watwarrior
Chemical
I was wondering if anyone had a general equation for the viscosity of crude oil based on its density and temperature.
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations TugboatEng on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Crude Oil Viscosity 6
- Thread starter watwarrior
- Start date
- Status
- Thread starter
- #21
watwarrior
Chemical
The feed gets flashed when its about 200 C, so in all of the exchangers that im looking at, this is a completely liquid feed, all the way up to 247 C.
And the pre-flash was a later addition to the plant, and was fit into an area were there is space, that is why it was put so late in the pre-heat stage.
But the crude that im tracking will stay liquid, the temperature and the SG are the only things which will change.
And the pre-flash was a later addition to the plant, and was fit into an area were there is space, that is why it was put so late in the pre-heat stage.
But the crude that im tracking will stay liquid, the temperature and the SG are the only things which will change.
All,
For what it’s worth Ashereng is absolutely correct. Oil gravity is a method to calculate the specific gravity of the oil and nothing else. Viscosity must be tested. It needs to be tested at the very least at two points. Without this test you could very well have a non-Newtonian liquid and not know it.
As a starting point Beggs & Robinson developed a formula for calculating dead oil based on gravity, but if you use it, it will get you in trouble sooner or later.
Beggs and Robinson dead oil viscosity in cP =
(10^(Temp in degree F^(-1.163)*EXP(6.9824-0.04658*API Gravity)))-1
If you actually test the oil you can use ASTM D341 standard to calculate the corrected viscosity at any temperature.
D23
For what it’s worth Ashereng is absolutely correct. Oil gravity is a method to calculate the specific gravity of the oil and nothing else. Viscosity must be tested. It needs to be tested at the very least at two points. Without this test you could very well have a non-Newtonian liquid and not know it.
As a starting point Beggs & Robinson developed a formula for calculating dead oil based on gravity, but if you use it, it will get you in trouble sooner or later.
Beggs and Robinson dead oil viscosity in cP =
(10^(Temp in degree F^(-1.163)*EXP(6.9824-0.04658*API Gravity)))-1
If you actually test the oil you can use ASTM D341 standard to calculate the corrected viscosity at any temperature.
D23
-
1
- #23
UmeshMathur
Chemical
watwarrior:
You indicated that you intend to estimate flow using pressure drop measurements in parallel pipes (the total flow before splitting having been metered). However, you will have to be very careful about estimating the resistance of pipe fittings, valves, bends, etc. to obtain a reliable flow this way. In my experience, estimating the resistance of partially open control valves is quite error prone.
Except in unusual situations, most pipeline flows are turbulent. Also, on the Moody chart (for a given e/D ratio) the friction factor falls rapidly with Reynolds number, becoming independent of Reynolds number beyond a threshold value. Then, the pressure drop is dependent only on liquid density, other things being equal.
A full range crude oil is generally hard to keep entirely in the liquid phase above 200 deg F, unless your line pressure is very high. Hence, I would recommend a few flash calculations along the path to ensure that you are not running into 2-phase flow. If you get 2-phase flow, it would be foolhardy to try to estimate the flow based on pressure drop measurements if pipe fittings offer a significant portion of the total resistance.
Further, did I understand correctly that you are pumping crude oil from 32 F all the way up to 477 F? This seems an excessive range. From where is your crude oil entering the system?
The Walther equation below is a useful relationship for viscosity v/s temperature for hydrocarbons, provided you don't have non-Newtonian behavior:
log(log(kv + C)) = A + B*log(T)
where kv is the oil kinematic viscosity (cS), and
T is the temperature (F), and A, B, C are constants found by regression. This can easily be done using the Excel Solver, provided you have at least three measurements.
NOTE 1: All logs are to base 10.
NOTE 2: Obviously, (kv + C) cannot be less than 1, so you must ensure that the range of C is restricted by the lowest viscosity from your data. E.g., if the lowest viscosity in your data is 0.3 cS, C cannot be less than 0.7. In most cases, one may assume C is 0.8 and then use two viscosities to get A and B.
As others have noted, it is not feasible to estimate crude viscosity even crudely (pardon the pun) from density data alone. However, it would seem to me that accurate knowledge of viscosity is unlikely to be critically important when your main interest is in estimating flow in pipe segments.
