Big Data in the Chemical Industry and the Need for More Use Cases

Wacker Chemie, BASF or Evonik – Germany’s industry is strongly rooted in the chemical sector. Described as conservative and slow-moving by many, the sector is prone for digitalisation and a more data-driven approach across various activities of the value chain. BASF recently launched a broad scale initiative under the umbrella “BASF 4.0”, others will probably follow with similar pursuits.
We talked with Sean Jones, former Vice President Information Services Business Relationship Management at BASF and Founder and Partner at Yukon Digital, a provider of digital solutions to companies involved in the Chemical, Petrochemical and Oil & Gas Industries, about data science, innovation cycles and the state of the chemical industry.

idalab: Can you tell us a little bit about yourself and what you’re doing with Yukon Digital?

Sean Jones: My business partner James Thomas and I founded Yukon Digital about two years ago in order to address the growing interest in digital analytics solutions for the chemical industry and other process oriented after industries like oil, gas, pharma.

In the last ten years, there’s been a lot of focus on process harmonisation and the standardisation of systems, trying to drive down costs as companies globalised the market. Most companies developed a lot of systems, to archive their records, capture information in the value chain and in marketing and sales, and storing their information in the manufacturing area. But, truth be told, there is actually very little optimization work with all the data they’re collected and so that’s what we’re trying to focus on.

idalab: What are the main areas you are focusing on?

Globally, there are about four different areas that are of interest to us when we work with chemical companies: manufacturing, supply chain, marketing & sales and new business models.

idalab: Could you briefly run us through these four core areas?

Sure. In manufacturing, it is basically all about automation and increasing the productivity of people. So we’re trying to understand how to take data and really increase performance, whether that’s through increasing reliability or decreasing maintenance costs.

In the supply chain, problems are of a different nature. Interestingly, visibility is one of them. My business partner and I always laugh at the fact that we can figure out exactly where our mail is, but we can’t figure out where in the world a chemical container is. That’s a rather odd observation, a decade ago DHL, UPS and FedEx developed these very transparent supply chains but those still don’t exist in chemical industry. This is primarily due to the vast amount of players in the chemical supply chain from production to the first transportation stage to shipping, to distribution and they’re disconnected.

When you have visibility on where your product actually is, then you can start doing more interesting things like optimizing inventory levels for required delivery reliability, or connecting both a better sales forecast with the production plans.

In the marketing and sales area, everything is about micro-segmentation and optimization. In the chemical industry, customers are very unique. So trying to better understand what are pockets of improvement, whether you can increase prices or improve market share or potentially even change a product is something that is done very manually, often still with tools like Excel. So we are helping companies to figure out from a very complex product and customer environment what are actual opportunities for improvement are.

And last but not least, there is the whole topic of building new business models. Essentially, the challenge is to take a classical chemical product and change the business model in order to increase loyalty and revenue streams. It actually tightly revolves around getting into a closer cooperation with the customer. How can you get data from the customer’s product usage to detect patterns? How can you better understand the drivers of usage and help your customers better solve whatever chemical or material problem they have? This can help inform future spending in product development, because it enables companies to allocate money to the right areas. The whole new business model area is extremely interesting but still very immature, so we see few topics which are pursued with urgency in this area.

idalab: That sounds fascinating. What’s your take on the reason that we see this apparently big disconnect between the ability to process data and to store data, the use of IT in general and clever ways of extracting value from these data and information. Any thoughts on that?

Well, this is definitely something we are trying to wrap our brains about, so we don’t have a real answer yet. In large parts, I believe, it is a cultural thing, so you know the chemical industry is about assets, “steel in the ground”. The topic of IT has always been one to avoid and most IT departments are still organised within the CFO function, which means it’s more about minimizing costs. The topic just appears to be not as “sexy” as others, but we have probably also not seen the real use cases in many areas outside of process optimization in the last five years. So, naturally, people rightly ask: where is the business case?

Presenting a business case is something that has not been done very effectively either by consulting companies or IT companies selling software so that’s something we try to focus on a lot when we engage with the client regarding innovative approaches to data utilization.

idalab: When you’re engaging with the client who are you talking to? Is this primary an IT topic or is it already such a strategic issue that you would discuss with the CEO, or strategy department?

We definitely focus on the business first and we always talk to business management. Depending upon the area of the project, we might then be talking to production and supply chain, marketing and sales people. There is usually a long phase of figuring out what their problems are and how these could be adjusted with a data-oriented solution.

