So I'm glad to see so many people here.

I see her seem to design student future
senior design students people of

like family friends
clients company clients.

You know if you see everybody.

For people who don't know who might not be

familiar as all the students
in the room senior design.

We started out this fall semester we had
twenty teams doing just a huge range of

engineering and engineering related
projects dealing with everything from

global supply chains all the way
to running things here on campus

and there are a lot of really nice
projects the three of the three best

overall from as
an engineering standpoint and

from the standpoint of their written
reports and presentations and

professionalism are the three
finalists will be presenting tonight.

So in alphabetical order
will see you in the team for

a G. energy and a team from Mars but
we won't see them in alphabetical order

because all three teams actually had
to sign nondisclosure agreements

to the three prohibit them
from being videotaped and so

the one team that that will be
videotaped and I interviewed I'll go to

the people doing video editing on the two
other presentations with a fairly

So they'll go first followed by eight X.
and then Mars and so I'm going to ask

each of those teams advisors just before
the team presents to get up and say a few

words about what it was like this semester
working with that team and thirty give you

a flavor of what that boy in the team
will get to tell you all about.

About the very things that they do so
the first team to present.

Like I said will be the team
that worked with G.E.

Energy and
professor you are the main third party.

Yeah I think only have you joy and

want with this team and
if you thought a lot about how away.

They were that you got something after
all that's not even you get more

think he's done the right thing.

I think I've got on really well hard
left and I mean Alpha yesterday.

I'm still of the old law now.

Yeah yeah but what would you guys

like a good evening and welcome everyone.

Thank you for joining us.

Thank you Dr Lee for this wonderful
introduction or a problem this called Arts

allocation Gorgie energy by my name
is there is a Money team over there

are these group one and both
are committed slightly nervous for me and

two of our members can I join us today.

But their names are most in the county and
so we decide.

We've been working with the energy
over the course of the semester and

we've been working with our contact and
real friends over the team and

yes we would like to thank them for all
the cooperation and all the work that they

put into this project and we'd also like
to thank our back the adviser Dr yelling

match for all this help support and
guidance throughout the semester.

We're going to get started as a kind
of give you a preview of what's to come

in the next few slides will give you
a little bit of background of Baiji energy

how things were buried now the goal of
our project was to pull the first was to

reduce the number of purchases the G.S.T.
makes in order to fulfill its contractual

obligations and the second was to
reduce the residual like a part.

Or the absence of the use in maintaining
these drugs off again will be using

some of the used throughout the battle
throughout the project the most notable of

which are the messenger programming as
well as the nonlinear programming and

the final look over G. energy that we
provided was a partnering tool and

the project valueless that made it
annually at eight point seven dollars So

that said we've been working with the
energy the energy the leading provider in

products and services in the energy
industry to provide everything from gas or

by steam turbines wind turbines generators
will be focusing on the gas turbines.

There are clearly more than six thousand
gas turbines being operated worldwide that

are maintained by G.E. Energy each of
these are maintained by something called

the contractual service agreements these
are the business contracts that G. energy

holds with its customers where G. energy
meetings these contracts were and will be

focusing on the hundred sixty one gas
turbines in the North American portfolio.

So this is a typical gas turbine that has
three major components the compressor

the combustion chamber in the middle and
the high gas prices on the right side.

Each of these components have specific
parts associated with them will be working

with the parts in the high gas prices
actually three of the nine parts that we

will be working with our pockets and
over the drought's each of the parts that

I mentioned before have two different
types of lifetimes the one is an hour and

the second is it starts Byard hours
the number of continues hours of operation

their part undergoes this was anywhere
between oxidation create corrosion and

strain on the part the next is a fire
starts and this is the number of times

a part of the used to start unit and
this causes thermal for

tea because a part heats up and cools down
very rapidly as a unit as he started down.

And we'll talk a little bit about the gas
turbines each gas turbine has one

of three Operation profiles either base
load peaking or sickly in a base load gas

turbine operating from the unit has
started up infrequently very few times but

it's run for a long time when
it is started in a peaking

is the exact opposite of the based on
where you start a very frequently but

it runs for shorter lengths of
time when it is started and

a simple operation profile is
a combination of the above two.

This is the energy's current system where
you have ANY GIVEN turbine turbine here.

Barger removed from the turbine
during certain maintenance intervals

to repair it was they're
sent to the repair shop

where either beam either reusable
in that specific turbine.

And if not they're scrap
what our project looks to do

is take these parts interchange
between different turbines

throughout the life of the part
thereby extending the portion of life.

That's used during during the contract and

finally when the maximum portion of
the life is going to use the participants

craft a very important
concept to understand

in order to understand our project is
this concept of tours and outages.

So this timeline will
representative of a contract.

We'll have started in the beginning
in this case two thousand

at a future point in time we
will have its first outage and

outages with the unit shutdown
you remove the parts.

So you can do the repair shop and replace
them with new parts or spare parts.

