Asphalt/bitumen 1

[jmw]

I need a pd pump for asphalt/bitumen to run at around 10-20l/min with a smooth output i.e. screw, vane or gear.
Does any one have any experience of pumps on this duty and can advise me of do's and don'ts?

 

[25362]

There are a number of websites that the internet gives on small gear pumps for asphalt transfer. One has to take into consideration that asphaltic bitumen may have very high viscosities. Thus, lines and pumps must be heated to overcome friction, and to avoid solidifying.

For pumping purposes refineries try to keep viscosities below 2000 cSt. For this purpose, temperatures could be as low as 50^oC for very soft penetration bitumens up to 220^oC for an 115/15 oxidized grade. Refiners sometimes use plunger (power) and piston pumps for the transfer of asphaltic bitumens.

Overheating can, however, cause cracking and coking in hot storage. Their low thermal conductivity (typically 0.15 W/(m*^oC) at normal working temperatures limits the (heat flux) rate at which the heating process can be carried out.

Bitumens are almost completely soluble in carbon disulfide, benzene and chlorinated hydrocarbons as carbon tetrachloride or trichloroethylene. Caution should be exercised when using these solvents due to their toxicity, flammability or environmental impact, whichever risk is applicable.

 

[jmw]

Thanks 25362,
my problem is with web searching i guess, since getting a direct hit is proving difficult.
I have to design viscosity analysers for paving asphalt at 330degF and for cutback and 150degF. So the line viscosities are well within limits for the upstream viscometer. I have to decide on a second, lower temperature to which i cool the paving asphalt for the second viscometer so that with two viscometers in series i get the viscosity at two different temperatures. This lets me solve the ASTM D341 equation and determine the viscosity at the reference temp of 210degF continuously and in real time.
But i usually find that small pumps are a problem to find and this has been the case so far. I need to meet the design criteria which calls for 550psig pressure and 500degF.
The pump will therefore be running at low viscosities and high temperatures. The second viscometer will be operating at as low a viscosity as the line will tolerate. My only other concern is that at low temperatures the viscosity increase will also change the behaviour from roughly Newtonian to pseudoplastic. It would help if i knew at what temperature this becomes significant. Help, but not critical, it just measn i have to take care with the FAT, SAT and OATs, and it may mean in situ calibration of the viscometers.
It's not a big problem and works exceedingly well (previous users say the end results are accuracies better than the lab, it's just a pain finding the right pump and the best type of heat exchanger for cooling. I want to avoid heating if i can. For the cutback i can either heat and cool or do neither and use a singleton viscometer but i prefer to heat and cool in refinery applications. If you have any other advice for me on this i would like to receive it.

 

[25362]

My experience of many years ago was mainly with refinery unblown asphaltic bitumens: vacuum bottoms and propane precipitated asphalts of various origins and their blends with extracts from furfural lube extraction units. Here are some thoughts from memory.

One of the real problems with in-line viscosity measurements would be determining the exact flowing temperature, a fact that is not so much of a problem in the lab. What kind of viscometer are you planning to use ? There are some makes offered on the web.

With cutbacks you must be sure the kero component doesn't partly evaporate at working conditions.

You are right about the viscosities of straight run bitumens at lower temperatures. Although at higher temperatures -well above their R&B softening points- all bitumens are essentially Newtonian liquids, bitumens are, as you say, visco-elastic materials, exhibiting a combination of viscous flow and elastic behaviour. This is apparently dependent on their colloidal nature wwhich is a result of their chemical composition.

For that reason three general classes of bitumens have been recognized at normal working temperature ranges:

a. SOL type. These include very soft penetration grades which always exhibit Newtonian or almost Newtonian behaviour.

b. SOL-GEL types. These are slightly elastic and non-thixotropic. If a small stress is imposed continuously their viscosity becomes almost independent of the rate of shear, however, under stresses imposed over short periods they behave as elastic materials. The majority of "penetration" bitumens belong to this category.

c. GEL type. Strongly elastic and thixotropic. They recover almost completely when the impossed stress for a slight deformation is removed. They also show a yield stress. These characteristics are typical of air-blown (oxidized grade) bitumens.

Besides, the majority of bitumens in use show flow properties which are influenced by their thermal history.

Speaking of road asphalts, one fact which is often overlooked is the hardening of bitumen during the mixing operation, and the consequent fact that its chemical and physical properties in use differ from those of the initial bitumen. That's the reason why many experts think it advisable to base bitumen specs on properties after aging in the lab, to better reproduce changes occurring in the mixing process. There are some crudes that contain high-melting microcrystalline waxes that may form exudates already at temperatures of around 60^oC.

Concerning the D-341 charts, I've found that whenever you have the viscosities at two temperatures you can extrapolate a bit; you can, of course, interpolate.
Besides, when dealing with the same refinery products and the same bitumen-producing processes, you may even rely on a previously-obtained "slope" and use just the higher temperature result for estimating viscosities at lower temperatures.

Gear and centrifugal conventional-pump manufacturers can help you in finding the right small pump for your purposes, you may find them searching the net.

