On-line Viscosity Analyzer

[YBrooman]

Hello All,

Has anyone knowledge in the use of on-line viscosity analyzer to a non-newtonian fluid?
I try to search for suppliers in the internet but I found just to newtonian fluid.
I'm in doubt if I use my agitator's power input or to install a on-line analyzer after the reactor.

Best regards

 

[jmw]

To be honest there isn't a single simple answer because the variety of non-Newtonian behaviours is quite a challenge but let's see if we can simplify thinking.

Simply put, so long as you control as many of the process conditions as is necessary, you can derive a repeatable number and you can calibrate the online sensor against a lab sensor... in many applications..... in some applications you may need to repeat this calibration quite frequently. In other snot at all.

The mystery of process viscosity measurement is often less to do with rheology than it is to do with the temperature viscosity relationship.

Don't believe it when manufacturers of rotational viscometers tell you you can't use vibrating elements viscometers for non-newtonian fluids. True, the answer you get based on facory calibration may not mean anything but in many applications what you get is a repeatable number you can use for process control and if it is necessary to know what the viscosity really is you can relate that to a lab measurement.

In many applications you don't actually want to know the viscosity, you just want to control the process. Viscosity is one of the most sensitive indicators of change there is and it is an excellent control parameter even as a repeatable number.

I think you need to give a bit more information for a sensible answer as there is no single simple answer.

JMW

http://www.ViscoAnalyser.com

 

[YBrooman]

Thanks for the answer

Well..
I want to control the flowrate of the reactants based on the viscosity of the outlet of reactor.
But sometimes the temperature can be XºC or YºC.
I can use the agitator's power input and send a signal to a control valve, but it will take a long to study how is the behavior of the agitator's power input due to the fluid viscosity.
Or I can use a analyzer at the outlet of the reactor that send a signal to a control valve.
The analyzer should be more expensive, but will improve a better quality to my final product as it has a trustable output value.

 

[JLSeagull]

Is there a close correlation with other properties such as density?

jwm is the go-to guy for viscosity.

 

[jmw]

First off, JL Seagull makes a good point but usually, where viscosity is concerned, the answer is no.
Density can make a good substitute for refractive index for example, or Brix. But usually viscosity is so very much more sensitive to change in quality than any other property that there is no substitute.
That is why so many viscosity critical processes are controlled using online viscometers where possible and by taking samples for lab viscosity measurement where not.

The good news is that in a continuous pipeline reaction we are measuring on-target all the time.
That means that when it is under control the viscosity (at a reference temperature) will be constant.

We usually expect viscosity to increase as temperature reduces but only if quality is constant.

So in a quench oil control application where the viscometer controls the heating of the residue before it is returned to the tower, a temperature increase will result in a viscosity increase. This is because the hotter quench oil condenses fewer light molecules back into the residue stream.

In plastics reactors increasing the temperature increases the molecular weight. The excursion could take two forms:
the temperature increases but the increase in viscosity due to quality is such that the viscosity also increases or the increase in temperature is such that the viscosity (at the process temperature) actually decreases.
If there is a temperature viscosity quality relationship then it is sometimes unnecessary to make a temperature correction... the viscosity change is all you need respond to to bring it back under control. The problem comes where the viscosity at the process temperature could increase or decrease dependent on the degree of change. Then a temperature viscosity correction may be needed to sort out the change in viscosity due to temperature from the change due to a change in quality.

So we need to know about temperature viscosity and quality.
It depends on the process but once you know the process and because of the link between temperature and quality it isn't too hard to sort out and you may be able to get away without a temperature correction.
You might think you could control using temperature... but it often isn't anywhere near sensitive enough.

How much hassle temperature will cause you depends on the process but in the worst case all you need do is establish a temperature viscosity relationship (2 - 4 points) for the on target product and establish what happens to the temperature and viscosity if you have a quality excursion high and low.

Some examples:
Methyl methacrylate polymerisation in batch reactions: The monomer is heated to initiate reaction. Some where along the line, as molecular weight increases, it goes exothermic and polymerisation rates accelerate.
The object is to find the end point and quench the reaction but flooding the jacket with coolant.
The reaction could take around 2 hours and the end point could be a window of 30seconds.

Fussy lab work takes too long even when you try to lot a reaction curve and try to project the quench time. Mostly crude offline viscosity measurements would be made and around 10% of batches would be reworked, though not totally lost.

A vibrating elements sensor with a suitable temperature viscosity algorithm, even though it can only give repeatable numbers, virtually eliminates rejects.
But note: in a batch reaction the viscometer has to track viscosity change across a range of qualities and temperatures. In reality, the only viscosity measurements that are critical are those made as the reaction approaches end point and this simplifies the viscometer set up.

In a continuous pipeline reaction the viscometer is generally monitoring at a set of constant conditions and the response to any excursion is to return the process to those conditions.. again, who cares what the real viscosity is when on target and who cares what the real viscosity is when off target?

The lab is there to determine viscosity based on samples. Let it look after quality assurance and use the repeatable number machine on line to look after control.

Worst case:
you can't use the same viscometer calibration setup all the time.
A parallel is in sugar inversion.
Density can be related to sugar brix but for each reaction they need to grab a sample, analyse in the lab and then work out what the target density is that is needed for control. Then they control to that density set point. This is more often a problem with organic products but the same might be done for viscosity... grab a sample, establish the set point number... or manually trim until the lab says the quality is fine then use the viscometer to maintain that condition.... even if the number today differs from the number yesterday.








