This thread originated from one started by Gordo so you can find the background there. However, I've copied and edited two of my posts from there to start the present thread.
I looked into engine simulation software in some detail to try to find the "best" one for simulating Gold Star engine performance. The good news is there are a lot of choices (including Engine Analyzer, Desktop Dyno (also versions called DynoSim and DynoMation), Engine Pro, GT-Power and Virtual Engine Dyno, and probably others). The bad news, is, there are a lot of choices.
It seems there are three price points, ~$100, ~$200 and ~$500 for three levels of capabilities. Unfortunately, comparing those capabilities is problematic. The manuals available for download for two of these are 323 and 314 pages, which would be a lot of material to have to read in detail to compare just those two programs. It's quite possible one program could have more capabilities than another but be missing some useful feature that the other has. Further, since there are no downloadable manuals for the other programs, all there is for them are one or two-page descriptions on the web.
OK, given the impracticality of determining the "best" simulation program, I returned to my initial thought of 'Engine Analyzer Plus' by Performance Trends. I've used two generations of their 'Port Flow Analyzer' with my flow bench and have been happy with the software and with their responsiveness in providing me with new software keys when I migrated to new computers every few years. Also, the company has been around since 1986 and their engine software has gone through a number of upgrades since then. Given this, I went through two rounds of questions with their support line and they satisfactorily answered my questions about the capabilities of 'Engine Analyzer Plus' for present purposes:
-------------------- For any accurate simulation, accurate head flow data is critical. However, there are some generic head and intake and exhaust examples for you to pick from, inside all our Engine Analyzer programs
Predicting changes in burn rate (which determines spark advance) is not something the Plus version deals with in great detail. Still, it WILL give you an estimate of a safe spark curve for different combos of engine parts with different octane fuels.
It is used mostly to predict the different torque and HP curves for these modifications. -----------------------
Translating the above, the first paragraph answers my question about whether the program has generic head options. It does, so this will let me use BSA's h.p. and torque vs. rpm curves to pick a generic head that reproduces the data, rather than having to input actual flow data on each head.
The second paragraph answers my question about fuels. Basically, it says that while it might estimate, say, 36-deg. for the spark advance for the stock configuration of my Competition (for which BSA says 39-deg.), it might then give a different estimate if I input a different CR. If that new figure were, say, 34-deg., that would indicate less advance would be needed with the new CR. While I couldn't necessarily rely on either value as being accurate, the relative difference would be useful information.
The third paragraph is the reason for doing this in the first place. Once the parameters fit the BSA curves for a given stock configuration, the effect of changing CR, cams, etc. on the engine performance will be determined. Instead of "I think the xx-xxxx inlet cam feels like it has more mid-range power," it will be clear whether or not it actually does.
I think I've caught and corrected all of my mistakes and now have reasonable input parameters for everything needed by 'Engine Analyzer Plus' (although at this point I'm continuing to work on those parameters). For example, a few weeks ago I measured the flow of my 1½" GP to be 85.1 CFM @ 3" H20. However, the simulation program uses what it says is the industry standard test condition, which is 1.5" Hg. The Superflow manual gives 2.58x as the factor to use to convert, giving 219.6 CFM @ 1.5" Hg, so that's what I entered into the program.
Another example is p. 22 of 'Goldie' shows cross sections of the ports of 350 cc heads. Scaling the 1956 drawing to the 1-5/32" ID of the port, the length is ~4.6". Although that length is for a 350 cc DB head, it should be essentially the same for a DBD. In any case, that's what I used.
As long as I didn't make any mistake with converting the cam specs to the way the program needs them as inputs this simulation program still leaves the exhaust system as an uncertainty. BSA gives data for 5000-7000 rpm for the Clubman DBD so nothing below 5000 rpm is available to "calibrate" the simulation program. Despite that, I started the simulation at 3200 rpm in the attached graph to show the effect of the exhaust system. Keeping the total length about the same but changing the details can easily move the dip centered at ~3000 rpm in this graph up to ~5000 rpm. Although the simulation has an input for straight pipes, so Scramblers and Catalinas can be modeled with more certainty, silencers or megaphones of any type, including 'twitter' silencers, will be an issue since the dyno data doesn't cover the lower rpm region where the silencers have an effect.
