Good.
The second finals of the night were worth
having the board of the car
in their records or another.
There's no fear of Muslims
if you know the U.S.
based on the
right.
Believe me.
Everyone thank you for
your time walking to team to board his
final senior design presentation by means
of an astronomy.
I'm joined by true to tell of your tell
each and we'll be every presenters today
we're also going by the rest of our
team Jason Yet you'll be talking about.
There's an emergent behavior.
I just figured we'd like
to take this opportunity to
thank our faculty advisor Dr Gupta for
all his help.
Throughout the semester.
Thank you.
Lot.
So let's dive right in their product.
What your body doesn't remember but
it is a Japanese company that has
a headquarters in Gainesville Georgia.
They specialize in manufacturing
agricultural natural equipment.
So in order to manufacture this equipment
they need parts from suppliers of
all across the United States.
So I think you see on this map they have
suppliers look at across the United States
and the different colors represent
the different modes of transportation and
suppliers so T.L. represents a mode
of transportation a Ford truck load.
So any supplier that ship with a full
truckload of would be green L.T.L.
represents less than a full truckload and
unassigned represents a supplier remoted
transportation hasn't
been determined before.
So at the moment your body has for
information for three months in advance
what that means is it gives them a lot of
flexibility when they're trying to decide
this is a large shipment what motivation
or the supplier schedule because
that data for three months and they have
approximately thirty shipments coming in.
And these shipments can be
classified into two different areas.
So there's a direct route and
there's a consolidated So
as you can see from this example I direct
route would be any route the comes
directly from the supplier
to supplier in this case and
would go directly to Cuba and get for
the same same applies to supplier B.
shipment directly from
supplier be coming into Cuba.
Now the difference between this and
a consolidated right would be the shipment
starts a supplier the truck travels over
to supplier he once it gets to be takes
a shipment from B. printed onto the same
truck and then carries on from B. to Cuba.
So that would be consolidated So
the problem there for our team
type of the beginning of this project
was redox in the transportation cost.
So we noticed two areas for
improvement two areas where we
could realize improvement and
be serious with one that the mode of
transportation at the moment is not a
value that consists of regular you know so
Viber transportation again I mean that's a
truckload full throttle so that was one of
the areas where we we thought improvement
could be our realized and the second thing
was the rights currently consolidated but
are being done so madly So
we felt that if we made this system more
systematic and used a program to do it.
We could realize a lot more
savings that they are doing so.
Currently so I deliver walls.
Based on the problem where we're going to
give our clients about open source tool
that assigns a transportation modes.
But does it on a weekly basis.
So like I mentioned earlier they
don't assign transportation modes
regularly enough.
So our program is going to do
this on a weekly basis and
tell them where the supplier should
be L.T.L. T.L. so and so forth.
Consolidation right so our program will
also tell our client about what route.
Each shipping should take.
So whether it should be direct or
whether it should be consolidated so
we'll tell them that on a weekly
basis to value guidance based
on running our program on a start.
Into our program.
There's a cost savings of the.
That would point three
percent on a weekly basis.
So this is an example of the order
information that a program takes in.
So I get this order information is
currently available to our clients again
it will be convenient for a time to use
our program because they're already have
all this data because it is new people and
so the different attributes so
we look at our supplier ID which is again
a unique identifier for use of the here.
The zip code is what we used to determine
where the location of the supplier is
the weight is again the ship
the shipping weight in pounds force but
is it is a unit that our clients who are
use to distinguish between the Ferrari and
trucks a specific shipment could only
have like say five corresponds out of
the total twenty six four spots from here
it is again the volume of each ship and
measured in meters Q. and
A delivery date is the date by which each
shipment needs to arrive about Capote I'm
not going to pass it on to Kelly
to talk about the methodology.
But if you are not.
So your approach to our problem is can
be separated into three separate phases.
Our first base is Route generation in
this space we generate every combination
of possible routes for
to get a problem during this phase.
It's possible to generate over a billion
different routes after this phase and
move on to our second phase
which is Route production.
There are just basically apply different
constraints that could better supply
the bus along with natural
constraints as truck capacity and
we eliminate these routes down
to orders of about the next
we do move on to our third phase which is
our opposition model at this point will
run are a couple of thousand routes and
put them into the oxidation model and
this model will generate our solution for
a given week so
we have an example of a chart of
course by using across a week for
suppliers the maximum out for
spots on a truck is twenty six.
So there's three different possible
methods of consolidating around
our first that it is.
Consolidating on a single day across
different suppliers our second method
is consolidating across different days
on a single supplier now in this battle
when you consolidate
across different days.
You assume that the earliest date
will be when the shipment goes in and
are therefore consolidation is actually
a combination of the previous two
where we consolidate across suppliers and
days.