JMW

http://www.ViscoAnalyser.com

 

[jmw]

Increasing the temperature can cause some fluids to behave in a more Newtonian manner. Since higher viscosity fluids are more likely to be non-Newtonian but because of their higher viscosity are often more suited to higher temperature handling, this is not always a problem.
If the viscometer is in a sample loop, then an inline heater will raise the temp of the sample only.
This is worth investigating for your fluid.

JMW

http://www.ViscoAnalyser.com

 

[siretb]

I have somewhat investigated the issue because we have the problem.
I have not found any 100% proof solution. The most promising device I have found, but not tested yet is manufactured by Hydramotion. See their we site, the XL7 model.
The Lamb wave based sensors are promising, but not commercial.
The main issue is to keep some liquid flowing around the device, and to avoid some deposits or scaling at the surface of the sensor.

 

[jmw]

Hydramotion sensors measure dynamic viscosity, like most such viscometers. e.g. Nametre, VAF etc.
A few also measure density which is necessary to obtain the kinematic viscosity.
These include the Emerson tuning fork viscometers and the Lemis-Process vibrating cylinder sensors (i think there may be one or two more such sensors today but for a some years it was only the Emerson).
Because they displace fluid rather than merely sheering it, their resonant frequency varies with density. Hydramotion, VAF ViscoSense etc. do not displace the fluid, they only shear it through a twisting action.

ASTM D341 is based on the kinematic viscosity.

A number of other functions in hydrocarbon processing also depend on kinematic viscosity and on density (not just for kinematic viscosity determination but as part of the function) e.g. molecular weight, Ignition Index viscosity gravity constant etc..

In refineries where they use the process capillary viscometer (a capillary in a temperature bath used to determine the viscosity at the reference temperature) it is necessary to also include a densitometer to measure the density at the same temperature and convert to kinematic viscosity.

The point about contamination is a good one.
In hydrocarbons, especially where blending a high viscosity crude or residual oil with a diluent, asphaltene precipitation is a common occurrence. Any coating will cause a false high viscosity reading.

Viscometers used in such applications necessarily have some form of anti-stick coating. The VAF twin cappilary (derived from the earlier Halikenen) has the capillaries machined from a block of PTFE - but still shows some sensitivity to fouling.

Most vibrating element sensors for these applications, including fuel oil heater control, have PTFE or PFA coatings.

Viscometers are generally very much more sensitive to the viscosity of the fluid at the surface than the fluid further away from the sensor.
Surface effects are therefore very important.
There would appear to be a practical relationship between the accuracy that can be achieved and the swept surface area. This means that the smaller the sensor the less accurate we might expect it to be. An exception is the TD Collaborative sensor which uses a pinch effect to enhance its accuracy (it is available with a paralene coating).

In any attempt to determine the viscosity at a reference temperature by indirect methods (calculation) sensor accuracy (and temperature measurement accuracy and speed of response) are both critical.

Vibrating element sensors can determine the reference temperature viscosity to accuracies of 1.0% of reading (varies between manufacturers).
Many viscometers (and the applications they are used in) can only measure the viscosity at the process temperature to 1-2% FSD accuracy.

The Biode (Sengenuity) sensor is an SAW acoustic wave sensor (the closest I get to a Lamb Wave sensor.. I'd be interested to learn if a different manufacturer is out their) which declares an accuracy of 10% (

http://www.sengenuity.com/tech_ref/ViSmart_Low_Shear_Bolt_sensor_spec_sheet.pdf)

They talk about the "Acoustic viscosity" and its relationship with density.
I confess, had thought they were more sensitive than that (perhaps 2%?)and maybe 10% is wrong?

In any event, at first glance, and if correct, 10% doesn't seem too good.

But that depends on the application.

Most viscosity measurement applications are "behavioural". That is, you need the viscosity at the process temperature.

These applications include spraying, coating dipping and atomising where the viscosity affects the spray pattern, coating thickness etc and since viscosity can vary very significantly with temperature, and since they tend to operate at a fixed viscosity value (if viscosity changes the temperature is modulated to bring the viscosity back to the optimum value), even relatively coarse accuracies are usable, repeatability at the control condition is more important.

For example, a heavy fuel oil could have viscosity vary from 380cst at 50C to 35 cst at 100C and the objective, for an injector, is to optimise the viscosity at 11cst, say. In which case if the viscosity is in the range 10-12cst (+/-)10%, it can provide a reasonably good measurement - except that in this application 1-2% was routinely achieved until the vibrating element sensors took over.

BUT: such accuracies would be unacceptable in an "analytical" measurement where the viscosity required is the viscosity at a reference temperature. This is because the temperature viscosity relationship.

In such applications it is essential to be able to separate out the effect on viscosity of a quality change from the effect on viscosity of a temperature change. That requires the best accuracy of viscosity and good fast responding temperature measurement. Even with today's very accurate vibrating element sensors those that measure density also) indirect or calculation methods have restrictions or limits on the relationship between the measuring temperature and the reference temperatures.






JMW

http://www.ViscoAnalyser.com

 

[JLSeagull]

JMW,
I hadn't noticed your British accent before.

 

[jmw]

????

JMW

http://www.ViscoAnalyser.com

 

[JLSeagull]

atomise
optimise

 

[jmw]

Ahah.
I understand.
That's the UK English spell checker at work.
I hadn't noticed it was working again, the last Mozilla update but one reverted to US English and wouldn't install the UK spell checker add on.
I guess the latest Mozilla update fixed that.


JMW

http://www.ViscoAnalyser.com

 

[gustipjr]

We do have a solution to offer to measure viscosity of non-newtonian fluids, as they flow, in pipes:

http://www.enico.com/technologies