Reliable Integrated On-Line Characterisation Tool for Thermoplastic Compound
NanoOnSpect has 3 development aims:
1) Newly-developed online characterisation techniques will be combined in a ""one-box"" characterisation unit. This will be able to determine dispersion, particle size, particle network formation and resulting composite properties online under demanding industrial conditions, with a (factor 2 to 10) higher resolution than conventional online characterisation tools.
2) An Artificial Neural Network and Expert System (ES) module will be developed to analyse sensor signals, predict the complex quality of the PNC and calculate new input parameters for the compounding line.
3) A breakthrough concept will be developed to adapt the mixing process during composite production, based on feedback from the ES.
The characterisation tool and process control technique will allow precise adjustment of the composite properties (e.g. dispersion-dependent properties) during production. This will improve product quality, allow the reliable production of PNCs with tailored functions and reduce waste production. The technology will be validated in existing production lines for two particle types. NanoOnSpect will also deliver in depth understanding of materials and processing, shortening material development cycles and formulation changeovers.
The NanoOnSpect consortium, consisting of 8 SMEs, 2 RTD performers and 2 associations, combines competences in measurement technology/automation, polymer processing technology, end use and demonstration, standardisation, dissemination and training. The project results will dramatically increase competitiveness and strengthen the market position of the mainly SME-based European PNC compounding industry."
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
Higher or Secondary Education Establishments
€ 752 664,32
Andrea Zeumann (Ms.)
Sort by EU Contribution
AIMPLAS - ASOCIACION DE INVESTIGACION DE MATERIALES PLASTICOS Y CONEXAS
€ 444 231,25
ATEKNEA SOLUTIONS CATALONIA, SA
€ 389 002,58
€ 301 160
FOS FASEROPTISCHE SYSTEME MESSTECHNIK GMBH
€ 215 173,75
HUKSEFLUX THERMAL SENSORS BV
€ 244 251,60
HBH MICROWAVE GMBH
€ 300 316
NEXXUS CHANNEL SRL
€ 260 034
€ 171 246
CMB Colorex Master Batches B.V.
€ 151 469,86
EUROPEAN PLASTICS CONVERTERS
€ 120 294,40
VERBAND TECHNISCHE KUNSTSTOFF-PRODUKTE EV
€ 49 710,24
Grant agreement ID: 263406
1 April 2011
31 March 2015
€ 4 703 754,10
€ 3 399 554
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
This project is featured in...
Quality control over polymer nanocomposites
Grant agreement ID: 263406
1 April 2011
31 March 2015
€ 4 703 754,10
€ 3 399 554
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
This project is featured in...
Discover other articles in the same domain of application
New products and technologies
Wood treated with wood: Foreco joins Bio4Products to develop bio-based preservative
Final Report Summary - NANOONSPECT (Reliable Integrated On-Line Characterisation Tool for Thermoplastic Compound)
1. Executive summary
The SME-driven project NanoOnSpect has developed an on-line characterisation tool onBOX and corresponding process control technologies to ensure the precisely-controlled dispersion of nanoparticles in composites, and reliable quality of the produced material.
Conventional on-line characterisation has insufficient resolution for polymer nanocomposites (PNCs) and is not designed to characterise their specific properties. There is also a lack of in-depth understanding of PNC processing and flexible processing equipment to adapt dispersion to material requirements.
NanoOnSpect had the following objectives in order to address this problem:
1) Combining newly-developed on-line characterisation techniques based on light-scattering evaluation, spectral analysis, realtime dialectric spectroscopy, viscometry, ultrasonic signal analysis and advanced optical, temperature, pressure and electrical and thermal conductivity measurements to accurately determine dispersion and compound properties under industrial conditions in a single, integrated measurement unit (onBOX)
2) New PNC processing technology: combination of the advantages of two compounding processes, twin-screw extrusion and Nexxus wetting technology
3) Intelligent Module: data obtained from the onBOX will be used in an artificial neural network and expert system to control the PNC processing to obtain consistent material quality
4) Validation of the network through the use of the newly-calculated parameters in the compounding line by a specially-developed automation system that can adapt the mixing process to the material requirements.
This automatic feedback loop ensures the reliable production of nanocomposites with tailored functions depending on the precisely adjusted dispersion state.
All developments have been validated in industrial case studies showing their feasibility under industrial conditions.
The consortium, consisting of 9 SMEs, 2 RTD performers and an association, combines competences in measurement technology/automation (6 industrial partners work in this field), polymer processing technology, end use and demonstration, dissemination and training.
Through successful implementation of the NanoOnSpect results, the European measurement and control industry can expect strong growth in the sales of high-tech sensors and corresponding software and automation systems.
Project Context and Objectives:
2. Summary description of project context and main objectives
The use of functional nanoparticles in polymers is one of the most promising but also one of the most challenging applications of this new material class. Low quantities of nanoparticles in PNCs improve barrier properties, fire resistance, electrical conductivity, optical properties and/or mechanical properties. In comparison with conventional fillers, the properties achieved with nanoparticles depend much more on a precisely controlled dispersion, deagglomeration and distribution of the filler in the polymer composite. Unfortunately the current state of the art in characterising these dispersion properties involves the use of time- and cost-intensive off-line methods. This results in comparably high technical and economic risks for the mainly SME-based nanocomposite manufacturers, as the most important properties of the composites can only be determined long after the production. Beside this, other important factors that prevent nanocomposites from reaching their full economical and technical potential are:
• Comparably high variations in nanoparticle input qualities, with especially the dispersability varying between particle production lots.
• The comparably high price of the materials, which results in a high economic loss if the later off-line characterisation indicates problems with the composite produced.
• On-line characterisation technologies currently available for compounding lines have insufficient resolution for PNCs and are not designed for the characterisation of specific PNC properties (e.g. conductivity).
• A lack of the necessary depth in understanding of the PNC preparation process and the flexible processing equipment needed to adapt the dispersion based on the on-line composite characterisation.
The NanoOnSpect project addresses these key problems with an integrated approach comprising developments in the following areas.
• Development of new on-line characterisation technologies capable of determining composite properties on-line under industrial conditions
• Intelligent Module analysing the different sensor signals, linking the information contained and calculating derived characteristics based on an off-line training period. This module consists of an Artificial Neural Network (ANN) which is able to calculate off-line material properties from properties measured on-line, and an Expert System suggesting suitable process parameters for the desired on-line-monitored property.
• Development of new compounding equipment enabling an enhanced use of elongation stress for particle dispersion
In work package 1 concerning the development of individual sensors and the onBOX, the following results have been obtained:
• Design, manufacture and testing of
o Miniaturised pressure sensor
o Multispot temperature sensor and IR sensor
o Thermal conductivity sensor
o Rheology sensing system
o Optical spectroscopy sensing system
o Ultrasonic sensing system
o Microwave spectroscopy sensing system
• Design and manufacture of the onBOX device
• Communication architecture between the onBOX and Intelligent Module (WP4)
• Controlling unit for the onBOX and IM
The onBOX and the sensors are available in a pilot version and have been tested for feasibility on an extrusion line. They can all be connected to the onBOX controller, which is able to display and store all sensor data during processing.
In work package 2 concerning polymer nanocomposite processing, the following results have been obtained:
• Design and construction of melting, mixing and degassing module of the Nexxus Channel
• Trials on PP/CNT system with CNT as aqueous suspension in the Nexxus Channel
• Optimization of conventional twin-screw extrusion for nanocompounding
• Development of the interface between Nexxus Channel and a twin-screw extrusion line
• Study of the distribution of the polymer melt flow from the extrusion line to the Nexxus Channel
In work package 3 concerning the off-line and on-line material and process characterisation, the following results have been obtained:
• Extensive studies on the processability of the CNT/PC and clay/PP systems
• Trials with sensors developed in WP1 and correlation of the material properties with processing parameters
• Data sets for the material properties and accompanying process parameters for the Intelligent Module
• Definition of rules for Expert System
In work package 4 concerning the Intelligent Module as a quality assurance technique for the nanocomposite process, the following results have been obtained:
• Development of the Artificial Neural Network Module
• Development of the Expert System Module
• ANN for prediction of material properties, shown on electrical resistivity – electrical conductivity
• Expert System for process control, shown for PP/Clay system
In work package 5 concerning the case studies in an industrial environment, the following results have been obtained:
5 successful case studies have been carried out under industrial conditions.
• 5.1 Case study: Implementation of the onBOX device into existing extrusion line for four different extrusion lines
• 5.2 Case study: CNT-based PNC, industrial validation at CMB Masterbaches for onBOX and Intelligent Module
• 5.3 Case study: Nanoclay PNC, industrial validation at Addiplast for onBOX and Intelligent Module
• 5.4 Case study for testing and evaluation, for ecological and economic evaluation in WP6
• 5.5 Case study: Three different polymeric composites defined by the Industrial Reference Group
In work package 6 concerning the ecological and economical evaluation, the following results have been obtained:
• Economic and ecological evaluation of the NanoOnSpect technology
• Eco-efficiency study of electrically-conductive composites
• Dissemination of standardization issues into standardization bodies and into consortium
In work package 7 concerning the exploitation and dissemination of the project results, the following results have been obtained:
• Development and update of the dissemination and exploitation plan
• Establishment of an Industrial Reference Group
• Development and update of the report on the market situation and exploitation possibilities
• Continuous dissemination of the project
• Preparation of dissemination material (flyer, newsletters and poster)
• Publications at conferences and in industrial magazines
• Attendance of nine fairs and conferences actively presenting the project
• Preparing a NanoOnSpect booth at the K-2013 fair showing the onBOX and related sensors in hardware
• Industrial workshop on nanocompound characterisation with hands-on trials
• Industrial workshop on standardization
• Training of the consortium on NanoOnSpect technology
3. Description of project context and main S&T results / foregrounds
3.1 WP1 On-line nanocomposite characterisation device
3.1.1 Main objectives
All sensor developments in the project were carried out in WP1, and the sensors were integrated into the onBOX unit with a central controller for all devices.
3.1.2 Pressure and temperature sensing
In polymer composite material development processing pressure and temperature are the most important basic physical parameters to be controlled in order to obtain reproducible and reliable material properties. This is especially true for the development of processes for new polymer nano-composites and high-temperature materials where the exact melt temperature information is needed to avoid degradation of the material. Beside this there is a need for smaller sensors. Especially when processing low quantities of expensive blends mini laboratory extruders are used, which have limited space for measurement probes.
Miniaturized pressure sensor
In NanoOnSpect FOS Messtechnik GmbH has developed a small size pressure sensor. The sensor is based on the standard FOS fiberoptical pressure sensor. The measuring principle is shown in figure 126.96.36.199 (see separate attachment). The displacement of a pressure membrane caused by the pressure is measured by means of a fiberoptical light reflection system. This configuration allows the construction of very small size pressure sensors for use at high temperatures. Figure 188.8.131.52 shows the developed miniature pressure sensor for extruders. The sensor properties are :Ø 3mm frontend membrane diameter, integrated temperature measurement, M5 mounting thread, extended temperature range of 450°C, pressure up to 300bars and a dry fiberoptical measuring system (enabling use in drug manufacturing). The small size avoids the generation of dead zones in the process chamber.
