Final Report Open-Source High Flow Sampler for Natural Gas Leak Quantification Colorado State University: Daniel Zimmerle Timothy Vaughn Kristine Bennett Cody Ross SLR International: Matthew Harrison Aaron Wilson Chris Johnson Prepared for California Air Resources Board Contract: 19ISD010 Date: 04/14/2022 Open-Source High Flow Sampler for Natural Gas Leak Quantification 2 Disclaimer The statements and conclusions in this report are those of the contractor and not necessarily those of the California Air Resources Board. The mention of commercial products, their source, or their use in connection with material reported herein is not to be construed as actual or im- plied endorsement of such products. Open-Source High Flow Sampler for Natural Gas Leak Quantification 3 Executive Summary High flow sampling remains the predominant method for measuring individual emission points. A High Flow Sampler (HFS) entrains the emissions in a stream of air and routes the combined flow through an instrument where flow and concentration measurements are used to compute the emis- sion rate. Developed in the 1990s, the method is widely adopted, but the most prevalent commer- cial instrument for HFS measurements (Bacharach® HI FLOW® Sampler or “BHFS”) is no longer manufactured. The California Air Resources Board (CARB) commissioned Colorado State Uni- versity (CSU) and CSU collaborated with SLR International (SLR) to develop an open-source architecture for future instruments and, if possible, to encourage the development of new commer- cial products. Open-source architecture: An architecture for an open-source high flow sampler (OS-HFS) was developed during the project. Key innovations were demonstrated, including: • Selection and demonstration of next genera- tion gas sensors, blower, and flow measure- ment sensors. Key issues with these subsys- tems were also identified, with advice for fu- ture users of the design. (Figure 1). No readily acceptable substitutes or im- provements were found for the existing BHFS appurtenances, such as the hose, bag, and other emission collection equipment. The prototype continues with the BHFS sup- plied equipment. • The OS-HFS demonstrates the concept of a sensor-flexible high flow platform that could adapt to new uses and challenges as they arrive. This architecture is achieved by isolating computation and display elements from sensing and blower elements. A commercial, off- the-shelf, single board computer (SBC) was used for computation and display. The SBC communicated to all sensors via industry-standard serial protocols. This architecture iso- lates the computer, data interfacing, human interface, battery management, and similar su- pervisory systems from the sensor systems. Adopting this approach will make high flow instruments more adaptable as needs change and expectations for data handling and communication evolve. • To the extent possible, the OS-HFS utilized open-source or easily sourced components, making the system accessible to anyone interested in constructing a copy or using the Figure 1: Cutaway view of the OS-HFS instru- ment design. Open-Source High Flow Sampler for Natural Gas Leak Quantification 4 design as the basis for a productized instrument. This includes use of platform-independent Python coding for all software, readily accessible sensor components, and use of a com- mercial high-volume blower used in another application (computers). • Illustrating a new package de- sign that fits better with in- field experience than the backpack packaging of the BHFS. Improvements include an improved graphical user interface, better placement of the user interface, smaller in- strument size, and longer and more stable battery life. (Fig- ure 2) The prototype demonstrates the concept of a sensor-flexible high flow platform that could adapt to new uses and challenges as they arrive. Section 5.1 provides additional detail summarizing the new open architecture. Testing pre-production HFS instruments: The project also tested three additional pre-produc- tion HFS instruments under development by three different companies. Testing indicates that these units should be capable of replacing the current BHFS when completed. Certain features of the CSU prototype were represented in some of the pre-production instruments, such as: 1) Modernized user interface and data collection. For example, several pre-production units added remote displays that were not tethered to the unit, and all provided some amount of data logging. 2) Two of the pre-production units utilized smaller packages that were more readily posi- tioned by an operator during measurement, while one closely mimicked the BHFS back- pack configuration. 3) Despite usability progress, field use and associated ergonomics remain challenging: All the instruments – CSU prototype included – are still bulky and heavy for use in the field, par- ticularly since no unit identified more compact or lighter weight hoses and enclosures for capturing the leak. However, none of the instruments included several key elements identified during this work (see Section 5), most importantly, support for multiple complementary sensors. Field experience Figure 2: Integrated OS-HFS instrument during initial Testing. Open-Source High Flow Sampler for Natural Gas Leak Quantification 5 indicates that it is operationally challenging to correct for varying gas mixtures in field conditions. The CSU prototype included one sensor as the primary measurement sensor and a second sensor to detect changes in gas mix, even if a composition-dependent correction was not possible without more information. The pre-production units were either methane-specific or utilized combustible gas sensors, like the BHFS. Additionally, pre-production units should undergo testing similar to what was conducted here prior to field use. All units exhibited problems of some nature – all likely fixable – but additional testing would raise confidence in these instruments for future use. High flow method and instrument testing: Incidental to the main project goals, the project per- formed extensive testing of the high flow method in a range of conditions. Controlled-release testing on realistic equipment at CSU’s Methane Emissions Technology Eval- uation Center (METEC) indicates that uncertainties for high flow measurements independent of any instrument are likely underrepresented in current literature, analysis, and reporting. Testing with the CSU prototype, one commercial instrument, and three pre-production instruments indi- cated substantially more variability and error in measurement results on realistic equipment than was seen when emissions were introduced directly into the instrument. These results indicate that user practices and environmental conditions may contribute substantial uncertainty to component- level measurements. Characterization of these uncertainties and development of measurement best practices are advised. Additional notes: The high flow method remains one of the few methods suitable for isolating individual emission points for direct measurement. Quality instruments that will play a key role in characterizing leak rates for many years to come are near completion. Open-Source High Flow Sampler for Natural Gas Leak Quantification 6 Contents 1. Introduction .........................................................................................................................9 1.1 Project Context and Objectives .....................................................................................9 1.1.1 High Flow Sampling in the Context of Other Methods ...........................................9 1.1.2 Current Market Situation for HFS Instruments ..................................................... 10 1.1.3 Project Objectives ................................................................................................ 11 1.2 Organization of This Report ........................................................................................ 12 1.3 The Technical Advisory Panel..................................................................................... 12 2. Overview of the High Flow Method ................................................................................... 13 2.1 Method Description..................................................................................................... 13 2.2 Sources of Instrument Error ........................................................................................ 14 2.3 Potential Method Errors .............................................................................................. 15 3. Open-Source Design of a High Flow Sampler .................................................................... 16 3.1 Architecture and Key Components .............................................................................. 16 3.1.1 Mechanical Architecture ...................................................................................... 16 3.1.2 Electronics Architecture ....................................................................................... 19 3.1.3 Software Architecture .......................................................................................... 23 3.2 Selection of Key Components ..................................................................................... 26 3.2.1 Blower ................................................................................................................. 26 3.2.2 Flow Measurement............................................................................................... 29 3.2.3 Concentration Measurement ................................................................................. 31 3.2.4 Background Concentration Measurement ............................................................. 36 3.3 Accuracy Analysis ...................................................................................................... 37 3.3.1 Lower Detection Limit ......................................................................................... 38 3.3.2 Quantification Accuracy ...................................................................................... 39 3.3.3 Gas Composition Testing ..................................................................................... 42 3.3.4 Practical Use Testing ........................................................................................... 44 4. Testing of Pre-production Prototype HFS Units ................................................................. 46 4.1 Lower Detection Limit ................................................................................................ 48 4.2 Quantification Accuracy ............................................................................................. 50 4.3 Gas Composition Testing ............................................................................................ 55 Open-Source High Flow Sampler for Natural Gas Leak Quantification 7 4.4 Practical Use Testing .................................................................................................. 57 4.5 Testing Summary ........................................................................................................ 61 5. Conclusions and Recommendations ................................................................................... 63 5.1 Key Conclusions: CSU Prototype OS-HFS ................................................................. 63 5.2 Key Conclusions: Pre-Production Prototype Instruments ............................................. 64 5.3 Key Conclusions: High Flow Method and Instrument Testing ..................................... 66 Appendix A Data Files and Additional Materials .................................................................. 68 Appendix B Recommended High Flow Method ................................................................... 69 Appendix C Testing Protocol ............................................................................................... 76 References ................................................................................................................................ 