Patrick J. Nugent Executive Director August 31, 2009 Via email dnesvaci@tceq.state.tx.us Danielle Nesvacil Industrial Emissions Assessment Air Quality Division, MC-164 Texas Commission on Enviromnental Quality 12100 Park 35 Circle Austin, Texas 78753 Re: "Upstream Oil and Gas Storage Tank Project Flash Emissions Models Evaluation," report by Hy-Bon Engineering Company Dear Ms. Nesvaci1: The Texas Pipeline Association ("TPA") appreciates the opportunity to submit the following comments regarding the draft report entitled "Upstream Oil and Gas Storage Tank Project Flash Emissions Models Evaluation," dated July 16, 2009, prepared by Hy-Bon Engineering Company, Inc. TPA is a trade association whose members gather, transport, and process natural gas and liquids through intrastate pipelines in Texas. As discussed below, TPA has many technical concerns about the report. The report in its current fonn is flawed and should not fonn the basis for policy decisions by TCEQ at this point in time. In addition, we have general concerns about the future use of the report and whether the consultant that perfonned the study had the necessary expertise to compare the captured data with the modeled results. TPA's general comments are as follows: 1. The applicability of any policies developed as a result of the report should be limited to "upstream" oil and gas storage tanks - as its title suggests. As an initial matter, we would note that the title of the report is "Upstream Oil and Gas Storage Tank Project Flash Emissions Models Evaluation." If and when the report is considered as the basis for possible policy decisions by TCEQ, the report's effect should be limited to VOC emissions from condensate tanks at upstream production sites. The report should have no effect on policies governing the calculation of emissions other than VOC emissions, nor should it have any effect on policies or rules governing storage tank sites, other than those at upstream production sites. TPA members' operations, which consist largely of mid-stream gas transmission, do not account for significant amounts ofVOC flash emissions because of the low throughput on midstream condensate tanks. We would like to take this opportunity to make the distinction between the likelihood and quantity of flash emissions occurring from condensate storage tanks located upstream versus those located downstream of a gas processing plant. 2 Downstream of a gas processing plant, flash emissions from condensate tanks are de minimis. Mainline transmission of natural gas usually commences at the outlet point of a gas processing plant. At this point, most if not all of the water and liquid hydrocarbons have been removed from the gas stream as the gas has been processed to dry, pipeline-quality sweet natural gas. The "dry" gas is then transported to a point of sale where it is used as fuel by an end user or is sold for distribution. At the delivery point for sale or distribution, the pressure in the pipeline is dropped and, if there are any liquids in the gas stream, they will drop out 'at this point and "flash." It is the experience of TPA members that flash emissions at this point in the transportation/delivery process are de minimis. In fact, those of our members who have been asked to perform flash emission calculations, as a follow-up to an Emissions Inventory Report or otherwise, have demonstrated that the VOC flash emission calculations are so small that they constitute a zero value. Accordingly, we would request that TCEQ recognize the distinction between the likelihood and quantity of flash emissions occurring from condensate storage tanks located upstream versus those located downstream of a gas processing plant in any future policies that may be developed by reference to this report. 