Hydraulic circuit design software 9

[frans]

Does anybody in the forum have experience with hydraulic circuit design software. I would like to simulate operating conditions in complex circuits and get results such as flow losses, pressures and temperatures.

 

[HCMechSys]

Thanks, I thought it was useful for me anyway.

1. Regarding the 35 bar vs. 70 bar - I saw your dp/2 equation but thought the software was asking for the full drop across the valve. I think you're right though...

2. Unfortunately the profile you can enter is limited to 10 steps, so I couldn't put in more points to fill in the cosine function a bit better. That is a fairly big limitation.

3. Yes because more of the flow is diverted to the accumulator early in the simulation and less at the end of the simulation, the cylinder manages to get back all the way to the cap end.

My question was referring to the Pa0 and Pb0 and yes I did notice that intially the forces are balanced. But what is the rationale behind the pressure values used? If we were running the machine with a relief valve setting of 2500 psi and lets say we had just retracted the cylinder, then I would expect 2500 in the rod side line and something less (2500Ab/Aa, I think) on the cap side...

 

[PNachtwey]

Actually, just about any set of pressures will do as long as the net force is zero. I have found from research ( playing ) with my hydraulic system, that the pressure does not reach system pressure on the rod side while just holding position without a load. In fact the steady state pressures will be different after an extend move and a retract move. This means that a move from 10 to 20 inches after a move from 20 to 10 inches is different from a move from 2 to 10 inches, stop, and then move to 10 to 20 inches.
When I say different, I mean the pressures will be different. The motion controller makes sure the net force and therefore the acceleration and velocities position the same.

My hydraulic system has pressure transducers on all the ports of the servo valve which is a Bosch NG10 -803. I can monitor position and the pressures at .5 millisecond time intervals and record, save and display the data. I can also use triggers to start the data acquistion like a logic analyzer. These features are built into the motion controller. I used these tools to gain knowledge of how a hydraulic servo system works.

ftp://ftp.deltacompsys.com/public/jpg/PA080005.JPG

I verified my model against my system. It is close, but my hydraulic system is just one system. I have been able to use my simulator to keep customers out of trouble by showing them the simulation of their design was not good. Just like the example above. I need to spend more time doing research ( play ). This is good stuff.

I/We make and sell hydraulic motion controllers. I am interested in the physics of hydraulics and how to really control it.

 

[HCMechSys]

Well I finally got around to modifying the BathFP analysis so that I could reflect the same initial conditions as the analysis in PNachtwey's post.

The updated inputs and outputs can be found at:

http://ca.geocities.com/hchesser@rogers.com/Ex-Eng-Tips2.pdf

Overall the results are much closer now than before. I suspect the only differences now are due to line sizes and minor things I didn't set properly in the BathFP objects.

PNachtwey, do you have measured results from your set-up (thanks for the picture of that by the way - nice equipment) to compare with this analysis? It would be interesting to see how close the analysis comes...

 

[PNachtwey]

That simulation was for a customer's system that didn't control very well. I increased the size of the cylinder in the model until the motion was stable and the A and B port pressure were in the middle third between system pressure and 0. They ended up with a 3.25 inch cylinder instead of the original 2.0. They didn't like having to redo their system.

[detour]
I find that sizing cylinders if very tricky. I don't trust any one set of equations. The natrual frequency method and the VCCM method have flaws. Choose the maximum between the two will generally get good results. However, I prefer the simualtions for the final word. It is too bad that it takes so much effort to get a simulation set up.
[/detour]

I can get the data for my system, in the picture, and I have a simulation for it but that will have to wait till next week as I am going home time. I will comment about HCMechSys's new simulation when I get home and have more time.

 

[EdDanzer]

PNachtwey,
The picture of the test device brings up some comments that may affect stability of the system.
1. The aluminum block that the valve is on will cause some volume variations if pressures run very high.
2. The tube running to the rod end is probably too small and too thin of wall for optimum performance.
3. The cylinder barrel wall is probably too thin for maximum performance.

Depending on the accuracy needed, back pressure caused by too small of plumbing to the cylinder ports can effect response time.
As the cylinder stroke becomes longer in relation to the bore, pressure expansion of the port tubes and the barrel will affect the stiffness of the system.

