Combustion Efficiency Performance Evaluation of a Variable Speed Drive Air-Assisted Flare using Passive FTIR

Paper from the AFRC 2016 conference titled Combustion Efficiency Performance Evaluation of a Variable Speed Drive Air-Assisted Flare using Passive FTIR

Located on the Cook Inlet, 60 miles southwest of Anchorage, the Tesoro Kenai Refinery can process up to 72,000 barrels per day (bpd). The refinery produces gasoline and gasoline blendstocks, jet fuel, diesel fuel, heating oil and heavy fuel oils, propane and asphalt. Pursuant to the forthcoming Consent Decree settling certain claims between the United States and Tesoro Alaska Company LLC ("Tesoro"), Tesoro contracted with IMACC, LLC to conduct a combustion efficiency performance test on the air-assisted Main Refinery flare at its Kenai Refinery. The main objective of this test was to prepare a combustion efficiency performance curve for the flare over a range of flare operating conditions. The performance test measured the air-assist rate provided by the current fan and fan motor, and assessed the combustion efficiency of the flare under varying vent gas flow rates of differing calorific values and different assist-air flow rates. During each test condition, the PFTIR remotely analyzed the resulting combustion gases in the flare plume to determine combustion efficiency. Data from the performance test are used to:;  Provide a detailed description of the extent to which the ratio of mass flow rate of air to the stoichiometric air flow for the vent gas (ṁair-asst/ṁair-stoich-vg) affects combustion efficiency;;  Provide a detailed description of the range of the ṁair-asst /ṁair-stoich-vg that the Kenai Air-Assisted Flare can operate to ensure a minimum 96.5% Combustion Efficiency (or as high an efficiency as reliably obtainable), taking into consideration variability in Vent Gas flow rate and composition; and;  Describe the maximum ṁair-asst/ṁair-stoich-vg at which Tesoro proposes to operate the Kenai Air-Assisted Flare.; The presentation will highlight the unique features of testing an air-assisted flare equipped with a variable speed drive under flare conditions of varying calorific values and volumetric flow rates.

AFRC 2016 Industrial Combustion Symposium Kauai, HI September 13, 2016 Combustion Efficiency Performance Evaluation of a Variable Speed Drive Air-Assisted Flare using Passive FTIR Authors: Troy M. Boley, PhD President IMACC, LLC Matt McCormick, PhD Marc Johnson Senior Scientist Environmental Engineer IMACC, LLC Tesoro Alaska Company, LLC [e-mail] (Presenter) Summary Tesoro Alaska Company, LLC contracted with IMACC, LLC to conduct a combustion efficiency performance test on the existing air-assisted flare at its Kenai Alaska Refinery. The test was in accordance with an approved EPA test plan. The main objective of this test was to prepare a combustion efficiency performance curve for the flare over a range of flare operating conditions. While the flare operating conditions were purposefully varied between test runs, each individual test run was conducted under stable conditions. The performance test measured the air-assist rate provided by the current fan and variable speed drive motor, and assessed the combustion efficiency of the flare under varying vent gas flow rates of differing calorific values and different assist-air flow rates. During each test condition, the PFTIR remotely analyzed the resulting combustion gases in the flare plume to determine combustion efficiency. Data from the performance test are used to: - Provide a detailed description of the extent to which the ratio of mass flow rate of air to the stoichiometric mass air flow for the vent gas (ṁair-asst/ṁair-stoich-vg) affects combustion efficiency; - Provide a detailed description of the range of the ṁair-asst /ṁair-stoich-vg that the Kenai AirAssisted Flare can operate within to ensure a minimum 96.5% Combustion Efficiency (or as high an efficiency as reliably obtainable), taking into consideration variability in Vent Gas flow rate and composition; and -Describe the maximum ṁair-asst /ṁair-stoich-vg at which Tesoro proposes to operate the Kenai Air-Assisted Flare. Assist-Air Control System Assist gas (ambient air) is purposefully added to the flare tip to