Seems simple but I am looking for some software to help calculate the linkage geometry needed to linearize different size butterfly valves driven by a motorized actuator. Actuator has slidewire feedback and would like to be close to linear as possible on the butterfly valve. Valve is controlling air flow to burners on a furnace. I basically only need a starting point.
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Butterfly valve linkage calculatulation 4
- Thread starter ray1761
- Start date
-
1
- #2
- Thread starter
- #3
On my furnace, a motorized actuator drives a butterfly air valve with a "connecting rod" or linkage. The air flow through the butterfly valve is not directly proportional to how far open it is. For example, when the valve is 10% open I may be getting 30% flow. when the valve is 20% open I may be getting 50% flow. By connecting the linkage in the correct location on the arms of the valve and the actuator, I should be able to get a linear flow output through the valve that corresponds with the actuator position. I was hoping to find software to assist in calculating this linkage "geometry". Searched the internet with no luck so just looking for something that could give me a starting point.
Thanks for the response....
Thanks for the response....
- Thread starter
- #4
-
2
- #5
Hello ray1761,
I think that you would be really fortunate if you could get a linear relationship between your actuator and the valve flow.
1)
A butterfly valve has an inherent quick open characteristic. This explains the aggressive flow curve you describe in your second post. You need to contact your supplier and ask for this characteristic. But it will only be valid for standardised pressures and densities.
2)
Then you should calculate the installed characteristic by calculating the real flows at the operating pressures and medium density. Note that for each pressure situation you get a different curve. Since the differences are not that big you can probably group the curves and make a 'best fit' characteristic.
3)
The linkage has its own curve which you can easily establish by measuring the angles and length of the lever. Changing the length and angle will change this curve and you are looking for the opposite of the flow curve. This will require some trial and error work.
4)
Finally you will have something that approximately works. But when the pressures or densities change you will still get an ofset.
Now my question: is this worth it?
Probably not and that is why nobody is doing this exercise. Especially not if you have >100 valves on a plant.
My advise would be to just accept that the relation is not linear. It looks like you already did a flow test and as long as you know that 10% movement changes the flow with 30% you know how to control your process.
Gr.
Terje
I think that you would be really fortunate if you could get a linear relationship between your actuator and the valve flow.
1)
A butterfly valve has an inherent quick open characteristic. This explains the aggressive flow curve you describe in your second post. You need to contact your supplier and ask for this characteristic. But it will only be valid for standardised pressures and densities.
2)
Then you should calculate the installed characteristic by calculating the real flows at the operating pressures and medium density. Note that for each pressure situation you get a different curve. Since the differences are not that big you can probably group the curves and make a 'best fit' characteristic.
3)
The linkage has its own curve which you can easily establish by measuring the angles and length of the lever. Changing the length and angle will change this curve and you are looking for the opposite of the flow curve. This will require some trial and error work.
4)
Finally you will have something that approximately works. But when the pressures or densities change you will still get an ofset.
Now my question: is this worth it?
Probably not and that is why nobody is doing this exercise. Especially not if you have >100 valves on a plant.
My advise would be to just accept that the relation is not linear. It looks like you already did a flow test and as long as you know that 10% movement changes the flow with 30% you know how to control your process.
Gr.
Terje
- Thread starter
- #6
Thanks for the post terje61. I have to agree that no matter what calculation is made, trial and error is how it ends up. Was looking for simple piece of software that did the calculations per your bullet point #3. I was trying to avoid remembering geometry class but appears I need to deal with it :
-
1
- #7
What has been said by Terje is right. However, you might help mitigate your situation by using a scotch yoke based actuator. Scotch Yoke's tend to travel more slowly at the ends of stroke than the center of the stroke.
Root around on this site and see if you can find any help. Beck DOES NOT make Scotch Yoke actuators, but they have to do some actuation tricks with their actuators to make them more effective, so they might have something for you.
Rotork makes Scotch Yoke electric actuators in their SM5000 series actuators. You might root around on their site and find something as well.
rmw
Root around on this site and see if you can find any help. Beck DOES NOT make Scotch Yoke actuators, but they have to do some actuation tricks with their actuators to make them more effective, so they might have something for you.
Rotork makes Scotch Yoke electric actuators in their SM5000 series actuators. You might root around on their site and find something as well.
rmw
- Thread starter
- #8
tr1ntx
Mechanical
- Jul 20, 2010
- 285
ray1761 - my 2 cents worth is that you would have to go to a more complicated linkage arrangement than a rod & lever arm in order to match the non-linear flow curve. Maybe I'm being Master Of The Obvious and you already know that, but it is more of an operation than what you mentioned of:
"By connecting the linkage in the correct location on the arms of the valve and the actuator, I should be able to get a linear flow output through the valve that corresponds with the actuator position"
You've got to have a mechanism which will match (or approximate closely enough) the exponential (or quadratic or partial sine wave or whatever it is) flow curve. Then, like terje61 said, if densities and pressures change, it will shift the curve. Maybe that is splitting hairs too much and you aren't interested in that tight of control, just in following the general shape of the curve and being "in the neighborhood".
