Modeling and Control

Well it is half time, and I have about 20 minutes to write something sensible about sensors for my blog while on vacation so off the top of my head I offer the following rules of thumb.

The best sensitivity from a temperature or pH sensor can generally be achieved by an installation where the tip of the thermowell or electrode is in the center of the pipeline. This is particularly important when there is a high viscosity fluid such as a polymer for temperature control or concentrated sulfuric acid reagent for pH control. For temperature, it is also desirable to maximize the insertion length in the center line to reduce the thermal conduction error from the tip to the flange. The insertion of the thermowell into an elbow affords this opportunity.

The location of thermowell or electrode should be at least 20 pipe diameters downstream of a heat exchanger or static mixer to provide blending of the exiting stream. When there are bubbles, the distance should be increased from 30 to 50 pipe diameters so that the bubbles have time to dissolve, which is important to insure a representative measurement and to avoid damage to the thermowell or probe from bubble impingement and collapse.

To make the response of the temperature measurement faster it is important to minimize the air gap between the sensor sheath and the inside thermowell wall. Sometimes the focus is more on the sensor than the total installation, which leads to erroneous conclusions, a popular one being an RTD is too slow. For a bare element installation the RTD might be a few seconds slower than a thermocouple but this is insignificant when other sources of lags and delays are considered (e.g. resolution limit of control valve). Furthermore for most chemical plants, a bare element raises safety concerns. An RTD with a tight fit will be faster than a thermocouple with a loose fit in a thermowell. Also, the actual response time depends upon the fluid velocity, process conditions, and the cleanliness, thermal conductivity, and size of the thermowell. The more important stuff is overlooked because it is not as easy to pigeon hole like sensor type or mass.

For both temperature and pH a pipeline location and size should provide a sufficient fluid velocity to reduce the fouling rate and increase the heat transfer coefficient for temperature and the mass transfer coefficient for pH. I go for about 5 fps but the probe installation and process conditions should be evaluated for each application because too high of a velocity can cause vibration failure and erosion from solids. For pH, the electrode signal gets noisy and the life expectancy shortened as the velocity is increased.

What impresses me the most is how the response time of a pH electrode can increase by an order of magnitude for low velocities (< 0.5 fps), low pH (< 2 pH), high temperatures (> 50 deg C), and thin films (> 10 mm). Thick or flat glass can aggravate the problem.

Why isn’t there information out there on these important installation effects? The information I have all dates back to Dupont and Monsanto technologists in the 1960s who had the luxury of picking fundamentals to investigate for long term importance. We can only hope manufacturers solve these problems so we get to know the rest of the story.

Since readers of last week’s blog may be waiting with baited breath, I offer the following mints.

Thermowells inserted into the side of a pipeline instead of an elbow may be sensing the effect of the wall or flange temperature. If the pipe wall temperature was the same as the process fluid temperature you wouldn’t need a thermowell, just clamp the temperature sensor to the pipe wall or flange. In particular, side entry thermowells for jacketed polymer pipelines may be seeing the jacket (e.g. Therminol) temperature more than the polymer temperature. Similarly, side entry thermowells for un-insulated pipelines may be affected by the ambient temperature. There are those exceptional cases where the fluid velocity is high enough (heat transfer coefficient is large enough), the cross sectional temperature profile is flat enough, and/or the pipe size is big enough where side entry is good enough. A trend chart of ambient and heat transfer fluid temperatures along with process temperatures can help detect improper installations.

Electrodes tips near the top of a pipeline or vessel tend to be noisier due to gas bubbles momentarily attaching themselves. Electrode tips near the bottom of a pipeline or vessel tend to be slower due to fouling from solids that hang near the bottom. Electrodes on the suction of a pump are more likely to see both bubbles and solids and the occasional wrench or welding rod. These considerations apply to temperature sensors although the consequences are usually less.

Since the actual pH of the process depends upon process temperature through changes in the dissociation constants of the acids and bases, improper electrode location can also lead to incorrect process temperature correction of the pH measurement ( not to be confused with the Nernst electrode temperature correction).

The side entry of sensors into vessels whose tip doesn’t extend more than a couple of inches into the fluid may be reading stagnant fluid and may be seeing the effect of jacket temperature. The sensors may be hidden from circulation by baffles and wall effects. Bulk velocities in even highly agitated vessels rarely exceed 1 fps so be prepared for more fouling problems and slower sensor response for sensors in vessels rather than in pipelines. Ideally, the sensor should be near but not too close to the tip of the impeller.

This blog is the only thing between me and dinner so here are some tidy tips on tips.

Sensor tips in a vessel located too close to a dip tube for pH or temperature and a sparger tube for dissolved oxygen result in a noisy measurement that may be responding more to the temperature or composition of the feed than the mixture of the volume. The probe should be on the opposite side of the vessel and at a different elevation than the feed entry, particularly for impellers that promote a radial circulation. The probes must always be immersed in the process fluid when the equipment is running but not in the bottom of the vessel where solids and wrenches hang out. For pH probes, it is desirable to keep them hydrated between runs and batches. When sensors are in an overflow line (not my favorite location), make sure the control loop is not trying to respond to an erroneous signal when the vessel level is not up to the overflow point. Other situations where sensors are not sensing the proper process temperature occur in condensers and kilns or dryers where the liquid and solids level, respectively, is not consistent.

When Monsanto was making the transition to a life science company, I had the opportunity to work on fermenter measurement and control for various genetically engineered products. Important opportunities identified then such as the application of mass spectrometers, dissolved carbon dioxide probes, and inferential measurements of metabolic processes have come to fruition today opening the door to more advanced process analysis and control techniques. Additionally the applications gave me a chance to apply my expertise in pH measurement and control in new ways and dig into the practical aspects of dissolved oxygen measurement and control.

These opportunities and practical considerations were documented in a book Biochemical Measurement and Control, which is now available free electronically via the link

http://www.easydeltav.com/controlinsights/biochemicalmeasurement/default.asp

This E-book also offers an introduction to importance of biotechnology and a perspective of the future from the past. My expression of gratitude in the Acknowledgement to KSHE Sweetmeats for the inspiration has initiated speculations over the years that are more interesting than the explanation that Sweetmeats is the mascot for a Saint Louis classic rock radio station that I listened to while writing the book.

My latest book Bioprocess Modeling and Control - Maximizing Process Analytical Technology Benefits published by ISA in 2006 provides an updated view and details on new tools for batch modeling, analysis, and control. This ISA book includes the development of neural network inferential measurements of dryer moisture by Washington University in Saint Louis and my first principle dynamic fermentor models for the National Corn to Ethanol Research Center. The book concludes with an excellent review of new technology for batch analytics by the University of Texas.

It is interesting how the past plays into the future. I how have the privilege to participate in a beta test of new PAT tools with Broadley-James for fermenter modeling, analysis, and control of BioNet systems. The beta tests offers a synergistic environment for combining the expertise of Dr Thomas Edgar and Yang Zhang at the University of Texas and many key people at Broadley-James and in DeltaV’s Future Architecture team including my coauthor Michael Boudreau. Another really neat thing about this beta test is that we can extensively share the details of the results.