News

Using AI to Optimize the Distribution of Sensitive Goods

News

Using AI to Optimize the Distribution of Sensitive Goods

Article

October 22, 2019

Next to distributed ledger technology (DLT), colloquially referred to as “blockchain”, and Internet of Things (IoT), Artificial Intelligence (AI) is the third key disruptive technology which is impacting the supply chain. Learn how AI can support you in optimizing the distribution of sensitive goods.

According to Gartner, AI is one of the key trends driving the competitive advantage in supply chain management. Using AI methods to generate actionable insights from large amounts of data and support the decision-making process, cannot only reduce the effort and cost involved in investigating non-conformities and deviations along the distribution chain, but can also help to mitigate risks by predicting events before they happen. The range of benefits is quite wide, some thereof being

  • Process improvements through intelligent issue detection
  • Liability identification based on automated root-cause analysis
  • Increased delivery flexibility from dynamic route planning
  • Better customer experience by optimizing delivery options on behalf of the customer
  • Improvements in supplier management through a clearer view of data driven automated performance analysis

Model of automated and risk-aware optimization

Image 1: Model of automated and risk-aware optimization

Modum provides an operational optimization engine running independently based on trusted data, which also takes a company’s risk appetite into account. It is important to include risk management into the model, as extreme optimization can often lead to a higher risk profile. In the case of sensitive goods distribution, this engine could be autonomously managing delivery schedule and modes, packaging selection, supplier evaluation and customer satisfaction regarding delivery.

Combining insights from past events (root-cause analysis) and future possible events (predictive analytics) with heuristics and rule sets provides the basis for machine generated suggestions, which simplifies and expedites decision making (prescriptive analytics). Based on the risk appetite of an organization, such machine-made suggestions can be further automated by enabling execution without human approval.

Shipment showing stationary segments identified using machine learning provides context to the temperature curves. The track & trace info is provided by the carrier. (1) indicates steep temperature increase due to unplanned stationary stop with vehicle most likely directly exposed to the sun.

Image 2: Shipment showing stationary segments identified using machine learning provides context to the temperature curves. The track & trace info is provided by the carrier. (1) indicates steep temperature increase due to unplanned stationary stop with vehicle most likely directly exposed to the sun.

In one practical case, using pattern recognition techniques to provide more context to the temperature curve of a shipment, we can identify deviations to the standard process which correlate directly to sharp rises or drops in package temperature. This is particularly relevant for shipments requiring ambient conditions (e.g. 15 degrees - 25 degrees), where, mainly due to cost pressure, goods are often shipped using only the external temperature forecast as factor to reduce the risk of temperature deviations.

Sample shipment showing correlation between predicted and actual temperature of package. A system based on prescriptive analytics can use package temperature predictions to suggest optimal packaging for a given shipment and its requirements. Predictions are made using a machine learning approach.

Image 3: Sample shipment showing correlation between predicted and actual temperature of package. A system based on prescriptive analytics can use package temperature predictions to suggest optimal packaging for a given shipment and its requirements. Predictions are made using a machine learning approach.

In another example, this time using predictive analytics, the customer experience can be enhanced and packaging costs reduced by having the system suggest the best suited shipment packaging based on a prediction of the package temperature over a given delivery window. The next step is to have the system choose the optimal packaging, route and carrier independently for a given shipment based on the user’s (e.g. customer) requirements, such as delivery date, goods value, product stability data etc.

Interested to learn more on how to use advanced analytics to gain insights and optimize your supply chain? Contact us today to book a Discovery Workshop and take a step further in digitalizing your supply chain.