The time.
In between these two outages it is known

as a torque and
during this tour a part can be run for

a certain number of hours and

a certain number of starts this process
repeats itself throughout the life of

the contract finally ending in this case
in two thousand and twenty two you can see

that there is a little bit of variability
in the number of hours in starts here.

However the base the basic operation
profile remains the same and

what important assumption that we made
in this project is that these outages or

these are parables are pre-determined
based on energy demand that the energy use

another bit of the will kind of talk about
who start to talk about on the methodology

to be used in tackling this problem and
killing so to approach the problem.

Can we want to compare that to its current
system with our product type system so

we used a network so to model time system
as you can see here we have two things one

terabyte one thing click and two peaking
turbines we lay out the Alexanders for

each of these turbines going
forward in time to write and

it's not here represents our necks and
he'd answer represents a torque now

at each out and
wax can either be purchased by scrapped.

So the article as is here has
been practicing or scrapping.

Parts.

Also parts have to be held spare
on the engine as so the car arcs

show the spare parts in the current system
we can see that a part would be used for

the first tour and they were being
repaired and health care for the second

tour would be running again the third
tour and then it would be held spare for

the fourth tour a counterpart
a turbine would do and exact opposite

in it's important to note here that no
current system there is no good quality of

sharing parts on them different turbines
to enable this system to parts.

We need modeled our proposed
system as a network

where we connected to
nodes to look like this.

Now a park would be used in a based
on turbine and it would and

a syphon of peaking turbine worse and
worse a park in stark in size and

I'm thinking turbine and
siphon a bit of the same not so

be used in this liquid turbines we
take this never thought process and

we designed the car sharing tool
the spark sharing to take certain and

well such as the gas turbine
operation profiles the Arab states

of the transaction and crime and
ensuring that she has for each time.

By the end not so done and turn it into an
optimization tool which is dismissed and

her for being jumped out of the stool
is to minimize the new card

purchases as long as there
is into life the part that's

up to my station tool uses a software
called Express and it provides us was

where specific will tell exactly when and
where to share parts.

It tells us there was a dual
life time the farts and

I guess is the cost associated with
each system specifically the purchasing

cost repairs transportation and
folding costs.

Not now turn over the money and he will go
more into the details of delays issue too.

I think you have the so before I go more
into the specific question this morning.

No doubt we considered three
different ject of functions for

the first one it was obvious one
was to minimize purchasing cost and

by doing this we had quick run times for
sharing tool but

however this does not address
the residual life going to the project.

The second subjective function to take
in consideration of residual life.

We have minimized the purchasing cost and
was a delight at the same time and

this directly addresses though the reject
is yet but however it is very complex and

takes a very long time to solve and so
to compensate for this we have made this

into as two step process where we first
knew I was the first in cost of the part

of the parts and then based on the number
purchased made from that objective.

We have minimized residual life.

And so using this two step process
here you can see the the next slide

you see the formulation of our purchasing
tool the decision here is X I J whether or

not for I uses it is a binary
decision decision variable.

So it's zero if that part is not used and
as one of the dozen you said and

the objectives here or again the step
one to minimize the person cost and

then step two based on the number of
purchases made your group minimized

the residual length of those parts and

there are many constraints that we've
formulated into this perjuring tool but

the main ones to know are the part like
time constraints mean that part cannot

be used past their design less This
occasions in our stars flow constraints

mean that whichever birds go into
an outage or node must come out and

pass the constraints meaning that every
unit was kind of are running it at all

times even after splitting of rejected
budget in two steps to reduce complexity.

We're still faced with this problem.

You can see here by this graph for
the number of units first the runtime of

the person and all that when we have
to run ten units ensure perjuring.

Well it takes around fifteen minutes so
if we go all the way to fifteen

units of our sharing tool it takes
a costly ten hours to solve and

if we put ALL hundred sixty
one of our turbines in here.

It would take extremely long time
solve and it's because you create so

many constraints in decision variables you
can see that over three hundred seventy

three thousand constraints will be made
and twenty six missing twenty six million

decision variables would have to solve to
solve the problem to solve the problem

of complexity and we have decided to take
a hundred sixty one here this is split

them up in the smaller groups which then
we would put in the workload individually.

We came up with three different proving
methods and Mark will talk more about.

Thank you much.

So he said our group of three
agreements first was a greedy

The second was a clustering and third was
and ration first of all of the greedy our

greedy album was an innovative
process which goes through and

selects the best of units to share parts
with removes those units from the pool

of turbines available and then repeats
this process over and over again.

So as you can see here on the map the
optimal group to select is removed from

the pool and this process is the repeat of
eventually grouping units will no longer

add value to with sharing parts so
that is when we start running this

process is not very flexible because the
groups are set and as units are added and

removed he has to rerun the greedier
the next method that we looked at was

the clustering analysis this method look
to treat similar turbines as one unit.