At later stages, of course, we work closely with the IT group, also from the architecture point of view, to understand where is the data, and how can we access and utilize it.

idalab: From the four areas you outlined, where do you think is the most potential in terms of data science or data driven solutions? And how does that potentially fit or not fit to the sort of responsiveness you get from executives? Is it easier to convince them to do something in certain areas even though the potential might be in another?

That’s exactly the situation right now. We are actually starting kind of in the back, gradually moving forward. So we’re starting in manufacturing, because it’s easier to explain and the use cases are somehow clear.

We, however, see the much larger value creation opportunities towards marketing and sales and of course new business models. But those are much more difficult to explain and also harder to prove. CEOs ask: “How are you going to really help me?” Marketing, pricing and sales are exposed to such a volatile environment, whereas manufacturing and supply chains are easier to be tackled. Thus, executive management is more open to get started in those areas.

idalab: Manufacturing, generally, has been a data-driven endeavor for years. In this environment, what are the most intense pain points of companies?

Well, in the manufacturing area, connecting different data sources together is still one of the biggest challenges – whether it is for internal or external processes. Additionally, the data is oftentimes not just stored in an operational storage but also in laboratory information systems. So bringing the different data with different formats together and then merging it into a data set that is adequate for predictive work is one of the biggest challenges.

idalab: What challenges does it bring when scaling solutions up to the entire production facility?

To be honest, we are not that far down the line yet. Usually, once we have done a small engagement to prove the point, then the next step is how would you scale this to plant or global level. That’s a bigger challenge and involves more ‘corporate politics’ and change management topics: how do you get the operational and management team to actually use the predictive model? That is a major challenge, as it needs to happen with ongoing operations.

idalab: With the whole topic of digitalisation and big data being so prominent that it is hard to avoid, I would suppose that the big companies all have really large programs in-house that focus on achieving on becoming more data driven. How far ahead are these endeavors and what do you think is their likelihood of success in this sort of internal set up?

Well, we see very few big programs honestly. BASF actually has a large program called “BASF 4.0”, which was communicated on various public channels. But if you take other companies, they are looking at business cases, trying to avoid larger programs. Small and medium sized companies, although they have the freedom to act because they’re smaller, they are just watching the larger companies and the industry. But change is very incremental, as the chemical industry is relatively conservative, especially in the IT area.

idalab: Going back to one point you mentioned a bit earlier, regarding the scalability of solutions to the entire production site and its requirements in terms of organizational change. What do you think could be done to still develop use cases or how could these large projects be broken down into smaller components in order to give better buy-in opportunities across organizational levels?

We usually try to address this problem with prototype design thinking phases, where people are brought together in order to form an idea about the potential impact of the solution. Classically, the chemical industry is not used to the agile principles of IT. You usually have to have a full plan upfront and then just apply the waterfall method for project management. It makes quite a tough endeavor to trigger change than a more data-driven approach with smaller, agile prototypes but this will be the way forwards.

idalab: What’s your impression on when there will be sufficient urgency on executive level to push initiatives in this regard? Or, asked with a different spin, why is there no pressure at the moment, is the chemical industry just doing that well?

I think it will probably take another two to five years for it to fully arrive on the executive agenda. In the chemical industry, margins are just bigger, compared to sectors like retail, which have already heavily adopted data-driven solutions and even transformed their business models accordingly. In the chemical industry, there is still no pressing demand and – to some extent – a lot of companies seem to just wait and see, sort of go with the flow. Eventually, it might take a couple of leaders, strong innovative companies, to go ahead and – once they can prove some initial success – the rest will follow. So, we are also primarily addressing the leading companies in the market, helping them to get their initiatives on the way. If we manage to help them to increase margins, other will notice and the market will take off.

idalab: If you take the industry a couple of years down the road, what is your vision for the industry, what kind of potential could be realised?

Pragmatically we see that on top of the larger IT platforms (e.g. Microsoft, Amazon, Google, SAP, IBM, General Electric) there’s probably going to be a lot of activity. These larger platforms are important as they provide the security and connectivity to the different data that you want to play with. People and companies, though, are craving for better micro-services, let’s call it algorithms, in order to do the optimisation that we have talked about in the manufacturing area. The need for customization will slightly decrease, but companies, which will position themselves successfully will move towards intelligently connecting those four areas that I outlined: manufacturing, supply chain, sales & marketing and new business models.

idalab: Thanks for the interview


Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.