In NanoOnSpect FOS has improved and tested its IR melt thermometer. This sensor measures the temperature-dependent amount of IR-radiation emitted by the polymer, resulting in higher accuracy and quicker processing times of the sensor. The key component of the sensor frontend is the pressure-tight sapphire window which transmits optical radiation in the range of 0.3 µm to 5.5 µm wavelength. The sensor with amplifier is shown in Figure 184.108.40.206 (see separate attachment). The standard ½”-20 UNF mounting thread enables the direct replacement of standard thermometers in a production line. Using IR radiation to determine the temperature of the polymer enables the absolute melt temperature to be obtained; this is due to the use of a maximum thermal resistance and not the thermal conduction between the measuring volume and machine housing. Figure 220.127.116.11 shows the IR-thermometer readout during a test run. The temperature reading is (in this case) 20°C higher than the barrel temperature. The IR-thermometer is also equipped with an internal thermocouple enabling the measurement and readout of the sensor frontend temperature. These two signals in parallel - the IR-temperature from the melt and the thermocouple temperature from the sensor front - provide new knowledge about the process. There is always a temperature gradient between the extruder wall and the melt inside. The temperature step in the IR-readout line (Figure 18.104.22.168 in the attachment) was caused by a sudden increase in screw speed, increasing the shear and therefore the energy in the system. Figure 22.214.171.124 shows the IR-thermometer mounted at an extruder die. With a response time of 10 ms this IR thermometer is up to 1000 times faster than a conventional thermocouple melt thermometer. These properties make this device very useful to obtain faster and more accurate temperature readings.
Melt Gradient Thermometer
The IR-thermometer enables a new approach to absolute melt temperature measurement; however the penetration depth into the polymeric material is limited to about 2 mm due to strong IR absorbance of the polymer melt. Therefore FOS has developed a new type of thermocouple thermometer with 2, 3 or 4 measuring points with a fast response time compared to standard devices. Figure 126.96.36.199 shows a typical frontend of this thermometer and the position of the measuring points. Each of these thermometers has a low thermal capacity and fast response time of about 1 to 3 seconds. Figure 188.8.131.52 (see separate attachment) shows the measuring set consisting of the sensor and the multi-channel amplifier with a USB computer interface. A LabView-software program allows the temperature readings and the temperature profile to be displayed on a screen. By measuring the temperature profile in a cross section of a melt stream more information about the process than just a single point measurement in the heating zone can be obtained. Figure 184.108.40.206 shows an application of this gradient thermometer at an extruder–nozzle. A low temperature gradient in the melt is needed for a stable viscosity of the material. This stable viscosity is a precondition for stable material throughput. Measuring the temperature profile in the melt can also avoid thermal damaging of the material or inhomogeneous mixing, especially when processing nanocompounds as not only the lower temperature at the barrel wall is monitored. Multipoint temperature measuring perpendicular to the melt stream is very helpful in combination with an inline viscometer like in the NanoOnSpect onBOX system where a non-existing gradient is the best precondition for accurate viscosity measurement. During test trials a homogeneous temperature distribution was measured.
3.1.3 Thermal conductivity sensing system
Hukseflux Thermal Sensors developed an on-line thermal conductivity measurement system. The purpose of this system is to serve as an optional module of the onBOX unit, which monitors the melt properties during production. Thermal conductivity is one of the possible “design parameters” of plastics with nanofillers. In theory nanofillers have the capability of significantly enhancing the thermal conductivity of plastics relative to their bulk properties. Higher thermal conductivity may be desirable to achieve improved passive cooling. This is particularly relevant for instance when working with high-power components such as the latest generation LEDs and high-power microprocessors. When processing plastics with nanofillers, on-line monitoring of the thermal conductivity of the melt is potentially useful to predict the thermal conductivity of the end product. Apart from this the thermal conductivity of the melt may also be used as a redundant monitor of the consistency of the mixing process, in addition to the standard viscosity measurement on the onBOX and the other measurements investigated in the work package.
The result attained is a system with the potential to provide a measurement of process consistency, but not an absolute measurement of the thermal conductivity of the melt or that of the end product. It remains to be seen if the benefit of implementing the measurement is sufficient for commercialization. The costs of the sensors and electronics are relatively high. The interpretation of the measurement results does not seem to be straightforward. The electrical resistance measurement (the DC component), also developed in the framework of the NanoOnSpect project, is a “competitor”, potentially offering similar information at a lower cost. On the other hand, in view of the importance of thermal conductivity in some upcoming markets such as LED technology, the need to monitor the consistency of thermal properties may increase and a higher cost level may be acceptable.
As anticipated, the development of a good measurement system proved to be difficult. The thermal conductivity measurement system including the measurement and control unit first went through a prototype phase. The issues with electronics, noise reduction and digital interfacing, although time-consuming, were practical rather than fundamental. The issues encountered with the sensor technology are more fundamental and can broadly be divided into three categories:
The first issue has to do with the working environment. The combination of high temperatures, high pressures and an electrically noisy environment was overcome after a number of trials. The consortium developed a prototype “flush” surface probe, NOS03, with the aim of making the sensor less vulnerable than the traditional thermal needle method which sticks out into the melt flow, illustrated in figure 220.127.116.11 (see separate attachment). In the latter configuration the needle is fully exposed to the flowing melt, which is a potential problem for sensor lifetime. The flush surface probe however proved insufficiently sensitive, and moreover the sensitivity changed depending on the exact way it was mounted. In the final design the consortium chose a short and relatively robust thermal needle, model NOS01 (see also figure 18.104.22.168). The problems with sensitivity and the electrically noisy environment were solved by using high-end electronics, and by the proper shielding and grounding of the sensor.
The second issue is process related. The measurement of thermal conductivity in principle requires a static, i.e. non-flowing, and thermally stable condition. The measurement of thermal conductivity is based on measurement of the reaction to a known amount of heating. The thermal stability of the environment must be compared to the heating introduced during the measurement of the thermal conductivity. The onBOX in which the NS01 is mounted offers a relatively stable environment but for the purpose of thermal conductivity measurement it still was a significant source of error. In addition the scenario to stop the flow of melt in the onBOX for a thermal conductivity measurement was rejected. Although technically possible the danger of locally overheating the melt was considered too large. The consortium tried to limit the effects of local flow by keeping the measurement interval short; however, this is only partially successful. It was possible to detect the consistency of the process e.g. trends of thermal conductivity (assuming a constant heat capacity), but not to perform absolute thermal conductivity measurements. The effects of thermal instability of the environment (the onBOX) could be countered by using relatively high heating as part of the measurement, at the same time keeping it low enough and the interval short enough to avoid overheating of the plastic melt. Figure 22.214.171.124 illustrates measurements in different melts, at different flow rates of the melt, compared to the ideal static condition (0 rpm).
The third issue is assumed to relate to the fundamental properties of plastics with nanofillers. The thermal conductivity of these plastics varies with the percentage of nanofillers (filling percentage). However this behavior is strongly non-linear. The plastics only start conducting significantly more than the bulk plastic at the moment the nanofillers start “networking” i.e make physical contact to one another, offering a uninterrupted preferred path along which heat can travel. This means that at low filling percentages the thermal conductivity measurement is not a “sensitive” or “high resolution“ indicator of the material properties or of the filling percentage. Also, unexpectedly, in the first phase of tests the correlation between the thermal conductivity of the solidified plastic and the thermal conductivity of the melt was not linear. Some filled plastics that did not exhibit a significant thermal conductivity increase relative to the bulk plastic when measuring in the melt, did show significantly increased thermal conductivity of the solidified end product (at room temperature), as can be seen for example in Figure 126.96.36.199 (in the attachment). In addition, although in one plastic type the trend is reasonably well correlated, the measurement in different plastics with the same room temperature properties also seems to have an unexplained bias. The present explanation proposed by the consortium, i.e. that this phenomenon relates to the networking of nanofillers, is still unproven. Further development will be carried out after the project in order to find the explanation by correlating the thermal conductivity measurement results with dielectric and rheological properties of the melt as well as measurements on the solid plastics at room temperature.
Results obtained during the project were disseminated during the K-Show in Düsseldorf, Germany, in October 2013 and at the ITCC 2011 conference (international thermal conductivity conference 2011) in Chicoutimi Saguenay, Canada.
Trials were carried out at Gneuss from 14 to 16 March 2012, at AIMPLAS from 27 February to 01 March 2013, at ICT Fraunhofer from 18 to 19 November 2013, at AIMPLAS from 22 to 25 April 2014, and at Colorex from 24 to 26 June 2014. Figure 188.8.131.52 (see separate attachment) shows a typical test setup.
After the prototype phase, a digital communication link was prepared for the measurement and control unit, so that it now can function both autonomously and as part of the general onBOX platform.
The combination sensor and measurement and control unit has meanwhile proven to survive both the operational conditions, temperatures up to 300 °C, pressures in the relevant ranges and repeated transport by air and road, as well as installation and operation by different users. In case of future commercialization the thermal sensor design will probably remain unchanged. The present design of the measurement and control electronics is probably too costly. It is now built from commercially-available high-end components, and would need to be downgraded to be manufactured at a lower price.
3.1.4 Rheology sensing system by capillary rheology
One of the most important parameters to describe the flow characteristics of a plastic melt is the viscosity function. In an extrusion line the value of the viscosity varies mainly with shear rate (due to the common shear thinning behavior of plastic melts) and the temperature, while the pressure plays a minor role for the viscosity under typical extrusion conditions. If theses parameters are constant the viscosity is determined by the material properties (e.g. chain length of polymer molecules, content of fillers) in a pure plastic melt and is in direct relationship to the physical properties of the material such as tensile strength and impact resistance. In compounding the value is influenced by the relative amount and type (e.g. particle size of solids) of the components that are processed in the line.
Gneuss GmbH developed a rheology sensing unit for integration into the onBOX characterisation unit for NanoOnSpect. With respect to an easy, straightforward and robust measurement system the following approach for realizing a viscosity measurement was chosen. By means of a high precision metering gear pump, a small part of the polymer melt is separated from the main melt channel. This polymer is pumped through a precisely manufactured slot capillary. Both the melt temperature and the melt pressure (measurement at 2 positions) are monitored. Figure 184.108.40.206 shows the design and measurement principle of the developed rheology sensing unit
For the capillary a rectangular shape with height H and width B was chosen. This rectangular shape allows a rheologically optimized design of the capillary in the region of pressure measurements when state-of-the-art pressure sensors for extrusion are used, because these capillaries have a flat surface which perfectly fits to the membrane of the sensors. If the two pressure measurements for melt pressures P1 and P2 are separated by a distance L for a given volume flow V ̇ the representative viscosity η_rep and shear rate γ ̇_rep are calculated from the mentioned parameters according to standard formulae by the control system attached to the viscosity measurement unit. The depth of the capillary slot is specified according to the material properties within a range of 0.5 to 2.0 mm. The body of the unit is designed in cylindrical shape. The idea is to fit it additionally into a given extrusion system in order to allow an on-line viscosity measurement. Therefore the existing and optimized line geometry should be as little disturbed as possible. Based on the chosen concept the system could be designed with an extrusion length of 140mm. This length is mainly given by the dimension of the melt pump which was selected with a size of 0.6 cm³ per revolution which also determines the maximum throughput for the capillary measurement system for the prototype with 40 cm³/minute. The interfaces to the upstream and downstream components were chosen with a round shape. The exact geometry of the bores and threads for mounting the system can be selected in a wide range, enabling the unit to fit to the given geometry of upstream and downstream components. In the ideal case the unit can be fitted into an existing flange connection.
