Constant Force Circuit 1

[lukerobert]

I have a project that I'm working on that requires a constant force (adjustable) to be applied by a double acting cylinder. Force only need be monitored in the extend direction (0-2500 lb f).

The setup looks very similar to a log splitter, but will have a load cell placed between the cylinder and its anchor point to monitor the applied force. This load cell will output a 0-10v signal that will be sent to a process controller.

I intend to place some sort of relief valve on the outlet side of the cylinder to regulate pressure and maintain constant force. This valve would need to accept a 0-10v or 4-20ma signal from the process controller so that the setpoint force will be maintained. It also needs to be fairly fast acting.

My question, does a valve of this type exist?

 

[PNachtwey]

We control these kinds of projects are done all the time. A relief valve is a safety device, not a control device. Usually a very small servo valve or servo solenoid valve is used depending on the rate of force change required and the volume of oil under compression. It looks like you could use a 2 inch ( 50mm ) bore cylinder to get the desired force and a D03, NG6 valve. The valve may be even smaller. The key thing is that the valve have zero overlap and low hysteresis. The other key point is that force will change rapidly with little flow in or out of the cylinder. The closed loop force control needs to be fast.

The key equations are:
P(t)=P(0)+∫(β* Q(t))/V dt
Q(t)=K*x(t)*sqrt(Ps-P(t))
P is pressure
βis the bulk modulus of oil
Q(t) is the flow into a volume of compressed oil
V is the volume of compressed oil
K is the valve flow constant
x(t) is the valve spool position
Ps is the supply pressure
P(t) is the current pressure of the compressed volume of oil
P(t)=P(0)+∫(β* Q(t))/V dt
Q(t)=K*x(t)*sqrt(Ps-P(t))

The net force is Pa*Aa-Pb*Ab-friction

I have some graphical data for sinusoidal testing of welds in the same force range that I can post when back in the office Monday.








Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

Thank you Peter. Yes, you are correct, I am looking for a servo valve that will be controlled by the process controller. I have found the following one manufactured by Moog.

http://www.moog.com/products/servov...o-valves-for-analog-signals-d636-d637-series/

I plan to use a 2" cylinder just as you suspected. But, I'm also thinking that moving to a slightly larger cylinder may help accomplish tighter control. The larger cylinder will be less sensitive to slight changes in flow, no?

Thank you for your help.

 

[PNachtwey]

The valve is a good choice but the 5 liter/min valve is the one to get. You should plug some numbers into my equations to see how fast the pressure/force can change.
A bigger diameter cylinder is good if you are doing position control moving heavier masses.
When doing force control
dP/dt=β* Q(t)/V
Multiply both sides by the area.
dF/dt=β* Q(t)/x
where x is the distance from the end of the cylinder assuming no dead volume.
You can see that the rate in change in force doesn't care much about the diameter so from a purely force control point of view it is a wash.
For laboratory or testing applications I care more about the seal frictions. If the seal friction is the same between the 2 inch and 2.5 or 3.25 inch cylinders then the seal friction will be a smaller percentage of the force when using larger diameter cylinders. I would stick to the 2 inch diameter cylinder and use the savings to buy a better quality cylinder with lower friction.

I doubt a process controller can respond quickly enough. The analog inputs and outputs must be fast too.
What exactly are you doing?

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

Peter, I am designing a drilling rig simulator. The client wants to build a machine that will drill different materials at various RPM and force (weight on bit). So, as the machine drills the weight on bit should remain constant. The weight on bit will be adjustable through setpoint on the process controller.

If not a process controller, then what type of controller could accomplish this?

 

[PNachtwey]

A drilling simulator will be tricky to do right. You need simulate the drill string turning and being disturbed vertically and torque wise by rocks. After the drill string gets long the pipe must be "lifted" to maintain a relatively constant force to offset the weight of the pipe. In other words there must be the ability to control force pushing and pulling. If your simulator can't turn then there must be a away of simulating disturbances at the "drilling end" of your simulator. A real system will have a long drill string that acts somewhat like a spring.

A small simulator system would need to respond very quickly in the millisecond range. A real system will be slower, much slower but much trickier than just trying to maintain a constant force.

I will look into this Monday. I am pretty sure we have customers doing force and torque limiting in drilling applications.


Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

Peter, for this project the drill string will be 5 ft max. They want to use the machine for testing out new bits and testing drilling through various materials (concrete, composite plugs, etc). I will be using a 20 hp constant torque electric motor for rotation controlled by a VFD drive. The max penetration on any drill cycle will be about 3 ft.

 

[Compositepro]

It is very important to define what is meant by "constant". Constant to what tolerance? I've seen many designs that in theory keep something constant but in reality do not compensate for friction or other factors. Of course, if no one defines the requirements then any design will "work".

