|Title||Hydrogen and syngas flares-comparison between experimental data and model predictions|
|Spatial Coverage||Houston, Texas|
|Subject||2014 AFRC Industrial Combustion Symposium|
|Description||Paper from the AFRC 2014 conference titled Hydrogen and Syngas Flares-comparison Between Experimental Data and Model Predictions by Derek Miller.|
|Abstract||Accurate prediction of thermal radiation from hydrogen and syngas flares is critical for safe operation in a number of industries. Published models available today are primarily based on natural gas and other hydrocarbon mixes. It is unclear how suitable these models are for other gases. Air Products is a large producer of hydrogen and syngas and this paper presents the results of full-scale tests recently completed by the company, along with a study of other previously published experimental data. Results are compared against predictions from published models, where significant deficiencies are identified. Results are also compared against predictions from a recently developed model, specifically designed to overcome these deficiencies.|
|Rights||No copyright issues exist.|
1 | P a g e Hydrogen and syngas flares-comparison between experimental data and model predictions Derek Miller, John Bedenbaugh Air Products and Chemicals, Inc. 7201 Hamilton Blvd., Allentown, PA 18195 email@example.com AFRC 2014 Industrial Combustion Symposium 1.0 Introduction 1.1 Safe flaring of both hydrogen and syngas (hydrogen / carbon monoxide mixtures) is essential to many industrial operations and having the ability to predict flame geometry and thermal radiation with reasonable accuracy is a critical aspect of flare design and operation. 1.2 Previous work by Air Products  identified weaknesses in the use of existing industry standard flare models ,  for both hydrogen and syngas, when compared against published data. These models are primarily based on data from hydrocarbon flares and do not do particularly well at predicting flame geometries or resulting heat fluxes for these other gases. Air Products has therefore decided to develop a new model called AP Flame and validate this model by collecting new data at one of the company's own facilities. Test results are presented here. Details of the model itself will be the subject of a subsequent paper. 1.3 Work was performed at an Air Products owned & operated hydrogen production facility in Louisiana, USA during a planned shutdown in January, 2014. Two tests were conducted, the first with pure hydrogen, the second with reformer syngas. Radiation levels were measured at two locations, one in the near field on an elevated platform close to the flare, and the other in the far field at grade at the facility fence line. Cameras were used to record visible and infrared still images along with high-definition video, to determine flame shape and tilt angle from vertical. Results were compared against industry standard models and the new model being developed by Air Products. 1.4 John Zink Hamworthy Combustion, the supplier of the flare, supported Air Products by provided equipment, personnel and expertise that contributed greatly to the success of the test. 1.5 The flare height is 175 ft with an effective tip diameter of 27". 2.0 Test details 2.1 Flare testing was performed during a planned facility shutdown, during which two 15 minute time periods were set aside for controlled flaring of hydrogen and then syngas. 2.2 Two locations were selected for the fixed radiometers, as shown in Figure 1, chosen to be orientated perpendicular to the nominal prevailing wind direction (East - West), to provide a side-on view of the flames (radiometers face north). Near field measurements (within about 2 flame lengths) have the advantage of higher heat fluxes, but far field measurements (further than about 3 flame lengths) are less sensitive to variations in flame geometry, wind etc. Consequently, one location was selected as being in the near field and the other in the far field. Video and IR cameras were located at the far field location. (in Figure 1, R is the distance from the radiometer to the predicted center of the flame, Lf is the predicted flame length) 2 | P a g e Figure 1. Overview of hydrogen plant, & flare highlighting near and far field data collection locations. Figure 2. Views of flare from near field and far field locations. 3 | P a g e 3.0 Special considerations 3.1 Obtaining images of non - luminous flames: Both hydrogen and syngas flames are almost invisible to the naked eye during day light hours (the tests were performed during the day) but both are quite visible in the IR spectrum, which is why an IR camera was used. Hydrogen and syngas flames do have some visibility at night. A comparison of images obtained prior to the test of smaller flames at night confirmed that IR and visible images are similar in both size and shape. 