The Need For Accurate Hydrocarbon Dew Point Determination

Share

Transcript

1 The Need For Accurate Hydrocarbon Dew Point Determination by Darin L. George, Ph.D., Andy M. Barajas and Russell C. Burkey, Southwest Research Institute, San Antonio, TX Figure 1 compares temperatures and pres- sures of hydrocarbon dew points measured in 2005 tests for the leanest test gas (1,050 Btu/scf) to curves computed using the certi- fied test gas composition and two common equations of state, Peng-Robinson and Soave- Redlich-Kwong (SRK). The error bars on the data points represent uncertainties in the dew point measurements, while the dashed lines on the predicted dew point curves reflect uncertainties in the certified test gas composi- tions from the gravimetric blending process. ithin the past few years, the nat- lytical methods for predict- ural gas industry has identified ing HCDPs. Because very a need that is vital both for cur- little useful HCDP data rent operations and for manage- existed for gases of interest ment of future gas supplies the accurate to the natural gas transmis- prediction of the hydrocarbon dew point of a sion and production seg- natural gas stream. ments, experiments were The hydrocarbon dew point, or HCDP, is the performed to collect HCDP temperature and pressure at which heavy compo- data on gases with a broad nents of the stream condense and begin to form range of heavy hydrocarbon liquids. It is not uncommon in some parts of the content at multiple pressures. country for ambient temperatures to cool a natu- The test apparatus was ral gas stream down to its hydrocarbon dew point built around a Bureau of and cause condensation within the pipeline. If Mines chilled mirror dew these liquids are not recovered, the heating value point tester, which is accept- they represent will be lost from the stream and ed by the industry for use in the liquids themselves may pose operational assessing gas quality and problems to equipment within the natural gas dew points. A digital video delivery system. In the future, accurate dew point camera was mounted to the Figure 1: Experimental and predicted dew point data for the data will be crucial to accommodating the intro- eyepiece of the dew scope to 1,050 Btu/scf test gas. duction of LNG and marginal gas supplies into record condensation on the the natural gas transmission network. As a result, mirror and temperature read- the HCDP is being considered as a current and outs in the field of view dur- future specification for custody transfer. ing HCDP measurements. This article presents a summary of recent The test gas blends were pre- research to improve the methods by which the pared gravimetrically by gas industry can predict hydrocarbon dew points standard manufacturers to using analytical gas compositions. As minimize the uncertainties in research continues, the results will be used by the gas compositions. the American Petroleum Institute to create an During the first experi- industry standard on HCDP determination. ments in 2002 and 2003, it was confirmed that common Origin Of The Research equations of state had signifi- Several different methods are available for cantly under-predicted the predicting hydrocarbon dew points from an dew points of the richer gas analytical gas composition. These include dif- blends. As a result, the system ferent software packages, different equations was redesigned to operate at of state and equation parameters, and different temperatures up to 250F. The methods of characterizing the heavy ends that modified test loop was suc- cannot be resolved by typical field gas chro- cessfully used in early 2005 Figure 2: Experimental and predicted dew point data for the matography. However, industry experience to complete the experiments. 1,522 Btu/scf test gas. indicated that these different methods could produce significantly different results, partic- Comparisons Of Measured Here, the confidence intervals of the meas- ularly as the percentages of hexane (C6) and And Predicted HCDPs ured data overlap the confidence interval of heavier components increased. When the measured dew points were com- the curve predicted using the Peng-Robinson As a result, Gas Technology Institute and the pared to values predicted by equations of equation, and the predicted curve is in good U.S. Department of Energy funded a research state, the agreement of the predictions with agreement with the actual dew points. The program at Southwest Research Institute actual gas behavior was found to depend both SRK equation predicted dew points 10-15F (SwRI) beginning in 2002. The purpose of the on the gas composition and on the pressure at higher than were observed in the tests. In research was to identify the most accurate ana- which the dew point was determined. comparisons to data from a 1,145 Btu/scf test Reprinted from Pipeline & Gas Journal / September 2005 / www.pipelineandgasjournal.com

