Balancing Engines

[petertha]

For each situation center of gravity was asked (One click)
Top 0 mm and 22.77
45 -1,18 21.42
90 -1,68 18.65
135 -1.18 16.61
180 -0 15.97

Thanks Neils. So to continue on...

I made a graphical plot of your CG points, (X=0, Y=22.7) then (X=-1.18, Y=21.42) ... etc. which correspond to the 180 deg rotation segment at 45 deg stops. I'm not sure what the curve itself would look like intercepting these points, I just drew a spline for demonstration. Presumably the remaining 180-360 deg rotation portion would be the same curve mirrored on the vertical datum line? Now what? By adding more or less counter weight to the assembly & re-calculating CG's & a new resultant curve, what are we tryin to achieve? A minimum area curve? A curve shaped in a certain orientation? An area 'lowered' to the 0,0 datum?

I've always thought the Cad power of 1-click CG determination of assemblies could be put to some powerful use for engine design, but I'm just not clear how!

 

[steamer]

Dave,
I worked the Sulzer low speed diesel ships because in the crankcase you could forget it was a diesel the lower running gear is the same as the steam engine days.

Notice I changed my avatar to an old snap shot of me in a Sulzer RLB 90. The piston is 900mm diameter and has a rod that goes to a crosshead guide just like a steam engine. The gland was used to keep the crankcase and the lower piston space seperate.

In the photo I was ridding the piston down so I would be at the right height to use a 4" angle grinder to grind the ridge to pull the piston which weighs 4.5 tons with the rod.

The top speed of the low speed Sulzers is 100-120 rpm depending on the model.

Dan

Hey Dan! Cool! Can you imagine an open crank pit with forced lubrication!...You have to go about yout duties with slickers on....YIKES!

Dave

 

[Niels Abildgaard]

Hello Peter

Computers cannot make experience and commonsense superfluous.If I was making a Lawnmover engine I would ad so much mass that travel in the two directions were equal.Worlds best motorbike MZ250 was balanced no traverse and a lot up and down .
Due to a rather clever motor mount rubber spring system it was as pleasant vibrationwise as a BMW.I have had both and prefer ed
MZ

 

[petertha]

...cannot make experience and commonsense superfluous.

...neither of which I have in significant quantity! ;D

I guess I'm saying... I'm drawing an engine in Cad anyway. So I have the ability to easily determine the CG's of individual engine components and their their relative movement paths. Does this ability help me in any way towards the general balancing guideline that is often referenced to -> "counterweight must balance half of all reciprocating mass, plus all of the rotating mass".

 

[Niels Abildgaard]

Hello Peter

Exactly
If you have identical travel sidevise and upwards You have balanced all rotating and half the oscillating.

 

[JorgensenSteam]

Balancing can be done on the drawing board.
Model Your engine in 3D Cad and eliminate the stationary parts.
Put the moving parts in let us say 8 positions around the working circle.
The puter can easily find the common center of gravity and you can then plot the trace during one revolution.
Ad mass (or remove) on counterweigth until center of gravity movement is minimum in the direction where vibrations will harm most.
I have tried to model real connecting rods,and compared machine calculated mass with measured mass.
Difference was usually within 5 parts per thousand.

I will try and clarify what Niels is saying above.
Niels will have to correct me if I don't understand it correctly.

Balancing an engine can be done using a 3D modeling program that calculated the center of gravity for assemblies. (And maybe in a 2D drafting program if that program calculates the center of mass of multiple objects).

Model only the piston, ring, connecting rod and nut, crosshead, crosshead pin and nut, connecting rod and brasses, crankshaft, and the proposed counterweights, and put these in an assembly.

With some 3D programs, depending on which options you have, you can output data for a revolving assembly, just as Charlie Docksteader does with his valve gear program. If you know how (I don't yet), you can find the X, Y and Z positions of each moving part, and also other data such as velocity, acceleration, and apparently center of gravity not only of each part, but of the assembly as a whole.

Have the 3D program rotate the parts for one revolution (360 degrees), and record the X and Y location of the center of mass at eight different points (or record continuously if your program will do that). I guess we could consider the Z axis, but lets keep it restricted to changes in the center of mass in the X and Y direction only.

As the parts rotate, the center of mass will probably shift around in an elliptical shape, if the location of the center of mass of the entire assembly is plotted in a polar plot.

Now adjust the mass of the counterweight(s) (some crankshafts have a single crank disk and a single counterbalance weight, and some have two crank disks and two counterbalance weights) so that the change in the location of the center of mass of the assembly is minimized in either the vertical or horizontal direction.

All balancing of engines is a compromise between vertical and horizontal balance, so pick which axis (X or Y) in which you want to minimize the vibration from the engine, or make the X and Y axis vibration equal if so desired.

If we remove 70 gram from counterweight as shown on picture center will not move sidewards but up and down 11.13 mm compared to 6.8 before.Perfect transverse and horrible vertical.

