Industrial Flare Performance at Low Flow Conditions. 2. Steam- and

Feb 27, 2012 - Full-scale tests of steam- and air-assisted industrial flares were conducted using low BTU content (lower heating value) vent gases at ...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/IECR

Industrial Flare Performance at Low Flow Conditions. 2. Steam- and Air-Assisted Flares Vincent M. Torres,*,† Scott Herndon,‡ and David T. Allen† †

Center for Energy and Environmental Resources, The University of Texas at Austin, 10100 Burnet Road, Building 133, M.S. R7100, Austin, Texas 78758, United States ‡ Aerodyne Research, Inc., 45 Manning Road, Billerica, Massachusetts 01821-3976, United States S Supporting Information *

ABSTRACT: Full-scale tests of steam- and air-assisted industrial flares were conducted using low BTU content (lower heating value) vent gases at low flow rates. A 36″ diameter steam-assisted flare with a flow capacity of 937,000 lb/h and a 24″ diameter air-assisted flare with a flow capacity of 144,000 pounds per hour were operated with mixtures of natural gas, propylene, and nitrogen or natural gas, propane, and nitrogen at flow rates less than 1% of maximum flow. Combustion efficiency (percentage of the flared gases converted to carbon dioxide and water) ranged from less than 50% to more than 99%. For the steam-assisted flare, combustion efficiency (CE) at low steam-to-vent gas flow ratios (0.5−1.0) was typically in excess of 95%. CE would gradually decrease as steam-to-vent gas ratio increased, to a point, after which CE would decrease dramatically. The steam-to-vent gas ratio at which CE would decrease dramatically depended on the heating value of the vent gas and the position of the steam injection. Higher heating values of the vent gas (600 vs 350 BTU/scf) and the minimization of steam coinjected with the vent gas, rather than injected at the flare tip, promoted higher CE. For the air-assisted flare, CE at low air assist rates ( 70% was used because flames with lower DREs/CEs were likely to display greater variability in DRE/CE depending on the location of the extractive sampler, while flames with higher DREs/CEs generally showed consistent DRE/CE values at a variety of sampling positions within the flame.4 The reproducibility is characterized by the ratio of the standard deviation divided by the mean, for both DRE and CE. This ratio averaged 0.021 and 0.024 for DRE and CE, respectively (standard deviation less than 2.5% of the mean) but increased as the variability in wind speed increased and as the overall DRE and CE decreased. Based on these data, it can be concluded that reproducibility is approximately ±2.5% of the mean. This represents an upper bound on the variability of true replicate tests because (i) vent gas flow rates and steam flow rates typically varied by ∼1% between replicate tests and (ii) ambient conditions, especially wind speed, varied between tests. Additional discussion of variability and uncertainty is provided by Herndon et al.3 Overall, analysis of replicate tests indicates that uncertainty in DRE and CE measurements are likely within bounds of ±2.5%. This is much less than the range of observed DREs and CEs, which included values less than 50% and greater than 99%. The analysis of replicates also indicates that the impact of wind speed on CE and DRE for steam flares is within ±2.5%, at least for the ranges of wind speeds in Table 1. More detailed analysis of the role of wind speed is being conducted using computational modeling of the flames6 and will be reported on in subsequent publications. Impact of Flare Operating Conditions on DRE and CE. Figure 1 shows the dependence of CE on steam to vent gas flow for test series S3−S11. For the steam flare tests with a 350 BTU/scf vent gas and approximately 500 lb/h center steam, CE drops below 90% at roughly a steam-to-vent gas ratio of 0.5 (Figure 1, upper panel). In contrast, at 600 BTU, a steam-to-vent gas ratio in excess of ∼1.2 is necessary to reduce CE below 90% (Figure 1 bottom panel). Using a ratio of total steam- (upper steam plus center steam) to-vent gas ignores the different impacts associated with center steam and upper steam addition. Figure 1 (middle panel) shows the CE as a function of steaming rate for test series with zero center steam. Comparison of upper and middle panels in figure shows that center steam has a greater impact in reducing CE than an equivalent amount of upper steam. For the 350 BTU/scf vent gas, at a steam-to-vent gas ratio of 1, with both upper and center steam, CE is roughly 30%. If the same steamto-vent gas ratio is used but all of the steam is upper, (zero center steam), the CE is roughly 80%. The differing impacts of center and upper steam are also evident in comparisons of test series S7 and S9 and test series S10 and S11. Test series S7 and S9 both had targeted steam flows of 1025 lb/h; however, in test series S9, no center steam was used. As shown in Table S8 and Figure 2 (upper panel), CE decreased more rapidly when center steam was used for half of the targeted steam flow (S7), compared to a case with no center steam but an equivalent total steam flow (S9). Comparison of test series S10 and S11 (Figure 2 lower panel) shows a similar pattern. In this case, test series S10 had zero center

