Better way to Optimise Logistics in Supply Chain

Traditional Route Optimisation: Focusing on Distance or Cost

Traditional route optimisation, particularly methods rooted in graph theory and operation research, revolves around minimising travel time or distance. Algorithms such as Dijkstra's algorithm, A* (A-Star) search algorithm, and genetic algorithms can quickly compute the shortest path between a series of nodes (i.e., delivery points). These techniques are fundamental for solving classic problems like the Travelling Salesman Problem (TSP), where the goal is to find the shortest possible route that visits a set of locations and returns to the origin point.

However, these algorithms operate on a fundamental assumption: that all delivery points (customers) are equal in importance. This can be a shortcoming in the real world, where customers vary not only in location but in demand volume. This difference in demand should ideally affect how we prioritise and allocate resources.

Weaknesses of Traditional Optimisation:

• Overlooked Demand Differences: When demand varies significantly between customers, optimising solely on distance might lead to less efficient resource use. A customer that is close in proximity but has little demand may be serviced before a distant customer with much higher demand.
• Resource Misallocation: Traditional approaches may result in fuel and time being disproportionately allocated to low-priority or low-demand customers.
• Profitability Imbalance: Focusing only on reducing travel time or cost may not maximise revenue because higher-demand customers, which contribute more to the bottom line, may be underserved.

Problem Context: The Complexities of Real-World Logistics

In the real world, logistics and supply chain operations involve much more than fixed routes with a handful of customers. Instead, you face a highly dynamic environment with:

• Variable customer demand: Different customers will have different levels of demand on any given day.
• Shifting delivery points: Routes are not static and often change daily or even hourly based on new orders, cancellations, or unforeseen events (e.g., traffic, weather conditions).
• Operational constraints: You may have limited resources (e.g., drivers, vehicles) and must balance time windows, fuel costs, and vehicle capacities while delivering.
• Unpredictable scenarios: You might be adding or dropping customers on the fly, dealing with last-minute customer requests, or addressing sudden delays.

Traditional route optimisation, which focuses solely on minimising travel distance or reducing fuel costs, falls short because it doesn’t account for variability in demand and the value of each delivery stop. It optimises for the shortest path or lowest cost without considering how much demand is fulfilled at each step.

Traditional Optimisation: Limitations in a Dynamic Setting

Traditional route optimisation methods such as Dijkstra’s Algorithm, Clark-Wright Savings Algorithm, or genetic algorithms typically aim to minimise total distance or time. They work well in static or deterministic environments but encounter several challenges in dynamic settings:

1. Demand Blindness: Traditional methods treat all customers as having equal priority or value, regardless of the demand at each stop. In reality, the value of delivering to a customer differs, with some customers having much higher demand (e.g., more items, higher revenue potential).
2. Capacity Underutilisation: When optimising purely for distance, a vehicle might visit low-demand customers first, leading to inefficient use of vehicle capacity. This can result in more trips, more fuel consumed, and wasted driver hours because high-demand customers are deprioritised.
3. Profit Inefficiencies: Optimising solely for distance can neglect profitability. For example, a short trip to a nearby low-demand customer might seem optimal by distance, but visiting a higher-demand customer farther away might generate more revenue per mile, maximising overall profitability.
4. Difficulty in Handling Dynamic Changes: As customer locations and demands change dynamically, traditional static route optimisations struggle to keep up. The route planned at the beginning of the day might no longer be optimal by midday when demand or route conditions have changed.

Demand Per Mile Optimisation: A Real-World Solution

As a chief optimisation scientist, it's crucial to incorporate both demand and distance into your route planning to better tackle the challenges of dynamic logistics. Demand per mile optimisation addresses these shortcomings by ensuring that every mile travelled delivers the most value in terms of customer demand. This methodology makes the route adaptive, balancing the trade-off between distance and demand to ensure that resources are used as efficiently as possible.

Here’s why demand per mile optimisation excels in real-world scenarios:

1. Handling Dynamic and Variable Demand

In logistics, demand can vary significantly from day to day. With demand per mile optimisation, demand data is used as a core component of route decisions. By factoring in customer demand, this method helps prioritise customers that yield the highest fulfilment value for each mile of travel.

• Example: Let’s say on Day 1 you have the following customers: A, B, C, D, and E. On Day 2, the route might include A, E, F, H, and J. Demand per mile ensures that, regardless of the changing route, your fleet is always prioritising customers who provide the greatest demand per unit of distance. This is especially valuable when your operations involve many small deliveries interspersed with high-demand stops.
• Dynamic Adjustment: Demand per mile allows for real-time dynamic adjustments based on fluctuating demand. For example, if new orders come in during the day, the route can be adjusted on the fly to ensure high-demand stops are prioritised.