provide additional flame zone turbulence and improved air entrainment, resulting in improved combustion efficiency and the 1 reduction of smoke formation. Care must be taken to ensure the combustible mixture of the combined gases is not diluted by the use of excess assist gas. Assist air is provided by a single fan equipped with a variable frequency drive (VFD). The air is pushed up the stack in the annular space between the outer wall of the flare stack and the vent gas riser. The assist air enters the combustion zone through the annular space around the vent gas riser at the flare tip and through three 4-inch inner tubes that enter the vent gas riser at the base of the flare tip. Per the manufacturer's specification sheets, the VFD air blower can deliver up to 18,000 actual cubic feet per minute (acfm). However, due to installation requirements and operational considerations, the actual amount of assist air supplied by the blower is less than full design. The supply of assist air can be adjusted remotely by the operators to prevent visible emissions. As part of this test, a temporary optical flow sensor was installed on the blower inlet to record the flow rate of air being sent to the flare tip. Passive FTIR (PFTIR) The instrument used to determine the gas composition of the flare plume is the IMACC Passive Fourier Transform Infrared (PFTIR) analyzer. The PFTIR operates on the principle of spectral analysis of thermal radiation emitted by hot gases. Passive means that no "active" infrared source is used; rather, the flare plume itself serves as the source of infrared light. This analytical approach is possible because the infrared emission spectra of hot gas is identical to the absorption spectra. Therefore, observing a flare with an appropriately calibrated infrared spectrometer allows for identification and quantification of each species through emission spectroscopy, just as in absorption spectroscopy. IMACC has helped pioneer the technique of using PFTIR for flare combustion efficiency studies, and has patented analytical software that can analyze the measured emission spectra of flare plumes. Gas Chromatograph (GC) The air assisted flare is equipped with a Siemens MAXUM II gas chromatograph (GC). A GC has the ability to speciate the vent gas components that contribute to its calorific value. A GC uses a mobile and stationary phase where the gaseous compounds being analyzed interact with the walls of the column which are coated with the stationary phase. Based on the difference in affinity and retention time, the gas components can be separated and analyzed. The Siemens MAXUM II GC on the air assisted flare has a typical sample and analysis duration of approximately nine (9) minutes; in no case during the test did a complete analytical cycle exceed 15 minutes. The GC is equipped with a thermal conductivity detector. Below is a list of all the compounds that the GC is capable of detecting.  Hydrogen (H2)  Oxygen (O2)  Nitrogen (N2)  Carbon Monoxide (CO)  Methane (CH4)  Carbon Dioxide (CO2)  Ethylene (C2H4) 2         Ethane (C2H4) Propane (C3H8) Propylene (C3H6) Isobutane (C4H10) Butane (C4H10) Butenes and 1,3-butadiene Pentanes plus (hydrocarbons with five carbons or more) Hydrogen Sulfide (H2S) The Main Refinery flare GC has two sample taps in the flare header, one being for sample extraction and the other for sample return. The length of the sample line from the point of extraction to the GC column is approximately 100 feet long. FTIR/Raman BTU Analyzer A unique feature of this test was the inclusion of a new instrument IMACC has developed for more rapid analysis of vent gas streams for BTU content. This instrument is the FTIR/Raman Dual Function Gas Analyzer. While FTIR is well known for its capability to quantify all the hydrocarbons and other multi-atom compounds, homonuclear diatomics such as hydrogen and nitrogen are "invisible" to FTIR detection. With the addition of Raman spectroscopic techniques, these molecules can be quantified, allowing for the complete analysis of the vent gas fuel stream. With this instrument, the fuel stream could be analyzed on a minute-by-minute basis. While the GC was used as the final indication of process stability, the efficiency of changes in process