You know the only thing that’s constant in agriculture is change. Milk, cattle, and many other agricultural products vary in composition, quality, and size throughout the year. In order to provide a consistent product for the customer and maximize returns for the company, the dairy supply chain manufacturing process must address this variation. Data science can help you with time varying challenges in a way your ERP can not. To help showcase how this is done I’ll use the production of cheese and whole milk powder (WMP) from liquid milk as an example.
As you know, milk is made up of 5 main components: fat, protein, lactose, minerals, and water. These components vary across the season depending on a number of factors including the point in the lactation cycle, type of feed, and weather (especially drought or heat stress). Finished goods, on the other hand, are of a standard composition. That means this variation needs to be taken into account during the manufacturing process. The process of collecting milk from multiple farms in a single tanker before delivery to a factory for processing does even some of this variation out, but not all. You are still left with significant variation across the season.
This is where the time varying bill of materials (BOM) comes in. Using the example of cheese: different cheese types require different ratios of fat and protein in order to produce the different characteristics of, say, a cheddar vs a Camembert (note the process and conditions also differ significantly). All milk coming into a factory is separated into a cream stream (typically 40% fat) and a skim stream which contains the majority of protein, lactose, and minerals. The cream stream is then added back to the skim stream in the right amount to get the target fat to protein ratio. As the protein varies across the season the amount of cream that needs to be added back will change.
What does all of this have to do with data science? Because the process can be so complex we need to have time varying BOMs. This is because the “cheese making recipe” must change fluidly as the composition of milk changes across the year. Many traditional ERP and planning systems will tell you that they can handle this…..when they in fact can’t!
Now I want to talk about multiple alternative BOMs which are another important feature that Austin Data Labs can handle but many other systems cannot. First, what do we mean by multiple alternate BOMs? Well, a BOM is essentially a “recipe” to make a finished good; multiple alternate BOMs just means there is more than one way to make the same finished good. A system capable of handling complexity, like ours, enabled you to choose the best “recipe” as variables change.
Let’s use an example of whole milk powder (WMP). WMP as a finished good has a standard protein content but there are multiple ways to get to this target protein level. Raw milk out of the cow has a protein level higher than the target so there are a number of milk-based ingredients that can legally be added to get us to the target level. The four main ingredients are permeate, lactose slurry, dry lactose, and milk permeate powder (MPP). On any one day at a particular WMP factory you may have all of these available or you may only have one. Each of these ingredients have different economics of use, plus a different shelf-life determining how long it can be stored. This means a choice or optimization must be made of which one to use, when, and for which product; this entails selecting between alternate BOMs. Not all systems can deal with this complexity, but ours was designed by experts in your industry to handle your specific challenges, just like this one.
So whats the take away from this? If you are an agricultural company in a disassembly business like the dairy supply chain, then you need to ask the right questions of any software provider to ensure they can actually handle the complexity and requirements of your business – otherwise you risk being disappointed.