2. TCEQ should examine whether it is appropriate for Hy-Bon to serve as anthor of a report such as this. Hy-Bon's expertise lies in detecting and capturing gas leaks. This expertise does not necessarily translate into an ability to analyze the captured data and compare that data with various emission models. TCEQ should examine whether Hy-Bon has any particular expertise in the subject matter of this report, i.e. evaluating the merits and demerits of various emissions models. In addition, TCEQ should consider whether Hy-Bon's financial interests are served by a conclusion that direct measurement is preferable to emissions modeling methods. TPA's specific comments are as follows: 1. The report assumes, without basis, that any inconsistency between measnred rates and modeled rates must be the result of modeling error. The report compares modeled emission rates with emissions estimates derived from measurements of tank vent gas flow rates and chemical compositions. Importantly, the report assumes that if there is a difference between the modeled emissions and the measured emissions estimates, then there must be an error in the modeled emissions figure. Hy-Bon ignores the possibility that the measured figure might be wrong. Hy-Bon's apparent belief in the infallibility of measured data is a fundamental flaw that undermines all aspects of this report. The report ignores the fact that other factors, such as measurement uncertainty and non-representative sampling, could just as easily be the cause of a given disagreement; measurement errors can and do occur. What is more, Hy-Bon does not appear to have attempted to justify its position that measured data are infallible (or at least are superior to modeled data). The report provides no assessment of measurement data quality, no quality assurance data, and no quality control data. In addition, the report fails to provide any 604 West 14th Street. Austin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: lexaspipelineassociation@yahoo.com 3 records of calibrations, description of calibration procedures, or traceability of standards for field or laboratory equipment; it fails to describe how the field or laboratory measurements were validated; it fails to provide any assessment of field or laboratory measurement data quality; and it makes no reference to any quality assurance project plan that might have been employed during the survey.! Errors in vent gas flow rate measurement, errors in laboratory analysis, and errors in the daily oil production estimate can all contribute to uncertainty in the measured emission rate. Yet Hy-Bon appears to have done nothing to ensure the accuracy of its measurements. In short, the report fails to provide any justification for its continuously applied assumption that any disagreement is due to model error, as opposed to measurement error, or for its conclusion that the "direct measurement method is more representative of the actual emissions" than modeling. Draft Report at 68. 2. Crucial input data were not tested or verified, but simply "assumed" to be correct. A basic and overarching flaw in the report is that crucial input data were neither tested nor verified. Rather, such data were provided to Hy-Bon by participating companies and Hy- Bon accepted the data without question: "[O]ne major variable that was not directly measured on site was the volume of oil produced through the tanks .... This data was provided by the participating company .... It should be noted that these volumes are critically important variables in the modeling software outputs, and were provided by the participating companies and will be used as factual production data." Draft Report at 10. "One of the key factors for each of the estimating methods is the production rates of the battery. These rates are a critical part of the process and could not be directly measured from most of the sites. This information was provided to the survey team by the operating company, and must be assumed as correct and valid for the study." Draft Report at 11. The report makes clear that crucial input data were neither gathered nor verified by Hy-Bon. Instead, Hy-Bon simply took production numbers supplied by the producers and "assumed" that those numbers were "correct and valid" for purposes of the study. This is a methodological error that calls the entire report into question. Production numbers are typically based on either tank gauging measurements or truck loading tickets. Therefore, very significant errors can be introduced inasmuch as the oiVcondensate rate is a critical fact when testing the results from VOC emissions models against measured vent rates. In addition, there may have been inconsistencies in the various participating companies' methods of gathering production data; yet it does not appear that Hy-Bon examined this issue. One or more of the companies may have had a turbine meter on the separator. Turbine meters cannot accurately distinguish between water, condensate, and gas; therefore, if the condensate dump valve might have allowed some water and/or gas to be mixed in, it would log as I In addition, the report does not attempt to explain the range of disagreement between measured and modeled emissions figures, even though the reported inconsistencies are so varied as to indicate that something other than the consistent modeling error assumed by Hy-Bon is occurring. 