Case studies – Wax precipitation
Introduction............................................................................................................................................... 1
Defining the wax model............................................................................................................................ 1
Multisolid model....................................................................................................................... 1
Coutinho wax model................................................................................................................. 2
Calculating wax appearance temperature (WAT)................................................................................ 2
Calculating wax precipitation.................................................................................................................. 6
This section is only applicable if your copy of Multiflash includes the Wax option.
Waxes are far more difficult to understand than pure solids because they are complex mixtures of solid hydrocarbons that freeze out of crude oils if the temperature is low enough. Waxes are mainly formed from normal paraffins but isoparaffins and naphthenes are also present. As with hydrates the formation of waxes is a serious concern in oil and gas processing.
Before discussing the modelling of wax deposition it is worth referring to a paper by Erickson et al. SPE 26604, (1983). which compares the results of measuring wax appearance temperatures (WAT) using three different experimental techniques. For twelve oils, where there were measurements made by at least two different techniques, there was only one case of complete agreement between two methods. Otherwise the minimum difference between techniques was 8 ºF, the maximum difference was 55 ºF, whilst the average difference was 24 ºF. It appears that the accuracy of WAT measurements has improved in recent years, but it is still difficult to measure; it is realistic when assessing results to assume that experimental error in WAT values may amount to several degrees.
There are two wax models in Multiflash. The wax model in earlier versions is a Multisolid model, whereas the more recent Coutinho model is a solid solution model. Both are described in more detail in “Modelling wax deposition” on page 60. We recommend the use of the Coutinho model but retain the Multisolid model for backward compatibility.
Multisolid model
This model is associated with a model for the liquid phase, typically a cubic equation of state. Originally we offered two fluid models from our standard model set, RKSA and PRA but recommended the RKSA model. RKSA is now the only fluid model accessible from the standard Model set window but the PRA based version can still be used based on loading the waxpra.mfc file..
With the Multisolid model the prediction of wax appearance temperatures is particularly sensitive to the characterisation of the fluid; the carbon number distribution and the properties assigned to each pseudo-component. As the PVT Analysis facility is common to several models we have not defined the characterisation procedure as part of the wax model, but allowed users full flexibility. You should be aware though that your choice of where to start the pseudocomponents and the number of pseudocomponents your plus fraction is split into can significantly affect the WAT predictions.
Our recommended procedure for the Multisolid model is to start the pseudocomponents from C6 and to split the plus fraction into 15 pseudo-components, regardless of the experimental analysis of the distribution.
We have made every effort to include constraints and corrections that stabilise the predictions. In general it appears from our tests that you should not split the plus fraction into less than 5 or more than 15 pseudocomponents. Starting the split from >C6 will usually tend to increase the predicted WAT but this will not always be the case. Similarly, using the PRA model in place of RKSA usually increases the WAT prediction by 3-4K.
Whatever the characterisation procedure chosen, if you have an experimental value for the WAT you can use the matching procedure to amend the properties of your pseudocomponents to reproduce the measured value.
Coutinho wax model
The Coutinho wax model is a solid solution model which requires more information than the Multisolid model. Predictions from Coutinho’s model are largely governed by the n-paraffin distribution. If no experimental data are available for this it can be estimated from the total wax content. If this data too is lacking then Multiflash has procedures to estimate this.
The n-paraffin distribution may be defined differently from that for the remaining liquid phase but we have found that starting the n-paraffin pseudocomponents from C6 and splitting the plus fraction into 15 pseudo-components is again a useful default.
With a limited data set it is not possible to make any definitive statements concerning the accuracy of the two models in predicting WAT, although it would appear that Coutinho's model will predict a WAT slightly higher than the Multisolid model. What is clear is that Coutinho's model provides a much improved prediction of amount of wax precipitated as a function of temperature.
The calculation of the wax appearance temperature (WAT), formerly known as the cloud point, is an example of a fixed phase fraction flash. To define the wax model, select the Waxes tab in the Select/Model set option. Two wax models are available in Multiflash..
Select one of the models and click Define Model button to define the wax model in the problem definition. Then Click on Close.
Go to the PVT Analysis form to characterise your fluid. The use of this is described in detail in “PVT Analysis” on page 190. For the Multisolid model, choose the Original Infonal1 method and complete the liquid composition table. Set the Start pseudocomponents to C6 and number of pseudocomponents to 15. Then Click on Do characterisation to characterise your fluid.
If you have selected the Coutinho model then you must choose the Revised method (Infoanal2) for the distribution. If data are available you should provide both the n-paraffin and other liquid hydrocarbon compositions and list these on the "Single fluid with n-paraffins tab". You must then choose the split for both liquid and n-paraffin distributions before completing your characterisation. If no n-paraffin distribution is available then the liquid composition should be entered on the Single Fluid tab and either the Total wax content entered in the appropriate text box or the box to estimate the wax content should be ticked. This will activate the n-paraffin distribution and both this and the liquid distribution should be defined before characterisation.
If you try to use a n-paraffin distribution with the Multisolid model then an error message will be generated.
