Editing and conveniently representing all the data involved in running an assembly line is a difficult task, and OptiLine takes off most of the burden with its unique graphical interface. However, once all the data have been gathered, there is still the difficulty of running the line, i.e., deciding who will do what and when at the line.

    Solving the latter task means finding a detailed assignment of operations to workstations and operators. In practice, this is a daunting task: even a modest line implies millions of possibilities of assigning operations to workstations and operators, and for large lines the figure quickly exceeds billions of possibilities. Obviously, not all those assignments are equally good, some yielding more efficient lines than others. But finding the best possible one becomes quickly unfeasible to do by hand.
    To free the line manager from the task of finding a good solution to the problem by hand, OptiLine offers an automatic line balancing feature that finds a high-quality assignment of operations to workstations and operators without manual intervention. To achieve that, OptiLine uses a unique optimization algorithm that finds the assignment with the least disbalance of workload among the workstations, while complying with all the constraints specified by the line manager.
    Finding a high-quality assignment of operations is provably extremely difficult (for those well versed in computer science: the problem is NP-hard). On the one hand, it is impossible to try all the possibilities (there are too many of them), while on the other hand, picking one at random is nearly guaranteed to fail (the probability of picking an excellent one is extremely limited).
    To achieve an unparalleled probability of success in a short time, a sophisticated optimization technique is necessary. To achieve that, OptiLine relies on its proprietary technique of Grouping Genetic Algorithms (on the Web), developed over more than a decade by Optimal Design (see publications, on the Web). The superiority of the technique is supported by the fact that it beats all the benchmarks ever published on the problem.