604 West 14th Street, Austin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: lexaspipelineassociation@yahoo.com 4 condensate. This would throw the numbers off. Alternatively, one or more of the companies may have measured condensate production by tank gauge or by the pump truck. If so, this is not a continuous measurement and could be very difficult to resolve into the exact 24-hr period for which the stack testing occurred. Further, the possibility of measurement inaccuracy is compounded by the fact that condensate dumping into the tanks is more of a batch than a continuous process. One day a separator could dump a large quantity of condensate into a tank and the next day very little. Therefore, if some type of average condensate production data were used, it is likely that the data were not representative of the 24-hour period when stack sampling occurred. Because oil production was not actually measured during the 24-hour flash gas measurement period, the authors of the report do not know what production flowed to the tanks during the test. Yet this is the information upon which the entire report is based. The fact that admittedly "critically important" data were not subjected to any verification, quality control, or assurance of consistency is reason enough to discount the report in its entirety. It would be inappropriate for TCEQ to base any policy decisions upon a report prepared in such a manner. TPA offers the following recommendations: . TCEQ should ask Hy-Bon to identify those sites where an actual measured oil/condensate production rate was available so that those sites can be evaluated separately. TCEQ should keep in mind that these test results are tainted by the lack of accurate data on oil/condensate production. It is wrong to assume that the "measured VOC emissions" are accurate and that results from empirical correlations, lab results, and process simulators can be compared accurately to the measured values. It might be possible to salvage some data by discarding data that are obviously in error and/or applying logical corrections to data that can be shown to be in error. 3. The report inexplicably fails to consider uncertainty and variations in daily production. Another error in the report is its failure to account for the fact that oil and gas production varies from day to day. The report assumes that production during one 24-hour period will be the same as production during any other 24-hour period. Hy-Bon assumed a constant rate of production and simply multiplied the daily emissions estimate by 365. This is an erroneous assumption. This flawed assumption undermines the entire report, because the entire report hinges upon that assumption. In practice, production rates vary substantially from day to day. If the 24-hour period used by Hy-Bon was exceptionally low or exceptionally high, then the conclusions reached for that site will be skewed accordingly and will be inaccurate. A measurement during one single 24-hour period may be far different from the average production at the site. If Hy-Bon used a daily oil production volume figure that was not representative for the site, then its annual emissions estimates derived from measurements will not be representative either. Such an error affects measurements, not models; indeed, models are more accurate than measurement in these circumstances because models take into account fluctuations in volume throughout the year. 604 West 14th Street, Austin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: texaspipelineassociation@yahoo.com 5 Hy-Bon's error in this regard undermines the measurements in the report. This is significant, because the report is based on the incorrect assumption that measurements are always correct and that all inconsistencies between modeled and measured emissions must be the result of a mode,ling error. It is disappointing to see this mistake being made, given that this same mistake was made, and pointed out by industry, in connection with the 2006 HARe report. 