Managing Long-term Production Disruption

Q&A with Tim Eschert

The below post is a summary of the Q&A that took place on the 15th April 2 pm GMT (during COVID-19 pandemic) with Tim Eschert, a world-renowned mechanical engineer, where executives asked their burning questions on managing their production disruptions. To watch the video version, click here.

Q: How to achieve greater flexibility in the production process?

The demand for flexibility is something we see a lot these days, and due to our current situation, it’s perfectly relevant. From what I see in companies that I have worked with, specifically heavy industries and chemical process industries, the demand for flexibility is massive. The larger picture of process mining and data analytics all feeds into this goal of increased flexibility. When discussing production flexibility, it’s important to remember there are many backgrounds and reasons for this pursuit: it could be a decrease in demand, a decrease in the available workforce, or a decrease in supply.

Instead of analysing one individual production process, it’s now becoming important to gather data for the entire supply chain. Once the holistic process view is gathered, all needs for flexibility can be addressed, and processes can be smoothed out. These principles are nothing new, but what’s new today is the tools we have which are more effective than ever. Thanks to increases in computational power and storage space, we can generate models and analyses of production processes that are much wider than anything from the last few decades.

One major way we can improve flexibility in modern times is by examining the vast amounts of data already available to us and harnessing findings to tweak and improve our supply chain processes. Another method would be analysing trends in both small-scale processes and the holistic bigger picture in order to forecast where things will go in the future. And finally, as a step beyond forecasting, it’s important to use in-depth analyses and machine learning to understand why things go the ways that they do.

Q: Is it a good time to introduce machine learning and data analytics tools? How can they help to sustain operations right now?

This is a perfect follow-up to talking about flexibility. Yes, I believe this is a perfect time to double down on analytic efforts that have already been started. There are various reasons why I think this. First of all, a lot of companies have already started building analytic capabilities or at least have some kind of manufacturing data foundation to work from. If that is the case for your business, you already have a competitive advantage over those who have not.

This would be the case with or without the current lockdown situation we are facing, but the next steps of doubling down on data analytics are perfect for remote work. It doesn’t matter where the database is: I could be in my Dusseldorf office and the data I work on could be held in Indonesia or Amsterdam. Over the last few weeks, we have really been reminded of how many ways there are to overcome this distance and allow effective remote working - I mean, who would’ve thought we’d be having this discussion entirely remotely? Building the analytics efforts on top of the collection efforts is just as doable as ever.

The other thing worth addressing is increasing efforts on data collection. Much of the engineering can be done remotely, and when we think of what our future investments will be, the answer that comes back time and time again is gathering more data. There’s no one perfect time to be doing this but if we have to prioritise which initiatives are worth betting on, building this foundation for forecasting and analytics is extremely important. While new challenges are being faced in our current times, it is important not to forget the range of modern challenges we were prioritising until recently: environmental challenges, emissions, energy efficiency. These priorities will come back when the COVID-19 crisis is over, and the data we will have gathered throughout this “downtime” will be very helpful and worthwhile.

Q: How to reduce costs during this time?

There are various ways of reducing costs at any time, not just right now. But there are some unique things to consider during our current situation. First of all, there are support programmes in place by governments to aid businesses. These differ greatly from country to country, so it’s important to know about the tens or hundreds of government support programmes available to your company.

Flexibility has already been discussed but it is, in my opinion, a low-hanging fruit. It allows us to keep operations up and running whilst mitigating losses in these volatile times. These days, production flexibility is one of the largest adjustable factors that can aid businesses in reducing costs. Along with production flexibility comes the flexibility of the workforce: in most places right now we see that it is becoming easier to either decrease or increase workforce numbers if needed. Obviously right now, the number of workers operating is lower than normal, but businesses are finding ways to make this work for them.

Q: What KPIs to consider to account for resilience in production processes?

This is an excellent question. Resilient processes are more important than ever before. If I were to design a new manufacturing process in these current times, I would come up with a measure for flexibility for sure. The ability to increase or decrease volume or quality over time, the ability to continue production with a reduced workforce: these are the flexibility KPIs I would use to ensure resilience throughout my process.

The next I would focus on is minimising disruption in the supply chain. If all of a sudden I cannot rely on my suppliers to deliver materials for production, then I need to be ready for this. The same applies to if demand goes down and I cannot rely on the customer, we need to be ready for that. These uncertainties are currently being experienced all over the world. Right here in Germany, automotive manufacturers are some of the main businesses facing supply difficulties.

There are also completely external factors that we cannot account for that need considering. This can include governmental shutdowns, new rules and regulations for disinfection or distancing in the workplace, and so on. In situations like these, having an already flexible production process will allow you to be agile and deal with whatever comes at you with agility and resilience.

Q. How can machine learning be used in predicting supply chain disruption using historical data?

When we talk about machine learning, we have two large branches that look at from a scientific perspective: supervised machine learning and unsupervised machine learning. When we are looking at historical data and learning from the past, we call that supervised machine learning. The problem with this kind of machine learning is that we are limited to predicting what we have seen before. With a crisis like the one we are currently in, we are facing completely new disruptions that past data could just not have ever predicted.

However, within the larger umbrella of data analytics, I see machine learning as the next step. It completely automates the data analytics process and this can be incredibly helpful with analysing trends and revising time lags between different steps of the supply chain.

Predicting a crisis with machine learning is as impossible as it is without machine learning. I would instead emphasise the data analytics part: looking at all of the available data from suppliers, customers, etc. and manually analyse the data in order to apply older findings to your current situation.

Q. How to effectively train your workforce to be production line-ready in a quick and remote way?

This is an interesting question as bringing people on-board in times like the current crisis is something I am yet to perfect myself. What we use to remotely bring people into the company are video webinars. This includes learning about subject matters, practical online courses, and a lot of reading material.

For theoretical onboarding, it’s difficult and slow, but it works. For practical and manual tasks, it’s even harder as we currently have no way of remotely training people for production lines and factory settings. We currently have no solution for remote practical training, but the webinars for theory and subject matter training are required for when new workers get to train practically so it’s all relevant and necessary.

Due to the current crisis and the amount of remote work being done right now, I think more attention than ever has been drawn to our onboarding process. This has led us to think: “Where do we have practical tasks and where do we have theoretical knowledge?” and “How can we separate them from each other?” Usually, these things are taught together which is obviously the most efficient way of onboarding. But while we are living in these disruptive circumstances, we can try to go over existing study materials and attempt to separate as much as is possible to be learnt remotely.

There are tools out there that tackle remote onboarding for repetitive tasks really well, as well as more video learning, webinars for process improvement tasks and six-sigma, and so on. For now, there appears to be no perfect method for making new workers production line-ready remotely.

Article Share Image Share this article on your TLN® Profile
Share this article TLN Linkedin Post TLN Twitter Post TLN Facebook Post