Camshaft design

[Charles Lamont]

Another example of sliding camshafts were Sentinel steam waggons. They had had poppet valves, with initially two and later three sets of forward cams giving different amounts of cut-off, one set for reverse, and one for 'neutral' with all valves open. Here is a good history:

 

[HMEL]

Is there any place where I can find information on basic cam design? Base diameter, working angle and lift are somewhat determined by the basic engine design. What I am trying to determine is how is the flank radius calculated. On the cam drawings that I have looked at the radius and the center seems to be a rather precise number so I assume that there is some way to establish this. One article states that it is easily established using a CAD program but I don't understand how.

Yes, Standard Handbook of Machine Design addresses this topic.(Schikley and Mischke)
But be prepared this is a special topic involving the follower as well. It can also be found in engineering textbooks usually at the snd year of study of Mechanical Engineering. I am not a believer a cad program can easily establish the profile. However much depends on the speed of the operating cam and the follower. You first need to know the lift of the follower and then you want to design the lifting curve with the right rise. In an ideal situation the flank is giving a smooth upward push and not loading the follower with side loads. But the above book or one like it will give a introduction of the parameters needed to consider.
If nothing else it will give you an idea of how deep you can go into to thia subject. I hope I have not confused you. HMEL

 

[Zeb]

I was just reading about cam design in my machinery's handbook. A lot of folks here should have a copy laying around. It was a nice Christmas gift. I keep mine at work to appear intelligent.

Here's a list of what it covers from an older version:

 

[peterl95124]

Everyone asks about camshafts but no one that I know of has ever asked what pressure valve springs to use.

gbritnell

for my model Merlin engine I measured the weight of the valves plus one third the weight of the rocker arms and computed how much spring force it would take to keep them on the cam during maximum deceleration at the maximum RPM I hoped to run the engine at, then ordered springs with the right size and spring rate, and they worked. for an engine with 48 springs I really didn't want to do this by trial-and-error, I also didn't want to wind my own because when I do it no two are ever the same.

There are (at least) two things that don't scale linearly for model engines, flywheels and springs. Maybe make that three things if you count RPM which has to go up as scale goes down.

Reynolds Number is literally a measure of scale, so for example flow in intake manifolds might not scale well if you cross a laminar vs turbulent boundary.

I've always wondered, but never worked out the details, about heat flow in water cooled engines, so the question is do scaled down coolant passages have enough surface area to conduct the heat away and at what flow rate of the coolant (and will a scale impeller create the necessary flow rate).


in other words scaling laws for model IC engines is an interesting subject, at least for me, and springs do fall into that.

 

[HMEL]

for my model Merlin engine I measured the weight of the valves plus one third the weight of the rocker arms and computed how much spring force it would take to keep them on the cam during maximum deceleration at the maximum RPM I hoped to run the engine at, then ordered springs with the right size and spring rate, and they worked. for an engine with 48 springs I really didn't want to do this by trial-and-error, I also didn't want to wind my own because when I do it no two are ever the same.

There are (at least) two things that don't scale linearly for model engines, flywheels and springs. Maybe make that three things if you count RPM which has to go up as scale goes down.

Reynolds Number is literally a measure of scale, so for example flow in intake manifolds might not scale well if you cross a laminar vs turbulent boundary.

I've always wondered, but never worked out the details, about heat flow in water cooled engines, so the question is do scaled down coolant passages have enough surface area to conduct the heat away and at what flow rate of the coolant (and will a scale impeller create the necessary flow rate).


in other words scaling laws for model IC engines is an interesting subject, at least for me, and springs do fall into that.

The Reynolds number is non dimensional number (has no units) and is a calculation based on viscosity of the liquid, its Velocity and the pipe diameter. Its practical use is often restricted to flow in pipes and across airfoils. Generally we want to know the pressure drop in the system. And the Reynolds number helps us get that. With respect to heat in the cylinder an estimate has to be made on what that load is. Most will be sent out the exhaust but the remainder has to be picked up by the coolant. We first calculate the the mass of water needed to take the heat away in terms of time. This is a function of the specific heat and the area available. Then we calculate what the pressure drop might be to move the mass in that time frame and then we can calculate the pump characteristics.
Now doing all this is work. But for the most part a model maker will do something and it will work or not work. A scaled impeller may not work as it may not generate the flow or the pressure. But I am of the mind set it your having fun and the model has worked before why go down a rabbit hole unless you have fun doing it.

