How to handle Supply Variance
The impact of the supply variance to the business is significant. The planner can receive the material later than confirmed date. It may happen because of the supplier, transporter or the customs. Or, it’s delivered on time but there’s a quality issue.
Let’s consider a production facility. It may have some technical issues and may delay the production. Moreover, the output varies day by day. Sometimes less, sometimes more. So, there’s an important uncertainty on the supply side. It can be on the duration, quantity and quality.
We need to figure out this uncertainty. In order to have an impact on the safety & order optimization, the supply’s variance should be combined with the demand’s variance. In this article, we discuss the supply variance and how to handle it with the Unified Supply Model.
Before getting started, it would be good to take a look at the articles Forming a Demand Function and Leadtime Demand.
Let’s start with the timing. When we look at the variance on the supply duration, we consider two things; how far is the actual mean and a deviation around it. For example, the supplier confirms orders with a given leadtime of 4 weeks. However, in reality, the average supply duration is 5 weeks. So, if we are not 100% sure how to decrease it to 4 weeks, we need to use the average actual leadtime, in our model. This important input is used in CSL Calculation and Forming the Leadtime Demand.
A very good practice can be comparison of the confirmed and actual leadtime, with a total supply chain cost function. Given the method on Optimizing Leadtime, we can observe the cost difference and discuss it with the supplier. They may observe the solid results of the delays, and come with a good proposal.
We handle the shift on the leadtime, but we have another issue; the deviation. Consider the same supplier above; they deliver in 5 weeks in average. However, there’s also a deviation of 0,5 weeks. Can we combine this deviation with the demand’s variance? If they have the same function type and independent of each other, yes we can.
We consider the normal distribution for the demand and supply.
We have a periodic demand (d), and standard deviation of it (sd1). We also have an actual leadtime (T) and standard deviation of it, (sd2). (sd2) is a percentage of the leadtime.
Then, the result is like below;
Mean of Leadtime Demand (D) = dT
Standard Deviation of Leadtime Demand (DSd) can be calculated with the formula below;
DSd = Sqroot[ (T)[sd1]* + (T)[(d)(sd2)]* ]
The operator [..]* takes the square, and the operator Sqroot[..] takes the square root. Before adding the supply deviation, first we convert it to a quantity with multiplying by (d). Then, we square it and multiply with leadtime (T). Finally, we combine it with the periodic demand deviation, and take the square root.
This complex formula delivers us the combined deviation, and we are ready to use it for Inventory Optimization.
How can we add the quantity & quality issues? We need to convert them to a time rate. For example, if we order 10 and received 9, we can consider the actual leadtime longer by 10/9. For more precision, we need to apply more statistics…
We can also create some scenarios and check the results. For example, due to winter conditions, there can be some delays on the road deliveries. At which point we need to change the transportation type? (i.e. air shipment) We can form two Total Cost Function and apply Leadtime Optimization Method. Then, by changing the supply variance, we can find the equilibrium, and understand when we can shift to the premium logistics.
With this article, we discussed the supply variance and how to combine it with the demand uncertainty. Unified Supply Model provides a proper methodology to analyse the result of this combination, and helps to create confidence on the execution.