With that background, the fact this initial simulation underestimated the h.p. of this DBD by ~7% is irrelevant. The important thing about this graph is how well it parallels the actual dyno data. First, the program's manufacturer only claims +/-7% agreement with actual measured dyno data. And second, I haven't necessarily found the final "best" set of input parameters so the graphs in this post should be regarded as "preliminary tests" at this point.
As an observation, it's been quite informative to work with the program. To avoid clutter I ran the "final" simulation twice so that the attached graph only would have one set of h.p. and torque curves. When working with the program, though, you can change a parameter, run the simulation, and the graph shows the latest result along with the previous one. This makes it very easy to learn what changes have what effects, and where those effects are in rpm.
As an example of what this simulation program makes possible, the second graph shows the calculated consequences of switching from a 1½" GP to a 1036 Concentric with a K&N filter on it. All other parameters of the simulation were kept constant and only the carburetor CFM was changed from 220 to 179 to correspond to my recent flow bench measurements.
Although the Concentric with filter flows 18% less air than the straight-through GP the simulation says using it only would knock ~2% off the maximum h.p. Personally, I was willing to give up even more h.p. than this not to have to deal with a GP, but I was pleasantly surprised to find there will be only a relatively trivial loss.
After taking time to check for the Nth time that I've correctly converted the data for the 65-2442 and 65-2446 cams I'll start converting the other relevant ones. I'll also continue checking to be sure all the other input parameters I've used are reasonable since, once I'm confident in them, knowing the parameters that work for the DBD Clubman should make it faster to set up simulations for other Gold Star models back through the ZB.
would it be possible to investigate what different ignition timing would do?.
Unfortunately, as the support staff warned me ahead of time, the 'Plus' version is limited in what it has to say about timing and knock. For that the $500 'Pro' version is needed. ---------------------------- 'Plus' Program sets spark to produce best power. Program also predicts when knock occurs and will reduce spark advance (which reduces power) to keep the engine from knocking. Very easy to use.
'Pro' You have 3 options: 1) To let program set spark advance which produces best power. 2) To let program predict Burn Rate (required spark advance) and modify it by a user defined spark curve with 4 break points. 3) To let you specify the spark curve with 6 break points, and tell the program this curve produces best power. This effectively lets you specify the engine's burn rate. Program calculates Knock Index, which gives the likelihood that detonation will occur. You can run with very low or high detonation. The programs time consuming calculations and method of calculation does not lend itself to letting the program automatically limit spark to a predetermined max allowable Knock Index. --------------------------------------
Originally Posted by Gordo in Comox
awaiting the data that will give me the best possible engines for today's conditions.
Be patient a little longer. I need to be 100% confident of the parameters in the DBD simulation before I move on to the other models (starting with your ZB) since the DBD parameters will help me determine the others.
