Deterministic inventory models: What are they and how are they implemented?

September 24, 2017

Let's remember that management models ofinventoriesthey are divided into dependent demand
and independent. Within the dependent demand, we find deterministic models.
of inventory. For independent demand, we also found besides the deterministic ones,
probabilistic.
This is an important clarification for the beginning of the post, as it will focus on the models.
deterministic inventory for independent demand.
And for items with dependent demand? Dependent demand is one that
"it depends" on the demand for another item, for this MRP systems are generated orplanning
of material requirements.
Let's get started!

What are deterministic inventory models?

These are those where it is assumed that we are certain of the demand. This can
to be given byrdemand forecaststhe real orders from customers.

Now, within the models, we can generate a subclassification if we consider

in addition to the assumption of certainty, the claim can be static, which is one where
this remains constant; and dynamic, where despite being known, it varies through the
time.
This generates the following models:
Inventories with static deterministic demand
We consider that the demand is known with certainty and is always the same.
Economic Order Quantity (EOQ–classic) models: Also known as the model
Harris–Wilson, the EOQ method seeks a balance between setup costs and costs
storage. It was a pioneering model that served as the basis for the development of other
variations of the model, such as EOQ with quantity discounts, EOQ with planned shortages,
EOQ with multiple items with storage limitations, etc.
EOQ with quantity discounts: Considers the reduction in the purchase cost of a
article when buying in bulk.
EOQ with planned shortages: It states that for a period, demand will not be met.
generating missing items.

Economic order quantity in production (POQ): Considering that the order can be
receive over a period of time, this model takes into account that the demand rate
and the production rate.
There are many more EOQ models with very specific parameters. For now, we...
let's consider the most studied.
Inventories with dynamic deterministic demand
We have a degree of knowledge about the demand but it varies over time.
This poses a challenge, which is the lot size, as the inventory costs depend on this.
they may be greater or lesser. To respond, methods or systems have been generated
plotting, such as the following:
Batch by batch: It consists of obtaining exactly what I need, which leads to having the
exact inventory required and with it a low maintenance cost.
Constant period: Arbitrarily sets the order intervals.
Economic Order Quantity (EOQ): The EOQ can also be used to determine the
batch size, however authors Jay Heizer and Barry Render do not recommend its use
when the demand is relatively constant and not dynamic.
Fragmented period balancing (BPF): Seeks to balance the costs of maintaining
inventory and ordering costs.
Silver-Meal Algorithm (SM): It is heuristic, meaning that through decision rules
seeks to provide a good (or optimal) solution to the inventory problem. It focuses on the
minimization of total cost (ordering and holding) per period.
Minimum unit cost (CUM): It focuses on minimizing the unit cost through
comparison of ordering and holding costs for different lot sizes, in order to
choose the one that shows the least difference.
Wagner–Whitin Algorithm (WW): Through dynamic programming, it seeks the
minimization of ordering costs and inventory holding costs

What is a continuous inventory review system?

It is also commonly referred to as the Q system, reorder point system, or quantity system.
fixed order. In this type of inventory control system, every time we make a withdrawal
from an article, the control of the remaining inventory is carried out, in other words, the level is reviewed.
of the resulting inventory taking into account that:
If the resulting inventory is below a certain level, a new order is placed.
If the resulting inventory is not below the level, it continues to be consumed.
That said, we constantly review the inventory due to uncertainty about demand or
of the delivery time, or having them variable which is the same. This is essentially what it means
a probabilistic scenario.
That is why a continuous inventory review system has 3 cases:
Variable demand and constant lead time.
Constant demand and variable delivery time.
Variable demand and variable delivery time.
What would happen if the demand is constant and the delivery time is also constant? We would be facing
adeterministic inventory model, What is it when we have certainty about both the demand?
like of the lead time. This is not the case.
In the exercises we will see, you will see me talk about average demand or delivery time.
average. We say average precisely because that data is going to vary. How much is it going to
to vary? It will be more or less variable depending on its standard deviation.

Remember
The standard deviation is a measure that indicates the dispersion of the data with respect to
the average. For example, the average for {7,9,6,8} is 7.5. The average for {10,5,10,5}
It is also 7.5. Now, the standard deviation of the first group of data is 1.29.
of the second data group is 2.88 being higher. This indicates that the data from
the second group is more dispersed, but that is noticeable at first glance, do you see it?

