It's gotten a bit long... We might need a separate topic in which to discuss a possible new performance model.

!! When I talk about a "(possible) new (performance) model", I am not saying there will be a new model. We can just think and dream at this point.

I agree that a practical, realistic model needs another boat-specific parameter, like weight. Since we're talking about boats, weight and water displacement are about the same, so together with the values in the polar (boatspeed), a realistic model can be made that uses ideas from real physics (momentum, kinetic energy).

The introduction of a new parameter is probably a lot of work: the database needs to be altered, some code-structures need to be changed. If we want the client to show the parameter (for sake of transparency), the client needs to be altered, including protocols between server and client, which might "break" (parts of) other software (like brainaid's).

This is assuming we keep using the percentage, as-is. That could be changed too. Since we're speaking about realism, momentum and/or kinetic energy could be a better base for the performance model, a percentage is not really realistic in the first place. Changing this is probably even harder, with more implications.

This is not a reason not to change, it's a reason to think well, do the calculations, discuss, simulate and try a lot before deciding upon "officially" presenting the idea to SOL/dev dudes. So, it's going to take a while, and we should take the time.

I don't think an intermediate (small) change (like removing the 93% rule) is wise, because:
- the current model is effective
- I'd like to keep "changing performance models" to a minimum. So when we change it, it'd better be for the long term.
So, (as far as I'm concerned), we're not changing the performance model until we have a long term alternative.


The 93% rule is not realistic, but it's definitely not the only thing, and probably not even the most important thing that makes it unrealistic.
I don't fully agree with Rods theory about slowing down. Of course, when in a straight line, heavy boats don't slow down as much as light boats. But we're talking about turning the boat here. I don't think slowing down in a turn is really influenced by the mass of the boat.

Before I go on and describe some of my ideas, I'd like to draw the attention to A2R's post, it's exactly how I think about the matter, and it's the main inspiration of the seconds of two ideas that will follow. Click for A2R's post @ 2013-12-20 20:18:19

----
Idea#1 (this is JUST an IDEA):

I propose to store with each boat/polar the weight of the boat. Ideally this should be the actual weight, but the boat/polar designer is free to change the weight of the boat so that its performance is more realistic.

The performance loss will be calculated using difference in cc and whether the boat has changed tack. With a 90 degree turn, there'll be a performance loss of 40% to 80% (so performance will drop to 60% to 20%). When the boat gybed or tacked an additional performance hit (of about 10% to 20%) is calculated. Performance cannot drop below 0%.

The recovery of this performance loss will depend on the weight of the boat. A light weight boat (like a laser dinghy) will have gone up to speed in the next server jump (10-15 seconds later. A heavy ship (like our 90m barque) will need some more time to get up to speed again. The recovery does not need to be linear.

Idea#2 (this is JUST another IDEA):

Alternatively, we can step away from the performance percentage, and allow for an even more realistic model. A function of mass and speed will give a number (like momentum, or kinetic energy). It'll cost time to change this number. Example:

We need the same weight parameter from the previous idea. We'll use momentum, to keep the calculation simple.

mass: 100
speed: 10
=> momentum: 10*100 = 1000
Now we change angle, in which we lose some momentum, let's say we lose 600. And the polar says that at this angle, our speed should be 14 knots.
So directly after the turn:
mass: 100
momentum: 400
=> speed: 4
After some time, when momentum had the chance to adapt to the new angle:
mass: 100
speed: 14
=> momentum: 14*100=1400
New, let's say we change angle again, but now to a very slow angle (upwind), at which the polar gives 4 knots. In the turn we lose, let's say, 500 "momentum". So directly after the turn:
mass: 100
momentum: 900
=> speed: 9
After some time, when the momentum had the time to change:
mass: 100
speed: 4
momentum: 4*100 = 400.

The drop in momentum could be a percentage that is determined as I described the the first idea.

The recovery of momentum is the main thing to figure out in this example.

__________
I hope this will trigger people to share (more) ideas, think about the consequences, and slowly results in a new performance model that we can then present to the dev dudes. It's quite possible that they say "Oh no.. we're not going to do that any time soon"

__________
I DON'T say I want the performance model changed, I am just trying to help those who what it to change to create a better model. I DON'T say that creating a better model will result in implementation of this model, I am just trying to explore the possibilities with you.

--- Last Edited by kroppyer at 2018-16-21 15:76:42 ---

I agree with you, Rod, with over 93%

I am sure you have made turning and tacking tests. I am 100% sure that the performance model is unrealistic. I just made one in the Christmas Race. Tack/Gybe from TWA 45 to -130.268 in TWS 5.87 knots the performance went to 96.08 from 100. After gaining performance to 100%, making a 100.268(the .268 is to be sure to get just over 7% performance to loose) degree turn from -130.268 to -30 TWA (without any boatspeed involved in the calculation ??) the holy 93% performance appears. Why has the boatspeed lost influence if you turn your boat?
Sailing VMG upwind in your VO74_v4 you need TWS 28.1(at 45.5 TWA)and downwind TWS 12.47 (at 138.68 TWA), then you reach the 7.01% performance-loss. Is a gybe that much more difficult to make than a tack. I am not trying to explain the performance model, just trying to understand.

