Hello,
I am trying to track and improve material shortages in our plant. I was wondering if there is any control chart that I could use for this. My objective is to show the current status of our system. The data that I have is just one column that tell the total number of shortages for each material produced.

I greatly appreciate your help on this. Thank You

Hello,
I am trying to track and improve material shortages in our plant. I was wondering if there is any control chart that I could use for this. My objective is to show the current status of our system. The data that I have is just one column that tell the total number of shortages for each material produced.

I greatly appreciate your help on this. Thank You

I found this elephant chart in another forum by Jennifer Kirley, but I'm sure you could customize it for your own. Track it for the week or month; by material, by part. It's quite versatile. Good luck :)

Hello,
I am trying to track and improve material shortages in our plant. I was wondering if there is any control chart that I could use for this. My objective is to show the current status of our system. The data that I have is just one column that tell the total number of shortages for each material produced.

I greatly appreciate your help on this. Thank You

There is a potential problem with tracking inventories on an SPC chart. Generally, the data for SPC should not be "auto-correlated", that is the most recent result should not depend upon the previous result. Inventories have a high degree of auto-correlation, since the most recent inventory is the previous inventory plus the goes-ins, minus the goes-outs.

I would recommend control charting the demand rate (goes-outs) and the receipt rate (goes-ins) and perhaps the change in inventory (difference between goes-ins and goes-outs) in order to analyse the potential for material shortages.

There is a potential problem with tracking inventories on an SPC chart. Generally, the data for SPC should not be "auto-correlated", that is the most recent result should not depend upon the previous result. Inventories have a high degree of auto-correlation, since the most recent inventory is the previous inventory plus the goes-ins, minus the goes-outs.

I would recommend control charting the demand rate (goes-outs) and the receipt rate (goes-ins) and perhaps the change in inventory (difference between goes-ins and goes-outs) in order to analyse the potential for material shortages.

It's unclear from the OP whether this is about tracking inventories, per se. It depends on what's meant by "material shortage." For example, in a metal-working operation, we might know that 500 pounds of steel is needed to produce 1000 parts. 500 pounds of material is "paid out" to the work order, but only 800 parts are produced and all of the material has been used. Where's the rest of the material? If we count this as one instance of "material shortage," it can be charted without fear of autocorrelation. It depends on the data, and what we want to know and/or predict.

True inventory is autocorrelated of course.

However, plotting the number of shortages (not how big they are or how close one comes to a shortage) as the OP asked might be useful as a busines emtric. This could be accomplished with a c chart - simple count of events each week or month. (at least that's what I've done)

One piece of unsolicited advice would be that giving management this data might result in unwanted action: the easy way to avoid shortages, in the short run, is to increase inventory. So you might consider how you will position this information and provide management with a better strategy on how to improve the number of shortages...

Charting the changes in inventory would be a great second level chart. Especially if you could determine how to chart the consumption rate vs replenishment rate...

Well I was thinking of using a 'c' chart or a 'u' chart, but am not able to establish a sample size, as they are all individual occurrences. All I am trying to do is to show that we have a situation here that needs attention and also trying to help our purchasing department in resolving these challenges.

I am also trying to find root causes of these shortages and try to develop a counter measure.

Thank You for all the suggestions.

If we are talking about a classic warehouse stocking problem, there are several things to consider in order to prevent shortages:

1. Yes, you can increase inventory, but that increases costs to store items.

2. You can work on controlling the use of the materials, and reducing unpredictability in receipt times. This is the basis for "Just In Time".

3. There are ways to set thresholds of how much minimum materials should be on hand before sending out an order.

Your variables are:
Usage rates (and uncertainty in usage rates)
Receipt rates (and uncertainty in how long it takes to be delivered following an order)
The costs of storing materials
The costs of making orders (affects the size of the order)
The costs of an stock-out (shortage).

a c chart will (most likely) work for you, then. simply plot the number of shortages that occur during each time period. (the time period is your subgroup). Not knowing how many shortages you are experiencing, I would think that monthly is probably about right. You should have an average count of greater than 1 per period. If you have less than one per month, then we should talk about why this is a problem, but tehre is an alternate chart that can be used.

Yes I was looking at the 'c' chart and I have started using it. What is the alternate method on this? Also Thank You everyone for your valuable suggestions.

Yes I was looking at the 'c' chart and I have started using it. What is the alternate method on this? Also Thank You everyone for your valuable suggestions.

the alternative should only be used when the average count is 1 or less and certainly when the UCLc <1. It is for rare events: you plot the number of days between shortages.

The control limits are:

center line = .7 X average number of succesful runs (or in this case, days)
UCL = -Average Runs X natural log(alpha/2) alpha is typically set at 1%
LCL = (alpha/2) X Average Runs

If you are concerned with the number of shortages, I'm assuming these are not rare events and a simple c chart will do.