Using Blair to spec a Cam

[SchmidtMotorWorks]

Mark,
Most of the time you get what you pay.

What I am hoping to do is put a free entry level set of equations here for anyone to use. (The more I analyze this I am not sure that can be done). This I believe would cover many of the readers (lurkers) while the others as needed would buy the software packages you listed.

Stan

Is the spreadsheet the limitation?

[GARY C]

What I am hoping to do is put a free entry level set of equations here for anyone to use. (The more I analyze this I am not sure that can be done). This I believe would cover many of the readers (lurkers) while the others as needed would buy the software packages you listed.

Stan

Considering that dyno prove equations get shot down here I am supersized that attaching the name Blair to a hypothetical equation is considered valid, even if you came up with a formula wouldn't it require testing to prove it valid?

For what it's worth here is my 5 minute garytheory cam 317 seat timing on a 111 LSA, I did not take the time to hunt down lobes to get more specific but I would probably go 10 to 12 degrees more on the exhaust, If this was a real world engine I would research more on what was needed and maybe choose something different and then I would test ICL and valve lash on the dyno to decide where to go from there.

[SchmidtMotorWorks]

What I am hoping to do is put a free entry level set of equations here for anyone to use. (The more I analyze this I am not sure that can be done). This I believe would cover many of the readers (lurkers) while the others as needed would buy the software packages you listed.

Stan

Considering that dyno prove equations get shot down here I am supersized that attaching the name Blair to a hypothetical equation is considered valid, even if you came up with a formula wouldn't it require testing to prove it valid?

For what it's worth here is my 5 minute garytheory cam 317 seat timing on a 111 LSA, I did not take the time to hunt down lobes to get more specific but I would probably go 10 to 12 degrees more on the exhaust, If this was a real world engine I would research more on what was needed and maybe choose something different and then I would test ICL and valve lash on the dyno to decide where to go from there.

Gary, the Blair work IS based on dyno results and fit to the data.
It is an example of; if you are going to do this by fitting dyno data, this is a comprehensive way to do it.
That said, it is a framework that should be tuned for any particular type of engine.
What makes is better than a simple formula is that it takes more inputs and does more computation that are essential to making any system comprehensive.
Still, it is far short of what a 1D system does.

[GARY C]

What I am hoping to do is put a free entry level set of equations here for anyone to use. (The more I analyze this I am not sure that can be done). This I believe would cover many of the readers (lurkers) while the others as needed would buy the software packages you listed.

Stan

Considering that dyno prove equations get shot down here I am supersized that attaching the name Blair to a hypothetical equation is considered valid, even if you came up with a formula wouldn't it require testing to prove it valid?

For what it's worth here is my 5 minute garytheory cam 317 seat timing on a 111 LSA, I did not take the time to hunt down lobes to get more specific but I would probably go 10 to 12 degrees more on the exhaust, If this was a real world engine I would research more on what was needed and maybe choose something different and then I would test ICL and valve lash on the dyno to decide where to go from there.

Gary, the Blair work IS based on dyno results and fit to the data.
It is an example of; if you are going to do this by fitting dyno data, this is a comprehensive way to do it.
That said, it is a framework that should be tuned for any particular type of engine.
What makes is better than a simple formula is that it takes more inputs and does more computation that are essential to making any system comprehensive.
Still, it is far short of what a 1D system does.

It will be interesting to see what cam numbers you present and the math you use to get there.

[SchmidtMotorWorks]

It will be interesting to see what cam numbers you present and the math you use to get there.

It is a framework that you tailor to dyno data, if it is different than the engines you develop, you adjust the numbers, that is how it is intended to be used.

[RevTheory]

So it doesn't scale?

[GARY C]

It will be interesting to see what cam numbers you present and the math you use to get there.

It is a framework that you tailor to dyno data, if it is different than the engines you develop, you adjust the numbers, that is how it is intended to be used.

Thats great, you preach Blair and write programs with the best state of the art programs so we are looking forward to your cam specs this engine and the math you use do derive said cam specs. It should be easy for you.

[hoffman900]

So it doesn't scale?

