Getting started with smart industry: how to reduce the distance to your machine?
We can distinguish between the following levels:
. Production line3
. Engineering and management
At each of these levels, we can go through an improvement cycle based on data, as described above. Together these levels form a spiral - hence the title of this article. Below we walk through all the levels, with this spiral behavior becoming clearer in each step.
"Back in the 1980s, it became clear that in an automated (robotic it was then called) production line, there is an optimal amount of 'work-in-progress' inventory (WIP) between production steps."
At the machine level, we can collect data related to the operation of the machine: number (or quantity) of products that pass through per unit of time, number of errors that are delivered, energy consumption, temperature progression, number of hours of downtime, changeover times, etc. This data can be made visible in a dashboard for each machine and can be used for calculations. If we have several comparable machines, we can also make comparisons. Based on these comparisons we can then make improvements to the machine, or the way it is used. This is in fact the smallest loop in the whole.
Smart industry distance to the machineFor a production line, made up of multiple machines, we can do something similar, with some different parameters (data) to interpret. Think about transportation time between machines, bottlenecks in the line (not every machine is equally fast), stock work-in-progress between production steps and total production time per delivered product. This data set will lead to different analyses, different calculations and comparisons, than those at the individual machine level. However, there will also be points where the two analyses may contradict each other. A simple example shows this. Back in the 1980s it became clear that in an automated (robotic it was then called) production line there is an optimal amount of work-in-progress (WIP) inventory between production steps.
No WIP at all means machines will shut down for lack of input, too much WIP means too little finished product will come out of the production line. Thus, this optimization for the entire production line will likely lead to a different WIP quantity than the one produced if we let each individual machine or production step produce at maximum speed. Indeed, because each production step has a different throughput, the latter leads to the least optimal combinations of downtime and bottlenecks. So at this point we need to look at both machine and production line level, and the inner part of the spiral emerges.
"At the machine level, we can collect data that relates to the operation of the machine"
Within a factory we may find several production lines. In addition, every factory will have somewhere warehouses, or other storage locations, for raw materials, semi-finished and finished products. If we start collecting data at the level of a factory, we will not only have to deal with the production process around the production lines but also with the logistics from, to and between the warehouses. Stocks must be coordinated with production and with that also the transport of materials.
When we produce in small quantities more orders go through a factory, but in many cases these will also be orders for production of different items. For example, consider a factory that produces different brands and types of pet food by order. If we produce in large quantities, fewer orders and for fewer different products go through a factory (a Campina milk factory). This calls for different internal logistics and warehouses and for coordination between production and storage - the second level of our spiral.
If we look at the entire enterprise of a manufacturer, then our spiral is again extended to a new level. Within such an enterprise we may find several factories, in different locations. These may or may not use the same raw materials, warehouses and perhaps even each other's products. This is where a whole new category of data comes into play: external delivery times and size of external deliveries, product stock between factories, reconciliation of stocks and times between factories, incoming orders (number, size, timelines), cash flows, etc. This data says something about the total operation of the company. This data says something about the overall functioning of the company, and this in turn can be linked to the data we have for production lines and factories.
This is also where coordination with purchasing and sales processes comes into play - and the data coming from those processes. After all, production has to be coordinated with warehouse capacity, but also with the capacity of the sales process and vice versa.
Engineering and management
The feedback loops described above clearly indicate the spiral, but there is another level to which this returns. Every enterprise, every factory and every production line is ultimately shaped by people, for the time being. The management of enterprises and factories and the engineers who design all the equipment and processes are links in a management process, from which we can also extract data and use it for improvement. This is not limited to financial performance or whether or not projects are completed on time and on budget. We can, in fact, also include all data from the other levels of the spiral: every shortcoming, but also every strong point of a machine, production line or plant has an influence on the performance of a company. Engineering and management are then the processes and links that need to be adjusted based on that performance. An engineering organization learns from every plant it builds, and a management team learns from every year it manages a company.
Smart Industry and data feedback play a role from the highest to the lowest level in this spiral
The fanning spiralEarlier we referred to the 1980s with respect to WIP, but these principles have been known for decades on other levels as well. With Smart Industry and its associated tools such as sensors, computers, data dashboards and Machine Learning, it becomes possible to align these levels faster and better, and to continuously monitor and adjust.
In two follow-up articles, we will explore this in more detail - one for the three inner levels (the inside of the factory), and one for the two outer levels (engineering and management).