Know when the end is near

Online measurement offers advantages
Many manufacturers use reactor-mounted viscometers (RMVs) with probes that are inserted in the reactor to monitor the viscosity. In practice, these instruments are unreliable because of unstable conditions due to chemical reaction, mechanical agitation, presence of gases, non-miscible liquids, solids, foreign matter, etc.

Continuous, online viscosity measurement (COVM), where applicable, is a more accurate and reliable method. With this approach, a small quantity of product is continuously circulated through a viscometer that is mounted in an independently temperature-controlled enclosure located outside the reactor. The ideal location for the viscometer is a few feet from the reactor to minimize the response time and installation cost.

With COVM systems, off-spec products are eliminated, process optimization is achieved, and full process automation is possible, thereby reducing batch time and operating costs. Another advantage of this technology is that product viscosity can be measured at temperatures different from the reaction temperature. For instance, at a reaction temperature of 300°F, the viscosity might not vary much with product composition changes, whereas at 100°F it might, allowing an accurate method of monitoring the reaction.

On the downside, this technology is not suitable for water-based resins or products for which temperature and chemical composition have little effect on viscosity. Furthermore, there are physical limitations regarding the retrofit of COVM systems on existing reactors. A flanged nozzle must be available on the lower section of the reactor, well below the liquid level. If such a nozzle doesn’t exist, it might be possible to modify the discharge pipe and relocate the discharge valve accordingly.

Bernard Seguy, president
RHE America, Lisle, Ill.

Use a vibrational viscometer
Without knowing more about the synthetic-resin mixing or reaction equipment, I’d recommend using a vibrational viscometer. There are plenty of designs out there, and they are easily calibrated. The viscometer can be inserted into a mixer, reactor or circulation line, and can be used to monitor the rate of polymerization.

John A. Purvis, process engineer
ATK-Thiokol, Magna, Utah

Try a coriolis meter
If the reactor has a recirculation loop, you might consider installing a coriolis meter that can determine the viscosity using a combination of temperature, flow and differential pressure across the meter. The vendor would need to check your application, but it is possible that you could get a fairly repeatable viscosity reading that could be used to determine batch endpoint.

If the reactor has an agitator instead of a recirculation loop, you can infer the viscosity and determine the endpoint from the agitator power draw. You’ll need an accurate and appropriately scaled power meter to generate a useable signal, but it does offer a fairly repeatable signal. This method might require some tweaking over time as the equipment ages or after motor maintenance.

There are other, more complicated viscosity meters, but they are generally expensive and are prone to plugging, creating a maintenance headache.

P. Hunter Vegas, P.E., automation consultant engineer
Avid Solutions Inc., Winston Salem, N.C.

Use an inline viscosity measuring device
An inline viscosity measurement will eliminate trips to the lab and has three significant benefits:

1. Reduction in batch cycle time due to elimination of offline testing time.
2. Consistent and improved product quality since the end point of each batch will be the same.
3. Elimination of waste.

There are a few inline viscometers available commercially. Since you want to check the reaction end point, it would be advisable to have a pump-around system for the analyzer, which can be solvent washed between batches. You will also need to heat-trace and insulate the pump-around piping to prevent any solidification of the resin.

Girish Malhotra, P.E., president
EPCOT International, Pepper Pike, Ohio

Correlate with temperature readings
I would use the distributed control system (DCS) to get readings from three redundant temperature probes in the reactor. You should have a viscosity versus temperature curve/table or a formula that can be programmed into the DCS for each product. The DCS then would display real-time data calculated from a two-out-of-three voting of temperature sensors.

The hard part is selecting the correct temperature sensor for the process. You might have a thermocouple inserted into the process stream through a compression fitting, placed in a thermowell, sensed by infrared through a window into the process stream, or a sample slip stream taken from process. Another important aspect is to have proper adiabatic flow across the temperature sensor. Using the thermocouple is relatively inexpensive and should give you give you good viscosity results.

Mark Svoboda, electrical engineer
Vulcan Chemicals, Geismar, La.

Time bubble rise
With practice, you can accurately and easily check product viscosity by timing the rate of rise of a bubble through the material. This method works best for liquids with a viscosity in the range of 500 cPs to 10,000 cPs.

During production, fill a bottle nearly to the top with the material to be tested. Cap the bottle and allow the product to settle. Invert the bottle and time the rise of the air bubble. The rate of rise is directly related to the viscosity of the liquid.

Daniel B. Farnham, P.E., senior engineer
Gemark Service Corp., Exeter, Pa.

Monitor power demand
Instead of sampling manually or installing an expensive inline viscometer, try monitoring the power demand of the mixer, pump or other flow equipment (assuming such equipment exists). You’ll need to take some data to correlate power demand (amps), mixer speed and temperature to viscosity.
Don’t expect laboratory accuracy since you are collecting industrial-scale data; however, the results might be good enough for what you are doing.

Edward Giugliano, team leader
AES Warrior Run LLC, Cumberland, Md.

Use a virtual sensor
A very good prediction of product viscosity at the end of a batch reaction can be achieved by training an artificial neural network (ANN) model. The model uses measured and known recipe parameters from historical batch process data as inputs, and gives the measured viscosity values as output.

Zvi Boger, president
Optimal Neural Informatics LLC, Rockville, Md.