Oil Viscosity Measurement for Oil/Diluent Blending 2

[soma0324]

i work in heavy oil and gas industry. i am working on a project to install dual viscometer to measure the viscosity of heavy oil blend with diluent in order to calculate the oil-diluent blending rate. once we have the measured viscosity at 2 different temperautres we can extraploate the viscoisty at the reference point and hence determine the blending rate of the diluent. I am strugguling with the temperature i need to measure the visocity on. theoretically speaking it doesnt matter what temperature i use but i would like to know if there are any limitations for heavy oil visocity measurement in terms of temperature, pressure and flow rate. i will take the slip stream of the blend lets say about 5-10 USgpm and measure the visocity on temperature of 40 C and other at 10 C. the reference temperature at the dilvery point varries between 18.5 C and 7.5 C throughout the year. anyone has experience working in such a project? please comment. i need recommendations for temperature and pressure of oil viscosity measurement.
i am new to this stuff. any book or reading material anyone could suggest?
Thanks and Regards

 

[jmw]

Can I ask which viscometers you are using and if you are having the system designed for you by the manufacturer or are fabricating it yourself?

The typical dual viscometer system uses two viscometers in a sample loop in series and separated by a heat exchanger and measuring the viscosity at two different temperatures. The reference temperature viscosity is then found by using the ASTM D341 relationship or similar.

In these systems the temperatures do not need to be precisely maintained but can vary (but not rapidly or extremely) with process conditions.

However, it isn't safe to say that it doesn't matter what those temperatures are.

A dual viscometer measurement must utilise an equation such as the ASTM D341 or Walther equation to describe the temperature viscosity relationship.

These have just two constants and two unknowns and hence can be solved by using two process viscosity measurements at different temperatures.

These are very good an workable expressions of the temperature viscosity relationship but they are not exact.

Therefore it does matter what temperatures are used to measure the vsicosity and what the reference temperatures are and the higher the viscosity and the more steep the temperature viscosity curve the more important this is.

More exact expressions for the temperature viscosity relationship do exist but they require too many different process measurements to be useful.

Then too, when the reference temperature was 100degC life for refiners was comparatively easy because the process temperature range included the reference temperature and is in a relatively flat part of the curve.

With 50deg C as the new reference temperature (ISO 8217 2005E) there are more problems for refiners than before because this is a significantly lower temperature than process and it is in a much steeper part of the curve.

If you download RMI ASTM D341.xls from the downloads page of

http://www.viscoanalyser.com,

you will see several fuel oil solutions pre-loaded and on a separate sheet the curves are plotted.

This spreadsheet will let you play with different measurement (instrument) errors to see what sort of reference temperature errors you end up with - even assuming this were an exact relationship - which it isn't.

Take care with temperature... I've seen to many installations where the metalwork of a thermowell and the exposed stem acted like a heat sink to drag the measured temperature down and remember also that the viscometer will respond virtually instantly to a change in viscosity but there is a lag in the temperature sensor response.

So if there is a fluctuation in viscosity due to ratio change between cold diluent and hot resid then there will be a temperature change associated with the quality change but you might also get a temperature change for other reasons. So the viscometer will see a viscosity change but won't see the temperature change. This means that results can be biased for temperature trends and should be smoothed for transients or noise.

If you'd like to post some process details and proposed operating temperatures, I can give more precise advise.

Note, for most fuel oil blending applications a single viscometer solution usually proves sufficient.
It is only usually refiners and large terminals that have the throughput that justifies dual viscometers but given the high price of HFO today some multi-stream terminal and large barge operators are now using them.

So while it is usually OK to be sanguine about the process temperature variations and reference temperatures, with a dual viscometer system it is that better accuracy (and freedom from assumptions about reference curves etc) that is being paid for and it is as well to optimise the design.

JMW

http://www.ViscoAnalyser.com

 

[soma0324]

Thanks jmw for your value comments.
wouldnt i just need viscosities at 2 temperatures in order to come up with the viscosity at the reference temperature? also the from ASTM D341 Charts viscosity vs temperature is a linear relationship while the spreadsheet you indicated is doesnt produce the linear relationship of temperature vs. viscosity. that is why I said that theoretically just by looking at the ASTM D341 charts it doesnt matter which 2 temperatures i use. I can always extrapolate the viscoisty if i know API of my oil.
please correct me if I am wrong.
Regards,

 

[jmw]

The ASTM D341 charts are log/log charts.

On the spreadsheet the graphs are deliberately shown as linear/linear to show this relationship most visually.
In the ASTM D341 standard the charts are most convenient as straight lines for oil blending purposes.

The relationship between the temperature and viscosity is clearly not linear but a log.log relationship:

log10/log10(v+0.7)= A-Blog10.T
V= kinematic viscosity in cSt and T = temp in deg Kelvin.

This means that at low temperatures the rate of change of viscosity with temperature is very much greater than at higher temperatures.

If you want to know the viscosity at 50degC and are measuring at say 100degC and 60degC you can set up the spreadsheet with data which uses these two temperatures as reference temperatures and then make sure that one of the column headings is 50degC.

