C6+ characterization for natural gas 9

[gaswell]

on our lines we have gas chromatograps measuring C1-C6 , the software calculates Cricondentherm, Wobbe, Dew Points etc. splitting C6+ in two fractions (according AGA method), our client (with their simulator and compositions from laboratory analysis) calculates different results (6-10 F in dew points), a chilled mirror returns intermediate values but closer to client's results. We had a long discussion about the procedure to use and there was no agreement, which is the usual practice ?

 

[PaoloPemi]

I am afrais there is not a best practice , in the sense that there is not a unique recipe to characterize C6+ , in-line gas chromatograps typically have C6+ design, measuring individual hydrocarbons up to C6 and returning the heavier components as C6+ measurement, when no other data on composition is available GPA 2261 suggest to split the C6+ fraction into several components, typically 50% n-Hexane, 25% n-Heptane, 25% n-Octane, in some cases this recipe may give reasonably correct results in others (when heavy components as C9, C10 etc. are present) the agreement with measured dew points can be poor.
When detailed laboratory analysis is available a common practice is to input those data in software package

http://www.prode.com/docs/proind.pdf

however, as no one knows what happens between two successive laboratory measurements the calculated values may be unreliable.
Apart from this a software package offers the advantage to calculate a large number of different properties with great accuracy.
a suitable equation of state (Peng Robinson, Soave Redlich Kwong , GERG or others) can calculate very accurate VLE values, for hydrocarbon mixtures typical errors are whithin 1-2 C of measured values.
For values as density AGA or GERG again are very accurate (typical errors < 0.3%).
But of course you must find a way to define the correct composition.

 

[zdas04]

That "correct composition" statement is really important. I once saw a case that a client didn't believe an extended analysis so they took another and it was considerably different. So they took a third and it was considerably different from both of the other samples. I heard they took a fourth with the same (not matching the earlier samples) results, but I wasn't on the project anymore by that time. The problem is that the heavy ends are so close to their dew point that you will get a different sample on a cloudy day than a clear day. Different on a windy day than a clear day. Different if you park your truck in a place that casts a shadow on the upstream piping than if you don't.

The two questions that you have to ask are: (1) What does the sales contract say? and (2) is the difference material?

David

 

[gaswell]

thanks for your help,

PaoloPemi
our EOS package (provided by GC manufacturer) splits C6+ in a certain way which possibly originates some errors, that's the point of arguing, your proposal to use the full compostion from laboratory analysis is intersting but I don't know if our sofwtare can do that.


zdas04
the contract specifies a maximum dew point temperature (at pressures below our delivery point), we must compare results periodically to manual visual measurements, the point is that with our std. distribution for C6+ component we get errors in excess of 15 F while the process simulator and laboratory composition has errors of 5-10 F , from a practical point of view this should have a very limited impact (we know it's a matter of ppm) but there is room for discussions.

 

[OldDoc]

if your gas contains a significant amount of octane, nonane, or decane you must consider all these heavy components when calculating dew points, differently you get uncorrect values, the characterization works only when all these components (with a significant amount) are included...

 

[gaswell]

OldDoc could you suggest a definition for "significant amount" ? We have 10-20 ppm of heavy components, I believe this is a usual amount but I have not much information about this and I do not know of standards.

 

[OldDoc]

everything mentioned in sales contract is significant, if not perhaps GPSA manual (or equivalents) may provide some hints.
The point is that specifications parameters may include compositions, contaminants, water content etc. in addition to dew point, as example we have a maximum content for heavy components, water content and dew point.

 

[JLSeagull]

Contact the manufacturer of the analyzer and ask about the cost to increase the measurement to include some components up to C10. Most of the simple plant chromatographs for gas plants are designed for C1-C6. More complex GC's can add other components or groups. Space may be an issue for a gas plant GC. It would be easier to measure C7's and C8's without separating all of the individual compounds. The additional components could add lots of money but perhaps not another $40k.

 

[gaswell]

JLSeagull,
that could be an alternative to laboratory analysis, I agree that adding a third column train to separate C6+ fraction would cost possibly 40K, however (as far as I know) a in-line automated chilled mirror costs more and I know little of other moisture analyzers (to monitor condensate and water dew point) using a variety of technologies (laser and so on) to recommend one, in addition we would prefer to maintain a single instrument so the GC solution is preferable over others.
However the problem we face seems different, sales contract specifies heating value, maximum water content and gas dew point, our client is calculating dew point with a process simulator and composition from laboratory analysis while we use our in-line GC (C1-C6+), results are different (10-15 F) and the manually operated chilled mirror which we use as reference returns intermediate values, possibly we have 10 ppm of condensate at different conditions, my point is that the difference is acceptable and within the limits of actual GC technolocy.

 

[OldDoc]

if the dew point is far from minimum operating temperature (say 3-4 times the max error) perhaps you may persuade the client to accept a correction with results from regular laboratory analysis, the same if there are really very small amounts of heavy components, differently better to budget a different solution.

 

[dcasto]

Is this gas straight from a well?

 

[gaswell]

dcasto,
pressure reducer + separation on each line.

 

[drilling53]

if that is the case (no processing) expect variable compositions (and dew points) depending from dp across choke valve, ambient temperature (as suggested by zdas04) and possibly other factors.

 

[dcasto]

Install a dew point control plant. Chill the gas down to 5 degrees cooler than the pipeline spec and you are on spec, not analyzer needed.

 

[MortenA]

Its easy to be a process enbgineer - would work in the plant im designing right now though...

Best regards

Morten

 

[drilling53]

a chilling plant would be a good solution, but the price could be higher than the analyzer...

 

[PaoloPemi]

also in most cases you have an analyzer as part of the fiscal metering system, it's not an additional expense.
And with a little additional price (can be less than 1000 $) you get many different properties which can be very accurate (see the above post).

 

[dcasto]

But you'll make money with the chill plant!!! It is a money generator.

 

[PaoloPemi]

the utilize of a refrigeration plant to meet transporting pipeline hydrocarbon dew point specifications is a very interesting option but (of course) it has a cost, when sufficient pressure is available one can also consider Joule Thompson effect, in some cases a little plant with one exchanger and a JT valve plus separator etc. can be economic and efficient...

 

[drilling53]

the basic components of a JT skid (pressure reducer + separation) are parts of std. plants, the heat exchanger can improve process efficiency but not too much according my experience, to effectively control the dew point I think a chilling plant (mechanical compression) would be much better, but expensive, see also GPSA chapter 14