3.2 Wind speed and bearing (direction): During the 15 minute tests, wind speed and direction were continually measured. Wind bearing was defined as being 90 degrees when in the east - west direction. Wind speed generally increases with elevation and a significant difference in wind speed was expected between the 10m measurement point and the top of the flare. Wind speed adjustment for elevation is usually made using the power law relationship, Equation 1. The power factor, α being a function of the local terrain. The location of this particular facility is such that the prevailing wind comes from an area of flat open ground and a factor of α = 1/7 (traditional value for open terrain) was used. u is the wind velocity at the elevation of interest, ur is the velocity at the reference elevation, h is the elevation of the top of the flare, hr is the reference elevation, in this case 10m. = Equation 1. 3.3 Background solar radiation: total thermal radiation measured at a location will be the sum of the radiation from the flare plus radiation from the sun. Hence, whenever flare radiation is measured, the background solar radiation must be subtracted, to obtain the true flare radiation. Conveniently on the day of the test, conditions were overcast (no rain) and the orientation of the radiometers was pointing away from the sun, no appreciable background solar radiation was measured and so no subtraction was required. 3.4 Determination of true flame length and tilt: In order to determine the true geometry of the flame from photographs, it is essential to have a reference vertical and a reference length. For these tests, the flare stack itself provided the vertical reference and the tip section of the flare, from the bottom of the molecular seal to the tip of the flare, provided the scale (for this flare ~ 21.5 ft). Given the relative orientation of the camera to the flare and the impact of wind bearing, it is important to correctly adjust the values obtained from the photograph for both flame length and tilt to allow for the impact of perspective. All data presented here has been adjusted. 3.5 Variation in conditions during test. As expected, some variation in wind speed, bearing and gas flow was experienced during both tests. Given the resulting uncertainties and complexities, it was ultimately decided to use simple averaged values over each 15 minute test as the basis for the subsequent analysis of results. 4 | P a g e Figure 3. Thermal radiation and IR image collection at far field location. Figure 4. Flare tip showing dimension used for scaling photograph images 5 | P a g e 4.0 Summary of test results 4.1 Key test parameters Table 1. Test result key parameters. Hydrogen test Syngas test Flare tip elevation 175 ft Effective flare diameter 27 in. Release rate 13,800 lb/hr (2.6 million SCFH) 103,400 lb/hr (3.7 million SCFH) Mol wt 2 10.7 Release temperature 100 °F 100 °F Total firing rate 702 million btu/hr (206 MW) 999 million btu/hr (293 MW) Initial jet velocity 193 ft/s 276 ft/s Reference wind speed at 10 m 5.1 mph 4.4 mph Wind speed at flare tip 9.5 ft/s 8.2 ft/s Wind bearing 90° 100° Ambient temperature 52°F 57°F Relative humidity 97% 90% x,y,z far-field radiometer (1) x = 0 ft, y = 4 ft, z = 240 ft x = -42 ft, y = 4 ft, z = 236 ft x,y,z near-field radiometer (1) x = 0 ft, y = 80 ft, z = 60 ft x = -10 ft, y = 80 ft, z = 59 ft (1) x horizontal downwind direction, y vertical distance above grade, z horizontal distance perpendicular to wind direction. For a fixed radiometer location, x and z dimensions are sensitive to wind bearing. Table 2. Gas compositions Composition (Mole %) Mol wt Hydrogen Syngas CH4 16.04 ------ 6.45 H2 2.02 99.90 73.36 CO 28.01 ------ 3.96 CO2 44.01 ------ 15.88 N2 28.02 0.10 0.13 H2O 18.02 ------ 0.22 6 | P a g e 4.2 heat flux 4.2.1 Hydrogen test Figure 5. Thermal radiation for far and near 4.2.2 Syngas test Figure 6. Thermal radiation for far and near 4.3 Flow 4.3.1 Hydrogen. The hydrogen flow reduced test. Superimposing normalized flow against far field showed an excellent correlation. only shows a relatively small amount of the full scale far and near field. field locations during hydrogen test far and near field. field locations during syngas test . somewhat about halfway through the normalized radiat Note: the vertical scale in the figure of 0 - 1. radiation at near and he 7 | P a g e Figure 7. Normalized flow and thermal radiation during hydrogen test. 4.3.2 Syngas. Syngas flow d 4.4 Wind speed and direction 4.4.1 Hydrogen. Wind speed at the reference elevation (10m) varie during the test, but