conditions and wind speeds over the course of each test (typically 7−10 min per test). Air-Assisted Flare. For the air-assisted flare, three types of test series were performed (A1-A2; A3-A6; and A7). Test series A1 and A2 were initial equipment test runs, conducted with pure propylene vent gas. Since the focus of the overall test program was on low LHV vent gas operation at low flows, these results will not be reported here; however, the results are available in the Final Report for the flare test project.4 Test series A3-A6 were conducted to examine the impact of vent gas flow rate, vent gas lower heating value, and air assist rate on combustion characteristics. For each test series, an LHV of the vent gas was selected (targeted values of either 340 BTU/scf or 560 BTU/scf). For each LHV, a test series was run at each of two vent gas flow rates (targeted values of either 900 or 350 lb/h). This combination of 2 LHVs at each of the 2 vent gas flow rates resulted in the four test series, A3-A6. For each test, a variety of air assist flow rates were used. Air assist was varied from a low value corresponding to the Incipient Smoke Point to a high value at which CE had dropped substantially below 80%. The procedure used to determine the ISP and the snuff point based on field observations is described in detail in the final report.4 Replicate tests (replicating operating conditions, not necessarily ambient conditions) were performed for at least one test in each test series. Table S4 summarizes operating conditions for the tests performed in test series A3-A6. Values shown in the table represent time-averaged values of flare operating conditions and wind speeds over the course of each test (typically 7−10 min per test). Test series A3-A6 were all conducted using a 4:1 mixture (volume basis) of propylene and natural gas as the vent gas. Test series A7 was conducted with a 4:1 mixture (volume basis) of propane and natural gas. The goal of test series A7 was to replicate tests series A5 but replacing propylene with propane. For test series A5, a lower heating value (LHV) of the vent gas of ∼360 BTU/scf was used with a vent gas flow of ∼370 lb/h. Table S5 summarizes the flare operating conditions used for the tests performed in test series A7. Values shown in the table represent time-averaged values of flare operating conditions and wind speeds over the course of each test (typically 7−10 min per test).



RESULTS AND DISCUSSION Steam-Assisted Flare. Tables S6−S8 present data on the combustion characteristics observed for each of the tests. The tables report combustion efficiency (CE, %), destruction/ removal efficiency (DRE, %), steam-to-vent gas ratio (mass basis), the ratio of CO to CO2 in the plume, and the fraction of the carbon in the feed propylene/propane that is converted to CO or CO2. For this final parameter, it is assumed that the CO and CO2 in the plume are due partly to the propylene/propane in the feed and partly due to the methane. The fractions due to propylene/propane and to methane are assumed to be equal to the fraction of carbon in the feed due to propylene/propane and methane. The results in Tables S6−S8 will be examined to identify impacts on flare combustion characteristics of parameters such as steam/vent gas ratio and fuel gas heating value. Before examining those issues, however, it is useful to examine the degree to which replicate flare operating conditions produced comparable combustion characteristics, even as ambient conditions, such as wind speed, varied. 12571

dx.doi.org/10.1021/ie202675f | Ind. Eng. Chem. Res. 2012, 51, 12569−12576

Industrial & Engineering Chemistry Research

Article

Figure 1. (Top panel) Combustion efficiency as a function of steam-to-vent gas ratio for vent gases with 350 BTU/scf and approximately 500 lb/h center steam (S4 − red square, S7 − blue diamond, except S4.4R1). (Middle panel) Combustion efficiency as a function of steam-to-vent gas ratio for vent gases with 350 BTU/scf and zero center steam (test series S8 − red square, S9 − green triangle, S10 − blue diamond). (Bottom panel) Combustion efficiency as a function of steam-to-vent gas ratio for vent gases with 600 BTU/scf and approximately 500 lb/h center steam (test series S5 − green triangle, S6 − blue diamond).

Figure 3 shows CE as a function of vent gas flow rate for these three test series. At vent gas flows below approximately 2,000 lb/h, there is little difference between the three tests. At vent gas flows below 2,000 lb/h, the results diverge, with more center steam resulting in a more rapid decrease in CE.

steam, while test series S11 had roughly a third of the total steam flow as center steam. Test series S7, S8, and S11 provide additional insight into the impact of center steam. In these test series (Figure 3), upper steam was held constant at ∼500 lb/h, while center steam was set at 0 (S8), ∼250 lb/h (S11), and ∼500 lb/h (S7). 12572

dx.doi.org/10.1021/ie202675f | Ind. Eng. Chem. Res. 2012, 51, 12569−12576

Industrial & Engineering Chemistry Research

Article

Figure 2. (Upper panel) Combustion efficiency as a function of vent gas flow rate for flaring with 1,000 lb/h steam; green triangles have no center steam (S9) and blue diamonds have 500 lb/h center steam (S7). (Lower Panel) Combustion efficiency as a function of vent gas flow rate for flaring with ∼825 lb/h steam; blue diamonds have no center steam (S10) and turquoise circles have ∼250 lb/h center steam (S11).