 

[PNachtwey]

EdDanzer, except for the LVDT feedback being inoperable right now my system in the picture is almost perfect.

ftp://ftp.deltacompsys.com/public/jpg/after%20tuning%20wizard%20bw%205pt1%20hz.png
ftp://ftp.deltacompsys.com/public/jpg/after%20tuning%20wizard%20bw%205pt1%20hz.plt

The links show how accurately I can move this system. The position units are on thousandths of inches so 4000 is 4 inches. Notice that the red line ( actual position ) covers the cyan line ( target position ). This shows the system is perfectly tuned. The lines are color coded. I used my tuning wizard ( auto tuning ) to calculate the gains for the motion controller.
The .plt file is a text version that can be imported into Excel or Mathcad to plot the way you want. This is how I do my research. The time periods are really increments of 1/1024 seconds so 4 means 4/1024 seconds. Notice also the sum of errors squared in the gray box. It is only 435. This is the sum of the error at each timer period squared and
summed for the whole graph. An error of 2 thousands would show as an error of 2 and add 4 to the sum of errors squared. Some people call this the ISE or integrated squared error. The goal is to minimize the ISE or sum of errors squared when tuning a system.

Ed, the simulations in the previous post above are for a customers systems, not mine. The links in this post above are for my system.

1. The aluminum block has the connections for the pipe leading to the rod end. It also has the P and T pressure sensors as well as the P and T ports. How do you get rid of the block and still have all these ports?

2. The pipe to the rod end look big compared to the ports. How big can the be? What choices do I have?

3. The pressure that I am operating at is only 1500 PSI. I doubt I am streching the cylinder in a noticeable way. At least the motion controller doesn't seem to care.

The stiffness of the system will vary a lot depending where you are in the stroke. The cylinder stretch does not affect this in any significant way.

ftp://ftp.deltacompsys.com/public/PDF/Mathcad%20-%20Natural%20Frequency.pdf

This is not for my system, I just want to assure you I know how to calculate natural frequency as a function of cylinder length.

I will get the HCMechSys data tonight hopefully.

 

[PNachtwey]

HCMechSys, your simulation look pretty close after you let the BathFP simulation reach a steady state condition before start the control sequence to the valve. There are a lot of place where BathFP and I could make different assumptions that would cause the rest of the difference. I think I do many more updates. My simulation looks a lot more rounded than the BathFP simulations. This could be cause I use 125 microsecond periods. The BathFP periods could be a lot longer.

You are right to use the sqrt(35bar) for calculating the pressure drop. If you go to the BoschRexroth website you will see the same thing. I appreciate your efforts. Are there any suggestions you have for me besides using SI units?
I know my simulation is not perfect. I don't think any simulation will be. I will try to show a simulation of my almost perfect hydraulic system.

EdDanzer, where would I get the capacitance of the cylinder or pipe. I don't see this number published anywhere.

 

[HCMechSys]

PNachtwey,

I looked at your previous posts and see that you are thinking of making this into a web-based tool. I like that as an idea - here are my thoughts as to how to go:

- on the technical side, I do think you should explore the parameter space of the model to see what are its limitations in terms of system pressure, flowrates, command signals, etc. This is where "playing" with the system you have would come in very handy. This would provide valuable advice to your users as to valid input parameter ranges.
- also, I would check your results against other software tools especially for parameter ranges not easily tested with your system. This is after all why one turns to simulation - to see what system changes produce better/faster performance without having to physically experiment - of course there are limitations to that as many have already noted above.
- the code itself can only analyze the behaviour of the valve-controlled cylinder, which of course is a very common circuit - but on the other hand it would be nice to be able to plug and play with components to make your own circuit - that's where tools like Automation Studio and BathFP are kind of nice
- one thought I had (since I've been "playing" with equipment at the college where I work) is that it would be handy to be able to tune the system parameters to match measured results - a sort of system ID tool, not sure what's out there that does that now - anyone?
- I assume the intention is to add a browser-based GUI, which sounds like "fun" (depending on your definition of course)

I'll look up that Bosch Rexroth valve and see what catalog data they have - hmm - this may be an "excellent" (again depending on your prespective) test question for my students actually....

 

[PNachtwey]

I was thinking of a Java applet. The trick is to get enough detail to make the simulation accurate enough with out making the simulation to difficult to setup with too many parameters. For instance, EdDanzer suggested the cylinder expands. Yes, it does. This has been mentioned to me before at the last IFPE show is Las Vegas. If I have a typical default valu, I would use it. This way a user could just use the default value unless he knows a better value. I am even more interested in the capacitance of pipe and hose because the capacitance of the hose and pipe is so much higher than the cylinder. I don't see capacitance ratings for those item either. How is one suppose to design a decent system without knowing all the facts?

-I would only simulate servo valve controlled cylinders. That is all I am interested in and simulating counter balance valves, sequence valves etc is just more parts that must be modeled.

-The graph of the motion of my system above used my auto tuning algorithm. I can ID systems as a gain and a time constant or a gain and a damping factor and natural frequency. This is good enough for auto tuning. I can then select the desired bandwidth and the tuning software will select either an overdamped or critically damped response that will achieve the desired bandwidth.