Over the next few days I plan to list all the relevant input data for 'Engine Analyzer Plus' that I'll use, along with the assumptions I make (e.g. 'hemi' for head type, vs. 'pent roof', 'wedge', or any of the eight other choices), so any mistakes I've made can be caught and any missing information filled in. Starting with the actual data for the DBD that are required inputs to the program:
DBD Clubman and Competition bore: 3.346" stroke: 3.465" rod length: 6.469" inlet port dia.: 1.5" inlet port length: 5.21" [*] inlet valve: 1.85" exhaust valve: 1.53" exhaust header length: 31" twitter length: 26"
DBD Catalina --same as Clubman, except: inlet port dia.: 1.188" (¿need to check this) exhaust pipe length: 49½"
[*] For the inlet ports I used the lengths shown by arrows in the first attached figure, extracted from p.22 of 'Goldie', that I calculated assuming the port diameters were the sizes of the carburetors supplied those years. Note that these are for 350 cc heads but, unless someone can supply more accurate data, I will be using these as "best estimates" for the various 500s. However, even if the inlet port lengths extracted from the figure were correct to the nearest 0.001" there is still wiggle room in the calculation. At full throttle the air resonating in the inlet tract can't tell any difference between the ~5" length of 1.5"-dia tract (for a DBD; smaller for earlier models) that's physically part of the head and the 6" length of the 1.5"-dia. GP that's bolted to it. At partial throttle the resonance is more complex but, like at full throttle, the tract of the carburetor alters the length of the "inlet tract." The point is, the quality of these simulations depends on as much on the quality of the assumptions that have to be made as it does on measured dimensions.
Once all the parameters are determined for each series ZB through DBD, the variables to test will be cams and compression ratios. It will take some time to convert all the cam data to the format used by the program. As the second figure shows,[**] by my count by the time we're done it will have required data on 5 inlet cams and 3 exhaust cams to cover the possibilities. But, that should take care of it even if Gordo builds his ZB to race on sand using alcohol (in his bike, not in him).
[**] The figure is a composite of tables in a May 1952 Instruction Book at the top and an August 1955 Maintenance manual at the bottom. Other than the lack of a Touring model the DBD cams were the same as the DB so this figure covers all models from ZB through DBD.
Visually, there doesn't appear to be a lot of difference between the Clubman and Touring cams, but looks can be deceiving. The area under the Clubman curve is 33% greater which potentially gives the engine enough extra fuel to make the difference between 30 h.p. and 40 hp. Similarly, the overlaps might not seem very different, but it is 72% greater for the Clubman cams so not all the fuel sticks around the combustion chamber to help with the h.p.
MM; In the BB engines the BB34 GS head (65-1501) is not as tall as my BB32 GS head and has a different inlet angle. It appears that the 32 was a complete development stage ahead. It was as if that version of a 34 head was just a carryover from the one piece head but with the change to the separate rockerbox.
To make sense of this I looked at your "Part numbers for heads" thread and see that you have for the BB engines Clubmans 65-1802 and 65-1805, Road Racing 65-1604 and 65-1501.
My 500 heads are 65-1501 and the 350 head is most likely a 65- 1802.(hard to make out numbers). That would mean that I am comparing Road Racing 500 heads with a Clubmans 350 head and find that the Clubmans head is more advanced, more like the later big fin heads. Have you sorted out the years/models that had both Clubmans and Road Racing versions for your simulations?
Does anyone out there have the BB Clubmans heads (65-1805) for comparison?
PS interesting cam graph
The roadside repairs make for the best post ride stories.
It appears that the 32 was a complete development stage ahead.
The author of 'Goldie', nominally A. Golland but apparently BSA employee Arthur Lupkin, makes it clear that until the mid-'50s the development effort went into the 350s with the 500s benefiting later by trickle down (trickle up?)
Originally Posted by Gordo in Comox
Have you sorted out the years/models that had both Clubmans and Road Racing versions for your simulations?
When I offered to do this I was oblivious to the fact I would be entering uncharted territory, at least as far as Gold Star engines are concerned. Engine simulation software has been around for several decades but apparently(?) no one has applied it to Gold Stars. Which is surprising given that the large number of factory cams and pistons, not to mention gearboxes and sprockets, makes it a customizer's dream. Anyway, although uncharted territory means the going is slow, the fact no one has been here before means the rewards are great.