81 Table of Figures Figure 1: Cutaway view of the OS-HFS instrument design. .........................................................3 Figure 2: Integrated OS-HFS instrument during initial Testing. ...................................................4 Figure 3: High flow instrument in use. Blue fabric encloses the leak source. ...............................9 Figure 4: A high flow sampler (HFS) quantifies the flowrate of an emission source directly ...... 13 Figure 5: A BHFS is used to measure an emission on a wellhead. ............................................. 14 Figure 6: Cutaway schematic of the integrated instrument. ........................................................ 17 Figure 7: Electronics architecture for the prototype unit. ........................................................... 20 Figure 8: Photos of key electronics assemblies. ......................................................................... 21 Figure 9: Software architecture schematic. ................................................................................ 24 Figure 10: Example GUI tab ..................................................................................................... 25 Figure 11: Fan curve comparison for the GVP-100 blower used in the BHFS ............................ 27 Figure 12: Inside the 3M GVP-100 blower motor used in the BHFS.......................................... 28 Figure 13: First OS-HFS flow meter calibration ........................................................................ 31 Figure 14: Second OS-HFS flow meter calibration .................................................................... 31 Figure 15: Sensor selection strategy for OS-HFS development. ................................................. 32 Figure 16: The SGX INIR-ME100 and Nevada Nano MPS Flammable Gas Sensor .................. 33 Figure 17: Test chamber for prototype sensor calibration and characterization. ......................... 34 Figure 18: Example data capture from sensor characterization. .................................................. 35 Figure 19: The INIR sensor shows a strong dependence on ambient pressure. ........................... 36 Figure 20: The Nano sensor output shows less pressure dependence.......................................... 36 Figure 21: Prototype OS-HFS INIR sensor lower detection limit. .............................................. 38 Figure 22: Prototype OS-HFS Nano sensor lower detection limit. ............................................. 38 Figure 23: Prototype OS-HFS INIR sensor quantification accuracy. .......................................... 39 Figure 24: Prototype OS-HFS Nano sensor quantification accuracy. ......................................... 40 Open-Source High Flow Sampler for Natural Gas Leak Quantification 8 Figure 25: Additional Prototype OS-HFS accuracy testing results. ............................................ 42 Figure 26: Prototype OS-HFS INIR sensor gas composition testing results................................ 42 Figure 27: Prototype OS-HFS Nano sensor gas composition testing results. .............................. 43 Figure 28: Prototype OS-HFS INIR sensor practical use testing. ............................................... 44 Figure 29: Prototype OS-HFS Nano sensor practical use testing. ............................................... 44 Figure 30: HFS units at METEC. .............................................................................................. 46 Figure 31: BHFS lower detection limit. ..................................................................................... 48 Figure 32: GFM lower detection limit. ...................................................................................... 49 Figure 33: Hetek lower detection limit. ..................................................................................... 49 Figure 34: Sensors lower detection limit. ................................................................................... 49 Figure 35: BHFS quantification accuracy. ................................................................................. 51 Figure 36: GFM quantification accuracy. .................................................................................. 52 Figure 37: Hetek quantification accuracy................................................................................... 53 Figure 38: Sensors quantification accuracy. ............................................................................... 54 Figure 39: BHFS gas mixture results. ........................................................................................ 56 Figure 40: GFM gas mixture results. ......................................................................................... 56 Figure 41: Hetek gas mixture results. ........................................................................................ 56 Figure 42: Sensors gas mixture results. ...................................................................................... 57 Figure 43: Pad 4 at METEC with equipment groups marked. .................................................... 57 Figure 44: METEC release points used for testing. .................................................................... 58 Figure 45: BHFS practical use testing........................................................................................ 59 Figure 46: GFM practical use testing. ........................................................................................ 60 Figure 47: Hetek practical use testing. ....................................................................................... 60 Figure 48: Sensors practical use testing. .................................................................................... 61 Open-Source High Flow Sampler for Natural Gas Leak Quantification 9 1. Introduction 1.1 Project Context and Objectives Accurate source emission measurement is key to many efforts by the California Air Resources Board (CARB), as well as industry efforts to understand and reduce emissions. CARB, NGOs1–5, and the US De- partment of Energy (DOE)6–8, have sponsored research programs to understand natural gas emissions. These field efforts have often used direct emission rate measurement techniques to obtain an accurate characterization of emissions at the component level on a diverse range of facilities from production and exploration through distribution. Ul- timately, CARB is most interested in mitigation of emissions. Practical implementation of mitigation programs often requires pre- and post- mitigation emission measurements. Whether for research or mitigation, measurement of component emis- sions is most often completed using the high flow measurement method using a purpose-built in- strument: A High Flow Sampler (HFS). Despite the varied needs and uses, there is now no com- mercially available HFS instrument. The most prevalent commercial instrument for HFS measure- ments to date (the Bacharach® HI FLOW® Sampler or “BHFS”), a 1990s design, was discontin- ued by its manufacturer 2 years ago. At a time when demand for measurement is growing, the devices are now unavailable. 1.1.1 High Flow Sampling in the Context of Other Methods While downwind techniques, such as tracer-flux 9,10 or mass balance techniques 11, can estimate emissions at the facility level, corrective action typically requires detection and measurement at the component level. In addition, measurements of component-level emission sources are needed to inform operators’ emissions mitigation programs. Regulatory programs are most interested in emission rates by component or process to better target emission reduction programs and to per- form cost/benefit assessments. All these efforts require measurement using a qualified, standard- ized device and an associated measurement protocol for its use. The scientific and industry standard for this type of measurement is the HFS, originally developed in the 1990s 12. This instrument has achieved regulatory acceptance and is routinely utilized for scientific studies 1–3,13–15, internal industry emission surveys, and regulatory reporting16. In recent years, a typical process is to detect leaks using optical gas imaging (OGI) or US Environmental Protection Agency (EPA) Method 2117, and then subsequently quantify those leaks using an HFS. Only when an HFS does not work (e.g., leak is too large) do teams resort to other leak measurement methods, such as in-stream anemometers or anti-static bags18. Figure 3: High flow instru- ment in use. Blue fabric en- closes the leak source. Open-Source High Flow Sampler for Natural Gas Leak Quantification 10 In brief, to use an HFS instrument, the operator loosely encloses the emission point. A blower in the HFS instrument then draws both the entire leak and surrounding air through the instrument and across sensors which measure both the concentration (% gas by volume) of the target gas species (e.g., methane or volatile organic compounds (VOCs)) and the total volume of air flow. A second sensor measures the background concentration of the same gas in surrounding air. Mass flow of the target species is calculated from the overall mass flow, background concentration, and emis- sion-point concentration. In summary, while many instruments are available to identify emission locations (optical gas im- aging, gas sniffers, laser-based detectors), and many instruments can produce an estimate of the concentration of the target species in air (ppm, or ppm-m readings), the only widely accepted, general method to quantify emissions from components is high flow sampling. 1.1.2 Current Market Situation for HFS Instruments While the technique is simple and well understood, the only commercially available instrument, the BHFS 19, has recently been discontinued. Many existing BHFS instruments are still in use, but long-term support is uncertain. The instrument has faced several challenges due to its usability (weight, bulk, potential lack of intrinsic safety) and concerns about its accuracy when utilized with wet gas20–23. Recent studies (spring 2019) have further illuminated problems with the BHFS. The National Physical Laboratory in the UK analyzed the instrument’s sensors in a recent paper24 and highlight several software and sensor performance issues. CSU also performed extensive testing in spring 2019 25 using the gas mixing and flow control capabilities at the Methane Emissions Technology Evaluation Center (METEC) at CSU. These tests revealed several previously undocumented issues with the instrument, including variations in indicated emissions with total flowrate through the instrument, and changes in indicated emissions when switching from high-to-low, or low-to-high, gas concentrations. A recent field campaign 26 paired a subset of BHFS measurements with con- current SUMMA-canister samples and found poor correlation in the so-called transition zone be- tween concentrations where the sensor operates in catalytic oxidation mode and thermal conduc- tivity mode – an indicated leak hydrocarbon concentration of 3-6 %. The instrument failed to tran- sition even though efforts were made to minimize the use of practices known to cause this un- wanted behavior. Several implementations of high flow systems have been developed for specialized circumstances, including measuring emissions larger than the BHFS capacity27; measuring small, distributed emissions28; measuring emissions from underground pipeline leaks3; and measuring emissions us- ing the “dynamic chamber” method29,30. Each of these developments has resulted in much larger instrumentation sets, typically vehicle mounted, and often using laboratory-grade instruments to provide the necessary sensor resolution. In October 2017, EPA’s Office of Research and Devel- opment (EPA/ORD), in conjunction with EPA Region 8, were awarded funding of $100,000 Open-Source High Flow Sampler for Natural Gas Leak Quantification 11 through the EPA Regional Applied Research Effort program to develop and bench test components of an HFS prototype 23. In summary, when this project started, no new commercial instruments were under