The Simulation of Gold Wing Engines and/or Understanding - Trevor White...continued......
Maybe Honda’s CB 750 from 1968/9 would be of interest. After all, it is often taken to be the original superbike. (Like everyone else, I tend to forget the post-war Vincent Rapides, Shadows, etc.. That begs the question of what defines a ‘superbike’. But that’s another story.) I couldn’t get all data to put in Table 2 but what there is looks curious. In spite of the CB 750 being crowned with superlatives, its cam timing is a real softy pussy cat – milder even than the ’75/77 Gold Wing. With a 28 mm carburettor, it may not have always gasped for breath but it wouldn’t have been able to climb high mountains of performance. Here, too, Honda was offering a very low stressed engine. Yes, in 1968, 67 horses were quite a performance, but retrospectively it was modest from a 750 cc four-cylinder engine.
Here I suspect that Honda was being honest with itself – by not getting more power out of that CB 750 engine. Was it recognizing that it could build pretty good engines but, like the other Japanese manufacturers, had in 1968 not yet learned to make the frames necessary to contain high engine performance – frames that did not easy turn into epileptic ironing boards on the road?
To illustrate what an aggressive timing can look like, one only has to turn to the Norton Commando parallel twin. Here valves have enormous open-durations of 304° and overlap a whopping 92°. So what? It only produces 56 HP compared to the contemporary CB 750’s 67, but this begs the question of how effective that HP could be. Those Commando horses had to carry around less than 400 lb (180 kg), whereas the CB 750 was saddled with 481 lb (218 kg). The BMW was lighter than this but it still carried 452 lb (205 kg). But why did the CB 750, and its many offspring, survive whereas the Commando (and the British industry) died?
One can only guess an answer. The Commando design-concept had its roots in the 1938 Triumph Speed Twin. This concept was, with much development and little creativity, flogged to death in the following 35-40 years. Just about everything had been wrung out of it – and it had nowhere else to go. It had become overtaxed and over-developed. It was not very reliable and it needed constant tweaking to keep it in trim. Perhaps the keyword about the CB 750 has already been used – under-stressed! Already Honda had established a hallmark in reliability. Yes, the CB still had to be maintained but every time the rider reached for the keys, it was not accompanied by a "Will it? Won’t it?" question. The rider just pressed the electric start button and headed for the horizon without leaving a pool of oil behind him.
One can probably assign the same virtue to that BMW 90S as to the CB 750. Even though this Beemer was the sportiest machine that BMW had produced up to that time, the specification provides decent power without busting a gusset to squeeze every last bit out of it. The flat twin and BMW survived. If it is survival we are talking about, then the Gold Wing must receive all the prizes. The four-cylinders stayed in production for 12 years, with only few changes of design principles of the motor. The 6-cylinder in the 1500 form lasted also relatively unchanged in the engine room for 13 years – and now comes a 6-cylinder as an 1800.
This is saying to me that Honda’s policy of conservative engine tuning has more than paid off over its long history. On the other hand, I would be willing to bet that another 20 or 30 horsepower could be pulled out of this engine. That would mean putting in a much more aggressive camshaft and feeding the beast with a wholly new carburettor system. It would be fun to do – but not very sensible to my way of thinking. First, for the money such modifications cost you could buy yourself a really hot sports bike. Secondly, as suggested, that Honda reliability and long life comes from lack of stress. Push up the ante, and that crankshaft, gearbox and valve train might throw in the towel sooner than you expect.
There is a side issue to mention. So far I have been looking at the mechanical specifications of these machines. However they do not contain all the information relevant to the pros and cons of performance required by a rider. It is one thing to consider the cost of modifying and engine to provide something special in respect to horsepower or torque. Yet there is the daily bread and butter aspect of the cost of running the bike in its standard form. For instance, information is not readily available about the fuel consumption of these machines. What is available, though, is the type of fuel required. BMWs then as now, call for a higher octane rated fuels than Hondas. In Europe these are 98-octane ‘Super’ and 95-octane ‘Normal’. Just to take Switzerland as an example, ‘Super’ fuel is about 7% more expensive than ‘Normal’. So, again, there are no free lunches. If power is obtained from higher compression ratios that demand high-octane fuel, you have to pay for it. If higher power is found by getting more Suck per intake – ramming more fuel mixture in – then you will have to pay for it. The main neutral operating-cost ways of upping performance would be either to improve the energy release – the combustion process; to improve the translation of this into the reciprocating action of the piston with a more efficient power stroke; or to reduce the power sapping effects of waste heat and gases. Only the latter might be reasonably cheap.
Other power losses can’t be reduced - the pumping losses that underlie the function of the engine. Other losses are also inevitable, even in a well-functioning motor – for instance the friction losses in the engine. The Motion Software texts note that about 70% of all engine friction losses are caused by the pistons moving in the cylinders. Their simulations provide a datum that allows the estimation of these losses. I used this to assess this power loss of the ’75/77 GL1000, up to operating temperature and pulling 80 HP @ 7500 rpm. This loss was about 12 HP! I now have a clearer idea why it is difficult to start a cold engine and have it running smoothly. Drained of lubricant, and when initially lubricated it is with a thick cold oil, the pistons have to overcome frictional forces way above that causing a 12 HP loss when up and running.
This also made me think of that poor little starter motor having to overcome not only this resistance but that of the crankshaft and valve-train as well. And then, what about that battery that has to feed all these demands? It sometimes makes you wonder that the whole thing works at all!