The diameter of the inner melt channel can be chosen depending on the given geometry of the extrusion system. For the prototype system in this project a melt channel diameter of 20 mm was chosen.
A gear motor drive is mounted directly to the body of the rheology sensing system. This allows the complete measurement system to be added without an additional support, if the extrusion line is stable enough to carry the weight of the unit, which has a mass of only 135 kg.
A disadvantage of this set-up is a possible heat transfer from the measurement unit body to the drive system. Since the viscosity changes with temperature it is important to control the temperature of the system carefully. In order to achieve a uniform temperature distribution of the measurement unit, the body of the rheology measurement unit is surrounded by jacket heater elements and temperature measurement ports for thermocouples. The heaters are controlled by the control unit attached to the system. With this equipment in a standard extrusion set-up it was verified that the temperature could be controlled with an accuracy of less than one K at typical extrusion temperatures of approx. 250 °C.
During the first phase of the project the whole mechanical setup was successfully tested with various polymers in order to proof the design concept.
One idea was that it should be possible to add the rheology sensing system to a given extrusion line. As discussed before, the mechanical setup was designed in such a way that the system could be added with simple adaptions into an existing flange connection of such a line, and, due to the small mass, in many cases without additional support. Moreover as part of this concept the control system must integrate all necessary functions for operating the device and should be able to communicate with existing control systems of the line. A control system based on Siemens S7 technology was chosen, which allows free programming for additional requirements. The control system incorporates the following functions: Temperature control of the whole unit including power supply for heaters based on an integrated temperature controller unit, control of the motor drive and speed of the gear pump, collecting all analogue signals from pressure and temperature transducers, calculating shear rate and viscosity based on measurement data, communication with the operator (human machine interface HMI) with a touch panel for inputs, displaying all relevant data on touch panel, communication with other devices via analogue signals, serial communication, profibus system or Ethernet communication. Figure 220.127.116.11 (see separate attachment) shows a picture of the prototype system and a screenshot of the HMI interface during operation
In the current design the system works with a constant flow through the capillary which means the viscosity measurement is carried out with a constant shear rate. The exact shape of the viscosity function is not detected. On the other hand this allows a higher measurement frequency which is important with respect to closed loop control applications as planned in the NanoOnSpect project. For future application it is planned to equip the system with a pump speed sweep which allows the complete viscosity function to be detected at reduced measurement rates.
During the NanoOnSpect projects two prototypes of the system were constructed. They were successfully used during material and process characterisation at the Fraunhofer ICT and Aimplas (see WP3) and at the sites of additional project partners during the industrial case studies (WP5). For dissemination the prototype system was presented during the K-Show in October 2013 in Düsseldorf, Germany.
3.1.5 Optical spectroscopy sensing system
The main aim of the work carried out in the topic “Optical spectroscopy sensing system” was to investigate the capabilities of the optical spectroscopy sensing system for the characterisation of nanocomposites.
Firstly the work focused on the implementation of the optical probe in the onBOX and carrying out on-line test measurements with CNT-nanocomposite materials.
The objective of the Raman measurement was to develop an evaluation method to determine the CNT-concentration. This analysis is essential to generate precise material properties of CNT-nanocomposite material.
The intensity of Raman scattering from transparent or non-absorbing samples can be described with an equation (Eq. 18.104.22.168and Table 22.214.171.124 in the attachment).
With constant spectrometer parameters the Raman intensity is proportional to the sample concentration. Black samples like the CNT dispersions affect the intensity of the excitation wavelength of the laser and the Raman scattering by absorbing the light. A non-linear relationship between the Raman intensity and the sample concentration (Eq. 126.96.36.199) is no longer applicable.
The absorption of the excitation wavelength (785 nm) can be described by the Lambert-Beer-Law.
In addition to the absorption of the excitation wavelength the Raman scattering of the sample is also reduced due to the absorption of the CNTs, which can be described by the Lambert-Beer-Law. An increase of the CNT amount causes an increase of the intensity of the Raman scattering and an intensified absorption of the excitation wavelength and the emitted photons from the sample.
Figure 188.8.131.52 shows the Raman spectra of the PP/CNT blends with CNT-concentrations of 0.1%, 0.2%, 0.3%, 0.4%, 0.5%, 0.6%, 0.8%, 1.0%, 1.2%, 1.5% and 2%. The Raman peaks assigned to the spectra are listed in table 184.108.40.206.
The structure and properties of the spectra were analyzed by partial least squares regression (PLS). The optimal regression model was achieved with the limitation of the spectral range to a range between a Raman shift of 530 cm-1 and 1800 cm-1, and with peak normalization. Moreover the analysis has shown that the model for whole the concentration range is evidently not applicable. For this reason the CNT concentration was divided into two concentration ranges. The first model was developed for the low CNT concentration range between 0.1 to 0.8% and the second for the high CNT concentration range between 1 to 2%. The optimal number of PLS factors was 4 for the low range and 3 for the high range.
Figures 220.127.116.11a) and 18.104.22.168b) in the attachment present the correlation between the CNT-concentration estimated from the calibration resp. cross validation model and the reference concentration. The blue points are the calibration points which were used for the generation of the regression model, and the red points are used for the generation of the validation model. The calibration model for the low range is characterised by a slope of 0.983 and an offset of 0.007.The R2 value of the calibration model is 0.982 REMSEC is 0.028 and the REMSEP is 0.053. The calibration model for the high range is characterised by a slope of 0.96 and an offset of 0.05.The R2 value of the calibration model is 0.96 REMSEC is 0.075 and the REMSEP is 0.182. Figure 22.214.171.124a) and 126.96.36.199b) show the results of the multivariate analysis of both concentration ranges. The average of the predicted absolute errors which were obtained by cross-validation is 0.05% for the low range and 0.18% for the high range of CNT concentration. Consequently the mean error of the predicted result for low concentration is bigger than for high concentration. With an increase of the CNT concentration the model becomes more accurate. The outlier between the concentration limits can be explained by the fact that the extruder may still have residuals from the earlier batch due to too low stabilization time while changing to higher CNT concentration .
The second focus in optical spectroscopy was the study of PP / nanoclay samples by VIS/NIR spectroscopy. The aim of this investigation was to obtain samples with different degrees of clay exfoliation and determine the clay concentration in the polymer matrix. Therefore the extrusion parameters were varied during the processing of the NC-material.
The NIR-wavelength range was used to determine the clay concentration (see Fig. 188.8.131.52 in the attachment). It was observed that the NIR signal and clay concentration have a defined relationship. The NIR signal decreases with increasing concentration; the effect is due to the higher scattering of light on scatter points/ clays. Figure 184.108.40.206 shows linear fits at 1317 nm at a concentration range between 0% and 5%.
In addition, systematic investigation of the degree of clay exfoliation has been carried out. Figure 220.127.116.11 shows the unprocessed raw VIS- spectra.
For the interpretation the turbidity theory was used. The turbidity theory has been described in detail inter alia in the following literature , .
The turbidity describes the wavelength radiation power attenuation of light passing through a disperse system. The degree of turbidity depends primarily on the shape, size, quantity, and the optical constants of the scattering particles and the matrix material.
Figure 18.104.22.168 shows a schematic view of the generation of a prediction model to determine the exfoliation degree of the clays by VIS spectroscopy. The prediction procedure can be summed up in the following steps: First, building of a calibration model with reference data which were determined by the turbidimetric method. Second: prediction of the exfoliation degree of clays by the PLS-model.
The partial least regression model (PLS) (s. Fig. 22.214.171.124) was used for the prediction of the optical average agglomeration particle size (exfoliation degree) of the remaining measured spectra (s. Fig. 126.96.36.199).
A total of 2469 spectra were evaluated by the PLS-model. Figure 188.8.131.52 shows the calculated results. It is shown that the multivariate model can make very precise predictions of the optical average agglomeration particle size.
FOS Fiberoptical probes for inline spectroscopy:
Inline spectroscopy applied to extruders in order to develop and process new nano composite materials can lead to a better understanding of the mixing process and even its chemistry. These optical probes and the fiberoptical inserts have to survive at permanent high temperatures up to 400 °C without any degradation of their optical properties.
In the NanoOnSpect project FOS improved and developed fiberoptical probes. Flat sapphire window probes: These probes are the standard optical probes for melt spectroscopy applications for e.g. melt color measurement in reflection and transmission mode. The aim was the construction and testing of fiberoptical inserts with high optical efficiency and safe handling. For measurements in transmission mode the best setup is to use a collimated beam with an appropriate beam diameter. A beam diameter of Ø4 mm gave the best results. The transmission is independent of the width of the measuring gap. Figure 184.108.40.206 shows this optical probe mounted at the onBOX-system. The corresponding measuring setup is shown in figure 220.127.116.11 consisting of a light source, optical fibers, probes mounted at an extruder and a spectrometer unit for the transmitted intensity.
FOS Flat sapphire window probes for reflection mode measurements
Most technical polymers are too opaque to transmit light through a few millimeters of material. In these cases the optical measurements have to be carried out in light reflecting mode operation. The principal setup is shown in 18.104.22.168. The reflected power using a basic setup is poor, which makes it difficult to obtain a good spectral response. Within this project FOS has developed fiberoptical inserts with improved optical efficiency.
FOS NIR Fiberoptical ATR-probe
NIR spectroscopy is a powerful tool for the off-line characterisation of polymers. However, up to now no fiberoptical probes are available for inline use in extruders. In NanoOnSpect FOS developed an NIR-ATR (Attenuated Total internal Reflection) probe. The principle function is shown in 22.214.171.124. The emitted light is reflected at the surface of the measuring tip and the optical wave penetrates the melt a few micrometers, giving a characteristic attenuation which can be detected. Figure 126.96.36.199.14 shows the design of the frontend of the sapphire ATR –probe and figure 188.8.131.52 shows a typical spectrum measured with this probe.
Figure 184.108.40.206 shows the fiberoptical ATR-probe mounted at an extruder for testing trials under real process conditions. All FOS fiberoptical probes can be used in the spectral range of 0.3 µm – 5.5 µm with excellent optical efficiency.
3.1.6 Ultrasonic sensing system
The density sensor developed during NanoOnSpect uses some of the known properties of ultrasound to provide a density reading in a very hostile environment like extrusion. The sensor withstands working temperatures up to 300ºC and pressures up to 100 bar. The sensor is designed in stainless steel to provide mechanical robustness and chemical protection. It is composed of three parts: the ultrasonic sensor itself, an adapter to be placed in the extruder and the processing system. The sensor was tested with several material like PE, PP, PC, TPE, PVDF and several fillers like calcium carbonate, carbon nanotubes, cloisite, etc.
