End to End Supply Chain Control Tower

Download Slides

When you look at traditional ERP or management systems, they are usually used to manage the supply chain originating from either the point of Origin or point of destination which all our primarily physical locations. And for these, you have several processes like order to cash, source to pay, physical distribution, production etc.

Our supply chain control tower is not tied up to a single location nor confined to a single part in the supply network hierarchy. Our control tower focuses on gathering and storing real-time data, and offers a single point of information related to all data points. We are able to aggregate data from different inventory, warehouse, production, planning, etc. to guide improvements and mitigate exceptions keeping in mind an efficient supply network operations in our end to end value chain.

Which allows us to do cross-functional data-based applications, one such example is like Digital Sales and operations planning. Which is a very powerful tool to align operations execution with our financial goals.

All this is possible, by using a future proof big data architecture and strong partnership with their respective suppliers such as microsoft and Databricks.

Speaker: Tarun Rana


– Good afternoon everyone. My name is Tarun Rana I am Corporate Senior Digital Transformation Manager in Henkel Global Supply Chain. Welcome to my presentation. I hope to give you some brief background on how we are generating value from big data. We termed it as end to end supply chain control tower. But before we go that, who we are. So Henkel started it’s journey back in 1876 it’s headquarters at Düsseldorf. We primarily operate in three business units, which is Adhesive Technology. We are number one there globally Beauty care and Laundry and Home care. I do Laundry and Home care business unit. So most of the use cases that you will see today will be belonging to that business unit. So [Indistinct] follow brands that you see Loctite, Schwarzkopf, Persil, et cetera. Now speaking about Laundry and Home care supply chain all the red countries are where we are operating our either manufacturing unit or warehouses. We have close to 7,000 colleagues worldwide, 33 factories 47 distribution centers where we serve the global market. Very quickly going on to our digital transformation the current state of penetration we have around 4,000 users of our Flight Deck Flight Deck we refer to our front-end tool, which is Tableau. We have around 28,000 excesses per day. So far in 2020, we have already saved around 25 million Euros in terms of savings coming from all the digital applications that we do we have more than 50 smart robots, always on and connected. We have around 4,500 sensors or IoT devices which are streaming data live to our Data Lakes. We have more than 10 billion data points which we are processing everyday. And in terms of data volume around 15 terabytes. Now we show you our district transformation landscape. When you speak about this topic you can immediately get lost various topics that come across. We primarily focus into four plus one areas. The four areas are Analytics, Visualization, Robotics, Sensorics and we will speak to you today about all these different areas some use cases, which will give you a bit of feeling how we digitalize our supply chain landscape. And of course the plus one End-to-End connectivity whether there’s a new use case new platform we try to embed it in our digital landscape, which is existing today. Now in terms of our supply chain, of course, you know, it starts from our suppliers then plant where we produce our products distribution center from where we ship our products to our customers and the customer themselves. Whenever we try to look at a use case, we look at the complete picture about our end to end value chain. We focus on all different four areas and we leverage our IT architecture and systems later on I will show you what architecture we follow what are the different physical and non-physical systems that we use. We follow a giant transformation culture. We believe without people, you cannot do transform anything. And on top you see a digital backbone which is the term that we refer to all our digital platform where we do our End-to-End analytics. We follow “Horizontal” and “Vertical” approach. What it means. We do POCs We do pilots. And if we find something which is very unique and can be scaled up we roll it out globally, quite fast. Now coming back to our digital backbone we started our journey back in 2013 not knowing that we are actually moving into a transformation journey. It was more to fulfill ISO 15,000 needs. Where we were asked to monitor our energy consumption across all our plants, across all our warehouses. And we decided that we will go online with this. Since then we have connected all our sites globally. As we already mentioned, we have various sensors IOT devices placed in the manufacturing area. We have online efficiency systems, quality systems. We have various machines which are streaming data life to 4G technology. We have more than 500 automated real-time dashboards in Tableau, which is serving all our colleagues globally and more than 50 machine learning pipelines in the cloud. Now coming to our four pillars Sensorics so this is what we started back in 2013, Real-Time Energy Metering System. We process here around 1 million data points per day global rollout, what it does is it collects and stream all the information back to our Data Lake. Very easy for our colleagues, especially in the sustainability area to benchmark different factories to each other. drill down to why different machines or let’s say similar machines in different factories are consuming more or less energy. What are