The first step is a good ol' exercise in Analytical Geometry, evaluating the flow curve to get a mathematical approximation. That is not too daunting a task for someone, unlike me, who remembers that stuff. Or some "CFD" program (Computational Fluid Dynamics) should be able to handle it. Then the next step is an exercise in Kinematics, designing a linkage arrangement to approximate your mathematic relationship. I don't remember if you can get there using a combination of circular arcs (lever arms). It is likely going to be a slotted plate/follower type arrangement, but the slot will need to be custom machined to generate the proper motion.
(Now just having said that, and having had another swig of Diet Pepsi this morning, it strikes me that surely someone already makes such a device. It would have to be adjusted/tuned to each installation, but we're not talking space travel here, just getting closer to the curve than the linear actuator.)
"By connecting the linkage in the correct location on the arms of the valve and the actuator, I should be able to get a linear flow output through the valve that corresponds with the actuator position"
You've got to have a mechanism which will match (or approximate closely enough) the exponential (or quadratic or partial sine wave or whatever it is) flow curve. Then, like terje61 said, if densities and pressures change, it will shift the curve. Maybe that is splitting hairs too much and you aren't interested in that tight of control, just in following the general shape of the curve and being "in the neighborhood".
The first step is a good ol' exercise in Analytical Geometry, evaluating the flow curve to get a mathematical approximation. That is not too daunting a task for someone, unlike me, who remembers that stuff. Or some "CFD" program (Computational Fluid Dynamics) should be able to handle it. Then the next step is an exercise in Kinematics, designing a linkage arrangement to approximate your mathematic relationship. I don't remember if you can get there using a combination of circular arcs (lever arms). It is likely going to be a slotted plate/follower type arrangement, but the slot will need to be custom machined to generate the proper motion.
(Now just having said that, and having had another swig of Diet Pepsi this morning, it strikes me that surely someone already makes such a device. It would have to be adjusted/tuned to each installation, but we're not talking space travel here, just getting closer to the curve than the linear actuator.)
tr1ntx
Mechanical
- Jul 20, 2010
- 285
Duh . . . I'm a little more awake now. Like rmw said, a Scotch Yoke is the way to go. Its output is a sine wave curve and can be adjusted/tuned/configured to be close to your valves' flow curves. They are used specifically for your purpose.
Otherwise you can split the flow curve into segments (at least 3) for linear approximation and configure a stepped sequential linkage system for the linear actuator so that it acts at different points on the lever arm at different portions of its stroke, giving you different rates of disc opening for the fixed stroke speed.
(Light bulb!) You said you have a motorized actuator, do you mean an electric linear actuator ("electric cylinder") like a Duff-Norton or Allied? If so, and it's PLC-controlled, you can program the PLC to move it different incremental amounts in relation to the input signal and thereby accomplish the linear approximation.
Alternatively, although I've forgotten most of what I once knew about motor speed control and thus don't know how or if this could be economically accomplished, you could vary the speed of the motor in order to get the varying rate of disc opening.
The jist of it is that you're going to have to open the valve fairly slowly over the first 0 to 20-25-30%, then at a faster rate over the middle portion of the range, then slow again the last 20-25-30%. (This is for a 3 segment linear approximation, with a PLC you can fairly easily use more segments and get a closer approximation. Or if you can generate the mathematical equation approximating the flow curve and if your PLC has the processor for it, you can program the equation directly and avoid all the complications.)
Otherwise you can split the flow curve into segments (at least 3) for linear approximation and configure a stepped sequential linkage system for the linear actuator so that it acts at different points on the lever arm at different portions of its stroke, giving you different rates of disc opening for the fixed stroke speed.
(Light bulb!) You said you have a motorized actuator, do you mean an electric linear actuator ("electric cylinder") like a Duff-Norton or Allied? If so, and it's PLC-controlled, you can program the PLC to move it different incremental amounts in relation to the input signal and thereby accomplish the linear approximation.
Alternatively, although I've forgotten most of what I once knew about motor speed control and thus don't know how or if this could be economically accomplished, you could vary the speed of the motor in order to get the varying rate of disc opening.
The jist of it is that you're going to have to open the valve fairly slowly over the first 0 to 20-25-30%, then at a faster rate over the middle portion of the range, then slow again the last 20-25-30%. (This is for a 3 segment linear approximation, with a PLC you can fairly easily use more segments and get a closer approximation. Or if you can generate the mathematical equation approximating the flow curve and if your PLC has the processor for it, you can program the equation directly and avoid all the complications.)
- Status
Similar threads
- Question