Online Tg estimation for blade manufacturing

Cost reduction is one of the main drivers in the production of wind turbine blades in the last few years. In this direction, the use of intelligent process monitoring to optimize the infusion, curing and bonding by measuring online the actual process can reduce energy and increase productivity by 20% while ensuring product quality. In this framework, durable non-intrusive and disposable sensors have been developed for the monitoring of resin arrival, the resin’s viscosity during infusion and the on-going Tg during curing by measuring the resin’s electrical resistivity and temperature.


Introduction
Nowadays, the need for efficient manufacturing in the wind blade industry is more evident than ever.State of the art in production involves only temperature measurements while DSC measurements are performed at the best, a few days after demoulding.Therefore, wind-blade production needs to improve its productivity, reduce production costs including repairs, ensuring product quality.The introduction of online intelligent process monitoring systems optimised for the wind blade production can help significantly in all of these aspects.Firstly, by monitoring the resin flow and the increase of viscosity during infusion, potential flow control actions can be activated, if necessary.Then during curing, the online estimation of the Glass Transition temperature can lead to significant decrease of the cycle times by ensuring the minimum cure quality.Towards that direction, Pantelelis and Bistekos [1] presented already in 2010, the DC-based dielectric cure monitoring technology with a clear focus on industrial applications.The Optimold system can measure directly the resin's electrical resistance and temperature using specialised sensors.Measuring electrical resistance from 10 5 Ohm up to 10 14 Ohm is capable to follow the polymerisation from very low viscosity at high temperatures to fully cured resins at room temperature.Comparison between the present DC and the conventional AC cure monitoring technologies has demonstrated the superiority of the DC technology.Furthermore, it is cheaper, very easy to use and very reliable and robust which has been proven in various industrial installations in wind energy, automotive and aerospace production lines [2, 3, 4].

Intelligent process monitoring: System Components
For more than ten years, Synthesites has been developing the components for a complete suite for monitoring the complete manufacturing process in wind turbine blades: from a dedicated system to provide online resin arrival and temperature, to online viscosity and Tg after infusion and during bonding including the corresponding intelligent software to calculate and communicate with controllers and PLC.The first attempt in using these cure sensors in wind blade production took place already in 2011 [5] but it was only in 2017 that this technology attracted serious interest from blade manufacturers including leading OEMs.Nowadays, these systems are being used in everyday wind blade production.The main components of this intelligent system are summarised below:

The data acquisition units
Both the Optimold cure monitoring (Figure 1a) and the Optiflow Resin arrival monitoring unit (Figure 1b) measure directly the electrical resistance of the resin by applying a DC voltage to the electrodes of the sensors.The resistance together with the resin temperature are necessary for the resin arrival as well as the online estimation of the resin's viscosity and glass transition temperature (Tg) which are the most important properties in composites processing.

The durable cavity sensors
A range of durable cure and resin arrival sensors have been developed to be used in direct contact with resin in the mould cavity or through the vacuum bag.Various sensor housings help to secure the sensors at the right location depending on the monitoring needs.In Figure 2, two of the available durable sensors are shown, οn the left the standard in-mould sensor which should be inserted flush-mounted in the tool.The Vacuum Bag (VB) sensor, Figure 2b, can be placed easily on the vacuum bag without mould modifications on top of the laminate, where the lagging curing area is found in a typical wind blade infusion set-up.Although the cure sensor provides the resin arrival instant, a dedicated resin arrival sensor in combination with the Optiflow unit can provide the resin arrival information at lower cost allowing the use of numerous resin arrival sensors which could be particularly useful in the large moulds for wind turbine blade production.

The durable inline sensor
The inline resin arrival sensor (Figure 2c) is connected to the Optiflow unit (resin arrival and temperature) and can provide a timestamp when the resin flows through the sensor, entering or exiting the mould cavity.The inline sensor can be easily integrated at the feeding or the vacuum lines without any modification in the mould.Following the resin arrival signal, Optiflow is capable for example, to trigger a valve for actively controlling the resin flow [4]. 3

The Optiview data-acquisition and visualisation software
The Optiview data acquisition software is capable for capturing, storing and visualising all the measurements from all the connected units and sensors.Optiview provides a real-time visualisation of the on-going process not only for the raw measurements but also for their by-products such as resin arrival, viscosity, degree of cure and Glass Transition temperature.For example, in Figure 3, the resin arrival timeline for sixteen resin arrival sensors of a typical RTM application is shown where the green bar indicates that a sensor is connected but dry which becomes orange when resin arrives at the sensor.The Optiview software also provides a Human-Machine-Interface (HMI) for the automated operation at production environment as can be seen in Figure 4, allowing to eliminate any interaction with the operator, while it handles also the communication with PLC and temperature controllers.