4. By-Bon's assnmptions regarding tank temperature and back pressure are incorrect. Hy-Bon used a tank temperature of 81 0 F for all E&P TANK models. Use of a correct tank temperature is crucial to an accurate result, yet not only does 81 0 F s'eem high, it is hard to see how it would have been appropriate to use the same temperature for all sites. As for the HYSYS models, Hy-Bon did not bring the tank temperatures up to 81 0 F to be consistent with its approach to E&P TANK modeling. To the contrary, Hy-Bon's HYSYS tank temperatures varied wildly. When all HYSYS cases are run at 81 0 F, the HYSYS and E&P TANK-RVP methods are generally consistent. In addition, Hy-Bon evidently did not account for the fact that a lot of upstream condensate tanks have a backpressure valve that maintains 1-3 psi of backpressure on the tank. In order to escape to the atmosphere, the vapor in the tank space must build up enough pressure to overcome the backpressure. For calculation programs such as E&P Tank, this backpressure should be added to the ambient pressure, yet this does not appear to have been done by Hy-Bon. 5. The report is not based on a representative sample of sites. Another problem with the report was that it is heavily skewed toward very low volume sites. The report acknowledges this; it states that producers' concern over confidentiality "resulted in a survey site pool of locations with relatively low vent volumes. Some of the models evaluated in this study may produce somewhat different aggregated results when modeling sites with larger vent volumes." Draft Report at 9. It is likely that use of a more representative selection of sites would have resulted in different and more representative results. This is another reason why the reports' conclusions are umeliable. 6. Vent gas flow rates appear to have been outside of the range of the flow meter nsed by Hy-Bon. For the direct measurement method to be reliable, there must be accurate monitoring of vent gas flow rates. It appears, however, that the flow meter used for this report was not capable of accurate measurement, because the flow rates were too low. (This, in tum, brings to mind the problem just discussed: failure to use a representative sample of sites.) It appears that a Daniel Model 3000 2" meter was used for the report. Table DM-2 (Draft Report at 21) indicates that flow rates were below the lower limit ofthe meter's range for most ofthe sites. 604 West 14th Street, Ausfin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: texaspipelineassociation@yahoo.com 6 7. The report contains numerous typographical errors and conflicting statements. For example: The report admits that the Vasquez-Beggs equation could not estimate emissions for 27 tank batteries because site operating parameters were outside the applicable range (Draft Report at 5), but in the same paragraph it is stated that Vasquez- Beggs underestimated emissions at 85 percent of those sites. The report states that a lack of substantive results precludes any analysis or assessment of the GRl HAPCalc model (Draft Report at 5), yet the report contradicts itself at Table ES-l (Draft Report at 7) by stating that GRl HAPCalc underestimated emissions 67 percent ofthe time. The report states that VB plus Tanks 4.09 overestimated emissions at 25 of 30 battery sites. Draft Report at 33. This statement conflicts with statements made elsewhere in the report. See, e.g., Draft Report 33 (underestimated 23 out of 27 times). The figures for GOR + Tanks in Table G-2 (Draft Report at 38) do not match the figures for GOR + Tanks in Results Table 1 (Draft RepOli at 64). Table G-2 shows GOR + Tanks method underestimating emissions 76.67 percent of the time. The figures in Results Table 1, on the other hand, indicate that GOR + Tanks overestimates 100 percent of the time (even though the summary box on that same table shows 76.67 underestimation). Table ES-l states that the GRl-HAPCalc method resulted in underestimation 67 percent ofthe time. Draft Report at 7. In fact, however, modeled emissions were within five percent of measured estimates in every case. Given that a five percent differential is likely to be within the range of measurement uncertainty, it is likely that the GRl-HAPCalc method actually reached results that were consistent with Hy-Bon's measurements. 