Outstanding presentation, heads and shoulder above everything else in 'KZfaq". Please keep them coming.

The basic problem with tank level controls is that the process (the tank) has already an integral function which makes a 90 degree phase shift. If you add a controller with an integral part you create a potentially instable control loop. In my opinion the best (most stable and secure against violation against hi and lo limits) control for a tank is a purely proportional control if you are allowed to a small deviation from the optimal setpoint. So for example: Outlet valve position= 33.3*(Level[%]-Level Setpoint[%]). The effect is that the outlet valve is fully open when level is 3% above the setpoint and fully closed when level is at or below the setpoint. As the controller has no time function it is completely stable and it is always assured that the tank does not overflow or empty. An additional subject which should be mentioned in this context is the shrink and swell of a tank at boiling temperature. Some tank levels need a more sophisticated control structure than just the level. In those cases the balance of outlet versus inlet flow should be considered.

Kevin, Thank you for posting these videos. It was a great refresher course!!!

Great review of tank level tuning. Be careful on process gain calculation (kp) at 9:23 in video. It should be kp = (delta Y1 / delta t1) / delta U1. It looks like the PPT formatting messed up the placement of the "t1" in the displayed equation. Knowing the tank volume and fill time is great way to simplify this.

with kevin and the katkimshow, you are ready for python AI-classes...wow good shows...:)

Sir really good informatiin thanks

Anyone know, how to include dead time in this method?

Hi sir... Can we use this tank level control method for close tank?? In your example outlet flow is the same, water. But, is this method will work if outlet is steam??

It appears as though using the M variable and the arrest rate and process gain simply mathematically cancel. According to the math here, you set your desired system response speed (M) and the controller gain is always the M value times 2? Shouldn't the process gain factor into the gain setting of the controller some how? If this process is used, your gain will always be 10,8,6 or 4. Or am I missing something?

Arrest Rate and Time is confused. I think you always mean Time when you say rate. You wouldn't represent a rate variable with the letter T.