the best practices that every factory follows and also share these best practices across different plants? The next example we have the latest edition that we included in our Data Lake of manufacturing is Digital Spray Tower Process Spray Tower Process is used to produce washing powders. And in terms of energy consumption, it accounts for around 60 to 65% of the global energy consumption that we do with our supply chain operations. very complex process, various factors. What we did with our Spray Towers, we install various devices like thermalcapacitors flow meters to judge to monitor different parameters that are operating the towers. These parameters were streamed globally from all our towers. We have 14 globally, and we have around 85% of them connected at the moment. And same as energy consumption. First of all, it allows you to benchmark different sites to each other. If for example, we have a golden batch which was produced. So it was giving you a maximum throughput and was produced with a very less energy consumption factoring in the weather effects and et cetera. We try to benchmark different sites to each other and try to improve what we are doing furthermore here, we are going into machine learning optimization approach We call it black box optimization where we are trying to find out optimal set of parameters to run a tower. Both from the area of energy consumption. So if you want to reduce your energy consumption what are the different set of process parameters to follow? And on the other side, the throughput if you want to increase your production output what are the different set of parameters? Keeping in mind, the boundary condition which is we want to operate our tower safely. The next example is coming from predictive maintenance. I think condition monitoring or predictive maintenance is a term that is widely used. We use here sensors, which are monitoring current voltage and other parameters of our motors online. You can imagine it’s a time series analysis. And if something is going wrong with the motors, you can do things like anomaly detection. You can recognize patterns and you can tell when your motors will enter into a breakdown or what is the time for a failure. Now, these motors, they drive various systems and various machines in our factories. For example, pumps compressor. I mean, think about a failure. If there’s an unplanned downtime because of an unplanned failure, it directly accounts for your efficiency losses in the plant. But if you smartly predict when a failure or breakdown is likely to happen and you organize your planned maintenance activities around it. So you make an unplanned downtime to a planned downtime, you can increase the efficiency of the line. This one is latest edition as well. We are piloting this in two factories in Western Europe. And I think so far we are monitoring 50 to 60 critical assets in our plants. If it goes well in terms of proof of concept, the business case justifies then we will scale it up quite first. The next example we have from Pouch Monitoring System, I think you are very well familiar with the washing powders that we produce, the washing liquids that we produce the latest technology of single dosage uses. So you just have a pouch, you open the pack take the pouch, throw the pouch in the washing machine. This is very innovative. And in terms of quality control, it is by far the difficult process. So what we do here, we have high resolution cameras. We are taking pictures of every plate. One plate generates around 20 pouches. So we are taking pictures of every plate. We are comparing this picture with, you can say a golden picture with highest quality possible. And we are trying to identify if there are any defects some of the defects could be, there is a much in the film, the two films are overlapping each other or there’s a leakage somewhere in the pouch. That is one topic which is supporting the quality control in-line quality control. The other topic is we store all these images in our data Lake. And then we can process these images with advanced pixel recognition system to try to find out what are the things that we need to check. Maybe some faults in the machine some faults in the camera, some plates need some maintenance or the plates are already old and try to reduce these scraps and production based here the scope is global and maturity is rollout and we have so far saved around 1.2 million Euros just by doing the pixel recognition technology. Next one in line is Demand Sensing. I think Demand Sensing or predicting or forecasting. I think it’s very easy to understand all the companies whether it’s FMCG company like us, or even companies like LinkedIn or Twitter, everybody is trying to sense demands. What customers will click Airbnb, what will be the demand at a certain location, Uber, what we are trying to do here we have our Demand Planning System which is there for last 10 years. This demand planning system is doing statistical evaluations and trying to tell us based on the customer ordering patterns in the past and taking into account things like seasonality cyclicity, this will be your demand for a particular SKU at this location. This technology was quite good five years ago but now we need, if we want to predict it accurately we need something on top. So what demand sensing technology does it takes into account first of all, the forecast which is emitted by our demand planning system it combines it with other pieces of information. For example, point of sales data, coming from our customers with the customers who are sharing the data with us, it combines it with other global phenomenon like weather information, economic parameters, financial parameters Covid is also quite a big one here. And it processes these demand signals