Challenge: Control a pump to produce varying pressure profile over time

[josborne]

Hi,

I am just starting a new project and could use some tips on how to design a solution.  Here's the challenge:

We want to control a peristaltic pump that is pumping fluid through a circuit (a loop).  To do this we are measuring the fluid pressure at two points in this loop (point X, Y).  And we want to control the pressure difference (Pxy) over time.  Basically, I want to allow the user to "draw" a pressure profile (Pxy versus Time) and then have my VI control the pump to make the pressure in the circuit follow this pre-determine profile over time.

Two gotchas:  We are trying to approximately model the internal pressure in a human heart.  So the pressure profile will be very short (maybe approx 1 second) and follow a repeating pattern.  Attached is a diagram that will give you an idea of how the pressure should look over time.

Any tips on a design?  I have only done basic on/off and PID control, so this project will be a challenge for me.  And I wonder if it might demand a RT environment? 

[nitad54448]

hi

for the profile type : I have been using a waweform made by the user and defining a starting time and a step. by checking the actual time, the setpoint value is found. This was for long time processes, like 10 days of so, any arbitrary profile built by the user.

Now, for such a short time (1 sec total time or response time ?) there will be many problems : the speed of a PID required ? How many steps ? the speed of addressing and answering of the pump.. and among other things, keep in mind that most liquids are not ideal...

regards

[johnsold]

First I would question whether a peristaltic pump is the right choice.  Peristaltic pumps are essentially constant flow pumps and the pressure will rise to the level required to force the flow through the system (or until the tubing breaks).

I would probably choose a constant pressure source and use a pressure regulating valve to adjust the pressure.  Because of your timing requirements it will probably be a servo controlled valve.  That is outside my expertise, so I cannot give you any specific recommendations.

If the servo control is mostly done in hardware (with software monitoring), you could get by with a non-real time system.  Use a D/A converter (Analog Output from a DAQ device) to set a voltage proportional to the desired pressure profile and have the servo system track that.   Thus, your profile is computer generated and easily changed.

Lynn

[josborne]

Thanks!

Johnsold:  you are right.  I just did some more research and I think we will probably do exactly as you reccomend ...  to use the pump to generate a constant pressure and then use a valve to vary the pressure. 

I guess I am still worried about the timing.  The profile will be about a 1 second length.  And the pressure will vary a lot within that 1 second.  I am just not sure if it is feasible to can change the set points quickly enough and have the valve (and the pressure) respond quickly enough.

But I guess I could try it using a standard LV environment, and then migrate to real-time if needed.

[Nemesis]

Hello Jasborne,

I see somebody try to face to do something what I am doing :). I am doing the same I mean I did. I am trying to improve still my control loop but generally is working well. First I am controlling the flow waveform and after pressure waveform after that I am going to put it together (I hope it will work 🙂 ). The PID controller is working well but I have just figured out that Internal Model Control is much better. I need some help if it is possible to implemented this to Labview and how. I have just found only simulations nothing about implementation to Labview. But about the equipment as it was said before you need two pieces of equipment one to give you mean flow and second to do fluctuations around the mean flow. You can use peristaltic pump to give you mean flow and depends on the system to set the mean pressure but the problem with the peristaltic pump is the sinusoidal wave of the pumping flow. In order to have it more constant you can use it two pumps heads with phase shift or pressure dampener. To play with the pressure wave you can use solenoid valve for example a company Burkert. They have response time 20-50 ms. I hope it will help a little.

Regards,

Kamil