So the final say.

Today the world the World Food Program.

All

working with the program and

a larger

world are very excited to be able to work

with their own I think

my favorite headquarters was when they
were more than one thousand miles away.

Do you believe

high little boys

live with it and make their

own their

first song you were doing right.

I mean there was also

something that

I was

really honest and

all of the people like you get

something and

I you know and

they also were like you never project.

So

I'm more into Live with
passion optimization project.

So today we'll begin with
an introductory video that discusses

the problems we identified you expect from

this company but just

got

the video.

Programs operations go from donations
to procurement to shipping commodities

to Overland transportation to country
offices where they're distributed

to different projects and served through
different non-governmental organisations.

Now let's take a look at the problems
we identified in this existing system.

The first one is variability in donations
this variability leads to an inability to

plan supply chain actions the procurement
and transportation activities are reactive

to the ability of donations and
this leads to more resource utilization.

This is a graph showing donations over
every month over a couple of years

where we see several spikes

that are international crises we can see
even in periods without crises that there

is definitely variability in
the donations that are there.

These variabilities would not be a serious
problem if one program had an agile

supply chain.

In short the times.

Unfortunately they also have long variable
be times due partially to the fact that

they shipped over long distances and
partially to their processes.

This is an inability to react quickly
when donations become available.

Therefore the variability indications
combine of these the times become

a serious problem here we see the times
in the standard deviations for

the top twenty most common use routes
as you can see the lead times average

around one hundred twenty five days in
the standard deviations in lighter blue.

Are usually a significant
portion of that lead time.

Finally we found that
the World Food Program has no

standardized inventory management.

This means that they get unpredictable
we sized orders at a medical intervals

combined with their challenges
in supply chain operations.

This is a serious issue for
their operations this

graph shows the stock levels in Malawi
three commodities cereals pulses of oil.

So as a function of the target one
hundred percent Internet as you can see

there are definitely periods
of extremely high stocks but

combined with periods where they are well
below the need to man in a month.

OK Let's now see how we
decided to solve this problem.

We first developed an inventory management
tool we needed an implementation

distorting minimizes
going to the office and

then three cost meeting
targets service levels and

it just helps to standardize the military
policy throughout the system and

next we develop a supply chain of
the relational model to redesign and

optimize their W.S.P.
supply chain by looking at

bridges ition in the most which
are distribution centers that

are strategically located through that as
well as the man in the valley time and

three offices and also looking at the use
of advance funding which is a mechanism

to borrow against forecasted
contributions to help minimize to have

variability in donations that our bank and
just model have done they come in and

terms that you can just
have the snap up better

than three management as this graph shows
this and entering into these go back but

this is another country office and country
project that part of the supply chain

have because of the variability
in lead times Amanda and so

many other constraints that have been
dispersed them it was very challenging to

find a model that could
be applied throughout.

We ended up developing what was never
in doubt and here our model then

developed seven different models and
you were dealing with W.F.

and this was I decided on because if
you are a this is you're not going

to be able to understand what's going on
and one minute in reality but it's going

to LAX another to capture all those
challenges and cost reading in the system.

Some of the inputs it takes
are demand information

lead time as well as cost that
he speaks about a quarter in

Unocal holding down the best
running out of mine and once again.

With all the required
endless It tells them

what how who are much to order
when to reorder what the safety.

Stocks are say several service levels
as well as overall annual cost

of more minimizes for

growth on costs and captures all the buns
range because they then buy in the system.

So just some commodities have minimum
order when it is required in order for

an order to be placed little orders
have to be ordered in batches.

And I don't mourn thing was that the
headquarters wanted to be able to specify

a part of the service level to meet
such as a proportion of psychosis or

a certain fraction of demand.

So without them.

Our time and people react with us and

this is just and what happened this year
was because of the way tomorrow morning

says that headquarters have to enters
learned that depends mation goes

the country officers are the ones that
want to mention the abuse in this forum

and they're non-technical so they enter a
lot of information in the background such

as specifying what model to use what
the fraction of various they want as well

as the expense of ordering as we see here
and this is a screen shot of the wall.

Once they enter all the required
information in the background.

They just act should instead be
put in front of the office and

that are going to be officers.

We just use a graphical user
interface that we develop.

So we don't have to worry about
everything else that's going on and

we'll actually go ahead and
demonstrate that for

you if you really believe

the city

is about you need to have a go.

So this is maybe what that leaving the
country of Susie and just really began.

And here is.

He enters what the project number of this
dealing with the particle commodity is

one of the stock price for a commodity or
a metric ton and where he's buying it from

and the best patient one for the particle
commodity and then his best buys to

her just like we just specify what office
there is processing the information so

we just decided to get
an inventory boss people serials.