So as you can see here here's a graph
of the different cost frames for

all the different terms.

So this is the first step in the cluster
analysis after we found the clusters

within a numerate all possible
combinations of clusters and

one another expressed program which
simultaneously finds a better best group

of units to share share parts with
our knowing the N. ratio the N.

ratio of a turbine is computed by taking
the ratio of the number of hours.

The number of stars run
to the turbine per tour.

We then rank from.

Highest to lowest based on the and ratio
of turbines and paired the top five in

the bottom five units and worked our way
in works until all units were grouped so

as you can see here on this chart in the
first debate the most baseload and most

people in units were selected and grouped
together to put into the network flow

this process was repeated through all the
different ratios on this until evolving.

Now we were comparing our three
different methods we can see both of you

have disadvantages in terms of flexibility
complex ability in the size of groups

in the flow for
the three different things.

Most importantly what our group of
that was the annual savings just for

the Stage two buckets using the cost of
capital thirteen percent to see the end

ratio has the highest in the savings
of two point six million dollars

Now that we have a network flow which
can model both the current and or

oppose system and also select our greedy
or grouping that are going to use it.

We can now align our system.

So as you can see here.

Here's an example of six
turbines in the current system

parts are currently not shared between
the different turbines and G.E.

interview incur a cost of just under eight
million dollars in our proposed history.

You can see the very complex exchange
of parts through of different turbines

throughout their C.S.A. using the system.

G. would incur cost of just two million
dollars giving a savings of one point five

million dollars This process is then
repeated for all the different parts and

all the different groupings of
turbines in order to obtain a final

I'm out here to go back over to
reduce will go over to little and

finally as part of the deliverable that
we handed to G.E. Energy was two fold.

The first was if I did a consolidation to
all this takes data from all different

sheets in Jena Geez I can just and this is
their data warehouse and consulting is in

a don't what Excel sheet that's then used
in the car sharing tool that we mentioned

before the models of models the current
system as well as the proposed system in

order to value our project we compare the
cost of the current system to the cost of

the proposed system and we did this in the
same way that Mark mentioned report back.

Bearing the cost of the current system
by running it through the model and

the same way for

the proposed system this start here shows
the number of savings that G.E. Energy

would have over the next eleven years here
the model those cash flows the net present

value of these cash flows is almost forty
four million dollars at a cost of capital

of thirteen percent which comes out to
anyone quibbling the five point seven

billion dollars This is the first part
of our of the value of our products.

The second part comes from
the reduction in residual bottom parts

these two charges charges here
show these two charge here

show the residual up of the parts in
the current system modeled in red and

the proposed system modeled in
green Using a man with the test.

We were able to show that a ninety
five percent confidence level that

these two values were
statistically significant.

This means there's nothing you are using
a higher portion of each part of life.

Most likely to start with and
this also adds

an additional three million dollars to our
value and the implementation of a system.

G. energy will incur costs there are three
major categories of cost for this system.

The first is the purchase of optimization
software to buy those expressed cost

twenty five thousand
dollars per commercial C So

any cost any license that you were
purchased with they would incur a cost.

Accordingly in the second largest one
is a contract renegotiation cost.

Are the customers must be willing to
participate in a car sharing system and

any discounts or concessions opportunity
energy to his customers could be looked at

as lost revenue and actually cost and
third and final is the G.E.

employee training cost of this
these would be incurred as D. and

as employees are trained in the
manipulation and execution of the new and

the current systems So
to summarize our project.

The objective was to reduce
the number purchased of the G.E.

Energy makes annually as well as
the residual of the parts that it uses to

maintain its contractual obligations we
will do this limited in your programming

in our network model as well
as non-linear programming and

other methods in our grouping model the
liberal was the partnering tool and our

estimated value of this project this eight
point seven billion dollars annually.

This concludes our presentation.

We hope you enjoyed listening to it
as much as we enjoyed working on it.

First it truly was an imagination
at work and at this point.

Open the door of any questions.

The audience rules

so what

is your back to your values.

I just wonder why that is so vague in
two thousand and twenty twenty five.

Absolutely.

There's two reasons for
this actually the first is that our

model does not take into account
the time value of money.

So it doesn't know that it's better to
believe the purchase of art as opposed to

buying it right now.

So that's why you may see
a lot of value from here here.

And secondly you see the sharp decline
because many contracts expire as

time moves forward.

So this contract expired the opportunity
to see exchange parts and

sure parts among different
turbines also expire with them.

So the value dropped significantly

just

to comment on but
the fact that that is true he said but

we also expect but it's not part of
the model to renew contracts or quarter.

New contracts as we go.

So we're hoping to see
the two thousand and

twelve type number continue as we
continue to renew the portfolio.

Thank you Michel.

Joe.
There is

one fraction of the thirty five million
where you don't have to Georgia

was going to go through all this
to renegotiate these are really do

you think you know where

you are just going very much.