Production Scheduling

Production Scheduling

Good production scheduling is an important prerequisite for the competitiveness, productivity, and resource efficiency of the processing industry. This page gives a brief overview of this topic.

What actually is production scheduling?

Die Produktionsplanung befasst sich mit der operativen, zeitlichen, mengenmäßigen und räumlichen Planung, Steuerung, Kontrolle und Verwaltung aller Vorgänge, die bei der Produktion von Waren und Gütern notwendig sind.
In der Prozessindustrie werden viele Anlagen in Chargenfahrweise betrieben, was bedeutet, dass die Produktion nicht kontinuierlich, sondern basierend auf Produktionsrezepten abläuft. Bei diesen Anlagen, die auch in OptiProd.NRW betrachtet werden, müssen die Verteilung der einzelnen Produktionsschritte auf die vorhandenen Anlagen und Apparate und die Reihenfolge der Abarbeitung der Chargen gut geplant werden, um die verfügbaren Ressourcen optimal einzusetzen und somit möglichst effizient und nachhaltig zu produzieren.
Der Einsatz der verfügbaren Produktionsanlagen, Ausgangsmaterialien, Mitarbeiter und anderer Ressourcen wird daher so geplant, dass alle Produktionsaufträge möglichst termingerecht und ressourceneffizient fertiggestellt werden oder, falls dies nicht möglich ist, die Produktionsverzögerungen zumindest minimal gehalten werden.
In der Praxis läuft die Produktionsplanung heute meist so ab, dass erfahrene Mitarbeiter auf Basis der aktuellen Auftragslage die Produktionspläne für die nächsten Tage oder Wochen erstellen, entweder komplett manuell oder mithilfe von Plänen, die aus einfachen virtuellen Modellen der Produktionsanlage generiert werden.
Die vollständig rechnergestützte Optimierung von Produktionsplänen ist heute allerdings noch nicht der Stand der Technik, wodurch in der Praxis viele Verbesserungspotentiale ungenutzt bleiben. Die folgende Abbildung zeigt einige der Herausforderungen, die sich in der industriellen Praxis ergeben:

Production scheduling deals with the operational, chronological, quantitative, and spatial planning, control, monitoring, and administration of all processes that are necessary for the production of goods. 

In the process industry, many plants are operated in batch mode, which means that production is not running continuously, but based on production recipes. Within these plants, which are addressed in OptiProd.NRW, the distribution of the individual production steps to the existing plants and equipments, as well as the order of the batches and operations, must be planned well in order to make the best use of the available resources and thus to produce as efficiently and sustainably as possible.

The utility of available production facilities, raw materials, employees, and other resources is thus planned in such a way that all production orders are completed in time and as resource-efficiently as possible or, if this is not possible, production delays at least are reduced to a minimum.

Today, production scheduling is usually done by experienced employees who create the production schedules for the next few days or weeks based on the current order list, either completely manually or using plans that are generated from simple models of the production plant.

The fully computer-based optimization of production schedules is not yet the state of the art, which means that in practice much potential for improvement remains unexploited. The following figure shows some of the challenges that arise in industrial practice:

What are the benefits of optimal production scheduling?

Für die Optimierung des Betriebs von Produktionsanlagen werden heutzutage bereits mathematische Modelle eingesetzt. So werden z. B. die Produktionsrezepte, das Anlagenlayout, die Eigenschaften der genutzten Apparate (wie Größe, Durchsatz, Betriebsgrenzen, etc.) und die Fahrweise anhand von Modellen entwickelt und optimiert. Für schon existierende Anlagen werden mathematische Modelle genutzt, um Engpässe („Bottlenecks“) zu identifizieren und zu beseitigen, um die bestmöglichen und preiswertesten Alternativen für Erweiterungen und Umbauten (z. B. zur Durchsatzerhöhung) zu ermitteln und um die Fahrweise im Sinne der Anlageneffizienz zu optimieren.
Die optimale Produktionsplanung hat höchste ökonomische und ökologische Relevanz  für die industrielle Chargenproduktion. Die folgende Abbildung stellt einige der Hauptvorteile der optimalen Produktionsplanung dar:

Mathematical models are already employed to optimize the operation of production plants. As an example, the production recipes, the plant layout, the properties of the equipment (such as size, throughput, operating limits, etc.) and the plant operations are designed and optimized using models. For existing plants, mathematical models are used to identify and eliminate bottlenecks, to determine the best and cheapest alternatives for extensions and conversions (e.g. to increase throughput), and to control the plant operation such that it optimizes efficiency.

Optimal production scheduling has highest economic and ecological relevance for industrial batch production. The following figure shows some of the main advantages of optimal production scheduling:
Optimal production scheduling has been in the focus of research for decades, and today, a variety of approaches are available that can compute optimal production schedules for plants of medium complexity and with simplified production recipes.
Many of these approaches are based on mapping the problem into an optimization model, which is then solved by appropriate optimization approaches. The solution of such problems using mixed-integer linear or nonlinear programming (MI(N)LP) is a widespread approach, for which the problem has to be transformed into a (non-)linear system of (in-)equalities. Other approaches use constraint programming or convert the problem into a network of timed automata for which the best solution is determined by reachability analysis.
However, there is still a large gap between the capabilities of these approaches and the requirements of real production plants.
For example, the optimization models currently used can only rerpesent the production plant and its mode of operation with considerable simplifications, which means that the computed production schedules cannot be directly applied to the real plant. Thus, it is not guaranteed that the computed production schedules provide optimal (or even just good) performance at the real plant, and they require significant manual adjustments. Another major problem is the effort that is required to create and continuously adapt the optimization model, and in addition, the corresponding tools are often academic and difficult to maintain. For these reasons, methods for optimizing production schedules have rarely been used in industrial environments to this date.
To avoid these problems, in OptiProd.NRW we use meta-heuristic optimization approaches, in particular evolutionary algorithms. Using a guided stochastic search, these methods automatically compute production schedules that, while not provably optimal, promise very good  performance. A major advantage of these approaches is that they can be coupled with models of the production plant of any complexity, e.g. models that highly accurately represent all relevant behaviors of the real plant. In OptiProd.NRW, material flow models from INOSIM Software GmbH are used for this purpose. These models can rerpesent even complex production plants very accurately and offer several advantages compared to the status quo:
The computed production schedules are thus directly applicable at the real plant and do not require any manual adjustments.
The optimization tasks solved in OptiProd.NRW provide very good solutions, but they also have some disadvantages:
  • A large number of simulations must be performed during optimization, which can take a lot of time due to the complexity of the models.
  • The optimization problems are complex and have a very large number of degrees of freedom, which makes it extremely difficult to find optimal or even good solutions within a short time.
In OptiProd.NRW, we solve these challenges ...
  • ... by extending the simulation environment of INOSIM Software GmbH with a new software component that allows a simulation of a large number of candidate solutions in parallelon powerful servers or in the cloud, and ...
  • ... by reducing the number of degrees of freedom through the systematic integration of production scheduling heuristics (as used by experienced production planners in their daily work) to such an extent that an efficient solution to the problem becomes feasible.
Share by: