Level in silo measurement - S7-1200

Arpf

Member
Join Date
Jan 2021
Location
Bremen
Posts
2
Hi, i have a problem with measuring level in silo using siemens S7-1200. On top of the silo there is an 0-10V radar level sensor, it works correctly, but i need more precision on my measurment.



attachment.php


So as you can see red line is the beam of the radar sensor. Somewhere in the middle of the beam there will be around 5V so lets say 50%. But the bottom cone of the silo can store less material than the "barrel/cylinder" part. Let's say it is 70% to 30% and im trying to write a piece of code to make the right scaling.


So far i have tried to separate the cone and the cylinder part to make two diffrent scaling functions. If it's less than 5V program will scale the gain in input in range of 0-60 so the 5V will be 30%. But i have no idea how to jump from that scaling to scale the cylinder part. Is there a way to measure only above that 5V? so then i can scale it aiming for that 70% and add both parts together? Or maybe there is another solution to that problem? I will be very thankful if someone can give me some advices on that.


Cheers:beerchug:

silo3.png
 
You mention measuring "level" but I'm thinking you might mean you want to measure volume? Is this correct?

If that is the case, I would look at the signal and when the level is below the conic section then just use the function for the cone. When the level is above the conic section then take the volume of material in the cone and add it to what you calculate for the cylinder.

Using your example of the cone ending at 50% signal then when you calculate the cylinder volume, use the total height minus 50%.
 
Last edited:
Say the height of the inverted conical part is Hcone, and the cylindrical part is Hcyl.


I assume you can interpret and convert the sensor data to get the height of material, Hmatl, from the bottom of the silo to the (assumed) level surface of the material; Hmatl will range from 0 to (Hcone + Hcyl).




If Hmatl is less than or equal to Hcone, then the volume of the material as a fraction of the full silo is (Hmatl/3)/((Hcone/3) + Hcyl).


If Hmatl is greater than to Hcone, then the volume of the material as a fraction of the full silo is (Hmatl-(2*Hcone/3))/((Hcone/3) + Hcyl)




Update: of course, then there is the shaky assumption of a level surface, and to deal with that we'd need to get into angle of repose (probably moisture-dependent), and whether the silo is filling or draining, to add or subtract another cone. Good luck with that.
 
Last edited:
Assuming you only want percent then you could have an array 0-99 and use this to compensate for the cone, not done it with a cone and cylinder components but used it on a pure cone i.e. each percent was calculated as a segment i.e. volume for each segment.
So use the analogue level in an integer 0-99 then indirectly point to the location in the array to get the value, I'm sure you could also get the volume per segment V= PIr2 x Height per segment.
so in effect you could profile the shape.
 
V_cone_bottom = V_frustum + V_cylinder = (1/3) * π * cone_height * ((diameter_top/2)² + (diameter_top/2)*(diameter_bot/2) + (diameter_bot/2)²) + π *(diameter_top/2)² * cylinder_height

To calculate the partially filled tank, just add the frustum part and cylinder part, depending on the level of the filled liquid, using the equations above.
 
If you can tell us the sensor model, might help more. A lot of the sensors have linearization (calculations for different shapes) built into them, you specify the change points and the distances, it calculates the rest. Siemens lr 250 (also 500's I presume) and the vega 65's for sure do this.

If you just want volume of material, I would allow the sensor to do the calculation, then use the scaled output as volume in the controller. Also, I don't think you'll get very accurate results with such an angle of sensing on the transmitter. Most require a fairly perpendicular orientation to the sensed media. If you can get the beam angle to cover the bottom of the vessel while the transmitter is square to the vessel, your results will be more accurate.
 
Last edited:
To increase accuracy, you need to use the volume of a cone equation and the volume of a cylinder equation. If measured distance equates to 5v or less, use the cone equation. If the value of distance is greater than 5v, then total volume is the volume of the cone plus the volume of the cylinder, where cylinder height equals full value measured minus 5. This value assumes your product is deposited as a flat surface. If it isn't, you have cones inverted such that their bases touch. Don't recall the formulae - high school geometry was 45 years ago!
 
