Abstract illustration of the mechanisms of the Container Loading Engine CLOE to find solutions to 3D packing problems.

Container loading engine

With CLOE, you get instantaneous answers to recurring planning tasks such as “Do these goods fit into a box or container?” or “How many vehicles are required to transport today’s orders?”.

The underlying problem of these tasks belongs to the family of three-dimensional packing problems. The goal is to determine a packing configuration for items that optimizes the usage of loading space while respecting numerous planning requirements.


Items may or may not be stacked onto one another depending on their attributes. For example, heavy goods must not be stacked onto fragile goods. If items can be stacked, their base area must be partially or fully supported by other items directly below. Some items may only be placed on the container floor.

Containers have an overall weight limit and may have weight distribution constraints, such as axle load limits.

Items may only be loaded or unloaded in a specific sequence, such as last-in-first-out (LIFO) loading. Items may also be grouped for others reasons, requiring to respect a given loading or unloading sequence.

Shipping containers can be loaded and unloaded from the front and the back. Truck trailers can be accessed from the back and sometimes from the sides. Access directions must be considered in concert with any loading sequences to guarantee the reachability of goods and to avoid time-consuming reloading efforts.

Items have a length, width, and height and may or may not be rotated to fit better into the loading space. For example, one requirement for Euro pallets or mesh box pallets is that they may usually be rotated around their vertical axis (yaw rotation), but in some cases, they may not. Items may also have a this-side-up constraint.


Vehicle routing algorithms

VERA assigns transport requests to vehicles and determines the sequence of locations to visit. It provides several tour plan suggestions. Dispatchers can choose the most suitable according to their specific business needs and given planning parameters. For example, one may want to

  • minimize transport kilometers to reduce fuel costs and emissions,
  • reduce the vehicle count to combat driver scarcity or to avoid using freight exchanges,
  • have robust delivery schedules to guarantee on-time delivery, reduce waiting time, and avoid contractual penalties for late delivery.

This makes it easy to select a well-balanced trade-off between different target criteria that satisfies the daily changing requirements and adapts to unexpected events in the fast-paced logistics industry.

Illustration of the mechanisms of the Vehicle Routing Algorithms VERA to find improving tours.


The cost function exposes multiple target criteria that can be parametrized. The parameters include costs for transport kilometers, penalty costs for waiting times or late arrivals, fixed costs per used vehicle and vehicle type, costs per stop, and more. If your requirements still need to be covered, feel free to get in touch.

An arbitrary number of time windows per location are respected. This allows to meet notified pickup or delivery windows and offers planning flexibility by selecting one available time slot. The dispatch time at the depot can be chosen flexibly within an interval without impacting costs or feasibility. Time windows can have soft or hard cutoff times to allow or restrict early or late pickup or delivery visits. You can specify setup and service times at locations to calculate ETAs accurately and reliably. Also, you can set a maximum route duration. The planning horizon is unlimited, so you can plan daily for the short term, schedule shipments together for a whole week, or even do strategic scenario planning.

Often, there is a single depot where goods are stored and loaded into trucks. Other times, there are multiple locations where goods must be picked up. This is commonly known as pickup and delivery, and can take the form of clustered or mixed pickup and delivery. The clustered version requires that all pickups happen before any delivery, whereas there is no such restriction in the mixed version. In industry, pickup and delivery is sometimes referred to as multi-depot planning. However, in the literature, multiple depots refers to the case when goods are redundantly available at different locations, such that an additional selection decision must be made: from which depot should the goods be picked up?

Common vehicles capacities are given by a maximum volume and weight limit. Other possible capacities are the maximum number of palettes or people that can be transported. Multiple capacity dimensions can be considered simultaneously, but this is distinctively different to considering 3D loading constraints. Considering the loading and unloading sequence is of great benefit, but only makes senses for vehicle routing with 2D or 3D loading constraints; see CLAIRE below for more info.

A variety of vehicles can be part of the same fleet, ranging from sprinters and small trucks to diverse categories of heavy goods vehicles with one or multiple trailers. They have different attributes, such as driving speed, size, loading space, weight limit, available loading aids, temporal availability, and expense factors, which must be considered during planning.

