volumes before any repair. See Table 2-2 for applied flow rates utilized in this work.

The standard error of each mean flow rate is estimated as using:

Comparison of 7G leakage to US average

• Scale up LDAR survey to all 7G sites for 2016 estimate

– Sites measured in 2016 assigned quantified emissions

• All gas processing and condensate processing sites, 5 out of 8 super pads,

18 out of 26 satellite pads

– Sites not measured are estimated

• log-log relationship between volume and leakage rate (wellpads) or simple

average of measured sites (gathering compressors)

12

US average Seven GenerationsSegment % of gross

prod.2016 volumes % of gross prod.

Production 1.3% 472 mcf/d 0.16%Gathering 0.5% 103 mcf/d 0.04%Processing 0.1% 215 mcf/d 0.07%Transmission and storage 0.3% - -Local distribution 0.1% - -Total loss 2.3% - -Total production 291,000 mcf/d 100%