And it's sort of like soap in this
is just to confirm that you enter

correct information and once that's
done he enters demand information

either out of an average monthly demand or
as forecast up to six months.

That's what he's doing now and he decided
way to forecast and on this here if you

can specify whether there is a minimum or
a point to meet and you can do it

which we just cited that there is
an over one thousand metric ton minimum.

So once he does this he could get order to
bottom find stuff appropriate information

that was already stated by headquarters
along with them may be good and

against the actual policy but it is as
you can see it summarizes them books and

bells them in the name and
it is from lot of people.

These were thousand five hundred metric
tons of them and repositioned reaches

eleven o nine and three times and
for headquarters we gave them and

I was just tools so that once they get
these really big and big on start level

stimulation to make sure that this then
Country policy is actually something that

will work if employment of Quest you're
seeing here is in the lines you're

seeing and then reposition over into real
period where the stock levels over a three

year period and a red line just shows the
demand and as we can see I'm just graph.

If this pharmacy that we saw in the
results were implemented we would be over

demand for this part of it and to make
sure that this will want to begin to

simulate again and you will generate
a new simulation based on the demand and

leave them you enter along with
them in three passes press.

To show you that it will look done and
not only aside from the results.

Next we are OK We

decided to determine the results of what
will happen if they invent three more

were used to do that we did
then was read by a volume

in the system as well as selected
then random countries and

we did this really best stimulation just
to make sure that we're not using one

realisation of the demand of our
information we obtain from them.

So we use country commodity
combinations as well as historical for

just ordering dates and
generated randomly times from nearly and

only by using constant demand and

once we have all this information we're
able to compare how they're doing.

We developed would have performed
under the same situations.

So we did a comparison and extrapolated
to savings of going to the entire system.

It's important to note that these savings
were not excessive because these twenty

countries already represented about
fifty four percent of the value W.

of the moves every And

once this was done we realized savings
in cereal splendid foods Falls also and

this one news item story about
seventy six million dollars a year and

the change of service level of negative
point one eight through just under

discrete events simulation we ran that
actually works because it's about

almost the same as a service level of
the actual system experience that year and

once we translated to seventy six
million dollars in actual beneficiaries

reach you see you four million four
hundred fifty nine thousand reach and

subtract in that from the negative one one
hundred percent translated into people so

hundred fifty three thousand give
us a net of beneficiaries reach

four million three hundred six
thousand and fifty which corresponds.

A net change in terms of having my purse.

That was and so I will talk a little bit
about our supply chain optimisation model

we model primarily the W.A.P. network from
Preacherman to country out of this but

then the model takes into account
the demand of the country level and

donations go to those parameters.

It's able to model the debt we
have the system over time so

the time we spend in never one model and

it was a billion to examine in
the pre-positioning So it does this so

the location makes it an extended to
program the objective function minimizes

total cost to meet demand given a penalty
cost for and that the men along with

other colleagues such as predicament are
purchasing cars the transportation cost

holding the costs of inventory any
borrowing costs between countries and

warehousing costs both its them
variable of this country warehouses and

depots and it's important for

the borrowing in our model too
because the problem was so large.

We actually had to aggregate countries
into somewhat larger regions which

took out most borrowing costs because the
borrowing would be with in these regions

so the model can account for
these it represents

the constraints of W.P. operations such
as donations but with directed and

non-directed donations were directed
to go straight to a country and

non-directed donations start out
in a pool like a whole bunch and

then the model actually optimally locates
these donations based on the objective

function along with this it takes into
account capacity of transportation

commodity purchases of how much you can
purchase at one location at a given point

and any depot in warehouse
inventory capacity constraints.

So from this model we did some
scenario analysis we began.

By the optimization with their current
network their current system and

we use this as somewhat of a validation
our optimization model give us

one hundred three percent of
the historical costs and a hundred and

thirty six percent of their service level.

The cost has to be expected because as I
said there's a penalty because from that

the man with the Model want to spend
as much money as it can in terms of

our donations the service level increase
comes because it's an opportunity to not

only it's deterministic so
we don't have variability in lead times as

well as the fact that we could make our or
any headquarters processing times and

I lead times because we didn't have
sufficient data for that we then took this

this the results from our current system
and compare them to other scenarios where

we used advance funding which again was
being able to borrow against forecasts and

donations and pre-positioning of
commodities where we locate the depots and

examined system performance there.

So the top of the current system
optimization became a benchmark

that we used to compare the performance
of advance finding and pre-positioning

some of the results of this we have here a
graph of the results of the events finding

on the bottom we see him out of the funds
we have available for this mechanism

as a percentage of annual donations for
W.A.P. on the right with the green

We have the cost difference versus
the benchmark which is the cost difference

of the model using events funding versus
our benchmark model and on the left.

