Hi everyone..

Currently I'm working on a blending automation project where there are a few fluid components going to flow simultaneously and then mixed inline before delivering to the destination tank. There will be Coriolis flowmeter running on Profibus DP to the PLC Controller (Siemens). However, the bus speed might be limited by Segment Coupler's fixed baud rate as they share the same bus. This is not confirmed yet though as alternative way might be opted to separate the flowmeter from the Segment Coupler.

Then there are Control Valve whose my supplier claimed they could achieve great flow rate accuracy if the opening is within 2-90%. Therefore, the turndown ratio, as claimed by them, is around 100:3 approximately.

The pump, which are electric pumps will be controlled by VSDs.

My questions are:
First: Do I really need to separate the Profibus DP flowmeter from the SC, which is running at around 45.45 kbit/s and change to DP's maximum speed as long as the length of the bus permits to improve the control and response time?

Second: Is control valve turndown ratio is really that good? What happens to the pressure built up inside the pipeline when the CV opening is at significantly small considering the pump speed remains the same?

Third: Will inverter contribute to a much more better control of flow rate accuracy in this case?

Final Question: What is the realistic flow rate accuracy I can expect from the system? Is the PID controller really going to perform for this system configuration?

I really appreciate and hope anyone who has experience in controlling flowrate using CV and PID controller can comment and share their invaluable knowledge on this forum.

Thanks & Best of Regards,
Marcus

 

Hi Marcus

Directly to your points:

1) A pneumatically actuated (?) control valve is a slow device. 12 Mb/s will not help that much your control loop. Fieldbus Foundation H1 uses 31kbps.

2) A control valve may have that turndown ratio (when new or properly maintained). But the control valve is just one element of your hydraulic system. The response of the whole system will determine your performance. The 'pressure X valve opening' may change accordingly the complete system characteristics. BTY, your pump is not a positive displacement one, right?

3) The inverter may contribute to a better control, but again your whole hydraulic system must be evaluated.

4) Look at #3 above.

Finally, talk to some control valve guys. They may indicate someone to visit your plant and calculate the flows and pressure drops in every condition. Then he will have the data to answer your performance questions.

These kinds of mistakes cannot be corrected later by network architectures or software parameters.

Regards

 

Marcus -

It is the process dead time that ultimately determines the quality of the control. For a flow loop, expect dead times of 1-5 seconds. Your controller should be running about 10x faster than the deadtime for good control. If your network is not heavily loaded, this should be within the specs of your system.

There are several factors affecting the valve vs. Variable-Speed Drive (VSD) decision. They are:

1. VSD's are more accurate than valves. With a valve, it is unlikely you can get within 1% of target flow.

2. VSD's are less subject to mechanical issues such as stiction and hysteresis

3. VSD's are generally linear over their range, and valves are not.

All these factors point to VSD control being a better way to control flow than the valve. (No, I do not sell VSD's!) If the VSD alone does not give you the turn-down that you need, you might consider using a mid-ranging control scheme.

When the valve is nearly closed, the biggest problem will be accuracy...the valve will stick and slip. Even a small amount of stiction, 0.5% of valve stem, would cause a large percentage change in flow if the valve is only 3% open!

Yes, PID is perfectly capable of managing this scenario. If you have multiple loops/multiple fluids for ratio control, you should coordinate their response times so the controllers do not fight each other. Check out a recorded webinar titled "Feedforward and Ratio Control" at http://www.expertune.com/r2.asp?f=AList&l=pastwebinars.html

Good Luck.
-George

 

Hi Marcos T.B.,

1) I believe higher communication network speed will decrease the dead time, and if the controller running and response times cycle are even shorter then it will directly reflect on the control loop. Having said that though I'm not sure what speed is good enough.

2) They are centrifugal pumps, which are not Positive Displacement (PD) type. As i can see you're trying to point out that PD pump like piston type will be almost impossible to control the flow rate, weren't you? With centrifugal pump, we shall get more consistent if not perfectly constant pressure on the same pump speed. Therefore, smaller CV opening will build lots more pressure inside the pipeline compared to bigger CV opening. This is the reason why I couldn't see flowrate can be controlled by CV alone even if the PI controller could overcome the inlinearity of the CV. Correct me if I'm wrong.

3) Agree.

4) I've talked to some experienced programmers with this kind of system, and he said 3~5% is a realistic accuracy figure though no guarantees. What say you?

Just found out that the CV supply is not within my team's scope anymore. Moreover, the plant is not ready nor easy to reach.

Thanks for your reply and with kind regards,
Marcus

 

Hi Marcus

1) I agree with Mr. Buckbee.

2) The pressure will not change that much, except if the flow is close to the pump specification. Pumps are oversized for this applications usually. If you have the pump model/size/manufacturer you can get the curves and estimate it.

4) You may reach better values than that for a properly sized system, but it is impossible to guess without much more information. And don't expect response times faster than some seconds.

A large turn-down seems to be a requirement. The declared range of your valve (2-90%) indicates a =% valve, so be prepared for non-linearity.

Regards