Logistics plays a key role in any supply chain as it pertains to the most important aspect of supply chain- movement of goods and products. That also means it is a major contributor to the carbon footprint a company generates.  With exponentially increasing focus on sustainable supply chains, companies today have access to a plethora of solutions to help them manage and run their supply chains in a sustainable way. With proliferation of Machine Learning (ML) technologies in the business world, there are opportunities to leverage ML algorithms to infuse sustainability in your logistics operations. This article shares some examples of how you can leverage ML to help support your sustainability quest in logistics.

Inbound and Outbound Transportation

As most companies today have a global footprint of suppliers, the carbon footprint of sourcing raw material and semi-finished goods from low-cost countries has increased. While it may be cost effective for companies, the carbon footprint attributable to inbound transportation has definitely increased for many companies. Unfortunately, despite all the claims of re-shoring or in-sourcing, organizations will keep sourcing globally, particularly from low-cost countries for a while. Analytics driven tools, powered by Artificial Intelligence (AI) and Machine Learning (ML) , can help companies optimize the balance among various elements like cost, footprint and performance. Therefore, despite the constraint of sourcing from global locations ,you can still leverage ML in a plethora of ways that help reduce carbon footprint : Reduce number of shipments: Algorithms can be designed to proactively balance ordering schedule and shipping schedule to minimize the number of shipments, specifically in case of ocean shipments. Depending on how granular your data is, you can incorporate aspects like parts packaging data so that the algorithm can accurately plan loads/containers. Optimal Mode Selection: In intermodal instances, optimal mode selection can be done by these algorithms, so as to minimize expedite shipments. Optimal Load Building: Sub-optimal load building not only results in more loads than required, but is also a driver behind transportation damages, which is another significant source of supply chain waste. ML technologies can help build loads that utilize the container or trailer space effectively, while ensuring product safety in transit. Optimal route optimization: On the outbound side, effective route optimization helps reduce miles and assets, among other metrics. And it is obvious that reduction in these metrics like miles traveled and number of shipments, helps reduce the carbon footprint. Traditionally, route optimization algorithms have been heuristics based, some of them that have been around for almost half a century. There are a few ways AI algorithms can help transform how we do route planning. The best approach in my perspective is to use a combination of Deep Learning (DL) and traditional heuristics, together with ML based predictive algorithms.

Inventory Management

Excess inventory is a major contributor towards supply chain waste. And while there is a procurement aspect to it which is not within the scope of this article, there is a logistics angel. Sometimes, the products may not have been placed or allocated to the optimal location. So overall, there may be a demand of the product in the network, but the product is placed in a region where the demand does not exist or is low. While there are traditional operations research based methods to optimize allocation, ML allows us to take this to the next level. Let us envision an example of an algorithm that leverages demand sensing to not only perform an initial allocation but does periodic review to evaluate opportunities to move inventory, balancing with cost implications of that movement, like transportation. There can be many variations of this algorithm. The key here is that ML can be really effective and can help meet sustainability goals by reducing excess inventory. In my opinion, no-one ML algorithm can present a complete solution but a combination of approaches, like reinforcement learning, supervised ML algorithms and DL, combined with some existing traditional operations research methods, have the capability to transform inventory planning and management.

Warehousing

Another driver behind wastage in the form of excess inventory is suboptimal warehouse planning. One aspect is the one already discussed in the inventory management section-Inventory is not placed at optimal warehouse location. Sometimes, warehousing footprint itself is sub-optimal but that is not within the scope of this article. A lot of product goes to waste due to damage incurred in warehouses. An ML algorithm can be designed to "learn" from the historical damage data and reduce these damages through suggesting improved rack designs, product placement, pick and/or store flow paths etc.

What Does This Mean for SAPinsiders?

For those who leverage SAP technologies, the good news is that SAP products are designed to help companies achieve their sustainability goals throughout the life cycle of products, from product design to reverse logistics and recycling. When it comes to building the capabilities described above, Intelligent technologies portfolio that comes integrated with SAP BTP , along with additional SAP tools can help you build these capabilities.