OCR Text |
Show 1 Predictive Flare Vent Gas Net Heating Value Through Determination of Average Molecular Weight Troy Boley, Ph.D.*, Derek Stuck, Nisarg Patel Sage Environmental Consulting, LP troy@sageenvironmental.com; derek.stuck@sageenvironmental.com; nisarg.patel@sageenvironmental.com *Presenter A presentation to the American Flame Research Committee (AFRC) of the International Flame Foundation (IFRF) on September 23, 2013 Abstract The research community is well aware of the recent investigations and results of industrial flare operations and the demonstrated severe impact on decreased flare combustion efficiency resulting from over-assisting (via steam or air) flare tip combustion zones.1 The petroleum and petrochemical industries use flare systems to safely combust excess process gas and emergency releases. In order to comply with current United States Environmental Protection Agency (USEPA) regulations as promulgated under New Source Performance Standard for petroleum refineries, 77 Fed. Reg. 56,422, September 12, 2012 (NSPS Subpart Ja) and in voluntary consideration of certain aspects of integrated flare control practices noted within recent flare Consent Decrees2, many refineries and chemical facilities have installed an ultrasonic flow meter to measure the total vent gas flow rate. Several facilities have also installed analytical instrumentation, such as in-line gas chromatographs (GC), to periodically measure the composition of the vent gas. The purpose of the compositional analysis was originally to quantify and assess flare emissions rates. However, in the Consent Decrees, these analytical devices are also being utilized to monitor and calculate the combustion zone Net Heating Value (NHV). The flare flow meters selected often contain a proprietary algorithm that utilizes the speed of sound in a gas to determine the gas density, and thus the average molecular weight (MW) of the vent gas. Anecdotally, those facilities with in-line compositional analysis of the vent gas have frequently reported that "similarities" and "commonalities" exist between the flow meter's MW algorithms and the MW determined analytically through calculations based on 1 Industrial & Engineering Chemistry Research, American Chemical Council, Vol. 51, October 2012 Special Edition, containing 18 separate flare related articles. 2 USEPA Flare Consent Decrees at several petroleum refineries are viewable at: BP Whiting: http://www.epa.gov/compliance/resources/cases/civil/caa/bp-whiting.html Marathon Petroleum Company: http://yosemite.epa.gov/opa/admpress.nsf/2467feca60368729852573590040443d/e841a5bbc6dd1082852579d7005b 6347!OpenDocument&Highlight=2,Marathon CountryMark Refining and Logistics: http://yosemite.epa.gov/opa/admpress.nsf/2467feca60368729852573590040443d/b511e565ba2af7f985257b20005cb 737!OpenDocument&Highlight=2,CountryMark 2 molar fractions measured by the in-line GC. When considering combustion efficiency and the need for accurate NHV calculations, the usefulness of the means by which one obtains NHV measurements are currently limited by the analytical lag time of the GC instrument, which prevents rapid, real-time monitoring of the net heating value, delays accurate and timely steam or assist air adjustments, and delays the facility's response time to smoking flare events. This paper will seek to explore the theoretical and practical connection between the average molecular weight as measured by state-of-the-art ultrasonic flare gas flow meters and the net heating value as measured by an in-line composition analyzer. Already, it is possible to voluntarily integrate the flow meter MW algorithm into a viable Steam to Vent Gas ratio (mass basis) for a continuous steam-assist control scheme. The ultimate objective is the potential demonstration that the ultrasonic flow meter's MW algorithm might be suitable as a predictor of NHV in the combustion zone, thus allowing automatic supplemental gas addition using a Predictive Emissions Monitoring System3 (PEMS) potentially negating the need for a continuous in-line GC or compositional analyzer during normal operations. Furthermore, the paper will look to incorporate this connection into the Automatic Flare Control Scheme to allow real-time adjustments to the net heating value. I. Introduction Refineries and chemical plants use flare systems to safely combust excess process gas and emergency releases produced during the refining process. Specifically, refinery flares are regulated under 40 CFR Part 60 New Source Performance Standards (NSPS) Subpart A and Subpart Ja. NSPS Subpart A requires that steam- or air-assisted flares be used only to combust vent gas with a net heating value of greater than or equal to 300 British thermal units per standard cubic foot (Btu/scf), while NSPS Subpart Ja requires (among many other requirements) the installation of a flow meter to monitor the flow rate of gas being sent to the flare at all times. In order to comply with the requirements of NSPS Ja, many refineries have installed ultrasonic flow meters such as the GE DigitalFlowTM GF868 ultrasonic flow meter upstream of the flare knock out drum. (Other ultrasonic flow meters are also available in the marketplace including the FLOSICK100 by SICK Process Automation and the FGM 160 by Fluenta). The GE flow measurement is based upon a Correlation Transit-Time™ method of