BIKE: observations from home on tri-rail

[alan_drake]

-- Mike Dahmus <mdahmus> wrote:
roughly the second half of this posting to my blog:

http://mdahmus.thebaba.com/blog/archives/000107.html

(I know we talked a bit about Tri-Rail here; see the article for what I 
saw during my 2-week trip home for Xmas).

- MD

One technical adjustment.  AFAIK, the US transit planning rule is that no pax are lost if they have to walk 1/4 mile (~1250' = 1') and there is a progressive loss from 1/4 to 1/2 mile.  Some % will walk .4 mile for example.

The EU rule is that each stop will draw walkers from 500 m away (with no   grey area), except in Iceland where they use 400 m (due to climatic factors :-).

Also, the incomplete double tracking of S. Florida Tri-Rail has improved on-time perfromance significantly (from dismal to OK) BUT has not had much impact on schedules yet (last N-bound train leaves Miami ~7:30 PM, frequency, etc as of June, 04).  On-Time performamce is supposed to be better still (OK to Good ?) with complete doubletracking.

Thus, only a fraction of the benefits of doubletracking are in place.  And there is a well known time delay (couple of years likely) before the public is aware of substantially improved service and adjusts their travel patterns accordingly.  Thus your condemnation of the improved, double tracked Tri-Rail is premature.

Perhaps the transit modeling for improved TriRail is off significantly (those #s look fairly good), perhaps not.  It is several years too early to judge.

Until reality resolves the issue in Austin, one can only go by the transit modeling.  Are the limited #s of pax in the model worth the investment ?  Are there flaws in the model ?

IMHO, that is where the debate should be.

Per my limited understanding, the original Tri-Rail models were off significantly, for the reasons that you stated.  

Have the flaws from that model been corrected in the Austin model ?  (Transit modeling is not a static art).  

It is my understanding that they have been for the Tri-Rail doubletracking model.

Alan