Hi all -
I'm working on project to control the viscosity of paint by controlling the amount of thinner, controlling the temperature of both the fluids in blending vessel, if there are links that have discussed this, pl suggest.
Thanks in advance
How are you measuring the viscosity? Can you do it in realtime? I could give some input on the control theory if you could elaborate a bit more on what you have / can do.
chuckyx:
How are you measuring the viscosity? Can you do it in realtime? I could give some input on the control theory if you could elaborate a bit more on what you have / can do.
I'm planning to have to 4 or 5 closed control loops - temperature, pH, Vibration, density, Angular velocity, need to calibrate the reading with standards of viscosity measurement ie., CUP ZAHN or FORD
Delta_G:
Controlling temperature has been done over and over. That one is trivial. Controlling the viscosity I haven't seen discussed in an Arduino context. But if you can describe how it is measured and controlled then I'm sure it could be done.
pl refer to my reply to chuckyx above, I've to work on 4 or 5 closed loops and need to calculate the amount of paint or lacquer to be mixed to get the required viscosity and should be measurement should co-relate to to CUP ZAHN or FORD that basically measure the time taken for a measure of the liquid to drain out
Please correct me if I'm wrong, but I assume you can't take measurements quickly enough to feedback your closed loop controllers like standard PI etc.
2 ideas that came to my mind: check model predictive control. Maybe if your control and prediction horizon are large enough on a time scale , it would suffice to quickly measure and feed it. However I'm quite unsure about the feasibility of this approach. But you wanted links to do further research so this maybe can help you.
My other idea, and I would prefer this one. Check out state observers. Basically the idea is that you can introduce states, that don't feature online measurement, like your viscosity, in your control system. However if I remember correctly you still need to develop a model. Preferably some of the variables you can measure have a relationship to viscosity that you can model.
Or maybe lastly you think about a combination of closed loop systems with an open loop system.
All in all I would think it is possible to achieve your goal. The main problem I would see lies within the theory. Developing and calibrating the model.
The actual implementation should be something that can be done with the experts here. Keep in mind that there may be, depending on your final implementation, more computational power required than an Arduino uno for example can provide. But I'd save this for later, there should be appropriate ICs out there (if needed).
Hi,
Interesting project, my work place last year helped a local paint manufacturer to install a many $1000 viscosity sensor, even had the distributors techs down to install and calibrate it.
Viscosity is a very hard quantity to measure, due to many variables, not just effecting the fluid but the materials in the sensor itself.
This sensor used acoustics and one of the measuring criteria was "NO BUBBLES".
It will be interesting where your investigations go.
Tom.... 
chuckyx:
Or maybe lastly you think about a combination of closed loop systems with an open loop system.
- can you elaborate a bit on your above idea
I could have sworn I stumbled upon a nice image for that and added it yesterday : Didn't find it now... And honestly I don't have too much time at the moment, sorry for that. I don't know the correct implementation for your problem either. So I threw together some links that can possibly guide in the correct direction.
The problem is, viscosity cannot be measured, so you cant design a closed loop system for that. Take a look at open loop and feedforward control systems. I think feedforward is what actually is closer to your problem.
But still you can design closed loop systems for everything that can generate feedback, like your temperature control. So you can combine your closed loop blocks for e.g. temperature with your control approach for your viscostiy.
I hope you find some of these helpful:
Couldn’t you measure something like the power consumption or rpm of a motor driving a paddle wheel?
I was thinking of something similar, like a spring loaded ‘stick’ dipped in the flow. Then measuring the resistance to deflection as the fluid passes by.