your letters | Readers Respond

Mind Your Ps and Is

As a specialist for many years in optimization of the base control layer, I read Rod Corbett's article "Mind Your Ps" (August InTech) with great interest.

Corbett has correctly identified many of the intricacies of controlling levels very well. However, he is completely

In 98% of most cases, cycling of level loops where P+I control is used is caused either by hysteresis in the valve, or else by bad tuning. Any integrating process, like level, will always cycle if it is tuned with P+I and if there is hysteresis in the valve. This is caused by the fact that integrating processes, as opposed to the simpler selfregulating type processes, start off with the process variable lagging the controller output with a phase angle of 0-90?. The I term in the controller will never give up if there is an offset (error) and will keep on trying to get the valve to the right place. Unfortunately due to the phase lag, there is always setpoint overshoot, and this results in a continuous cycle.

The second most common reason for cycling is due to bad tuning, and this is because few people indeed really understand the characteristics of integrating processes, or the practicalities of tuning them.

It is fairly common knowledge that when tuning self-regulating processes like flow loops, one can get the integral setting correct and then make adjustments to the control response merely by adjusting the P setting. (This is provided that one is using a controller with a Pl algorithm like the ISA Ideal, where the P term multiplies through both I and D terms.) However, this doesn't work on integrating processes, where there is a fairly simple basic relationship between the P term, the I term, and the process gain. The product of all three of them must equal or exceed a certain constant. Therefore for example, if you decrease the P term, you must also increase the I term proportionally.

In general after optimizing literally many thousands of integrating processes in my career, I find most people generally set the controller gain (P term) far too low, and the I term far too fast, which also causes cycling. (Generally, the I term needs to be set very slow on integrating processes as opposed to self-regulating processes.)

Corbett is also correct in stating when it comes to levels, one has to often use a filter on the PV due to the noise in the process being multiplied by the high P gain in the controller. This can result in the valve bouncing around, thus shortening its life. He is also quite correct that one can get away with very much bigger filters on integrating processes than you can on self-regulating processes.

Michael L. Brown, South Africa

Response:

Our success has been in using very low gain (high Proportional action), and no integral action. Of course there are no absolutes. The point of my article was to summarize our experiences and to get process control specialists to think again about treating level control as the very simple process that it is. Most of the time, unnecessary complications like Integral action are detrimental to the final control performance. Simplest is always best.

Rod Corbett, P.Eng.

What is NPSH?

I was reading the questions in the CIMM certification review section of the June 2006 InTech (page 60), and I don't agree with the answer for question 2.

Here is the question: What is Net Positive Suction Head (NPSH)?

The answer InTech has is "the height of liquid available to feed the pump." This is a very incomplete answer; the definition is really in the name. Net Positive Suction Head is the difference (NET) between the head at the pump impeller (SUCTION HEAD) and the saturation pressure for the liquid being pumped. If you no longer have a POSITIVE number, cavitation will occur. The simple equation is as follows:

NPSH = P^sub stat^ + P^sub sys^ - P^sub sat^ - P^sub head loss^

Where:

P^sub stat^ is the static fluid pressure.

P^sub sys^ is the system pressure.

P^sub sat^ is the saturation pressure of the fluid.

P^sub head loss^ is the amount of pressure loss due to piping resistance, fittings, etc.

This only defines the Theoretical NPSH; pump manufactures will provide a Required NPSH, based on losses in the pump casing, etc. The pump will actually require a bit more than theoretical NPSH to prevent cavitation because the pressure at the eye of the impeller is going to be considerably less than indicated suction pressure due to the increase in velocity.

Ron Gilmer

Texas Tech University

Real People

Over the years, InTech editor Gregory Hale and I have talked about the trends and changes in the controls industry. Today, I am writing to comment on the great people in this industry.

I am involved in the Wisconsin Kidney walk this year and sent out an e-mail to many people I have known over the years in this industry, including competitors, to ask for sponsors. This is not something I have ever done and was not sure how people might receive this solicitation.

I was amazed how much monetary support I received and the warm notes of encouragement. This speaks volumes about the quality of people we associate with in this industry.

The reason I became involved is because my son-in-law's kidneys shut down in 2004 shortly after Jeff and my daughter, Sarah, came back from their honeymoon. Because of this, I became very aware of the impact of kidney problems. In June of last year, Jeff had a transplant with his brother Michael bravely donating a kidney. Jeff and Michael are doing great. He and my daughter Sarah are chairing the Wisconsin Kidney Walk this year.

I am like most people where issues like this are kind of fuzzy until they touch your family.

In the U.S., 20 million people have chronic kidney disease, and another 20 million are at risk.

Jeff was lucky; many people must be on expensive dialysis for years.

It is great to be in an industry dealing with new technology, ideas, concepts, and in addition great people.

Bill Lydon

Applied Marketing Concepts