 

[PNachtwey]

I was hoping to have some actual examples from real drilling applications but I couldn't find any.
Since this is a small model the controller will need to be fast.A PLC may be able to do the job but a hydraulic motion controller would be better especially if you want to collect data and connect to data analysis software like LabView.



Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

Peter, do you have a recommendation for a make/model of Hydraulic Cylinder for this application? I'm looking for something with low friction just as you recommended earlier in the thread. I'm trying to eliminate the possibility of any "stick/slip".

 

[lukerobert]

"It is very important to define what is meant by "constant". Constant to what tolerance? I've seen many designs that in theory keep something constant but in reality do not compensate for friction or other factors. Of course, if no one defines the requirements then any design will "work". "


There are no specific specifications, but naturally the closer to setpoint, the better. I'm hoping to see results in the +/- 3-4% range.

 

[PNachtwey]

Parker bought out Miller Fluid Products many years ago. They made teflon coated cylinders. I would ask Parker.
However! Your designed of a solenoid moving a bellows looks good to me since there will be no seal friction and the solenoid can retract to reduce pressure to 0.

You should be able to get MUCH closer than 3-4% if you use a load cell for force feed back. You should be able to get within 1% if the force is to be constant. We do so all the time.
It is import that the servo valve have 0 overlap or be slightly underlapped ( v-spool ) and have low hysteresis.


Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[Compositepro]

To simulate a system you need to accurately understand or model the system. That would include the elasticity and inertia of the drill string. You could easily get resonance effects due to the spring-mass effect, which may be one of the most important factors to simulate.

 

[lukerobert]

"Parker bought out Miller Fluid Products many years ago. They made teflon coated cylinders. I would ask Parker.
However! Your designed of a solenoid moving a bellows looks good to me since there will be no seal friction and the solenoid can retract to reduce pressure to 0.

You should be able to get MUCH closer than 3-4% if you use a load cell for force feed back. You should be able to get within 1% if the force is to be constant. We do so all the time.
It is import that the servo valve have 0 overlap or be slightly underlapped ( v-spool ) and have low hysteresis"

Thanks again Peter. I will take your advice and find the nearest Parker dealer and go sit down with them. The Moog Valve I'm looking at is supposed to be available with zero overlap and very low hysteresis.

On another note, I noticed that Moog recommends placing an adjustable metering orifice across the head and rod end of the cylinder (after the Moog valve). The say it acts as a dampner and will increase accuracy of system. Ever seen something like that in action?

 

[PNachtwey]

On another note, I noticed that Moog recommends placing an adjustable metering orifice across the head and rod end of the cylinder (after the Moog valve). The say it acts as a dampner and will increase accuracy of system. Ever seen something like that in action?

Yes, this is a common trick. This isn't necessary if you get an underlapped valve because the oil will leak through the valve. The orifice trick is necessary if there is a small over lap on the spool. The orifice will cause the pressure to drop or leak from the pressure side to the tank. The controller will need to always be metering in oil to maintain pressure or force. The spool will only need to move about 1% or so to maintain force. The orifice makes it so the valve spool does not need to be shifted through the overlap.

BTW, the 1% assumes there are no outside disturbances.

To simulate a system you need to accurately understand or model the system. That would include the elasticity and inertia of the drill string. You could easily get resonance effects due to the spring-mass effect, which may be one of the most important factors to simulate.

Yes, in a real system the natural frequency will drop as the drill string becomes longer and heavier. Controlling a constant force is really impossible because it is like trying to control the force applied by a stretched out slinky. The average force can be controlled but it won't be constant because the the disturbances. I don't see how one can physically model as drilling rig with a small lab type model and get everything right.



Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

Peter, we are going to give it our best effort. Keep in mind, this piece of equipment will actually be pretty big. It will have a 15'x15' base with a 22' mast. It will weigh in at around 20 tons. The main use of this machine won't be for exact simulation of drilling, it's more geared toward testing different bits for plugs in the wellbore. The customer understands that there will be 'kick' from the drill string and that this will cause vertical disturbances.

For the adjustable metering orifice, are we talking about something like this?

http://www.surpluscenter.com/Hydrau...-GPM-PRINCE-WNV-400-NEEDLE-VALVE-9-7961-4.axd

 

[PNachtwey]

I would try without the orifice. It looks like you are selecting a good servo valve that shouldn't need the orifice.

So how are you going to measure applied force? There is the load cell option and the differential force method. The load cell option is the most accurate but it doesn't work if the load cell can be damaged easily. The differential force methods requires a pressure sensor on both ends of the cylinder. NetForce=Pa*Aa-Pb*Ab. This method ignores seal friction.

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[lukerobert]

I plan on using this load cell placed between the drilling motor and the drill string. It is a multi axis (axial + torque) cell.

http://www.interfaceforce.com/index.php?5600-Axial-Torsion-Load-Cell&mod=product&show=189

I will likely be contacting you soon to get a quote on one of your single axis controllers.