conditions was improved as a result of the faster response of the FTIR/Raman. Assist Air Optical Flow Sensor To provide a measurement of the true air assist rate obtained by matching the RPM output of the fan to its calibration curve, an OFS-2000 optical flow sensor from Optical Scientific, Inc, was installed in the air intake pipe upstream from the fan. A heated rod upstream of the sensor induced turbulence into the air flow and the linear velocity was measured via the scintillation of a laser passing through the turbulent air. The optical sensor recorded the actual assist air flow rate during the PFTIR test. Due to discrepancies between the manufacturer's fan curve and the air flow measurements, the refinery has chosen to develop a new fan curve based on the measured flow of assist air at various set points (100%, 85%, 50%, and 20% of maximum air assist output). At 100% output, the OFS flow sensor reported 14,700 scfm of air flow. The flow rate is not linear with the VFD's RPM reading. This new site specific fan curve will be used to determine the amount of air being supplied to the flare tip. The fan curve is subject to change, based on modifications to the VFD blower or operational changes at the Kenai Refinery and a future permanent flow meter may be required. Flare Test Program The primary objective of the flare performance test was to measure combustion efficiency over a range of operating conditions. Passive FTIR was used to measure the combustion efficiency during each test condition. The test was conducted at an operating refinery and, thus, certain operational 3 constraints required adjustment to the original test scenarios presented in the approved flare test protocol. Tesoro and IMACC targeted ten operational scenarios consisting of low, average, and high vent gas flow rates. Within each flow regime, the characteristics of the vent gas were systematically changed to provide varying NHVvg by adjusting:    refinery operations (e.g., process unit purges that vent to the flare system), the direct addition of nitrogen to lower the NHVvg conditions, or the direct addition of LPG or natural gas to create higher NHVvg conditions. The ten scenarios are laid out in more detail in Table 1. Target Scenario NHVH2=274 Vent Gas Flow 1 Air Assist Output Setting 2 3 4 5 T1 340 Low 100% 85% 75% 50% 20% T2 560 Low 100% 85% 75% 50% 20% T3 1000 Low 100% 85% 75% 50% 20% T4 1500+ Low 100% 85% 75% 50% 20% T5 560 Medium 100% 85% 75% 50% 20% T6 1000 Medium 100% 85% 75% 50% 20% T7 1500+ Medium 100% 85% 75% 50% 20% T8 560 High 100% 85% 75% 50% 20% T9 1000 High 100% 85% 75% 50% 20% T10 1500+ High 100% 85% 75% 50% 20% Table 1. Ten Operational Scenarios The original intention of the test plan was to have test scenarios where the low, average, and high vent gas flow rates were all approximately equivalent while varying the NHVvg. However, as a result of operational constraints, this plan was altered such that the NHVvg was comparable between the low, average, and high flow rates, and that the flow rates of each condition were clearly differentiable. Thus the low NHVvg cases had higher vent gas flow rates than the average NHVvg cases; to achieve the desired calorific value, additional nitrogen was intentionally used to lower the NHVvg, resulting in a higher overall vent gas flow rate. The original test plan used NHVvg target values using 274 BTU/scf for the NHV of hydrogen as this was how the NHVvg was to be calculated by the GC during the test. However, the NHVvg has been recalculated using 1212 BTU/scf for the NHV of hydrogen. These values are used throughout this paper and are represented using NHVvg. To provide a comparison of the test result with the target listed in Table 1, a net 4 heating value calculated using NHVH2 = 274 BTU/scf will be provided for completeness. These will be represented using NHVH2=274. The test results show that combustion efficiency varied predictably as a function of vent gas heating value and the volume of excess assist air that was introduced into the combustion zone. Analysis of the data allow general conclusions to be drawn relating to the recommended operation of the air assist fan as a function of vent gas heating value and flow rate. Unlike in steam-assisted flares, there is no