Fleet routing is a problem that has been tackled by logistics companies for several years. Yet with the soar of e-Commerce and dynamic supply chain networks, this topic has become critical for a wide range of industries.
Accordingly, retail, logistics and mobility companies should investigate the role of advanced analytics and BigData, towards optimal fleet dispatching and routing, and workforce management.
What is the added value of advanced analytics comparing to standard approaches for fleet routing? What are the benefits of advanced analytics in terms of fleet management and staff costs?
This article aims at answering these research questions through a practical use case.
Use Case: Urban Logistics Operator
The covid pandemic has ignited the consumers' habits towards e-Commerce. Taking into account the increasingly high demand for daily delivery of consumer goods, the biggest logistics challenge is played at a city level, through decentralized supply chain management.
In this use case, one portraits Wiimer as a Urban Logistics Operator, having a local warehouse in Porto, Portugal.
Fig. 1 - Illustration of a urban logistics vehicle
I. Simple Vehicle Routing Problem
In a simple vehicle routing problem, the company's goal is to find optimal routes for multiple vehicles delivering client orders to a set of locations. In truth, this problem is a generalization of the Traveling Salesperson Problem, considering more than one vehicle.
The operating model is focused in minimizing the traveled distance by passing each destination (client) once. Although the benefit of having a simple model, it frames the problem of urban logistics in a too theoretical way, since it does not consider variables such as different demand levels at each destination and the fleet's capacity.
II. Vehicle Routing Problem with Time Windows
In this operating model, the logistics operator ought to commit the deliveries and pickup times for specific time windows. This constraint embed the clients' time slots availability to receive the order.
III. Vehicle Routing Problem with Pickups and Deliveries
In a Pickup and Delivery Vehicle Routing Problem, each vehicle picks up orders in several sites and drops them off in others. The problem is assigning routes for picking up and delivering all the items, while minimizing the length of the longest route. This is the typical operating model of on-demand urban mobility operators (e.g. taxi service).
BUSINESS CASE: VEHICLE ROUTING WITH CAPACITY CONSTRAINTS
For this example, one assumes Wiimer's office is the local warehouse, holding a fleet of 3 vehicles with an individual capacity of 10 packages.
Wiimer has to carry out a work order list that consists of picking up packages in 10 clients (different locations).
Fig. 2 - Wiimer's work orders
In the optimal routing algorithm one has to include the number of packages per client (i.e. location) and distance between sites.
Afterwards, Wiimer's dispatch center runs the algorithm to achieve optimal routes for the concerned fleet, seeking to minimize the total distance covered by the 3 vehicles.
Fig. 3 - Result of the routing algorithm (one map per vehicle dispatch)
One can conclude that the algorithm clusters the city in 3 segments and distributes the routes accordingly (although some routes have more pickups than others). This is indeed an illustrative case that demonstrates the potential of such an algorithm for urban logistics.
Still there is a missing point in this analysis: what is the added value of optimal fleet routing? In this use case one has assumed a vehicle capacity of 10 packages each. What if the logistics operator has the ability of designing the optimal vehicle capacity? What would be the result if the logistics operator has a vehicle capacity of 15 packages?
Fig. 4 - Result of the routing algorithm with vehicle capacity optimization
The increase of the vehicle capacity has led to a 4 km reduction of the total covered distance, and non-dispatching one of the vehicles (for the same service level). Although this is an optimal outcome for a given day, one could also perform a weekly, monthly and annual simulation (with work order forecasting). In short, these figures can be easily quantified (in fuel consumption, carbon emissions and workforce allocation) and the benefits monetized: lower fleet CAPEX or OPEX (e.g. leasing); vehicle fuel saving; and optimized staff costs (in this case, from three to two drivers).
Fig. 5 - Mapping decision-aid models for urban logistics
To conclude urban logistics could seize short-run benefits from advanced analytics, tackling the next levers:
Demand forecasting (of work orders);
Vehicle sizing and fleet investment planning;
Optimal fleet routing;