 

[peterl95124]

The Reynolds number is non dimensional number (has no units) and is a calculation based on viscosity of the liquid, its Velocity and the pipe diameter. Its practical use is often restricted to flow in pipes and across airfoils. Generally we want to know the pressure drop in the system. And the Reynolds number helps us get that. With respect to heat in the cylinder an estimate has to be made on what that load is. Most will be sent out the exhaust but the remainder has to be picked up by the coolant. We first calculate the the mass of water needed to take the heat away in terms of time. This is a function of the specific heat and the area available. Then we calculate what the pressure drop might be to move the mass in that time frame and then we can calculate the pump characteristics.
Now doing all this is work. But for the most part a model maker will do something and it will work or not work. A scaled impeller may not work as it may not generate the flow or the pressure. But I am of the mind set it your having fun and the model has worked before why go down a rabbit hole unless you have fun doing it.

since we don't change the viscosity of the coolant from full size to scale model the Reynolds Number becomes a function of the scale factor.

what fascinates me is not computing coolant requirements from scratch, but rather from scaling laws.

here's an example, say you're wondering about required tensile strength of head bolts. our scale models operate at the same compressions and combustion pressures as full size, so the force on the head is proportional to the area of the head (PI * R ^ 2), but the strength of a bolt is also proportional to its cross sectional area (PI * R ^ 2), so if you use the same alloy bolts you can scale them down to your model size without having to do any calculations from scratch about pressure, force, and tensile strength. You are relying on the fact that both force on the head and the strength of the bolt have the same R squared scaling. Nice, Neat, Tidy.

But there are places where the scaling laws indicate issues, for example the spring rate formula has a wire diameter to the fourth power and spring diameter to the third power as factors, so the wire diameter doesn't scale to the same third power as the mass of the valve and rocker arm do. so when it comes to springs you do have to resort to doing the math.

so the question is not how to calculate coolant requirements from scratch, but whether the scaling laws work in your favor or not.

 

[HMEL]

since we don't change the viscosity of the coolant from full size to scale model the Reynolds Number becomes a function of the scale factor.

what fascinates me is not computing coolant requirements from scratch, but rather from scaling laws.

here's an example, say you're wondering about required tensile strength of head bolts. our scale models operate at the same compressions and combustion pressures as full size, so the force on the head is proportional to the area of the head (PI * R ^ 2), but the strength of a bolt is also proportional to its cross sectional area (PI * R ^ 2), so if you use the same alloy bolts you can scale them down to your model size without having to do any calculations from scratch about pressure, force, and tensile strength. You are relying on the fact that both force on the head and the strength of the bolt have the same R squared scaling. Nice, Neat, Tidy.

But there are places where the scaling laws indicate issues, for example the spring rate formula has a wire diameter to the fourth power and spring diameter to the third power as factors, so the wire diameter doesn't scale to the same third power as the mass of the valve and rocker arm do. so when it comes to springs you do have to resort to doing the math.

so the question is not how to calculate coolant requirements from scratch, but whether the scaling laws work in your favor or not.

Well except for the temperature of the heat source does not scale it remains at essentially the same combustion temperature. And the delta T driving force for the heat transfer will be the same only with smaller material thickness which should yield higher rates of heat transfer. That is the model is smaller but the combustion temps remains the same so I would not expect the cooling flows to be linear scaling factors. Now there are ways to lower flame temperatures and I have used a few of the techniques to reduce nitrogen oxides but they are not practical for models. So one answer to the question has anyone built a model that overheated or are the cooling systems just oversized in making a model.

 

[Charles Lamont]

Because the power output and fuel consumption of a model scale as the cube, and the cooling surface area scales as the square, the heat transfer to the coolant should be easy enough. The thickness of the cylinder/head wall does not matter too much, as most of the thermal resistance is in the coolant boundary layer at the wall of the water jacket.

I have been doing some experiments with cooling and small centrifugal pumps (1/2" impeller diameter) but I am not yet ready to write up the results.