2 8. The report leaves certain important questions unanswered, even though the answers could affect the reliability of the report. The text ofthe report raises various unanswered questions, such as: For the laboratory GOR method, at what temperature did Caprock hold the sample at atmospheric pressure? Was the temperature held at 81 0 F in order to be consistent with the computer models? If not, the temperature was most likely less than 81 0 F (lab temperature) which would result in GOR flash emissions being lower, which in turn could help explain why GOR was reportedly underestimating emissions 77 percent ofthe time. This would be another error in the report. 2 This brings up another problem with the report: it fails to provide an assessment of measurement uncertainty. In light of the low flow rates encountered, the range of uncertainty is probably high in many cases. 604 West 14th Street. Austin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: texaspipelineassociation@yahoo.oom 7 It appears that pressurized liquid samples had an analysis pressure of 500 psig, which suggests that the samples were put under high pressure in order to ensure condensation to liquid prior to analysis. If there was an error in sampling technique, and if some of the separator gas stream got into the liquid cylinder during sampling, this would result in a more volatile sample, because applying 500 psig would condense light components of the gas. This would help explain the high modeling emissions from HYSYS and E&P TANK. 9. Notwithstanding its pro-measurement bias, the report supports use of HYSYS and E&P TANK methodologies. If the report is to be believed, the HYSYS and E&P TANK methodologies overestimate emissions and thus are conservative models. As such, they should continue to be available for use by industry participants. 10. Analysis of field test data. Finally, to properly evaluate the field test data, the results of the data should have been presented in terms of a VOC emissions factor (VOC EF) in pounds of VOC per barrel of oil/condensate throughput (lblbbl). This would have allowed the results to be more easily compared and potential errors in the data more easily recognized. A. Calculated VOC emissions factors. Table 1 (attached) presents the field test data along with a calculated VOC emissions factor ("EF") for each data set. Two VOC EFs were calculated, one based on total measured VOC emissions and one based on total measured VOC emissions less any contribution from working and breathing losses (Tanks 4.09 correction). However, it is recommended that working and breathing losses generally not be considered since working and breathing losses are only significant when there are very few flash VOC emissions, such as for heavier oils letdown from low separator pressures (see Group A, referred to below). A general review ofthe data is presented below. B. Grouped VOC Flash Emissions Factors and Data Variability. In Table 2 (attached), the field test data were divided into logical groups based on API gravity and separator pressure. Measured variables, except for oil/condensate production, were then averaged. The variability of each measured variable is shown in Table 2 by actual plus or minus from the average and by the percentage of plus or minus from the average. This methodology can help spot erroneous data. It can also show how variable some ofthe calculated VOC EFs are for data sets with very similar measured field data. Comparison of the data sets demonstrates just how much error can occur in 24-hour measurements of VOC emissions. Even with some variation in separator liquid compositions, stock tank temperatures, and stock tank oil API gravities, one would not expect more than a 1.5 to I range ofVOC EFs within each data set group. The measured data, however, show a range of 604 West t4th Street. Ausln. Texas 7870t Phone: (512) 478-287t Fax: (512) 473-ll476 Email: 8 VOC EFs of 20 to 1 for Group B, 13 to 1 for Group C, 3 to 1 for Group D, 5 to 1 for Group E, and 2 to 1 for Group F. Only Group A provides results