Uncool and unLean taxi batching at Logan Airport

One of the things you learn from Lean is that batch processing is often inefficient and wasteful compared to continuous flow processes.  I have sometimes used this blog to illustrate this phenomenon, like at this bakery and, ironically, even at the food service line of a Lean conference.

I saw another example when I arrived at Logan Airport late at night this past week and was waiting for a taxi.  Here's the scene behind me.  Counting those in front of me, at least 50 people were in line at Terminal B, with more showing up every minute.

Taxis were arriving in groups of three to five as they were sent over from the parking lot at which they are required to wait.  In between, the dispatcher required all passengers to wait behind a chain.  (By the way, note the signage!)

Then the taxis arrived and parked at the curb. 

Only then would the dispatcher allow the batch of three to five passengers groups past the chain.  It would take them about a minute to reach their cabs, load their luggage, and sit inside.  Since the cab at the front of the line was furthest from the chain, its passenger would have the longest distance to walk and take the most time to load. 

Only after that first cab was filled and ready to leave could any of the ones behind it proceed to leave.

In between batches, the dispatcher had nothing to do. He even had time to visit a passing MassPort supervisor driving by in his SUV.

I was about number 40 in line when I arrived at the scene.  It took me over a half hour to get into a cab using this approach.  Each batch had three to five cabs, arriving every three to five minutes.  Loading took a minute for each batch.  I had to wait for eight to ten batches. 

Imagine a simple improvement to this system.  Assuming the taxis have to come over in batches from their parking lot, why not have the dispatcher get people ready to load by standing at pre-marked spots along the sidewalk, ready to load luggage and hop in as soon as a car pulled up.

But imagine a further enhancement.  I asked my cab driver, "Did you have to wait a long time at the remote parking lot before being freed up?"  "No," he said, "only about 15 minutes."  I was stunned that this could be the case on a very busy night, with unmet demand at each terminal.

Apparently the overall dispatcher at the parking lot sends a clump of cabs to each terminal, in sequence, when it is a busy night.  With five terminal locations, every fifth group will show up at any given terminal, always in a batch.  That batch takes time to exit the parking lot, and then the next group moves up the line from their parking places.  What if, instead, the dispatcher at the parking lot sent each single cab, on a continuous basis, to the next terminal in sequence?  A steady flow of individual cabs would arrive at each terminal, separated by less than a minute, to be entered by a waiting passenger at the curb.

I'll let our process flow engineers do the math, but I guarantee this approach would have reduced waiting times considerably, both for cab drivers trying to make a living by getting as many trips as possible and for passengers trying to get home as quickly as possible.