So now it is on into around Option Base.
We have several different constraints and
the one you see here is one
of geographic constraints.
So we went through all the suppliers.
I proposed and
separated them into ten different groups.
The reason why we separated this is
if we try running to two hundred
suppliers in our problem.
It will end up taking a few days
to actually broaden the model
by separating into groups we actually
cut down from a couple of days of proton
down to two hours at maximal.
So we think we base these groups off of
the locations relative to other suppliers
along with locations
relative to major highways.
So if you direct your attention
to the north east of the U.S.
in the group that's five.
You can actually see that
this is an example of how
all the suppliers run across the major
highway that runs down the East Coast.
Our second Geographic constraint
is one of route direction.
So we ensure that any route
that we run if it happens to
be North Dakota it must head
southbound and any route South Dakota.
That's had northbound we didn't
want into our capacity constraints.
There's a limited amount of space that we
can we have to truck to be our supplies.
So the three concerting
factors are weight.
All you an area which is four slots
and now we do get into
our cost constraints.
So there are two different
methods of calculating cost for.
The voters run time.
So we have a lasting trouble or
cost that we can calculate from taking
the four inputs the origins of code
the destinations of the weight of a given
shipment and the date that the shipment
must arrive after both this is
then put into a website that could
go to users to receive estimates of what
the L.T.L. shipment would be at that time
the second way of calculating cost
is there are a truckload costs.
This is based off of the distances
of the routes right.
We take but the we take a mileage
rate that voter has that and
we multiplied to the distance along with
adding a seventy five dollars stoppage B.
for each additional stop made on
the route along with multiplying it by
the National Fuel cost.
Now now four routes of only one
supplier will calculate both the L.T.L.
routes and cost and the cost for
routes of more than one supplier
will calculate only the truck costs.
So in order to use this as a constraint
will look at each supplier on each
individual day and take all the routes
that touch that supplier on that day and
eliminate all routes greater than
the average and then after this.
We are able to eliminate down
from an order of magnitude
of a billion routes down to six
thousand to five thousand pounds.
Now passed off to Drew for the next steps.
So we started with more than billion
rounds and then removed the rads
was rather not visible to our I have about
six thousand routes these six thousand
routes I've taken them have that input or
optimization model along with their cost
the optimization model outputs
the total transportation cost.
This is minimized at this point and
the optimal set of grounds.
When it says yes then it should be
chosen the next I am the optimist.
Equation refers to the rug and the C.I.
reference to the cost of
the eyebrows the optimization
equation price to minimize the term
transportation costs as you can see here
and there's still by ensuring that each
shipment has been picked up as we
think that only once the decision will
extract is binary meaning that it's
either one episode or not is chosen or
zero and their route is not chosen
the output of the optimization model.
Now needs to be checked for
compatibility of routes amongst themselves
a pair of trousers said
to be incompatible.
If he leads to other teams in the sequence
of I Will of shipments and K.M.'s.
So take for example Subway has any
shipping since you have a Monday Tuesday
and Wednesday.
Do you Route one.
So I beez Westminster Venice for sure and
one Monday which is incompatible with
Route two because these wins this event as
a landing before beast used to ship and
now this is an example of a compatible
routes and to eliminate this from
happening and to ensure the sequence
of what I love to menses maintain we
generate our program generates constraints
dynamically to a way this from happening.
So in this case I constrain will be
generated and the model will be redone
with this constraint to make sure these
two routes are not chosen at the same time
to summarise the optimization process we
take a set of inputs which is made up of
feasible routes and their corresponding
cost and these actors the interest of
the optimization model which outputs
the optimal routes along with it all
minimised past the output of
the optimization model vendors check for
compatibility of routes of the routes and
compatible with each other they constrains
that dynamically generated and then
wonders really run with these constraints.
Now this continues to happen.
I'm till all the routes
are compatible with each other and
this point the model
is there to be solved.
Here's an example of how our model
consolidates shipments across days and
suppliers for four years an example of for
showing that for four different suppliers
A.B.C. and the cell supplier be
in seeing how their shipments for
Monday picked up by a T.L. product and
as yet drug respectively the green
frog represents the T.L. through whereas
the red for a presence at the end for
on Monday itself the struck from software
B. then continues as a black day and
then picks up its Mondays Tuesdays and
Wednesdays shipment they are going to
continue on its way to
campaign on Wednesday so
glad the ships it's Wednesday
shipment as a want to transportation
to came in short on this graph and
the cost for the three weeks up there
that we received from my client and
these are the orders of course which
the arc of the thing with YOU HAD TO has
happened for I find in the past three
weeks when I want to was run for
the same three weeks the fun pass
savings were eleven person which
is approximately fourteen thousand for
week one twenty three per cent for
week two which is approximately
thirty thousand and
two percent for week three Vince's
two thousand approximately for
deliverables will be given I primed and
I sell macro around generation program
which together solves
the whole model the Excel
macro the whites the data into four
different into different groupings.