You indicated that you intend to estimate flow using pressure drop measurements in parallel pipes (the total flow before splitting having been metered). However, you will have to be very careful about estimating the resistance of pipe fittings, valves, bends, etc. to obtain a reliable flow this way. In my experience, estimating the resistance of partially open control valves is quite error prone.
Except in unusual situations, most pipeline flows are turbulent. Also, on the Moody chart (for a given e/D ratio) the friction factor falls rapidly with Reynolds number, becoming independent of Reynolds number beyond a threshold value. Then, the pressure drop is dependent only on liquid density, other things being equal.
A full range crude oil is generally hard to keep entirely in the liquid phase above 200 deg F, unless your line pressure is very high. Hence, I would recommend a few flash calculations along the path to ensure that you are not running into 2-phase flow. If you get 2-phase flow, it would be foolhardy to try to estimate the flow based on pressure drop measurements if pipe fittings offer a significant portion of the total resistance.
Further, did I understand correctly that you are pumping crude oil from 32 F all the way up to 477 F? This seems an excessive range. From where is your crude oil entering the system?
The Walther equation below is a useful relationship for viscosity v/s temperature for hydrocarbons, provided you don't have non-Newtonian behavior:
log(log(kv + C)) = A + B*log(T)
where kv is the oil kinematic viscosity (cS), and
T is the temperature (F), and A, B, C are constants found by regression. This can easily be done using the Excel Solver, provided you have at least three measurements.
NOTE 1: All logs are to base 10.
NOTE 2: Obviously, (kv + C) cannot be less than 1, so you must ensure that the range of C is restricted by the lowest viscosity from your data. E.g., if the lowest viscosity in your data is 0.3 cS, C cannot be less than 0.7. In most cases, one may assume C is 0.8 and then use two viscosities to get A and B.
As others have noted, it is not feasible to estimate crude viscosity even crudely (pardon the pun) from density data alone. However, it would seem to me that accurate knowledge of viscosity is unlikely to be critically important when your main interest is in estimating flow in pipe segments.
- Thread starter
- #24
watwarrior
Chemical
We pump the oil from a tank that is at ambient conditions. which right now can be anywhere from -10 to 10 deg C. and the process is suppossed to bring the oil up to about 248 deg C, before it goes into the heaters.
This is all done using the products off of the distil. tower, so we dont have to spend as much money burning fuel to heat the crude to the point that it will give us the products we are looking for.
As for the valves, unless the exchangers are off line, the valves are wide open, and i have been provided with pressure drop curves over the valves relative to flow, density and viscosity of the fluid.
This is all done using the products off of the distil. tower, so we dont have to spend as much money burning fuel to heat the crude to the point that it will give us the products we are looking for.
As for the valves, unless the exchangers are off line, the valves are wide open, and i have been provided with pressure drop curves over the valves relative to flow, density and viscosity of the fluid.
Good point from UmeshMarthur,
the spreadsheet I gave the link to is for viscosities above 2cst:
log10.log10(v+0.7)=A-B.log10.(T+273)
where v is kinematic viscosity in CST and T is temperature in degrees C.
For lower viscosities the number of "constants" increases to 3 or even 4.
Note, The spreadsheet uses log10 though I have had a comment that it should be logn i.e. Naperian rather than Briggs. Without a copy of the ASTM D341 standard to hand I can't verify which is correct though it would appear that log10 is consistent with the Walther Equation provided by UmeshMarthur.
Anyone know for sure?
JMW
the spreadsheet I gave the link to is for viscosities above 2cst:
log10.log10(v+0.7)=A-B.log10.(T+273)
where v is kinematic viscosity in CST and T is temperature in degrees C.
For lower viscosities the number of "constants" increases to 3 or even 4.
Note, The spreadsheet uses log10 though I have had a comment that it should be logn i.e. Naperian rather than Briggs. Without a copy of the ASTM D341 standard to hand I can't verify which is correct though it would appear that log10 is consistent with the Walther Equation provided by UmeshMarthur.
Anyone know for sure?
JMW
Watwarrior,
I was interested in D23's comment on the Beggs Robinson equation and did a search.