After a bit of frustration I finally realized (...I think...) that the opening/closing data for BSA cams aren't for the cams themselves, which is the common way of specifying things in the automotive world, but rather are for the valves. Since the lifters are 1"-dia. a valve will start opening sooner in angle than if the lifter were the diameter of the tip of a dial indicator as used to generate cam data, and will close later. So, my current top priority is to overcome the camshaft obstacle, requiring "translation" from the BSA "valve spec." data to industry-standard "cam spec." data.[*]
Originally Posted by Gordo in Comox
PS interesting cam graph
Ignore that graph since I'm now pretty sure it's a "valve lift graph," not a "cam graph."
[*] So far I've not found an on-line calculator for this so am presently dead in the water. Since all the cams have the same base circle, if someone has "real" cam data on any of the BSA camshafts that would be enough to know what constant to subtract from the opening and closing of all the valve specs to convert them to cam specs. I'll see if I have a loose BSA cam in the garage, in which case I'll be able to measure the offset. However, I'd much rather not spend the time doing that so I hope someone comes up with actual data in the next day or two.
Last edited by Magnetoman; 05/20/199:56 pm. Reason: added [*]
Increased cam follower radius is similar to a steeper face on the cam . How you factor that it in I dunno,change the cam max open time parameter , if it cant handle ignition, probably not. I have a graph ( chart) showing the difference somewhere , has most effect at mid lift. A flat cam follower makes the sine wave fatter,
71 Devimead A65 750 56 Norbsa 68 Longstroke A65 Cagiva Raptor 650 MZ TS 250 The poster formerly known as Pod
What's the point in having a garage full of rusty BSA parts if you don't make use of them? Thanks to my labeling system ("Rusty BSA Parts Shelf #1, Rusty BSA Parts Shelf #2, ...) I quickly found a box with a half-dozen cams. But, they weren't necessary because when I then dragged a set of B34 cases down from the shelf it already had cams in it. Less conveniently they turned with difficulty, but a little penetrating oil quickly dealt with that.
I'll be making a more careful measurement soon, when I have more time, but for now I counted the no. of teeth on the cam between the tappet just starting to open, and just finishing closing. That was 12.5 teeth, times 10-deg./tooth = 125-deg. x 2 = 250-deg. engine. The part number is on the cam, but since these are just interim measurements I won't take the time to try to find where their specs. might be listed to see how well it matches. I'd estimate my ability to count fractional cam teeth is within 1/4, corresponding to an uncertainty of ~+/-5-deg. engine.
I then removed the tappet and installed a 2" dial indicator. Conveniently the ID of the tappet guide is the same as the OD of the indicator making this easy. With the tip of the indicator rather than the 1"-dia. flat of the tappet the two sides of the lobe were 11 teeth apart = 220-deg. engine. So, for draft simulation purposes I just have to subtract 30 degrees from the duration of each valve spec. to get the duration of the respective cam.
After doing this, inputting the cam values to the program, and tweaking the parameters to get a reasonable, but not perfect fit with Clubman cams, it showed ~40 h.p. at 7000 rpm. Simply switching to Touring cams dropped this to a reasonable ~30 h.p.
Don't memorize these numbers because they'll change once I enter more precise cam data as well as spend some more time tweaking the parameters in light of them. The important thing is, thanks to rusty BSA parts shelves, I'm no longer dead in the water.
Note the lack of data as yet on any pre-DBD Clubman, so keep those dyno sheets coming (emphasis added).
The more data I have to narrow down the range of parameters, the more accurate the simulations that will be produced. As a simple example, the inlet tract and exhaust pipe lengths both affect the calculated results (although not in exactly the same way), so it's possible to get "reasonable" fits to a single dyno curve for a range of both of these. However, having dyno data on different configurations narrows the range of parameters that give "reasonable" fits.
I'm still in the stage of "training" the software so the more data I can feed it, the better. If, say, there were data on both the CB Scrambles and Clubman (same head but different cams and exhaust), and if the parameters were adjusted to fit both, confidence in having a correct result if simulating, say, substituting Road cams would be increased. So, keep those dyno sheets coming.
