Author Topic: Radial 3-cylinder steam project  (Read 10449 times)

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Radial 3-cylinder steam project
« on: May 07, 2012, 06:37:26 PM »
This is a project I’ll devote some time to, and there will be a lot of waiting around for parts as I’m not going to do the machining myself. In other words, don’t expect the build to move rapidly.

Plans started with a Cygnet Royal and some computations. Figured I needed a 2.5:1 scale in order to get 1HP at ~2000rpm and 2HP at ~3000rpm. I won’t need much more than 1HP, but some parameters are guesstimates, so I stay on the safe side. Also, it would be nice not having to push it towards its max all the time (I’ll team it with a 1HP boiler).

I built a CAD model (SolidWorks, attached pic) after scaled drawings, played a bit around with it to learn the operating details like the orbital valve, and now I’m starting to build a second model I’ll use as a blueprint when machining the parts.

Some changes are to be made thou, especially with the crank mechanism  :scratch: crank arms offset from piston centre worries me, and with a master/slave arm system I just can’t seem to get enough room for the shaft dimensions and bushings I need without increasing stroke length  :bang: I’ve had a look at the “fork & blade” system used on many V-engines, and a triple one is probably doable but I’m pretty sure a noob like me would mess it up. Best way I see is plain narrow crank arms and offset cylinders. The symmetry of the looks isn’t all that important to me, and I avoid over complicating things. Is there a big vibration penalty to this? We’re talking about 1cm offset between them  :scratch:

I’m also building a primary balance model for it  :smart: in Matlab (see attached pic). I’ll treat piston and wrist pin as one unit, crank arm as one, then there is the offset crank shaft, and the orbital valve (out of phase with the crank). Plan is to have the model make vectors of locations of the centres of gravity over one full rotation, in order to assign mass to them and plot the total centre of gravity over the cycle, which I can hopefully use to calculate how to shave the counterweight  :thumbup:

Crank casing will be 1st part btw, I've located some 6" diameter 6082 alu round bar for it  :ddb: But some more drawing will have to be done before I can cut  :coffee:

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #1 on: May 08, 2012, 02:14:31 AM »
Had an hour this morning to play with Matlab,

Black squares are CoG of the pistons, offset out along the line from the wrist pin by a variable.
Yellow squares are CoG (axis) of the orbital valve and supporting pin
Red squares are CoG (axis) of the offset crank shaft

The rest of the coloured lines are the crank arms, and the square marker on them identifies the CoG of that crank arm.

Inputs to the model will be material weights and CoGs as computed by SolidWorks, and there will also be tolerances, but it should get me as close as I can get.

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #2 on: May 08, 2012, 03:20:51 AM »
Back to the crank, by "fork and blade" setup I'm referring to something like the attached scribble, only built differently, probably aluminium with bronze bushings. I haven't seen such a triple setup anywhere, I don't know if that's for a good reason. I'm still a bit undecided on what to do here  :scratch: any input on the subject would be highly appreciated  :poke:

fork & blade vs offset cylinders...

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #3 on: May 08, 2012, 08:01:40 AM »
I just had a revelation skimming through Elmer's radial engine  :coffee: It uses way thinner shafting than the Cygnet, at the same cylinder volume, perhaps not advisable to run at the same max RPM, but it was reassuring reading. I did a sketch with my shafting dimension references somewhere inbetween the two, and was able to fit a bushed master rod large enough to accommodate bushed slave rods. Main shaft remains 20mm, crank shaft 1/2", and slave pins 5/16". Back to not eliminating any options at all, I'm more undecided than ever. This one would actually be my favorite ATM.

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #4 on: May 08, 2012, 11:17:14 AM »
Geometry of the primary balance model for the master/slave rods gets even more cryptic than before  :smart: Red is the master rod, cyan and magenta is the slave rods. But now I can combine the entire crank into the same body mass as the master rod. Only other bodies are the two slave rods, the pistons/wristpins, and the orbital valve/pin.

Offline PekkaNF

  • Hero Member
  • *****
  • Posts: 2523
  • Country: fi
Re: Radial 3-cylinder steam project
« Reply #5 on: May 08, 2012, 01:58:56 PM »
Very analythical approach. I like You probably have stubled on this (or related) documents: http://www.enginehistory.org/NoShortDays/Vibration.pdf

I have this faint memory about W or semiradial engine, but as far I can remember there was no magick on cylinder angles.

Like the master/slave conrod, I think someone said me that it is nowhere perfect, but good enough to get job done.

My daughter insists on this one:
 :coffee: :wave:

Pekka

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #6 on: May 09, 2012, 04:37:59 AM »
Hi Pekka, and thanks for your response, it was some quite interesting reading.