ultrasonic flow measurement technology. The basic configuration of a single-channel flow meter system consists of two transducers with associated fittings, valves, transformers, preamplifier, electrical components, electronics console and interconnecting wiring. The flow meter measures the speed of sound in the gas which translates in to transit time and thus the velocity of the gas. Utilizing the fixed pipe cross-sectional area one can easily calculate the volumetric flow rate. The pressure and temperature sensors then enable the actual flow rate to be corrected to standard conditions prior to reporting the data. The ultrasonic flow meter provides the volumetric flow rate and utilizes the well known fact that the speed of sound in a gas is inversely proportional to 3 Performance Specification 16 for Predictive Emissions Monitoring Systems and Amendments to Testing and Monitoring Provisions, Federal Register / Vol. 74, No. 56 / Wednesday, March 25, 2009. 3 the gas's average molecular weight, as generated by a proprietary GE algorithm, to report the flow rate on a mass basis. In addition to the regulations promulgated within 40 CFR Part 60, the United States Environmental Protection Agency (USEPA) has continued its flare enforcement initiative.4 While each consent decree contains certain requirements unique to each facility, the USEPA has been consistent in requiring facilities to automate the addition of both steam and supplemental gas to the Covered Flares and to use these automated control systems in some cases to maintain the vent gas net heating value (NHVvg) above 300 Btu/scf (as required by NSPS Subpart A) and the combustion zone net heating value (NHVcz) above its flammability limit (NHVcz-limit) based on the highly variable composition of the vent gas. Meeting these heating value requirements is meant to promote good combustion efficiency (≥98%) at the flare tip, thus ensuring that harmful organic compounds are prevented from reaching the atmosphere. Figure 1 shows the spatial relationship of each of the NHV requirements and the gases which are to be considered at each location. Figure 1 - Flare Combustion Efficiency Parameters Generally, flare consent decrees have required determining the net heating value to be accomplished using a gas chromatograph (GC), which speciates the vent gas into the components shown below in Table 1, or, in certain circumstances, a continuous BTU analyzer or calorimeter. Using a BTU analyzer presents a number of problems with respect to the calculations the USEPA provides in each of the enforcement initiative CDs. Specifically, without individual component concentrations, lower flammable limit (LFL) calculations are not possible. Also, 4 EPA Enforcement targets Flaring Efficiency Violations, Enforcement Alert, EPA 325-F-012-002, August 2012. 4 without a GC, the specific concentrations for propylene, hydrogen, and total VOC are not available and thus the combustion efficiency multipliers used to calculate NHVcz-limit cannot be estimated. Each of the refinery flare consent decrees thus far includes an appendix which details the calculation procedures for determining NHVvg, NHVcz, and NHVcz-limit. i j Compound NHVi (Btu/scf) MWi (g/mol) LFLi, (volume fraction) 1 1 Hydrogen 274 (or) 12121 2.02 0.040 2 - Oxygen 0 32.00 ∞ 3 - Nitrogen 0 28.01 ∞ 4 - Carbon Dioxide 0 44.01 ∞ 5 - Carbon Monoxide 316 28.01 0.125 6 2 Methane 896 16.04 0.050 7 3 Ethane 1595 30.07 0.030 8 4 Ethylene 1477 28.05 0.027 9 5 Acetlyene 1404 26.04 0.025 10 6 Propane 2281 44.10 0.021 11 7 Propylene 2150 42.08 0.024 12 8 iso-Butane 2967 58.12 0.018 13 9 n-Butane 2968 58.12 0.018 14 10 Isobutylene 2928 56.11 0.017 15 11 trans-2-Butene 2826 56.11 0.017 16 12 cis-2-Butene 2830 56.11 0.016 17 13 1,3-Butadiene 2690 54.09 0.020 18 14 Pentane+ (C5+) 3655 72.15 0.014 19 - Water2 0 18.02 ∞ (1) Facilities are permitted to use either the real net heating value of hydrogen (274 Btu/scf) of the effective net heating value of hydrogen (1212 Btu/scf). Unlike other flammable vent gas components, hydrogen is not a hydrocarbon and consequently has different combustion characteristics than hydrocarbon compounds. Specifically, hydrogen has a wide flammability range and a higher adiabatic flame temperature than most hydrocarbons. As a result, the conditions under which hydrogen combusts are less restrictive than most hydrocarbons and its combustion is more efficient than most hydrocarbons. When hydrogen is combusted in the same mixture as hydrocarbons, it has been demonstrated to increase the effective NHV of that gas mixture. (2) Water is not measured by a GC, but can include in calculations of net heating value if the vent gas is high in moisture content. Additionally, the NHV of water is used in the calculation of NHVcz. Calculating the NHVvg is a straightforward calculation based the concentrations measured by the facility's GC of both calorific contributing and inert species and is shown below: !" = !×! ! !!! Equation 1 Where xi and NHVi are the molar (or volume) percent concentration of each component in the vent gas and each component's individual net heat value, respectively. 