and provide us a demand on every sq level and on every warehouse level. here, the scope is Europe, plus North America. We rolled it out last year in North America. And so far, we have not only seen significant benefits in the forecast improvement. We are targeting for every SKU plus minus 20%. But in terms of better forecast, you immediately capture this benefit in all the other areas of supply chain. We share with you here, inventory impact minus five days that we have realized if you convert this number into a Euro number, you will get somewhere around seven to eight million Euros of networking capital, which is actually your tied up cash in terms of inventory. In terms of Visualization the end to end KPI flight deck. So what we did, we connected all our operational systems whether it’s SAP passport management system warehouse management system, we connected all these systems to our data Lake and streaming this data directly to our visualization layer, which is Tableau. It is real time and what Tableau allows you you must be quite familiar with it. It allows you drill-down capabilities. So we are not only tracking real time. What is our KPIs but also why some KPIs are going down. So drill-down on single order level, drill down on single SKU level. This we started I think back in 2017, we have more than 90 reports which is not only tracking our golden triangle of KPIs, which is Cash, Cost and Service, but also things going beyond these KPIs. As you see here, we are tracking OAE. We are tracking quality of labels. We are attracting service level even the savings that we generate, et cetera. The next example, in terms of End-to-End connectivity I think you will be able to understand the concept of data silos. So earlier we had data silos in our planning system, we will use a lot of parameters which is pertaining to what happens on the shop floor. What happens onto the production line. We wanted to bridge that gap. So imagine just example, if we are producing, let’s say 200 bottles permitted on our line. And we are planning with this, so when we are doing production scheduling we are scheduling our lines with the capacity of 200 bottles per minute. Now, if this number is actually not the reality if the number is I don’t know, 150 bottles per minute in reality, or 180. So if it, or if it’s 250 230, if the number is overstated, then it leads to delivery delays or service level losses. Because in reality, you cannot produce what you plan. On the other hand, if this number is understated then of course it leads to high cost of production. And if you think about 33 production lines 33 factories, more than 400 production lines and we lose 1% OE everywhere it could lead us to even wrong investment decisions. The other topic is we are synchronizing. So synchronizing the End-to-End supply chain or synchronizing your production and planning. It’s quite a hot topic at the moment. We want to close the loop from our production to our planning system which would have an impact on all the supply chain KPIs. And we will steer the supply chain based on facts, not the assumptions. I wish I would do two examples now, which is by far the biggest implementation of big data. And it goes beyond the walls of supply chain. First one is a sales and operation planning or sales and operation execution. I think every FMCG companies for aligning their financial goals or aligning their strategy, they do a monthly cycle of S&OP. Now until last year before we rolled this system out people from different departments, Production, Planning, Marketing, Sales, Customer operations, they would all come together. They would discuss primarily what happened in the past, how the KPIs. So if it’s focused accuracy how was the focus accuracy two months ago, or one month ago? How is the inventory going? There was no forward looking approach and especially in S&OP or effective S&OP what you need to do is look forward because what happened in the past has already happened. Good to know, but the perspective should be forward looking. And it’s very important to tie everything back to the financial goals of the company, because at the end of the day, we need to bring profit into the bottom line. So here there are four or five components, which i mentioned on top. First of all, we financialized our demands. So whether the demands in our system are actually able to cover our financial goals or not. We check the demands on in short-term basis and mid term basis and long-term basis. Then we do the second cycle, which is supplier review. If our capacities in our supply network are sufficient to cover those demands or not as per our financial goals. And then the third one is product review which is very important. We will not doing this at all. At the moment, there were several teams which were doing portfolio analysis a complexity analysis. We built this bridge and tie up marketing into the sales and operation process. And we are doing now a complete product review. Whether our portfolio is healthy enough to drive our financial strategy to drive our financial gains. And we identify gaps in all of these [Indistinct] these gaps are then shared in the executive review where the decisions are taken on new innovative, new promotions that we want to run, et cetera. The second one that I want to show is Cost to Serve. So supply chain the biggest lever that we have is the cost of the supply chain and cost to serve can we loosely define into what are the costs that you incur to serve your customers? So it’s very important here that you break down the complete cost to serve to smaller components. Yeah. So cost drivers that leads and some of the cost drivers are, for example, whatever is happening in the warehouse in terms of picking in terms of handling of the pallet, then what