Figure 3. Combustion efficiency as a function of vent gas flow rate for flaring with 500 lb/h upper steam; 500 lb/h center steam (S7) shown as blue diamonds; 250 lb/h center steam (S11) shown as turquoise circles; and 0 center steam (S8) shown as red squares.

Impact of Vent Gas Composition on CE. Test series S3−S11 were conducted with vent gas mixtures of propylene, natural gas, and nitrogen, with propylene and natural gas

maintained at a 4:1 volume ratio. To assess the impact of vent gas composition on flare combustion characteristics, test series S12−S14 were conducted with mixtures of propane, natural 12573

dx.doi.org/10.1021/ie202675f | Ind. Eng. Chem. Res. 2012, 51, 12569−12576

Industrial & Engineering Chemistry Research

Article

Figure 4. Comparison of CE, as a function of steam-to-vent gas ratio for propylene/methane vent gases (S7, S8, and S11; blue diamonds) and propane/methane vent gases (S12 − S14; red squares).

Table 2. Data for Air Flare Tests Where Replicate Flare Conditions Were Used in 3 Tests test series A3.1 A3.4 A3.6 A4.1 A4.3 A4.4 A4.5 A5.1 A5.3 A5.5 A6.1 A6.3 A6.4

average standard DRE deviation/ average (%) average DRE CE (%) 98.9 76.6 90.8 97.9 95.2 91.1 87.8 96.4 83.7 93.9 99.6 97.1 92.9

0.008 0.003 0.017 0.005 0.014 0.006 0.007 0.015 0.040 0.017 0.002 0.008 0.024

98.3 72.0 88.2 97.1 93.6 88.5 84.5 96.0 80.8 92.7 99.3 95.6 89.2

standard deviation/ average CE

range of wind speeds (mph)

0.010 0.005 0.022 0.007 0.018 0.006 0.007 0.014 0.046 0.021 0.002 0.013 0.029

10.3−13.4 (Δ=3.1) 10.4−12.7 (Δ=2.3) 11.9−13.1 (Δ=2.2) 9.1−12.4 (Δ=3.3) 9.2−15.2 (Δ=6.0) 9.6−14.3 (Δ=4.7) 9.5−14.5 (Δ=5.0) 2.7−7.2 (Δ=4.5) 2.5−5.3 (Δ=2.8) 2.7−5.6 (Δ=2.9) 13.8−16.0 (Δ=2.2) 15.1−15.6 (Δ=0.5) 12.3−15.4 (Δ=3.1)

gas, and nitrogen, with propane and natural gas maintained at a 4:1 volume ratio. Test series S12−S14 were designed to parallel test series S7, S8, and S11. For each test series, vent gas LHV was maintained at ∼350 BTU/scf, and upper steam was held constant at ∼500 lb/h. Center steam was set, for the three test series, at 0, ∼250, and ∼500 lb/h. Within each test series, multiple vent gas flow rates were used. Figure 4 shows comparisons between the six test series (3 propane and 3 propylene). The behavior of the flare CE for propane and propylene dominated vent gases is consistent at all levels of CE when steaming at 500 lb/h is used for both upper and center steam. At lower center steam flows, CE is consistent above 90%, but in test series S13, the propane vent gas showed a more rapid decrease in CE than the propylene vent gas as steam-to-vent gas ratio increased. This difference between propylene and propane dominated vent gas is restricted to a single test series (S13.4), but three replicates were conducted for this test series. Overall, comparable test series with propane and propylene vent gas produced very similar flare combustion characteristics, especially at CE > 90%. Air-Assisted Flare. Tables S9 and S10 present data on the combustion characteristics observed for each of the air-assisted

Figure 5. (Top panel) Combustion efficiency as a function of stoichiometric ratio for vent gases with 340 BTU/scf (A3, ∼900 lb/h, red squares; A5, ∼350 lb/h, blue diamonds). (Lower panel) Combustion efficiency as a function of stoichiometric ratio for vent gases with 560 BTU/scf (A4, ∼900 lb/h, red squares; A6, ∼350 lb/h, blue diamonds).