Where do you teach?

 

[HCMechSys]

PNachtwey,

I believe the capacitance of the hydrualic line is related to the stiffness of the line - just as an electrical capacitor stores charge/voltage, the hydraulic line stores pressure by expanding (like your cylinder). The line stiffness is related to the material (as well as diameter and wall thickness) of the pipe in question. This is basic pressure vessel stuff - I'll see if I can find a readily available reference for you on this. If the line has significant bends, etc then it gets more complicated. But for a basic tool a simple model should be sufficient as a starting point anyway.

For your GUI, you could have a range of picks for line models - like Schedule 40 steel pipe, SAE 100R, etc. Maybe as a first cut, you could also offer a "rigid line" or "ignore line capacitance effects" pick.

I agree that from a control systems standpoint the circuit can be simplified to a first or second order block, but what is more useful for hydraulic designers is to leave the details of the components in. The power of the Automation Studio or BathFP or tools like it is that it allows the designer to see the effect of swapping in new parts - not a 4-digit accuracy prediction of what happens, just enough to know if you're on the right track.

I teach at a technical college in the Toronto area, we use Parker stations to run experiments on - the biggest limitation I find is not being able to load the circuit...

 

[PNachtwey]

I know what hydraulic capacitance is. I just don't have values for pipes, hoses, and cylinders. Hydraulic capacitance is defined by

volume
-------
bulk modulus of oil

This is for the fluid itself in a vessel that doesn't expand. A cylinder, pipe or hose should have a multiplier like 1.005 for a cylinder which wouldn't expand much or 1.4 for a hose that does. It is this multiplier that the manufacturers don't provide. The capacitance value would allow one to easily combine capacitances in parallel or in series by using the same rules as for electrical capacitors.

BTW, the units or dimensions for capacitance is length^5/force or m^5/N in SI units.
The change in pressure is equal to the change in volume / capacitance.
The rate of change in pressure is equal to the flow / capacitance. Try it! You will see the units work out.

Simulating a simple second order transfer functions is not enough. It is important to do the hydraulic modeling because of the non-linearitiess. I use the model to tell me if the system will not work. If the model works that does not mean the real system will work because there are things like hose that aren't taken into account. However, if the model doesn't work, the real system will not work. If the simulator indicates the system just marginally works, the model will give a pretty good idea of what needs to be improved.

Not being able to load the system is a problem. My problem is that I have only one hydraulic system. I need to use modeling to test on different systems.

We will look at the BathFP software. It looks good enough to keep people out of trouble.

 

[EdDanzer]

Sorry to take so long to respond, and I have to make this short, but will try to provide more information over the weekend.
PNachtwey,
I was not trying to say the system didn’t work, or was poorly designed. These were just observation of possible areas of concern when tuning, and over all system stiffness.
Some of the capacitance of a hydraulic system comes from size change due to pressure, and flow restriction through plumbing, fittings and hose ends. Size change from pressure is mostly diameter increase. This link will show diameter change in a barrel for a seal application:

http://www.danzcoinc.com/html/step_lock_piston_seal.html

 

[PNachtwey]

Oooh. Good info. Now I will must figure out how to use it. I will report back if it makes a signficant difference or not. At least I will know and can put this nagging question in the proper perspective relative to pipe and hose. I still think the affects of cylinder swell are small.

BTW, do you have the same kind of data for hose and pipes? I can just use these values as typical/default values.

I figure you must sell the seals? If so you must be on top of the subject of cylinder expansion. I may be back with questions. What do you know about pipes and hose swell? Thanks, another star for you. Maybe, I am giving them out too easily both both cylinder epansion and BathFP links are gold. Now I just have to fiquire out what to do with them.

All works stops, I figure this out now.

 

[PNachtwey]

My conclusion is that cylinder capacitance is too big to be ignored but it can be taken into acount by fudging the bulk modulus of oil and making it a little smaller. This is normally done anyway to account for air in the oil.

Although the cylinder capacitance changes with pressure, the change is not enough to make a difference and it can be ignored.

ftp://ftp.deltacompsys.com/public/PDF/Mathcad%20-%20cylinder%20capacitance.pdf

Ed, I noticed that you must use Mathcad too.
Back to work.

 

[kcj]

I can open page 2, but barely read it. Should I download to anothe rapplication somehow?
k

 

[PNachtwey]

Page 2 of what? My files are .pdf files. You can expand them using the Acrobat reader.