Fixed Phase Fraction Flash - at specified P (Mole Fraction):
*** ERROR 15406 ***
The petroleum fractions are not suitable for the Multisolid wax model.
*** ERROR 10022 ***
Model cannot be initialised
The WAT at any pressure can then be
calculated by setting the pressure and using the WAT button, 
button, the fixed phase fraction
flash (at specified P) with Wax as the selected phase with zero mole fraction.
or the phase envelope can be used to plot the wax boundary.
Our particular example is based on a supplied problem set up file called wax.mfl. This particular fluid has a reported WAT of 45 DegC at 1 bar. The initial problem is set up to use the Multisolid model.
The predicted WAT at one bar is 45.5 DegC.
Now return to the model selection and this time specify the Coutinho model. In the PVT Analysis form select the Revised analysis method (nfoanal2). Tick the estimate wax content box and set the n-paraffin distribution to start from C6 with 15 pseudocomponents and complete the characterisation. As you are using Infoanal2 you will also see a plot comparing the experimental and fitted SCN distribution.
The WAT at 1 bar is now predicted to be 47.3 DegC. Return to the PVT analysis and, instead of estimating the wax content, enter 8.1% as the Total wax content and complete the characterisation. This reduces the WAT to 45.9 DegC.
You may wish to vary the pseudocomponent split for the Multisolid model and the n-paraffin distribution for the Coutinho model and see the effect on the predicted WAT.
The wax phase boundary for both models can be generated and compared using the phase envelope plotter.
If you have measured values for the WAT then you can tune the pseudocomponent properties for either model to match these values. This is done using the Tools/Matching/Other phases option. The Matching option is described in “Matching using petroleum fraction properties” on page 121.
For this particular example we know the WAT at 1 bar is 45 DegC.
The matching facility will amend the values for the melting temperature and the change of enthalpy on melting for all the pseudocomponents until the experimental WAT is reproduced. The amended fraction properties are echoed in the main window. For the Coutinho model it is the properties of the n-paraffin fractions that are modified.
In this case we only have a single experimental WAT. The resultant wax phase boundaries now show agreement at the matched WAT at 1 bar, but still exhibit entirely different behaviour with pressure.
The wax boundary for a live oil has a different shape.
The full range of flashes is available for the Wax models.
As with any other phase the amount and composition of the wax phase is determined as part of any flash calculation. Given the uncertainty of the WAT from some experimental techniques and the sensitivity of WAT calculations to the characterisation of the heaviest fractions a better picture of wax precipitation can be derived from calculation of the wax precipitated as a function of temperature.
Using the Windows version of Multiflash
you can carry out a series of PT flashes to see how the wax builds up as the
heavier components solidify with decreasing temperature. This has been
simplified in MF3.5 with the introduction of the Wax precipitation curve
button, 
. Clicking on
this produces a table of the wax weight fraction precipitated as a function of
reducing temperature. The starting temperature is 0°C, or the equivalent in
other units, and the finishing temperature is the calculated WAT. The maximum
number of points is twenty but the actual number of points will depend on the
WAT, the units used and a sensible step. The pressure will be taken as that
specified in the pressure text box or 1 bar if no pressure is specified.
|
Wax Precipitation Curve |
|||
|
Pressure: |
1. bar |
||
|
|
|
|
|
|
T (degC) |
Wax Weight fraction |
||
|
0 |
4.62E-02 |
|
|
|
5 |
3.63E-02 |
|
|
|
10 |
2.81E-02 |
|
|
|
15 |
2.10E-02 |
|
|
|
20 |
1.50E-02 |
|
|
|
25 |
1.01E-02 |
|
|
|
30 |
6.35E-03 |
|
|
|
35 |
3.51E-03 |
|
|
|
40 |
1.46E-03 |
|
|
|
45 |
0 |
|
|
.If you require additional control of the table configuration then you can use a command entered in the Tools/Command box.
The format of the command is
WAXPC pressure tstart tincrement
For example the following table has been set to calculate at 10 bar, starting from -10°C in steps of 4°C, finishing at the WAT
|
Command line: WAXPC 10 -10 4; |
||
|
|
|
|
|
Wax Precipitation Curve |
||
|
Pressure: |
10. bar |
|
|
|
|
|
|
T (degC) |
Wax Weight fraction |
|
|
-10 |
6.64E-02 |
|
|
-6 |
6.01E-02 |
|
|
-2 |
5.13E-02 |
|
|
2 |
4.22E-02 |
|
|
6 |
3.49E-02 |
|
|
10 |
2.84E-02 |
|
|
14 |
2.26E-02 |
|
|
18 |
1.74E-02 |
|
|
22 |
1.31E-02 |
|
|
26 |
9.42E-03 |
|
|
30 |
6.45E-03 |
|
|
34 |
4.10E-03 |
|
|
38 |
2.26E-03 |
|
|
42 |
8.55E-04 |
|
|
45.1649 |
0 |
|
However, if you have the Excel add-in, you van also do the calculations in a spreadsheet to see what is happening.
We have supplied two spreadsheets, wax.xls and wax_new.xls, showing the calculations in detail for the Multisolid and Coutinho models.