Let’s start this blog with: Why the heck small lot sizes?
We as consumers want products and solutions that are built exactly to meet our requirements. It means ordering the exact car we want with the right color, right engine, and so on. That will logically lead to smaller lot sizes, or in extreme cases to a lot-size of one, a concept we know from discrete manufacturing. This concept is applicable for B2B and for paper and packaging companies, too.
Production planners may not like the idea of needing to manage many, but small lot sizes. But paper manufacturers can differentiate if their offering can meet this kind of customer demand in a flexible manner. They can use technology to better organize their sales order processing and production, so that they can combine production processes, align schedules, fill up their scheduling blocks, etc. to ensure each process is as profitable as possible.
The first thing they need to deal in this process is more sales orders requiring smaller lot sizes. Companies really want to automate the sales order processing as much as possible as labor is expensive especially in developed countries. The sales order process includes a wide range of activity including transportation scheduling, credit check, costing, pricing, and so on, Let’s consider, for example, finding the right product configuration. With help from algorithms and machine learning, for example with SAP Product Configuration Intelligence, customers can be guided to “their” appropriate product configuration already in a web-shop.
Figure 1. showing a workflow during sales order creation, with both default product configuration and recommended product attributes calculated by SAP Product Configuration Intelligence based on historic sales data
Producing and delivering profitable smaller lot sizes means being very good at organizing the production – and applying the latest technology in the context of Industry 4.0 is the key to this. Manufacturers start toward this goal by getting all information about customers, products, production, and logistics connected in a digital manner. They can use this to analyze demand and improve production & logistics processes. For example, they can easily and precisely understand what the customer requires, and in case of late order changes, they can check production status, and decide instantly if changes are (still) possible, and re-schedule transportation.
Once all data is connected, technologies such as predictive analytics can not only recommend the best product configuration of a product as mentioned earlier, but also the best way in which to fulfill its production and distribution. When we look closer to production, understanding and correcting flaws early will impact the ability to capacity.
I assume that paper producers will have to deal with more frequent change-overs of base weight & grade in the future, probably more than they like, and a better understanding of the production process can help to stabilize the process faster after a change to achieve maximal output rates again.
Let me conclude this blog with some real-world examples: There is Koehler Paper Group, a specialty paper producer based in Germany with some 800M Euro in annual revenues and 1,800 employees. Koehler is now working with SAP using IoT and big data techniques to do predictive (production) quality in their mills. They started with the goal to predict paper breaks on the machine. The sensor data now enables them to predict paper breaks, but not yet early enough in the process to really avoid all breaks. But they learned so much about their manufacturing processes that they can predict product quality parameters and they now correct production processes before issues occur. This means they can stay within the allowed quality interval and no rework is required. Koehler says this approach paid off instantly.
Figure 2. showing a “decision tree” modeling environment of SAP Cloud Platform. Once predictive analytics has calculated that e.g. product quality issues may occur at a certain probability, multiple measures can be defined to fix the issue or mitigate the risk, for example sending an email, creating a maintenance order, or communicating with a process control system for changing process parameters.
Another example scenario is Fibria Cellulose (now Suzano). This South American pulp producer is improving pulp production in using Predictive Analytics, too. They analyzed more than 1.6 billion records and over 650 variables in one huge data model, to identify how best to configure and adjust their digesters. The model then advises set-point parameters for better production yield.
Of course, the value of IoT and Industry 4.0 is not limited to manufacturing. In my next blogs I’ll talk for example about how it can help to increase worker safety, or how to improve asset management beyond predictive maintenance.