Alpro optimizes planning for its plant-based alternative
As published in Business Logistics Magazine, December 2018;Axon trawls through data in search of effective quality release time When it comes to fresh products, every hour counts. Because the fresher a product is when it arrives in the shops, the better. Alpro, manufacturer of plant-based alternatives for dairy products, knows this too well. The firm […]
As published in Business Logistics Magazine, December 2018;
Axon trawls through data in search of effective quality release time
When it comes to fresh products, every hour counts. Because the fresher a product is when it arrives in the shops, the better. Alpro, manufacturer of plant-based alternatives for dairy products, knows this too well. The firm wants greater visibility over its network and is looking for ways to optimize its planning. To this end, it has teamed up with Axon, Bluecrux’s digital supply chain platform. As proof of concept, the quality release time has been analyzed, which resulted in insights into the real value and variance of this parameter. This has led, in turn to more accurate safety stock calculations. In addition, operational areas for improvement have been identified. For example, it emerged that there was a significant difference in timing between a product that was produced in the factory in Wevelgem and the same product that was made in Kettering (UK). What is more, Alpro is convinced that even more intelligence can be gleaned from its data.
The range of yogurts or fresh drinks in the refrigerated section of the supermarket is almost endless. We pick out our favorite brand and flavor and quickly check the expiry date. If we have the choice between a pot whose use-by date is still two weeks away and one that needs to be consumed within a week, we opt for the one with the longest shelf life. So the message to the manufacturer is clear: make sure that your products arrive at the shops as fresh as possible. Alpro is all too aware that effective planning plays a crucial role in this, which is why they are currently implementing the Plus version of OM Partners’ supply chain planning tool – OMP Plus. The operational planning is already live, and the Sales & Operations environment is scheduled for launch before the end of this year.
In order to optimize the planning of the finished products and ultimately be able to make better decisions, the target stock must be fed into OMP. This parameter is calculated on the basis of the service level, demand uncertainty, and lead time. The first two elements are closely monitored, but the actual lead time (and the quality release time in particular) is something that Alpro has little insight into.
To get to grips with this, Alpro has been working with Bluecrux since autumn 2017, a supply chain and operations consulting bureau that has developed the digital platform Axon, a platform on which to carry out digital supply chain planning.
Peter Decroos, Int. Planning Process and Tools Manager at Alpro, explains the purpose of the collaboration with Axon: “Our ERP system – SAP – contains a huge amount of transactional data. Unfortunately, it is no simple matter for us to get information out of this database. With the proof of concept that we have developed with Bluecrux, we are examining whether we can extract intelligence from our data with Axon. Now we have focused on a small part of the lead time, the quality release time: the time between the production of a pallet and its release by the quality department. But at a later stage, we also want to collect more information about the total lead time and to know when a production order was planned, when the planned order was created, when it was released to become a process order, when it was packed, when it was finally released, and even when it left the warehouse for its final destination.”
Quality release time as proof of concept
Before an Alpro product arrives on supermarket shelves, it has already come a long way. Alpro is best known for its soya-based products, but for some time, it has also been processing other plant-based ingredients such as almonds and hazelnuts. If we look at the production process for soya drinks, we see that a bean-based formulated product is made, the semi-manufactured product. This is ultimately packaged by one of the numerous filling machines.
A few days after packaging, the quality department checks whether the pallets can be released. The department checks the flavor and whether the product has been correctly packaged. When part of a production run is not released at the expected time due to additional checks, the subsequent shortage can be met using safety stock. This is calculated based on the lead time, the time it takes from making a planning decision to having that volume available in the warehouse. This is why it is so important to identify this lead time so that the safety stock (also known as target stock) can be correctly determined.
It is immediately clear why, in the proof of concept, Alpro and Axon are focusing on the quality release time forecast, known as LeadTime ForMore. The manufacturer wants to know how many products per product group and type are released at the expected time after checks and how many are delayed. If there are variations between theory and practice, then the logical next step is to adjust the parameters.
Peter Decroos: “You can enter parameters based on theory, but the reality is often different. SAP contains a considerable amount of transactional data (which reflects the reality) and static planning parameters, but these types of data are not aligned with one another.”
In SAP, a product is given Q status, which stands for quality after production. After a few days, the product can be released by the quality department, after which a member of staff actively needs to remove the Q status. As long as a product has Q status, the product cannot leave the warehouse. These status changes are registered as material movements within SAP, and it is this data that Alpro supplies to the Axon program in the cloud. For the proof of concept, this involved 4.55 gigabytes of data in the period from January 2015 to February 2018.
Network crawler trawls through masses of data
Around three years ago, Bluecrux invested in its own platform, Axon. The platform always sets up a virtual twin for its projects so that the physical supply chain can be presented digitally. Businesses no longer need to describe their distribution centers, factories, production lines, and possible customers or suppliers because this information is directly extracted from the data. Therefore time no longer needs to be invested in describing the supply chain network or – crucially – in keeping it up to date. The virtual twin forms the basis for the calculation of smart parameters. For this, Axon unleashes a ‘network crawler’ on the transactional data.
In Alpro’s case, this transactional data comes from SAP. Valérie Vandenbroucke, Product Manager at Axon, regards SAP as a system of records, a system that keeps track of all the information. “We unleash the digital crawler on the transactional data from SAP”, she explains. “Smart algorithms and machine learning enable us to construct the virtual twin and calculate the actual quality release time. Next, we compare the current planning parameter to that suggested by Axon. We regularly send the latter through to OMP, an advanced planning system that can ultimately generate smart plans with accurate parameters. The difference between SAP and OMP, and our augmented platform Axon, is that we try to fill in the white spaces and thus move towards an autonomous supply chain. One of these spaces is, therefore, smart parameters, whereby we allow other systems such as SAP and OMP to work with dynamic rather than static data.”
The advantage of the digital platform is that when it has access to SAP, large quantities of data can be calculated in real-time. You no longer need to describe in advance what your supply chain looks like, and there is no need to think out what you want to be able to see, as is the case with older technologies. A large amount of data is calculated in advance. By the time everything has been calculated and thought out, the reality is completely different. The platform in the cloud also works incredibly fast and can already deliver results after a week. However, once the results are there, they do need to be discussed and interpreted and may need to be fine-tuned.
The difference in quality release time in Wevelgem and Kettering
It was apparent from the analyses from Axon that there was a difference in quality release time for a product that was manufactured in both Wevelgem and Kettering. Nevertheless, all the parameters in SAP were the same. Closer investigation revealed that one particular task was executed every hour in Wevelgem, whilst in Kettering, it only took place every twelve hours, always at the start of a new shift. Peter Decroos explains what the consequences of this could be: “You can easily lose an average of six hours on the standard lead time. It may happen that certain departments achieve local process optimization, but that might be a disadvantage for other departments, such as planning.”
At first glance, six hours may not appear to make a big difference, and for some products with a shelf life of more than nine months, that is indeed the case. But when it comes to products with a shorter shelf life, like the section of the Alpro range that you find on the fresh produce shelves, the issue takes on a different dimension. Each additional day in the network means one day less that the product is on sale in the shop. Peter Decroos: “Imagine that the release of goods in the factory is extended by one day to reduce time pressure. And in the distribution center in Wevelgem, the decision is made to take an extra day to allow more time for loading lorries. This is a whole chain in which many of the links would like to build in a buffer, but at the end of the chain, the difference can add up to several days.”
That is why it is important to have the correct parameters for the service level, costs, and freshness. Alpro continues to focus on freshness by keeping buffer stock at the right level and manufacturing the products with sufficient frequency. Peter Decroos: “We did make efforts to offer fresher products in shops, but that move had no effect at the retail level. Research showed that our efforts were negated by the higher buffer that the logistics provider built-in. The retailer turned out to be offering bonuses to the service provider if it succeeded in improving service. So, the service provider was delighted with the fresher products it received because that enabled it to increase its buffers in order to offer a better service. So, we need to look beyond Alpro itself and take account of the entire supply chain.”
Co-creation relies on sharing data
Businesses are often reluctant to share their data, but the Axon platform is well protected. Moreover, a confidentiality statement is signed, and of course, a certain amount of trust is required too. The collaboration between Alpro and Bluecrux is a co-creation in which a huge amount of information is exchanged.
Valérie Vandenbroucke: “We do have insights from the supply chain, but we don’t always know how a planner wants to look at the data. For example, Peter Decroos asked us about a number of specific visualizations and correlations with other data elements which we would not immediately have thought of.”
This means that the consultants are constantly learning new things. Alpro is making Bluecrux smarter, but the opposite is also true. Peter Decroos, who now has twelve years’ experience at Alpro, is not afraid of sharing his knowledge with Bluecrux: “Readers of this article can also learn from our project. I think that you achieve more by sharing and searching for solutions together than by concealing your data from everything and everyone.”
Toward a systematic refresh of all the data
Alpro aims to use Axon in two ways. On the one hand, it wants to attune its planning to reality, but on the other hand, it does not want to be limited by what is regarded as the truth. It wants to challenge this truth by asking how real data can provide even more intelligence. For example, in the future, Peter Decroos would like to examine the loading of lorries. “SAP knows exactly how many tonnes of goods are loaded onto a lorry,” he explains. “Based on this data, the average load factor of a lorry during a journey from – let us say – Wevelgem to the Netherlands can be calculated. We can then examine how close we are to the limit and whether it might be possible to load an extra half pallet or even a full pallet onto the lorry.”
Measurement is the key to knowledge, but Alpro still needs to spend too much time making these measurements. If all the data held in SAP could be supplied to Axon on a weekly basis, then Alpro would no longer need to spend time making measurements. Valérie Vandenbroucke: “Today, there are many advanced planning systems (APS) available, but as a planner, you still need to supply the data yourself. And by the time you are able to do anything with the results, they are often out of date already. In the future, the computer will send the analyses to the planner, who will then only need to look at the results. More time will be freed up for activities that add value and informed data-driven decisions.”
A possible next step might be to carry out simulations. This would make it possible to map the supply of finished products, the lead time of raw materials, any stock held by suppliers, etc. Peter Decroos: “It would be great to no longer have to rely on your gut instinct, but to be able to go to Sales & Operations Planning with evidence. To use simulations to unleash a sensitivity exercise on our buffering strategy for our raw materials. SAP knows our target stocks and the lead times that have been agreed with our suppliers, but what increase in sales over and above the forecast can we absorb with this? Up to what percentage on top of the envisaged forecast is our buffering strategy designed to work?”
Valérie Vandenbroucke: “As well as the transactional data, data from suppliers, for example, will also be incorporated into the Axon platform. OMP will need to be supplied with parameters that are as accurate as possible. And this won’t just happen once: we need to do it time and time again so that greater value can be extracted from OMP. There needs to be constant monitoring of whether the supply chain can follow the planning and the forecast. If not, adjustments need to be made.” Peter Decroos: “At this point, we want to build strong foundations for our parameters. That is the starting point. But ultimately, this is a story that never ends. Over time, a systematic refresh of all the data needs to take place because the world is constantly changing. In the supply chain, it is a challenge to find the right balance between cost and service. We want to use data analysis to be able to make objective choices.”