After picking the type of combustion chamber, CR, and valve layout, the variables the program uses for the intake of a head are:
valve dia. average port diameter port length flow efficiency
The first two are determined by measured dimensions but, for reasons I discussed in a previous post, the optimum value of port length won't necessarily be exactly one of the values shown by a red arrow in a post of two days ago. Flow efficiency is a percentage that the program says can vary from 25% for a 'bad flathead' to 55% for a 'racing head' to 65% for an 'excellent racing head' to 75% for the 'best racing heads available today'. For the particular simulation I have set up in the program at the moment, changing from 55% to 65% has a ~5 h.p. effect at 7000 rpm.
Anyway, I'm speculating at this point, but I predict that after fitting the data I'll find the best-fit port length and flow efficiencies of, say, BB Clubman and Scrambles heads to be pretty close to each other. That said, as the 55/65% example in the previous paragraph shows, "pretty close" still results in significant differences in the output. Which is why the more measurements I can feed the program, the better the simulations will be. Looked at another way, it will be interesting to see if systematic increases in flow efficiency are shown by the simulations when going from ZB through to DBD.
Today's task is to make careful measurements of the cam duration with and without the tappet so I can have accurate data for the BSA cams to enter in the program.
The first photograph shows I made a spacer for a cam and used rubber cement to glue a protractor to it using a rod with a small dimple from a center drill to locate it. I then glued the spacer to the cam, again using rubber cement. Since I had the engine on its side neither joint had to be very strong as long as I didn't bump it, which I didn't.
As can be seen from the second photograph, using a fine wire in a stud let me read angles to 0.25 degrees (0.5 degrees engine). I found conflicting information on whether BSA's cam specs are from measurements made at 0.014" or 0.018" lift so I measured both. It makes a 5.0 deg. difference in duration depending on which value of lift is used so it's important to figure out which is the correct one.
The third photograph shows how I made the measurements with the lifter in place, and the fourth photograph shows the measurement with a dial indicator having a point tip (using 0.014" for both of these photographs). Taking several measurements each and averaging I arrived at 28.5 +/- 1.0 deg. (0.014" lift) or 33.5 +/-1 deg. (0.018") as the angles to subtract from BSA's valve data for duration to convert to cam duration.
For example, BSA shows for the 65-2442 inlet cam that the valves open 65-deg. before TDC and close 85-deg. after BDC for a total duration of 330-degrees. To convert this to the cam duration you need to subtract either 28.5 or 33.5, depending on which lift is the correct one. I hope someone can point to a piece of BSA factory literature for the correct information.[*]
Since the cams are symmetric, to find the opening and closing angles of the cams you would subtract half this value. For example, since the valve opens at 65-deg BTDC the cam opens at 65 - 28.5/2 = = 48.25-deg BTDC (if the 0.018" value is the correct one to use).
So, today's burning question is, 0.014" or 0.018?[*] Accurate simulations await the answer.
[*] It's 0.018". I found it in a 1955 BSA 'Gold Star Data Book'.
Last edited by Magnetoman; 05/22/193:59 am. Reason: added [*]
"After a bit of frustration I finally realized (...I think...) that the opening/closing data for BSA cams aren't for the cams themselves, which is the common way of specifying things in the automotive world, but rather are for the valves."
I'm surprised that your simulation software wants to use cam data rather than valve lift data to predict performance. Isn't the valve lift data more relative to performance than what happens at the cam lobe? Megacycle states that their cam data is measured at the pushrod for pushrod motors, so even those numbers are not what happens at the valve which depends on tappet radius as well as rocker ratio. For example, Megacycle's data for the BSA B50 X4 cam is measured at the pushrod using Megacycle's specified 3/4" radiused tappets, but you can get them to grind tappets at 1-1/4" radius to get more duration as well as overlap.
You could, of course, degree one of your known BSA Goldstar cams to compare with the published data to see if it correlates.