that seem reasonable taken as a whole. Group A Group A contains two data sets from Crane County that demonstrate very little variability of both measured data and calculated emissions factors. The data are for heavier oils (29 0 APD of similar composition and for a low separator pressure (23 psig). GroupB Group B contains ten data sets that demonstrate fairly low variability of measured data, but great variability in calculated VOC EFs (minus 90 percent from average to plus 180 percent from average). The data are for light oils (39 0 APD and low separator pressure (22 psig). In general, one would expect that the VOC EFs for these light oils flashed from a pressure of 15 to 28 psig would be slightly higher than the VOC EFs for the data sets in Group A. However, two of the data sets have a calculated VOC EF that appears to be a little too low and three have a calculated VOC EF that appears to be much too high. There will always be some variability in the VOC EF from oils that have different compositions, but there should not be such a wide spread for this group of data. This indicates problems in either the stock tank gas measurements or oil production rate assumptions. Group C Group C contains six data sets that demonstrate low variability of measured data, but higher variability in calculated emissions factors (minus 80 percent from average to plus 167 percent from average). The data are for light oils (41 0 APD and fairly low separator pressure (41 psig). Based on the separator pressures, data set WTB #22 has a calculated VOC EF that appears to be much too high (8 lblbbl). Data set NTB#2 has a calculated VOC EF that appears to be much too low (0.6 lblbbl). GroupD Group D contains only three data sets; however, the data demonstrate low variability of measured data but much more variability of calculated emissions factors. The data are for light oils (46 0 APD and medium separator pressure (68 psig). Based on the separator pressure, data set WTB#4 has a calculated VOC EF that appears to be too low. GroupE Group E contains five data sets that demonstrate low variability of measured data, but higher variability in calculated emissions factors (minus 73 percent from average to plus 45 percent from average). 604 West 14th Street, Austin, Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: texaspipelineassociation@yahoo.com 9 The data are for condensate (57 0 API) and separator pressures of approximately 140 psig. Based on the separator pressure, data set NTB#13 has a calculated VOC EF that appears to be too low. Group F Group F contains two data sets that demonstrate very low variability of measured data, but extremely high variability in calculated VOC EFs (16 Iblbbl to 34 Iblbbl). The data are for condensate (64 0 API) and high separator pressure for condensate production (228 psig). The separator liquids were of similar composition. Given the similar separator liquid compositions, separator pressures, and stock tank temperatures, one would expect to see very similar calculated VOC EFs for these two data sets. Obviously, the measured vent gas flow is in error, most likely for data set NTB#11. Group G and Group H Group G and Group H both contain only one data set such that no average emissions factor can be calculated. 11. Conclusion. In conclusion, TPA is very concerned that the report, in its current state, might serve as a basis for T C E Q ~ s rejection of commonly used and accepted modeling techniques, and as a basis for TCEQ's imposition of increased monitoring requirements. The identified errors, flawed assumptions, and inconsistencies should be resolved before any weight is given to the report. We appreciate the opportunity to provide you these comments on the draft report. Very truly yours, PaJM.CI< ~ . ~ / cA. Patrick J. Nugent Executive Director Texas Pipeline Association Attachments (Tables 1 & 2) 604 West 14th Street, Austin. Texas 78701 Phone: (512) 478-2871 Fax: (512) 473-8476 Email: texaspipelineassociation@yahoo.com Table 1. Calculated VOC Emissions Factors - With and W/O Tanks 4.09 Correction Stock Tank Sevarator Flash VOC Site Oil Gas OF Psig Gas, Total Measured Total less Tanks 4.09 Bblld API OF MW Frac. VOC Sp. Gr. Tpy Lblbbl Tpy Lblbbl WTB#I 976 44 64 42 0.77 64 70 0.81 II40 6.4 1100 6.2 WTB#2 450 39 79 40 0.67 89 25 1.1 660 8.0 