sharp roll-over in the combustion efficiency - instead, a seemingly steady decline in the combustion efficiency with increasing air volume was observed. This was a trend generally present in every test condition. A sufficient number of tests were performed to duplicate at least 50% of the individual test conditions (e.g., 75% blower output in Test Series 2). In addition, every test condition reported met the requirement of ± 15% stability for both vent gas heating value and vent gas flow rate. A few sample test series are included below to highlight the findings of the performance evaluation. Test Series 3 - Low Vent Gas Flow, Medium NHVvg Test series 3 (T3) represented a case where the flow rate of the vent gas was to be as low as feasible while having the heating value of the vent gas be approximately equivalent to the sweep gas. The average NHVvg during the T3 test was 1,221 BTU/scf with an average flow rate of 114,000 scf/day. The target BTU value of the test was set at 1,000 BTU/scf. Figure 1. Plot of CE with respect to air assist output for Test 3. The results of the combustion efficiency as a function of air assist are shown in Figure 1. The flare's average combustion efficiency for each test condition was above 96.5%, with only occasional measurements reading less than 96.5%. Once again, only the 20% air assist setting reliably achieved CE's above 98%. The overall "slope" of the CE vs. air assist trend appears to be somewhat less pronounced compared to the tests previously described. It is worth noting that there was a 2.5 hour gap between the 85% test and the 75% test, and the fuel stream composition had changed during that time, even though the BTU content was relatively similar. This may explain some of the disparity 5 between the higher fan speeds and lower fan speeds. No visible emissions were observed at the 20% fan setting. Figure 2. Plot of CE versus stoichiometric ratio for Test 3. Figure 2 shows the average combustion efficiency from each test condition plotted against the stoichiometric ratio calculated from the air assist. Based on the equation of the trend line from the test results, the CE drops below 96.5% at a stoichiometric ratio of 11.2 or higher, and it drops below 98% at a SR of 4.41 or higher. Because of the discrepancy caused by the time delay, the two high air assist conditions were not factored into the trend line. If they are included, the predicted SR for achieving 98% efficiency is 1.38, which is clearly incorrect. The results from this test condition show that when the flare is operating at its base load condition with medium NHVvg (1,220 BTU/scf), the stoichiometric ratio should be less than 11.2 for 96.5%, or less than 4.41 for 98% CE, to be achieved in the combustion zone. These SRs correspond to a fan blower setting of approximately 90% (or 13,500 scfm) for 96.5% CE and 30% (or 5,100 scfm) for 98% CE. Test Series 6 - Medium Vent Gas Flow, Medium NHVvg Test series 6 (T6) represents a case where the flow rate of the vent gas was to be at the typical rate used at the refinery while having the heating value of the vent gas be approximately equivalent to the sweep gas. While the fuel flow rate was less than in T5, it was set to be higher than T3. Two tests were conducted back-to-back on the same day, with each test running through all five air assist rates. The average NHVvg during both T6 tests was 1,233 BTU/scf with an average flow rate of 144,000 scf/day. The target BTU value of the test was set at 1,000 BTU/scf. 6 Figure 3. Plot of CE with respect to air assist output for Test 6. The results of the CE as a function of air assist are shown in Figure 3. The flare's combustion efficiency was found to be generally equivalent or slightly higher than that measured during the T3 test series, indicating again that the higher vent gas flow rate resulted in less dilution from the air assist. Most of the data shows CEs above 96.5%, with only one 85% test showing an average CE below 96.5%. The 20% air assist settings were the only ones that uniformly achieved CEs above 98%. Both data sets appear to show the 100% air assist as being more efficient than the 85% air assist, which does not make intuitive sense. It is not immediately clear why this is the case. In the first set of data, the analysis of the data showed CO2 values which were higher than most of the other test conditions, even though the other components of the CE calculation seemed to track each other, which would inherently increase the calculated CE. The second set of data actually show similar results between the 100% and 85% settings, but a couple of low CEs pulled the average of the 85% air assist down. No visible emissions were observed at any of the test conditions. Figure 4. Plot of CE versus stoichiometric ratio for Test 6. 