I'm surprised that your simulation software wants to use cam data rather than valve lift data to predict performance
The simulation software allows a person to explore effects of rocker ratio, lifter type and diameter, and valve train type (i.e. three types of pushrod/rocker, which I haven't fully explored as yet). To do this requires having raw cam data. Since BSA's valve data has the 1"-dia. tappet baked in it wouldn't be possible to explore the effect of different lifters if the program used valve lift rather than cam data.
Originally Posted by koncretekid
You could, of course, degree one of your known BSA Goldstar cams to compare with the published data to see if it correlates.
That thought crossed my mind, but only for a microsecond before I dismissed it. I'm not going to pull the cams from any of my engines for this project. Also, I degreed my Ariel's cams so I know how to do it, as well as know how long it takes to do it. So, I'm not even going to look through my boxes to see if I have any Gold Star cams (the ones I used for the measurements over the past few days have different numbers so are likely B31 or B33).
Originally Posted by Kerry W
This would explain why, 30 years ago, when I plotted data on many cams to find the ones closest to the factory specs, none came close.
To be fair to BSA, they were consistent about calling it "valve timing." That said, every time I've seen tables of those tabulated values I assumed it was cam data, and as a result it was a frustrating few hours of running simulations whose output was counterintuitive before I realized the source of the problem. It does make sense that BSA listed the values the way they did because 99% of the riders who cared about swapping cams would have wanted to know their effect on the valves. The raw cam data would have been useless to them.
Maybe someone has written about BSA's "valve vs. cam" issue previously, but I don't remember reading about it, and I couldn't find anything on the web. So, it was another shoal in these uncharted waters.
Two other shoals I'm dealing with at the moment are the Catalina head and the nature of BSA's dyno data itself. If I had a DBD Catalina and Clubman head on my workbench I could measure the length and diameter of the inlet tracts. Comparing how much the measured length of the Clubman's tract differs from the value I used in the simulation would tell me how much I needed to adjust that of the Catalina. I also need the diameter.
I've mentioned in at least one other thread that the BSA dyno data is too good to be true. Some of the dyno sheets are plots, but others contain what are supposed to be the raw data. The attached curve is a plot of that raw torque data for a CB Scrambler. Note that the first three data points lie precisely on a straight line. If any of the three measured torques had differed by as little as 0.05 (i.e. 0.16%) from the plotted values a straight line through any two points would miss the third. The same would be the case if the measured rpm for a given data point had differed by as little as 50 rpm (also ~0.15%) from the precise values of 2500, 3000, etc.
OK, now imagine yourself in front of an engine dyno in a noisy, smelly room with a 40 h.p. engine vibrating on its mount while you adjust the throttle to reach each new setting of rpm and then read the analog torque gauge. Do you think you could adjust the throttle of this vibrating engine to hold it at precisely 3500 rpm to within 0.15%? And, even if you did, do you think the torque reading would fluctuate less than 0.15% and that you could read it to that precision on an analog gauge? OK, the answers are no, which means the data on these engine dyno tests are fake. Well, not completely fake, but "dry labbed."
I can only guess but a plausible scenario is a "master engine" in a given configuration was measured and the data, complete with fluctuations, was plotted. The draftsman then drew a smooth curve that was best-fit to that data, along with a set of parallel curves somewhat higher and lower than that curve. From then on "production" engines would be tested at whatever rpm had been decided on for this configuration (say, aim for 6500 +/-200) and the torque measured to whatever accuracy was achievable (maybe 1%). Whoever was in charge of producing the dyno sheet found the plot for the relevant configuration in a notebook, noted the measured rpm (which wouldn't have been precisely 6500) and torque, located the parallel curve on the master plot that was closest to that data point, and then "dry labbed" the entire curve based on it.
The reason this is relevant for this thread is, among other choices I have to make, I have to decide how well the parameters I choose need to fit a "measured" BSA curve across its entire range of rpm.
