On the afternoon
and evening of November 2, a trial run of the Rapid Transit
Challenge route was conducted. The partial route ran from
Station X to Station Y using 21 trains and visiting 134 stations,
representing the first 25%-30% of the planned route. (Note:
Station names are omitted as our route is valuable intellectual
property!)
The main goal
was to test whether the model worked for estimating the time it
would take for the route and the results were very
encouraging. After six and a half hours, the actual time was
exactly 14 seconds faster than the model projected. The status
during the trip ranged from as much as 11.5 minutes behind schedule
to 1.5 minutes ahead. The actual running time (on the train)
was slower than the model (300 minutes compared to 292.5), but there
was less waiting and transfer time than expected (90 minutes
compared to 98).
Over a longer
period of a time, there is a greater risk of falling behind
schedule, especially since missing a connection overnight could cost
as much as 20 minutes (the worst delay was about 10 minutes at Z
Station when the train was held outside the station while another
train left first). Also, a larger group traveling together will
likely be slower. Even with that, we're very confident about
the model. We could very realistically beat the record by over
two hours.