Like the master/slave conrod, I think someone said me that it is nowhere perfect, but good enough to get job done.

The paper had that very same point of view. Seems like the only practical approach when designing with many cylinders, no doubt, but I end up asking myself whether it can possibly be an optimal solution to my simple problem.  :scratch: I would think offseting cylinders and using plain simple crank arms would introduce a lot less fuzz to the system. The forked option hasn't quite left my mind yet either.  :scratch:

My daughter insists on this one:
 :coffee: :wave:

You shouldn't give her too much coffee, it gets addictive :D take it from an addict  :coffee:
Just joking.. Greets from your neighbour to the northwest  :wave:  :wave:

Offline PekkaNF

  • Hero Member
  • *****
  • Posts: 2523
  • Country: fi
Re: Radial 3-cylinder steam project
« Reply #7 on: May 09, 2012, 05:42:03 AM »
Hello,

Happy you found the reading entertaining.

My gut feeling is that the fork system looks odd with three cylinders, you increase complication and risks to trade off a little weight and lenght saving over all conrods over same journal arragement. Also on your sketsch the fork is pretty wide and I don't see how it could compete with master/slave arragement in mechanical sense of benefits? Unless, if you are drawn to fork/blade system. You could stack (interleave) the fork differently to reduce the span of the single fork.

I had a Harley and it had fork, no doupbt it's fine for two cylinders under certain conditions. I think only you can put prorities on order. Similar conrods on common axle will be simplest solution, but will make crankcase a little longer and will introduce some prize moment there, I believe old times that was main reason to use fork. You could make the crank just a little beeffier for torsional forces.

Just my gut feelings, I have never build motor from scratch.

RGDS,
Pekka

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #8 on: May 09, 2012, 06:53:35 AM »
Thanks for the input Pekka,
I made a different fork sketch that I plan to simulate with some forces to see if I can strengthen any weak spots, only problem is I haven't done such simulations since school, and I don't remember exactly how to. Anyways, this would be the outter rod, if that one can be done right, the other two probably can as well. No closer to any decision  :bang:

Kjetil

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #9 on: May 09, 2012, 09:22:49 AM »
I attached a visualization of the forked crank. It doesn't look all that bad, the worry is ofcourse the outter fork.

Offline BillTodd

  • Hero Member
  • *****
  • Posts: 1102
  • Country: 00
  • Colchester Essex (where the lathes were made)
Re: Radial 3-cylinder steam project
« Reply #10 on: May 09, 2012, 10:41:43 AM »
Is there any reason why you cannot off-set the cylinders by the width of the big-end? - It would mean all the con-rods could be identical and simpler to construct.

Bill
Bill

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #11 on: May 09, 2012, 11:22:13 AM »
Is there any reason why you cannot off-set the cylinders by the width of the big-end? - It would mean all the con-rods could be identical and simpler to construct.

Bill

Nope, no reason whatsoever, it's probably the way to go. Just had to skim through the options first. Think it'll actually be easier to balance properly too, with all rods being identical.

Offline PekkaNF

  • Hero Member
  • *****
  • Posts: 2523
  • Country: fi
Re: Radial 3-cylinder steam project
« Reply #12 on: May 10, 2012, 03:52:47 AM »
I attached a visualization of the forked crank. It doesn't look all that bad, the worry is ofcourse the outter fork.

Is there a reason why do you must place the fork/blade such that one will span the whole distance on the crank pin? If you allow a little offset or cylinder and do all con-rods forked and anly middle one a little wider fork? Should probably C-O-C one, because I'm not sure if my explanation does any good.

PekkaNF

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #13 on: May 10, 2012, 08:22:07 AM »
I understand the suggestion Pekka, and it would perhaps be a better choice than the setup I sketched, but I think my mind is settled now, on plain conrods and cylinder offset. Just makes the most sense.  :beer:

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #14 on: May 10, 2012, 12:38:27 PM »
So, the crank casing got a bit deeper, so did the balancing eccentrics, but crank casing width/height could be reduced some without the crank arms colliding with the cylinder wall, and I was able to also shave some millimetres off the crank arm length due to the reduced size of the crank mechanism.

From a cosmetic point of view, a cylinder exterior like this one would even hide the cylinder offset from eye. Just a thought.

