351W stroker, Dyno 2000 vs. Dyno test

I have received the dyno data from Keith Craft concerning the Dart 351W stroker motor (details at website listed below). I was hoping for better agreement between Dyno 2000 and the test data. Here is what I have:
power.png

torque.png


If you have any questions about either let me know.
 
Oh yeah, I'll be disappointed too if my engine dynos out at 50 hp above what Dyno2000 predicts
tongue.gif


Did you use one of the built in options for the heads, or did you have a flow file specific to the actual heads? And what about the cam? In my case, the specific cam that's in my engine isn't in their database so I had to choose similar ones based on lift & duration. But the similar ones that look to have identical specs produce different results...
 
That's very interesting...congratulations on making 615 naturally aspirated, carbeurated horsepower.

The measured HP curve is shifted to the left (by about 500 rpm). It would be interesting to compare measured cam profiles against the profiles used in the Dyno 2000 simulation.
 

Adam C.

GT40s Sponsor
Sorry guys, Dyno2000 is NEVER going to give you good results. It is simply a curve fit progam that uses a bunch of dyno data and interpolates between the points.

Only a true simulation code will predict engine performance. These are one-dimensional, nonlinear, time domain codes that take upwards of 5 minutes to do a 3000-7000 sweep for a V-8. Any code that gives you the curve instantly is doing nothing.

True simulation codes are
Wave http://www.ricardo.com/
GT power http://www.gtisoft.com/index.html
Virtual Four Stroke http://www.optimum-power.com/

These codes all run on the exact same numerics originally developed by Ford. the history is that one of the guys that helped develop the Ford code left and formed Wave. Some guys left wave and made GT Power. I'm not sure where virtual came from. All of them are of course extremely expensive.

Gary, I have access to more than one of these codes. I could run your engine if you want just for jun. I need REALLY specific data however.

Adam

[ January 14, 2003: Message edited by: Frankenstang ]
 
Frankenstang, you sound like a salesman for those other products
smile.gif


Dyno2000 is best used for comparing different possible engine builds, or getting a feel for what effect a change might make (e.g., what if I raise the compression by a point?). That said, what do you think constitutes "good results?" It looks to be within about 10% of the actual engine to me, and as we've already discussed, part of the differences may be due to not having the exact parameter settings to match the actual engine. In any case, +/- 10% seems pretty good to me.

As far as how Dyno2000 works, there is a bit of discussion of that on their web page. And according to that, it is a true simulation. Now, all simulations make tradeoffs in the interest of speed or whatever vs. accuracy. So many folks may consider it a toy, but it's still kind of fun to compare the output with reality. btw, Dyno2000 says my engine (a 289) will make 370 peak HP, and I don't believe it, but then I don't *really* know what my compression ratio is, and like I said before, my specific cam is not in their list. I'll be happy if the engine makes 325HP...

Quote from Dyno200 web page:
<BLOCKQUOTE><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><HR> Filling-and-emptying models repeat this entire flow analysis for each rpm point at which horsepower and torque are to be predicted. As a result, computer programs that use this technique, like the Dyno2000 from Motion Software, commonly "build up" engine power curves by calculating and then drawing them one power point at a time. The math needed to predict each power value involves several million calculations. Despite these oppressive computational requirements, a modern PC equipped with a math co-processor can develop an entire power curve, consisting of 10 to 15 data points, in less than a second. <HR></BLOCKQUOTE>
 
<BLOCKQUOTE><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><HR>Originally posted by Frankenstang:
Gary, I have access to more than one of these codes. I could run your engine if you want just for jun. I need REALLY specific data however.

Adam

[ January 14, 2003: Message edited by: Frankenstang ]
<HR></BLOCKQUOTE>

I would be interested in seeing the results of a more accurate simulation. I have some fairly specfic data (head flow, basic cam profile specs, rod length, etc.) Other stuff may be lacking. Send me an e-mail so that I will have your address and I will send you all the data that I have, etc.

Thanks
 

Adam C.

GT40s Sponsor
Steve,
Sorry for sounding argumentative. I am not a salesman for the other products. In truth, I am part of a research group that has spent much of the last decade refining the model that Ford uses. This causes me to be a little jaded when it comes to these junk packages. I have also written my own code that uses the the same finite differencing scheme (although it does not have a valve model yet, it is the most difficult).

As a consequence I am intimately familiar with what it takes to actually simulate an engine. We are somewhere in this paragraph on the page you referenced:

"Virtually all of these initial efforts were developed by corporate engineers working in cooperation with university research staff. The software was written for mainframe computers at a cost of hundreds of thousands to millions of dollars, and it featured the addition of accurate and comprehensive thermodynamic, combustion, kinetic, and frictional models to the Mesh-Method of analysis"

This Mesh-Method is the 1-D finite differencing scheme that we use. It takes quite a bit of time to process, and requires more inputs than the average bench racer can provied. There is another approach called method of characteristics, which works just as well, but is a little more abstract and tricky.

I'm sorry, but just because 2000 makes a bunch of calculations does not mean that it simulates anything.