2. Improved Resource Allocation

By focusing on demand per mile, you ensure that your vehicles and drivers are allocated more efficiently. Vehicles are less likely to make trips that fulfil small amounts of demand, which would be inefficient when considering capacity constraints and fuel costs.

• Optimised Vehicle Utilisation: Demand per mile ensures that high-demand customers get served first, helping to maximise the capacity of each vehicle before heading to lower-demand stops. This reduces the total number of trips required and improves fuel efficiency, thereby cutting operational costs.

3. Increased Profitability

Demand per mile optimisation directly impacts profitability because it maximises value delivery for every mile travelled. By prioritising high-demand customers (who may be more profitable), the business can focus on serving customers that generate more revenue per trip.

• Revenue vs Cost: While traditional methods may minimise costs by reducing distance, they often fail to optimise for revenue generation. Demand per mile balances the cost of travel with the potential revenue gained from serving high-demand customers first. This results in higher overall profitability.

4. Flexibility and Adaptability

Demand per mile optimisation thrives in dynamic environments. When routes change frequently, the algorithm adapts to optimise routes for both distance and demand on a day-to-day or hour-to-hour basis.

• Example: Let’s say on Day 2 you have customers in new locations. You can quickly recalculate the demand per mile for each possible route and dynamically adjust to prioritise high-value customers, even if they are farther away. This makes your routing strategy highly adaptive, ensuring you’re always making the most efficient decisions as new data becomes available.

5. Risk Mitigation

Another benefit of demand per mile optimisation is the ability to mitigate risks such as delays, traffic congestion, or vehicle breakdowns. Since this method focuses on maximising demand fulfilment earlier in the route, it reduces the risk of under-serving high-demand customers if disruptions occur later in the day.

• Example: If a delivery truck breaks down or there’s an unexpected traffic jam, demand per mile optimisation ensures that the most valuable customers (in terms of demand) have already been served, minimising potential losses.

Mathematical and Computational Insight

Demand per mile optimisation can be formalised as an objective function that combines demand weighting with distance. Let’s assume:

• Di = Demand at customer i
• distij = Distance between customer i and j

The goal is to maximise the cumulative demand per mile and minimise cumulative distance:

max∑i(Didistij) & min∑i,j(Djdistij)

This objective function can be solved using metaheuristics like Simulated Annealing, Genetic Algorithms, or Ant Colony Optimisation, which allow for dynamic adaptability in the face of changing customer locations and demand. These methods perform well in large, complex, and variable real-world scenarios.

Real-World Application: How Demand per Mile Delivers Value

Let’s consider a real-world example. Suppose a large distribution company operates with a fleet of delivery trucks, serving hundreds of customers per day. The company starts with traditional distance-based route optimisation:

• They notice that trucks are frequently returning with unused capacity because some trips, while short, fulfil only small customer orders.
• Higher-demand customers are often served later in the day, leading to missed revenue opportunities when unexpected delays occur.

By switching to demand per mile optimisation:

• The company begins to serve high-demand customers earlier, ensuring that their trucks are fully utilised before returning to the warehouse.
• They reduce the number of trips overall by maximising the efficiency of each mile travelled, saving on fuel and labour costs.
• Profit margins increase as they prioritise high-value deliveries that contribute the most revenue.

Conclusion: Demand Per Mile is the Key to Unlocking Efficiency in Dynamic Logistics

In today’s complex logistics and supply chain landscape, traditional route optimisation focusing purely on distance falls short when customer demand and delivery points are highly variable. Demand per mile optimisation offers a superior solution by balancing distance and demand, enabling companies to:

• Maximise efficiency and resource utilisation
• Boost profitability by prioritising high-demand customers
• Remain flexible in a dynamic environment
• Reduce operational risks

Call to Action

To fully unlock the potential of demand per mile optimisation, it’s critical to leverage Trusted Data Technologies 's platform Optihub and Grigora . Implement real-time data collection, predictive analytics, and dynamic optimisation algorithms to ensure that your logistics operations remain adaptable, profitable, and highly efficient.

Take the first step towards transforming your logistics by adopting data-driven, demand-centric routing strategies—because in the world of modern logistics, it’s not just about how far you travel, but how much value you deliver with every mile.