development. Revision or redevelopment of an HFS has been impeded by the expected small lifetime volume (≈1500 units over 20 years*) and required development cost. While several scientific groups have developed their own one-off HFS instruments, none of these designs represent a reproducible, compact, or standardized design necessary for practical deployment and widespread use. A prac- tical, reliable, compact, and versatile version of an HFS instrument is necessary to achieve wide- spread use and acceptance of results, and is critically needed for industry, governmental and sci- entific purposes. Several technological advances since the 1990s are now available and can improve the operation and packaging of future HFS instruments: • Gas concentration sensors have advanced substantially since the 1990s, including sensor-in- ternal analog-digital (A/D) conversion, new sensor elements, and higher sensitivity. • Auxiliary sensors for flow, power supply monitoring, and other purposes have both advanced and decreased in cost since the 1990s. • Computer control of air flow (i.e., blower speed) is more practical using readily available speed controllers, including controllers internal to the blower assembly itself. • Computational and computer memory improvements now support time series logging of sen- sor readings for quality assurance, method requirements (e.g., ensuring stabilized readings), and capturing changing emissions. • Current lithium batteries provide more compact and stable power supplies than were available in the 1990s. 1.1.3 Project Objectives The stated objectives of this project were to develop an open-source design that reduces the risk of producing a new product generation, i.e., for a commercial entity to develop a new HFS instru- ment. Project deliverables were: 1) Designs for key components which are not currently available off-the-shelf. 2) Consider designs that retrofit existing instruments with improved sensors and electron- ics – if retrofits emerge as a viable path forward. 3) Pre-production software embodying the method and quality control. 4) Recommended component selection or design for intrinsic safety. 5) Sourcing and construction instructions to build the instrument. * Estimate based upon anecdotal data from private conversations with a company distributing BHFS units over the last decade. Estimate does not necessarily account for any increasing interest in methane emissions mitigation due to regulatory, company, or certified gas programs. Open-Source High Flow Sampler for Natural Gas Leak Quantification 12 Note that the open-source design would still require additional product development and safety certification to become a saleable instrument. During the project, several commercial entities announced plans to develop new HFS instruments. Drivers for this sudden increase in commercial interest are unclear but are likely due to increased interest by operators and regulators in replacement instruments for the BHFS. It is also possible that this project both stimulated additional interest and helped companies identify HFS instruments as a commercially viable opportunity. By spring 2021, it was increasingly clear that the commer- cial efforts would arrive on market near the end date of this project. Therefore, the project team engaged with three manufacturers and tested each of their pre-production units using the same protocol used for the prototype OS-HFS being developed herein. The incumbent BHFS was also included in testing as a point of comparison for all instruments under development. 1.2 Organization of This Report Chapter 2 provides an overview of the high flow method, including a qualitative summary of po- tential error sources in the method and in instruments used for the method. Chapter 3 describes the OS-HFS design created by this project, with emphasis on selection and evaluation of key instru- ment components. Chapter 4 summarizes prototype testing and Chapter 5 provides conclusions and recommendations for the high flow method and instruments. Appendices provide auxiliary written information. Test data, software, mechanical design, and PCB layouts for the OS-HFS are included in attached files. 1.3 The Technical Advisory Panel The project also included a Technical Advisory Panel (TAP) to review progress of the project. The TAP included approximately 50 contributors in the oil and gas arena, including representatives from oil and gas companies, methane detection technology companies, and regulatory agencies. The panel met several times via Zoom or Teams in the early stages of the project in 2020 (April 16th, May 21st, June 29th) to discuss options for blowers, sensors, and other design elements desired in a next-generation HFS. Subsequent meetings in 2021 (January 20th, July 14th) and 2022 (March 29th) provided updates on progress, presented prototype information, and solicited feedback from the TAP. These meetings occurred via Teams. Additionally, the project team met with several of the TAP members on an individual basis to get more detailed input and understand the needs and desires of the likely cus- tomer base. Two members of the project team also participated in a “customer discovery” process during early 2021 that solicited feedback from HFS users, potential suppliers, and other groups. Learnings from these stakeholders impacted the OS-HFS design in many tangible and intangible ways. Open-Source High Flow Sampler for Natural Gas Leak Quantification 13 2. Overview of the High Flow Method 2.1 Method Description The high flow method quantifies the flow rate of an emission source (𝑄𝑄𝐶𝐶𝐻𝐻4) by entraining it com- pletely within a flow stream and simultaneously measuring the flowrate of the total stream (𝑄𝑄ℎ𝑓𝑓𝑓𝑓 = 𝑄𝑄𝐶𝐶𝐻𝐻4 + 𝑄𝑄𝑎𝑎𝑎𝑎𝑎𝑎) and the concentration of the emission (𝐶𝐶𝐶𝐶𝐻𝐻4) entrained. The two measured parame- ters are then used to calculate the unknown flowrate of the emission source, as given in equation set ( 1 ). 𝐶𝐶𝐶𝐶𝐻𝐻4 = 𝑄𝑄𝐶𝐶𝐻𝐻4 𝑄𝑄𝐶𝐶𝐻𝐻4 + 𝑄𝑄𝑎𝑎𝑎𝑎𝑎𝑎 𝑄𝑄𝐶𝐶𝐻𝐻4 = 𝐶𝐶𝐶𝐶𝐻𝐻4�𝑄𝑄𝐶𝐶𝐻𝐻4 + 𝑄𝑄𝑎𝑎𝑎𝑎𝑎𝑎� 𝑄𝑄ℎ𝑓𝑓𝑓𝑓 = 𝑄𝑄𝐶𝐶𝐻𝐻4 + 𝑄𝑄𝑎𝑎𝑎𝑎𝑎𝑎 𝑄𝑄𝐶𝐶𝐻𝐻4 = 𝑄𝑄ℎ𝑓𝑓𝑓𝑓𝐶𝐶𝐶𝐶𝐻𝐻4 ( 1 ) Figure 4: A high flow sampler (HFS) quantifies the flowrate of an emission source directly by entraining the emission source (QCH4) completely within a bulk flow of surrounding air (Qair) and measuring the concen- tration of the emission species within the total flow through the HFS (Qhfs). The following assumptions are intrinsic to the method: • That the air and methane behave as ideal gases. • That the air and methane are perfectly mixed. • That the concentration measurement is perfectly accurate from 0-100 %. • That the flow stream measurement is perfectly accurate over its defined usage range (Qhfs,min to Qhfs,max), and unaffected by the relative proportion of air/methane. • That the emission source is entrained completely by the flow stream. This implies that QCH4 must be <= Qhfs,max. Open-Source High Flow Sampler for Natural Gas Leak Quantification 14 It is also important to note that concentration measurements made by instruments vary in the specificity of gas sampled. While some units may use sensors that are sensitive only to methane, other instruments utilize sensors that are more gen- erally sensitive to all hydrocarbons, or potentially to all combustible gases. The archetype commercial offering for natural gas component leak quantification has been the BHFS shown in use in Figure 3 and Figure 5. A blower contained within the backpack creates the flow stream drawing in the emission source and excess air. In both Figure 3 and Figure 5, the operator has used the bag attachment to help block the wind and direct the emission into the BHFS. The bag is draped loosely to avoid restricting flow to the BHFS and allow excess air to be drawn in and mixed with the emission source. Sensors inside the back- pack measure the gas concentration and report the measured flow, measured concentration, and calculated emission rate to a hand-held screen. The results can also be stored as text data and retrieved over a computer interface. An HFS could be constructed in a variety of ways, and several groups have implemented their own versions using different techniques for flow and concentration measurements (see 27,31). Other ap- proaches are also possible. The configuration of a specific HFS would be guided by the application of interest, usage model, required accuracy, desired cost, safety concerns, available resources, and other factors. Converting the principles of this simple method into a useful, real-world measurement device is non-trivial and subject to potential limitations and sources of error, some of which are discussed in the following sections. 2.2 Sources of Instrument Error Gas composition, cross sensitivities, sensor saturation, and sensor poisoning can all contribute to errors in readings reported by HFS instruments. Environmental effects such as temperature, pres- sure, and relative humidity may also influence reported results. Properly accounting for each of these factors and specifying the appropriate (and inappropriate) use cases for the instrument can solve most of these issues, either by design or procedural approaches. Gas composition and typical emission rates will vary greatly across the natural gas supply chain along with environmental conditions. For example, using an HFS instrument with a methane spe- cific detector to measure tank thief hatch emissions may provide an accurate methane emission rate but will likely under report the whole gas emission rate and give an inaccurate picture of VOC Figure 5: A BHFS is used to measure an emis- sion on a wellhead. Open-Source High Flow Sampler for Natural Gas Leak Quantification 15 emissions present. Depending on the end goal of the measurement this may or may not be prob- lematic. 2.3 Potential Method Errors Two potential shortcomings of the high flow method are incomplete capture and “over-capture”: Incomplete Capture may occur when attempting to capture a source where: • The flowrate is larger than the HFS instrument is capable of measuring. • An emission source has an exit velocity that cannot be overcome by the HFS blower. • Wind disperses the emission source such that the HFS blower cannot entrain the emissions. • Enclosure of the emission source is incomplete, and some emissions are emitted outside the enclosure and not entrained into the instrument. Practitioners can guard against incomplete capture by observing the measurement with an OGI camera to ensure that all visible emissions are entrained into the instrument.25 Over Capture occurs when the motive flow of the HFS withdraws an emission at greater than its natural emission rate, effectively sampling “stored emissions” rather than “emission rate.” Over capture may occur when insufficient dilution air is admitted along with the emission source and the HFS effectively “vacuums” the source from its containing vessel. While this error may occur in a number of situations, it is most prevalent when measuring openings on tanks, i.e., when sampling a large volume of stored emissions at near-atmospheric pressure. Open-Source High Flow Sampler for Natural Gas Leak Quantification 16 3. Open-Source Design of a High Flow Sampler The CSU prototype open-source high flow sampler (OS-HFS) was meant to be a replacement for the BHFS that addressed that instrument’s shortcomings, expanded its capabilities, and refreshed component selection for long-term supportability. Performance targets for the OS-HFS were guided by the specifications of the BHFS as a baseline. The project sought to reproduce the general functionality and maintain the existing usage model of the BHFS but with improved performance and user experience. Overall goals were to: • Reduce the size and the weight. • Improve battery performance. • Expand the lower and upper ends of the measurement range. • Improve overall system accuracy. Preliminary designs showed that this would be possible with careful selection and integration of appropriate off-the-shelf components. 