We are not certain about the demand and/or the delivery time. We always order when the
inventory is at a certain level (reorder point), therefore the lead time varies, which
this means we must have safety stock.
And what would the reorder point be?
If the number of items reaches a specific level in the inventory, we will place a new one.
purchase order and we will call it 'Q'. That specific level is what we know
as reorder point or reorder level.
Let's keep in mind that this 'Q' will always be fixed (and yes, therefore we are going to calculate it),
but the time between one and the other "Q" is variable; if it were not so we would be facing a model
deterministic, which is when we know with certainty the demand and the lead time.

How does safety stock come into play here?

This scenario presents us with uncertainty about demand (probabilistic scenario), it is because
that even if we are always going to ask for the same, we will have safety stock available and
We will constantly review the inventory level in order to avoid shortages.
Thus, the safety stock or inventory consists of the storage of units
additional measures to avoid shortages.

But how to choose how much safety stock to have? The decision aims to have a
balance between the level of customer service and the costs of maintaining inventory.
One option is to propose cost minimization models, but it is a complex task since
involve the determination of the cost of stockouts and backorders.
Another more common option is to define service level policies for the inventory and based on
there to calculate the safety stock according to that policy.
Inventory Service Level Policy
Establishing an inventory service level policy is nothing more than defining the probability
to avoid running out of inventory during the time an order is placed until
that this arrives (lead time or waiting time).
For example, we can define an inventory service level of 95%, which means that
There is a 95% probability that demand will not exceed supply. In other words
In this way, the probability of shortages is 5% (100%-95).
However, demand may not behave uniformly over time.
wait (lead time), as it may be so.
Let me explain: If the variation in demand with respect to the average is not large, we can
to give ourselves the luxury of having a small safety inventory. On the contrary, if the demand
varies greatly from one order to another, it is advisable to have a large safety inventory for
avoid shortages.
This variability leads us to talk about probability distribution and with this to include two
very important concepts in inventory management: mean and variance.
Calculating the safety inventory
To express the safety inventory, we have the following formula:

With a view to the graph below, let's assume

that the demand is normally distributed during the lead time. The central line is the
average. If we set the inventory service level at 50% (that the probability of having
shortages are 50%) the reorder point will coincide with the central line. If this were the
In that case, the safety stock would be 0.

The reason? The reorder point is equal to the average demand plus the inventory of
security. Clearing the security inventory, we have to:
Reorder point = Average demand + safety inventory
Reorder point - average demand = Safety stock
Now, if we want to provide a service level higher than 50%, the reorder point must be
greater than the average demand during the lead time, which would imply including the point of
rearranged to the right of the central line of the Gaussian bell.
Let's assume we want an inventory service level of 90% (that is, that the
the probability of having shortages is 10%). How do we calculate the inventory of
security?

In this way:
At higher values of z, we will have a greater inventory service level and therefore
greater safety inventory. If z=0 there would be no safety inventory, which would indicate that
the reorder point is equal to the demand and we would have stockouts 50% of the time.
Enough theory, let's land what was said in a small exercise.
Exercise #1: Calculating z
It is important that you know how to determine z. We will discuss the theory about the normal curve in
another post. For the moment, find a regular table. There are plenty on Google, just type in regular table
and go to the images section.
Let's suppose that a company wants to have an inventory service level of 95%. To
To calculate the safety stock, we need the z value, what would it be?
What we will do is search in a normal distribution table for the number closest to
0.95. Look...
That number is in yellow, it is 0.9505. Moving to the header of the columns.
we found 1.6 and in the header of the rows 0.05. When added, we get a z=1.65.
With this clear, you already know how to calculate z, something vital for the exercises we will see.

Q system with variable demand and constant lead time

If we talk about variable demand, it is because we need to estimate the demand distribution.
during the lead time.
How is that done? Well, in the following paragraphs I will explain where the formula comes from.
of the distribution of demand during the lead time. It is not essential that you know where from
It comes, so you can go straight to the example.
How do you arrive at the formula?

