Despite a wealth of research and data that show the very large errors associated with calculating hydrocarbon dew points (HCDP) from GC-based field measurements, this practice continues in some parts of the natural gas industry. It is proven by researchers across the globe that using a C6+ (the most common GCs in the natural gas industry) to calculate the HCDP is no better than a guess which could be off by as much as ±50 °C (±90 °F). A C9+ GC does not really improve this guesstimation by much, eliminating only one of many sources of error.
So why does this practice continue?!! And who pays the cost associated with this error?
INTRODUCTION
The HCDP of natural gas is an important quality that needs to be measured and controlled from processing, to transport, distribution, and use. It is the temperature at which natural gas undergoes a phase transition from the vapor phase to liquid phase’ i.e. dewing out. Formation of hydrocarbon liquids in pipelines, and in compressors and gas turbines creates significant measurement uncertainties as well as operational and safety issues. Controlling the HCDPs can also significantly reduce pigging operations resulting in cost savings. A significant portion of the lost and unaccounted-for gas in pipelines is a result of the gas-to-liquid conversion (phase transition) taking place inside the pipe.
Chilled mirrors are the only way one can measure a dewpoint, whether it is an HCDP or water dew point. All other technologies are inferred calculations that provide some approximation of the true value.
WHO PAYS THE COST OF WRONG MEASUREMENTS
Pipeline Operator:
Let’s say you are a pipeline operator. You have studied your pipeline network, temperatures, and pressures it gets exposed to, and have determined that the highest HCDP you can allow into your system is -15 °C (+5 °F). Your gas vendor gives you a gas that has a “calculated” dew point of -20 °C (-4 °F). However, in reality, it has a dew point of only -5 °C ( 23 °F). That means that as soon as the pipe temperature drops below -5 °C (23 °F), liquid hydrocarbons start to drop in the pipeline. So what are the consequences?
The liquid, sloshing through the pipe, will reduce the effective diameter of the pipe, resulting in increased pressure loss. This requires increased compression to move the gas along.
The liquid formed will interfere with all measurements of gas flow and properties.
The liquid formed needs to be removed through pigging, adding additional costs for the pipeline operator.
The liquid that drops out contains BTU (heating value) that the operator paid for, but will not turn into revenue on the other end. This is a notable part of the LUAF (Lost and unaccounted for) gas.
Power Generation Companies:
You operate a gas-based power generation facility. It is well known that liquids entering the turbine will cause significant damage. Thus you have instituted a maximum dew point tariff. But do you measure it, and how? If you are relying on GC-based calculations, you are rolling the dice. Every year, scores of turbines are damaged by liquids causing millions of dollars of damage in each case.
Local Utilities:
You receive gas from a pipeline relying on their data on what the HCDP needs to be. Having liquids enter your local distribution network can cause significant operational issues. How do you control that? What is the cost?
Metal Smelters and other large users of gas
If you have liquids coming into your burners, you will not be able to control your combustion process. What is the cost?
SO, WHAT IS WRONG WITH A GC-BASED CALCULATION of HCDP
GC-based calculations of HCDP have several sources of error. Each one of these errors is significant by itself. Taken together, they render GC-based calculations, completely useless as an analytical measurement.
These errors are:
a- Distribution error of heavy components (C6+ and C9+)
b- Pressure reduction error
c- GC Uncertainty error
d- Equation-of-State error
e- Sample collection error (for laboratory GCs)
We discuss each one of these sources of error in detail.
a- Distribution of Heavy Components Error
Most GCs field-deployed in the natural gas industry cannot measure all the components of natural gas. The most widely deployed natural gas GCs only measure components to C6. In other words, they measure N2, CO2, and C1-C5. They lump all other heavier hydrocarbons as C6+. While this practice is reasonable for calculating heating values, it is completely unacceptable when calculating HC dew points. The HC dew point of any natural gas depends heavily on the concentration of the heavier components.
An early attempt was made to account for the fact that C6+ components really include C6-C12 (or higher) components. The popular “C6 splits” were employed. In some case a 60/30/10 split was used; meaning that 60% of the C6+ components were hexane, 30% heptanes, and 10% octane. In some other cases, a 47/35/17 split was assumed and used in calculations.
This practice is erroneous for two reasons; i) there are typically heavier components beyond C8 in a natural gas stream; ii) These splits may represent only one specific natural gas mix. It cannot possibly apply universally to every other natural gas mixture. Therefore calculation of HC dew points using these splits produces very large errors. This fact is even acknowledged by some GC manufacturers[i]. The referenced paper from a GC manufacturer shows that using a C6+ GC will have an 80°F (~45°C) error.