We have the service level
difference versus the benchmark.

So you can see the cost increase because
we're actually allowing the countries to

purchase more since they get donations in
time to purchase and meet their demand and

we see a corresponding increase of
service level of around five percent.

It's important to know that for
this trial and for

all the rest when we talk
about cost with the model.

It's purchasing cost as well as
the capital cost of using the funds.

So when we say ten percent of
annual donations are in this fund

it means that we have one hundred
million dollars and we've taken the cost

of capital of having that hundred million
dollars two included in the cost.

Here's a great little packs of service.

I was in Cox and response to De Profundis
So we're now talking about using

prepositioning for the model we
just opened which ever do close

it feels is optimal given the amount of
the programs available here on the bottom.

What deep problems are is
the amount of money.

We have available to actually operate
these prepositioning the photos and

purchased commodities to stop them with
an import on the right in green We

have the cost difference versus our
benchmark and on the left in blue.

We have the service level difference as
you can see here this is a zero percent

cuts there is zero cost difference and
it's in their lives.

So in most situations you can
actually see a net savings.

Even though we're accounting for the cost
of capital of these funds the depots

allow us to purchase more effectively
because the depots don't have

the constraint of waiting on donations and

we see another increase in service
level of around five percent.

So once we saw the depot

implementation was effective or
using people as was affected.

We wanted to see if you like what order.

We would want to open these
the input was since W.B.

probably can't go open
seventy POWs at once so

we begin with Senator Mozambique and
eastern Africa.

From there we go to Western Africa and
Senegal then to Thailand

back to eastern Africa and with Kenya
then to Panama City Cameroon and

finally in Sudan these locations
were chosen by the model actually

eight possible locations we get invited
to be the only one not open to Pakistan.

It was chosen by the model because
again it produced overall costs and

increase service level.

This is a graph of the effects of
opening these the pros on the bottom

here we have the order of opening up so
in Mozambique one Depot is open

Senegal Studios and it's the second
depot again Green is the cost difference

versus our benchmark on the right and on
the left and the Lu is the service level.

So our zero cost is in fact
at the top on the right

which is beginning with our benchmarks
we see the cost decrease to finally

we get a savings of around
four million dollars and

that our service level increases
to about five percent.

So this is kind of a summary of
what we would like to see W.P.

do we like to see them
open seventy in prose.

Which were the most and he threw suit
on that we showed you in the crowd or

in the map we would use thirty
million dollars in funds

to operate these depots and
purchase inventory for them.

That would come to around using these
things we get around the seventy nine

percent service level in our model when
comparing that to the benchmark of our

model of their current system we
see around a five percent increase

in service level which equates to about
a order to have million in various and

beneficiaries or our total additional
cost is actually negative we

say around the corner in dollars and that
equates to and around eighty nine percent.

Savings per additional bit
best theory survey you know.

So that's the marginal cost
of serving these people for

a year we save eighty nine cents a person.

So in summary we first had
to write down price some

of these problems with the W V
supply chain and operations and

then to who address these problems we
created an inventory management tool

to minimize cost the target service I was
free countries and the supply chain model

to sat in their beds different ways to
smooth variability in donations and we

used to model different possible scenarios
to wow Governor Peter plan ahead and

actually open depots and prepositioned
commodities are there any questions

like every

yes you have too many different countries
and where I was with in those countries or

whatever is in charge of managing their
money at that where I was not a forecast

for the six months or they can just get
the average man just right so I have

no other world forecast.

I guess it was a dangerous for

us to use because we didn't get into
any of their demand forecasts and so

we were kind of depended on that it was
those words is not the demand is a very

specific term for what we use here but
because it's my flights going to purchase.

There is always more of
a rhythm of the majors so

demand in the way the value
is how many how many metric

tons of commodities that they want to
be able to provide for the product.

Well Mr capital that
where the mission was me

the cost of capital we
used was twenty percent.

And as far as where it was made.

No I like who was made to
work him from where I was and

with the use of their company.

So there's there's the two types
of direct donations directed

an underactive of that and then they come
from all over they come from companies

they come from countries they come
from individuals so directly donations

go specifically to countries that can't
be used for anything else not directed

donations already then just W.B.
decides how to use them the best way.

So you know where the cost
of capital is not the kind

of story

a lot

to do.

I mean there's actually you know.

Just for

you this year story just got me either.

They were expediting which various
things about what I was or

if they were part of the break
actually does happen.

They don't have you know there are just
locally or borrow they will ration

rations out to the people in order
to get to as many people but

not what they would want because they
have got a certain specific number of for

us very kind of my post is about right for
everybody together.

So instead of just getting to
one person with a meeting to.

Where Roger things were from Mandy
holding it up and ordering coffee.

So we really have just

got models because there's a lot of.