5 Although gas chromatography is currently the most commonly utilized technology to determine the net heating value of flare vent gas, it is temporally limited because the process of separating individual hydrocarbons by gas chromatography takes a short, but significant, amount of time, which impacts the responsiveness and degree of control a facility can expect to have over the NHVvg of a given flare. Technological advancements and chromatographic techniques (temperature and carrier gas adjustments) have decreased the overall instrument lag time to approximately six to seven minutes in some cases, and while this may seem like an insignificant delay, it can take an automated control system approximately 30 minutes to respond to a low NHV flaring event. Figure 2 depicts a theoretical scenario comparing the "real-time" NHV of a flaring event to the NHV reported by a GC with a seven minute lag time. Figure 2 - NHV Instrument Lag As Figure 2 shows, a GC with a seven minute lag time may require fourteen minutes to report that low NHV vent gas is being sent to the flare. This prevents the facility from responding to the flaring event in real-time, meaning that by the time the GC reports a problem, the low NHV flaring event may be already concluded. The system then spends the next 14 minutes technically out of compliance, which, depending on the actual low NHV, and thus may trigger stipulated penalties at facilities with lodged CDs. The delayed response is due in part to the simultaneous reporting of the previous analytical cycle and sampling of the next cycle. This prevents the system from responding (with the addition of supplemental gas, for instance) to the results of the previous run before sampling for the next 0 50 100 150 200 250 300 350 400 0 7 14 21 28 35 Net Hea'ng Value (Btu/scf) Time (min) RT NHV Reported NHV 6 analytical cycle. This is especially true at facilities that add supplemental gas upstream of the GC sampling point. In these cases, since the supplemental gas has not yet reached the GC when the sample is taken, its effect on the NHV will not appear for compliance purposes until the next analytical cycle is reported in 14 minutes. It may be beneficial for such a facility to offset the reporting and sampling cycles by one to two minutes to allow the system to respond to the previous analytical cycle before taking a new sample that fails to accurately represent the true NHV of the vent gas after supplemental gas addition. Alternatively, the only compliance option is to add more supplemental gas for the duration of the next analytical cycle, regardless of true demand. This option potentially adds a significant amount excess natural gas and resulting operating cost to the facility. Optimizing the analytical cycle time helps decrease the lag time, but it does not completely eliminate the problem. In gas chromatography, condensing the cycle time too aggressively may actually cause a compression of the chromatogram, improper peak resolution, and create calibration errors and resulting distortions. Short duration, low NHV flaring events may still go unnoticed by the GC, resulting in the potential emission of harmful hydrocarbons into the atmosphere. As such, anything less than a true real-time method for determining the NHV of flare vent gas potentially falls short of what is desired in the highly dynamic environment of the petroleum refinery due to the significant volumes and potential hazards of gases being analyzed. For the purposes of refineries, real-time NHV calculations could prove to be invaluable in responding to flaring events before they result in more serious problems. On-line calorimeters, such as the COSA Xentaur 9610 Wobbe Index analyzer, has a 5 second response time; however, it lacks the ability to incorporate the various EPA combustion efficiency multiplier factors. Generally, for refinery applications, EPA has thus far not agreed to a consent decree negotiation allowing an on-line BTU analyzer. Current alternative methods for determining real-time NHV using techniques such as FTIR may be in preliminary stages of conceptualization and the cost comparison to GCs is currently not available. II. NHV and Molecular Weight Correlation Ultimately, the easiest, most straightforward way to determine the net heating value of the flare vent gas in real-time would be to utilize the ultrasonic flow meter installed on many flare headers and correlate its MW to a prediction of NHV. The ultrasonic flow meter generates a number of pieces of data which may be useful for continuously determining the net heating value of the vent gas. Specifically, the ultrasonic flow meter calculates the molecular weight of the vent gas based on the speed of sound though the gas. Since the measurement is made continuously, the molecular weight would be an ideal parameter on which to base an alternative monitoring plan for compliance with the vent gas NHV requirement found in NSPS Subpart A. There are substantial obstacles to implementing this plan. In general for refinery applications, as molecular weight increases, so does the NHV of the vent gas in question. Figure 3 demonstrates this relationship, while also revealing the problem with this approach. 