The sensor is still under development, but a commercial product is expected within the next three years. A complete picture of the whole measurement system can be seen in Figure 220.127.116.11.
The sensor works using the known relationship between the acoustic impedance and density of a material. The acoustic impedance measures how a material reacts to the flow of an acoustic wave. The units for the acoustic impedance are Rayls and they are equivalent to 1 Pa·m-1·s. If using the absolute value of the acoustic impedance, The absolute value of the acoustic impedance can be calculated as the product of the acoustic speed in the material and its density.
To measure the density, it is therefore necessary to calculate the acoustic speed and the acoustic impedance. The acoustic speed can be measured precisely. However, measuring acoustic impedance is more complicated and the precision in the measurement is lower than in the case of the acoustic speed. The method that Ateknea has developed is based on the reflection coefficient between the polymer and the acoustic sensor. However, first of all, the acoustic impedance of the sensor has to be calibrated and well known, since the acoustic properties of the materials change significantly with temperature, and to a lesser extent with pressure.
One of the most critical issues in the development is the temperature of the ultrasonic transducer. If the temperature of a piezoelectric material (from which the ultrasonic transducers are made) exceeds its Curie temperature, it loses its piezoelectric characteristics. In the design of the ultrasonic sensor, special care was therefore taken to reduce thermal conduction from the extruder to the ultrasonic sensor. The material used is stainless steel, for several reasons. The thermal conductivity of stainless steel is quite low compared with other metals, like aluminium or copper. Stainless steel 304 has a thermal conductivity of 17 W·m-1·K-1, while aluminium thermal conductivity is 237 W·m-1·K-1.
In Figure 18.104.22.168 it can be seen how the ultrasonic transducers have been separated more than 150 mm from the extruder, but maintain the mechanical contact to allow the acoustic waves to travel from the ultrasonic transducer to the polymer melt.
During the trials made during the NanoOnSpect project, several different blends and pure materials were tested. The pure materials, with known solid viscosity were used to calibrate the system, while the performance of the system was tested with blends. The consortium also tested the degradation of materials with the density system. Most tests were carried out on LDPE (low-density polyethylene), which is widely used in the polymer industry. The partners tested pure polypropylene, polycarbonate and polyvinylidene fluoride. Some of these results can be seen in Figure 22.214.171.124 (see separate attachment). The diagram shows how the densities changed with the material following the expected behavior. The most remarkable density change occurs when there is a change from PP to PVDF, since the solid density of PP is around 900 kg·m-3, and the solid density of PVDF is around 1700 kg·m-3. Regarding the blends, several different additives were tested, namely carbon nanotubes, calcium carbonate and cloisite. The difference of density created by the additives is usually imperceptible if the amount of additive is lower than 2%. The clearer example is shown in Figure 126.96.36.199 where it can be seen that the density remained the same for pure PP and PP with 1% cloisite. However, when 3% cloisite was used, the change was more visible.
The work presented here is just the beginning of a development involving mechanical design, electronic design, signal processing and material engineering. The sensor has clearly demonstrated its ability to measure density in very harsh environments, but the robustness and the temperature stability of the sensor have to be improved before the sensor can be commercialized.
3.1.7 Microwave spectroscopy sensing system (ICT)
In order to measure the content and distribution of nanomaterials in a polymer melt two different microwave sensors and a measurement system were developed within this project. Both sensors are based on off-line measurement techniques used to determine the dielectric properties. Those measurement techniques were adapted and designed to be used in an extrusion line. The developed measurement system generates and measures the microwave signals and enables a data communication via RS232.
Cavity perturbation method: The cavity perturbation method is a volumetric measurement in transmission. The signal is irradiated into the cavity, passes the sample and is measured on the other side of the cavity.
A cavity has a magnitude of the resonance peak (quality (Q)) and a resonant frequency, which change when a material is inserted (Figure 188.8.131.52 in the attachment).
The resonance frequency shifts to a lower frequency depending on the dielectric constant of the sample. The Quality Q shifts to a lower amount depending on the dielectric loss of the sample. The material can be characterised based on the changes in these factors.
Corbino method: The Corbino method is an open coaxial cable setup suited for near-surface measurements in reflection. The signal is irradiated into a coaxial cable, indicating that the sample and the reflected signal have been measured (Figure 184.108.40.206).
The material can be characterized based on changes to the magnitude and phase of the reflected signal.
Sensor and measurement system setup
Resonant cavity sensor setup: In Figure 220.127.116.11 (see separate attachment) a schematic diagram of the resonant cavity sensor is shown. The lower part of the cavity is a sintered boron nitride block with a through hole for the melt flow. The upper part is a manually-compressed boron nitride powder in which the antenna extends. The system can be tuned using a shorting plunger. The whole setup is covered by a robust metallic housing.
The resonant cavity sensor is shown in Figure 18.104.22.168. The sensor can be mounted in the extrusion line via flanges and does not disturb or constrict the melt flow. The resonant frequency of the system is between 2 and 3 GHz. The connection to the measurement system is established via a DIN 7/16 coaxial port.
Corbino sensor setup: Figure 22.214.171.124 shows a schematic diagram of the Corbino sensor. The setup is an open coaxial conductor. The inner and outer conductor consist of copper. The dielectric consists of SiO2 ceramic.
The Corbino sensor is shown in Figure 126.96.36.199 and can be mounted into the extrusion line via ½” 20 UNF thread which is similar to conventional temperature and pressure sensors.
The Corbino sensor can be used from DC to 3 GHz and can be connected to the measurement system via SMA coaxial port.
Measurement system setup: The measuring system (Figure 188.8.131.52) operates in the range from 1 to 3 GHz. Both dielectric sensors can be connected to the measuring system via an SMA coaxial port. The measuring system converts the signals to processable data and provides the measurement data to the ANN via an RS232 interface.
Resonant cavity sensor on-line trials:
Measurements on neat PP: In Figure 184.108.40.206 the effects of extrusion temperature changes with neat PP on the resonance peak is shown. The temperature variations have a little effect on the shift of the resonance peak and no effect on the quality.
Variations of the rotational velocity with neat PP have no effects on the shift of the resonance peak and the quality (Figure 220.127.116.11).
PP/Clay material system: In Figure 18.104.22.168 the effects of the variation of clay content on the resonance peak are shown. The resonance frequency and quality factor changes when the clay content is varied. Material with the highest clay content showed the lowest resonance frequency and the lowest Q-factor.
By changing the rotational velocity the resonance frequency remains constant and the quality factor changes (Figure 22.214.171.124). With higher rotational velocities better dispersions should be achieved leading to a lower Q-factor. It is therefore assumed that the dispersion of the clays influences the quality factor.
PC/CNT material system: The resonance peak from neat PC shows a resonance frequency at 2.5 GHz (Figure 126.96.36.199). In the presence of CNTs the resonance peaks shift to higher frequencies in contrast to the clay measurements. It is therefore assumed that the presence of CNTs changes the electromagnetic field mode in the cavity. Nevertheless the CNT content can be measured. The resonance frequency of PC+0.25% CNT is at 2.7 GHz. With increasing CNT content the resonant frequency and quality factor shift to lower values.
With increasing rotational velocity the resonance frequency and quality factor shift to lower values (Figure 188.8.131.52). As sawn in the extrusion trials, better dispersions are usually archived with higher extrusion velocity. Therefore a relationship between dispersion quality and resonance frequency and quality factor is assumed.
Corbino sensor on-line trials:
PP/Clay material system: In Figure 184.108.40.206 the Corbino sensor measurements for the variation of clay content are shown. The magnitude in decibel of the scattering parameter S11 is plotted against the frequency. As can be seen the Corbino sensor cannot measure the clay content variation.
PC/CNT material system: With the Corbino sensor the variation of the CNT content can be measured (Figure 220.127.116.11). The measurements from 0% CNT content to 2% CNT content are an exact match.. With higher CNT contents the magnitude of S11 shifts to lower values. As increasing CNT contents should lead to better conductivities it is assumed that low magnitudes indicate high conductivities.
In Figure 18.104.22.168 the effects of the variation of rotational velocity on the reflection signal are shown. The measurements with 300 rpm and 600 rpm show no difference in the reflection signal. The 1100 rpm measurement seems to have a better conductivity, leading to a lower signal S11. It is therefore assumed that the dispersion of the CNTs influences the reflection signal.
HBH Microwave GmbH and Fraunhofer ICT have successfully designed and manufactured two different dielectric sensors for application in an extrusion line, based on a measurement system working from 1 to 3 GHz. Both sensor systems can be permanently used in extrusion processes.
Tests were successfully performed with nanoclays and CNTs. The resonant cavity sensor can measure the integral characteristics of the polymer melt and works for the CNT/PC as well as the Clay/PP material system. The Corbino sensor measures the polymer melt at the polymer surface and works for the CNT/PC material system. The content and distribution of clays in polymer nano composite can be detected. Content and distribution of CNTs in the polymer nano composite can be detected.
3.1.8 Development of the onBOX device (Gneuss)
In accordance with the concept and objectives of the NanoOnSpect project a one-box characterization unit (onBOX) was developed for the on-line characterisation of nanocompounds. The requirements were as follows:
- Integration of all measurement systems developed in work package 1 into one single measurement unit
- The onBOX should be easy to adapt to a common existing extrusion line
- Collecting and recording all data from the measurement units
- Transferring sensor signals and measurement data to the Intelligent Module (Artificial Neural Network/Expert System)
- Optional: Transferring calculated machine parameters from the Intelligent Module to extrusion equipment.
Moreover the design of the onBOX should be modular. In order to achieve a cost-efficient solution it should firstly be possible to add only the measurement systems from work package 1 that are necessary for an arbitrary application and secondly to add additional measurement units which are not yet considered in this cooperation project.
It was therefore agreed that all measurement systems from wok package 1 should be developed as stand-alone systems. This means that any measurement system should be able to be used as a stand-alone unit. If possible such systems should use sensor probes that fit to 1/2” UNF drilling which is commonly used in extrusion technology to place sensor probes like temperature and pressure transducers. In order to integrate these measurement systems into the joint control system of the onBOX, the measurement systems should contain a signal exchange interface based on analogue signals (4-20 mA or 0..10 V), serial communication (RS232), profibus or ethernet.
As in all measurement applications it is important that the measurements have the smallest possible effect on the process to be measured. Especially in extrusion technology a rheologically optimized melt channel system is important, because many polymers and compounds tend to degrade thermally: for this reason a minimal extrusion length and the avoidance of dead spots in the melt channel system should be achieved if possible.
When using the viscosity sensing system (section 3.1.4) a small part of the polymer melt is separated from the main melt channel by means of a high-precision metering gear pump. This polymer is pumped through a precisely-manufactured slot capillary. All sensing techniques which do not detect process conditions but rather material properties can be placed in this side stream which is generated by the rheology sensing system and afterwards fed back to the main polymer flow. This is carried out by a special sensor block which allows placing probes. The advantage is that no additional sensor drilling in the main channel is necessary, so additional costs in retrofit applications, and the potential need for additional extrusion length due to mechanical reasons, can be avoided. Apart from an additional extrusion length of 140mm when using the rheology sensing system, no additional disadvantage for the rheological design of the main polymer flow is created when an additional sensor is placed in the rheology sensing system. Figure 22.214.171.124 shows how various sensors (yellow) are placed into the sensor block of the onBOX.
