Joint International Combustion Symposium Paper, October 11, 2004, Maui, HI
Practical Implications of Prior Research on Today's Outstanding Flare Emissions Questions and a Research Program to Answer Them
- James Seebold, ChevronTexaco (Retired)
- Peter Gogolek, Natural Resources Canada
- John Pohl, Virginia Polytechnic Institute and State University
- Robert Schwartz, John Zink Company LLC
The external combustion of hydrocarbon gas mixtures by any means, including flaring, literally manufactures and subsequently emits to the atmosphere traces of all possible molecular combinations of the elemental constituents present either in the fuel or in the air including the ozone precursor highly reactive volatile organic compounds (HRVOCs) and the carcinogenic hazardous air pollutants (HAPs). In the case of flare operation, this is probably particularly true not only of over-steamed flares but also of Best Practice no-flaring purge-and-pilot-only hot-standby operation. But these trace emissions are hard to measure and in prior related research on hydrocarbon gaseous jet-mixed diffusional combustion have been shown to be trace enough that they pose no threat whatsoever to the public health and welfare.
Although recently it has been treated as such by some researchers, regulators and environmentalists, it is hardly a revelation that even burning methane pure as the drifted snow and in the best possible well-mixed way produces trace emissions of ethylene, propylene, butadiene, and all the other highly reactive volatile organics; formaldehyde, benzene and benzo(a)pyrene, the class-archetypal hazardous air pollutant carcinogens; and all the other hydrocarbon compounds in the gas phase up through 300 mw coronene. This will be illustrated by the emissions to atmosphere from the diffusional combustion of natural gas. The severely over-aerated condition (stoichiometric ratio 4.5; i.e., four-and-one-half times theoretical air supplied) may be typical of severely over-steamed flares while the severely under-aerated condition (stoichiometric ratio 0.75; i.e., three-fourths theoretical air supplied) may be typical of Best Practice no-flaring purge-and-pilot-only hot-standby flare operation. In the Petroleum Environmental Research Forum's Project 92-19 this issue was met head-on. The landmark measurements showed 1) that all of those emissions are indeed detectable in the stack plume if the investigators are good enough at the detecting and 2) that the trace concentrations under a broad range of normal operating conditions are trace enough that they pose no threat whatsoever to the public health and welfare. This knowledge forms a foundation upon which reasonable approaches to the needs of the public for affordable products and employment and the public need for a clean environment can be structured. Such was the case when industry regulatory advocates, environmental activists and government regulators worked together to give special consideration to process heating and steam raising operations in the EPA's Industrial Combustion Coordinated Rulemaking.
Now a foundation of knowledge needs to be constructed regarding flaring operations. Critically needed is resolution of these issues not just by arm-waving but quantitatively and systematically, comprehensively and unambiguously just was done for process heating and steam raising in the PERF 92-19 Project. Both what we know and what we need to know about flare emissions will be described. The presentation closes with a brief description of a new flare emissions program that will produce the new knowledge that will resolve the outstanding issues and support sensible flare regulations.
The mid-90s saw completion of the Petroleum Environmental Research Forum (PERF) Project 92-19 entitled "The Origin and Fate of Toxic Combustion Byproducts in Refinery Heaters: Research to Enable Efficient Compliance with the Clean Air Act." Some interesting things were learned during the course of this project about hydrocarbon gaseous external combustion.
Much, but not all, of the experimental work was conducted at the Burner Engineering Research Laboratory ("BERL"), Sandia National Laboratories ("SNL") located in Livermore, California. Other experimental work was conducted at the UCLA Chemical Engineering Laboratory and at the Stanford Thermosciences Laboratory. The 4-year, 20-participant $7-million project produced 7 volumes. But a "Final Report", really an extended executive summary, can be found at the following Website:
Some results from the SNL research furnace (dubbed "Baby BERL") are shown below. These results illustrate that even when burning laboratory grade methane pure as the drifted snow, traces of higher molecular weight compounds such as benzene and toluene that are not originally present in the fuel are nevertheless found in the flue gas.
The combustion reaction zone behaves like an effectively dissociated highly reactive elemental soup in which all possible molecular combinations of the elements present are formed in accordance with their chemical kinetic propensity to do so. Furthermore, there being no zero in nature, traces of all possible molecules remain in the flue gas for the detection if you are good enough at the detecting. In short, Cs and Hs and Os (Oh My!) beget Cs and Hs and Os and the hydrocarbon fuel composition doesn't matter very much.
Full-scale burner trials were carried out in "Big BERL", the SNL Combustion Research Facility's (CRF) Burner Engineering Research Laboratory illustrated below.
