In my last post I looked at different ways to store material. This post is a continuation, looking at where to put materials. I want to give an overview of the different options to help you choose one that is suitable for your situation. This first post looks at storage with fixed locations, and why this is usually not so hot.

## Goals

Usually, any kind of storage unit has three main parameters of interest. Depending on your storage approach, different methods may have different trade-offs. The three main KPI are:

**Retrieving and Storing Cost**: How much cost is needed to get material in and out of storage. Often, the cost in retrieving and the effort in storing is similar, except that smaller quantities are retrieved but more frequently. Storing larger quantities may require lifting equipment (forklift, pallet lift, etc.), whereas retrieving may be possible by hand. Storing and retrieving may also be on different sides for rolling racks. Of course, the goal is to reduce the cost.**Increase Retrieving and Storing Frequency**: How much time is needed to store and retrieve an item. Or, how many items can you get per hour. This is related to the cost, since you can easily increase the frequency by adding more workers or equipment. The goal here is to get the maximum number of items for a given number of workers or a given set of equipment. The main factor here is the distance traveled for retrieving.**Space Usage**: How efficiently can you use the space (i.e., how much material can you fit into the storage). Do not fall for the fallacy that you can use 100% of your storage space. You always need a few empty spots to add material before you have retrieved other material. Assigning fixed locations to your material usually is a much worse usage of space than having no fixed location.

## Fixed Location

One of the main decisions for your storage strategies is if you have **fixed locations** for your items or not. The storage spaces are reserved for certain items, and you must not put any other item in these spaces. In effect, every storage slot is assigned to one part type, and cannot hold another part type.

The advantage is that it is easier to find items without the help of a computer. Since the location for each item type is fixed, you can easily write this on paper. The shelf can be labeled with the item type and not just the slot number, and workers will soon remember which item is where. Overall, it reduces the effort needed to find out where an item is, since it is always at the same slot.

On the other hand, it significantly decreases your utilization of available space. You need enough storage slots assigned to each item type to cover the maximum storage need. This means most of the time you have empty slots that you cannot use for other items. Overall, you need much more storage space than if you have no fixed locations.

### Example Calculation of Fixed Location

Let’s look deeper into this. First, we start with fixed locations. Assume you have an **average** storage need assigned for one product of μ = 10. If you now build storage for exactly 10 parts, you will be almost always out of space, because your storage needs fluctuate! For mathematical simplicity, let’s assume a normal distribution with a standard deviation σ of 50% of the mean, or σ = 50 parts. If your inventory has only 10 spaces, then this will be sufficient only 50% of the time, but surely this is not good enough for you.

If you want your inventory to be good enough 80% of the time (still not really impressive), you would need to add 0.84 standard deviations (i.e., you would need space for 14 parts). The more fluctuations you want to cover, the more storage space you need. Unfortunately, this is not linear, and for higher values the required storage increases drastically. A 100% availability would require infinite storage space. Even only 99% would more than double your needed inventory space to 22 slots.

Before you worry about this “infinity” too much, reality is (a tiny bit) better. You do have different distributions, and due to the limits of reality, you will never need “infinite” storage space. On top of that, you can also counteract and sell more (if you are lucky), order less (if you can), or simply rent more storage space (if you can’t sell more or order less). But, even if you don’t have to worry about “infinity,” you still certainly need more storage space than your mean storage need to cover for fluctuations.

Let’s assume a more realistic reliability of 95%. In this case you would need 18 slots assigned to this part. If you now have 1000 identical part types, all with a mean storage demand of 10 and a standard deviation of 5, then you would need 1000 times these 18 slots or storage space of around 18,000.

### Example Calculation of NOT Fixed Location

However, this looks very different if you don’t have a fixed location assigned to each part. In this case, an excessive demand by one part may be offset by a lower demand by another part. Let’s stick with our example of 1000 parts, each having a mean of 10 and a standard deviation of 5 parts. The combined mean of all part types will still be 1000*10 = 10,000 parts. However, the standard deviation σ is calculated differently as shown below:

and

Hence, the combined standard deviation of 1000 parts with σ = 5 is not 1000 x 5 = 5000 but only 158.

Hence our mean storage need of 10,000 parts has a standard deviation of only 158 parts. Calculating the storage quantity that covers 95% of all cases, we need only 10,260 parts. Compare this with the 18,000 parts that we would have needed for a fixed storage location.

Below is the overview of our example. The y-axis shows how much inventory you need to cover a certain percentage of the parts. The x-axis shows how many different part types we have. The total average inventory needed is always 10,000 parts, and the standard deviation of the fluctuations is always half of the mean value (if you change this, you still get a very similar image). The top horizontal lines are for fixed locations. It does not matter if you have 10,000 of one part type or 10 each for 1000 part types, you always need the same inventory for the same required percentile we want to cover.

It is different, however, if the items do not have fixed locations. In this case only one part with 10,000 items is identical to the fixed locations. The more we increase the number of different part types, the lower our inventory requirement gets, approaching the sum of the means of 10,000 storage slots.

### Caveat of the Theoretical Model

It looks like you can save almost all the excess space to cover for inventory fluctuations if you do not use fixed locations. That almost sounds too good to be true.

Well, it is too good to be true. First, this is a simple mathematical model, with quite a few assumptions. The randomness of the fluctuations in your case is probably not normally distributed, for one thing because the boundaries of a normal distribution are ±∞, and you can not have negative inventories.

Second, we assumed that these variables are independent of each other. In reality, however, the inventory level of one part is frequently dependent on another part. For example, if it is high season, you probably have more of everything than in the low season. Hence, your storage requirements without fixed locations are probably a lot higher than my theoretical model. However, it will be still much lower than with fixed locations. How much? That depends on a lot of things, and is hard to calculate. Just try it out.

## When to Use Fixed Locations

So, when should you use fixed locations? Well, it is easier to answer when not to. If your storage space is tight and you are constantly running out of space, then you should avoid assigning fixed locations for your part. Similarly, if you produce on order and frequently have many custom parts in your inventory, then it makes no sense to assign this “once per year” part a fixed slot.

However, if you have plenty of storage, then … I would still be careful about using it. It may be an option if you have plenty of space, items with higher quantities, and a cumbersome system of tracking parts (i.e., if you use paper to write down which part is where, then any change is a lot of effort with a risk of mistakes). The less you change your locations, the easier it may be for you. Similar applies if you still manage your locations in Excel.

Overall, fixed storage locations are usually not so hot. There are some uses in mixed systems, however. I will discuss mixed and other systems in **my next post. Until then, stay tuned, and go** out and organize your industry!