Construction Today - Volume 16, Issue 2 - 63
This is a resulting scheduling in ALICE that matches the logic and duration
of tasks from a contractor's schedule. Notice the resource utilization graph,
with the peaks and valleys - there is a mismatch of crews.
This is the same schedule rules as in the photo above, but with an optimal crew
mix run in ALICE. The schedule is resequenced with this more efficient crew mix,
creating a faster and less costly project timeline with a more gradual curve.
product did not go into full release until
Applying AI to construction scheduling
required a number of breakthroughs before
it was ready to use on a real project. The first
was figuring out the building sequence that
the schedule should follow. ALICE needed
to break down the individual components
of the building to understand the scope of
the project. Fortunately, existing building
information modeling (BIM) tools already
contained most of the data that ALICE
required, such as the location, quantity and
types of materials that were going into an
individual project. "There's a lot of useful
information encoded in BIM," Morkos says.
Beyond the information available in BIM,
the software understands the more subtle
complexities of scheduling. For instance, it
is aware that it takes time to move a crane
from one location on the job site to another,
and it accounts for those realities as it produces the project timeline.
By accessing the BIM, ALICE could determine the pieces it needed to build a project,
but it still had to figure out everything that
went into assembling those pieces. That's
where Morkos hit on his other big innovation: parametric scheduling.
Instead of telling ALICE how to build
every individual column, window or girder,
parametric scheduling uses a "Recipe" system. Engineers only need to tell ALICE how
to build a single column and the program
will then extrapolate that process to all oth-
VOLUME 16, ISSUE 2 CONSTRUCTION-TODAY.COM