Heat was extracted from the combustion zone by water-cooled walls to precisely duplicate the conditions in the radiant sections of typical process heaters. That capability was already present when the project began. But PERF Project 92-19 added a "Convection Section Simulator" at a cost of about $½-million to precisely mimic the entire heating and cooling profile typical of industrial process heaters. Most importantly, the temperature vs. time cooling profile could be tailored to mimic a variety of designs encountered in the field. As an aside, it is perhaps interesting to note that added capability to tailor the cooling profile would have become important in subsequent research, has it been carried out, on the generation and fate of the polychlorinated dibenzodioxins and dibenzofurans in hydrocarbon gaseous external combustion. While that research was never carried out the capability remains intact at the laboratory and could be utilized in future research.
Some results are shown above. PERF Project 92-19 research proved conclusively that while trace emissions under the very broad range of stoichiometries that characterize normal operating conditions are detectable in the flue gas, if you are good enough at the detecting, the concentrations are so low as to pose no threat whatsoever to the public health and welfare.
But the research also showed, as illustrated above, that there exist both substoichiometric and superstoichiometric regimes in which the emissions from hydrocarbon gaseous external combustion are markedly increased. In our judgement these observations from prior research may have important implications on flaring operation and best practices. That is why we proposed new research and put together the research consortium.
In beginning to sort out the implications of prior research on flare emissions, the crucial question is, "How might flaring operation produce the adverse stoichiometric mixing regimes that proved to be undesirable in the prior research on industrial burners?"
There have been identified three elevated flare reacting flow mixing regimes; viz., inertia-, buoyancy- and wake-dominated mixing. These regimes are illustrated below.
In the inertia dominated regime nothing is important except the jet-mixing inherent in the high velocity jet. Because of the strength of the flare jet momentum flux the combustion zone is very little affected by any other mechanism and combustion efficiencies are invariably high.
But as the flare jet velocity is reduced there is a changeover to mixing dominated by buoyancy in which some possibility of eddy quenching or eddy stripping apparently would emerge. Nevertheless, we know that the combustion efficiencies remain high in this regime as well due largely to the very powerful bouyant engulfment mechanism of air induction and the overwhelming reactivity of the reactants.
Eventually, as the flare jet velocity is reduced or the crosswind speed increases, the wake-dominated mixing regime emerges in which the flame is sucked down and stabilizes itself in the vortex trail off the stack. While extremely stable, this regime is one of potential low efficiency. But the operating condition boundaries lack definition today.
What we do know about the wake-dominated regime is illustrated in the chart above and we conjecture that wake-domination leads to potentially low-efficiency substoichiometric eddies as suggested in the illustration below. While this deleterious effect has been clearly identified by recent research, quantification, operating condition envelope, and governing parameters all lack definition today.
Another low-efficiency mixing regime has been postulated. A very effective and well established operating pracice has been the introduction of steam in elevated flares to suppress smoke. The main effect is to enhance the aeration of the combustion zone.
But we postulate that excessive use of smoke-suppressing steam may lead to over-aeration and the production of superstoichiometric eddies as illustrated below and there is evidence to suggest that over-steaming compromises combustion efficiency in certain circumstances as shown in the chart below.
Key knowledge gaps that today stand in the way of sensible regulation and the enumciation of operating parameter envelopes that would delineate "good operating practice" and ensure the high efficiency operation of flares:
To what extent are prior research results on off-stoichiometric jet-mixed diffusion indicative of the emissions from elevated flares that might not be "well-operated"?
Like jet-mixed burners, can elevated flares be operated in such a way that they pose no threat whatsoever to the public health and welfare?
So the question comes, "How will we get the key flare emission knowledge that we do not have today to support sensible regulation?" The answer is, of course, further research.
Direct measurement of the in situ combustion efficiency of full-scale flares is both difficult and dangerous. Remote measurement techniques are under development but are not proven. Today's means of ensuring the high efficiency of flares is to specify the operating conditions that, in direct measurement programs such as the USEPA's mid-80s Study of the Efficiency of Industrial Flares, produced high efficiency. But to date the operating parameter efficiency envelopes have not been developed to account for wind nor for the chemical properties of the gases flared nor for the amount of smoke-suppressing steam employed. Recent studies of hydrogen or inerts in the flared gases have demonstrated that energy content (Btu/scf) alone is a poor descriptor even though it is relied upon in the USEPA's 40CFR60.18 "General Requirements for Flares".
These facts highlight critical knowledge gaps that stand in the way of enunciating measurable operating parameter based "Flaring Best Practices" to ensure high efficiency operation. That is why the authors proposed a new flare emissions research project that would study the effect of flare gas flow and composition; steam assist and flare gas mass ratio; wind and flare gas momentum flux ratio; and wind turbulence structure on the combustion efficiency of flare flames focusing on speciated emissions of the highly reactive volatile organic compounds ethylene, propylene and butadiene; and the class archetypal hazardous air pollutant carcinogens formaldehyde, benzene and benzo(a)pyrene.