Moreover the small dimension of the channel geometry of the side stream allows an easy and straightforward realization of transmission measurements at the small cross section areas. Also probes that have to be placed longitudinally to the flow direction can be placed easily, because the flow direction in this system has several 90° changes in direction. The sensor block can be modified in a wide range regarding the shape for the sensor drillings. Unfortunately, for various reasons (e.g. size of the probe), not all measurement techniques can be used in the side stream. One example is the probe for the resonator method in microwave spectroscopy (section 3.1.7).
In the case of these probes the best method is to add an additional disc-shaped flange containing the probe upstream or downstream of the rheology sensing unit with the sensor block.
Each measurement system can be connected to the joint onBOX control system based on analogue signals (4-20 mA or 0..10 V), serial communication (RS232), Profibus or Ethernet.
The joint control system of the onBOX allows the measurement data of these control systems to be displayed. Moreover control commands can be entered into the onBOX control and are distributed from there to the control systems of the measurement units.
This modular concept allows additional sensors to be added which were not considered in the original project proposal. One example is the electrical conductivity sensor developed by the ICT. This sensor can be placed in the sensor block and via the abovementioned interfaces the signals can be fed into the onBOX control. However, this has not yet been demonstrated in the project.
If necessary the flow through the side stream channel of the onBOX can be controlled by the onBOX depending on the requirement of the measurement modules. For example the viscosity measurement requires different pump speeds depending on the selected shear rate for the measurement. Another application is to stop the melt flow for thermal conductivity measurements.
The data exchange between the modules was described above. In order to organize the communication to the Intelligent Module and to the extrusion equipment the following architecture was chosen. The control panel of the onBOX system is equipped with a programmable logic controller (PLC) and a human machine interface HMI (touch panel display). The PLC allows analogue signal exchange and simple digital data exchange. For enhanced digital communication adapters for profibus standard and ethernet communication are included in the control cabinet of the control panel. With this equipment and the necessary programming, communication via profibus connection and profibus protocol and via ethernet and profinet protocol is possible. Both standards are currently widely used in extrusion and machine equipment. In order to bridge the gap to other computing technologies and to meet the latest standards for machine controllers the system is also equipped with an OPC server. This allows additional communication with an ethernet connection and OPC UA protocol. During this project communication with the Intelligent Module developed in work package 4 (section 4.3.) was achieved.
With the developments described above and carried out during the project it was possible to meet the requirement for the modular mechanical and electrical integration of all sensors developed in work package 1 into one system, and the requirement for data exchange with the Intelligent Module and extrusion equipment. The solution allows a simple and straightforward adaption to existing extrusion lines. This was successfully demonstrated during the industrial case studies in WP5.
3.2 WP2 New compounding process technology and automation
3.2.1 Main objectives
WP2 covered work on the new compounding device Nexxus Channel, on the optimization of conventional twin-screw extrusion for nanocompound processing and on the combination of both systems together as a new flexible compounding device.
3.2.2 Nexxus compounding technology
Nexxus Channel has developed a special method and equipment to fabricate CNT nanocomposites based on thermoplastic carriers. (Fig. 126.96.36.199 in attachment)
The novel system aims to overcome the well-known limitations in conventional technology in relation to agglomerate dispersion into polymeric liquids as well as the safety conditions when handling nano-powders in a dry state so as to minimize or completely eliminate the risks associated with VOCs (Volatile Organic Compounds).
Nexxus core technology replaces the well-known bi-dimensional flow occurring in a screw extruder with a simple annular, mono-dimensional flow. In a screw extruder the melt flow is chaotic as it travels either parallel to the screw flight or cross wise, according to a flow distribution which is difficult to predict and cannot be controlled. The relative motion of the smooth wall vs. the channeled wall is reversed in Nexxus. The helical channel is replaced with an annular, static channel machined into the barrel while the driving wall is a smooth rotor, which drags the melt flowing down the static channel. As a consequence all the main processing functions (e.g. melting, mixing, degassing, pumping) take place in Nexxus in a shorter time and more effectively. (Fig. 188.8.131.52a 184.108.40.206b)
Outstanding advantages in polymer processing are:
• Reduced SEC (specific energy consumption). Tests have demonstrated that between 30T and 50% less energy is needed to melt the most popular volume polymers (e.g. polyolefins) (Fig. 220.127.116.11a)
• Reduced residence time which minimizes the thermal degradation risks (Fig. 18.104.22.168b)
• Very effective mixing capability, especially tested with glass fibers
• Outstanding degassing capability (22.214.171.124c)
• Extra narrow quality distribution (close to zero) due to the mono-dimensional flow (Fig. 126.96.36.199d)
• Exceptional capability to control all processing parameters like speed, temperature, pressure, residence time, etc. in all critical process locations
• Compact design and low space consumption
• In parallel with Nexxus technology a novel method has been developed to enhance the nano-agglomerate dispersion by some orders of magnitude, while preserving the environmental safety. It has been recognized that all nanoparticle agglomerates in dry powders diffuse easily in the environment or require very special protective clothes. This circumstance poses serious handling problems, and special rooms are required to operate the equipment. (Fig. 188.8.131.52)
• Nexxus Channel has developed a novel system:
o To disperse nanoparticles in water. This step is easy and effective for both dispersion efficiency as well as dusty powder suppression.
o To mix the above waterborne dispersion with other polymeric substances that render the solid content flowable once the water is removed
o To remove the water
o To add further polymers where needed to dilute the nanoparticles to the desired concentration
• The most critical issues arising during the development of the method were:
o The selection of the proper surfactant when preparing the waterborne dispersion of nanoparticles
o The selection of the polymer substance to add to the waterborne dispersion
o The water removal
o The water removal was dramatic as more than 75 % wt water (Fig. 184.108.40.206) has to be removed in many cases. The issue of water removal can be appreciated considering that the vapor pressure at a polyolefin is of the order of several bar and thus the vapor develops at a very high speed, carrying many nanoparticles with it. To avoid this a novel Nexxus-Vapor Melt Separator (VMS) system (Fig. 220.127.116.11) was added. Using this system outstanding water fractions can be readily stripped under controlled vapor speed, with zero particle drag.
• To test the validity of the system two parallel systems were constructed:
o One with VPS in combination with a Nexxus Degassing Unit (Fig. 18.104.22.168)
o One with VPS in combination with a co-rotating twin-screw extruder (Fig. 22.214.171.124)
• In both arrangements a multimeter carried out on-line measurements of the melt resistivity (Fig. 126.96.36.199).
• Optical analysis reveals the superiority of CNT dispersions in products achieved by Nexxus technology (Fig. 188.8.131.52)
3.2.3 Optimised conventional extrusion techniques
NanoOnSpect also used twin-screw extrusion as a conventional compounding process for the nanocomposites. During the project duration four different extrusion lines from three different twin-screw extrusion line producers were used. Depending on the optimal process parameters needed, different dosing systems and allocations were considered. The screw configuration of the compounding line was the key for the material properties.
The compounding extruders are modular machines. For this reason, numerous factors must be considered in the machine configuration. The expertise of the NanoOnSpect partners meant that the screw configurations used in the project were limited.
The following parameters were considered in order to obtain a general understanding of the whole process and its relation with the final properties of the obtained compounds.
Adjustable parameters: Screw design, barrel design (material feeding protocol), temperature and throughput
Fixed compounder properties: L/D ratio, max. engine power (KW), max. rpm
Resulting process variables: Torque, die pressure, residence time, dispersion quality, melt temperature, SME (specific mechanical energy)
At AIMPLAS the Coperion ZSK 25 was used, at ICT the Leistritz ZSK 27, at Colorex the Berstdorf ZE 25 and at Addiplast a Coperion ZSK 40 for the nanocomposite processing.
Modifications during processing were made on the dosing unit (allocation and dustless dosing) of the nanoparticles and the screw configuration.
Dosing: The processing of nanocomposites requires modifications of the extrusion line for safety and health reasons. The dustless dosing and processing of the nanofillers is ensured by using closed dosing and feeding systems. The feeder is filled separately prior to the trials and closed, and all dosing pipes and fittings are fitted to be tightly closed to the particles. Depending on the nature of the nanoparticles, the dosing is carried out either into the main feeder or via a secondary side feeder. When the particles are added to the main feeder they are mixed with the still solid polymer granules, resulting in a milling step which is advantageous for very hard and stable particles. Using more sensitive particles the side feeder is preferred. In this manner the particles are dosed carefully into the molten polymer.
Screw configuration: Different screw configurations, resulting in different process variables, were designed and tested during compounding. The screw design was chosen in order to ensure high shear stress for the particle deagglomeration and distribution. The screw design was simulated for the material system using Ludowig software, which also enables up-scaling for the later production-scale processing.
3.3 WP3 Material and process characterisation
3.3.1 Main objectives
The main aim of WP3 was to develop processing and material understanding for two of industrially-relevant polymer nano-composites (PNCs).
To achieve this objective and to correlate the final composite properties and processing parameters it was necessary to carry out the following steps:
• Define PNC case studies: polymer selection and percentage definition of nanoparticles (CNTs and nanoclays).
• Definition of both experiment designs including compounding parameters, polymers types and nanofiller percentages.
• Development of compounds according DoE, and their characterisation to determine the effect of the process parameters on the properties.
The results of this WP were used to feed the Intelligent Module developed in WP4 with data on processing parameter modifications and their impact on nanocomposite particles. The main objective was to obtain good correlations between on-line and off-line properties. The sensors developed in the project need to be able to measure a wide range of values. The correlation between on-line and off-line characterisation needed to be performed with the best and the worse process conditions.
Compounding is a very complex process with many different parameters. The range of compounding parameters that can affect the final PNC properties is high. These parameters were taken into account to understand the whole process; its relation with the final properties of the compounds obtained was also considered.
Polymer selection was based on two polymeric materials (PP and PC), and the nanoparticles were CNTs and nanoclays, one black and the other colorless materials. Different percentages were used to assess the final PNC properties.
3.3.2 Polycarbonate – CNT material system
A co-rotative twin-screw extruder was used for the trials. The first DoE was used to investigate the CNT percentage necessary to achieve the percolation threshold for electrical conductivity. After that, a complete dispersion analysis was carried out with 1% CNT samples. The goal of dispersion analysis was to study the effects of compounding parameters in CNT dispersion and distribution as well as the correlation with electrical conductivity in the pressed plates. The parameters employed to process CNT compounds were selected taking into account previous partners’ know-how in conductive plastic materials. The speed of the screw, throughput value and screw configuration were selected as processing parameters to assess the material properties. A characterisation protocol was established in order to measure the electrical conductivity and dispersion.
A clear effect of the screw configuration was detected in the electrical conductivity. When the screw configuration was designed with elements which provide high dispersion of CNTs, a greater influence on the electrical conductivity was observed. In the case of more distributive elements, few differences were observed in the electrical conductivity.
The influence of polymeric matrix polarity on the filler dispersion and consequently the electrical conductivity was also observed.