flare tests. Table 2 presents data on replicate tests with CE > 70%. The reproducibility, characterized by the ratio of the standard deviation divided by the mean, was less than 1.5%., with no distinct pattern associated with either wind speed or DRE/CE. This result is slightly more precise than observed for the steam-assisted flare, likely because it was possible to more precisely control the air assist rate than the steam assist rate, reducing the variability in operating conditions for replicate tests. Impact of Flare Operating Conditions on DRE and CE. Figure 5 shows the dependence of CE on air to vent gas flow using the stoichiometric ratio for test series A3-A6. The stoichiometric ratio is the ratio of the actual amount of air assist 12574

dx.doi.org/10.1021/ie202675f | Ind. Eng. Chem. Res. 2012, 51, 12569−12576

Industrial & Engineering Chemistry Research

Article

Figure 6. Comparison of CE, as a function of stoichiometric ratio, for propylene/methane vent gases (A5; blue diamonds) and propane/ methane vent gases (A7; red squares).

Table 3. Steam Flare Products of Incomplete Combustion for Vent Gas with 4:1 Propylene/Natural Gasa species propylene CO acetylene ethylene butene isomers formaldehyde acetaldehyde propanal acrolein methanol acetone propylene-oxide methane ethane a

98 < DRE 95 < DRE < 98 80 < DRE < 95 DRE < 80 1 1.57 0.0228 0.0388 0.000321 0.0588 0.0372 0.00109 0.0289 0.00132 0.00131 0.00164 0.132 0.0103

1 0.671 0.0164 0.0278 0.000274 0.0384 0.0226 0.00100 0.0233 0.00135 0.00120 0.00151 0.132 0.0103

1 0.434 0.0162 0.0199 0.00208 0.0265 0.0153 0.000853 0.0125 0.00109 0.00102 0.000711 0.132 0.0103

1 0.166 0.0104 0.00907 0.00732 0.0102 0.00541 0.000532 0.00585 0.000780 0.000638 0.000266 0.132 0.0103

lb per lb propylene in flare plume.

(lb/h) to the amount of stoichiometric air (lb/h) required for combustion of the vent gas. For the air flare tests with a 350 BTU/scf vent gas and approximately 360 BTU/scf, CE drops below 80% at roughly a stoichiometric ratio of 25 (Figure 5, upper panel). In contrast, at 560 BTU, a stoichiometric ratio of 30 is necessary to reduce CE below 80% (Figure 5 bottom panel). Comparison of the two panels in Figure 5 shows that the vent gas flow rate had a smaller effect on CE than vent gas LHV. Impact of Vent Gas Composition on CE. Test series A3-A6 were conducted with vent gas mixtures of propylene, natural gas, and nitrogen, with propylene and natural gas maintained at a 4:1 volume ratio. To assess the impact of vent gas composition on flare combustion characteristics, test series A7 was conducted with mixtures of propane, natural gas, and nitrogen, with propane and natural gas maintained at a 4:1 volume ratio. Test series A7 was designed to parallel test series A5. Figure 6 shows comparisons between the test series. Overall, comparable test series with propane and propylene vent gas produced very similar flare combustion characteristics, especially at CE > 90%. Products of Incomplete Combustion. Table 3 provides data on the mass concentrations of PICs relative to the mass concentration of propylene in the flare plume for the steam flare test series conducted with propylene. Data are sorted by DRE. As shown in Figure 7, as DRE decreases, the formation

Figure 7. Percentage of carbon in propylene feed going to CO2, unburned hydrocarbons, CO, and other PICs, for steam-assisted test series S7.1, S7.3, and S7.4.

of both PICs and CO increases. However, as DRE falls below 90%, PIC and CO formations plateau. Similar characteristics were observed for the air-assisted flare. As the DRE decreases below 90%, the major non CO2 species is propylene and not PICs and CO.



CONCLUSIONS

Full-scale tests of air- and steam-assisted industrial flares were conducted using low BTU content vent gases at low flow rates. Combustion efficiency (percentage of the flared gases converted to carbon dioxide and water) in the flare tests ranged from less than 50% to more than 99%. For the steam-assisted flare tests, CE at low steam-to-vent gas flow ratios (0.5−1.0) was typically in excess of 95%. CE would gradually decrease as steam-to-vent gas ratio increased, to a point, after which CE would decrease dramatically. The steamto-vent gas ratio at which CE would decrease dramatically depended on the LHV of the vent gas and the position of the steam injection. Higher LHVs of the vent gas (600 vs 350 BTU/scf) and the minimization of center steam promoted higher CE. 12575

dx.doi.org/10.1021/ie202675f | Ind. Eng. Chem. Res. 2012, 51, 12569−12576

Industrial & Engineering Chemistry Research

Article

For the air-assisted flare, CE at low air assist rates (