Just on the first page came some very interesting links which you might wish to follow up:
(also see the home page for a discussion)
(this appears to be an online calculator doing what you originally asked)
The two pdf files were most interesting.
JMW
I was interested in D23's comment on the Beggs Robinson equation and did a search.
Just on the first page came some very interesting links which you might wish to follow up:
(also see the home page for a discussion)
(this appears to be an online calculator doing what you originally asked)
The two pdf files were most interesting.
JMW
jmw,
I did a quick scan of the 2 pdf files in your post.
WOW! and thanks for the links (Star for you.)
It seems that a lot of scientists went to a lot of trouble to try to develop the correlations for their fields crude oils (and within fields as the HC distribution changes).
watwarrior,
In order to develop the correlations that I think you are thinking of, this is the level of empirical data collection and development required to get a useful set of equations/charts.
Haveing said that, the authors of these studies are looking at a very large range, I am hoping that your range of interest can be localised (for example, my appliations is limit to 3 temperature points, and within a very narrow range at that), then the level of work can be simplified.
Often times, we can make simplifying assumptions on lots of things since the uncertainty in another part of the overall "problem" may overshadown, in this case, your viscosity uncertainty.
Hope this helps.
I did a quick scan of the 2 pdf files in your post.
WOW! and thanks for the links (Star for you.)
It seems that a lot of scientists went to a lot of trouble to try to develop the correlations for their fields crude oils (and within fields as the HC distribution changes).
watwarrior,
In order to develop the correlations that I think you are thinking of, this is the level of empirical data collection and development required to get a useful set of equations/charts.
Haveing said that, the authors of these studies are looking at a very large range, I am hoping that your range of interest can be localised (for example, my appliations is limit to 3 temperature points, and within a very narrow range at that), then the level of work can be simplified.
Often times, we can make simplifying assumptions on lots of things since the uncertainty in another part of the overall "problem" may overshadown, in this case, your viscosity uncertainty.
Hope this helps.
All,
Please remember that the oil API is a method for calculating the specific gravity only. There is not any relation between specific gravity and viscosity. Any correlation you use can only be a starting point. Testing the oil is the only way to know the viscosity.
As a example I looked at a well several months ago where the oil API was 12.4 and the bottom hole temperature was about 100 degree F. If I calculated using Beggs the dead oil viscosity should have been about 715 cP. The actual test at 100 degree F was over 2700 cP. Designing lift required and friction losses using calculations verses actual test can result in a catastrophic error.
The ASTM formula for calculating viscosity at different temperatures does not account for pressure.
V = [log.log.(v+07)=a-b.log(t+273)]
If the produced liquid is non-Newtonian this formula will not work.
Please remember that the oil API is a method for calculating the specific gravity only. There is not any relation between specific gravity and viscosity. Any correlation you use can only be a starting point. Testing the oil is the only way to know the viscosity.
As a example I looked at a well several months ago where the oil API was 12.4 and the bottom hole temperature was about 100 degree F. If I calculated using Beggs the dead oil viscosity should have been about 715 cP. The actual test at 100 degree F was over 2700 cP. Designing lift required and friction losses using calculations verses actual test can result in a catastrophic error.
The ASTM formula for calculating viscosity at different temperatures does not account for pressure.
V = [log.log.(v+07)=a-b.log(t+273)]
If the produced liquid is non-Newtonian this formula will not work.
UmeshMathur
Chemical
As others have already noted, the units of temperature in the Walther equation should be absolute [K] or [R], rather than [C] or [F]. Otherwise, one woukld get log errors below 0 degrees.
In response to JMW's question, I'm pretty sure (from work in grad school in the 1970s) that the original form of the Walther equation required use of logarithms to base 10. Of course, if one selected Naperian logs, one would get numerically different coefficients but the statistical fit would be equally accurate.
The reference I have is:
C. Walther, Proc.World Petrol.Congress 2, 419 – 421 (1933)
In response to JMW's question, I'm pretty sure (from work in grad school in the 1970s) that the original form of the Walther equation required use of logarithms to base 10. Of course, if one selected Naperian logs, one would get numerically different coefficients but the statistical fit would be equally accurate.
The reference I have is:
C. Walther, Proc.World Petrol.Congress 2, 419 – 421 (1933)
Thanks UmeshMathur.