Next try changing some of the settings to represent measurement errors... 1% for viscosity and 0.5degC say for temperature.
With two measurements you can generate a range of errors and then see what this does to the viscosity at 50deg C.
Then repeat with different reference temperatures.

You should find that the calculation is most sensitive when you are calculating from higher temperatures (60 and 100degC) to lower temperatures (50degC).

The closer the two measurement temperatures the less the accuracy and the greater the difference between the measurement temperatures and the reference temperature the greater the errors.

This may seem alarmist but in practise the dual viscometer system, when properly designed and commissioned, can produce results that are very difficult to discriminate from the lab readings.

Most often, once installed, operators find they have to upgrade their lab equipment or review their sample collection and management procedures and their laboratory procedures, simply because the results are so good.

So don't be off put by my comments, I'm just fussy, but at the same time do be aware that while very good process viscosity measurement is possible and especially with the dual viscometer system, at the same time process viscosity measurement is very sensitive to poor system design, implementation and commissioning.

The temperature viscosity relationships is the main reason why process viscosity measurement has proven so difficult until now, far more so than any consideration about the fluid's rheological properties and whether it is Newtonian or Pseudoplastic or whatever.

It isn't through lack of effort that it has taken so many years to find a way to successfully replace the process capillary viscometer with all its faults, it is because viscosity is a difficult measurement to do well, but when done well, it can be very very good.

Oh yes, and fuel oils are among the most benign applications as it happens, despite being apparently nasty dirty fluids. (Ah, but do be sure you have chosen viscometers with PFA or PTFE non-stick coating and you have decent sample flow rates; when blending residual fuel with diluent (or crude oil with cutter stocks in pipelines), the diluent can dissolve the resins that normally are bonded to asphaltenes causing them to be precipitated which can contaminate the viscometers causing false high readings.

So, anyway, please use the spreadsheet to gain some appreciation for the influence of measurement errors.
You may be surprised how sensitive the calculation is to errors and especially where both measuring temperatures are above the reference temperature and the reference is down in the steep part of the curve.

But, in a good process set up the measurement temperatures will not vary very greatly under normal process conditions because there shouldn't be such significant variation in process temperatures that the heat exchangers can't manage a reasonable control of the measurement temperatures.

The errors can then be addressed as two components:
1) systematic error which is overcome by managing the actual measurement temperatures to be reasonably constant, and calibrating the sensors and system against the lab (where necessary - the errors do not always require this, it depends on what is considered significant) and
2) random errors or noise which are effectively minimised by averaging. Incidentally, most labs are used to taking a sample, managing it indifferently and then running one sample through a singe capillary... it may be an ASTM D445 capillary but one measurement an hour doesn't encourage much fussiness... but in some refineries the laboratory will have multiple capillaries in the temperature bath and will discard highest and lowest values then average the rest.

Finally, the benefits of modern inline viscometer systems are virtually 100% on stream factor, the lack of any need for re-calibration, especially with the dual viscometer system, except under exceptional circumstances and speed of response.

The lab can give very accurate measurements - mainly because the ASTM D445 set up will give very precise temperature control but if using the lab to control the process the time delay between sampling and the result allows plenty of time for product quality deviation which is why this can really only be done with relatively stable processes.
Fuel oil quality varies to much for this which is why process capillaries have been used for heavy fuel ils despite, of all hydrocarbon products this is the least suitable for this technology, and why here is so much maintenance and down time.

The capillary gives something upto 30minutes response times.

The modern viscometer is giving something like 20seconds.
So in terms of product quality control the combination of accuracy and response times gives much improved product quality.

PS I note you say "heavy oil blend with diluent" and not Heavy Fuel Oil, not that it makes any real difference except that it affects the emphasis on accuracy and the need to be fussy.
Dual viscometer systems have been used with both heavy fuel oils and pipeline blending applications where the heavy crude has to be diluted to meet pipeline density and viscosity limits (in such applications a separate fiscal density meter is often used as the primary parameter is density and once inside the density limits the viscosity can be maximised).

In crude oil blending applications fussy is worth while because these systems can pay for themselves in a matter of weeks and even a small accuracy improvement is worth pursuing.


JMW

http://www.ViscoAnalyser.com

 

[SNORGY]

Soma0324:

Further to this thread, you might find the viscosity prediction methodology outlined in ASTM D7152-05 to be of some use to you.

Regards,

SNORGY.

 

[jmw]

Sorry, I should have added that for blending, a useful program is the Shell Bunker Calc.
DNV PS also have Bunker Master which is free from them and Exxon also had one.
The Shell is easiest to find which is here:

http://www.shell.com/home/content/m...nd_services/products/fuels/bunker_calculator/

A caution, it is designed for heavy fuel oils so if you input viscosity data for a component at a different temperature to the reference temperature then it will convert assuming VI for fuels.
To be safe, enter the viscosity data at the reference temperature and determine this using viscosity temperature relationship known to work for the fluid.



JMW

http://www.ViscoAnalyser.com

 

[soma0324]

Thanks all for your help.

 

[SNORGY]

Thanks jmw.

A colleague had emailed me bcapp.zip a while back but I could never get it to work because when it unzipped the executable went missing.

Downloading directly from the link you provided fixed that problem for me.

Regards,

SNORGY.