I. Introduction: A Brief of Logistics and E-commerce Relationship
The world is continuously evolving and has become more and more interconnected than before indeed. This increment in interconnectivity is triggered by many factors, but the most apparent one is arguably the e-commerce boom. The convenience of staying at home while having your orders delivered was modestly appreciated, even when it gained some momentum in the early 2010s. In late 2019 and early 2020, however, the momentum was then driven by the pandemic situation that had an impact on the increment of online sales for retailers, while having their offline sales fallen drastically at the same time.
According to Statista.com, Global retail e-commerce sales 2014–2023 see a comfortable increase which peaked in 2017–2018. From 2019 to 2023, however, the sales increase is forecasted to be an average of 17% with low variance around the mean, indicating a constant increase even in an ever-changing business landscape. Meanwhile, Indonesia’s e-commerce revenues almost quadrupled in 2020 after only 3 years, reaching 30,309 million USD, showing strong growth. E-commerce success in Indonesia relies on the fact that it plays a huge role as it facilitates small-medium businesses to find more customers with ease.
Ultimately, this growth in online shopping through e-commerce sites also boosts the demand for logistics. Indonesia is an archipelago with many islands, making logistics more essential to support this growth. According to Badan Pusat Statistik Indonesia (Statistics central agency of Indonesia), the Logistics and warehousing segment received the fastest rebound compared to all other business segments. It had a 0.85% increase in the GDP proportion of Indonesia, almost 4 times higher than the rebound of the segment in the 2nd place, accommodation and FNB (food and beverages) at 0.22%. These results justify the hype around logistics and spark interest in how to implement IoT in this field.
II. Implementing IoT in Logistics for E-commerce
Image 1. Route planning of a Vehicle with an IoT Device Attached.
Visibility — possible to know the location of each vehicle and its route in real-time. Geo-fencing also helps in alerting in case of misdirection. Visibility eases supervisors’ work in monitoring the activity of their assets.
Traceability — Each route is registered and recorded. By collecting data such as average speed, average time spent at certain drop points, average ton/route, vehicle status, and other pre-defined metrics would help an optimization algorithm to suggest better and more actual route recommendations.
Competitiveness — it is widely understood that logistics cost is more important for low-value items. Since e-commerce sites allow customers to buy in small batches with wide varieties, understanding how much to charge will help logistics providers to profit while still considered by customers. The data earned from IoT devices would help to aggregate goods in order to use fewer vehicles while accommodating the required volume.
Image 2. A Simple Graph of a Middle Mile and Last Mile
We will now deep dive into one specific activity in logistics: the middle mile. The complexity of the middle-mile logistics relies on this main trade-off between time and vehicle saturation.
Time — Time is essential since it determines the service level. There is a window of which the drivers operate (normally 8 hours per day) and the items must arrive at Tier 2 during operating hours. Using more vehicles to ship is safer, but how much would your chargeable price be?
Variables: operating hour, working hour, traffic, distance.
Vehicle saturation — as mentioned, logistics cost is more essential on low-value goods. Thus, ensuring full optimization of vehicles — or even to use bigger ones if the volume and weight justify it, is key to drop down cost. Using fewer vehicles to ship is cheaper, but would you arrive on time?
Variables: vehicle capacity, the volume of goods, the weight of goods, number of packages, vehicle type availability, vehicle type costing.
To solve this problem, we can rely on an optimization algorithm to find the best fit for us. An example of that algorithm is VRP or Vehicle Routing Problem. VRP is a type of optimization algorithm to find optimal routes for multiple vehicles visiting a set of locations. This algorithm uses the Distance Matrix (distance between place A to B) of each place and finds the minimum route for vehicles, along with the capacity constraints. It is also important to note that every vehicle type has different cost rates.
III. What is the Reward?
Recent simulations show that there is an advantage from implementing the IoT devices to record data and feeding this data to an optimization algorithm. This advantage comes from two factors: reduced total daily distance [km/day] and reduced total daily vehicle deployed [vehicle/day]. The cost-reduction driver for distance is having less fuel used while the number of vehicles deployed is the reduced daily rent or operating cost, plus driver cost.
The results show somewhere between 33%-57%* of vehicle deployed reduction and 23% to 45%* of distance reduction after the optimization algorithm is implemented. An important insight also lies in the fact that such a widespread % reduction is due to a fluctuating number of vehicles deployed (related to demand) and a fluctuating number of drop points to be reached on that particular day (related to demand concentration). It seems that maximum %reduction can be achieved when the drop points are more clustered, closer to the central warehouse, and have higher and stable daily demands. When the demand is low and unstable, such aggregation is harder to be achieved.
*(Note: %reduction may vary depending on the spread of warehouses and the demands. results may vary on different landscapes and cases)