As far as what is good results. I can typically predict peak engine vol. eff. within 100 rpm and easily within 5% on a first hack, and get closer with some refinement.

The real problem is when you stray away from a configuration similar to your average small block chevy 350. I've used Dyno2000 in the past, and could make it give some wild predictions pretty easy.

Still, as you say, most people are interested in the difference between one cam and another, or a certain manifold and another. Almost universaly, most codes are better at giving these deltas than absolute power figures. This is true for our codes as well.

I guess the big thing I wanted to get across is to be careful in interpreting the results from those programs, don't build an engine based soley on those graphs. They are the Avenger kit cars of simulation codes. (Meat don't get started)

I'll get with Gary and we'll do the pepsi challenge with this stuff.

Adam
 
Frankenstang,

I think this is a good discussion - no offense taken and I hope I'm not coming off as a jerk or a salesman for Dyno2000...

I did some modeling of computer systems in my youth. What we did was simplify everything down to a set of queues and processing delays. Thus, to model access to a disk, you might just say that disk access time is distributed about a certain delay. Or you could spend more time and create a more detailed and accurate model of the disk (file system). Depending on what you're doing the simulation for, the simple model may give you good enough results. Either way, I think you can say that the disk access is being simulated, though the model may not take into account the exact behavior of the system. But it's true, you always have to keep in mind the assumptions and limitations of the model in use.

Most of the market for Dyno2000 IS guys building standard small block Chevy 350s, or something similar. I think it serves it purpose for that market fairly well. I can understand that FoMoCo would need a better, more accurate simulation and would be willing to pay the price required to get it.

<BLOCKQUOTE><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><HR> I guess the big thing I wanted to get across is to be careful in interpreting the results from those programs, don't build an engine based soley on those graphs. <HR></BLOCKQUOTE>
I agree.
 
My engine (a 289 with Weber 44IDFs for those who haven't been paying attention
smile.gif
) went on the dyno at Kroyer Engines yesterday. Actual results vs. Dyno2000 predictions are shown below:

dyno.JPG


Dyno2000 is dead-on at 3000 RPM but then something goes terribly wrong... I'm not sure if the dyno numbers are "corrected" for altitude or not - my engine is set up for high altitude operation (4000 ft). That might account for some of the difference...

Can anyone explain the dip in torque between 3000 and 3500 RPM? Could it be a Weber thing? Transition from idle circuit to main jets?

[ February 28, 2003: Message edited by: Steve Toner ]
 
Gary, your set up is pretty well dialed in. My 351/408, which I have toyed with for 2 tears, was 620 hp, 565 lft lbs. In a '66 Mustang chassis the only problem is there is no way to keep the rears from going up in smoke when rear wheel hp gets over 425... My throttle response got great when I went with a 750, although I gave back 30 peak hp. Part throttle response and torque were much better with the 750, after I dialed it in correctly. IMHO the weber thing may be giving you some reversion to see your torque fall off. You can usually see the fuel droplets coming back up through the carbs if it is a severe case.
 
The heads are Edelbrock Performer RPMs with the large (2.02") intake valves. Compression is upposed to be 9:1, but I don't know what it really is - I didn't build the engine, and the guys who did build it didn't give me any kind os build sheet. If they used flat-top pistons and just bolted on the heads without milling them, that would result in somewhat less than 9:1 on a 289 with these heads...
 

Ron Earp

Admin
Obviously the dyno doesn't lie, but you might not be able to expect much more if the compression is very low and the cam isn't all that hot.

What cam are you running?

What is your initial ignition timing, advance rate, and total advance?

Did you collect A/F data during the run, what does it show?

If you are pig rich you are handing off a lot of power for sure. Also, if you aren't getting enough total spark advance you'll be shorting yourself too, which would really show up once all the advance was in.

Ron
 
Gary,

I used the file Ford_SB_Edelbrock_6025.flw, which I found somewhere. For exhaust I used Small Tube w/mufflers. There is a column header that says STPCor Factor - if I multiply this by the results, then peak hp is 332, which is pretty much what I expected (I said before in another thread that I didn't believe the 374hp number that I was getting out of Dyno2000). But if they've already applied it, then I'm a little disappointed in the results (although it's still plenty of hp as far as I'm concerned)...

Ron,

The cam is a Federal Mogul CS1231R, which is not listed in the Dyno2000 cam database. So I used one one that appeared to have the same specs (214/224 duration, .472/.496 lift). It's a hydraulic, non-roller cam. As far as spark advance, I think it's all in by 3000 RPM, where they started measuring. 36 degrees total advance. They didn't collect A/F data so I can't tell if it's leaning out above 5000rpm or what...
 
Steve,
Most dyno data is corrected to standard conditions which should correspond to Dyno 2000's goals. In Dyno 2000 what did you use for head flow numbers? Where did you get them? What about exhaust specs.
 

MWGT40

Supporter
Anybody else got any experience of the Dart Smallblocks and the power outputs that are available from these engines with decent flowing heads (such as AFR or Edelbrock Victor Jrs)?

Martin /ubbthreads/images/graemlins/ooo.gif
 
Back
Top