ov comparing video of the flame reported from the local weather meter that the meter appeared to be giving an inaccurate reading with an offset of about 30 degrees. For purposes an average Figure 8. Wind speed and bearing during hydrogen test 4.4.2 Syngas. Wind speed during this test showed a similar trend to the data collected during the hydrogen test, although with a slightly lower average value. the observed meter and the video of the flame heat striations, for purposes of analysis, an average bearing of 100 degrees was . during the test was relatively steady (not shown here) . varied overall was reasonably consistent. It was found, when visible heat striations with the wind bearing n bearing of 90 degrees was used. . error between the wind bearing reading from the local weather used. uring d as expected . of analysis, Given 8 | P a g e Figure 9. Wind speed and bearing during 4.5 Infrared flame images (examples) 4.5.1 Hydrogen. Figure 10. IR images of hydrogen flame 4.5.2 Syngas. Syngas flames have low tilt and are fairly "tight". syngas test . Flames have significant tilt and are fairly wide and "loose" . 9 | P a g e Figure 11. IR images of syngas flame 5.0 Comparison between measured values and model predictions. 5.1 Hydrogen. Table 3. Comparison between test data and model prediction for hydrogen test Data AP Flame Chamberlain  API 521  Brzustowski & Sommer  (4) Flame length, ft 85 84 83 100 345 Flame tilt, deg 50 40 22 48 76 Near radiometer, btu/hr/ft2 410 350 610 440 92 Far radiometer, btu/hr/ft2 69 68 120 95 63 10 | P a g e 5.2 Syngas. Table 4. Comparison between test data and model prediction for syngas test Data AP Flame Chamberlain  (2) API 521  (3) Brzustowski & Sommer  (4) Flame length, ft 82 84 84 120 165 Flame tilt, deg 9 7 31 37 30 Near radiometer, btu/hr/ft2 240 310 790 513 232 Far radiometer, btu/hr/ft2 54 62 160 122 64 (2) Note that in previous work  it was identified that when using the Chamberlain model, the use of the simplified method for determining fuel mass fraction presented in the paper is not appropriate for non paraffin fuels and that the full stoichiometric calculation need by performed instead (method used in this analysis). This is clearly stated in the paper , but the simplified equation is still currently used in some commercial software for non paraffin fuels. (3) API 521 does not provide a value for the radiant fraction for syngas and so the same value as recommended for hydrogen is used (4) Brzustowski & Sommer method provided as alternative in API 521. Radiant fraction same as API 521. Assume Lower Flammable Limit of syngas same as pure hydrogen. Model only predicts location of flame radiant center. Flame tilt determined from this location, relative to flare tip. With the assumption that radiant center is 2/3 along flame, overall predicted flame length is determined. 6.0 Conclusions 6.1 The flare testing was successful both in obtaining good near and far field radiation measurements and also in recording flame geometry for both hydrogen and syngas flares. 6.2 As far as the authors are aware, this is the first time a full set of data has been collected and published on full-size hydrogen or syngas industrial flares. 6.3 Although the syngas tested contained over 70% hydrogen, flame shape and radiation were quite different than 100% hydrogen. The density of this syngas is about five times as high as hydrogen and along with a higher initial jet velocity, has much higher initial vertical momentum. Although syngas firing rate is higher than for hydrogen, measured heat fluxes are lower, indicating that the radiant fraction for syngas is lower than for hydrogen. 6.4 Comparison with predictions from currently available flare models confirms previous observations that these models are not really adequate to fully model either hydrogen or, in particular, syngas 6.5 The newly developed AP Flame model performed quite well against this test data, with acceptable accuracy in both near and far field. AP Flame takes account of initial jet momentum and determines radiant fraction as a function of flame temperature and combustion product composition. 7.0 Acknowledgements 11 | P a g e Air Products would like to thanks John Zink for their extensive help and on-site support during this test program. 8.0 References  Miller, D., Lutostansky, E., Jung, S., "Applicability of Currently Available Flare Radiation Models for Hydrogen and Syngas," AIChE 2014 Spring Meeting.  Chamberlain, G.A., "Developments in Design Methods for Predicting Thermal Radiation from Flares," Chemical Engineering Research and Design, Volume 65, 1987.  API STANDARD 521, Pressure-Relieving and Depressuring Systems, 6th ed., American Petroleum Institute, Washington DC, 2014.