The Evolutionary Process: Driving Smarter Scheduling Through Measured Improvement
Every time a strategy is executed in EvoAPS, our Evolutionary Algorithm generates a schedule and evaluates its quality—referred to as its fitness—based on how well it aligns with the selected business strategy.
This process doesn’t stop at a single result. Instead, EvoAPS applies small, deliberate changes—known as mutations—to explore alternative outcomes. Each new schedule is measured against the current best-performing result. Only those that demonstrate an improvement are retained. Over thousands of iterations, the system continuously evolves, refining the schedule towards better alignment with strategic goals. This iterative process is what gives the algorithm its name: evolutionary.
Once completed, all viable results are made available in the cloud, allowing users to review, compare, and select the most appropriate schedule. Any selected schedule can then be seamlessly loaded into Opcenter APS for execution or further refinement.
Conclusion: Shifting from Subjective Decisions to Objective Strategies
This evolutionary, strategy-based approach enables faster and more effective implementation of scheduling solutions like Opcenter APS. Instead of relying on traditional, subjective methods, businesses can now evaluate thousands of schedules per minute—objectively and consistently—against clearly defined business strategies.
This not only accelerates deployment but also empowers teams to explore multiple scenarios, identify what works, and continuously improve planning outcomes with confidence.
Defining Strategies and Managing Risk in EVOAPS
One of the greatest challenges businesses face when scheduling is balancing competing priorities. For example:
· “How do we maximize machine utilization while still delivering every order on time?”
· “Can we reduce changeovers without compromising delivery performance?”
· “How far can we pull orders forward without increasing operational risk?”
These are not simple trade-offs—and there is rarely one “correct” answer. That’s why EvoAPS empowers users to define business strategies that reflect their unique goals, constraints, and risk tolerances.
Strategy Configuration Through Intuitive Weighting
Within EvoAPS, strategies are configured using a slider-based interface that allows users to assign relative weightings to multiple criteria. These criteria—such as on-time delivery, resource efficiency, changeover minimization, and more—determine how the system evaluates the fitness of each scheduling result.
By adjusting these weightings, users can simulate different business scenarios and control how EvoAPS prioritizes outcomes. This enables users to create customized, flexible strategies that reflect real-world business decisions.
Each user can create as many different schedules as required, allowing the process of ‘What if?’ scheduling to be performed over both new data and any archived schedules that are available against the profile.
New strategies can be built by simply selecting the weighting criteria value and adding it to the strategy. Then the slider is used to set the desired level.
Interpreting Result Strength and Schedule Risk
As strategies are executed, EvoAPS provides real-time feedback on the strength of each result. Strength is a measure of how closely a result aligns with the defined strategy—essentially, how “fit” that schedule is in the context of the business’s goals.
While EvoAPS does not claim to find a mathematically optimal result—something often impractical or computationally unfeasible—it continually evolves and improves its results. When a solution holds the top position through successive iterations, it becomes a strong indicator that further improvements may no longer be realistic within the defined constraints.
Quantifying Risk and Providing Insights
Each schedule is also evaluated for WIP (Work In Progress) risk, helping users understand how tightly coupled operations are across the schedule—and where potential bottlenecks or process risks may emerge.
EvoAPS also offers insight dashboards alongside results, enabling users to interpret outcomes and make informed decisions, not just based on raw data, but through a strategic lens.
The Trained Model: Capturing and Scaling Knowledge
In traditional scheduling systems, businesses often rely on what the industry refers to as a “trained model.” This means the scheduling logic—whether built using standard APS rules, sequence preferences, or fully custom algorithms—has been configured by someone with deep domain knowledge to achieve a specific, desired result.
Regardless of the method used, the outcome is inherently subjective. It reflects the experience, preferences, and decision-making patterns of a domain expert who guides the system’s behaviour, often in collaboration with the implementor.
But this raises a critical question: how do we capture and scale this knowledge?
Let’s consider a highly experienced planner who has built five schedules per day, five days a week, for 30 years. That equates to roughly 35,350 schedules crafted over a career.
With EvoAPS, we can build and evaluate over 2,000 schedules per minute. In just under 20 minutes, the system can simulate the equivalent of three decades of human scheduling experience—objectively, consistently, and without fatigue.
Conclusion: Scaling Expertise Through Intelligent Automation
EvoAPS enables organizations to move beyond the limitations of individual expertise by capturing strategic intent and applying it at scale. This doesn't just replicate expert knowledge—it amplifies it.
By combining evolutionary algorithms, objective strategy evaluation, and high-speed processing, we can gain deeper insights, faster—helping businesses implement smarter, more adaptive scheduling solutions.
Why Not Use Machine Learning?
Machine Learning (ML) is a widely recognized form of artificial intelligence that enables systems to learn from data, identify patterns, and make predictions—without being explicitly programmed. It’s a powerful tool used across industries from customer service to fraud detection and advanced analytics.
But when it comes to scheduling, the question isn’t whether ML is powerful. It’s whether it’s practical.
To effectively use ML in scheduling, you would need vast amounts of high-quality historical data that accurately reflects the real-world decisions and trade-offs planners have made over time. Not only must this data exist, but it also needs to be clean, complete, and contextually consistent—a challenge for most organizations.
So, where would this data come from?
Can we rely on historical actuals to define how we should schedule in the future?
Often, actual results reflect compromises, manual overrides, and non-ideal circumstances—not best practices or strategic intent. This makes historical data a risky foundation for training an ML model intended to drive future performance.
Conclusion: A Heuristic, Not a Black Box
While EvoAPS always seeks to deliver the most effective scheduling outcome possible, it doesn’t promise the single best answer—because in most real-world planning environments, the “best” is a moving target shaped by evolving business goals.
That’s why we use heuristics—practical, strategy-driven methods that provide good, workable results quickly, rather than chasing theoretical perfection through a black-box algorithm.
What is evolution?
Evolution is the process by which species of organisms change over time through variations in their genetic makeup. These genetic changes can occur due to mutations, natural selection, gene flow, and genetic drift, and they gradually accumulate, leading to new traits or sometimes new species.
At the core of evolution is the idea that organisms with traits better suited to their environment are more likely to survive and reproduce. Over generations, these advantageous traits become more common in the population. This is called natural selection, which was first proposed by Charles Darwin.