Think about the following: A case where we know the average demand and its deviation.
standard over a period of time (let's call it t). Consider that the average demand and the
standard deviation is constant for each period t.
Now consider the lead time (waiting time) which can be a multiple or fraction of that.
waiting time. Consider for example that t=1 week and LT=3 weeks.
Thus, during that LT, the demand could be described as the sum of the demands during
each period t. If t=1 week and LT=3 weeks, the demand D for LT would be d+d+d, or also
d*LT. Are you following me? The same would happen with the variance of the demand distribution during
the lead time, that is to say:

A graphical view for what I just mentioned is the following:

Source: (Krajewski, Malhotra, & Ritzman, 2008)

Considering the previous image, notice how the demand distribution develops during
the lead time based on the individual distribution of weekly demands. If d=75, σt=15,
If LT = 3 weeks, we will have an average demand during the waiting time of 225.
units (3d=75*3) and a standard deviation of 15√3=25.98.
Decided? With this we go for our second exercise:
Example 2: Variable demand and constant delivery time
In this example, we are going to consider a variable demand and a constant delivery time.
In this example, we are going to determine all the values of the inventory system. Keep in mind
that just as we developed it in this example, we do it for the upcoming examples,
exception of the reorder point and safety stock formulas, which do suffer from
modifications.
Hinca is a company that imports microphones and sells them. Its average demand is 60.
units per week with a standard deviation of 15 units. The time since it was issued
The order takes two weeks to arrive.
The ordering cost is $145. The holding cost is $36. If the company establishes a level
For a 90% service level, what should the safety stock and the reorder point be?
Reorder? What is the total cost of the inventory system?
Let's see:
σ= 15
d=60 units per week
LT=2 weeks
Step 1: We obtain the z value for a 95% inventory service level, as we did
in the previous example. In this case, the z value for 90% is 1.28.
Step 2: We calculate the reorder point and the safety stock:

Remember to work with the same units when using the formulas. In this case, we are
working in weeks, but if the delivery time were given in days, the demand
we must convert it to days.

Step 4: We cannot calculate the total cost of the inventory system Q if we do not have the
optimal order quantityEOQ. To do this, we first need to determine the annual demand.
Then we can to do the calculation del cost total.
Notice that the formula to calculate the annual cost is very similar to that of the classic EOQ model.
The only thing that changes is the part of the annual maintenance cost. This is calculated
considering that the safety stock is available all the time, even when
sometimes there is demand higher than the average demand during the delivery time, and sometimes
smaller. So, during the year, we assume that the safety stock will be available.
How are these results interpreted? When the inventory level is at 147 units,
we will place an order for 159 units that will arrive in two weeks. During that time the
The risk of having shortages is 5%. That said, the total cost of the Q system is $6679.28.

Q system with constant demand and variable lead time

In this case, we are certain about the demand but not about the delivery time, as this is
variable. How to determine the reorder point then?
The reorder point would be given by the following formula:

Other variables such as total cost are calculated in the same way we addressed in the example.
previous
Example #3: Constant demand and variable delivery time
Let's go with another simpler example:
Higgi is a company that sells televisions. It sells 20 televisions a day, being almost
always constant. When Higgi places the order, the television manufacturers have a
average delivery time of 10 days with a standard deviation of 4 days. Because of the
the cost of shortages is very high (well, it's a television) the company has estimated a level of
98% service.
We are going to calculate reorder point and safety stock.
20 televisions per day
10 days
4 days
98%
Step 1: Let's locate z of 98% just like we did in the first example. Checking in a table.
normal, we obtain that z=2.055
Step 2: We already have everything to calculate the safety stock and the reorder point.
Interpretation: When the inventory level is at 364 units, we will place an order that
It will take about 10 days to arrive. During that time, the risk of having shortages is
2%.
System Q with variable demand and waiting time
Finally, we address the most realistic case. In practice, this is the case that occurs most often.
in companies, where we have uncertainty both about demand and time
wait.
This situation is often addressed through computer simulation. But in
In this post, we will limit ourselves to calculating the point of the inventory system as it is
we have done in the previous examples.
That is to say, if you have understood the previous examples, you can already deduce the formula to
employ for this scenario.
Example #4: Demand and variable delivery time
A company that sells All in One equipment presents a demand.
with a normal distribution of 200 computers on average with a standard deviation of
20 units. The delivery time from your supplier also has a normal distribution of
18 days with a standard deviation of 8 days.
What is the reorder point and how much is the safety stock if the company has a
95% inventory service level policy?
Step 1: As always, we determine z of 98%. We obtain 1.645.
Step 2: We calculate safety stock and reorder point.
Interpretation: When the inventory level reaches 6236 units, we will place an order that
It will take about 18 days to arrive. During that time, the risk of having shortages is
5%.