Digital twins and simulation have emerged as powerful tools that have the power to significantly improve supply chain operations. They are particularly beneficial where product freshness is concerned. They enable supply chain managers to monitor and predict real-time status, test various strategies in low-stakes and non-spoil environments and optimize their operations. This article will explore the importance of digital twins and simulation in supply chain management and how they can create a competitive advantage.

Digital Twin or Simulation – What’s the Difference?

It is important to recognize that many people use the concept of a digital twin interchangeably with simulation. The reality is that it differs from simulation. According to ChatGPT, a digital twin is a virtual replica of a physical entity that mirrors its real-time status, working condition, or position. The critical difference is that it enables real-time monitoring, diagnostics and prognostics.

In the refrigerated world, think about constantly monitoring each trailer's temperature conditions, location, distance to unloading, traffic conditions, etc. – and making decisions based on playing that forward over time. Those decisions may include working the load into a dock schedule that has it slotted earlier or later. The key here is the ability to integrate transportation and warehouse decisions. It’s like having a living model of the operation or network that you can explore in-depth without interfering with operations.

Simulation is more strategic or tactical. It is a model of a real system upon which experiments can be conducted to understand its behavior or evaluate various operating approaches. It does not require a physical counterpart or real-time data to function.

Their similarity lies in their purpose: both are used to understand, predict and optimize systems to achieve better outcomes. However, while simulations are typically used for analysis and testing in the design phase, digital twins serve throughout the entire lifecycle of their physical counterparts. They complement each other, with simulations often feeding into the creation and operation of digital twins. Simulation in supply chain management predates digital twins predominantly because sensors and computing power have only recently made real-time monitoring economically viable. There are many applications of simulation in supply chain management.

Using Simulation to Determine the Optimal Location for Warehouses and Plants

The optimal location for warehouses and plants is a significant cost and service driver in supply chain management. Simulation has been used to determine the best location for these facilities, considering transportation costs, labor costs, proximity to customers, and access to raw materials. By simulating different scenarios, such as fuel-cost changes or volume growth, electric rates, etc., supply-chain managers can evaluate the impact of other variables on the location decision and make informed decisions that will optimize customer service, supply chain cost and capital picture while mitigating risk.

A supply chain simulation can show the behavior of a logistics network over time. Each successive period is dependent on the previous one. The logical rules of a supply chain are represented in a simulation model and then executed over time, making the simulation dynamic. This helps supply chain managers understand how the logistics network will behave in different scenarios and conditions. They can use this information to optimize inventory positions and prepare for possible eventualities. Simulating in small time increments, it is possible to understand the impact of inventory freshness on service.

Supply chain simulations can help design a robust network to handle demand fluctuations or other events. For example, Monte Carlo simulations can simulate various demand, weather, electricity, or fuel cost scenarios.

Using simulation to provide an objective function, mathematical programming can be overlaid to determine the best options, such as which warehouses to open or close. It is also possible to simultaneously consider multiple objectives, such as service, freshness, and cost. Additionally, the new field of reinforcement learning in artificial intelligence can use Simulation as a guide for determining how to come up with the best solution. This is just an extension of testing many scenarios in an automated way.

Supply chain simulations can also be used as educational tools. For example, the Supply Chain Game is an online supply network simulator in which students are divided into teams and compete against each other in assignments.

Using Digital Twins in Supply Chain Management

Digital twins, as people choose to define them, are slowly taking hold. Forty years ago, companies started to model industrial processes. On a computer screen, plant staff could see a schematic of a large and complex operation showing all the pieces of equipment and their operational status, as well as the real-time work-in-process inventories waiting to be processed. It was, and continues to be, an excellent means of bottleneck identification and operational management. Since its early days, this visibility has expanded to other parts of the supply chain.

Digital twins can be used to map transportation routes and optimize logistics. This is an extension of what we have on our phones today—Waze constantly updates the fastest way home based on traffic bottlenecks. Waste Management, the large trash hauler and recycler, has recently implemented this technology. They call it “Waze on steroids” and recognize significant productivity improvements. Refrigerated and frozen delivery, as well as other multi-stop shipments, will not be far behind.

Digital Twins for Planning and Forecasting

Some believe that digital twins can help with planning and forecasting. The concept is that monitoring the environment or sales can trigger action. For example:

• Watching the weather and traffic flows makes it possible to predict how much ice cream is sold at an individual store.
• As sales are recorded, the demand signal is sent to the factory to make a replacement unit.

Since real-time replenishment is not an immediate opportunity for most companies, the value of immediate information needs to be questioned.

While most warehouse management systems rely on a skilled manager to release work to be performed, a new class of digital twins has been developed that does this work. With a complete vision of all activities, tasks, inventory and labor availability, the twin can optimally determine the best course of action within the capacity constraints. Actions include:

• Reallocating inventory.
• Determining the best dock door to load/unload –for example, to facilitate cross-docking.
• Moving staff to where they are needed to meet deadlines or ensure the correct temperature/quality balance.
• Adjusting schedules to reflect truck arrival time data. The expected truck arrival times can be gleaned from visibility companies.

Since any action, such as changing a dock time, can impact other shipments and receipts, any updates must be holistic.

Many of the warehouse digital twin’s capabilities also help manufacturing, where line supply issues or the failure of a second or third-tier supplier can be mitigated. In both these cases, the reaction is generally not real-time but updated on a regular but short cadence, for example, every 30 minutes.

Digital twins with optimization can be used to rapidly scale capacity, increase resilience and drive more efficient operations. We talked about this above with the warehouse digital twin. The reality again is that this needs to run on a short-cycle-time cadence, not real-time. And that makes sense because events are generally discrete, and there needs to be a period of “stability” to ensure activities can be initiated and executed. Put another way, you can’t get staff or equipment to change what they do every two minutes. In most cases, they would likely achieve nothing. Applying optimization requires a steady state.

Consider supply planning, where replenishment plans have traditionally been created without concern for cost or operational constraints. In general, they are pretty volatile. There is a new breed of optimizing digital twins, effectively solving this problem. Working inside real-world constraints, the digital twin gathers a substantial lift of data and simultaneously optimizes all the flows in the network. The word simultaneously is important here; adjusting volume on any lane in a distribution network generally impacts other lanes. For example, if a receiving warehouse fills up and the manufacturing site that supplies that location must push volume, the volume must go to another warehouse.

Optimizing digital twins comes with its own set of challenges:

• Fragmented/dirty data landscapes make it difficult to obtain near real-time data, which generally comes from many sources.
• Lack of in-house talent that can understand and handle complex systems.
• In addition to the effort required when setting up a digital twin, the ongoing operational costs can be high due to the significant computing power needs.

Two Powerful Tools

They can significantly improve cold chain operations, enabling managers to monitor and predict status, test various strategies in a low-stakes environment and optimize operations. While simulation is used for analysis and testing in the design phase, for example, determining the number and location of coolers or freezers, digital twins serve throughout the entire lifecycle of their physical counterparts.

To achieve the full potential of digital twins and simulation in logistics management, cold chain managers must take a comprehensive approach to implementing and optimizing digital twins or switch to simulation, with or without optimization. This approach should involve strategic planning, skilled personnel and robust data management practices. By adopting digital twins and simulation, companies can create a competitive advantage in the industry and improve their operations.