Expedition Navigation Software 
1 of 1- pjl
- Expert
Offline - From: Lake St. Clair, MI
- Registered: 11/09/2015
- Posts: 128
Optimizer use of polar data
Can anyone please explain how the optimizer obtains the data from a polar file?
My question is, is this process a table lookup or an algorithm solution?
Another way of expressing my concern is: Will a polar with more discrete entries be more accurate than one that has only, for example, points at TWA = 45, 100 and 170? Which leaves a lot of landscape in between that is either open or estimated.
I'm trying to compare the routing results from Predictwind, Sailgrib and various gribs used by Expedition. PW and SG use canned polars and I'm trying to eliminate variables in this analysis.
- Nick
- Moderator
Offline
- Registered: 9/16/2015
- Posts: 2,964
Re: Optimizer use of polar data
Exp fits a surface, but with obvious constraints about the targets.
Accuracy is more a function of what numbers you use.
I don't know about SG, but PW tried to copy Exp which is why the outputs similar. They also used the Exp polars.
- pjl
- Expert
Offline
- From: Lake St. Clair, MI
- Registered: 11/09/2015
- Posts: 128
Re: Optimizer use of polar data
Interesting. Do I read correctly that the values are obtained by solving the curve equation for the specific TWA and TWS? That has implications that mean the instrument calibration has an effect on the ability of the polars to produce a good optimization, I think.
I have edited the PW polars to match the ORC numbers I use and still get deltas in the solution. That suggests that the cause for widely different solutions is in the grib, but that is much harder to detect because PW only tells me graphs on the grib data.
I am much more comfortable with the Exp approach because of the ability to adjust the gribs to local actual, but last Mac race here the PW boats that used it "blind" got the sweet spot on the course and beat us.
Thx, pjl
- •
Thread Starter- Nick
- Moderator
Offline
- Registered: 9/16/2015
- Posts: 2,964
Re: Optimizer use of polar data
Instrument calibration has no effect on the polar used in the routing.
It all depends on the data you feed into it and you need to decide on the day what is the best. I never found the GLERL forecasts much good for the Mac races, although they have the only current forecasts I know of. The NAM was generally best and Chris' models were good too.
- pjl
- Expert
Offline
- From: Lake St. Clair, MI
- Registered: 11/09/2015
- Posts: 128
Re: Optimizer use of polar data
Thx, Got it! The router takes the polar data and the grib input and calcs the optimum solution.
Need help with the reference to Chris. Is he the PredictWind principal?
Do u subscribe to the practice of making adjustments to the grib input on the course to better match actual and then running scenarios? In other words, has this worked for u?
TIA, pjl, Expedition Bigot.
- •
Thread Starter- Nick
- Moderator
Offline
- Registered: 9/16/2015
- Posts: 2,964
Re: Optimizer use of polar data
Chris Bedford. He would have given the weather briefing in Chicago and provides the race with forecasts and grib files.
The GLERL currents generally seemed quite believable.
I don't use the specific optimal course as much as the isochrones and timing. The value in the routing solution is using it with your interpretation of the weather and figuring out where you would be in relation to the weather features you think are important and how to deal with them.
Will's book is also a worthwhile read.
- pjl
- Expert
Offline
- From: Lake St. Clair, MI
- Registered: 11/09/2015
- Posts: 128
Re: Optimizer use of polar data
I have followed the GLERL (GLCFS to me) for years and found them to be good on winds, but that's a result of our using them in visual analysis mode before Exp came along. We had a dyslexic guy on the boat who was uncanny in looking at the progression of the wind patterns they displayed and picking the course. Maybe all hoocky, but we won several Macs that way.
The knock on GLCFS today is that the grib files are large, covering all the Lakes, unless I have missed something in being able to select a subset.
I've done Will's book, but too often find a discussion is more nourishing and that is hard to do in this area as the locals are quite protective of their proprietary experience.
Thanks for ur comments.
pjl
- •
Thread Starter