7 Figure 4 shows the average combustion efficiency from each test condition plotted against the stoichiometric ratio calculated from the air assist. The two data points from the 100% air assist tests have been omitted because the trend line including these points would have implied that there was no stoichiometric ratio that would have achieved 98% combustion efficiency. Based on the equation of the trend line from the test results, the CE drops below 96.5% at a stoichiometric ratio of 12.4 or higher, and it drops below 98% at a SR of 3.06 or higher. The results from this test condition show that when the flare is operating at its normal condition with medium NHVvg (1,230 BTU/scf), the stoichiometric ratio should be less than 12.4 for 96.5% CE, or less than 3.06 for 98% CE, to be achieved in the combustion zone. These SRs correspond to a fan blower setting of approximately 85% (or 13,000 scfm) for 96.5% CE and 30% (or 5,100 scfm) for 98% CE. Test Series 8 - High Vent Gas Flow, Low NHVvg Test series 8 (T8) represents a case where the flow rate of the vent gas was to be set at a rate higher than what is typical at the refinery while keeping the heating value of the vent gas relatively low. As with T2 and T5, the flow rate of the vent gas was artificially inflated because a moderate amount of nitrogen was required to dilute the vent gas to the desired BTU value (560 BTU/scf), but the average flow rate was higher than that used in T5. Two tests were conducted back-to-back on the same day, with each test running through all five air assist rates. The average NHVvg during the T8 test was 663 BTU/scf with an average flow rate of 273,000 scf/day. Figure 5. Plot of CE with respect to air assist output for Test 8. The results of the combustion efficiency as a function of air assist are shown in Figure 5. The flare's combustion efficiency was found to be similar, but somewhat lower than that measured during the T5 test series; the consistency of the data during both tests is quite striking. Only the 50% and 20% settings showed average CEs above 96.5%, with no 85% or 100% air assist measurement above 96.5%. The 20% air assist settings generally achieved CEs above 98%, along with some of the 50% air assist settings, while the higher assist air conditions were clearly below 98% CE. No visible emissions were observed at any of the test conditions. 8 Figure 6. Plot of CE versus stoichiometric ratio for Test 8. Figure 6 shows the average combustion efficiency from each test condition plotted against the stoichiometric ratio calculated from the air assist. Based on the equation of the trend line from the test results, the CE drops below 96.5% at a stoichiometric ratio of 8.18 or higher, and it drops below 98% at a SR of 4.84 or higher. The results from this test condition show that when the flare is operating at high vent gas flow conditions with low NHVvg (660 BTU/scf), the stoichiometric ratio should be less than 8.18 for 96.5% CE, or less than 4.84 for 98% CE, to be achieved in the combustion zone. These SRs correspond to a fan blower setting of approximately 60% (or 9,800 scfm) for 96.5% CE and 20% (or 3,500 scfm) for 98% CE. Conclusions Two consistent patterns were identified from the entirety of the test scenarios: 1. The combustion efficiency of the flare as a function of the air assist rate varies approximately linearly for a given vent gas NHV and flow rate. This has been observed before in air assisted flares, and is in stark contrast to the performance of a steam assisted flare where the combustion efficiency appears to drop sharply once too much steam is added. 