Why don't you be more like your buddy Gary who I think is a good guy (he's reached out to me and we've chatted on the phone)? Blair's method is to develop a way for it to scale from a weedwhacker to a Rolls Royce Merlin V12. As such, it's a bit more complicated than using a simple rule of thumb. It however wasn't pulled out of thin air.

It will be interesting to see what cam numbers you present and the math you use to get there.

It is a framework that you tailor to dyno data, if it is different than the engines you develop, you adjust the numbers, that is how it is intended to be used.

Thats great, you preach Blair and write programs with the best state of the art programs so we are looking forward to your cam specs this engine and the math you use do derive said cam specs. It should be easy for you.

The difficult part is taking the time area and producing a meaningful valve lift curve that is realistic within the constraints of the mechanical limitations of a particular valvetrain.

[pcnsd]

2. Setting the target values, the article provides a set of values that were fit to a small set of engines and those engines had a wide range; from garden tools to racing engines.
When the values for each of the engines are shown as points in a chart they do not form a straight line or even simple curve.
So the data was fit to the best approximation.

For someone that wants to use this for a particular engine type, they would begin by testing it and then comparing it to known engines that perform well. Then you would adjust the values in the formula to fit the result that you will want to duplicate in the future.

You guys have mentioned this a couple times......I would think that is a deal breaker as it shows the work was incorrect and only produces a rough approximation that is 80-90% right at best.

If you want to have a system that does consider all of the intake, exhaust dimensions and more, then you need to step-up to a 1D software like EngMod4T authored by Vannik, who is posting in this thread.

http://vannik.co.za/

This really seems the the best input in the thread so far. Time has moved on, the days when simple equations that get you 80-90% there where really helpful are gone. A program like EngMod4T or Dymomation (that I use because its a bit simpler but less control) just makes "testing" your ideas so quick and painless, buying a copy was some of the best money I've ever spent. PipeMax is even cheaper and quicker...this it probably the best collection of equation solutions I've seen, again less flexible than the 1D sims but 5 minutes and you're in the right ballpark.

Sorry for the diversion, back to Bliar.

Mark,
Most of the time you get what you pay.

What I am hoping to do is put a free entry level set of equations here for anyone to use. (The more I analyze this I am not sure that can be done). This I believe would cover many of the readers (lurkers) while the others as needed would buy the software packages you listed.

Stan

Stan,
I am thrilled at the prospect of a spreadsheet with reasonably sorted equations from a known and respected source. Having the outputs goes a long way to understanding the inputs and their relationship to each other in a system. Looking forward to your results and wish you better luck than myself on your quest. If I could make a request, it would be to indicate those unfixed constants by some method.

[hoffman900]

As a slight aside, this is an old article (2008), but shows what can be done: https://www.researchgate.net/publicatio ... SAE_Engine

The students took a known engine and inputted and validated their model in a 1D program. They then set the program to adjust a multitude of dimensions (intake and cam lift, intake and cam duration, intake and exhaust centerline angles, exhaust pipe length, intake runner length, and restrictor diffuser length - all from different minimums to maximums at a defined increment) for a given rpm range (maximize torque from 5000-12000rpm). This resulted in 1230 engine combinations, which the computer ran for 4.5 days. The program then converged on a best combination. Then they performed a sensitivity analyses to see what parameter had the greatest influence, and rank them. The program recommended a valve lift profile, which was then brought into 4stHEAD and tweaked so it could mechanically work. This required some compromise. They had the cams ground and tested the engine.

I could do the same with Neels program, but would have to do so manually. IE: develop a matrix to keep track of the combinations, then analyze the data. This isn't hard to do, just takes work. However, Neels' program also doesn't cost as much as a house (depending what you drive, his costs just a couple gas fill ups). I pretty much do this anyway, except my thousands of engine combinations take a little longer, but having one cylinder helps.

When you hear about GM running millions of hours of simulation work on piston tops (for DI), this is the kind of thing they're doing. A computer can run all the different combinations and pick out trends way better than a human could, and do it in a fraction of the time and price. Once the computer converges on an ideal engine, you build it, establish a baseline, and work from there.