I can't remember now, but I've seen a function that would take in an input array, output array, in and out signal and would map inbetween.

This allowed you to map, say, 0-20% of the input to 0-30% of the output. It was mostly used for this type of application or at times to add some weird output response fairly quickly. perhaps in the Siemens Open Library... although I could swear it came with TIA Portal.
 
I can't remember now, but I've seen a function that would take in an input array, output array, in and out signal and would map inbetween.

This allowed you to map, say, 0-20% of the input to 0-30% of the output. It was mostly used for this type of application or at times to add some weird output response fairly quickly. perhaps in the Siemens Open Library... although I could swear it came with TIA Portal.


There is the Polyline instruction for 1200/1500 now as well (Technology -> PID Control -> Auxiliary Functions). It's a little tricky though, you need to change some tags in the instance DB, not just inputs.
 
Wow, a big thank you for all of the people writing here. I did not expect that kind of feedback, thank you very much.



Monday i will try to apply your advices to my program i will report what i've come up with.


Once again, thank you very much.
 
..... it works correctly, but i need more precision on my measurment.

No-one has asked to what degree of "precision" you need.

Everyone has assumed the silo contains a solid, granular, material.

On the assumptions that the silo is cylindrical, with a conical cylindrical bottom section, the calculations are fairly straightforward, as you have been advised.

So, to get "precision" in your measurement of volume, there are things to take into account.

Already mentioned is the "angle of repose" of the silo contents. This will give you a false calculation of contents and will depend on where the laser level is sighting. Having it angled as you have drawn really complicates the calculation, as it will be measuring the range to a varying position up the angle of repose. That variance could probably be incorporated into the math somehow, but not by me, lol.

Ideally the laser "beam" ought to be vertical to counteract that variance, but this may not be possible due to the position of the silo inlet ? So if it is offset from the center, and you elect to go for a vertical laser beam, then there will come a point at near empty, when the laser is not sighting the material level, but a point part way up the cone.

Another already mentioned problem is caused by that angle of repose, which will be positive while filling, but as soon as you start emptying, it will quickly reduce to zero, then go negative (look at the lower and upper bulbs on a sand timer to understand this). Whatever calculation you had for volume in a static state would most likely suffice while filling (the dynamics are the same), but start emptying and it all goes (inversely) "pear-shaped".

On the subject of emptying versus filling, will you be using the contents measurement in the delivery of the product to the process ? If you are, then you will have to inhibit simultaneous filling, otherwise there will be a road-tanker full of excess in your process delivery !

A large proportion of my career has been in the brewing industry, and silo contents "volume" is never, and has never been, attempted using laser, sonar, infra-red, or any other means of any sort, unless they just wanted a rough estimate. Depending on the design of the silos, some facilities were able to incorporate load cells into the build, but most could not, many silos were concrete cavities, and had odd and irregular shapes.

The vast majority of silo contents measurement was derived from delivery weight (all breweries had, or had the use of, a weighbridge for delivery vehicles in/out), and weight delivered to the process, either from integration of inline check-weighers, or load cells on the receiving vessel. In other words, contents = "measured in" - "measured out",

I am not saying that what you are attempting is impossible, just "difficult", and every step of the way introduces "factors" to be taken into account, but it all comes back to the question of how much "precision" you want to achieve ....
 

Similar Topics

Customer does not want any silo penetrations. Hates level sensors on top of silos. No capacitance sensors either. Would be great if it had...
Replies
6
Views
1,484
I am looking for a way to determine level in a silo. The material is called Perilite which has a very light density. Bulk density could be...
Replies
10
Views
2,722
Can I use a touch panel like magelis (schneider electric) or Panel View (AB) with a stand alone PLC to monitor the level of a silo in real time or...
Replies
1
Views
1,257
Hi, I am working on automating an industrial fabric shrinkage tester to replace its outdated electronics with a PLC. To get the tank's water level...
Replies
14
Views
502
Hello, Haven't been on in a while. I need to generate a bit level pdf of the I/O for RSLogix 500. I can generate a report but it just shows the...
Replies
1
Views
115
Back
Top Bottom