 

[HCMechSys]

For tube capacity, the correction for bulk modulus is the same as for the cylinder. I found the correction formula in the Keller reference on pg. 63. As PNachtwey points out the hydraulic capcitance is volume/(effective bulk modulus), with the effective modulus corrected to account for tubing and I guess the cylinder as well. The correction is based on radial strain of a cylindrical pressure vessel.

For line models, BathFP uses a similar bulk modulus correction formula given in Keller. The BathFP formula doesn't involve Possoin's ratio, so all you need for the tube material in question is the Young's modulus. BathFP comes with 3 predefined materials - steel (E=207 GPa), a Aluminum/Nickel/Brass alloy called Tungum (E=117 GPa)- used in marine systems, or "flex hose" (E=4.3 GPa).

I was going to try these different values in the BathFP model I did of PNachtwey's system. But the performance is already not that great and it sounds like he doesn't have measured performance of it anyway. Can you supply more appropriate values (maybe for the system you've got at your office), PNachtwey? I'm going to change the line diameter to 3/8" anyway since this is a more typical value than the 1" I have now. Perhaps I should change the cylinder diameter to 3.25" instead of 2"?

One other thing I spotted in your Mathcad simulation, the formula:
MinRodVol=MinCapVol+0.5*CylLen - not sure the second term is dimensionally correct...both of volumes - Cap and Rod - refer to line volume, correct?

 

[kcj]

pg 2 I was referring to EdDanzer's link

k

 

[PNachtwey]

"But the performance is already not that great and it sounds like he doesn't have measured performance of it anyway."

Actually, I do. I am trying to resolve some differences. It seems my perfect system works OK for motion control but as a research systems it is lacking. Ed was right. The pipe to the rod end is too small. The outside diamter is .75 inches.
There is also too much hose between the accumulator and the system. All this hose is 3/4inch inside diameter hose. One other problem is that the system is on a stand and the stand vibrates. This vibration adds another frequency to the pressures or net force.

My model and the VCCM formula predict my system should travel at 52 inches per second while extending. The maximum that it can really do is only 47 inches per second. My model does not include loss for hose and piping. I need to get my stuff together before continuing.

"
Can you supply more appropriate values (maybe for the system you've got at your office), PNachtwey? I'm going to change the line diameter to 3/8" anyway since this is a more typical value than the 1" I have now. Perhaps I should change the cylinder diameter to 3.25" instead of 2"?
"

My system has a bosch NG10 -803 spool 27 GPM valve. The cylinder is a Purakal 2 x 24" with a 1.375 in rod. The load is 600 lbs that rolls on wheels.

The reason why I like the idea of capacitance is that is takes into acount the volume. If the bulk modulus of pipe is used, it still needs to be scaled by the length of pipe used.

Here is a link to a simulation of my system
ftp://ftp.deltacompsys.com/public/PDF/Mathcad%20-%20hydsim-delta.pdf

Reality differs from my simulation in two ways. The simulation goes faster. There is nothing wrong with my calculations. The VCCM formula predicts the same speed. I just don't my calculations are complete because it doesn't take into acount the hose ( too small and too long ) between the accumulator and the valve and the pipe between the valve and the cylinder's rod end. If I lower my models system pressure I can get the same results. Another difference is that the simulation doesn't show the effect of the stand vibrating. At least I think it is the stand. Finally, the graphs of the data is much different extending and retracting. The extending looks similar to the simulation. Notice there is no overshoot in the simulation and there isn'[t in the actual system. However, there is ringing when the system is stopped. Why. Does the system go from being underdamped while stopped to overdamped while moving? In the retract direction the actual system overshoots and looks quite different. Why? I think it is the stand and how it is braced.

I can post lots of .jpgs of my physical system. It would be easier for me just to post the raw data. This way you can import it into Mathcad, Excel or our own software that is free on the web and analyze it instead of just look at it. I need to fix some things before I continue.

 

[EdDanzer]

Guy’s,
I was not able to put together the information over the weekend, to many other unfinished projects so this will be short also.
Kcj,
Try blowing up the page by hovering over it and using the expand button or print it.
PNachtwey,
The expansion for the port tube, is done the same as the barrel. Be sure to multiply the diameter expansion by the length to obtain the volume change created by pressure change. What is the wall thickness of the ¾” tube?
When the cylinder is retracting the pressure in the rod side will be higher than then when extending, yet the flow is less. Depending on the direction of movement the back pressure of the oil exiting the cylinder, on each end of the cylinder will be different through the valve because of area differences and plumbing differences.
The vibration could be coming from the valve modulating to maintain the speed demand.
A 27 gpm valve seems large for a 2” bore cylinder, if the valve has to do a lot of metering at full speed the bore to rod flow differences could cause some problems.
It is best to hard plumb the accumulator with large passages as close to the valve as possible for the best performance.