If I had a camshaft business, I'd solicit all the feedback and dyno results I could. I could create a database and through some statistics and algorithms, I could really start honing in on some trends. You could probably hire a data scientist intern or a co-op for a summer to do this, and it would be quite easy for them as they're trained to work with datasets with MILLIONS of data rows. Bring on the machine learning!

[BigBro74]

If I had a camshaft business, I'd solicit all the feedback and dyno results I could. I could create a database and through some statistics and algorithms, I could really start honing in on some trends. You could probably hire a data scientist intern or a co-op for a summer to do this, and it would be quite easy for them as they're trained to work with datasets with MILLIONS of data rows. Bring on the machine learning!

Sounds exactly what happened in some engineer/authors' life that (will hopefully still) post on speed talk.........................
Just sayin

[SchmidtMotorWorks]

This may be helpful for those that are not familiar with machine learning.

The 7 Steps of Machine Learning
https://www.youtube.com/watch?v=nKW8Ndu7Mjw

[hoffman900]

If I had a camshaft business, I'd solicit all the feedback and dyno results I could. I could create a database and through some statistics and algorithms, I could really start honing in on some trends. You could probably hire a data scientist intern or a co-op for a summer to do this, and it would be quite easy for them as they're trained to work with datasets with MILLIONS of data rows. Bring on the machine learning!

Sounds exactly what happened in some engineer/authors' life that (will hopefully still) post on speed talk.........................
Just sayin

Er, kind of, but let's stay on topic.

I suggest everyone to watch that video Jon just posted. This stuff is going on everywhere. All my friends with masters (and PhDs) degrees in physics, stats, math, engineering, and science are all working on machine learning algorithms, either for employment or to make their lives easier. An engineer friend of mine (we're all 30-mid 30s) just hired a "kid" out of school and his mission statement was "I want to automate every repetitive task in my job".

Anyway, back to the topic...

[GARY C]

So it doesn't scale?

Why don't you be more like your buddy Gary who I think is a good guy (he's reached out to me and we've chatted on the phone)? Blair's method is to develop a way for it to scale from a weedwhacker to a Rolls Royce Merlin V12. As such, it's a bit more complicated than using a simple rule of thumb. It however wasn't pulled out of thin air.

It is a framework that you tailor to dyno data, if it is different than the engines you develop, you adjust the numbers, that is how it is intended to be used.

Thats great, you preach Blair and write programs with the best state of the art programs so we are looking forward to your cam specs this engine and the math you use do derive said cam specs. It should be easy for you.

The difficult part is taking the time area and producing a meaningful valve lift curve that is realistic within the constraints of the mechanical limitations of a particular valvetrain.

That seems to be the issue I see with both, rule of thumb or models, you have to have some prior knowledge as to what your doing and then you have to have real world data to decide where you are going after that.

[hoffman900]

So it doesn't scale?

Why don't you be more like your buddy Gary who I think is a good guy (he's reached out to me and we've chatted on the phone)? Blair's method is to develop a way for it to scale from a weedwhacker to a Rolls Royce Merlin V12. As such, it's a bit more complicated than using a simple rule of thumb. It however wasn't pulled out of thin air.

Thats great, you preach Blair and write programs with the best state of the art programs so we are looking forward to your cam specs this engine and the math you use do derive said cam specs. It should be easy for you.

The difficult part is taking the time area and producing a meaningful valve lift curve that is realistic within the constraints of the mechanical limitations of a particular valvetrain.

That seems to be the issue I see with both, rule of thumb or models, you have to have some prior knowledge as to what your doing and then you have to have real world data to decide where you are going after that.

The beautiful part about having a pushrod engine with common tappet diameters (.842-.904") is that there are tens of thousands of camshaft lobes out there developed. It's a matter of creating the performance box you need and finding what fits. Likely, it's already been design and multiple iterations with varying amounts of aggressiveness exist.

The rest of the engine world isn't so lucky. The bucket lifter OHC builders are in better shape than the OHC rocker or finger follower builders, but nothing like what the pushrod engine builders have. I have some simulation tests going on. I'll post results later this evening.