So if I understand the above correctly, if you remove 70 grams of mass from the counterweight in the example, then you get a flattened ellipse, which becomes a vertical line.

Computers cannot make experience and commonsense superfluous.

This is my favorite line. My slant on it is "The data that your computer produces is only as good as the data that the person who input the data, and who programmed the computer, or in other words, garbage in, garbage out. Neils solution for balancing engines is a simple and elegant solution, and yet not one I would ever have thought of. Yes, experience and common sense can easily trump everything else.

If you have identical travel sidevise and upwards You have balanced all rotating and half the oscillating.

If you look at the individual center of mass of each separate component, and the movement of the same, then I would think you would have to do a vector sum of the movement of the center of mass for all of the rotating parts. I think this is what the computer program is doing for you though (calculating the movement of the center of mass for the assembly as one entity).

Pat J

 

[petertha]

Exactly...If you have identical travel sidevise and upwards You have balanced all rotating and half the oscillating.

So for graphical representation using the same sketch.... Example Engine A has the resultant curve using your example x,y data. The red arrow is the vertical extent, green arrow is the horizontal extent (what you call travel). Example Engine B we have made some internal changes, like counterweight size, different material density, dimensional layout or whatever. Now it's resultant curve shape is defined by the orange line for example. It's vertical extent is now reduced to a value closer to the horizontal value. So this means we have acheived a net improvement in terms of balance? And the area within the curve or shape is of no real consequence value for these purposes?

 

[JorgensenSteam]

I am guessing here, but it seems like you have to do a polar plot, and use the entire 360 degrees, and the shape has to be like an ellipse?

The ellipse does not necessarily have to center on the 0,0 point, but if you take the center point around which the combination of moving part masses are moving, and put that at 0,0, then you should seen X and Y numbers that alternate between positive and negative?

Am I correct on this?

I think the area under the curve always has some meaning, but I can't remember in this case what it means.
The slope of the line of velocity is acceleration (rate of change of velocity).

 

[petertha]

I am guessing here, but it seems like you have to do a polar plot, and use the entire 360 degrees, and the shape has to be like an ellipse?

I'm not sure myself, thats why I was asking for clarification!

To me, a polar plot would allow a series of 2 dimensions: an 'amount' & some corresponding angle, like '7.5' at 135 deg. I *think* he is saying at a particular crank angle position, say 45 deg, the combined CG the entire assembly occurs at X=-1.18 & Y=21.42. So I'm not clear how a polar could display 3 parameters simultaneously? (degrees, X and Y). But I'm certainly no math whiz & rarely use plots like this.

I'm now thinking maybe the 'half plots' I was volunteering earlier as a graphical representation of his example data (representing 0-180 deg rotation) might be misleading & incorrect though in terms of balance. Presumably the other half of data would be symetrical though (the 180-360 deg crank rotation portion). So maybe the correct extent of the horizontal (green arrow) spans the symetrical shape curve & right away looks closer in amount to the vertical. But maybe the goal is still to get the horizontal & vertical as close to equal?

In terms of your comment that the "shape has to be like an ellipse"... I really can't say. Hopefully more input will reveal the light!

 

[JorgensenSteam]

I read about the data that Alibre will produce from a motion study, and you can get position (X,Y), velocity, and acceleration, in a printout like a spreadsheet, and also get a graph of any of these valvues.

I am sure you could import the data into a spreadsheet, and calculate a number of values.

I remember F = m A, (force = mass * acceleration) from years ago, and I guess you could calculate the total forces in the X and Y direction as the engine rotates, but I am not sure exactly how.

I would think the plot would have to be symmetrical, since the same thing happens at TDC and BDC?

I also know that the piston velocity drops to zero at TDC and BDC, and maximum acceleration of the piston will be where the graph of the velocity is steepest.

I will read up on Solidworks, and see what it can calculate.

Pat J

 

[Niels Abildgaard]

Hello Peter and Pat

Pat is rigth.The plot must be a polar plot to be of any use.
Pat is wrong as it will not be symmetrical top and bottom.
If the conrod was very long it would be symmetrical ,but a short conrod makes the motion more violent in top dead centre than the low one.
Again for a new engine I would try to makecenter of gravity travel equal.

 

[JorgensenSteam]

Ok, I think I understand what Neils is doing.

In Solidworks, if you open an assembly, such as Niels has shown, with the piston, connecting rod, and crankshaft, and go to Tools, Mass Properties, the center of mass of the entire assembly is shown in X, Y and Z coordinates.

If you have your constraints set correctly for your mates for the parts, then you can manually rotate the crankshaft, and the rod and piston will follow the movement.

For each new location that you rotate the crank to, you can hit the "Recalcuate" button in the Mass Properies window, and the X, Y and Z center of mass coordinates for the entire assembly will generally (but not always) change to different values.

I will look and see if Alibre will do the same.

Pat J

 

[JorgensenSteam]

Looks like if you open an assembly of the moving parts in Alibre, and select Tools, Measurement Properties, and then pick the Calculate button in the pop-up dialog box, you can also get the X, Y and Z coordinates for the center of mass.

