A practical approach to LDAR effectiveness evaluation

Abstract As a growing number of oil and gas facilities worldwide implement leak detection and repair (LDAR) plans, the owners and the public are interested in better understanding their effectiveness, i.e. how total facility leak emission is decreased as a result of LDAR implementation. For this objective, it is critically important when calculating total mass leak emission to consider an estimated leak lifetime for the leakers found. Obviously, more frequent monitoring and shorter leak repair time should return lower leak lifetime and less emission. This simple yet important factor is mentioned but not detailed in existing guides. Some guides suggest formulae with the same equipment operating time both for leakers and non-leakers, which is not quite accurate since the actual leak lifetime could be significantly less than the total equipment operating time due to the proper LDAR implementation. In this article, we address this gap. The LDAR effectiveness evaluation approach utilizing leak lifetime is suggested and illustrated using examples.


Introduction
Leak detection and repair (LDAR) plans are widely recognized as an approach to control and reduce fugitive leaks of pollutants and greenhouse gases in oil, gas, and chemical facilities.The requirement to implement LDAR plans is included in the global Methane Guiding Principles [1], supported by all industry majors.The US Environmental Protection Agency (US EPA) published the best-practice guide [2] for voluntary LDAR implementation.The American Petroleum Institute (API) published a comprehensive Compendium [3] of equipment leak quantification methods.The United Nations (UN) Oil & Gas Methane Partnership (OGMP) publishes the Framework for Emission Reporting and Technical Guidance Documents (TGDs), including TGDs for unintended equipment leaks [4].The UN program 'Clean Development Mechanism' allows, under certain conditions [5], recognizing decreases in leak emissions as Certified Emission Reductions (CERs) or carbon credits, which can be later traded on marketplaces.
The nature of equipment leaks is that they are usually unintentional, small and hard to detect, and may exist for a long time.An unintentional leak might happen at any time due to wear and tear on any equipment connection: on flanges, valves, pumps etc., all named 'components'.If not found and repaired, unintentional equipment leaks could essentially contribute to the total facility emission when combined with process vents and flares.
Control effectiveness for an LDAR program (or LDAR effectiveness) is defined as an absolute and relative fugitive emissions reduction due to LDAR program implementation.According to the Texas Commission on Environment Quality (TCEQ) [6], LDAR effectiveness varies from 30% to 97% depending on LDAR frequency, leak definition threshold and repair requirements.The baseline equipment leak emission (components population count) and actual equipment leak emissions should be calculated to evaluate LDAR effectiveness.
In this paper we suggest an approach for actual leak emission calculation utilizing a leak lifetime time basis for a more accurate LDAR effectiveness evaluation.Below we consider this step-by-step approach in an example.We suggest for the purposes of clarity the following assumptions: Onshore gas production facility is considered, Methane (CH 4 ) emission is calculated, LDAR monitoring is implemented using optical gas imaging (OGI) cameras with leak sensitivity of 60 g/h.

Baseline emission factors
Emission factors (EFs) depend on component and service type and are defined in Table 1.
Baseline emission is calculated using the following equation: where:

Sample calculation
Input data 2  There are 100 gas valves and 250 flanges in a stream at a gas production site that contains 80% methane by weight (w ¼ 0.8).We calculate the methane emissions produced in a year.
Calculation methodology: Emission is calculated using emission factors from Table 1

Leak lifetime definition
The leak lifetime T leak is a sum of awareness time T A from leak appearance to leak detection and repair time T R from leak detection to leak repair: An unintentional leak may appear at any time ([3], 7.1.5"Time Basis of Equipment Leaks").Therefore, we suggest: According to OGMP recommendations [4], if monitoring is done for the first time or no data on previous monitoring exists, any leak found is suggested to have existed from the beginning of the reporting period.If repair is not done, the leak is considered to exist until the end of the reporting period.