7 Figure 3 - NHV vs. Molecular Weight Generally, for hydrocarbons, the NHV increases linearly with the MW of the vent gas composition. However, the four selected compounds along the x-axis of Figure 3 represent the problem with using molecular weight to the approximate the NHV of a gas. Inert gases like nitrogen, carbon dioxide, and oxygen increase the molecular weight of the gas while depressing the NHV as their respective concentrations increase. Nitrogen and carbon dioxide are particularly problematic in industrial settings. Both have the same molecular weight as combustible hydrocarbons (ethylene in the case of nitrogen and propane in the case of carbon dioxide), making it especially difficult to utilize this technique to differentiate between the combustible hydrocarbons and the inert gases using only molecular weight. III. Past Research Over the past few decades, a number of methods have been proposed to use the physical properties (i.e. molecular weight and/or specific gravity) of a gas to determine the heating value of the gas. These methods take advantage of existing (for NSPS Ja compliant flares) ultrasonic flow meters to measure these physical properties on a continuous basis. These measurements are then used in calculations to determine the heating value of the gas. One such method, as envisioned by a US patent filed in 1981, uses a comparison between the behavior of sound in two gas mixtures to determine the specific gravity, or relative density, of each gas mixture5. The patent establishes a correlation between the specific gravity of pure hydrocarbon gases such as methane, ethane, and propane and expands this correlation to the 5 Haruta, M. (1981). Patent No. 4,246,773. United States. 8 specific gravity of a hydrocarbon gas mixture. The results of this method show a linear relationship between the specific gravity of a mixture and the actual heating value. Applying this invention and its concepts to the ultrasonic flow meters, one could mathematically determine the specific gravity of the vent gas in real-time and thus provide real-time estimates of the net heating value of the flare vent gas. However, the method in the patent was only tested on mixtures containing flammable components. Flare vent gas generally consists of a mixture of flammable volatile organic compounds (VOCs) and inert compounds, such as nitrogen and carbon dioxide. It is at those times when the inert compounds represent significant or dominant components of a gas mixture that one requires supplemental means to estimate the NHV of industrial flare gas. Another potential method for determining the NHV of flare vent gas was proposed in an International patent filed in 20086. This method combines the physical properties of the gas as measured by the ultrasonic flow meter with a Monte Carlo simulation method. Specifically, the method generates a wide range of potential gas compositions (using a range of possible concentrations for each component) which will equal the molecular weight being reported by the flow meter. A separate range of possible gas concentrations is also generated to match the specific gravity (or speed of sound) as measured by the flow meter. These simulated gas compositions are then compared to each other. Ideally, the combined set of gas compositions which match both the molecular weight and specific gravity of the vent gas consists of only one likely gas composition. From this "most likely" gas composition, the NHV can be determined within a known correlation or error range, just as would be done from the analytical results from a GC. To the author's knowledge, there has been no industrial application of the 2008 Huang patent concepts. IV. Future Research and Testing The aforementioned patents provide a general idea of potential avenues for future research, although they are by no means the only potential methods to be investigated. Assuming the ultimate goal to be the real-time, continuous and accurate estimation of the NHV of flare vent gas, future research should also include Fourier Transform Infrared (FTIR) spectroscopy instrumentation alongside calibrated and validated GC and on-line BTU analyzers. A PEMS demonstration per USEPA Performance Specification 16 could be made at a willing facility. These instruments are commercially each available and have instrument lag time on the order of seconds instead of the typical 12-15 minute lag experienced by GCs, and each may require less down time for daily calibration drift or span checks. Additionally, both instruments may be cheaper to operate over the life of the instrument than a similarly capable GC. Further benefits include more accurate usage of supplemental gas addition. The benefits of real-time NHV monitoring are numerous and significant. Monitoring and accurately predicting the NHV in real-time allows for a facility to respond to flaring event faster and potentially allows for the facility to use less supplemental gas in responding to a low NHV 6 Huang, Y. (2008). Patent No. 2008/073672. International. 9 flaring event. At facilities operating under the auspices of a CD, faster flaring event response should also allow the facility to avoid stipulated penalties. Finally, the potential use of existing instrumentation, such as the ultrasonic flow meter, to monitor vent gas density and MW, so as to accurately derive a NHV within certain ranges of error, provides facilities with a cost-effective alternative to purchasing expensive additional instrumentation. For further discussions or inquiries, please feel free to contact the presenter at: Troy M. Boley, Ph.D. Vice President and General Manager Sage Environmental Consulting, LP Kennesaw, Georgia Mobile 770-883-7082 Email troy@sageenvironmental.com |