Artificial intelligence (AI) and machine learning (ML) algorithms have a significant new role to play when it comes to optimizing transportation, logistics and delivery. Getting the right product in the right quantity at the lowest price sounds simple in theory, but it comes with data flows too large for human operators to manage, constant disruptions in distribution chains, and fluctuating fuel prices. , the existence of multiple suppliers for the same product, and ever-changing and unpredictable levels of consumer demand.
Therefore, all areas of logistics are leaning toward machine learning (ML) to predict future inventory needs. Machine learning (ML) is the branch of AI that makes machines smarter by feeding them data so they can “learn” from it what to do. But nowhere is the need for ML more acute than in shipping. Here is one practical application of AI best practices for transportation. Predictive maintenance and spare parts management.
Optimization of parts management
This case study focuses on the need for predictive maintenance on ships and is related to component optimization work with a new oilfield drilling and exploration company. The company uses a vessel called FPSO, which stands for Floating, Production, Storage, Offloading. FPSOs are vessels used in the petroleum industry far from shore where oil and gas pipelines cannot reach. Spare parts management for this type of vessel must take into account that the vessel is a traveling warehouse with very limited space.
The company’s main objective was therefore to avoid stock-outs, to increase the availability of spare parts and to avoid the so-called “dead stock”, i.e. the storage of materials that unnecessarily take up space on board.
ToolsGroup began by conducting a preliminary audit by collecting and verifying spare parts and vessel master data, inventory levels, consumption history and other statistics on parts consumption and demand.
Next, we developed an artificial intelligence algorithm that responds to “what if” maintenance needs beyond conventional preventive maintenance. So the AI we employed played a role in enabling forecasting and scenario planning. In doing so, we have effectively created a new business model for shipping companies, giving them better control over the process of forecasting the required spare parts for each ship, taking into account all logistical constraints. bottom.
This process started by analyzing the current performance of these FSPO vessels, but it also included a completely new set of business analysis and best practices that allowed us to capture “what if” scenarios and evaluate different options for resolving them. I was able to propose a transformation model.
Traditional preventive maintenance is usually an assessment of all factors related to periodicity and past events. However, by considering multiple contingencies, the system was able to anticipate the need for specific replacements beyond normal maintenance, as well as anticipated failures and time-bound obsolescence. AI can be used in this way to predict or predict which spare parts will be needed and which spare parts should be proactively kept on hand, optimizing inventory levels and spare parts transportation. I made it. In this case, we have developed a form of machine learning with self-adaptive and self-learning algorithms specifically for the maintenance, repair and operation of these vessels. The system can also calculate advanced consumption forecasts for parts. Therefore, by optimizing the inventory levels of machine spare parts and consumables and setting the inventory levels accurately based on predictive algorithms, in addition to avoiding running aground at sea, the needs of vessel safety and convenience are met. I responded.
The supply chain planning software adopted by the shipping company took a step-by-step approach. That is, we introduced implementations in a conscious order, gradually replacing old systems, processes and methodologies. We used probabilistic forecasting and machine learning technology designed to work together seamlessly and automatically. ML engines have started with historical data about demand and applied machine learning technology to existing historical data to improve baseline probabilistic forecasts. This allowed us to generate a more robust and reliable baseline forecast that accurately models the phenomena that shape demand. The tool then leverages additional external data sources to layer more advanced machine learning.
That said, ToolsGroup’s experience shows that predictions are not entirely based on machine learning techniques. Instead, a solid statistical backbone is needed to deal with the changing and often random nature of demand. In this case, we recommended using a hybrid approach that employs probabilistic forecasting and machine learning techniques that seamlessly and automatically work together.
To do this, we introduced a self-adaptive model for probabilistic forecasting using detailed historical demand. For this carrier and others, this approach has proven to be critical to their success when using advanced machine learning, and provides significant benefits in and of itself. Applying machine learning technology to existing historical data further improves probabilistic forecasting, resulting in a more robust and reliable baseline that accurately models the phenomena that shape demand. From there, the system can perform more advanced machine learning using external data sources such as weather forecasts, nautical indicators, availability at distributors and stores, social media and online searches, and the Internet of Things. increase.
The machine learning engine therefore improves the calculation of the factors that influence demand. For this transportation company, ML produced more accurate forecasts into the future, resulting in reduced costs, optimized inventory of required parts, and reduced risk of downtime.
Quantitative, Qualitative and Green Benefits
Besides helping solve some common industry problems, optimizing the delivery supply chain also has wider implications. In the project described here, the benefits were more quantitative than anything else, as inventory optimization happened at the same time as waste reduction. This approach also avoided the existence of two common risks in logistics: out-of-stocks and excess inventory. There are also qualitative benefits. For example, better planning reduces downstream interventions (and thus costs from renegotiation with suppliers). Finally, efficiency gains are the source of sustainability, which is determined by both waste reduction and the reduction of potential toxic events. Enhanced forecasting forestalls corrective action that could address additional, and therefore more costly, and polluting transportation.
In general, one of the strengths of AI-powered technology is its ability to process multiple demand variables and automatically generate reliable demand forecasts. This “self-tuning” approach allows the system to predict demand behavior much more accurately than considering demand history alone. Supply he chain experts understand the importance of accurate demand forecasting, but modern demand planning is so complex that this is a daunting task. Increased forecasting complexity and rapid changes in consumer demand are often compounded by a myriad of causal factors such as seasonality, new product introductions, promotions, weather and social media. High-level automated machine learning is an ideal application for improving forecast accuracy in supply chain planning. ML also supports the development of more resilient supply chain planning practices to ensure the system as a whole responds to changes and disruptions in a timely manner. Companies using ML-enhanced supply chain platforms can leverage real-time data to take immediate action and become more resilient and future-proof.
Author’s note: This case study was presented at the recent conference “Digital Infrastructure and Predictive Logistics: Strategies, Risks and Opportunities in Transportation Supply Chain Data Exchange” held in Genoa, Italy. This event was sponsored by the Logistic Digital Community. His Confcommercio-Conftrasporto initiative in collaboration with Federlogistica and Consorzio Global.
