(Automatic) Optimisation of the Barrel Temperatures in Extrusion

Screenshot: SHS Extrusion Control Unit (Retrofit)

As already mentioned in the previous article, it is always advisable to optimally parameterize the temperatures of the extrusion cylinder to avoid production problems and to ensure economical and productive production.

Optimizing cylinder temperatures while production is running is not an easy task and is therefore often avoided due to the inertia of the system, the mutual influence of adjacent cylinder zones and the fact that waste material can be produced as a result. This means that many extruders are operated at sub-optimal operating points, leaving hidden potentials untapped.


Ways to optimise the different barrel-zone temperatures

As with any optimization task, it must first be clarified what the degrees of freedom of the optimization (the adjustable parameters) are and what the (measurable) optimization goal is.

The adjustable parameters are the setpoint temperatures of the different heating zones, but it must be taken into account that for each individual zone there are critical (limit) temperatures which either cannot be realistically reached due to a lack of power (e.g. cooling capacity) or which should not be achieved as this would lead to process problems (burns, degradation). In addition, it must be noted that in most cases the setpoint temperature set on the controller does not correspond to the mass temperature prevailing in the zone, but only to the temperature at the sensor’s measuring head.

Various criteria can be considered as optimization goals. Common target criteria are the mass temperature, the minimization of temperature fluctuations, the minimization of pressure fluctuations, but also other parameters such as the surface quality, the gloss/haze of the product or its mechanical properties.

In order to enable process optimization, it must be possible to adjust the process parameters (this is usually the case), but the result of the variation must also be measurable, especially in the required frequency and accuracy (e.g. in the event of pressure fluctuations; this is usually not possible with standard equipment).

It must also be taken into account that, depending on the size of the extruder, periods of time between a few minutes (small extruders) and hours may elapse before the new process state has settled and is in equilibrium.

From a scientific point of view, the most common way to identify an optimum for such a problem would be to conduct a complete series of experiments according to the DOE principle (e.g. fully factorial), whereby all cylinder zones are varied and the system response is recorded. However, such a DOE examination can take up an entire working day even with small extruders, so that this procedure is usually of no importance in practice.


Due to the complexity of a DOE, manual process optimization in practice therefore usually takes place in the form that individual zones (only one zone at a time) are reparameterized with regard to their temperature in small steps (e.g. +/- 5°C). After the system has settled and returned to a stable operating state, the process quality is evaluated. In the event of an improvement, the change is accepted if no improvement has occurred, either the original parameterization is reset and another zone is varied or the change is retained and yet another zone is adapted.

The disadvantage of this approach is that the mutual influence of different zones is so difficult to grasp and that an intelligent selection of which zone should be changed depends on the machine operator’s know-how or wealth of experience. In addition, such an optimization can also take many hours if the machine operator does not intuitively choose the correct sequence for re-parameterization. In addition, it is difficult to determine whether a local or global optimum was found when using this method.

Another method for identifying an optimal temperature profile is dynamic optimization (e.g. according to Dr. Chris Rauwendaal). With this method, a significant change in a zone temperature (e.g. around 20-40°C) is set and the course of the temperature as well as the course of the target variable is recorded continuously (if possible automatically). Because a change does not occur abruptly, but the system slowly approaches the new target state, information is continuously generated that describes the behavior of the system.

The evaluation of such a process clearly shows at which temperature the target variable has reached an optimum and provides valuable information about the interrelationships of the process.

Such and similar (further) algorithms are nowadays made available to the plant operator as automatic functions by some modern machine controls. The controller then automatically and intelligently varies the zone temperatures (e.g. automatic zone temperature optimization – ZTO) and determines the optimum values for different zones and different target variables. Monocriterial as well as multicriterial optimization algorithms are used.

A further variant of this simple gradient based optimization are self-learning algorithms. Such systems continuously record the actual situation of the system by sensors and store all process states in which the system has once been in a database. This provides the system with a large amount of data (from real production practice) from which special algorithms of “artificial intelligence” can “learn” system behavior.

Without the plant operator/operator noticing any of this, the system continuously learns from the real situation and is thus put in a position to provide the machine operator with information on which temperature setting will lead to which result from the knowledge acquired. The appropriate keyword here is “virtual assistance systems” and represents the future of extrusion from our point of view.


Register here (free) as a premium member and get access to our download area. There you will find, among other things, an Excel calculation program for designing the temperature profile of an extruder including an expandable material database. In addition, as a premium user you will always be informed about new contributions.


Excel Tool zur Berechnung des Temperaturverlaufs im Premium Download-Bereich

(P.S. The development of self-learning systems is one of the main activities of SHS plus GmbH and is used in our virtual assistance systems and our extruder controls for retrofitting to existing systems (retrofit).

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.