 

[Steamchick]

Back to cam-shaft design, one of the things most modellers do not cover is structural stiffness, never mind total strength. (I am not an educated expert, but a few of you are, so please correct my gaffs herein!).
I worked with a Doctor of Maths, who had redesigned cam and follower profiles on a large truck engine maker's engines, as the cycle of motion of the cams through followers, push-rods, rockers, valves, and the stiffness of the block, head, rocker mounts, etc. all made the whole set-up a spring that stored energy as the cam lifted the valve (In addition to the energy stored in the valve spring) that was released as the cam released the valve. This "additional" stored energy was re-configured (timing of addition and subtraction of the energy from the cam) following his study of the stiffness of everything so the energy usefully operated the return of the valve, instead of hitting a resonance that was destroying engines at a high engine speed, just short of truck cruising speed.
When he applied the same modelling strategy to my design of a high speed electrical circuit breaker we trebled the lifetime without servicing the equipment, which was a huge cost saving on equipment that is in service for 40 or 50 years.
When relating this to models, often the models are hugely stiffer than the full sized engines, and subject to other constraints become way over designed. But sometimes they become too weak to function, or not stiff enough to be machined, or function... So maybe that is why models are never truly to scale? And I have no idea how to re-configure cams so the valves don't bounce, or whatever...
But maybe models work well without such complexity because they are relatively stiffer in structure than the truck engines?
K2

 

[peterl95124]

the only data point I have with water cooling not scaling well is that to date every model Merlin engine that I know of is under cooled and overheats, but this probably has nothing to do with scaling laws and everything to do with not being able to replicate the coolant paths cast into the full size head.

 

[peterl95124]

Because the power output and fuel consumption of a model ange scale as the cube, and the cooling surface area scales as the square, the heat transfer to the coolant should be easy enough. The thickness of the cylinder/head wall does not matter too much, as most of the thermal resistance is in the coolant boundary layer at the wall of the water jacket.

I have been doing some experiments with cooling and small centrifugal pumps (1/2" impeller diameter) but I am not yet ready to write up the results.

Charles, I like your logic. here's one humorous data point about pumps, I have a 1.25" impeller in one of my engines, which I had doubts about, but when running the engine at a show last summer it blew the hose off the pump and sprayed the coolant over 6 feet away, I was pretty surprised !!!

 

[Zeb]

All the 1:1 scale Merlins I know of have cooling issues on the ground, employ big oil and water coolers, and spin a giant fan.

I would thinkpf that after you had your baseline cams profiled, you would need to run your engine to tune valve overlap and catch any harmonic transients. I would expect someone to make at least two or three sets of profiles before settling.

 

[HMEL]

since we don't change the viscosity of the coolant from full size to scale model the Reynolds Number becomes a function of the scale factor.

what fascinates me is not computing coolant requirements from scratch, but rather from scaling laws.

here's an example, say you're wondering about required tensile strength of head bolts. our scale models operate at the same compressions and combustion pressures as full size, so the force on the head is proportional to the area of the head (PI * R ^ 2), but the strength of a bolt is also proportional to its cross sectional area (PI * R ^ 2), so if you use the same alloy bolts you can scale them down to your model size without having to do any calculations from scratch about pressure, force, and tensile strength. You are relying on the fact that both force on the head and the strength of the bolt have the same R squared scaling. Nice, Neat, Tidy.

But there are places where the scaling laws indicate issues, for example the spring rate formula has a wire diameter to the fourth power and spring diameter to the third power as factors, so the wire diameter doesn't scale to the same third power as the mass of the valve and rocker arm do. so when it comes to springs you do have to resort to doing the math.

so the question is not how to calculate coolant requirements from scratch, but whether the scaling laws work in your favor or not.

There is a mathematical way of chasing down scaling factors and one is the Buckinham Pi theorem. But there is no way the Reynolds number does not change so the skeptic in me says its not going to work using a scaling factor. The reynolds number is used to find the friction factor which depends on the surface roughness. I have not seen any math concerning the other scaling factors so I will take your word on those. However it would interest me to know whether these come from trial and experience or based on inference from the design formulas.

However I think a close examination of smaller pumps is very worth while and I would expect them to follow the same laws as larger pumps only harder to keep the tolerances.