A comparison between the on-line resistance and off-line resistivity was carried out, with the help of the compression of the pellets obtained from the on-line measurements. On-line resistance techniques provided very accurate measurements which have more statistical significance. Also, other on-line devices such as thermal conductivity sensors, VIS/NIR, microwave spectroscopes, ultrasonic sensors etc. were used to measure direct and indirect plastic properties
Off-line characterisation was carried out to measure: electrical resistivity/conductivity, dispersion (optical microscopy/SEM), rheology, thermal conductivity and permeability. Electrical conductivity and resistivity were measured with injection and compression samples.
The dispersion analysis allowed the study of the effect of compounding parameters and also the correlation with electrical resistivity. AIMPLAS developed some trials with 0.25 % up to 5 % of CNTs. It could be observed that the polarity of the polymeric matrix influences dispersion and consequently the electrical conductivity.
Under the percolation threshold of the electrical conductivity compounding parameters have more influence on the resistivity compared with the first off-line trials. The main objective was to reduce the CNT amount to obtain the percolation threshold in a PC matrix. Capillary rheology characterisation was used in order to assess the best amount of CNTs in the polymeric matrix.
The viscosity measurements showed a more uniform behaviour with compounds developed for higher production rates. The on-line resistance and off-line resistivity showed a parallel responses at high reproducibility. The resistance / resistivity are higher when the masterbatch is used. Better dispersion is obtained with compounds obtained from the masterbatch, but electrical conductivity decreases by one order of magnitude, which probably means that the conductivity is impaired since the CNTs are broken when high shear is introduced.
A technical data sheet was prepared at the end of the project with the best formulation obtained with PC and CNT.
3.3.3 Polypropylene – Clay material system
A co-rotating twin-screw extruder was also used for the processing of PNC polypropyle/clay. The aim was to improve the viscosity and the permeability of the compound and to gain data sets for the Expert System used in WP4 and validated in WP5.
Trials were planned using Design of Experiment and carried out using Fraunhofer ICT’s extrusion line Leistritz ZSK 27 using closite 15A in concentrations between 1 and 3%. Neither the screw configuration nor the throughput had a significant influence on the viscosity of the product, which was caused by the good dispersion of the platelets at all processing parameters, can be seen for example in Figure 184.108.40.206 (in the attachment).
As the compounding process of nanoclays could be optimized for a polypropylene (PP) matrix, this material system was studied in more detail for the interpretation of the newly developed sensors. Especially the microwave resonator sensor, the optical spectroscopy and the capillary rheometer were successful in characterisation of the clay compound at different processing parameters, as described more in detail in WP1.
A technical data sheet was prepared at the end of the project with the best formulation obtained with PP and nanoclay.
3.4 WP4 Intelligent Module
3.4.1 Main objectives
The main objective of the Intelligent Module in NanoOnSpect is to provide advanced process control by carrying out the automatic loop control for the extrusion line during the compounding process. The Intelligent Module gathers all the values coming from all the sensors, machine parameters, actuators, and self-correlated values of desired off-line properties in order to carry out the process control loop and on-line parameter adjustment of the extrusion line.
The Intelligent Module performs the loop control by first reading all the input values from the extrusion line through the onBOX. After this all input values are passed to the trained Artificial Neural Network (ANN), which performs the on-line correlations of the desired property values (resistivity and conductivity). Finally, all input values from the extrusion line and the values coming from the ANN are used by the Expert System in order to perform the automatic adjustment of the compounding process (Fig. 220.127.116.11 in the attachment).
The communication between the Expert System and the extrusion line is carried out using OPC Server-Client Protocol through an Ethernet network. The Expert System first communicates with the extrusion line HMI, and the HMI pass the data to the extrusion line. This communication was developed using the OPC protocol (Fig. 18.104.22.168).
3.4.2 Artificial Neural Network
The Intelligent Module contains an Artificial Neural Network Module to calculate the in-line property values corresponding to the desired off-line properties, such as resistivity and conductivity. These desired off-line compound properties are normally calculated in the lab using pellets obtained from the compounding trials. These pellets are made from compounds which are compressed and moulded, and are used for the resistivity and conductivity measurements. A software sensor capable of calculating these property values in-line is consequently of significant interest for the compounding industry. It would save a lot of valuable time by eliminating the need for many different trials, and it would considerably decrease the amount of waste produced in the compound trials.
In order to develop the Artificial Neural Network (ANN) for in-line correlation of the desired off-line properties, several steps were accomplished as described in Fig. 22.214.171.124 (in the attachment).
• First, a large data set was provided for the training of the ANN. This data set was created using all the trials performed in the lab, and contained all the value parameters of the process such as values from the sensors, machine parameters and machine actuators, together with the values correlated off-line. In NanoOnSpect, the partner in charge of the characterisation process was AIMPLAS. AIMPLAS collected all the data from the different trials and carried out the off -line measurement to produce the data set used for training the ANN.
• Secondly, a training application was used to look for the ANN that could carry out the best correlations. This is an interactive process in which different ANN architectures and learning parameters are used in order to simulate the results of the correlations using a training. The outcome of the training stage is a trained ANN.
• Third, the trained ANNs with the best scores (smallest training error) are tested using a test data set with a test application. This test application also allows us to simulate future correlation values using input values from different trials. The outcome of the test is to rank the best ANN that provides the smallest error when correlating real input values.
• Finally, the best ANN is selected and integrated into the Expert System GUI in order to be used for the in-line calculation of the desired property values (resistivity and conductivity).
This procedure was carried out using different training and testing data sets. The results of the case studies carried out at Colorex allowed the consortium to test and validate the results of the ANN in a real industrial scenario (Fig. 126.96.36.199).
3.4.3 Expert System
The Expert System module is in charge of adjusting the extrusion line parameters in order to obtain the desired properties from the compound process. The ES receives all input parameter values coming from the sensors, extrusion line and ANN module.
From a top level point of view, the ES is the brain of the Intelligent Module and uses all the input values from the extrusion line system. It receives the correlated parameter values from the ANN and uses these values in order to adjust the extrusion line parameters so that the desired compound output is achieved.
The Expert System module has a very close relationship with all the other modules of the Intelligent Module:
1. It directly receives the correlated parameters from the ANN.
2. It reads all input values of the IM module and communicates all the output values to the extrusion line using the OPC communications module.
3. It is integrated in the GUI, and all input and output values are represented in the GUI of the IM module. Within the GUI there is an explanation box that shows a track of the changes performed by the ES.
The values of all these inputs are read by the ES engine and matched against the rules that compose the control logic of the ES. These rules these rules regulate adjustments to the extrusion line parameters. They compose the basic logic for the reasoning, which is the same type of reasoning that an expert compounder will carry out by looking at the current state variables of the compounding process (Fig. 188.8.131.52).
The creation of the rules is performed using knowledge inputs from the compound experts: in NanoOnSpect the knowledge base was created using feedback from ICT, AIMPLAS and ADDIPLAST. Each package of the rules corresponds to a certain type of outcome in terms of compound properties. The desired rule package file can be selected from the Intelligent Module graphic user interface and the ES system module can be launched once the selected file has been loaded from the user interface. As a result the ES will carry out the automatic adjustment of machine parameters during the compounding process in order to keep the process within the certain range of production values.
The ES works in three different modes: automatic, semiautomatic and manual. The automatic mode carries out the parameter adjustment automatically, the semi-manual suggests values to the compounder and the manual mode allows the operator to fully control the process parameters (figures 184.108.40.206 220.127.116.11).
3.5 WP5 Application development and industrial case studies
3.5.1 Main objectives
In WP5 the developments made in WP1-WP4 were validated in an industrial environment.
3.5.2 Case study polycarbonate - CNT
The case study was carried out on a Berstorff ZE 25 extrusion line installed at Colorex Master Batch B.V. in the Netherlands, shown in Figure 18.104.22.168 (see separate attachment). The CNTs were fed with the polymer into the main feeder.
The onBOX developed in the project was used (shown in Figure 22.214.171.124) equipped with the following sensors: thermal conductivity, electrical resistivity, capillary viscosity, temperature, pressure, Raman spectroscopy, microwave spectroscopy.
A central control system collected all the sensor signals and an Artificial Neural Network was used to calculate the electrical conductivity.
The scope of the deliverable was to test the onBOX on a small production unit and evaluate its feasibility under harsh industrial conditions such as vibration, exposure to dust and parallel processing with other processing lines.
The characterisation system functioned as planned and it is definitely feasible to work in a production environment. The benefit is on-line monitoring of the stability of the material properties, enabling quality control. In detail the monitoring of the main parameters temperature and pressure, and also the advanced parameters such as viscosity, thermal conductivity, electrical resistivity and microwave and Raman spectra could be monitored and used as quality target values.
The on-line control will not only enable constant material quality, but also accelerate the development time of new materials, as it is possible to adjust the process parameters and monitor the dependence of the values on-line.
The significant advantage of the onBOX is that it is not limited to one set of materials. Colorex also foresaw the possibility of further developing the onBOX for use on polymers containing flame retardants, to measure dispersion or, for a masterbatch and compounds, to measure the color and color dispersion. This case study therefore demonstrates just one example of the different options that can be measured: more options for the sensors are available.
In addition the onBOX is easy to use, and once installed requires little maintenance. As far as Colorex could determine, the onBOX does not cause any disturbance in the extruder, which still performs as intended. Colorex has also not observed any influence of the extruder on the onBOX.
As far as personnel is concerned, Colorex believes the machine operator of the extruder can handle the onBOX in a safe manner and provide accurate data. To summarize from industrial point of view the onBOX will be a valuable addition to an extruder and can provide essential on-line data.
3.5.3 Case study polypropylene – nanoclay
The second industrial case study of the project was carried out at the ADDIPLAST plant in France. The goal of this case study was to implement the onBOX system on an industrial compounding line and to validate the Expert System of the Intelligent Module. The case study demonstrated the reliability of the system and its capacity to be used in an industrial environment as a decision tool for process optimization.
This case study was carried out on a Coperion ZSK40 co-rotating twin-screw extruder shown in Figure 126.96.36.199. The onBOX developed in the project was equipped with the following sensors: thermal conductivity (Hukseflux), capillary viscosity (Gneuss), temperature (Gneuss), pressure (Gneuss), NIR spectroscopy (ICT), microwave spectroscopy (Ateknea), 3-points temperature sensor (FOS), IR temperature sensor (FOS) and an ATR IR sensor (FOS).
Most of the sensors were set in a bypass channel which was continuously filled with the main polymer flow through a gear pump. The speed of the gear pump and the flow speed in the channel can be adjusted depending on the product.
Most of the data from the sensors can be seen and recorded on the onBOX controller, while experimental process parameters can be set manually: composition, extruder output and screw speed.
The compound chosen for this case study is a polypropylene (PP)/nanoclay(NC) compound. The PP is a homopolymer with melt flow index of 4 to 9 at 230°C under 2.16 kg. The NC is Cloisite 15A commercial grade from BYK Additives.
The process parameters were chosen in order to validate the project results, especially for the Intelligent Module.
The screw configuration is the intellectual property of Addiplast, generally used for industrial filled compounds.
The goal was to scan a wide domain of process and compound compositions to analyze the responses of the sensors and onBOX.