Actual EFs for monitoring with OGI cameras
OGI cameras can detect a leak and help to localize the leak source, but cannot reliably measure the leak rate.As a result of facility monitoring with OGI cameras, the component population is divided into two classes: class 'no-leak', including the majority of components, where leaks were not detected by the monitoring tool; and the minority class 'leak', where the leaking components were detected.For this kind of monitoring a special approach to total facility mass emission quantification was developed ([3], 7.1.4.1 "Leaker Factors Using Leak Screening Surveys"), and EFs for classes 'leak' and 'no-leak' were derived.It should be noted that EFs for the 'no-leak' class are not equal to zero, although they are very small.This way the approach acknowledges the fact that small leaks still may exist below the threshold of the instrument sensitivity.The EFs for monitoring with OGI cameras depend on the OGI leak rate sensitivity and are shown in Table 2.
Note that emission factors in Table 2 do not depend on the service type.
The leak detection thresholds for leak detection instruments are found experimentally and discussed with the industrial community before acceptance by regulatory agencies.For OGI cameras, sensitivity thresholds depend on gas temperature difference, windspeed and distance.Although in laboratory conditions an OGI camera with a cooled matrix can demonstrate perfect sensitivity up to 0.35 g/hr, for field survey reporting purposes a conservative threshold is usually taken, which is 60 g/hr [7].
The actual emission is expressed using the following equation: and the sum is performed over all leaking and non-leaking components correspondingly.
Noting that: and that EFs for the class 'leak' are much greater than the corresponding EFs for the class 'no-leak': we can transform expression (2-3) into the more practically convenient form where M zeroline is calculated the same way as M baseline using Equation ( 1) with emission factors for class 'no-leak' from Table 2, and M leaks is calculated over leaking components using emission factors for class 'leak' from Table 2 and leak lifetime as defined by Equations (2-1)-(2-2).Emission volume M zeroline can be thought of as the lowest achievable equipment leak emission level for a given facility and a given leak threshold definition.

Input data
On the same site as described in Example (1.2) above with 100 gas valves and 250 flanges, with methane weight share w ¼ 0.8, regular LDAR monitoring with OGI cameras is performed.The reporting period is one year.At the first observation, performed on 1st of April, no leaking components were detected.At the second observation, performed on 1st of October, one leaking valve was detected.The leak was repaired and verified using the leak detection method in 10 days.We calculate the actual methane emissions and compare them with the emissions baseline from Example (1.2).

Effectiveness definition
As mentioned above, LDAR effectiveness is defined in terms of relative emission reduction: It should be noted here that the actual emission M actual does not have to be less than the baseline emission M baseline : It might happen that the actual emission appeared to be greater than the baseline, especially in the first few cycles of LDAR monitoring, when a lot of hidden leaks may be found.From the formal point of view, at this moment the LDAR efficiency will be negative.This should not be taken as a reason to stop the LDAR implementation.After a few LDAR cycles most of leaks will be repaired and the actual effectiveness will stabilize essentially below the baseline, according to TCEQ [6].
The parameters such as EFs and leak lifetime (and in some cases even the population count), which are used to evaluate effectiveness, are values with uncertainty.Therefore, the effectiveness is also a value with some level of uncertainty.The relevant uncertainty estimation and error propagation techniques are described in the API Compendium, in the sections '3.7 Emission Estimation Quality/3.7.1 General Statistical Approach to Calculating Uncertainty' [3].

Input data
As in the examples above.

Conclusion
In this paper a simple approach to LDAR efficiency evaluation is suggested.The approach is illustrated by examples for LDAR OGI methane monitoring on onshore facilities, but it can be applied equally to other types of monitoring and facilities.For example, LDAR programs for air pollutants can also use the approach described.Owners and operators using this approach can better understand and correlate the influence of monitoring frequency, repair time and other factors, such as definition threshold, on overall LDAR effectiveness.Therefore, operators and owners can achieve better results in terms of LDAR costs.

Disclosure statement
No potential competing interest was reported by the authors.