3.1 Architecture and Key Components 3.1.1 Mechanical Architecture This section provides an overview of the electronic and mechanical architecture of the instrument. The mechanical architecture of the instrument was designed for a specific usage model, based upon field experience of the project team in prior projects 1,2,25,26. Previous instruments: The BHFS packaged the instrument into a metallic box placed in a backpack for ease of carrying. While the backpack could potentially hold some attachments for the instru- ment, there was generally insufficient space in the backpack for the hose and larger capture equip- ment. As a result, the operator(s) would need to make multiple trips to bring equipment to each leak. Additionally, field experience indicates that, in actual use, the operator typically placed the backpack on the ground, placed the display in a convenient position using the magnetic attachment, and then arranged inlet equipment to capture the leak (Figure 5). The team’s field experience in- dicated that the backpack design was not well suited to practical field conditions. The backpack is difficult to clean and picks up dirt and oil from the surrounding area. Attachments likewise become dirty, and it is problematic to place these in the backpack with the instrument. Finally, the cord and display, while often convenient, were also inconvenient to stow and, in many cases, could not be magnetically attached near the point of measurement. In practice, many field teams utilize two operators for high flow measurements; one to enclose the emission source, and one to monitor and record the data from the HFS display. Current design: The mechanical architecture of the OS-HFS is based upon the following use model: • Move instrument and required attachments near the leak. Open-Source High Flow Sampler for Natural Gas Leak Quantification 17 • Place instrument unit on a nearby surface and open lid. • Arrange collection attachments (hoses, bags, etc.) to capture the leak. • Monitor the display screen to initiate and record the measurement. • Remove attachments, close lid, move to next measurement location. To move the instrument, a shoulder strap may be fitted to the instrument and swung over one or both shoulders to free up hands for carrying other equipment; plastic case components are utilized to minimize the instrument weight. While not completed in the current prototype, case components may need to be spray-coated with a conductive metallic finish to improve grounding and reduce static discharge. Figure 6 provides a schematic cutaway view of the OS-HFS. Principal components are labelled. A cosmetic cover enclosing the electronics and display on the top shelf was not designed and is not shown. The prototype uses a Pelican™ Pelicase™ 1430 as a case. Figure 6: Cutaway schematic of the integrated instrument. Principal mechanical components are shown in Figure 6. An overview of the instrument operation is provided here with details of key components in the next section. Air flow path: In operation, the inlet is equipped with a hose-attachment fitting that rapidly con- nects to a hose. Multiple hose types can be used; the OS-HFS utilized the hose provided with CSU’s BHFS instrument. The selection of the length and type of hose influences the maximum flow rate through the instrument. In general, shorter hoses have less flow resistance, leading to Open-Source High Flow Sampler for Natural Gas Leak Quantification 18 higher air flow rate through the instrument, which is important to measure larger emissions. An inlet screen is also required, either as part of the hose attachment or at the inlet to the instrument. The blower draws in the emissions through the flow tube and orifice plate, and discharges into the sensor chamber. By placing the sensors directly in the flow path, no sample pumps are required to extract part of the flow stream for measurement, reducing the complexity of the instrument. The flow is directed out the flow exit. In operation, this port is equipped with a screen to keep foreign matter from the sensor chamber. In some cases, it may be useful to attach a discharge hose to direct the captured emissions away from the instrument and operator. This can be accomplished using the same fittings and hose type as the inlet. Increased flow resistance caused by the discharge hose will reduce the maximum flow rate through the instrument and the maximum emission rate that can be measured. Sensing: Total flow is measured by pressure drop across the orifice plate in the flow tube, using a differential pressure sensor. The flow tube dimensions provide sufficient stability for the orifice plate to measure the flow rate. Changes in these dimensions should be analyzed thoroughly. The size of the flow tube and blower practically set the horizontal dimension of the instrument, and all other components are positioned relative to the volumes of those components. The sensor chamber includes mounts for gas sensors. The prototype instrument used two sensors. More/fewer sensors could be installed in the same chamber with some re-engineering. Pressure fluctuations from turbulent air flow at the blower exit may affect sensor readings. Some sensors may need to be sheltered or placed in separate sensor chambers supplied by sample pumps. Many gas sensors also report temperature, pressure, and relative humidity which may be useful for ad- vanced calibration corrections and for translating measurements to standard conditions. As a general rule, the architecture favors the use of sensors with digital outputs. This approach eases the integration with other instrument electronics (see next section), and the integrated am- plification, A/D conversion, and correction reduces sensor noise issues presented by using separate Wheatstone bridges and A/D conversion. It does, however, limit the choice of sensors and the speed at which the sensor can be read. In practice, these limitations did not unduly restrict the design of the system and are highly recommended for derivative devices. Electronics: Electronics include an SBC that provides supervisory control, user interface, and data management for the instrument, and auxiliary electronic boards that integrate necessary compo- nents for integrating sensors and battery management. Enclosure: The enclosure uses a commercially available, water- and dust-tight, case. All compo- nents are inserted into the enclosure from the top of the unit. This restricts access to the components but maintains the integrity of the enclosure. Open-Source High Flow Sampler for Natural Gas Leak Quantification 19 The main battery, which powers the fan and several other components, is located on the bottom of the enclosure to keep the center of gravity low, helping stabilize the unit when stationary or carried. The SBC utilizes a separate battery and will continue to function when the main battery is dis- charged. Charging of the main battery occurs in situ, without removing it from the case, using appropriate connections. The auxiliary SBC battery charges from the main battery. In the proto- type, charging is done with pigtails inside the box, while a productized unit would likely have bulkhead connectors installed in the case. The airflow path is sealed, and sealed to the case, to minimize gas intrusion into the electronics area. A complete seal is, however, impractical. Evaluation of possible gas intrusion and ignition sources are part of the safety review required for any completed instrument. 3.1.2 Electronics Architecture This section discusses the design and interconnections of electronics components. See PCB files in the data archive for details. A primary goal of the electro-mechanical architecture was to allow easy replacement (or substitu- tion) of sensors and controls. The architecture utilizes an industry-standard SBC (Raspberry Pi™) and the Python programming language. This class of computer is inexpensive, powerful, relatively low power, and integrates well with other computational systems, including networking and Blue- tooth devices. The programming language is transportable to other, similar computer platforms, and the I/O ports of any similar SBC can be readily mapped to talk with the sensor systems via common serial port standards. Finally, this class of computer supports quality display screens, with backlights, that enable a modern user interface and make diagnostics, data upload, and field oper- ation easier than the 4-line LCD display utilized in the BHFS. Additionally, the move away from proprietary controllers supports fast upgrading of the instru- ment’s supervisory control components; SBCs evolve rapidly over the life of an instrument and building around a proprietary microcontroller may retard adaption of the instrument to evolving use patterns. The inability to upgrade the existing BHFS instrument makes time-series data collec- tion difficult, prevents effective measurement automation, and effectively prevents the integration of new or different sensors. As a key example of integration, many companies are moving toward integrated data tracking systems for their leak detection and repair (LDAR) programs. Automated integration of leak meas- urement to these tracking systems would reduce data errors, provide better tracking of detection and measurement, and reduce regulatory reporting burden. A complete redesign would be required to adapt a BHFS to this type of communication. In contrast, the OS-HFS’ SBC controller could be readily adapted to upload leak measurements to a server, using standard data interchange protocols. Figure 7 provides an overview of the electronics and sensing architecture of the OS-HFS. Open-Source High Flow Sampler for Natural Gas Leak Quantification 20 Figure 7: Electronics architecture for the prototype unit. Overview: Primary control is accomplished by an SBC. A Raspberry Pi was selected for the pro- totype, as it has sufficient general-purpose (GPIO) pins and runs a robust version of the Linux operating system. The SBC also drives the display screen, using an HDMI standard driver with a touch screen add-on. A GPIO header connector expands the GPIO pins for access to multiple other components. In general, each of these is on a separate board, and dispersed throughout the instrument to suitable locations for sensing or control. The GPIO break-out is a commercial product (Raspberry Pi Cob- bler from Adafruit*). In a production unit, the breakout would be mounted on an instrument-spe- cific PCB that would then fan out connections for other subsystems. For the OS-HFS, a plug-board was utilized, as shown in Figure 8. In the prototype, there is no cover over the electronics assem- blies; a production unit would have a sheet metal or conductive plastic cover. PCBs: Both sensors utilize a 5-pin mounting that required a custom PCB to securely hold the sensor and mount it in the sensor chamber. Both sensors performed A/D conversion, calibration corrections, and noise reduction internal to the sensors, and communicated results via a serial con- nection (see Figure 7 for the types of serial connection used). Therefore, the PCB provided only serial and power connections; no other active components were required. The PCB must also be * https://www.adafruit.com/product/914 Microcontroller (Raspberry Pi) Interconnection Board PiJuice Power Board Pressure Sensor Assembly Temperature & Humidity Sensor Sensor Carrier Board (SGX) Sensor Carrier Board (Nano) GPS Sensor 5V Regulator A/D Converter GPIO Header Connection G PIO Header Connection I2C Analog UART 1 UART 2 UART 3 12 V PWM On Board Speed Controller 12 V 12 V LiIon Battery 12 V Ch ar ge Po rt Blower https://www.adafruit.com/product/914 Open-Source High Flow Sampler for Natural Gas Leak Quantification 21 designed to mount securely to the sensor chamber and provide space for a gas-tight seal around the sensor. Figure 8: Photos of key electronics assemblies. Main photo shows the top of the prototype with key electronics assemblies visible. Inset photo shows the mounting PCB for the sensor elements. Blower speed control was accomplished by sending a pulse width modulation (PWM) signal to the blower’s on-board controller. This method simplifies electronics complexity, as no H-bridge or similar motor drive is required, and provides better noise suppression, as all