I have a Catalina head sitting on my workbench, where it awaits installation after I pluck up the courage (and find the time) to install my shiny new Pearson crank in the cases and actually build the engine. So I will be happy to do the measurements for you this weekend.
Also, regarding carbs, since my Catalina air box is also sitting on a shelf, I can send that to you for your flow test. Seems a shame to take yours off when you have your Catalina running so sweetly, and before I get a chance to break it.
1949 BSA ZB34 'Bitsa' 1959 BSA DBD34 Catalina 1973 Norton Commando 850 R 1974 Norton Commando 850 R (I know, one too many) 1975 Honda TL250 Trials, a new addition to the family 1998 Montesa Cota 315 HRC 2004 Ducati M1000ie
since my Catalina air box is also sitting on a shelf, I can send that to you for your flow test.
Thanks for that offer, but since it is held on by only two bolts I can remove mine in less time than it would take to re-box yours and take it to the UPS office.
Basically, air flow has a linear effect, but the inlet tract creates a resonance so its effect is non-linear. Because of this the air flow through the carburetor and filter (if any) has an effect on the simulation, but much more significant are the dia. and length of the inlet tract that correspond to the red arrows in a previous post..
The second image is from a test sheet someone sent me, for a BB with Touring cams. As can be seen, the h.p. increases smoothly to ~5500 rpm but then hits an inflection point. My Catalina simulation shows the same type of behavior. Changing the carburetor air flow raises or lowers the overall curve (the effect increases with increasing rpm), but changing the port length or diameter moves the inflection point to higher or lower rpm.
I easily can pick parameters that move the inflection point to higher than the 6500 rpm upper limit of the Catalina test sheet (which doesn't show this in the measured curve, so it's higher than 6500 rpm), but the program has enough parameters that I probably could make a h.p. vs. rpm curve look like the New York skyline if I didn't impose physical constraints (like actual valve diameters, inlet tract dimensions, etc.). Hence, I eagerly await your Catalina measurements.
It's entirely Gordo's fault that I'm spending my time on these simulations so it's only fair that I inflict a draft ZB Scrambler simulation on him without further delay. I'm still labeling it as 'draft' but I hope to take that qualifier away as soon as I have the Catalina head data to use in another check of my parameters.
All I did to turn the DBD Clubman simulation into a ZB Scrambler simulation was to change to the ZB's 8:1 piston, cams, 49" straight exhaust, and valve diameters. In addition, I used:
1.21"-dia. inlet (given in earlier post by Gordo) 3.6" port length (from red arrows in drawing in a previous post) 5% less flow for the 1-5/32" TT than I measured for a 1-3/16" TT 1-5/8" exhaust diameter (is this correct for a ZB?)
For the attached graph I simply traced the BSA data in a new layer in PhotoShop so it would remain visible and then overlaid the simulation with its axes scaled. As I think people would have to agree, the fit is remarkably good, reproducing the shape very well and the absolute value to ~8% (the fit of the program to measured automobile dyno data is claimed to be +/-7%). I should note that the simulation also over-estimates the total h.p. of the DBD so the fact it does as well for the ZB is consistent.
Again, I first worked out a simulation for a DBD Clubman using physically reasonable parameters. I then simply changed only the relevant measurements (compression ratio, valve dimensions, cam specifications, etc.) and the result is shown in the graph. With each such result my confidence in the output of this program grows.
Of course, the point of doing this isn't to reproduce curves we already have, but to investigate the effects of making changes for which we don't have curves. That will come shortly, as soon as I have enough confidence in all the parameters to remove 'draft' from the simulations.
MM: My ZB head measures about 1.75 to 1.78 inches going around where the pipe goes into the head. Can't accurately measure if the opening is tapered or measure the actual opening that the pipe would butt up against.
The roadside repairs make for the best post ride stories.
MM: My ZB head measures about 1.75 to 1.78 inches going around where the pipe goes into the head.
