Many industrial problems can be modeled as optimization problems , whether it’s planning production, organizing delivery routes, or managing inventory. Effective methods already exist, such as constraint programming and local search (have you heard of Oscar.cbls ?). Today, a new approach joins this family: Decision Diagram based Optimization (DDO) .
How does it work? What are its advantages for your operations? And how can it be useful to you? This article explains it all simply.
Date: 17 October 2025
Asset
About projects
To find the best solution to a problem, DDO explores a diagram where each "node" is a crossroad. This crossroad represents a specific state at a given moment. This state contains all the information needed to generate the next possible paths.
Let’s take a simple example: a home service route. You have a technician and several clients to visit, each with their own availability. The objective is to find the shortest route while respecting everyone’s time constraints.
For example, a state could be:
“I am currently at client 2, it is 10:00 AM, and I still have clients 1 and 3 to visit.”
From this state, the decision to be made will be:
“Knowing the travel times, which is the most relevant next client to visit?”
The graph below represents a small instance of this problem. For each node, we have an availability time window and the travel time to the other nodes. The decision diagram represents all possible paths, and the green path shows us the optimal (in this case, the shortest) sequence.
Of course, in reality, with dozens of clients, the number of possible paths explodes, leading to a gigantic and incalculable diagram. To avoid building a huge diagram, DDO calculates a lower bound (the tour will take at least X hours) and an upper bound (it will take at most Y hours) based on smaller diagrams. In successive steps, it refines these two bounds until they converge, leaving only one possible path: the optimal solution.
Industrial problems are often characterized by a large number of business constraints. On a delivery route, you have to manage vehicle capacity, client schedules, and also precedences (picking up a package at A before delivering it to B), technical skills, etc..
For many optimization methods, each new constraint adds a layer of complexity. The room for maneuver shrinks, and it becomes difficult to find a quality solution.
DDO, on the other hand, works the opposite way:
It leverages constraints to speed up its calculations.
As it builds the solution step-by-step, as soon as a partial path violates a constraint, the entire branch of explorations stemming from it is pruned. Constraints are no longer obstacles, but guides that prune the search tree to find the solution faster.
DDO particularly excels at solving sequencing and scheduling problems. Here are a few concrete use cases:
For example, within the European project Deep Scheduling, we will use DDO to plan the manufacturing stages in metallurgy to obtain high-quality steel while reducing CO2 emissions and energy consumption.
Convinced by this method’s potential, we are developing DDOLib, an open-source software library in Java, in collaboration with UCLouvain (Prof. Pierre Schaus from ICTEAM and Roger Kameugne). The goal is twofold:
If your operational challenges involve complex sequencing problems with numerous constraints, DDO technology offers a new and powerful perspective. By turning these constraints into allies, it paves the way for faster and more robust solutions. With open-source tools like DDOLib, this approach is now accessible for optimizing the core processes of your industry.
View online : DDOLib