The Cure Simulator
The first version of the Cure Simulator was developed within an EU funded project [7] to provide a nonintrusive solution for monitoring the curing in autoclaves and other applications where it is difficult to introduce a cure sensor.In the wind-turbine blade production, the Cure Simulator has been employed for monitoring the curing at areas that they are either not easily accessible, such as the bondlines or other sensitive areas to sensor's intrusiveness such as the spar caps and the shear webs.The Cure Simulator comprises a small heated resin cell equipped with a cure sensor.The principle, depicted in Figure 5, is that the resin sample placed in the resin cell of the cure simulator, will experience the same curing as the composite at the location of the input thermocouple.The temperature of the resin cell follows accurately and instantly the input of that temperature sensor using an advanced temperature controller.
The capability of the Cure Simulator to reproduce the same curing as the laminate has been verified in several applications in aerospace and wind energy applications.A very important detail in this procedure lies on the right location of the input temperature sensor which should provide as much as possible the real temperature of the resin under curing.Besides the cure sensor elimination, the Cure Simulator can help to eliminate the operator's interaction with certain cure sensors, allowing its use in everyday production.

Figure 5:
The Cure Simulator concept for RTM: The temperature measurement becomes continuously the setpoint for the temperature controller of the heated resin cell of the Cure Simulator.

The Online Resin State (ORS) software
Last but not least, the Online Resin State software translates in real-time the Optimold measurements of resistance and temperature to useful resin properties such as the resin's viscosity, glass transition temperature (Tg) and degree of cure.The ORS software can be combined with the Optiview HMI software to automate effectively the curing process.The ORS software is using a calibration function for a specific resin.The calibration campaign consists of manufacturing a series of coupons made of the resin under investigation cured at specific conditions.The generated data of the resistance and temperature are correlated to the evolution of viscosity and glass transition temperature measured experimentally to produce the calibration function, which then can be used to obtain the same properties in real-time.Similar calibration procedures have been developed to check the resin mixing ratio and the storage aging, onsite.In the last 7 years more than 70 resins, all the kinds of epoxies as well as polyesters, vinylesters, reactive thermoplastics, phenolics, PFA, benzoxaxines, cyanate esters and others have been studied successfully.

Results
Since 2011, several industrial projects in wind blade manufacturing have been realized, verifying the accuracy of the Online Tg estimation at industrial conditions.In this framework, numerous epoxy resins and adhesives from several resin manufacturers have been studied, firstly in the lab and then at industrial conditions.In Figure 6, the resistance curves during several isothermal cure cycles of a typical epoxy adhesive used in turbine blades are shown.As expected, when increasing the curing temperature, the reaction speed increases resulting in the resistance curve reaching faster its final value.On the other hand, as temperature increases, the resistance decreases.In order to test the new durable sensor and the Online Tg module at realistic conditions, numerous laboratory scale trials were performed [3] in simulated production conditions i.e. representative cure cycles and laminates from a variety of wind-blade components have been tested at the laboratory.In Figure 8, a typical vacuum infusion set-up of a thick laminate together with cure and temperature sensors can be seen.Following the successful lab-scale trials, the system was tested in the production with very good results.
In Figure 9Figure 10, the first such trial in 2017 is shown.A durable sensor was placed during the infusion of a half-blade on top of the vacuum bag at the root section.In Figure 10, two representative lab-scale trials are shown: one 'almost' isothermal at 80 o C and one highly non-isothermal from curing a rather thick laminate by heating only from the bottom of the laminate.In both cases the online Tg estimation is also depicted calculated in real-time during curing based on the real-time values of electrical resistance and temperature measured by Optimold.Especially in the non-isothermal case (Figure 10a) it is obvious that without the temperature, the electrical resistance alone cannot provide the necessary information about the curing.So with the appropriate resin calibration, the Online Tg can provide an accurate and robust estimation of the on-going Tg/degree of cure.In Table 1, a comparison between the real-time estimated Tg and the Tg measured by DSC after demoulding are shown for various representative isothermal and non-isothermal cases.The isothermal trials include several trials with resin cured at 60, 70, 80 and 90 o C such as the trial shown in Figure 10a with different duration.The trials at real blade components (non-isothermal cases) include shear web and half-shell trials such as the one depicted in Figure 10b.As can be seen in the last column, the difference between these two values for the isothermal and the non-isothermal cases are within the DSC accuracy.
In order to facilitate the use of this system in wind blade manufacturing, Synthesites developed a new durable sensor specifically for vacuum bag applications.As can be seen in Figure 11a, the durable sensor includes a wide disc to allow its attachment on the vacuum bag, secure and easy.In order to eliminate the effect of the sensor's mass on the curing of the laminate underneath, a heated version of the Vacuum-Bag sensor was also developed as shown in Figure 11b.In both cases, the sensors were used consistently during production of blade components for acquiring the online Tg [6].In the bonding of a blade, for the monitoring of the Tg, the Cure Simulator has been employed as can be seen in Figure 12a while a disposable Pt temperature sensor was inserted at a specific depth in the bond-line (Figure 12b).The Cure Simulator can provide a unique way to simulate and measure the curing in the middle of a bond-line which on one hand is the latest area to cure since it is quite difficult to heat from the mould or the hot air and on the other hand it cannot be accessed easily to acquire a DSC sample before it is demoulded i.e. much later than the end of heating.A typical bonding cycle depicted in Figure 13, shows that although there is a continuous heating of the adhesive by hot air to 95 o C that temperature was hardly reached at the end of the cycle.However based on the Tg that was calculated online, after 180 min the maximum Tg was reached and cooling could have started saving significant time and energy.The accuracy of the online estimated Tg has been confirmed by DSC [6] as it is also shown in Figure 13.