There are two reasons to split orders: 1) forced splitting and 2) splitting for economic reasons. Forced splitting occurs if the items of a single order do not fit into the loading space of a single vehicle, so the goods must be distributed onto multiple vehicles. Splitting for economic reasons can drastically reduce transportation costs but must be applied with reasonable splitting restrictions because splitting too much risks producing fragmented tours and reduces the bundling of goods.

There may be incompatibilities between concrete vehicles or vehicle types, drivers, locations, goods, and orders. For example, some goods may only be delivered with refrigerated vehicles, or some goods should not be combined with others, such as food and chemical products.


container loading and integrated vehicle routing engine

Does your TMS provide enough decision support to dispatchers? Suggesting automatically generated routes that respect all planning restrictions is a must-have. What is routinely neglected is three-dimensional load planning: if freight does not fit into the vehicle, the suggested routes become useless.

Volume approximation misjudges actual loading capacity

Conventional vehicle routing solutions approximate vehicle capacity by volume, load meters, or weight. As a result, loading capacity is used inefficiently and it cannot be guarantee that goods actually fit into the vehicles as planned. To avoid overloading vehicles, a volume capacity limit must be chosen that is less than the actual vehicle capacity. But how do you determine that limit? No matter how it is determined, it leads to a waste of resources:

  • If it is chosen too low, the actual capacity is underestimated. So, trucks drive with unused empty capacity.
  • If it is chosen too high, the actual capacity is overestimated. This means that freight cannot be transported as planned.

Underestimation of loading capacity

Volume capacity limit of 70% underestimates actual vehicle capacity. Load checking by volume approximation determines that two additional pallets don’t fit onto the truck, whereas in reality they would fit.

Overestimation of loading capacity

Volume capacity limit of 80% overestimates actual vehicle capacity. Load checking by volume approximation determines that two additional pallets can be loaded onto the truck, whereas in reality they don’t fit.

There is no good threshold value for the volume utilization limit

Loading feasibility must be determined on a case-by-case basis. It depends on the dimensions and specific characteristics of the transported goods and vehicles. Even the sequence in which the goods are delivered can be decisive: they might fit in one sequence but not in another, for example, due to LIFO loading, stacking restrictions, or item fragility. The loading problem and routing problem heavily influence each other in intricate ways.

Avoid these pitfalls by planning with CLAIRE and unlock the full potential for your transportation planning.

Benefits of integrated planning

Beyond the advantages of conventional vehicle routing solutions, CLAIRE’s integrated planning capabilities

  • allow determining the required number of vehicles exactly,
  • achieve higher vehicle utilization and reduce the required number of vehicles,
  • prevent reactive and sub-optimal rescheduling,
  • reduce hidden costs resulting from not considering the interactions of loading and routing during planning.

Degrees of integration between load planning and vehicle routing

The loading of goods must be reassessed whenever tours are changed. This can be achieved either in a sequential or integrated manner. In the sequential approach, tours are first created and then evaluated for their cargo loading feasibility. In the integrated method, the cargo loading feasibility is continuously assessed during the tour planning process. There are five levels of integration between load building and route planning:

  1. By far the most common case: any form of volume approximation,
  2. Sequential manual planning,
  3. Sequential, computer-aided manual planning,
  4. Automated sequential planning,
  5. Fully integrated planning.

Whether sequential planning is done manually or in an automated manner, much time is wasted on constructing and checking routes where goods don’t even fit into the vehicles. More importantly, the quality of generated tour plans is considerably below an integrated planning approach. In “The value of integrating loading and routing“, vehicle routing researchers demonstrated that fully integrated planning consistently produces better solutions, with an average cost reduction of 7% over automated sequential planning.

Fully integrated planning

Illustration of the mechanisms of the Container Loading and Integrated Routing Engine CLAIRE to find improving routes while considering the packing of cargo-space.

CLAIRE offers the highest level of integration and takes into account all interactions and dependencies between both planning problems.

The Tech Lead of Operations Research at Google says that vehicle routing with three-dimensional loading constraints is “exceptionally hard” and “just absurd”. And it really is, but the rewards of solving this problem are substantial. We have taken on the challenge so you don’t have to. Let’s discuss how we can help you unlock the full potential of transportation planning.

Scroll to Top