The introduction of C9+ GCs was an attempt to address the distribution error. However, they only partially address the problem, as all the components heavier than C9 are lumped into the same bucket. Although this reduces some of the error, it still represents an error as high as 35°F (~20°C)[ii].
b- Pressure Reduction Error
GCs operate at close to atmospheric pressure (~10 psig). However, most gas operations including pipeline transport, storage, and compression are performed at elevated pressures, up to 2000 psig (~ 150 barg). Therefore one needs to reduce the pressure of the gas from the operating pressure down to 10 psig. This pressure reduction can change the composition of the gas significantly unless it is performed in multi-stages with significant external heating. If not done correctly, heavier components (as well as moisture) will condense out, introducing significant error in the composition of the heavies, in turn introducing significant error in the calculation of the HC dewpoint using the GC[iii].
c- GC Uncertainty Error
Like all other instruments, GCs have some inherent uncertainty. This uncertainty is larger for the heavier components. GC manufacturers typically mention an uncertainty of 2% on concentration measurement. Even at 2% uncertainty on C6+, the uncertainty introduced in the calculation of the dewpoint is tens of degrees. However, the typical field-installed GCs actually have an uncertainty of 5% or more, depending on the condition of the GC columns, the sample composition, the frequency of calibration, etc.[iv]
The chart below illustrates this issue. The blue line is the phase diagram of a typical natural gas with components up to C8. The green line is the phase diagram of the same gas if we add the 2% uncertainty of the GC to the C8 concentration. The red line is the phase diagram if we subtract 2% GC uncertainty from the C8 concentration. It is clear that the uncertainty of the GC, at its best case 2%, will add 20°F uncertainty to the value of the dewpoint at cricondentherm.
If we use a more typical 5% uncertainty for the GC, the error in the dewpoint calculation is even higher. Furthermore, if we include the uncertainty in all the other components of natural gas, the error will be greater yet.
The following graph takes a C6+ GC analysis and plots a phase diagram. Various plots correspond to various assumptions for the distribution of the heavy components and uncertainties associated with them. As you can see, by just changing some assumptions, you can get any results you want. That may be a good thing if you are a gas vendor, but not so good if you are a pipeline operator. While the true value is about 20 °F, the range of values from the GC analysis using different assumptions spans more than 100 °F!!!! If it has not done so already, this should really scare you!
d- Equation-of-State(EOS) Error
There are quite a few different Equations-of-State (EOS) for the calculation of gas properties[v]. Some of the more popular ones are Ping-Robinson (PR), Soave-Redlich-Kwong (SRK), and the GERG model amongst others. Some of these models work better at higher pressures, and some at lower pressures. Some perform better for lean gases, and some for rich gases. It is possible to use two different EOS models and get two distinctly different results for the calculated HC dewpoint[vi].
The example below corresponds to a very simple gas mixture compromising only C1-C3 components plus N2 and CO2. The phase diagram of this simple gas mixture was calculated using two different Equations-of-State. We can observe a deviation of 10°F (6°C) even at a relatively low pressure of ~270 psia (~18 bar). Gases with components with C4 and above will have far greater variance between different equations of state.
e- Sample Collection Error (for laboratory GCs)
Given all of the above errors, which cumulatively could be well in excess of 90°F (50°F), it is tempting to sample the gas and use a higher-end C12 process GC. While the C12 GC will address some of the heavy distribution errors, it adds another set of errors, namely sample collection and transport. Sampling natural gas at a high-pressure sample point is difficult. Multiple experiments have shown that a representative sample is difficult to obtain, particularly in terms of the heavy components. Moreover, when the sample is transported and re-heated from use in a lab, additional errors can be introduced[vii].
Conclusions
Hydrocarbon dewpoints cannot be calculated from GC-based measurements and have to be measured using an automated chilled mirror device in the field.
ZEGAZ Instruments CEIRS™ is the most advanced automated chilled mirror in the World. It is the only device that uses Infra-Red spectroscopy to measure the HC dew point analytically, without any assumptions, equations, or approximations.
References [1] Shane Hale, Pipeline & Gas Journal, Feb. 2013
[1] Darin L. George, Ph.D., Andy M. Barajas and Russell C. Burkey, Pipeline & Gas Journal, Sep. 2005
[1] Sohrab Zarrabian, NGSTECH, January 2014
[1] Sohrab Zarrabian, Mahmood Moshfeghian, GPA Annual Convention, Apr. 2013
[1] Giorgio Soave Chemical Engineering Science, 1972, Vol. 27, pp. 1197-1203.
[1] K. K. Shah, G. Thodos, Ind. Eng. Chem., 1965, 57 (3), pp 30–37
[1] Eric Kelner and Darin L. George, American School of Gas Measurement Technology, Sept. 2008. For more information contact ZEGAZ Instruments: ZEGAZ Instruments 7340 Executive Way Suite-M Frederick, MD 21704 Tel : (877) 934-2910 info@zegaz.com www.ZEGAZ.com
Comentarios