Based on experiments carried out during the tests, the rules for the Intelligent Module were determined and set in the system. The goal of the Expert System in the Intelligent Module is to provide the right process parameter adjustments (screw speed, output or composition) to maintain the product properties as defined in the production specifications.
The Expert System uses all the input parameters from the process that are passed to the Intelligent Module through the onBOX, and applies the rules and recommendations to determine the new value parameters for the process, if the product quality is out of the specifications. When the Expert System operates in automatic mode, it can directly adjust the process parameters in order to achieve the desired product specification.
For example, as a demonstration in the presence of the Project Technical Advisor Prof. S. Harris, the consortium ran a trial with a target composition of 3% NC, but the feeder recipe was set at 2.9 %. Following the data recorded by the onBOX (mainly the viscosity), the Intelligent Module recommend to set the NC content at 3% to keep the product within specifications. Furthermore, if the viscosity of the product was not the desired one, the Expert System adjusted the speed of the process in order to increase or decrease the viscosity value.
Thus, the consortium demonstrated that the Intelligent Module Expert System is able to detect small errors in the composition through the product characteristics measured in the onBOX, and give the corrections to adjust the process.
This demonstration on an industrial compounding line provided a variety of results: First of all, it was proved that the onBOX system can operate under industrial conditions with a nominal output. It was possible to run the system in a range of 80 to 160 kg/h with a PP/nanoclay composition. The limit was set by the workload of the extrusion line for the given compound, not by the capacity of the onBOX.
The sensors developed in the project can easily be added to the system on the onBOX channel or on the adaptor flanges.
Interesting results were also demonstrated with the sensors themselves:
The high sensitivity of the viscosity measurement (Gneuss) to the composition.
The potential of NIR sensors (ICT) to follow the filler content, and even the filler dispersion at low feeding levels.
The potential of the ultrasound sensor (ATEKNEA) to follow melt density variation.
The IR temperature sensor (FOS) allows a more accurate monitoring of the real melt temperature than regular thermocouples.
Finally, the potential of the Intelligent Module Expert System to follow, survey and correct the composition and the process to keep the product within pre-defined specifications was demonstrated.
3.6 WP6 Standardisation and technology evaluation
3.6.1 Main objectives
WP6 followed two objectives: standardization issues were disseminated and a technology evaluation was carried out.
3.6.2 Economic and ecological evaluation
In WP6 an evaluation of the ecoefficiency of the case studies (WP5) was carried out.
In order to evaluate the technology developed, NanoOnSpect collected data during different tasks of the project.
1. NanoOnSpect carried out several case studies, including two showing the industrial applicability of all the foreground developed.
These case studies were reported in detail in the deliverables D5.1 D5.2 D5.3 and D5.4 all of which were disseminated as public documents
These case studies showed the feasibility specifically for
a) Implementation of the onBOX device into an existing extrusion line
b) Case study 1: Quality control and quality prediction for electrically- conductive CNT based PNC, small throughput
c) Case study 2: Quality control and process control via Expert System for nanoclay based PNC, high throughput
d) Case Study 3: Testing and evaluation of the technology used, carried out for electrically conductive material, including an extensive study of eco-efficiency of the material and process
2. In order to ensure fast transfer to the market two industrial workshops were also carried out showing the developed technology and discussing standardisation topics.
3. Subsequent to the industrial hands-on workshop on technology development an Industrial Reference Group was established. Four companies joined this group and signed an NDA with the consortium, each introducing their own small application case studies for a feasibility study.
4. An impact analysis was carried out summarising future opportunities for new market implementations.
The economic validation highlighted:
• Reduction of the necessary particle content in the composite and its economic impact
• Analysis of the off-spec material production volume compared to standard processes
• The effect of the shorter material formulation changeover time
• Analysis of the investment costs of the NanoOnSpect process
• Reduced waste production
The ecological validation covered:
• Safety and resulting benefits due to reduced particle loading in the composite and better dispersion
• Evaluation of the benefits of specially designed PNC compounding equipment, especially the new powder feeding
• Reductions in off-spec material production
Eco-efficiency of the electrically conductive material and its process
• Comparison of CNT and carbon-black-filled polypropylene for electrically conductive materials
An economic evaluation always involves a comparison between investment costs and financial benefit due to reductions in material, work effort, safety costs and energy costs.
NanoOnSpect focusses on nanocomposite processing. The materials and their processing are very specific for each application. The controlling sensors developed in the project are built up in a modular manner enabling an individual set-up for different materials. The Intelligent Module hierarchy is easy to adapt to the material as well. Therefore each application (material as well as the later product) has to be evaluated separately concerning its benefits in the later product.
NanoOnSpect firstly developed a new quality controlling and monitoring technology consisting of sensor hardware and a controlling Intelligent Module. Secondly a new compounding module was developed, enabling high elongation stress to the material.
This process will give economic benefits in
- Saving material:
o Lower time for material changes (stabilization time for a new product, cleaning of the processing line after production)
o Quick on-line assessment of material properties in the material development instead of subsequent analysis
o Optimal amount of filler needed for the later product property due to better dispersion and on-line adjustment of the property
- Saving time
o Quicker material change
o Higher screening of material during DOE for a new formulation
o No remanufacturing of off-spec material
- Saving energy
o Lower processing times (lower stabilization times)
o No remanufacturing
o Assessment of the process at low specific energy input to ensure the correct dispersion
o Assessment of the minimal amount of filler needed for the later property (energy needed for raw material production)
NanoOnSpect has an impact on ecological issues as well, as the developments aim to use a lower content of nanoparticles and the required technology intends to lower any emissions of nanoparticles. Specifically, the following technologies developed in the project will lower the impact to human health:
• New process development of the Nexxus technology using nanoparticles in a aqueous suspension: the nanoparticles are kept in a matrix in the whole value chain
• Guidelines for dosing of nanoparticle powder in a closed system
Due to the use of optimized material formulations, NanoOnSpect technology can reduce the environmental impact of the nanocompounds
• Reduced content of nanoparticles used, as the formulation can be adjusted on-line and therefore lower off-spec material production
• Optimised amount of energy needed for production of the nanocomposite shown in 2.1.
• Lower amount of filler produced with high energy demand needed
• Lower waste material and energy demand due to monitoring and control on-line as well as lower stabilization and cleaning time for the processing
Eco-efficiency of the electrically conductive material and its process
The NanoOnSpect approach is evaluated for the case study of electrically-conductive polymeric material produced using CNTs and carbon black as functionalized fillers. State-of-the-art material and the material developed in the project are compared in terms of the development effort required to achieve the same electrical properties. The benchmark material used is a state-of-the-art material from Orion Engineered Carbons, which is a member of the project’s Industrial Reference Group. The study was carried out in the framework of a Master thesis and is published by the Karlsruhe Institute of Technology KIT (Claudia Seidel, „Öko-Effizienz-Bewertung der Produktion von elektrisch leitfähigen thermoplastischen Compounds am Beispiel von CNT und Ruß als Füllstoffe“, Masterthesis, Institut für Fahrzeugsystemtechnik, Karlsruher Institut für Technologie, Nr: 14-L-0057, December 2014).
The technology developments in NanoOnSpect were presented to a working group of TecPart in order to evaluate influences to standardization issues. The workshop took place at Gneuss in Bad Oeynhausen on March 18th 2014.
An application-compatible expert workshop was prepared by TecPart, based on the IRG workshop. In addition to the consortium, IRG and members of standardisation bodies, companies were contacted, which are active in the industry and can take a broad view of the subject. About 50 invitations were sent, and 10 industrial representatives (20% of those invited) attended, indicating the high relevance of the project. The aim was to introduce the project and to develop criteria for the standardization.
During the project the opportunities to standardise new developments were outlined and advice concerning already existing standards which have to be taken into account in the exploitation was provided.
• Evaluation of whether the valid standards can be applied to the NanoOnSpect developments
• Evaluation of whether NanoOnSpect technologies can be standardized
• Evaluation of opportunities for standardization of processes and related parameters for specific applications
During the project an evaluation on the standardization of on-line characterisation of (nanoparticle filled) polymer melts was carried out. In a workshop held in Würzburg on 25 + 26/09/2014 the TecPart expert group, made up of quality managers, discussed the relevant and currently accepted values on material data sheets of polymers and listed the relevant standards.
4 Description of potential impact, dissemination activities and exploitation of results
NanoOnSpect aimed to develop novel soft sensors for measuring process property parameters that are currently not available in in-line processing in the polymer nanocomposite (PNC) industry. Thus, the project consortium developed an intelligent process control loop that integrates the knowledge of the expert process operators in the close control loop so as to carry out the adjustment of the extrusion lines automatically.
The Artificial Neural Network developed in NanoOnSpect can be included in the category of soft sensors (software or non-physical sensors) for advanced process control (APC) in process industries, but up to now there is still a lack of industrial control applications in the compounding industry that allow the measurement and correlation of interesting process property parameters. The process property values that the Intelligent Module has addressed have non-existent methods for the value in-line estimation (correlation).
There is therefore a lot of potential for the prediction of the desired property values in the compound industry as well as in any other industrial process. The sensors the consortium developed can easily be adapted to a different scenario with a similar pattern of processing. This means that as long as there is enough input to create a knowledge data set that can be correlated to some desired parameters the Neural Networks could easily be trained and adapted to tackle the problematic related to parameter estimation.
The NanoOnSpect consortium developed two different modules included in one Intelligent Module in order to carry out the advance process control of the PNC.
The ANN Module aims to correlate off-line desired values (as explained above) and the Expert System is designed to automate the process control. The Intelligent Module integrates an Expert System that codifies all the logic used by experienced operators in the process control so as to adjust the process machine parameters in an automatic way. The Expert System acts as the brain of the intelligent process control: it receives all the values coming from the sensors, soft sensors and the process parameters, and carries out the necessary machine parameter adjustment to obtain the desired property outcomes in the compound process.
The Expert System carries out the inference in the process using the logic provided in the knowledge data base and decides the values of the control variables of the extrusion process lines in order to achieve the desired end properties in the process. This control loop continues until the Expert System decides that the difference between the on-line state variables and the desired properties is below a defined threshold.
The ANN and ES complement each other to provide a holistic automatic control loop, which gathers the information from the process, determines desired process parameters and carries out the suitable reasoning to adjust the machine parameters. However, these two modules are independent and can be applied separately to different types of industrial process.
4.1 Market of nanomaterials
4.1.1 EU-definition of nanomaterials
The 2011 Commission Recommendation on the definition of nanomaterials defines ‘nanomaterial’ as “a natural, incidental or manufactured material containing particles, in an unbound state or as an aggregate or as an agglomerate and where, for 50 % or more of the particles in the number size distribution, one or more external dimensions is in the size range 1 nm-100 nm. In specific cases and where warranted by concerns for the environment, health, safety or competitiveness the number size distribution threshold of 50 % may be replaced by a threshold between 1 and 50 %. […]”
The definition is intended to be used by Member States, European Union agencies and companies. The Commission will use it in EU legislation and instruments of implementation where appropriate. Where other definitions are used in EU legislation, provisions will be adapted in order to ensure a consistent approach, although sector specific solutions may remain necessary.