is happening in terms of logistics transportation. So we’ve built several modules here which is giving you the complete scorecard End-to-End up the customer what are their ordering behavior, their ordering patterns. And then furthermore, we enhanced the scorecard of the cost drivers, we enhanced it with the predictions and prescriptive analytics. As you see, we build order size optimizer. So looking at the size of the order that the customer is asking for handling optimized and looking at specifically the picking cost or the handling cost of the warehouse co-packing activities the third one frequency optimizer frequency means how many times in a week you are sending the goods to the customer let’s say Monday, Wednesday, and Friday. And if you cut one day, it has a big impact in terms of cost. The last one is Network Optimizer. So maybe you are doing in several regions maybe you are doing capillary distribution. So you have a big customer, they have a smaller warehouses. You send one truck to a bigger warehouse. And then from that warehouse, you then send smaller truck or smaller weekers to different warehouse. And we tie up everything back to the cost drivers. So what are the costs for all the activities that we do? Furthermore as, this tool is built for negotiating with the customers we also include sustainability impact for example, carbon dioxide emissions. So if the customer is not ordering full track but only half truck what would be the carbon dioxide emissions that we need to that we incurred because we are not utilizing the truck fully. in terms of rollout. It is Europe and North America because the supply chains and supply network there are quite aligned. And in terms of maturity complete rollout. That those were the use cases and as I mentioned we will also speak about IT systems and infrastructures. We speak here about Henkel Data Foundation which is our virtual infrastructure, primarily consists of Azure Data Lake, where we store our data connected to all our operations systems all our machines and assets from the plant then data bricks, which is doing big data processing for us or streaming the data directly then dremio is used to watch analyze this data because you cannot read the data directly from the Data Lake. In the second, in the middle part, we use Front-end Tableau We are using Tableau since roughly 2015, with the startup the energy nutrient system. And now since last year we are also utilizing power BI as most of the other departments outside of supply chain they don’t have Tableau and for bringing them on board on any of the use cases that we have connection to them we need to build on power BI. The third one Kony is actually a mobile app development platform especially designed for our colleagues blue worker colleagues on the shop floor. They also need insights. They conduct various activities like shift handle overs, change overs, or cleaning inspection lubrication. And we are trying to digitize and trying to capture all this information from a mobile device. That data stream started again to our digital backbone. And of course, everything that we do connects back to factories and we required a very healthy physical infrastructure network capabilities servers, and 4G to bring everything together and any new use case that we bring as I already mentioned, any new platform it all ties up back connecting the physical and virtual infrastructure. Now in terms of transformation of people. So no transformation is possible by just buying a tool outside. We need to always keep along the development of the people development of the organization. We follow a very simple governance model. We have a small international team sitting in Amsterdam. We have 50 local digital single point of contacts or change agents, responsible for every factory or every warehouse where we operate. We meet with them regularly every six weeks. We follow top down approach in terms of our overall digital strategy about the technology standards the architecture, when it comes to ideas on what is the real need on the shopfloor different pilots, different POCs conducting rollout full scale implementation. We rely on our change agents to then take our ideas further and then roll it out. It’s very important to upskill the organization at the same time because we don’t build our tools for our consumption. So we don’t take care of any KPIs ourselves. It’s actually our colleagues. So we conduct regular webinars with them. We have a global supply chain Academy install and this is installed actually since last 10 to 11 years every new colleagues has to join this. We are of course, doing on the job trainings. And we have building now with HR, additional capability framework, leveraging networks in an efficient and agile way. Of course, in all our journey, we are not alone. We always partner with technology partners with conferences, with the consultants. We are very well connected with universities, especially in and around Amsterdam. We work with analytical partners, et cetera. And now we will share with you a video. We are very proud of it. We won award from World Economic forum this year for one of our factory, this lighthouse please enjoy. And then we open the Q and A session. Thank you very much.

Watch more Data + AI sessions here
Try Databricks for free
« back
About Tarun Rana


Tarun Rana is based in Amsterdam and have been with Henkel for 5 years. He holds a bachelor’s degree in mechanical Engineer, master’s in international business and an MBA. As Corporate Senior Manager in Digital Transformation, he leads topics around Big Data and Industry 4.0 in Laundry & Home Care business unit in Henkel. He is passionate about innovation and in continuous pursuit of ways to introduce digital tools to optimize end to end value chain.