switching occurs close to the fan. It is highly recommended that any production units identify fans with internal speed control functionality. The interconnection board primarily mapped incoming serial ports, or the outgoing PWM signal for the blower to the GPIO ports on the controller, although some level shifting from the Raspberry Pi’s 3.3 V to 5 V was implemented for some interface functions. GPIO mapping was accomplished in software. Data communications: With the exception of the pressure sensor, all sensors were selected to com- municate over serial ports. There are two advantages to this strategy. First, sensors with integrated A/D conversion, scaling, and calibration tend to be more stable and less susceptible to electrical noise in the instrument. Since these critical functions are inside the sensor module, this also sup- ports fast replacement of any failed assemblies. Second, sensors can be positioned as needed within Open-Source High Flow Sampler for Natural Gas Leak Quantification 22 the instrument, and do not need to be located on the main PCB. This eliminates supporting equip- ment, such as sample pumps for gas sensors, and reduces the amount of electrical shielding re- quired, although shielded connection wire is recommended. The pressure sensor (Honeywell HSCDRRN002NDAA5) used a commercially available A/D con- verter board (Adafruit carrier board* for the TI™ ADS1115 I2C A/D converter chip) to perform A/D conversion and communicate results back to the SBC using serial (I2C†) communications. I2C versions of this pressure sensor are available but were out of stock during prototype develop- ment. Since any differential pressure sensor will require tubing from the pressure ports in the flow tube to the pressure sensor, the sensor unit can be located on the same circuit board as the GPIO breakout. The GPS module provides both time and location to the SBC. Since data files are often processed using time stamps to combine multiple data streams, accurate timing is essential. The GPS module uses an auxiliary antenna mounted within the plastic instrument enclosure. If a metallic enclosure type is utilized in derivative units, some provision for an external antenna may be required. GPS may also be required to integrate the location of measurements into companies’ data tracking sys- tems. Communication speeds vary between sensors (see Section 3.2 for key sensors), but since none of the sensors needs to be sampled at more than a few Hz, serial communications are sufficiently fast to simultaneously read all sensors. Power management: All power was provided to the prototype using the 12 V lithium-ion battery. The battery was purchased with a charger. The blower is connected directly to the battery, and the instrument will not function without sufficient charge in the main battery. The SBC was powered from an onboard lithium-polymer battery (a PiJuice™ battery pack‡) via a 12 to 5 V regulator. The operational plan for the power system is: 1) Operator charges the main 12 V battery in the unit overnight and possibly during breaks, such as when driving from facility to facility. 2) The PiJuice pack is charged from the main battery and requires no direct attention from the user. * https://www.adafruit.com/product/1085?gclid=Cj0KCQiAmKiQBhClARIsAKtSj-mfjMTuiiBfQe9myI45V6en- MEQ9GxW7dEGcl1MfewyX5R6_fjnoFu0aAqAXEALw_wcB † I2C or 𝐼𝐼2𝐶𝐶 is a commonly utilized serial communications protocol that can accommodate multiple slave components on one serial connection. See, for example, https://en.wikipedia.org/wiki/I%C2%B2C ‡ https://uk.pi-supply.com/products/pijuice-standard?_pos=21&_sid=4a1f56936&_ss=r https://www.adafruit.com/product/1085?gclid=Cj0KCQiAmKiQBhClARIsAKtSj-mfjMTuiiBfQe9myI45V6enMEQ9GxW7dEGcl1MfewyX5R6_fjnoFu0aAqAXEALw_wcB https://www.adafruit.com/product/1085?gclid=Cj0KCQiAmKiQBhClARIsAKtSj-mfjMTuiiBfQe9myI45V6enMEQ9GxW7dEGcl1MfewyX5R6_fjnoFu0aAqAXEALw_wcB https://en.wikipedia.org/wiki/I%C2%B2C https://uk.pi-supply.com/products/pijuice-standard?_pos=21&_sid=4a1f56936&_ss=r Open-Source High Flow Sampler for Natural Gas Leak Quantification 23 3) The SBC remains powered when the main battery is disconnected or discharged, for 1-3 hours, allowing data upload and other maintenance operations. This also prevents loss of settings, etc. if the main battery runs low during use. 3.1.3 Software Architecture The software running the OS-HFS is written in Python, a platform-independent interpreted lan- guage. GUI elements are currently implemented using PyQt5, but the modular software approach allows for easy substitution of other GUI libraries. The software architecture closely resembles the hardware architecture as it is implemented using an object-oriented approach (see Figure 9). “Reader” classes communicate directly with each sensor or peripheral device based on manufac- turers’ datasheets and any undocumented behaviors observed during testing. Each reader class maps appropriate lower-level commands and responses specific to each sensor into a common command/response queue for use in other instrument functions. A “Hi Flow” class orchestrates the critical functions of the OS-HFS by issuing commands, receiving responses, performing cal- culations, and updating the GUI. This class also handles file I/O and measurement data storage. The current prototype software has been supplied to CARB and is identified in Appendix A. Since the OS-HFS is implemented on an SBC with a full Linux operating system rather than a microcontroller, display and device interfacing is streamlined, and networking capabilities are in- trinsic to the SBC’s operating system, including Wi-Fi, Ethernet, and Bluetooth. The current im- plementation is built as a desktop application, but the SBC could easily serve the application on a web interface where it would be available to any mobile device with Wi-Fi and a web browser. Open-Source High Flow Sampler for Natural Gas Leak Quantification 24 Figure 9: Software architecture schematic. Custom Python classes interact with native libraries or with physical sensors directly. The software architecture follows the modular approach of the physical architec- ture. Open-Source High Flow Sampler for Natural Gas Leak Quantification 25 The prototype improves the user experience by providing detailed information to the user on a color touch screen display. From the display, the user can view all instrument parameters, calibrate sensors, and perform system checks. The user can also save instrument data elec- tronically using an automated measurement cy- cle or start and stop a “manual” measurement that captures time series data without interven- tion from the instrument. At the end of each measurement, leak rates and summary data are provided along with data quality indicators. By using an SBC, data can be logged to standard file formats, and accessed using file transfer and connectivity with other devices. This also opens the possibility of real-time data transfer by pairing the OS-HFS with auxiliary devices over Wi-Fi or Bluetooth®. A screenshot of the OS-HFS GUI is shown in Figure 10. Time and location data are always shown in the header box at the top of the screen. This data is supplied directly by the GPS unit, and is thus always correct, with minimal offset from the true time. Several tabs provide access to the various functions needed to operate the device including direct access to the instrument’s sen- sors (named the “INIR” and “Nano” throughout this report; see Section 3.2.3) to perform zero, span, and offset calibrations. General device parameters are displayed as well, and other command- based interactions are available. The measurement tab (“Meas”, Figure 10) is of most interest during measurement activities. Here, the operator can enable or disable the blower and change the setpoint of the blower. The current PWM setting along with blower speed in rpm are shown. The measured sample flow is also shown along with the concentration measured by each sensor, and the gas molecular weight classification provided by the Nano sensor. Finally, the leak flowrate is shown, and plotted on a time series graph. Stability of concentrations shown on this graph provides an invaluable tool for experienced HFS practitioners. Figure 10: Example GUI tab as shown on heads-up display or a virtual desktop accessible by a Smartphone, table, or remote PC. The screenshot was taken during a debugging session, not during an actual measurement. Open-Source High Flow Sampler for Natural Gas Leak Quantification 26 Auto, Manual, and Stop buttons are used to save data to a delimited text file that includes all available input parameters and the resultant calculations. The Auto button starts a timed measure- ment cycle that automatically stops and saves a measurement file when completed. The Auto mode implements a typical fugitive measurement, and the timer is set to a two-minute- long measurement period by default. Future extensions to the software could automate more of the measurement process following the outline in section Appendix B including automatic flowrate adjustment. This type of implementation could potentially reduce training requirements for oper- ators. The Manual button allows the operator to run a longer measurement but requires the Stop button to be pressed to conclude the measurement. A default timeout can be set to intervene if a measure- ment is not stopped manually. This mode is intended for “long-term” measurements, for example, to estimate emissions from time varying sources. A representative sample time can be used at the operator’s discretion up to the length of the battery life of the unit. All leak flow data are shown on the time series graph. During an active measurement (either Auto or Manual) the data points are shown in yellow to differentiate them from idle points on the graph that were collected prior to the start of the measurement, and to indicate that a measurement is underway. In the version 1 prototype the INIR sensor was considered the main sensor and the estimated emission rate was calculated based on its output, as can be seen in Figure 10. Note: The project team made every attempt to utilize only open-source software libraries for the prototype unit. However, any party adopting the OS-HFS design should check all libraries utilized to ensure that their use is compatible with the licensing of the libraries. 