Incidentally I have noted that in several learned articles the equation is quoted without showing the base for the logs and that they also define T as temperature without stating the units.
This is really poor practice and to be deplored.
I will start by putting myself on the list of guilty parties for my own omission of the base for the logs on the ASTM D341 spreadsheet I referenced above.
(I'd better correct this.)
JMW
Incidentally I have noted that in several learned articles the equation is quoted without showing the base for the logs and that they also define T as temperature without stating the units.
This is really poor practice and to be deplored.
I will start by putting myself on the list of guilty parties for my own omission of the base for the logs on the ASTM D341 spreadsheet I referenced above.
(I'd better correct this.)
JMW
UmeshMathur
Chemical
d23:
Two points regarding your post:
(1) Crude oil behavior is generally pretty close to Newtonian
(2) The effect of pressure on liquid viscosity is significant only at very high pressures (thousands of psi, typically). Correlations in the API Technical Data Book demonstrate this pretty clearly.
I would think that watwarrier's application (preheating using crude oil from tankage) is unlikely to be at high pressures on the crude oil side.
Once again, I think we are straying a bit from the issue raised by watwarrier's original problem: is the Reynolds number high enough for viscosity to be a factor in pressure drop or not? At very low temperatures (around 32 F), it is quite likely that the oil is so viscous that Reynolds number drops drastically, since viscosity increases dramatically at low temperatures (by virtue of the log-log relationship), and the friction factor needs to be evaluated below the transition to full turbulence.
Therefore, as others have mentioned, there seems to me to be no way around a couple of decent viscosity measurements in this situation.
Two points regarding your post:
(1) Crude oil behavior is generally pretty close to Newtonian
(2) The effect of pressure on liquid viscosity is significant only at very high pressures (thousands of psi, typically). Correlations in the API Technical Data Book demonstrate this pretty clearly.
I would think that watwarrier's application (preheating using crude oil from tankage) is unlikely to be at high pressures on the crude oil side.
Once again, I think we are straying a bit from the issue raised by watwarrier's original problem: is the Reynolds number high enough for viscosity to be a factor in pressure drop or not? At very low temperatures (around 32 F), it is quite likely that the oil is so viscous that Reynolds number drops drastically, since viscosity increases dramatically at low temperatures (by virtue of the log-log relationship), and the friction factor needs to be evaluated below the transition to full turbulence.
Therefore, as others have mentioned, there seems to me to be no way around a couple of decent viscosity measurements in this situation.
UmeshMathur, Your last statement got me thinking to one of my intern jobs.
Most companies have physical data of the materials they use. Actually, they have stuff no one would think of. For example, I needed to estimate heat loss in a particular tank (complete through insulation, etc.) Now I admint, I am not the best guy at heat transfer calcs, so I asked around, and sure enough, there was a set of binders with heat loss empirical data for each and every tank on site. My classmate in a different department evidently had the same task assigned as well, and did the calcs from estimates and first principles. We compared answers, and were off by about 5%-10%, depending on surface size.
Watwarrior, have you asked around to see if the viscosity data you need is already available somewhere?
Most companies have physical data of the materials they use. Actually, they have stuff no one would think of. For example, I needed to estimate heat loss in a particular tank (complete through insulation, etc.) Now I admint, I am not the best guy at heat transfer calcs, so I asked around, and sure enough, there was a set of binders with heat loss empirical data for each and every tank on site. My classmate in a different department evidently had the same task assigned as well, and did the calcs from estimates and first principles. We compared answers, and were off by about 5%-10%, depending on surface size.
Watwarrior, have you asked around to see if the viscosity data you need is already available somewhere?
- Thread starter
- #33
watwarrior
Chemical
The oil is flowing through these pipes at 900 k#/hr, which makes this flow extremely turbulent (8" - 16" pipes).
And unfortunately, the previous people who have tried this, assumed that the viscosity of crude was constant over the temperature range, and used the old values (from early 70s) that were used as rough estimates to size the first exchanger in the sysem.
And unfortunately, the previous people who have tried this, assumed that the viscosity of crude was constant over the temperature range, and used the old values (from early 70s) that were used as rough estimates to size the first exchanger in the sysem.
- Status
Similar threads
- Locked
- Question