2 gas (not shown here), the SRK equation of Lumped C6+ method: The amounts of all points can be accurately predicted to accommo- state tended to predict the measured values hydrocarbons with carbon numbers < 6 date anticipated changes in the natural gas sup- more accurately, while the Peng-Robinson are added together and input to the soft- ply. Current methods of predicting dew points, equation under-predicted the dew point by 10- ware as normal hexane. Many field GCs particularly software packages using the Peng- 20F at pressures of 700 psia and above. provide a complete analysis of a gas Robinson and Soave-Redlich-Kwong equations On the other hand, for richer gas blends, stream through pentane, but provide a of state, are increasingly likely to under-predict both the Peng-Robinson and SRK equations lumped value of hexane and heavier the hydrocarbon dew points of a gas stream as under-predicted the experimental dew points components that is equivalent to this the line pressure and the heating value of the gas by as much as 25F. Data collected on the 1,522 characterization. stream increase. However, for leaner gases, accu- Btu/scf gas blend, shown in Figure 2, were 10- Lumped C9+ method: Hydrocarbons such racy of predicted dew points can be improved 20F above the predictions of both equations of as normal hexane, benzene, normal hep- when a proper characterization of the heavy ends state at all pressures, from 125 psia up to 1,250 tane, toluene, etc. are specified individual- is used. psia. Overall, the comparisons provided evi- ly in the gas composition used for dew Although the exact composition of a gas dence of a trend among predicted hydrocarbon point calculations. The amounts of all stream may not be available, research has dew points: as the line pressure and heating hydrocarbons with carbon numbers < 9 found that obtaining and using accurate data on value of a natural gas stream increase, existing are added together and assigned to nor- the gas composition through C9 will allow pre- equations of state are more likely to under-pre- mal nonane. diction of true hydrocarbon dew points to with- dict the HCDP of the stream. GPA 60/30/10 method: This method treats in 25F in many cases. Evidence to date sug- the lumped C6+ amount as 60% n-C6, gests that the best characterization method may Further Evaluation Of HCDP 30% n-C7, and 10% n-C8. This method depend on the actual composition of the gas for Prediction Techniques was developed for accurate calculations of which the dew point is being predicted. Clearly, improvements are needed in the tools natural gas density, but is often used for Additional work in 2005 is being funded by available to the natural gas industry for predict- predicting dew points as well. Pipeline Research Council International to ing hydrocarbon dew points. The data complete the evaluation of analyti- collected during the experiments at cal dew point prediction methods. SwRI and data found during earlier This work will investigate possible literature searches are now being used causes for the differences between in further research to identify the experimental and calculated dew most accurate prediction methods points. When the evaluations are available. Related work has already complete, the results will be used to begun at SwRI to assess the accuracy create industry guidelines for pre- of various methods for predicting dicting analytical or operational hydrocarbon dew points and to pro- hydrocarbon dew point tempera- vide this information for use in future tures. industry standards. The findings of the research pro- Some key information has gram to date have been published, already been gathered from the and are available to benefit the study. While common equations of industry in such areas as LNG state can under predict hydrocarbon accommodation and accurate spot dew points even with complete sampling techniques. Interested knowledge of the gas composition, readers can obtain a complete listing other details of the dew point calcu- Figure 3: Comparisons of dew points com- of all research reports and technical lation may lead to additional sources of error. puted for the 1,145 Btu/scf test gas using papers from the GTI website (www.gastech- For instance, several different software pack- several common heavy-end characteriza- nology.org) or the Metering Research Facility ages are available to the industry to perform tions and the SRK equation of state. website (www.grimrf.org). dew point calculations, and each of these may Authors: Darin L. George, Ph.D., is a yield different predictions using the same gas Figure 3 shows one example of the calcu- Senior Research Engineer at Southwest compositions, equations of state and pressure lated dew point curves produced by these dif- Research Institute (SwRI) in San Antonio. conditions. ferent characterizations of the same gas Georges background as a nuclear and mechan- To investigate the impact of using different stream. Using a C6+ characterization instead ical engineer spans 15 years of work on the software packages, dew points were computed of a full characterization containing all known measurement and control of liquid, gas and using several common programs. Given the components of the gas was found to change multiphase flows. He is project manager of the same gas compositions, line pressures and the computed dew point by as much as 70F, gas quality and gas sampling research program equations of state, all the software packages and invariably led to under-prediction of the at SwRIs Metering Research Facility. examined by SwRI predicted dew point tem- dew point. The characterization method that Andy M. Barajas is a Principal Engineer in peratures that agreed to within 5F of one most closely predicts experimental dew points the Multiphase Flow Section at SwRI with another a relatively small variation when varied from one test gas to another. professional interests in the areas of heat compared to the 25F margins of error seen Based on the comparisons to date, however, transfer, fluid dynamics, and multiphase flow. earlier. Similarly, differences in dew points the C9+ characterization most often appears to Barajas has designed and built a general pur- predicted by the Peng-Robinson and SRK predict measured dew points to within 25F. pose, higher-pressure natural gas multiphase equations of state, all else being the same, This is consistent with the opinion of many flow facility at SwRI, where he has conduct- were within 5F. researchers that the composition of a gas stream ed research on flow assurance, wet-gas meter- One detail of dew point prediction known to must be known through at least C9 to accurately ing, and multiphase flow metering. have a large impact on accuracy is the method predict HCDPs. Some researchers advocate ana- Russell C. Burkey is a Research Engineer in used to characterize the distribution of heavy lyzing the gas stream through C12 to compute the the Multiphase Flow Section at SwRI with pro- hydrocarbons in the gas stream when the exact dew point accurately, though this requires analy- fessional interests in mechanical design and composition beyond C6 cannot be resolved by sis with a laboratory GC instead of a field GC. fluid mechanics. Burkey also serves as the field gas chromatography. Characterization Quality Coordinator of the Flow Component methods commonly used to fill in the blanks Conclusion Testing Facilities (FCTF) at SwRI, providing of an unknown composition have also been test- Work is continuing to advance the industrys quality system support for the API surface and ed in studies at SwRI. These include: knowledge of natural gas HCDPs so that dew subsurface safety valve testing program. Reprinted from Pipeline & Gas Journal / September 2005 / www.pipelineandgasjournal.com

Load More