3.2 Selection of Key Components 3.2.1 Blower The user manual for the BHFS19 states that the device can produce a maximum of ~10 SCFM (283 LPM) at initial flow with full battery. The measurable leak rate is listed as 0.05 to 8.0 SCFM (1.42 to 226 LPM), though the device will display readings down to 0.1 LPM when the display is in the LPM mode setting. Testing shows that while the BHFS display reports readings below the stated lower limit, the accuracy is questionable, and it may be that the stated 1.42 LPM lower limit is that which still meets the stated system accuracy (± 10 % by volume leak rate reading, methane). Pre- liminary calculations made using various blower and sensor combinations under consideration in- dicated that 0.1 SLPM was a reasonable target for a lower usable limit in the OS-HFS. Similarly, the target upper limit for the OS-HFS design was set at 12-15 SCFM (336-420 SLPM). This target would expand the BHFS capabilities by ~50 % while still allowing the prototype to be a more compact device. Open-Source High Flow Sampler for Natural Gas Leak Quantification 27 The BHFS uses the 3M™ GVP-100 Motor Blower, to create the flow stream which draws the emission and excess air into the device. The GVP-100 is a component of a Powered Air Purifying Respirator system marketed by 3M that consists of the blower, a power cord, and a battery pack. The respirator system has been classified as intrinsically safe for use in hazardous locations. Per- formance curves for the GVP-100 blower did not appear to be readily available, so a test rig was created to develop a fan performance curve as the basis for identifying suitable replacements. A spare (non-working) BHFS unit was provided to the project in kind by Heath Consultants Inc. The blower was removed and tested by applying full power and gradually blocking the outlet and mon- itoring the static pressure produced as the flow rate decreased. This procedure is a simplified ver- sion of the formal test procedure used to characterize fans and blowers in industry32. The resulting fan curve is shown in Figure 11. Similar fan curves can be obtained from fan manufacturers; other units under consideration are shown in Figure 11, for comparison. Figure 11: Fan curve comparison for the GVP-100 blower used in the BHFS vs. other models under consid- eration from San Ace and Delta Fan. The yellow trace (San Ace (fan)) shows the performance of an axial fan, for comparison. The performance curves for the GVP-100, Delta, and San Ace are typical of centrifugal blowers. When selecting a fan for a given application, the fan should be chosen such that it is generally operated below the stall region (the region where, starting from maximum flow and following the curve, the curve begins to flatten). Thus, the desired flow rates should be available when the needed static pressure is at or below the stall region. In Figure 11, the curve for an axial fan (San Ace (fan)) is shown alongside performance curves for centrifugal blowers – two common fan designs used in electronics cooling applications. The shape of the GVP-100 curve indicates that it is likely a centrifugal blower, which was not obvious from observing the design of its housing. For a similar flowrate, package size, and power consumption, centrifugal blowers are generally able to produce a higher static pressure at a given flowrate compared to axial fans. Observing the yellow curve (San Ace (fan)) the axial fan shows a more drastic reduction in flow with increasing static pressure. Thus, centrifugal blowers should provide a more stable flow stream that is less sensitive to changes in back pressure. This characteristic may be advantageous during field measurements where a wide Open-Source High Flow Sampler for Natural Gas Leak Quantification 28 variety of measurement situations will be encountered and will likely require a variety of attach- ments to collect the emissions. Special consideration must be given to ensure that the blower will be safe for use in direct contact with gas/air mixtures that may be within flammability limits. This automatically points to brushless motors suitable or certified for use in hazardous locations (e.g., UL Class/Division, ATEX, Haz- Loc, IECEx, etc.). To better understand the characteristics of an intrinsically safe blower, first the GVP-100 from the spare BHFS was disassembled to study its construction. As shown in Figure 12, the blower was constructed primarily of plastic with a metal shaft and small electric motor. All the electrical connections directly in contact with the flow stream were potted (encased) in epoxy, and the inside surfaces in contact with the flow stream were sealed from the outside. The blower was a centrifugal design, in line with results of the fan curve testing. A search for DC centrifugal blowers certified for use in hazardous locations did not identify any units available commercially off the shelf. Conversa- tions with several manufacturers revealed one com- pany willing to make them on contract with a speci- fied minimum order and production period, along with fees for engineering services and certification. Two other manufacturers who recently started pro- ducing ATEX rated fans were also identified. Both manufacturers were producing axial fans for use with AC mains power. When asked about the availability of DC centrifugal blowers with ATEX certification, both manufacturers stated that the intended applica- tion for the ATEX axial fans was in new commercial refrigeration units that use hydrocarbons as the work- ing fluid. They stated that a DC centrifugal blower that was certified was not likely to be produced unless there was a similar new, widespread market demand. The ATEX rated axial fans from both manufacturers had similar part numbers designating non-ATEX units with otherwise identical specifications. The non-ATEX units were IP68 rated against water and dust ingress and featured encapsulated electrical connections and non-sparking materials, much like the GVP-100. To compare rated and non-rated units, one of each model of fan was acquired and disassembled for study, revealing little, if any, difference between the two units. The only difference observed was so minor, that it may be attributable to manufacturing variance. The manufacturer declined to comment on this matter. Based upon this result, the project team made two assumptions. First, the DC units were likely not ATEX certified due to lack of demand, and second, the required engineering to make an ATEX- Figure 12: Inside the 3M GVP-100 blower mo- tor used in the BHFS. The blower is certified for use in hazardous areas. All the internal electri- cal connections are encapsulated. The internal housing (upper left) is characteristic of a cen- trifugal blower design. Open-Source High Flow Sampler for Natural Gas Leak Quantification 29 compliant unit was likely small. The team chose the San Ace 97W, an IP68 rated, DC powered centrifugal blower for use in the prototype. Production units or subsequent prototypes may require the use of other blowers or contracting with the manufacturer to develop the correct certifications. Two units were acquired, and one was disassembled for study. The project team found encapsu- lated electrical connections and components, with components made of aluminum, stainless steel, brass, and plastic. However, the housing was not gas tight and had vents or drain holes in the housing. For use in the prototype, these were sealed with silicon adhesive to completely contain the flow stream within the housing. 3.2.2 Flow Measurement An orifice plate flowmeter is a standardized, well-documented flow measurement technique with a simple design and low component count that is relatively easy to implement. An orifice plate flowmeter works by measuring the pressure drop caused by the increase in flow velocity in the presence of a sharp-edged flow restriction within a closed pipe. Several documents describe the design considerations and uncertainties associated with orifice plate meters, including AGA Re- port 3.133 and ISO Document 516734. An orifice meter was also used in the BHFS. Flowrate meas- urements reported by the BHFS were compared to a recently calibrated laminar flow element over a range of flowrates between 100 and 250 LPM and found to agree within 5 %, as specified in the BHFS manual. Given the performance of the BHFS flow measurement and the ease of implemen- tation, an orifice plate meter was selected for use in the OS-HFS. The flowrate and associated uncertainty reported by an orifice plate flowmeter constructed and implemented according to the ISO and AGA standards should be well-characterized; measure- ments made without additional calibration should fall within the specified tolerances, by design. Often, a meter calibration is used to account for any unavoidable deviations and to characterize a specific meter, in situ. The resulting “meter factor” characterizes the deviations of the meter under calibration relative to flow through a reference instrument. The meter factor is then applied to the in-use readings to align them with the reference standard. The OS-HFS uses this type of independ- ent calibration and correction. The ISO 5167 standard offers detailed guidance on the construction of an orifice plate flowmeter. However, the guide does not cover orifice plates for use in pipes less than 50 mm in diameter or at Reynolds numbers less than 5000. Reynolds numbers calculated for OS-HFS target flow rates generally exceeded 5000 in a 2 in diameter pipe, making this a suitable starting point for the design. The 2 in pipe also supported the desired increased flowrate relative to the BHFS, while supporting a similar or smaller package with similar power requirements. The practical accuracy achievable with an orifice plate constructed in accordance with the stand- ards will be limited by the accuracy of the differential pressure measurement, and any environ- mental factors present that influence that accuracy33. The accuracy can also be limited by the un- certainty in the discharge coefficient, which is determined experimentally or from empirical Open-Source High Flow Sampler for Natural Gas Leak Quantification 30 equations based on large numbers of experiments. The ratio of the uncertainty in the empirical discharge coefficient (𝐶𝐶𝑑𝑑) at a given Reynolds number (Re) to the uncertainty in 𝐶𝐶𝑑𝑑 at an infinite Re is minimized when the diameter ratio β is ~0.56; these dimensions were utilized for the OS- HFS. During the fan flow curve testing for the GVP-100 it was observed that the maximum static pres- sure produced was ~1.5 in H2O. The original BHFS was equipped with a differential pressure (Δ𝑃𝑃) sensor with a ~15 in H20 range (MPX5004DP). The fan curves published by the blower manufac- turers showed a maximum static pressure of 1.5 in H2O at zero flow. This meant that the full range of the sensor used in the BHFS would not be needed in practice, and that the resolution of the instrument could be improved by substituting a sensor with a reduced range. The OS-HFS uses a sensor with a range of ±2 in H2O to improve resolution, which covers expected pressure differen- tials but protects the sensor from over pressure in the event of a complete exhaust blockage. The chosen sensor has an accuracy of ± 0.25 % (versus ± 1.5 % for the BHFS’ sensor). The combination of restricted range and better accuracy improves the resolution and the overall accuracy of OS- HFS’ flow measurement relative to that of the BHFS. OS-HFS flow measurement was calibrated using tracer gas dilution in accordance with ASTM designation E2029-1135. CP grade methane (Airgas, Inc.) was metered at a constant flow of 13 SLPM* through a critical orifice and introduced at the inlet of the OS-HFS sample hose. Upstream pressure and temperature were monitored for consistency, and the downstream flow was verified with a volumetric prover (Bios DryCal) reporting in standard units. A methane analyzer (QC- TLDS, Aerodyne Research Inc.) measured the methane at the exit of the OS-HFS by drawing a sample from an inlet placed in the center of the OS-HFS sensor chamber. The methane analyzer measurement range is 0-200 ppm methane with a 1-second precision of 0.3 ppb. During testing, the PWM controller of the OS-HFS fan was used to vary the setpoint between 20 % and 100 % duty. The methane tracer gas flowrate was held constant. methane readings ranged between 30 and 60 ppm during testing; background methane concentrations (in ambient air) were taken between each PWM setpoint to calculate the enhancement due to tracer gas flow. This calibration procedure was performed twice. Both tests were performed in Fort Collins, CO at an elevation 5,003 ft above sea level at a station pressure of ~12.2 psia. The first test was performed on 7/9/2021, at an ambient temperature of 27 °C using a 2 in diameter antistatic vacuum hose on the OS-HFS inlet. Results are compared in Figure 13, first for a constant 𝐶𝐶𝑑𝑑 based on the empirical formulation found in the standards, and second with a meter-specific 𝐶𝐶𝑑𝑑 developed as a function of Δ𝑃𝑃 – i.e., by solving for the unknown 𝐶𝐶𝑑𝑑 at each test point. For the parameters of the OS-HFS design, the empirical 𝐶𝐶𝑑𝑑 was 0.637, while the meter specific 𝐶𝐶𝑑𝑑 varies between 0.674 and 0.656 for Re ranging from 6.8x103 to 12.6 x103. The maximum Δ𝑃𝑃 achieved during testing was 0.72 in * Standard conditions are 25 °C / 1 atmosphere throughout. Open-Source High Flow Sampler for Natural Gas Leak Quantification 31 H2O. In this configuration the OS-HFS was able to achieve a maximum flowrate of 442 SLPM, which was well aligned with the target design goal. In subsequent testing, the antistatic vacuum hose used during the first test proved too stiff for practical use when tested at CSU’s METEC fa- cility. Therefore, the OS-HFS was modified to use the hose from the BHFS. A second calibra- tion was conducted using the BHFS hose on 12/19/2021 with an average ambient tempera- ture of 19 °C. In this configuration the OS-HFS was able to achieve a maximum flowrate of 316 SLPM, as shown in Figure 13. In this configu- ration the 𝐶𝐶𝑑𝑑 varied from 0.677 and 0.651 for Re ranging from 4.6x103 to 9.8x103. The maximum Δ𝑃𝑃 developed was 0.36 in H2O. As can be seen in both Figure 13 and Figure 14, the empirical discharge coefficient differs from those ob- tained specifically for the OS-HFS by direct cal- ibration. In both cases 𝐶𝐶𝑑𝑑 varied with flow (and thus Δ𝑃𝑃) and results were improved by apply- ing 𝐶𝐶𝑑𝑑 as a function of Δ𝑃𝑃. In each case a sin- gle 𝐶𝐶𝑑𝑑 of 0.667 would likely be suitable over the flow ranges tested and would not degrade the accuracy greatly. 