Again, you have to have the constraints (mates) set correctly on the parts of the assembly, so that when you rotate the crankshaft, the connecting rod and piston move correctly.

I will try and post an Alibre file for an example here.

Pat J

 

[steamer]

Another plus for longer conrods is that the maximum piston velocity should be lower with a long rod vs a short one, to say nothing of side forces on the cross head guide which will be lower with a longer rod. With an infinitly long connecting rod the piston velocity profile would be sinusoidal. A real rod will have angularity affects
These affects can have a adverse effects on a crankshaft with a large rotating mass, like a propeller, which could be attributed to balance but in actuallity, are torsional in nature. All the operator will hear is all the bearings hammering.
A length of 4-5 times the crank throw (2 -2.5 times the stroke) is fairly long


Dave

 

[petertha]

...
...open an assembly...with the piston, connecting rod, and crankshaft, and go to Tools, Mass Properties, the center of mass of the entire assembly is shown in X, Y and Z coordinates....For each new location that you rotate the crank to, you can hit the "Recalcuate" button in the Mass Properies window, and the X, Y and Z center of mass coordinates....

Pat, thats exactly what I assumed Neils was showing in his example. Using his data, at a specific crank angle (45 deg), the assembly centroid was determine to be (X= -1.18mm) & (Y=21.42mm). And by assembly, I am assuming it encompasses the individual contributions from all the parts you reference. So the curve I have volunteered as a graphical representation simply connects the X,Y dots in a cartesian manner, but corresponds to the respective 0 to 180 deg crank angle 'stops'. That yields what I loosely called the half-curve. Then if we assume it's symmetrical (left to right, not up & down) it yields the curve I show in post #28 which encompasses the whole 0-360 deg rotation. Point 0,0 reference is far below the screen & doesn't seem to matter I assume.

So if this is correct up to this point, I guess I'm asking for confirmation: is the objective for ideal balance to make the horizontal & vertical 'amounts' EQUAL (as I've shown in colored arrows)?

If these plots or the procedure is incorrect somehow, please elaborate & correct me.

 

[petertha]

Hello Peter and Pat
Pat is rigth.The plot must be a polar plot to be of any use.

Okay... maybe you could elaborate a bit. What IS a polar plot? How does one construct one? I'm assuming this graphical representation of your example assembly [Angle,X,Y] centroid data is therefore NOT a polar plot?

I feel like the apprentice baker who wants to make nice bread... so far we have 100% agreement on 'flour, water, eggs & salt'. ;D
Just kidding, this has been enlightening & interest to me. I just don't quite have a complete understanding.

 

[mklotz]

What IS a polar plot?

In an x-y plot we plot a point by going out to the x value on the x-axis and then up/down to the y value on the y-axis.

In a polar plot we have a distance r and an angle theta. We plot a point by finding the angle on the angular scale and then going out along that angle line a distance equivalent to r.

The relations between x,y and r,theta are:

r = sqrt (x^2 + y^2)
theta = arctan (y/x)


x = r*cos(theta)
y = r*sin(theta)

Some functions are simpler to plot in one system; some in another. For instance, one would need to specify a large number of x,y pairs to plot a circle on an x-y (Cartesian) plot. On a polar plot, a circle is simply r=constant for all values of theta.

Similarly, plotting a rectangle in polar coordinates would be nightmarish. In Cartesian coordinates it would be dead simple.

 

[Niels Abildgaard]

Hello Peter

The ideal balance for a single cylinder engine must be the one where broken screws,carburetors etc are minimum.
This varies from case to case.It does not make any sense to go outside the envelope of no balance for oscillating parts (pure up and down forces) and full reciprocating(meaning only transverse forces.
All this is not completely true due to the finite length of connecting rod but this starts to get out of hand.
I still think Your onion shape curve is wrong but I do not know how it should look.Could we make it free for all?

 

[Niels Abildgaard]

Hello again

The ideal comic strip would be a small animation showing a slowly revolving crank,rod piston assembly with an arrow superimposed showing force magnitude and direction.We shall have three cases .No reciproc balance ,half and total.
Then the role of the socalled second order forces can be shown as well.
My wife is ill so please feel free to enter Your thougths.My soul is sometime a way.

 

[JorgensenSteam]

I guess my interest in balancing steam engines came from this engine that was built by my Dad, and with no counterbalances.

Nice little engine, but it literally hops up and down on the table when it runs, and so that started my thinking about how I could find some sort of semi-accurate method of dynamically balancing a steam model.

I considered adding counterbalances to this engine, but I don't really want to do that without some sort of scientific approach, and I want to balance other engines also in a predictable fashion.

Pat J

Edit:
Peter Quote: I feel like the apprentice baker who wants to make nice bread... so far we have 100% agreement on 'flour, water, eggs & salt'.

Peter, I could not agree with you more. Pat J