But the more I read on the topic of crank arms, the more confident I get that this is the best solution for me;

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #15 on: May 10, 2012, 04:36:29 PM »
OK, I did some programming, and plotted the centre of mass over a cycle. Individual component centre of mass is rough guesstimates for now, I didn't bother to make exact numbers as they may well change. All I wanted to see if there was a symmetry. The black squares in the middle marks the centre of mass over the cycle. Looks pretty good if I got it all right, i simply took R=1/M*sum(r*m) where R is the overall centre of mass, M is the total mass, r is component position and m component mass.

This is what the code looks like in Matlab;
Code: [Select]
% MATLAB R2010b
figure(1);
hold;
syms v

nr_of_steps = 18; % number of positions to simulate for one revolution

axis_length = 130; % length of displayed piston axes [mm]

L=72; % length from chank shaft to wrist pin along the crank arm [mm]
ca_offset = 30; % crank arm centre of gravity, outwards from crank [mm]
p_offset = 3; % offset of piston CoG axially outwards from wrist pin centre [mm]
valve_pin=[1.95*2.5;-0.34*2.5]; % initial coordinates of the valve pin [mm]

% masses
m_valve=214.5; % [g]
m_valve_pin=3; % [g]
m_wrist_pin=22.6; % [g]
m_wrist_pin_bushings=3; % [g]
m_piston=206.4; % [g]
m_crank_arm=33; % [g]
m_crank=50; % [g]
m_crank_bushings=5; % [g]
m_total=m_valve+m_valve_pin+m_crank+3*(m_wrist_pin+m_piston+m_crank_arm+ ...
    m_wrist_pin_bushings+m_crank_bushings);

% compute piston offset vectors;
p_offs_1 = [0;p_offset];
p_offs_2 = [cos(2/3*pi) -sin(2/3*pi);sin(2/3*pi) cos(2/3*pi)]*p_offs_1;
p_offs_3 = [cos(4/3*pi) -sin(4/3*pi);sin(4/3*pi) cos(4/3*pi)]*p_offs_1;

% plot diagram piston axes, 120 degrees angle to each other
plot([0 0],[0 axis_length],'-k')
plot([0 -axis_length*sin(2/3*pi)],[0 axis_length*cos(2/3*pi)],'-k')
plot([0 -axis_length*sin(4/3*pi)],[0 axis_length*cos(4/3*pi)],'-k')

% loop for each angular position:

for a=-pi:pi/(nr_of_steps/2):pi

    A=[cos(a) -sin(a);sin(a) cos(a)]*[0;17]; % A = rotated crank shaft position
    cg_c=A;
    t=asin(A(1)/L); % angle from crank shaft to x=0 (top piston)
    B=[cos(t) -sin(t);sin(t) cos(t)]*[0;L]; % rotate top crank arm in place
    AB=[A(1)+B(1) A(2)+B(2)]; % AB = position of top wrist pin
    plot(A(1),A(2),'bs')
   
    cg_ca1=[A(1)+B(1)*(ca_offset/L);A(2)+B(2)*(ca_offset/L)];
    plot(cg_ca1(1),cg_ca1(2),'rs')
    plot([A(1) AB(1)],[A(2) AB(2)],'-r') % plot top crank arm from A to AB
    cg_p1=[AB(1)+p_offs_1(1);AB(2)+p_offs_1(2)];
    plot(cg_p1(1),cg_p1(2),'ks')
   
    % valve mass centered around valve pin
    cg_v=[cos(a) -sin(a);sin(a) cos(a)]*valve_pin;
    plot(cg_v(1),cg_v(2),'-ys') % plot orbital valve indicator
   
    % the top crank arm was the easy part, now we make a linear function of
    % the other two piston axes, then solve the equation for angle where it
    % intersects the wrist pin position of the crank arm. We get two
    % solutions as we intersect at two positions, determine the correct one
    % based on sign.
   
    % tan(30deg)=y/x, tan(pi/6)=y/x, y=x*tan(pi/6)

    % x=0:1:axis_length;
    % plot(x,-x*tan(pi/6),'-c') % this is the right hand cylinder axis
    % x=-axis_length:1:0;
    % plot(x,x*tan(pi/6),'-m') % this is the left hand cylinder axis