640 7.8 WTB#3 16 37 76 28 0.40 86 28 0.81 8.8 3.0 7.5 1.2 WTB#4 34 50 71 43 0.77 70 75 0.94 13 2.1 5.3 0.9 WTB#5 18 48 78 42 0.90 76 90 0.95 53 16 48 15 WTB#6 176 37 72 31 0.51 68 46 0.73 87 2.7 81 2.5 WTB#8 104 39 78 36 0.57 74 17 1.0 33 1.7 31 1.6 WTB#10 350 34 86 35 0.54 96 15 1.0 690 II 680 II WTB#11 250 43 86 40 0.69 94 20 1.0 72 1.6 53 1.2 WTB#12 45 29 88 36 0.33 90 23 1.1 22 2.7 19 1.1 WTB#13 94 29 88 42 0.57 90 23 1.2 56 3.3 48 2.8 WTB#14 137 38 72 29 0.34 61 24 0.9 250 10 247 9.9 WTB#15 332 41 78 38 0.71 70 17 1.0 99 1.6 80 1.3 WTB#17 166 41 80 41 0.68 82 22 1.0 13 0.4 4 0.13 WTB#18 28 40 78 38 0.54 80 25 1.0 7.7 1.5 2.6 0.5 WTB#19 1980 43 74 39 0.73 78 60 0.81 1790 5.0 1745 4.8 WTB#20 II9 38 76 32 0.54 76 40 0.76 51 2.3 44 2.0 WTB#22 85 40 78 38 0.69 78 40 0.80 124 8.0 II3 7.3 WTB#23 327 43 80 35 0.62 88 40 0.91 94 1.6 82 1.4 NTB#l 69 45 74 38 0.64 78 48 0.91 37 2.9 34 2.7 NTB#2 74 45 68 31 0.55 80 34 0.75 8.3 0.6 7.1 0.53 NTB#3 98 42 56 33 0.55 54 26 0.79 6.9 0.4 1.2 0.07 NTB#5 50 48 66 43 0.79 68 30 0.77 155 17 149 16 NTB#6 13 56 74 36 0.60 74 144 0.76 19 8.0 15 6.5 NTB#7 14 59 80 36 0.61 77 147 0.71 38 15 33 13 NTB#8 34 59 78 40 0.71 72 125 0.73 100 16 94 15 NTB#9 16 55 82 40 0.70 80 153 0.71 39 13 37 13 NTB#II 12 64 78 35 0.60 76 231 0.73 72 33 70 32 NTB#12 5.0 64 78 38 0.68 76 225 0.73 15 16 II 12 NTB#13 62 56 76 41 0.68 60 125 0.76 39 3.4 31 2.7 Table 2. Grouped VOC Flash Comparisons and Data Variability Stock Tank SeDarator Flash VOC Site 011 Gas of Psig Gas, Total Measured Total less Tanks 4.09 API OF MW Frac. VOC Sp. Gr. Tpy Lb/bbl Tpy Lb/bbl GrouDA WTB#12 29 88 36 0.33 90 23 1.1 22 2.7 19 1.1 WTB#13 29 88 42 0.57 90 23 1.2 56 3.3 48 2.8 A v e r a ~ e 29 88 39 0.45 90 23 1.2 3.0 2.0 Variance -O/-Hl -O/-Hl -3/+3 -.12/+.12 -O/-Hl -O/-Hl -O/-Hl -.31+.3 -.91+.8 % Variance -O/-Hl -O/-Hl -8/+8 -26/+26 -O/-Hl -O/-Hl -O/-Hl -10/+10 -45/+40 GronpB WTB#2 39 79 40 0.67 89 25 1.1 660 8.0 640 7.8 WTB#3 37 76 28 0.40 86 28 0.81 8.8 3.0 7.5 1.2 WTB#8 39 78 36 0.57 74 17 1.0 33 1.7 24 1.6 WTB#10 34 86 35 0.54 96 15 1.0 690 11 680 11 WTB#11 43 86 40 0.69 94 20 1.0 72 1.6 53 1.2 WTB#14 38 72 29 0.34 61 24 0.90 250 10 247 9.9 WTB#15 41 78 38 0.71 70 17 1.0 99 1.6 80 1.3 WTB#17 41 80 41 0.68 82 22 1.0 13 0.4 4 0.13 WTB#18 40 78 38 0.54 80 25 1.0 7.7 1.5 2.6 0.5 NTB#3 42 56 33 0.55 54 26 0.79 6.9 0.4 1.2 0.07 Averal!e 39 77 36 0.57 79 22 0.96 3.9 3.5 Variance -5/+4 -211+19 -12/+5 -.23/+.14 -251+17 -71+6 -.17/+.14 -3.5/+.7.1 -341+7.5 % Variance -13/+10 -27/+12 -33/+14 -40/+25 -32/+22 -32/+27 -18/+15 -90/+180 -97/+214 GronpC WTB#6 37 72 31 0.51 68 46 0.73 87 2.7 81 2.5 WTB#20 38 76 32 0.54 76 40 0.76 51 2.3 44 2.0 WTB#22 40 78 38 0.69 78 40 0.80 124 8.0 113 7.3 WTB#23 43 80 35 0.62 88 40 0.91 94 1.6 82 1.4 NTB#1 45 74 38 0.64 78 48 0.91 37 2.9 34 2.7 NTB#2 45 68 31 0.55 80 34 0.75 8.3 0.6 7.1 0.53 Averal!e 41 75 34 0.59 78 41 0.82 3.0 2.7 Variance -4/+4 -7/+5 -3/+4 -.08/+.1 -101+10 -71+7 -.09/+.11 -2.4/+5 -2.2/+4.6 % Variance -101+10 -91+7 -10/+12 -14/+17 -13/+13 -17/+17 -111+13 -80/+167 -811+170 Stock Tank Separator Flash VOC Site on Gas of Psig Gas, Total Measured Total less Tanks 4.09 API OF MW Frac. VOC Sp.Gr. Tpv Lb/bbl Tpv Lb/bbl GroupD WTB#I 44 64 42 0.77 64 70 0.81 1140 6.4 1100 6.2 WTB#4 50 71 43 0.77 70 75 0.94 13 2.1 5.3 0.9 WTB#19 43 74 39 0.73 78 60 0.85 1790 5.0 1745 4.8 Averal!e 46 70 41 0.76 71 68 0.89 4.5 4.0 Variance -3/+4 -6/+4 -2/+2 -.03/+.01 -7/+7 -8/+7 -.04/+.09 -2.4/+1.9 -3.11+2.2 % Variance -6/+9 -91+6 -5/+5 -4/+1 -10/+10 -12/+10 -5/+11 -53/+42 -76/+55 GroupE NTB#6 56 74 36 0.60 74 143 0.76 19 8.0 15 6.5 NTB#7 59 80 36 0.61 77 147 0.71 38 15 33 13 NTB#8 59 78 40 0.71 72 125 0.73 100 16 94 15 NTB#9 55 82 40 0.70 80 153 0.71 39 13 37 13 NTB#13 56 76 41 0.68 60 125 0.76 39 3.4 31 2.7 Averal!e 57 78 39 0.66 73 139 0.73 11 10 Variance -1/+2 -4/+4 -3/+2 -.06/+.05 -13/+17 -14/+14 -.02/+.03 -8/+5 -7/+5 % Variance -2/+4 -5/+5 -8/+5 -9/+8 -18/+10 -10/+10 -3/+4 -73/+45 -70/+50 GroupF NTB#II 64 78 35 0.60 .76 231 0.73 72 34 70 32 NTB#12 64 78 38 0.68 76 225 0.73 15 16 II 12 Avera2e 64 78 37 0.64 76 228 0.73 25 22 Variance -0/+0 -0/+0 -2/+1 -.04/+.04 -0/+0 -2/+3 -0/+0 -91+9 -10/+10 % Variance -0/+0 -0/+0 -5/+3 -6/+6 -0/+0 -11+1 -0/+0 -36/+36 -45/+45 GroupG WTB#5 48 78 42 0.90 76 90 1.0 53 16 48 15 GroupH NTB#5 48 66 43 0.79 68 30 0.77 155 17 149 16