I think every SOLer wants a game as realistic as possible, as fair as possible. A2R, has really something there with the boat specific momentum coefficient.

It seems to me that we may start a performance brainstorming thread

There is an aspect of the gybing part of the PL problem that has not been considered so far. I speak from the experience of radio-controlled model sailing yachts, specifically the Soling One Meter.
When gybing, one usually pulls in both main and jib sheets (using a fairly powerful 'sail winch' servo, capable of pulling in the sheets in a wind up to approx 20 kn). One then turns down wind, usually with little or no loss of boat speed, and the sails promptly gybe. The sheets are then let out to the required amount, as the boat continues to turn onto the new course. All of this occurs with little or no loss of boat speed.
If someone has a similar description of the gybing behaviour of small, medium, and large yachts, as well as 90 meter clipper ships, perhaps their knowledge might be incorporated into any model that may be chosen in the future to describe a realistic Performance Loss for SOL

Depending on the type of boat, and the wind strength, gybing and tacking can be without noticeable loss, sometimes even with a gain (roll-tacking/gybing in dinghies).

I am encouraging a performance model in which we have performance big performance losses, and fast recoveries. Rather than a gybe, in which the speed does not drop below 90% of the max speed, but it still takes over 20 minutes to get back up to speed, as is the case now. More realistic is a performance hit of 50%, and a recovery of about a minute (of course depending on the boat an other factors).

Attached, description, graphs and other stuff about the current performance model.
Also, the "source", so that people can change the document and use it for other purposes if they want. No strings attached, as long as you're not going make money of my work, by selling it (that is basically the summary of (correct me if I'm wrong) I think GPL)

Great thanks Kroppyer, for making us nice instruction files.

The mass * boatspeed = momentum idea, let's think further of that.
Which variables have greatest impact on recovery?

I am very pleased to see the direction this thread has taken lately. Now we are discussing realistic changements to SOL instead of preventing "tacking gamers". Thank you.

Is it possible that the "dev dudes" correct current gybe/tack formula to the more accurate
e = 1.0-(aftergybe.bs*current performance)/200.0.

--- Last Edited by Klyvarn at 2013-12-22 10:39:41 ---

I was preparing time ago some thoughts about this subject.
The attached file has my words about it, and they go with Kroppyer.

IMHO, the current performance system has the following pros:
1. It is same for all boats
2. The basics are relatively simple
3. My understanding is that from a game engine perspective it is easy to implement

and the following cons:
1. The extent to which it simulates reality is limited
2. While the basics are simple, estimating the impact of any given maneuver is not

Now, whether the system should be changed, that's another story. To me the system is one part of the game mechanics, the rules, and as such it does not seem to function too badly. If it ain't broken, at least think what you're trying to fix.

Hi SOLts!

Does anyone know if there is any minute advantage gained by sailing CC vs sailing TWA.

What I really want to know is this "Does sailing on TWA, where the game engine makes continuous small changes, cause any small performance loss? Thus, over a 6 hour period all the very small hits on performance loss adds up to a loss of a yard?"

Been sailing here for years now and I still dont know if this is of any concern or not. So, be honest and let me know

Not entirely sure about whether performance loss is calculated when sailing normal twa or cc.

I actually think that performance loss is calculated from twa, so cc is the type of command with the invisible performance loss.

We sailing cc in a windshift, so that the boat auto-gybes, performanceloss is calculated. Other than that, the performance loss from continuous twa changes (or cog changes) will be insignificant. Before you worry about such small performance losses, you will have so many DCs, following an exact course, that is doesn't matter anymore whether you use cc or twa.

I can give you the numbers and everyting, but I think the conclusion is more interesting.

Disclaimer: though I am an admin, I have no knowledge about the sailonline code (including performance loss) that isn't available to normal users, so the above is might be wrong (i would be extremely surprised though).

I have been thinking (deeply, of course) about the practical aspects of the performance loss problem, and have concluded that one aspect is missing (It is possible I could calculate it, but not instantaneously!!)
The info really needed is the angle of turn which would produce a perf loss of about 8% at any given speed. This would allow the SOLer to divide any turn into two parts, so that a tack or gybe would result in no perf loss greater than c. 8%.
This seems like an obvious tool to use, if it was in the form of a graph, showing a line for each of tack and gybe, with the margin (8% rather than 7%) of turn needed to allow for the keyboard entry of the second part of the turn before the perf loss is decreased to below 7%. SOLers could place this graph beside their computer work stations as a sheet of paper, or have it as an extra Window on their computer screens along with the SOL map.
I would do this myself, but would prefer it done by someone a bit younger and more mathematically inclined---like Kroppyer????

--- Last Edited by Rod at 2014-07-17 16:47:37 ---

--- Last Edited by Rod at 2014-07-17 18:19:39 ---