And these groupings our initial looping
through is fairly mentioned earlier which
is the way of the suppliers into according
to their location and your graphic goal
location the ragin research program.
Finds the optimal routes
along with a total.
Minimize cost and
it does still by using an optimization
over which is that we use currently So
while there are hundreds of program
is used to create it using Python and
the optimization model that we use is
created using the little bit our open
source software is so our fans will not
have to incur any implementation process
both of these are really a local and
did currently have excels on the computers
to summarize the pope
process we start with
to tackle the problem of using
the domestic Ambassador we start with
generating all feasible routes and then
who want to renew using those routes Mr
and I have visible through certain certain
certain actions the the visible
routes are then
taken as inputs to the optimization model
with the outputs the total transportation
cost along with the said optimal
routes West should be chosen.
I promise I found savings of eleven
with the person for a week and.
This is approximately five hundred
thousand a year after I and has expressed
their interest in implementing a program
that's very measured from them and
open to any questions
that you guys may thank
you so
the question was when we refer to
their regenerated constraints how to
do change only this is
an often strange solve or so.
What happens is the model does run again
the opposition models run
time is less than a second.
Once the routes are generated so
we find that the running it isn't
taxing on any resources or time.
So it's rerun
with the police were
called pulling the one for
example though where we were right
on the report of the Let It Go.
So the question was how do we take
all the costs into account for
our for our problem we did not
take only costs into account.
We spoke of codes talk about
storage space usage for
a high point cost and from what they told
us they do have extra storage space for
his overflow of inventory and
as far as into the pipeline cost.
We've found it to be minimally taxing
since the the time line is only
across two very days that
the inventories at just
three boxes
of human development go on.
So the question was how do we justify to
constraints that we have applied as and
you can take the total
volume of the shipment and
put it into told Boy that's a good
question because all obviously
all the shipments aren't liquid so
you can't just pour it all in put truck.
We actually along with our
four spots constraints and
our point constraints and
the weight constraints.
It creates you can essentially
look at area you can.
To look at four spots and
You Tube it combining together
a way of forming specific boxes that you
can kind of just stack into the truck.
So it's not just a calculation of
the volume and going flat out it's kind of
the integration of those three
constraints those items are question.
They're just
one users were far less
were they were wrong
you repeat the question I think you
said why are things where you go out.
So it doesn't require twenty six.
There are spots.
It's a set maximum so
we can't exceed the number of four spots.
If we ever did that it would
just cost them not to be able
to fit all the shipment onto a truck.
So it's in or goal of it's it's pretty
integrated into our problem that
we need to be constraints and
be sure that's along the lines of
where you're asking or not you know.
So.
Are you asking about supplier delivery.
They're run times in the one zero
zero zero zero zero zero zero zero.
I think I can answer that.
So folks forgive let's say the shipment
was originally scheduled for
Tuesday and then I consolidation would
do only what only considered Tuesday or
immediately or
it never considers a later date.
So the shipment due on Thursday.
First of many and there is is going to
consider when they've used them and
then there is their delivery
date possible delays and
I'm proud of that if I'm
going to consider that
while Codey uses industry standard
trucks and especially for
sponsors essentially term that only
the boat uses in their practice
it doesn't actually mean anything.
You talk to any other company it's
essentially I believe the actual number
for what a horse farm represented was
a four foot by four foot space and so
after you split down those times.
There's no truck that
exhibits the results.
It's not my kind
of background experience of the person
who's using this to look over here and
what kind of training they need to
be effective in getting a solution.
So our product essentially has made it so
that whoever has to run the program has to
put as little effort into it as possible.
The only points where
they would need to apply.
That's a bit of brainpower is
the changing of the cost multiply.
Or at least that previously
we took the average
of the cost of all the routes touching
a certain suppliers and day but
to go further into that we
actually also multiply it with
a archery multiplier that we set in
the beginning of our user interface and
this number has to be adjusted the pending
on the number of combinations of routes
created and this of course depends on
several different factors as in the size
of the shipments the number of suppliers
there are shipping during his day.
So this most part is to just constantly.
So in terms of their running it to receive
optimal efficiency is
a little bit of a move around
as Actually there is an errant out how if
you're too low and there's an errant out
you're too high so that number just needs
to change and it will still present.
So is the composer or from using it here
are they are they a warehouse where her
using this to the person who we
believe to be using it is delicious.
Or is the logistics planner ants.
But then again any worker can use
it as long as they have a computer.
Through a.