Conclusions-Next steps
Windblade production is one of the most mature composites production even though automation has not been advanced considerably due to the size of the parts.At the present work, the complete suite for the real-time Tg estimation and cure control have been analysed.The new technology was proved robust and accurate enough even at the toughest processing conditions in the manufacturing of several blade components.The use of the intelligent system in everyday production is under way by a large OEM and is expected that besides the improved quality assurance and traceability, one hour bonding time can be saved.The next step in the intelligent platform will emerge through the TURBO R&D project which is led by DTU and involves SGRE in Denmark and it is partially funded by the European Commission.In this project, a more complete intelligent platform targets in fully automating windblade production and further advancing productivity by eliminating the needs for repairs in post-production.Lauter C, Jaquemotte K-P, Pantelelis N, Improvement of productivity and quality in the wind energy industry through the use of an advanced sensor system, Sampe Journal 53/6 6-10

Figure 1 :
Figure 1: (a) The Optimold cure monitoring unit and (b) the Optiflow resin arrival unit.

Figure 2 :
Figure 2: (a) The durable sensors: the In-mould sensor, (b) the Vacuum-Bag (VB) sensor and (c) the inline resin arrival sensor for the feeding and the vacuum lines.

Figure 7 :
Figure 7: Correlation between Resistance and Tg from TDS of a popular epoxy infusion resin during isothermal cure at 70 and 80 o C. Similar behaviour for the resistance can be observed in the curing of a popular epoxy infusion resin (Hexion RIM 035-037) as shown in Figure 7 for 70 and 80 o C together with the corresponding Tg evolution.

Figure 8 :
Figure 8: A typical vacuum infusion set-up of a laminate together with cure and temperature sensors on top of the vacuum bag.

Figure 9 :
Figure 9: A Vacuum-bag sensor was used for the first time to monitor the curing on top of the vacuum bag in real production at Carbon Rotec [3].

Figure 10 :
Figure 10: Typical ORS graphs showing the resistance together with temperature and Tg (a) Isothermal case, (b) Strongly non-isothermal real case.

Table 1 :
Overview of the various test cases and the difference between the Tg estimated online and measured right after demoulding by DSC [3].

Figure 11 :
Figure 11: (a) The new durable vacuum bag sensor; (b) Installed at production

Figure 12 :
Figure 12: (a) An Opticure system with the first version of the Cure Simulator at production, (b) A special temperature sensor placed in the bond-line in blade bonding [6].

Figure 13 :
Figure 13: Typical cure cycle for the adhesive at the centre of the blade's bond-line, the resistance (right vertical axis) together with temperature and Tg (left vertical axis) and the Tg evolution that was measured by DSC (dashed line).
Bistekos E 2010 Process monitoring and control for the production of CFRP components SAMPE USA [2] Meier R, Pantelelis N, Hauber M, Wolf C and Drechsler K 2015 Process monitoring and control for an aerospace application International Conference in Advanced Manufacturing of Composites [3]