Hello to the Smart SMT Factory Forum community:
The summer is almost over, and I hope you’ve had a relaxing and enjoyable summer vacation with your beloved ones and are now back at work with fully charged batteries. I certainly am, as my vacation was great, but not entirely without any work. Because regardless of whether you climb mountains, relax at the beach, do some cycling or whatever, there is always time and room to think about things which are important to you. In my case it was about the question “How to start with the smart factory?”. Not so much the question of how to start from scratch on a greenfield, but more how to start in a given, heterogeneous environment with all the restrictions and limitations that apply there.
I’m pretty sure this is a question that keeps many of you busy, so let me share my thoughts on the subject with you. I’d also be happy to get yours so that we can eventually start a vivid and beneficial discussion about it. First of all, in order to think about getting started with the smart factory, there is another question to be answered, and that is: why would I want to do this at all? Why would I need a digital factory in which the machines talk to themselves and make their own decisions? What are the benefits of such a thing? This is definitely something that can’t be answered in a few sentences, so let’s move this question to a separate discussion and let’s assume for the moment that we have good reasons for implementing a smart factory, and that we decided to get started now.
Thinking about how this can be done, it is necessary to define the key elements of a smart factory and the sequence in which these elements need to be implemented. For me, the answer to this question is:
Now, as this sounds like “meaningless buzzwords”, let me explain in the following what’s behind these “buzzwords”.
What I mean with communication is the ability of all participating elements in the digital factory to talk to each other. Elements can be machines, equipment such as automation devices, or warehouses, software solutions, and also materials. And not to forget: human beings. The “ability to talk to each other” is more or less the creation of interfaces between these elements and of common protocols. Or to put it metaphorically: interfaces are the eyes and the ears of the protagonists that allow them to listen and to talk, and protocols are the languages they speak.
So, the first concrete task of the smart factory project is to ensure that every important element in the factory – in the first place the machines – is a connected device that has its own IP address. Most of today’s machines have that anyway, but older ones might need to be retrofitted. There are now companies in the market that specialize in turning traditional, standalone machines or devices into connected devices.
Once all the players have been equipped with mouths to speak and ears to listen, the next step would be to create a room for them to communicate and to collaborate. This means creating a network over which every element can communicate. In the past and up until today, this has been the classic firewalled LAN environment with WLAN extensions. In the future we might need something different, because we may want to integrate elements that are not part of our proprietary network and that are physically farther away and hence out of reach of our own wireless network. One of them could be the newly developed Long Range Wide Area Network (LoRaWAN).
Also, thinking about interfaces, we might need new solutions as well. One example: in the past, the keyboard and the mouse were the only interface between a human being and the machine. Then came the touchpad, and now we have voice input, just as we know it from our smartphones. Companies like IBM are already offering APIs that allow machines to understand the spoken word.
Furthermore, we need to make sure all elements in our smart factory speak the same language(s), meaning to implement common protocols, so that everyone can talk to everyone else.
Now we come to the most important part of creating a smart factory: applying intelligence to the network. Most of today’s machines possess already a certain level of built-in intelligence, but that is typically restricted to the operation of the machine itself. What we need now is the intelligence of the entire system or network. Again, metaphorically speaking, you can call this the intelligence of the flock. I’d rather call it shared intelligence.
A factory in which the machines can talk to the warehouse to find out which parts are in stock and which are out of stock might be the first step. But in order to become “smart”, it requires the ability to draw conclusions and make decisions. And this is where the latest IT technologies come into play. Things like artificial intelligence (AI) have been around for a long time, but smart mathematical algorithms alone are not sufficient. What is needed is data. Big data, to be precise and to use another “buzzword”. The ability to obtain all kinds of (unstructured) data from all kinds of sources and the ability to analyze this data with smart algorithms allows the owner of the network – or the network itself – to draw conclusions and make decisions. You want an example? Here you go: Have you ever wondered why you get certain ads on the right side panel of your web browser, while your wife, husband, colleague or child gets something completely different on the same page? This is because companies like Google and Amazon know exactly what you are interested in (because of your age, gender, hobbies, preferences, etc.) and what the person next to you is interested in. How do they know this? Because they have big data (the all-time summary of your digital footprint), and they possess the algorithms to analyze your unstructured data and draw conclusions.
And exactly the same kind of intelligence is needed for our smart factory. We need to have access to all kind of data about our customers, suppliers, markets, ordering behaviors, etc. from whatever sources, including social and professional networks. And we need to have access to smart algorithms to analyze this data. Only then we will be able to predict what the customer wants next and instruct our factory to produce it.
Once we have all of the above in place, we need to ensure the best possible flow in our factory. Flow in this context means providing the customer the ordered product in the shortest possible time (and, needless to say, at the best price, with the highest quality, in the quantity ordered, and at the right place). This requires perfect mass customization. Mass customization requires operational excellence and the right approach to flexibility.
Today’s classical line structures of PCBA factories might not be sufficient for this requirement anymore. Instead, cellular structures in factories should be thought of. But this subject alone is so comprehensive that it would require a separate discussion.
Now, these are my thoughts on how to get started with the digital factory.
I’m keen to hear from you how you see the things and what your approach would be.