That sounds consistent with a 1-5/8" ID for the pipe.
In a previous post I showed evidence that BSA faked their dyno data. Even though no one came to BSA's defense when I made that accusation, I'll go further. The graph attached to this post is the same as the graph in my previous post except with the simulation displaced down to match the h.p. curve. As can be seen, the h.p. curves overlap almost perfectly. That's a good thing, right? But, what's the deal with the torque curves below 5000 rpm?
Torque and h.p. are linearly related through (torque x rpm) / 5252 = h.p. According to p. 52 of 'Goldie' the BSA dyno worked by balancing the engine output against a variable weight with (W(lbs.) x rpm)/4500 = h.p. The text doesn't give details but that weight would be at the end of a lever so would give the torque once the dyno was calibrated, which is where the constant 4500 comes from. Again, though, the torque and h.p. are linearly related. What this means is, if the h.p. curves overlap, the torque curves must overlap as well. If they don't, as is the case for the attached plot, something is rotten in Birmingham.
The three BSA dyno pages I've taken the time to check all have issues of the torque and h.p. curves not being correctly related to each other. So, do I use parameters that best fit the torque curves, which is what the BSA dyno actually measured, or the h.p. curves that they calculated from the dyno data? The problem is, once you know some data is faked, all data is subject to suspicion. I won't take the time to explore this issue further here, but it's yet another factor to contend with in trying to decide on the parameters that best simulate BSA's "data." Oh what a tangled web BSA has weaved.
Have been away for a couple of days and just catching up with 'Dynogate'.
30 years ago it struck me that the curves were all too 'nice' to be true and that the reality must have been data points scattered above and below the charted line...though I assumed the general trend reflected the engine it was related to, especially the peak power. I've never seen a BSA graph that showed the power peaking and dropping, with RPM pas peak power RPM, as I'm used to seeing with my 2 strokes. The impression is that there is more power at higher RPM, if you're brave enough to go there, or can accept the consequences ($$).
No generalisation is wholly true, not even this one. Oliver Wendell Holmes
A. Golland, the nominal author of 'Goldie', has been identified as being the BSA development engineer Arthur Lupton. Lupton says about the dyno testing that "The engine was run under moderate load for about an hour and then the standard exhaust pipe would be fitted, when power tests were undertaken at various engine speeds. ... a series of readings were completed in a very few minutes..." Lupton's description tells the reader that the dyno room produced careful, accurate measurements of every engine over the full range of rpm.[*]
At the height of production in the mid-1950s Gold Stars were despatched at an average rate of 5-7 per day. Bolting an engine in place, warming it up for an hour, testing it, then unbolting the hot engine and removing it from the stand barely allows time to meet that average production rate, let alone higher short-term rates when batches were needed for overseas orders.
Given the above, the dyno room would have been the bottleneck. If the guy locked in that hot, smelly, noisy dyno room knew a, say, good Clubman engine produced 31 h.p. at 5000 rpm as shown in the curve on p. 55 of 'Goldie' why not warm up an engine for just 30 min., run it up only to 5000 rpm to avoid any risk of damage, and measure the output at that single rpm? If it produced at least 31 h.p., use the time while it was being unbolted from the stand to drylab the full curve from the sheets in the test notebook, move on to testing the next engine in order to meet the production demands, and give the day's stack of dyno sheets to the supervisor at the end of the day to sign.
I'm not saying this is how the dyno "data" were produced. But, as the examples I've included in previous posts show, the "data" are too good to be true, and the torque and h.p. values cannot both be correct. This means something is amiss in the way those sheets were produced.
Again, the relevance for these simulations is having to decide at what point a given simulation fits faked data well enough to have confidence to use the parameters from that simulation to fit other faked data. The normal situation would be to develop a simulation that does a reasonable job mimicking known reality. Here, there's a good chance the simulations are more real than BSA's "reality."
[*] The attached image is the signature on a 1961 dyno sheet.