Weight Rate of Change Calcuations

[Michael Pemulis]

Hello,


I'm trying to take input from a group of load cells and compute throughput (i.e. tons/hour, kg/hour, etc.). Currently the load cells are wired in to a common load cell transmitter that then outputs the current weight on a 4-20mA scale to the PLC. The transmitter itself is pretty bare-bones and is not capable of rate of change calculation.


I'd like to manipulate this signal to approximate throughput. Before I start trying to reinvent the wheel, I wanted to see if there was an existing solution to this issue. I'm using Studio5000 and I see that there's a built in TOT block, however this seems to do the opposite of what I'm trying to accomplish. If I understand correctly, TOT integrates a rate to come up with a total for a given period of interest. I believe I need to differentiate in my case.



Major beginner here so any help is appreciated. Thanks for your time!

 

[danw]

Subtract one reading from the next and you got your dy.

 

[Ken Moore]

DanW is correct. Just take a sample now, and another in x time. Then extrapolate a rate per, hour, etc. You will have to base your time between samples to match your process.

 

[Michael Pemulis]

Thank you both for your input. That seems simple enough to implement. I appreciate the responses.

 

[Taylor Turner]

I have done this for velocity and temperature by calculating instantaneous rates of change. Using a timer preset to determine the measurement derivative and writing a sample position at time = 0 and when the time is finished calculate the following.
rError := ABS(rPositionSample - rPosition);
IPM := rError / ((Timer.PRE/1000)/60);

With a slow moving feedback the time derivative needs to be longer. Otherwise you'll be jumping all over the place. That is when you use an exponential moving average. I have not implemented this and is on my to do list. But my reference is as follows.
EMA := alpha * Yt + (1.0-alpha) * St1;
If you are able to implement this I would like to know what you think of the results.

 

[Michael Pemulis]

Thanks a lot Taylor, that's incredibly helpful. I'll see if I can incorporate that into my project.

 

[danw]

You didn't mention why you want a rate-of-level change, so I"ll pontificate on why it doesn't work for tiny leak detection.


I was once asked to use rate-of-change to detect and alarm on a minuscule leakage rate on several stock tanks. The tanks would sit a constant level for days on end and the idea was to turn off the rate-of-change leak detection alarm during discharge or filling when there would be a detectable and definite rate-of-change but let leak detection operate during the idle periods. Silly, silly, silly, it could never work.



There was no way to distinguish between the level transmitter noise and the level dy/dt for leaks of several cc's over days in a 10,000 liter tank. Not that I attempted, I did a quick calc that showed there was no way to distinguish noise from change.



In fact, a check of their level recordings showed that the temperature changes over a couple days would alter the level readings several orders of magnitude greater than an equivalent miniscule level change due to density changes in the liquid from the temperature change.



Dy/dt might work for catastrophic leaks, but it doesn't work for minuscule leaks.