4.1.2 Benefits of nanomaterials and their contribution to growth and jobs, innovation and competitiveness
The total annual quantity of nanomaterials on the market at global level is estimated at around 11 million tonnes, with a market value of roughly 20 bn EUR. Carbon black and amorphous silica represent by far the largest volume of nanomaterials currently on the market. Together with few other nanomaterials they have been on the market for decades and are used in a wide variety of applications. The group of nanomaterials currently attracting most attention are nano-titanium dioxide, nano-zinc oxide, fullerenes, carbon nanotubes and nanosilver. Those materials are marketed in significantly smaller quantities than traditional nanomaterials, but the use of these materials is increasing fast. Many of the new nanomaterials under development are used in innovative applications such as catalysts, electronics, solar panels, batteries and biometrical applications including diagnostics and tumour therapies.
There are also many newly founded SMEs and spin-off companies in this high technology area. Currently direct employment in nanotechnology is estimated at 300.000 up to 400.000 jobs in the EU.
Nanotechnology has been identified as a Key Enabling Technology (KET), providing the basis for further innovation and new products. In its Communication ‘A European Strategy for Key Enabling Technology- A Bridge to Growth and Jobs’ – the Commission has outlined a single strategy for KETs including nanotechnology built upon three pillars: technology research, demonstration and competitive manufacturing activities.
The applicable legislation must ensure a high level of health, safety and environmental protection. At the same time it should permit access to innovative products and promote innovation and competitiveness. The regulatory environment affects time to market, marginal cost structure and allocation of resources, especially for SMEs.
4.1.3 Safety aspects
Since 2004, the Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) and the Scientific Committee on Consumer Safety, the European Food Safety Authority (EFSA) and the European Medicines Agency (EMA) have been working on the risk assessment of nanomaterials. The identified hazards indicate potential toxic effects of nanomaterials for man and the environment. However, it should be noted that not all nanomaterials induce toxic effects. Some manufactured nanomaterials have already been in use for a long time (e.g. carbon black, Ti02) showing low toxicity. The hypothesis that smaller means more reactive cannot, therefore, be substantiated by the published data. In this respect nanomaterials are similar to normal chemicals / substances in that some may be toxic and others may not. As there is not a generally applicable paradigm for nanomaterial hazard identification, a case-by-case approach for the risk assessment of nanomaterials is still warranted. EFSA confirmed in its 2011 scientific opinion that the risk assessment paradigm used for the evaluation of standard food products is also appropriate for nanomaterial applications in the food and feed chain and that a ‘case-by-case’ approach is needed.
Harmonization and standardization of measurement and test methods to support the risk assessment of nanomaterials is being promoted through the OECD and by a Commission Mandate to the European Standards Organisation.
4.2 Business plan - Turn-key solutions for compounders and masterbatchers
4.2.1 Introduction and background to advanced process control for compounders
The developed services should be exploited both in the production of polymer nanocomposites and in applications which are not directly linked to nanomaterials but where advanced process control will allow compounders and masterbatchers to economise through (1) lower material quantities required, (2) better control over properties of compounds and (3) reduced energy consumption.
The logics of advanced process control: ‘Adaptive control’(AP) or ‘advanced process control’ (APC) uses (1) input of data obtained from sensors and (2) special neural network algorithms which allow the incorporation of response-behaviour of the system which is subject to control. The difference with classic PID controllers is that APC allows anticipation of how the installation will respond when a specific path is followed. This enables the variance to be reduced, and thus allows more precise control of the system.
Advanced process control technology has been transferred towards many different fields of application, including (non-exhaustively) micro-electronics, high-speed printing, chemical process industry, glass industry, ceramics industry, food processing industry.
This project has been successful in developing sensors for measuring a wide range of parameters right at the output of the compounding line, and thus will allow APC to be transferred to compounding and masterbatching.
Existing APC providers allow clients, for instance the chemical process industry, to implement sensors on existing lines in order to yield typically (1) energy reduction, (2) higher throughput, (3) less variability in quality. One important parameter under consideration when APC is implemented as a project is ‘PAY BACK TIME’. How long does it take to pay back the system when being implemented?
Vision, goals, objective, leverage points: This business plan highlights the potential impact of spinning off NanoOnSpect results to a new venture, able to mobilise the necessary resources to transfer advanced process control with the sensors developed for in-line measurements to existing manufacturing lines of compounders and masterbatchers, thereby enabling them to optimize existing processes (1) and enable new processes, (2) enable in-line measurements to serve quality control and (3) to implement APC in identified niche-applications such as PNC production, the latter with specific knowledge on polymer nanocomposite compounding.
Turn-key solutions for compounders and masterbatchers
Benefits from APC to compounding and masterbatching:
• Optimisation PID loop controls;
• Distributed control system reconfiguration work;
• Instrumentation recommendations;
• Increased production output;
• Reductions in energy consumption;
• Implementation of sensors to control (1) melt viscosity, (2) electrical conductivity, (3) thermal conductivity, (4) density, (5) pressure, (6) temperature, (7) state of dispersion
APC for compounders and masterbatchers: The project NanoOnSpect succeeded in bringing innovation partners together which have developed (1) sensors and (2) an integrated adaptive control unit and system implementable on compounding lines (onBOX).
4.2.2 Porter analysis
Threat of new entry: This sector, which is under development, has niche market characteristics. Due to the highly-specialized skills needed (combination of (a) knowledge on compounding technology, (b) neural network heuristics and programming and (c) access to unique sensor technology) it is difficult for existing competitors to acquire the same set of skills.
The project consortium has a clear focus on opportunities occurring for high-tech applications such as micro-manufacturing where applications can also be customized to specific needs (micro-injection moulding requiring high level of precision and control).
There are, however, competitors on the market. It is a trend that APC is transferred from micro-electronics (printing) to other domains. The chemical industry has been adopting APC gradually, so APC providers are continuously looking for new markets and applications. Those typically include continuous processes such as cement kilns, glass and ceramics and metallurgy, but also batch processes such as beer brewery.
Suppliers: Within the context of NanoOnSpect collaboration specific skills and elements are in place to produce hardware including sensors and software. Further integration of access to these supplies, key competitive advantages may result from direct access to the sensors.
It is important to notice that other big APC providers are often also integrating exclusive hardware supplies into their business model.
Clients: More than 300 masterbatchers and compounders are active in Europe. Their common characteristic is their nearness to customers and their ability to position in specific niche applications.
Government: The future of polymer nanocomposite applications will be influenced by future legislative changes which will regulate the use and implementation of nanomaterials in Europe. It is therefore important to diversify action purely from PNC, also because the volume of PNCs on the market is currently low, and thus to explore other options to implement APC on compounding and masterbatching. However, the technology is also an enabler that has strong relevance for accelerating market uptake of PNCs in specific applications, thereby increasing process reliability and safety of the compounds, due a to higher level of control of the material properties.
Key characteristics of product / service: The potential service aims at a turn-key solution for prospective clients by customising the service to their existing equipment.
This must be implemented by a team qualifying to (1) make a study on existing compounding lines, to determine the benefit APC can bring, (2) to implement sensors and onBOX to existing infrastructure, (3) to implement, using both optimized software (4) to run initial tests with minimal off-time, (5) to train employees on how to use the instrumentation.
The product / service: This business plan envisages the commercialization of turn-key APC solutions for compounders and masterbatchers. Turn-key solutions imply that there is a team able to make a free study (no improvement, no pay), install an onBOX, sensors and an Intelligent Module, run tests which are limited in time (limitation of down-time for the customer) and train the employees on how to use the system.
4.3 Technological results
Figure 4.3.1 (in the attachment) summarizes NanoOnSpect’s exploitable results.
4.3.1 Short description of the onBOX
The onBOX developed in the project was equipped with the following sensors (Figure 4.3.2):
• thermal conductivity
• electrical resistivity
• capillary viscosity
• Raman spectroscopy
• microwave spectroscopy
A central control system collected all the sensor signals.
An Artificial Neural Network was used to calculate the electrical conductivity. The development and process integration of several sensors for the on-line measurement of polymer nanocomposite (PNC) properties was carried out.
In the area of nanocompound characterisation the onBOX has been developed in a pilot version and the first sensors have been adapted to the system. The tool is now able to measure viscosity, thermal and electrical conductivity, pressure, and melt temperature at different locations in the melt flow.
Microwave, ultrasound and optical spectroscopy sensors have also been developed and are now being adapted to the onBOX
4.3.2 Short description of the Nexxus Channel
Nexxus compounding device processes (Figure 4.3.2) any type of thermoplastic polymer. The consortium processed different resins compatible with polypropylene (PP) or any other resin, none of which were affected by water during the process.
The pumping group delivers a special waterborne dispersion including carbon nanotubes (CNTs), surfactant and a special polymer-based lubricant to impart flowability to the compound. Nexxus Degassing allows for water removal. Up to 50% weight and more water can be effectively removed, in order to obtain a substantially water-free compound at the discharge port. The compound exiting from Nexxus Degassing enters in a co-rotating twin-screw extruder (cTSE) to undergo the final dispersion and venting.
It has been found that the best temperature processing conditions approach to the boiling point of water at atmospheric pressure.
In general it has been proven that the water-based route is very successful in granting a top quality CNT/polymer dispersion. The exceptional results achieved allow use of the concept in many other nanoparticles applications beyond CNTs, such as graphene oxides, nanoclay, etc.
In general a similar water-based approach is felt to be useful for treating not only nano but also macro systems, thus including all silica additives with micro scale dimensions, for example pigments, melamine cyanurate, CB, etc.
4.3.3 Short description of Intelligent Module
The Intelligent Module gathers all the values coming from all the sensors, machine parameters, actuators, and self-correlated values of desired off-line properties in order to carry out the process control loop and on-line parameter adjustment of the extrusion line.
The Intelligent Module performs the loop control by first reading all the input values from the extrusion line through the onBOX. Then, all input values are passed to the trained Artificial Neural Network (ANN), which performs the on-line correlations of the desired property values (resistivity and conductivity). Finally, all input values from the extrusion line and the values coming out from the ANN are used by the Expert System in order to perform the automatic adjustment of the compounding process.
The communication between the Expert System and the extrusion line is carried out using OPC Server-Client Protocol through an Ethernet network. The Expert System first communicates with the extrusion line HMI and the HMI pass the data to the extrusion line. This communication was developed using the OPC protocol. This process can be used for closed loop quality control of the compounding process. Variable parameters can be implemented into the ANN system if their off-line – on-line relationships are clear. Different rules defining the quality range can be integrated into the Expert System.
The obtained signal can be used for identification of necessary operator interventions or moreover for direct control of production parameters by a control system.
List of Websites:
Grant agreement ID: 263406
1 April 2011
31 March 2015
€ 4 703 754,10
€ 3 399 554
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
This project is featured in...
Deliverables not available
Publications not available
Grant agreement ID: 263406
1 April 2011
31 March 2015
€ 4 703 754,10
€ 3 399 554
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
This project is featured in...
Grant agreement ID: 263406
1 April 2011
31 March 2015
€ 4 703 754,10
€ 3 399 554
FRAUNHOFER GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.