3.2.3 Concentration Measurement The BHFS used a pellistor-based sensor (e.g., SGX Sensortech VQ549ZD series) which was operated in both catalytic oxidation and ther- mal conductivity modes. The sensor is not me- thane-specific, and its output varies with gas composition. The catalytic nature of the sen- sor makes it vulnerable to both saturation and poisoning by other gas species that may be en- countered in field use (e.g., higher hydrocar- bons, oil mist, H2S, etc.). The precise implementation of the sensor within the BHFS is a trade secret, but many have ob- served a “transition behavior” around the 5 % concentration, which is generally assumed to be the Figure 13: First OS-HFS flow meter calibration performed in July 2021 using a 2 in diameter anti- static vacuum hose. Figure 14: Second OS-HFS flow meter calibration performed on 12/19/2021 using the inlet hose from the BHFS. Open-Source High Flow Sampler for Natural Gas Leak Quantification 32 transition between catalytic oxidation mode and thermal conductivity mode, or the transition be- tween calibration curves. This transition may be noted by observing the number of digits displayed on the BHFS screen; no other indication is displayed. The goal of selecting sensors for the OS-HFS was to (a) alleviate identified sensor issues, (b) modernize communication between the SBC and sensors, (c) mitigate other use issues while still selecting readily available, moderate cost, sensors. A market survey indicated that component-type sensors and modules targeted at industrial applications met these requirements, as shown in Figure 15. Figure 15: Sensor selection strategy for OS-HFS development. Avoid low-cost sensors with poor accuracy, and avoid high-cost sensors with great accuracy, but increased size. Focus on compact, readily available industrial sensors with reasonable accuracy and affordability. Two sensors stood out based on manufacturer specifications, availability, and cost (Figure 16): 1. SGX INIR-ME100*, a non-dispersive infrared (NDIR) sensor. Key datasheet claims include a “Triple Range” (0-1 %v, 1-4 %v, >4 %v) operation, a resolution of up to 10 ppm, and an LDL of 100 ppm. 2. Nevado Nano MPS flammable gas sensor†, which the manufacture calls a “molecular prop- erty spectrometer” with some ability to differentiate between gas species (or mixtures) based on average molecular weight. Key datasheet claims include a methane detection * https://sgx.cdistore.com/products/detail/inirme100-sgx-sensortech/578557/ † https://nevadanano.com/mps-flammable-gas-sensor/ https://sgx.cdistore.com/products/detail/inirme100-sgx-sensortech/578557/ https://nevadanano.com/mps-flammable-gas-sensor/ Open-Source High Flow Sampler for Natural Gas Leak Quantification 33 range of 0-100 % LEL, a resolution of 0.1 % LEL, an accuracy of ± 3 % of LEL from 0- 50 % LEL. In operation the sensor could also report by % volume (as opposed to % LEL) and was used in that mode in the OS-HFS. Both sensors are integrated packages with onboard diagnostics and control, requiring only power and communication connections. The initial design intent was to use the wider range and higher resolution of the INIR sensor as the primary sensor, and the Nano as a secondary sensor to provide corrections or additional calibrations based on the type of gas being measured. For example, the Nano can distinguish between 6 classes of gases or gas mixtures. This would enable switching between a family of built-in calibrations for the INIR sensor to improve the accuracy of measure- ments made at various locations in the natural gas supply chain with varying compositions. Figure 16: The SGX INIR-ME100 and Nevada Nano MPS Flammable Gas Sensor were chosen for use in the prototype OS-HFS. To understand the behavior and performance of the sensors, a sensor test rig was developed to hold 2 each of INIR and Nano sensors as shown Figure 17. At the time of testing, it was discovered that the INIR sensors were using different firmware versions (2.25 and 2.33); users of the OS-HFS design may wish to control this variable, as the performance and interfacing may differ between firmware revisions. Testing characterized each sensor's performance over a range of gas concen- trations and varying atmospheric pressures. The INIR was found to have a consistent, non-negli- gible pressure dependency, as indicated in the datasheet. The Nano was found to exhibit much lower variation with pressure. Open-Source High Flow Sampler for Natural Gas Leak Quantification 34 Figure 17: Test chamber for prototype sensor calibration and characterization. During testing, methane/air mixtures were generated using choked flow orifices (O’Keefe) and calibrated volumetric provers (Bios DryCal). Gas mixtures were introduced into the sample cham- ber between 0.5 and 1 SLPM. A back pressure regulator was used to monitor the pressure within the chamber. Pressures were varied between 85 kPa and 100 kPa for each gas concentration blend introduced. Tests were performed with nominal concentrations of 200, 500, 1000, 2500 and 5000 ppm, 1, 1.5, 2, 3, 5, 10, 15, 20, 25, 35, 50, 60 and 80 % methane by volume in “shop-air” which had been filtered and dried by typical coalescing and particulate filters. Prior to each test the INIR was zeroed and spanned according to the manufacturer’s guidelines, with the chamber held at 90 kPa. The Nano does not require user calibration and is factory cali- brated for five years. During testing it was observed that the Nano appears to “self-zero” during power up, and the presence of gas or ambient pressure can affect the output depending on the conditions present at power on. As a result, care must be taken to present only background air concentrations to the sensor at power-up. An example time series from one of these tests is shown in Figure 18. Open-Source High Flow Sampler for Natural Gas Leak Quantification 35 Figure 18: Example data capture from sensor characterization. The INIR sensor output varies with ambient pressure to a greater degree than the Nano sensor. Briefly, a methane-air mixture at 1.5 % by volume was introduced into the sample chamber at 85 kPa until the sensor readings stabilized. Once stabilized, data were recorded for 5 minutes. Next, the pressure was changed to 90 kPa and the readings were allowed to stabilize before recording again. This procedure was repeated at 95, 100, and again at 85 kPa for a variety of gas concentra- tions. This testing took place over the course of several days during February and March of 2021. It is evident from the testing that the INIR response is linear over the range of concentrations tested, and that there is a strong dependence on pressure, as shown in Figure 19. The readings are consistent at the calibration pressure of 90 kPa, 10 % low at 85 kPa, and 10 % high at 95 kPa. Therefore, the INIR sensor requires a calibration and correction for the prevailing atmospheric pressure. The INIR sensor was tested both with and without an offset calibration point. As the manual states, this step does not appear to be needed with the current firmware and does not seem to offer a significant improvement in performance. Open-Source High Flow Sampler for Natural Gas Leak Quantification 36 Figure 19: The INIR sensor shows a strong dependence on ambient pressure. A 5 kPa change from the 90 kPa baseline used for calibration results in a 10 % change over the tested range. The Nano response to changes in pressure is less pronounced than the INIR, reading 1% lower than the 90 kPa baseline at 85 kPa, and 2 % higher at 95 kPa. The Nano under-reports the concen- tration present considering the full-scale results, as shown in Figure 20. It was also noted that the Nano reports 0 % concentration when the actual concentration falls below 1500 ppm and reports readings when the actual concentration is above 1500 ppm. Figure 20: The Nano sensor output shows less pressure dependence but seems to consistently underesti- mate the gas concentration. 3.2.4 Background Concentration Measurement The use of a background sensor is intended to account for any elevated gas concentrations in the ambient dilution air drawn into the HFS instrument along with the leak source being measured. Typically, the background gas concentration is subtracted from the measured gas concentration within the instrument to remove its contribution to the reported emission rate. This situation may occur, for example, on a small leak within an enclosed building with poor ventilation. If the Open-Source High Flow Sampler for Natural Gas Leak Quantification 37 operator fully captures the leak source, the contaminated air being drawn in along with it will induce an additional sensor response, artificially inflating the reported measurement. A back- ground sensor on the instrument could subtract the background contribution and report the true emission rate from the captured leak. This example assumes that the background sensor is meas- uring a truly elevated background, and not a parcel of air containing remnants of the emission source, other nearby emission sources, or the exhaust from the HFS itself. More typically, an in- complete capture is picked up by the background sensor, further lowering the reported emission rate from the source. A background sensor located near the sample inlet could be useful as an indicator of incomplete capture but would be difficult include properly within the measurement strategy in all possible scenarios. For this reason and the fact that HFS quantifications are normally coupled with a pri- mary means of detection, such as a handheld sensor or OGI camera, the project team decided to eliminate this feature from the prototype OS-HFS. 3.3 Accuracy Analysis Once the prototype OS-HFS was assembled, several rounds of testing were conducted to quantify its performance and identify improvements needed. Tests were aimed at quantifying the lower detection limit, evaluating overall quantification accuracy, determining effects of varying gas com- positions, and conducting practical use testing on “real equipment” at CSU’s METEC facility. For each of the tests, readings from each of the two onboard sensors (INIR and Nano) were used to calculate results independently rather than using the Nano as an auxiliary sensor as originally planned, although use of the Nano as an auxiliary gas-type sensor could be useful with further development. Several follow up tests were performed in an attempt to identify the biases observed in the initial testing. Testing methods are described in more detail in Section 4. Open-Source High Flow Sampler for Natural Gas Leak Quantification 38 3.3.1 Lower Detection Limit Figure 21: Prototype OS-HFS INIR sensor lower detection limit. Figure 22: Prototype OS-HFS Nano sensor lower detection limit. Lower detection limits (LDLs) were tested with the OS-HFS sample flow set at ~150 SLPM. Sev- eral data points were tested near the expected LDL and a logistic regression was performed on detect vs non-detect of the source emission. It should be noted that LDL as reported here only indicates the lowest test gas flowrate at which the HFS reported a non-zero emission rate and says nothing about the accuracy of the reported value. Based on the sensor data sheets, the INIR has a stated LDL of 100 ppm, roughly equivalent to an LDL of 0.015 SLPM. However, during testing these sensors did not detect concentrations with meaningful accuracy much below 1,000 ppm, Open-Source High Flow Sampler for Natural Gas Leak Quantification 39 leading to an expected LDL of 0.15 SLPM. The LDL observed in system testing was close to this figure: 0.178 SLPM. The Nano sensor’s datasheet states a detection range of 0-100 % of LEL (5 % vol/vol for methane) with an accuracy of ± 3 % of LEL. Interestingly, during testing it was observed that the Nano sensor would report 0 once the test concentration fell below 1,500 ppm and would resume reporting values once the test concentration was above 1,500 ppm. This would lead to an expected LDL of 0.225 SLPM, and the observed LDL was somewhat higher at 0.370 SLPM. 3.3.2 Quantification Accuracy Figure 23: Prototype OS-HFS INIR sensor quantification accuracy. Open-Source High Flow Sampler for Natural Gas Leak Quantification 40 Figure 24: Prototype OS-HFS Nano sensor quantification accuracy. Quantification accuracy tests were performed across the usable range of the prototype OS-HFS. Each emission rate was measured at two flow settings: 1) Emission rates lower than 30 SLPM used flow rates of 175 and 275 SLPM to improve sensitivity. 