Referring to the cooling of a hot asphalt stream, I'd say the subject is site-specific. Tempered soft water, or condensate, in parallel-flow cooling, or even steam generation may be used, among others, for heat exchange. Then, again, it all depends on location and availability of coolants.

I hope these thoughts may be of some help.

 

[jmw]

25362, thanks for your excellent advise and information which is extremely useful. (a star awarded!)
Temperature, as you point out, is one of the keys to online process viscosity measurement and is one reason why the process capillary has survived for over 40 years while many newer tehnologies have failed; the capillary and the temperature bath compliment each other. Attempts with other technologies to use a temperature bath have largely failed. I know Sofraser have introduced their vibrating element technology in a temperature bath, but i have no actual knowledge of how well it peorms.
There are other factors, of course. To use heat exchangers is a problem because of the problem of sustaining the temperature stable enough, though this problem was resolved using a dynamic mixing system (blending hot and cold sample streams and adjusting the ratio to maintain the temperature; a method devised by Mark Shelley of Fluor Daniel and described in a paper at the Texas A&M Instrument 2000 symposium)which could, potentially, open the door to a number of technologies using these "direct" methods.
My choice is a vibrating fork viscosity transducer which i have worked with for many years. Also, i don't look for stable temperatures, just that they don't vary to quickly. Using two viscosity transducers, the actual temperature of measurement doesn't matter, just that it is accurately known and doesn't vary too fast. The ASTM D341 equation allows "indirect" (calculation) methods to be extremely effective if used with a viscosity sensor which is very responsive and very accurate, otherwise the direct method (control the temperature to the reference temperature) is appropriate.
Successful measurement is more than the sensor, the method of determining the reference temerature viscosity, and more than a good installation. It also depends on the nature of the fluids and the best use of the laboratories. Using the fork sensor systems equal or beter in performance than the process capillary are possible, and are successful in applications where the process capillary cannot be used, such as bitumens, tars, vacuum residues, quench oils (in ethylene cracker applications), bitumens and asphalts, heavy fuel oils and so on. However, each application must be investigated with care in order to design the system properly and never to assume that one application is just like another.
For non-newtonian fluids it is essential to ensure that the shear rates are constant which includes ensuring that the flow is constant irrespective of change in viscosity due to composition or temperature. Whatever viscometer is used, the "measured" viscosity will be different for each technology so the laboratory is essential in matching the measured value with the lab value. This then allows the process measurement to be used to control the process while allowing the lab measurement to perform its true functions of overchecking quality and the process instrument and not try to use it directly as a process control measurement.
The caution of the industry in accepting new technologies must be matched by the caution, and care, that the supplier invests in offering systems. Your comments on the hardening and heat treatment effects need to be carefully considered.
Your comments on the ASTM D341 charts are, of course, apt. The key is to understand when and at what conditions you can use the viscosity at two temperatures to infer the viscosity at a third temperature. The estimation of errors is vital here. In some instances a 1degC error in temperature measurement at the process conditions can result in substantially higher than 1% errors in the inferred viscosity at the refrence temperature.
You also comment on using a known slope to infer the properties of the actual process fluid. Software solutions based on this approach are very effective in a number of applications but in refineries, the two viscometer approach is better because there are no assumptions about continued relevance of the reference curves. In multi-product processes the extreme variability of the process conditions from one product to another makes the dual viscometer approach preferable.

 

[25362]

Thanks for the undeserved star.

Europeans didn't rely as much on viscosities as in the USA and continued in using needle penetrations at two temperatures and softening points as indirect measures of viscosity at low temperatures; they instituted the Penetration Index (PI) to define the sensitivity of asphalts to temperature. Europeans preferred to apply viscosities to define low-viscosity products.

Of course you are right on temperature errors, in particular when inferring the low-temp viscosity from a high-temp measurement. However, it appears to me that in-line mixing of two viscous -colloidal- streams adds another hurdle: it needs to be thoroughly done, the mixing streams must be of the same overall asphalt sample for the sake of accuracy and reproducibility. Question: if one of the streams needs to be cooled anyway, why not do it to the right temperature level in the first place, so the mixing step is avoided ?

I also agree with you in that dual viscometer measurings are a necessary and sufficient minimum number for a reasonably good interpretation of the viscosity-temperature relationship of petroleum products, in general, and of bitumens, in particular.

Good luck in your project.

 

[jmw]

Oh, I think deserved. Your information is invaluable.
I don't yet have a set of questions answered by the vendor so i only have proces stream data. I assume that the cutback application is blending but the paving asphalt application may be for reactor temerature control. This sort of information will be vital.
I am also assuming from the product data provided that the cutback data comes from a lab capillary (showing very good corelation with ASTM D341). The paving asphalt data from a cone and plate viscometer (less good corelation). But all needs clarification. It will tell me if i can allow the measurement temperatures to "float" a bit with the process temperature variation or if i need to stabilise the measurement temperatures to within a couple of degrees irrespective of the process temperature changes.