2. Increasing the vent gas flow rate increased the combustion efficiency for a given air assist and vent gas heating value. The increased flow rate counteracts the dilution effect of the air assist and results in improved combustion efficiency. Table 2 provides a summary of the combustion efficiency for each test condition. The CEs are colorcoded as per the legend to help provide a visual guide as to the vent gas and air assist conditions that would allow the flare to operate above 96.5% CE. 9 Table 2. Summary of average CE's observed at each test condition with color-coding to indicate measured CE relative to 96.5% CE T1 T2 T3 T4 Fan Output Setting 100% 85% 75% 50% 20% average efficiency 94.47% 94.98% 96.15% 96.50% 98.19% standard deviation average efficiency 0.30% 1.60% 0.50% 0.38% 0.25% 95.60% 96.37% 96.17% 96.35% 97.86% T5 Fan Output Setting 100% 85% 75% 50% 20% 100% 85% 75% 50% 20% 1.04% 0.58% 0.86% 0.68% 0.21% T6 average efficiency standard deviation average efficiency standard deviation 96.86% 97.20% 96.61% 96.82% 98.41% 0.94% 0.49% 1.07% 0.52% 0.53% 96.65% 96.70% 96.82% 98.15% 98.40% 0.41% 0.85% 0.81% 0.22% 0.42% T7 average standard average standard average standard efficiency deviation efficiency deviation efficiency deviation 96.82% 96.11% 97.27% 97.80% 99.14% 0.45% 1.79% 1.00% 0.85% 0.34% T8 Fan Output Setting standard deviation 97.85% 97.11% 97.00% 97.38% 98.30% 0.31% 0.55% 0.54% 1.03% 0.34% T9 96.83% 96.76% 96.97% 98.48% 98.99% 0.69% 0.97% 0.90% 0.35% 0.30% T10 average standard average standard average standard efficiency deviation efficiency deviation efficiency deviation 94.00% 94.91% 96.09% 97.53% 98.22% 2.11% 0.93% 0.85% 0.90% 0.36% 96.06% 96.62% 97.49% 98.09% 99.11% 0.99% 0.73% 0.57% 0.38% 0.31% 98.19% 98.60% 99.02% 98.81% 99.01% 0.28% 0.44% 0.16% 0.19% 0.30% LEGEND CE < 96.5% CE < 96.5%, but STD extends above 96.5% CE < 96.5%, but STD extends below 96.5% CE > 96.5% A summary of the SRs and blower settings for each test are provided in Table 3. From the results obtained through this performance test, Tesoro was able to conclude that the blower should rarely be operated at 100% output (14,700 scfm) in order to achieve greater than 96.5% CE. Only conditions where high BTU content of the vent gas (NHVvg > 1,550 BTU/scf) allow for the blower to be operated at maximum output. In addition, the high NHVvg tests showed are lower bounds on 10 the blower output to prevent visible emissions. Looking through the table, it can be seen that the blower could be set at an output of 50% (8,300 scfm) and the requirements of greater than 96.5% CE and no visible emissions could both be achieved (over the range of conditions covered in this testing). Test SR @ Equivalent SR @ Equivalent Blower Setting Test NHVvg 96.5% Blower 98% Blower to Avoid No. Condition CE Setting (%) CE Setting (%) Emissions (BTU/scf) 1 Low Very Low 10.9 60 5.15 20 -2 Low Low 16.5 50 2.91 10 -3 Low Medium 11.2 90 4.41 30 -4 Low High 12.2 95 5.36 50 30+ 5 Medium Low 18.4 75 10.5 40 -6 Medium Medium 12.4 85 3.06 30 -7 Medium High 7.57 90 4.15 60 30+ 8 High Low 8.18 60 4.84 20 -9 High Medium 10.6 85 5.81 50 -10 High High 12 100 5.52 100 40+ Table 3. Summary of stoichiometric ratios required and blower settings to ensure 96.5% CE, 98% CE, and no visible emissions for each test scenario. Test Flow Condition (lb/hr) Conversely, if the flare were to operate with a requirement of 98% CE, the restrictions on blower operation are tighter. Under low NHVvg conditions (NHVvg < 800 BTU/scf), the amount of assist air being sent to the flare needs to be kept at or below 10% output (3,500 scfm) in order to ensure a CE greater than 98% irrespective of the vent gas flow rate. However, for vent gas flow rates below 200,000 scf/day with high NHVvg above 1,550 BTU/scf, the blower settings are tightly bounded on two sides, between 30% and 50% (5,200 - 8,300 scfm), to achieve 98% CE while avoiding visible emissions. Operating the flare while requiring 98% CE would generally see the fan output kept below 50%, except in cases of high vent gas flows (> 375,000 scf/day) and NHVvg > 1,550 BTU/scf. The stoichiometric ratio is highly dependent upon the composition of the vent gas. Although the summary above focused on vent gas flow and NHVvg, the actual vent gas composition will play an important role in determining the appropriate amount of air assist to be provided. 11