Wednesday, August 15, 2007

Tour de Madeira 2007 planning

Well it's time to start thinking about Tour de Madeira 2007. Well, actually it has been time to be thinking about it for quite some time now, and not to worry - I, your faithful blogger, have been doing just that.

Although the 2006 T de M was very successful (my take - PJs take), I felt that the route could definitely stand some improvements. All in all, it was a solid route, but there were a few areas that merited a second look. As always, feel free to familiarize yourself with the ground rules.

And I am pleased to report that I have (at last) finalized a route for the 2007 edition. So let's take a brief look at the changes / updates:

  1. After my Tour de Fairport Harbor planning, I had a breakthrough involving Eulerian paths and cycles and such, and so of course I began applying this knowledge to improve the Madeira route. There are 44 "bad" nodes in Madeira, not including the start and end nodes of Kenwood and Camargo. While I did not exhaustively prove this to be the case, I did quite a bit of study on the matter, and I believe that the shortest way to turn these bad nodes good is to apply the following 23 artificial edges
    • Kenwood / Dawson Kenwood / Shawnee Run
    • Dawson / Eleck Dawson / Rosecrest
    • Strifler / Springcrest Euclid / Pineneedle
    • Euclid / Wallace Euclid / Maple
    • Longfield / Camargo Euclid / Sanoma
    • Euclid / Hosbrook Euclid / Summit
    • Southside / Summit Southside / Fowler
    • Southside / Wallace Laurel / Miami
    • Osceola / Maxfield Longfield / Maxfield
    • Sanoma / Osceola Sanoma / Iuka
    • Sanoma / Sanoma Rita / Sanoma
    • Miami / Greenbriar Thomas / Greenbriar
    • Thomas / Mapleleaf Thomas / Tances
    • Thomas / Locust Dee / Thomas
    • Berwood / Jethve Dee / Britten
    • Juler / Wesley Juler / Dee
    • S Timberlane / Fowler Miami / Juler Ct
    • Fowler / S Mingo Fowler / N Timberlane
    • Hosbrook / Miami Hills Hosbrook / Shewango
    • Miami / Locust Miami / Shewango
    • Berwood / Homart Homart / Thomas
    • Rathon / Thomas Homart / Thomas
    • Thomas / Sanoma Homart / Thomas

    23 edges for 44 nodes instead of only 22? That is to handle the bad node at Thomas / Homart, and its neighboring bad nodes of Homart / Berwood, Thomas / Rathon and Thomas / Sanoma (the last 3 entries in the above list). You can think of it as 2 edges going from Homart / Berwood to Thomas / Homart and Thomas / Rathon to Thomas / Sanoma (going through Thomas / Rathon). I chose to break it out as 3 edges as shown above - distance-wise it's the same

    So, in comparison to the duplicate edges of the 2006 route, only duplicating these edges is approximately 0.4 miles shorter than last year's route. By my calculations it was 2441 feet, but it is a bit silly to give numbers that precise when my mode of measuring is an online mapping tool that can't hardly get that accurate I'm sure. Seems like a lot of work for less than 1/2 mile savings. But hey, 1/2 mile is 1/2 mile. And actually, given that there's at least one new street since last year. I think this will be the optimal route until and unless Madeira adds any more bad edges (most new streets are just going to be out and backs and therefore no changes to this route are necessary)
  2. In addition to shortening the distance, there were a couple of other tweaks that I tried to put in. First off, I tried to eliminate as many out and backs as possible. Obviously, all the culdesacs (and lo, there are many) are going to require going out and stopping / slowing to turn around and come back. But last year's route had (by my count) 9 additional out and backs. Every time you do this you slow down - it's much more efficient to design a route that takes them out. This year's route only has 1 non-necessary out and back, and that was left in to avoid the next point
  3. Avoiding left turns and traffic lights. I left an out and back on Miami Hills from the Timberlanes to Miami because without it, there was no way to avoid having to make a left turn onto Miami, and so I figured this was a good trade-off. The 2007 route tries to avoid having to turn left on to or off of major routes, and to have any intersection with a traffic light consist of 2 right turns at different sections of the route (to avoid potentially being stopped at a red light)
  4. Where possible, avoiding uphills. This is probably on par with the 2006 route - there's only so much you can do.

I hope to try this out within the next few weeks, probably ear-lie on a Saturday morning. The current record is 5:43:33, and I think that it's not out of the question that I can get it done in under 4 hours.

Sunday, August 05, 2007

Every Larosas

Yes, as mentioned in my last post I have new plans! Every Larosas!!

There are 60 Larosas, mostly in Cincinnati, but also ranging up to Dayton and one (my foe) in Batesville, Indiana.

Here is a map with a tentative route plan. It starts up in Dayton and then ends at the Mt. Healthy Larosa's (which is one of 4 that is open till 1 a.m.) Batesville is a 63 minute drive from Oxford. Normally I'd try to start or end there, but since this has hours of operation to deal with, it's not really possible - since the Batesville Larosa's is open only till 11 a.m. (And opens at 11 a.m. as opposed to some of the other ones which open at 10:30)

So recently I did a test run of part of the route, starting in Landen and heading to Montgomery, Loveland, Goshen, Milford, Mt. Carmel, Amelia, Anderson, Mt. Washington and Mariemont. I had calculated the times between each of the stops through Google Maps. As it stood, to hit all the stops, it would require doing so with only 1 minute 53 seconds of time at each store. That includes all time to find the store, park, get out, order, pay, get back in the car, and hit the road again.

Obviously that's not a whole lot of time, so I wanted to try the route in real time to determine whether or not the times were accurate. If I could pick up a minute or so per store, that could mean the difference between success and failure.

The good news after my real-world route simulation: The times were not that accurate. The bad news: They were not accurate in the wrong direction. Most of them were pretty close - within a minute or so, excepting some traffic problems that (hopefully) wouldn't be repeated on a Saturday, though you never know when you'll have a broken down car on Beechmont Avenue backing up traffic, as I did.

I also missed the Mt. Washington Larosa's, which is on a street with no parking lot or anything. Definitely can't be having any of that.

One positive is that it someone at church told me that you can pay for your order when you make it, which makes sense. But that could definitely pick up some time at each store.

I haven't decided if I'll keep trying for this or move on to something else. Another option would be to just do the ones in the "Cincinnati area" whatever that means. The more I think about it, the more I think that should be defined on a county basis. So even though there are areas of Butler county that I don't consider part of Cincinnati (Middletown / Hamilton / Oxford), that at least is consistent with the Cincinnati / Hamilton / Middletown consolidated statistical area

I will keep you posted as usual

The demise of ECF

I have done some thinking about ECF, and I have decided to put it on hold. The more I thought about having to limit it to 275 just made it definitely lose some "cool factor". But don't worry, I have other plans...