MOTION SOFTWARE SIMULATIONS
Let us be clear about one thing in respect to engine performance simulation. It is impossible to simulate any engine exactly. The purpose of a simulation – it can be looked on as model building – is to provide a tool. It allows the user to ask "What if …?" questions – such as about what might happen if I could fit a Norton Commando camshaft in Gold Wing GL1000, or about the possible consequences of fitting GL1200 carburettors to a GL1100. Such a simulation delivers no absolute answers – only relative ones. One thing is clear, though - the better the simulation program then the better the tool it is.
There is another limitation. Internal combustion engines have been around about 125 years but, in spite of enormous advances recently, the processes going on are still not fully understood. What is known is often associated with very complicated mathematics that challenge the capacity of even the most advanced computer. However, such a computer could not close the gap over what is simply not known. Therefore, even the most exotic program starts out by having to make assumptions and approximations. By necessity, a program run on a desktop PC introduces more approximations. Yet, with such PCs having a performance undreamed of just a few years ago, these additional assumptions stay within reason.
That last issue deals with the theoretical problems of simulation. There are then practical limitations. Even professional-standard programs can’t take the absolutely finest detail into account, purely because of the difficulty of getting the required input data. Take the case of camshafts in general, and of their grind (or shape) in particular. The timing specification is relatively easy to determine and these data are basic information. The amount of lift imparted to the valve by the cam lobe (with the cam followers) is also often available. Such information gives the when and how much of valve opening and closing, but it doesn’t give the how.
The ramps on the cam lobe may have a very gentle slope, opening and closing the valve very mildly. On the other hand, the ramps may be very aggressive, have a very steep slope so that the valve operation is very quick and forceful (with, by the way, those forces also stressing the valve-train more than a mild grind). It wouldn’t be far wrong to suggest that the camshaft of any engine is the most individual, unique part. The mathematics, and the data to use, to model every conceivable camshaft is virtually impossible. Therefore, a simulation programmer has to take a limited number of general options. These are based on current practice and common use in automotive application. They range from cams grinds commonly found of stock vehicles oriented towards modest performance, long life and economical running to hot cams whose only purpose is to produce maximal performance on the track.
A similar approach has to be taken towards the exhaust system after the valve. Again, commonly used configurations of exhaust manifolds, pipes and collector/mufflers are available for choice. Some data for the induction system upstream of the valve are available, such as the carburettor venturi diameter and, maybe, the induction flow rate of the whole system. Again, though, generic configurations of intake manifolds are used to give approximations of the almost infinitely varied details of individual manifolds.
The specifications provided in the various Gold Wing workshop manuals did not include induction flow rates. No doubt Honda or Keihin have the results of practical flow-bench studies in a drawer somewhere – but they aren’t available to the hoi-poloi like you and I – well, on asking Honda (Switzerland) I was told that such data, if available, would be treated as company secrets. Rather than guessing (for which I had no basis whatsoever), I made what I thought was a logical estimate.
Earlier on I calculated an estimate of the amount of charge that got rammed into the cylinder. (Formulas (5) above gave a volume of 19.79 in3 or 324 cc). This could only get into the cylinder while the intake valve was open. Taking 7’500 rpm as my reference point, it is no great hassle to calculate the time that an intake valve is open during each cycle – the IVOT, intake valve opening time. With the volume and the time, it is an easy step to work out the average flow rate though the intake system of each cylinder. Multiplying by four gives that flow rate for the whole engine. Also from the flow rate and the venturi dimensions, the average velocity of the air through the carburettor can be calculated. These various estimates, with the exact GW specifications from Table 1 were fed into the Motion Software simulation program to obtain simulated power and torque curves over the rpm-range. I have listed the calculated estimates and the peak HP and torque values in Table 3. (I only use peaks as representative because of the mountain of data that makes up such curves.)
I must admit that I never really thought about some things before I started playing with these things. I mean, I never imagined that valves, on the ’75 GL1000 for example, are only opening for about 1/100th second – 10.44 ms to be exact – when running at 7’500 rpm. (My curiosity awakened, I took this calculation a step further – just for fun. The valve starts at zero kph/mph. It accelerates up to travel its 8.5 mm of lift, stops at 0 kph/mph, then accelerates down to reseat at 0 kph/mph. That is 17 mm in 10.44 ms; or 5.86 kph (3.66 mph). That is the average speed, with the peak reaching double that – 11.72 kph (7.3 mph) – and that 3’750 times a minute. That is quite a beating that the valve train gets. Some Gold Wings have been doing that for more than 25 years!)
Moving on to the ‘78/79 model, the intake valve opens fractionally shorter, but the smaller venturi makes the mixture move faster. This higher induction velocity gets more mixture into the cylinder through that ramming effect. Moving on to the 11 and 1200, they have more capacity pulling in mixture and now a 30 mm venturi. This continually increased the mixture velocity and, through ramming, the induced volume as reflected in the induction flow rate. If the common wisdom is correct, the main consequence of improving induction is to increase torque. (As discussed above, this could also increase horsepower – but that would depend on the portion of the rpm band where the torque increased.)
Turning now to the actual simulations – it is clear that, in comparison to the published performance data, the simulated values are all too high. Now, I reckoned above that Honda built engines that are very conservatively configured. I set the conditions for the simulations as conservative as I could. On the advice of Motion Software’ support group I also toned down what I thought were the best estimates of the GW cam-form. My values came down to those given in Table 3, but still remained high. So, that engine really must be a mild pussy cat! - milder than even Motion Software allow for.
But this didn’t worry me unduly. I had no problem in accepting Motion’s advice that the idea was not to reproduce performance exactly (though that would be nice), but to get an idea about the performance of different engine lay-outs and/or tweaks. ‘Reading’ the specifications suggests that the first Gold Wing started out as a torque machine and not a horsepower wizz – and that this principle was extended all the way through. The simulation shows that, from the ’75 GL1000 to the ’87 1200, peak horsepower increased by 117%. The