    % first solve equation, then evaluate expression, then remove imaginary
    % component
    g=real(eval(solve('L*sin(v)+A(2)=(L*cos(v)+A(1))*tan(pi/6)',v)));
    % evaluate both answers to determine the correct one
    v1=[cos(g(1)) -sin(g(1));sin(g(1)) cos(g(1))]*[L;0];
    v2=[cos(g(2)) -sin(g(2));sin(g(2)) cos(g(2))]*[L;0];
    if (v1(1)<0 && v1(2)<0) B=v1;
    else B=v2;
    end
   
    cg_ca2=[A(1)+B(1)*(ca_offset/L);A(2)+B(2)*(ca_offset/L)];
    plot(cg_ca2(1),cg_ca2(2),'ms')
    plot([A(1) A(1)+B(1)],[A(2) A(2)+B(2)],'-m')
    cg_p2=[A(1)+B(1)+p_offs_2(1);A(2)+B(2)+p_offs_2(2)];
    plot(cg_p2(1),cg_p2(2),'ks')

    g=real(eval(solve('L*sin(v)+A(2)=-(L*cos(v)+A(1))*tan(pi/6)',v)));
    v1=[cos(g(1)) -sin(g(1));sin(g(1)) cos(g(1))]*[L;0];
    v2=[cos(g(2)) -sin(g(2));sin(g(2)) cos(g(2))]*[L;0];
    if (v1(1)>0 && v1(2)<0) B=v1;
    else B=v2;
    end
   
    cg_ca3=[A(1)+B(1)*(ca_offset/L);A(2)+B(2)*(ca_offset/L)];
    plot(cg_ca3(1),cg_ca3(2),'cs')
    plot([A(1) A(1)+B(1)],[A(2) A(2)+B(2)],'-c')
    cg_p3=[A(1)+B(1)+p_offs_3(1);A(2)+B(2)+p_offs_3(2)];
    plot(cg_p3(1),cg_p3(2),'ks')
   
    % Centre of mass R of a system of particles of total mass M is defined
    % as their average of positions, r, weighted by their masses m.
   
    R=1/m_total*(cg_v*(m_valve+m_valve_pin)+(cg_p1+cg_p2+cg_p3)* ...
    (m_wrist_pin+m_wrist_pin_bushings+m_piston)+(cg_ca1+cg_ca2+cg_ca3)* ...
    m_crank_arm+cg_c*(m_crank+3*m_crank_bushings));
    plot(R(1),R(2),'ks')
end

axis equal
hold;

Offline PekkaNF

  • Hero Member
  • *****
  • Posts: 2523
  • Country: fi
Re: Radial 3-cylinder steam project
« Reply #16 on: May 11, 2012, 05:54:10 AM »
Looking good. Radial moments should look symmetric anyways. What kind of moments do you get between the crank disks? I can't remember the english word for Kippmoment (DE). Long time ago I spoke to people who actually designed engines and if I remember correctly they were not that concerned balansing or vibration, but crankshaft wobling wildly and wearing damper out or snapping off somewhere. Don't remember much.

I'd like to see that engine build.

PekkaNF, annother coffee addict. I have cut down to four before lunch - on my workking days :beer:

Offline Kjetil

  • Jr. Member
  • **
  • Posts: 16
Re: Radial 3-cylinder steam project
« Reply #17 on: May 11, 2012, 08:20:43 AM »
Pekka,
Kippmoment translates to "overturning moment" by google translator, but I'm not quite sure that's an accurate translation. Searching around brings up mainly building stability articles and this how-to, amongst other things referring to the Miles equation, but I can't immediately make much sense of it (then again, I'm not exactly a mechanics wiz). I text messaged a German friend of mine with the term but she's probably at school right now, don't know if she can help me out though, she's not a technician.

Another thing that boggles me when it comes to rotational physics and moment of inertia etc is the crank arms and pistons, in the sense that they are linked to the rotating system, but not directly a part of it as they don't rotate. All I've done so far is look at the mass locations over time, not the forces at play (centrifugal, piston acceleration, etc).

As a curiousity; In the Matlab plot I posted previously, the overall centre of gravity seems to be a circle centered around the origin, but it's actually off center by a tinsy bit (x=-0.0870mm, y=0.3352mm)  :zap: I'm not sure why this is, if it could be a numerical error somewhere. I averaged (min+max)/2 values for both axes over a cycle with 0.25 degrees resolution, since I noticed the vector length from origin to each plotted point varied. It would not pose a problem to perfect primary balancing though, but I'd like to understand why it's not perfectly centered.

Anyways, the crank casing is coming nicely together in SolidWorks, and I'll be ready to have it cut in not too long. The first small step of the actual hardware work  :beer:

And good luck on your coffee cutdown, I know I too should set myself such a goal one of these days  :coffee:

EDIT: the plot is not a perfect circle even when centered around it's mean value. Analyzing the circle (no offset added) I see i get a circle with a small hump where the crank shaft nears a cylinder. Or at least that's what I think I see judging from the numbers running down my screen. radius varies from 7.3mm between pistons to 8mm when crank shaft is near a piston (and thus the piston is at it's outer extremity).
« Last Edit: May 11, 2012, 08:45:28 AM by Kjetil »