The New Research Project
The "International Flare Consortium" has been formed by the author Principal Investigators (PIs) to eliminate the knowledge deficit that today stands in the way not only of enunciating operating-parameter-based best practices but also sensible regulation. The project has a long title; viz., "The effect of flare gas flow & composition; steam assist & flare gas mass ratio; wind & flare gas momentum flux ratio; and wind turbulence structure on the combustion efficiency of flare flames focusing on speciated emissions of the highly reactive volatile organic compounds (ethylene, propylene, butadiene) and the class archetypal hazardous air pollutant carcinogens (formaldehyde, benzene, benzo(a)pyrene)."
The project kicked off August 10-11, 2004, at the Natural Resources Canada (NRCan) CANMET Energy Technology Centre (CETC) Flare Test Facility (FTF), Ottawa, Ontario, Canada where most testing will be conducted. Additionally the PIs expect that several CANMET FTF conditions will be duplicated at full industrial scale at the John Zink Flare Test Facility (JZFTF), Tulsa, Oklanoma, and "sampled" by a remote measurement technique; e.g., 1)no steam with smoke, 2) optimum smoke suppressing steam and 3) severe over steaming, all at the same composition and flow rate. These verification tests at the JZFTF will be conducted "blind" with the dual purposes of providing much-needed validation of the chosen remote measurement technique (FTIR or LDAR) that appears best and confidence in the broad range of comprehensive and unambigous results to be obtained in the CANMET FTF.
Phase I of the new research program will investigate experimentally and theoretically the combustion efficiency and speciated emissions of natural gas; refinery fuel gas; HRVOC- and BTEX-spiked natural and refinery fuel gas; and low-Btu production gas under a broad range of wind vs. flare mass flux ratios and steam assist vs. waste gas mass ratios. Phase II will repeat the critical conditions found in Phase I to effect combustion efficiency and speciated emissions of flare flames with a broader range of gas compositions. Phase III (TBD and not included) will provide both full-scale verification of the CANMET FTF experimental results and validation of remote measurement techniques at the John Zink Flare Test Facility.
The following are among the technical or regulatory drivers that led to the formation of the International Flare Consortium:
Emissions from flares in the Houston Galveston Area - particularly of the highly reactive volatile organic compounds like butadiene, propylene and ethylene - are currently of great interest to the Texas Commission on Environmental Quality and to the industrial community.
The same can be said of California's South Coast and Bay Area Air Quality Management Districts and the industrial communities there and elsewhere.
The World Bank's Global Flare Reduction Initiative seeks to reduce the emissions of greenhouse gases from flares.
The program will provide a much-needed check on and validation of remote measurement techniques which may develop to monitor flare flames.
If development of remote techniques to measure flare combustion is delayed, or if remote measurement techniques are found to lack practicality or adequate detection limits, this program will provide independent and unambiguous resolution of today's pressing issues.independent and unambiguous resolution of today's pressing issues.
The program as currently envisioned is shown above. Since the program will be jointly and continuously managed by the Principal Investigators and the representatives of the Supporting Entities who sit on the Technical Advisory Board, the program is subject to change as the research unfolds.
The "Initial Scoping Matrix" shown in the program plan is detailed above. At press time the project had just kicked off but we expect to have some early results to discuss in the live presentation in October.
To briefly describe the selection rationale for the conditions included in the Initial Scoping Matrix, tests A1, A2 and A3 are intended to scope the range of wind effects as characterized by the momentum flux ratio for all fuels in the absence of steam injection and will therefore be of broad interest to all of the Supporting Entities represented on the Technical Advisory Board. Tests B1, B2 and B3 will scope the range of steam effects at constant momentum flux ratio for fuels of interest to the downstream (refining, petrochemical and chemical) Supporting Entities that are represented on the Technical Advisory Board. Typically the upstream (production) entities do not employ steam injection so fuels 5 and 6 are omitted in the initial scoping matrix. Finally tests C1, C2, C3, C4, C5 and C6 scope the extremes of momentum flux ratio and steam injection for any surprises and therefore are limited to natural gas (fuel 1) and refinery fuel gas (fuel 2) in the initial scoping matrix. The results of the Initial Scoping Matrix will enable PIs and TAB in their 2nd meeting to more rationally lay out the test conditions that will smoothly fill out our understanding during the completion if the 1st efficiency campaign.
IFC program deliverables include comprehen-sive and unambiguous quantification of:
speciated emissions concentrations from elevated flares as a function of flare gas composition, flare/wind momentum flux ratio and steam rate;
flare/wind low momentum flux ratio inef-ficiencies in downstream "best practice" no flaring purge and pilot only operation and in upstream field flare operation;
over-steaming inefficiencies in down-stream operations;
and identification of measurable operating parameter based "Best Practices" to ensure high efficiency operation of upstream and downstream elevated flares.