PID design and tuning simuation software 1

[CChulo]

Know about any good free PID design and tunning simulation software? I've heard there are a few running around, at least a demo version but I just haven't been able to find one. I even heard you can even see the temperature control valve acting/regulating in this demo. I'll appreciate any input on this.

 

[VisiGoth]

You can play with this in MATLAB and you can get the advantage of doing a more formal classical control loop analysis. If you are going to tune this yourself and do not want to analyze it first, then the traditional advice is to run the PI first and then sloqwly and very carefully add in a little D to speed things up only if needed.

 

[djs]

You might try Expertune. I do not know if they have a free demo.

 

[PNachtwey]

All you need is Excel. Look at the zipped excel spread sheet files in this directory.
ftp://ftp.deltacompsys.com/public/PID/
One of these may fit your application. The T0P4 is a type 0 system with 4 real poles. A temperature system typically has two but with some dead time.
If you have a SLC or micrologix then the Hotrod.zip will also be of benefit. I have simulated a temperate system in ladder. The PLC can then control the simulation using PID blocks and the results are shown in a trend.

"You can play with this in MATLAB and you can get the advantage of doing a more formal classical control loop analysis."

Matlab is good for getting answers without know how they are arrived at. Mathcad is better if you want a true understanding but you must be willing to do some homework.
Neither of these packages are free.

"then the traditional advice is to run the PI first and then sloqwly and very carefully add in a little D to speed things up only if needed."

The "traditional" advice is naive. Each system has its own formulas or techniques depending on the type and the number of poles and zeros. One must also know the desired response. I like crtically damped or over damped responses as most of my applications can't tolerate overshoot.

If you have to tune a large temperature system the trial and error method is not practical. It just takes too long to see the results.

 

[EEbody]

In reality, you will not have a formula representing the system you wish to tune. How can you simulate a system without the formula. Trial and error is the only way is it not?

I studied control at university, and I still do not understand how any of it can be used in a real situation. Perhaps sombody can shed some light on the subject.

 

[PNachtwey]

Whose reallity? I calculate system models by comparing the PV with the control signal. Trial and error is not the only way. One can put some of that control theory you learned in college to use. As a bare minimum one can use the ISE or ITAE to give one an idea of how good the tuning is. Graphs of the PV as a function of the control signal are a must. Otherwise reality .....

If you have RSLogix you can do a pretty good job with the RSTrends, an excel spreadsheet ( I prefer Mathcad ) and a little knowledge.

First and second order systems can be modeled pretty easily by just looking at some graphs of the PV as a function of the control signal. I know this is covered in many test books. Basically that is all a tranfer function is. Things get messy when the mechanical guys design non-linear systems, one with too many poles, and with dead time.

We use least squared system identification to calculate a model. Actually we try to fit the data to many models and then play them back using the recorded control signal and then we compare the estimated values with the actual values. The model that generates estimated values closest to the actual values is the model we choose.

Control class in university is just to get you exposed. You have to spend much more time than just a few hours a week to get good. I do this for a living so I have many hours invested.

The point of the different spread sheets is that it show that tuning different types of systems require different techniques. What 'winds up my integrator' are all the people on the forums that think that just because they can tune a system that all system are the same and can be tuned the same way.

If you remember there are
type 0 systems ( velocity and temperature )
type 1 systems ( position )
type 2 systems ( I can't think of any now )

First order lag ( 1 real pole )
Second order lag ( 2 read poles )
Second order underdamped ( 1 complax pair of poles )

Now start mixing and matching and there are many different combinations. Each combination has a different set of formulas for calculating the closed loop gains. Yes, you can calculate the closed loop gains. You don't have to guess if you have the model.

 

[manmeetahuja]

Hi,

There are lots of demo available on net. Pconsim is one of them . I dont know if the page is there or not.

Top class PID simulators are Intune from ControlSoft(demo may not be available on net), expertune, others are there like controlstation, tc

But PID tuning is by trial and error

you may need to calculate the PID values to your process

Manmeet