2) Emission rates greater than 30 SLPM used flow rates of 275 and 300 SLPM to ensure complete capture. Open-Source High Flow Sampler for Natural Gas Leak Quantification 41 Four separate experimental setups were employed to generate “true leak rate” reference flows as a basis for comparison: 1) For the low test points (also included in LDL testing) leaks were generated using a critical orifice which was checked with a volumetric prover (Bios Drycal) before and after each test. 2) A dilute gas standard (2.5 % methane, balance air, Airgas Inc.) delivered by a mass flow controller (Alicat, Inc). 3) For the mid test points, 100 % methane (ME CP300, Airgas Inc) was delivered using a mass flow controller (Alicat, Inc). 4) For the high test points utility-delivered natural gas was metered using a thermal mass flow meter (Omega Inc). Accuracy of both sensors was relatively poorly at low emission rates, although it should be noted that once the Nano began reporting values, they were very well correlated with the true leak rate. The Nano output switching behavior discussed in the LDL section can be seen in Figure 24 (top right panel). Both sensors produced well-correlated linear responses in the mid test points, with good repeata- bility, though both were biased low. This result was unexpected based on the results of the sensor testing and flowrate testing and will be investigated further. The INIR sensor became saturated for two of the three high test points and reported a 93 % leak concentration when the true leak concentration was 25-40 % for the three high test points. The high test points were conducted using utility-delivered gas with 85 % methane, 9 % ethane, and 1 % propane, while the sensor was calibrated using 100 % methane. This result highlights im- portance of considering gas composition, and the need for a built-in error checking to help alleviate similar issues in field deployed HFSs. The Nano sensor showed linear and repeatable results for the high range testing, although was biased slightly lower than for the mid-range testing. Two additional accuracy tests were performed in December 2021 and February 2022 immediately following calibration of the INIR sensor, as shown in Figure 25. The results indicate that the INIR sensor had drifted since the prior accuracy testing was performed, and that the Nano had remained stable. Again, these tests showed a low bias in predicted leak rate from both sensors across the range of testing. The cause for this low bias remains unclear. Taken as a whole, accuracy testing indicates that any productization of the OS-HFS design will require additional testing, and possibly control/calibration to maximize the accuracy of these sen- sors. While performance fell below expectations from earlier sensor-only testing, results indicate that: • This class of sensor is capable of supporting an HFS instrument. Open-Source High Flow Sampler for Natural Gas Leak Quantification 42 • Use of multiple sensors, using different sensing mechanisms, shows promise to identify errors, gas composition problems, or other accuracy issues that would not be caught by a unit using only one sensor or even multiple sensors of one type. • Use of sensors with built-in A/D conversion and serial communication works well. Figure 25: Additional Prototype OS-HFS accuracy testing results. 3.3.3 Gas Composition Testing Figure 26: Prototype OS-HFS INIR sensor gas composition testing results. Open-Source High Flow Sampler for Natural Gas Leak Quantification 43 Figure 27: Prototype OS-HFS Nano sensor gas composition testing results. For gas composition testing, 21 gas blends were generated with varying proportions of methane, ethane, and propane. For the OS-HFS, all blends were delivered at 20 SLPM. Three replicate tests were performed for each blend, varying the flow setpoint using nominal flows of 180, 240, and 280 SLPM. The INIR sensor was calibrated with methane prior to testing and reported elevated emission rates for all blends with even a small amount of non-methane hydrocarbons (NMHC) present in the gas. The response increased dramatically with increasing NMHC content, illustrating the need for cal- ibration on the gas under test for this type of sensor to avoid erroneous results and possible sensor saturation. The dynamic range of the sensor should be tested when gas blends with higher levels of NMHC are being measured to understand if, and at what level, the sensor becomes saturated. The Nano sensor can differentiate between gas species, or blends of gas species, based on average molecular weights. For gas blend tests with a methane content less than ~50 % the Nano reported the presence of a “Light Gas” which is defined as a gas blend with an average molecular weight between 25-75 g/mol and a density between 1.2-2.5 kg/m3, with a typical number of carbons be- tween 1-4. For gas blend tests with a methane content greater than 50 %, the Nano reported the presence of “Methane” which is applicable to methane and light natural gas mixtures with an av- erage molecular weight of 16-19 g/mol, average density of 0.6-0.9 kg/m3, and typical number of carbons between 1 and 2. In this testing, gas test blends with an average molecular weight of 17- 25 g/mol were reported as “Methane”, and gas test blends from 25-33 g/mol were reported as a “Light Gas”. Over the range of gas test blends considered “Methane” the Nano, the output of the INIR sensor varied by a factor of five. When “Light Gas” was reported by the Nano, the INIR output varied by a factor of 2. This result indicates that an immediate accuracy improvement in reported emission Open-Source High Flow Sampler for Natural Gas Leak Quantification 44 rate is not possible by applying a correction factor to the INIR data based on the gas classification provided by the Nano. Calibrating the INIR sensor on a gas other than methane would likely impact the results, although how much is not clear. 3.3.4 Practical Use Testing Figure 28: Prototype OS-HFS INIR sensor practical use testing. Figure 29: Prototype OS-HFS Nano sensor practical use testing. Open-Source High Flow Sampler for Natural Gas Leak Quantification 45 For practical use testing, the prototype HFS was deployed in a mock field trial at CSU’s METEC facility where its performance was evaluated against known emission rates on “real equipment” intended to closely mimic the challenges encountered in actual field use. Tests points with nominal true leak rates of 35 SLPM were tested on in early October 2021 (2S-11, 2T-11,1S-11, 1W-15, 1S- 15) while the remaining emission points were tested in November 2021 while testing other pre- production HFS units. For the INIR sensors, all of the emission rates were overestimated due to the fact that the emissions were natural gas while the sensors were calibrated on methane. For both sensors, the October test data shows a wide range of variability which may indicate that the challenges associated with field measurements (i.e., method accuracies) have a much larger influence on overall results than the instrument accuracy of the HFS itself. For the Nano sensor, all test results were biased low, though the overall bias was influenced by the October tests which showed a greater low bias than the November tests. These results further reiterate the importance of sensor calibration for the INIR sensor, but also indicate that performance from the Nano can be reasonable. Open-Source High Flow Sampler for Natural Gas Leak Quantification 46 4. Testing of Pre-production Prototype HFS Units In addition to the prototype constructed during this project, three commercial pre-production pro- totype units were also tested. The complete test protocol is provided in Appendix C, and a photo of the units is in Figure 30. Note: All pre-production units tested here were under development at time of test- ing. All test results should be considered preliminary, and readers are encouraged to consult the developers of these instruments for the most current performance data. Device descriptions are included as provided by manufacturers and specific claims have not been verified by CSU or CARB. Four test sets were performed. The first three test sets (Sections 4.1-4.3) were performed by intro- ducing the emission directly into the instrument by inserting the outlet of the gas release equipment into the inlet of the instrument. This approach ensures that all emissions were captured by the instrument. The final test (Section 4.4) measured representative leaks on real equipment. The final test was per- formed on modified field equipment at METEC at CSU. Five leak locations were captured and measured. For all tests, instrument readings were recorded from the screen of the instru- ment. Data recordings produced by the instrument were reviewed but were not extensively analyzed. Code used to characterize the CSU prototype was also used to generate the analysis and charts seen below. The following instruments were tested: • Bacharach® – HI FLOW® Sampler: (“BHFS”, herein): This instrument was included for reference, as it is the most commonly utilized high flow instrument. Manufactured by Bacharach, it was recently discontinued and is no longer manufactured. • Sensors, Inc. – SEMTECH® HI-FLOW 2: (“Sensors”, herein) Sensors, Inc., 6812 State Road Saline, MI, 48176 USA Figure 30: HFS units at METEC. Left to right: Sensors, Inc. (gas sensor module, handheld unit attached by cord), Bacharach HI FLOW, Hetek Flow Sampler, and the AddGlobe GFM. Open-Source High Flow Sampler for Natural Gas Leak Quantification 47 Description (from manufacturer): •Sampler - Handheld device with a high-volume vacuum sampling fan and total flowrate monitor •Analyzer - Portable control module (which can be carried, placed on the floor, or mounted to a backpack) housing the gas sensor technologies, control electronics, and battery pack The combination of these two components (with a variety of sampling adapters) allows the entire fugitive methane emission to be captured, diluted, and quantified accurately. By utilizing Tunable Diode Laser Absorption Spectroscopy (TDLAS) for the accurate measurement of the fugitive methane, the dynamic range for concentrations can accurately span 4 to 5 orders of magnitude and moreover without any cross-interference from other gases present in the captured leak. Coupled with an accurate measurement of the extracted flow (methane leak and ambient air) the volume- and mass-based leak rate of the fugitive methane can be determined with high accuracy over a wide range (for example 0.001 to 25 CFM). Productization Plans: Q2 2022 • HETEK Solutions, Inc. – Hetek Flow Sampler: (“Hetek”, herein) Hetek Solutions, Inc., 2085 Piper Lane, London, ON N5V 3S5 Canada Description (from manufacturer): The Hetek Flow Sampler is a portable, intrinsically safe, battery-powered instrument that accurately measures fugitive emissions in the field at natural gas components. This is accomplished by sampling at a large flow rate to com- pletely capture all the gas leaking from the component, and by accurately measuring the flow rate of the sampling stream an