You are on page 1of 22

Optimization of Heat Transfer in Carbon Fiber Epoxy Composite Tooling By: Tariq K.

Maksoud

Abstract The National Research Council (NRC), Institute of Aerospace Research has developed a new out-of-autoclave, conductive based process for manufacturing preimpregnated composite material identified as the Integrally Heated Conductive Tooling. The study contained herein focuses on understanding and analyzing key parameters that may have an effect on heat transfer from the tool to the composite part in order to optimize the performance of the Integrally Heated Conductive Tooling. Experiments were performed on sample models and were later validated through CFD simulation models.

The main objective is to develop an accurate modeling capability using Computer Animated Drafting (CAD) to design the IHCT. In order to do so, certain parameters must be known so that they are used in developing the CAD model of the IHCT. The focus of this project has been on determining these parameters, quantifying them, and determining the range of accuracy that should be sought. Parameters to Investigate: The convection coefficient of air in the vicinity of the IHCT model, and heat transfer behavior as the IHCT model is oriented differently. Effect of bonding tape thickness between heating element and the copper plate on heat transfer from the heating element to the copper plate Effect of thermal conductivity of heating element on heat transfer between the heating element and the copper plate. Heat transfer behavior in the composite tool as the silicone bonding layer between the copper plate and the composite tool is varied. Copper Plate Heating Assembly In the copper plate heating assembly, the heating element was fixed on top of the center of the copper plate using a combined layer of Kapton and 3M double sided tape. The 4 thermocouples were placed at distinct locations (defined in figure 3-5) to record temperature values every second as the copper plate was being heated. The heating assembly was constructed to study the effect of different parameters on heat transfer from the heating element to the copper plate and atmosphere. The parameters that were investigated are summarized in table 3-1. Figure 3-1 depicts an illustration of the CFD model of the copper plate heating assembly.

Experimental Phase . The heating assembly was held by a clamp which was mounted onto a tripod. The thermocouples were connected to the 4 locations on the heating assembly from one side and to a computer on the other side, where LabVIEW was used to record temperature readings. The assembly was setup with four thermocouples were located at pre-determined coordinates to collect temperature readings. Figure 3-3 depicts the locations of each of the thermocouples. As stated previously, depending on the orientation of the plate and heating element, different convection coefficients may arise. The orientations were divided into two categories, 2 vertical orientations, and 2 horizontal orientations. The 4 orientations are illustrated in figure 3-2.

The temperatures of the air and initial temperature of the copper plate were measured to be approximately 296K. The heating power supplied to the heating element was 25W. During each cycle the assembly was heated for 300 seconds, and was subsequently cooled back to room temperature by natural convection. For each orientation, 3 experimental runs were performed and the average of the temperature values of the 3 experiments was used in the results. During

experimentation, the edge of the copper plate was held by a clamp that was attached to a tripod.

Variability Between Trials For each orientation, 3 trials were conducted. The average of the 3 trials (a, b, and c) was calculated and used to plot the following charts. This section was dedicated to demonstrating the variability or consistency between trials. Figures 3-4 consist of a plot with 3 curves. Each of the curves represents a trial.

For all the trials for different thermocouple measurements, the margin was estimated at worst case to be approximately 4 degrees. An error of 4 degrees was considered to be satisfactory. Experimental Results Trends and Observations in the Horizontal Orientations A repetitive trend in the heating and cooling cycles of each of the two horizontal orientations was witnessed. Starting with the cooling cycles, a similarity between the curves of orientation #1 and orientation #2 was clearly apparent. In inspecting figures 3- 10 and 3-13 for example, it was discovered that each of the thermocouples was initially at a different temperature. Regardless of their locations, all thermocouples cooled to the same temperature after an average of 90 seconds. Since all thermocouples were at different temperatures, it is known from heat transfer theory that uniform temperature throughout the assembly is achieved through conduction (equilibrium state), and therefore the heat was distributed by conduction through the entire assembly until all thermocouples reached equal temperatures. Conduction was seen to occur faster than natural convection of heat out of the plate. Conductive heat transfer occurs between two solid mediums where the contact between particles is superior when compared to the contact between a solid medium and a surrounding fluid such as air. The entire assembly then cooled further for an average of 337 seconds through natural convection until it reached room temperature. The time needed for thermocouples to reach equal temperatures is very small in comparison to the total cooling time. It is therefore rational to conclude that conductive heat transfer is much faster than convection.

At this point, it has already been established that in the scope of this project, heat prefers to dissipate through conduction rather than convection. Moving on to the heating cycle curves of figures 3-9 and 3-12, a considerable temperature difference was previously noted between thermocouple #1 and the rest of the thermocouples. the volumetric heat capacity of copper is noticed to be approximately 1.5 times greater than that of the heating element which is made up of silicone rubber. Assuming that 10 J of energy is required to heat 1m3 of silicone rubber, it would require 15 J of energy (1.5 times that of silicone rubber) would therefore be required to heat 1m3 of copper. This means that 1.5 times more energy is required to heat up a volume of copper than it would if the volume was made up of silicone rubber. The fact that the volume of the copper plate is already 7 times greater than the heating element further explains the reasoning behind having thermocouple #1 heat up to a higher temperature than the rest of the thermocouples. Trends and Observations in the Vertical Orientations A similar behavior to the horizontal orientation cooling curve was witnessed in the vertical orientation cooling curves. The conclusion that was attained that heat

transfer is faster through conduction in the plate than to the atmosphere by convection also holds in the case of the vertical orientations. Similar heating cycles of the horizontal orientations were also observed in the vertical orientations. The substantial temperature difference between thermocouple #1 and thermocouple #4 was also noted in the vertical orientations; hence the observation that was made in that regard for the horizontal orientations also applies to the vertical orientations. After completing experiments for all 4 orientations, it was evident that there was negligible difference between thermocouples #2 and #3 in all heating cycles. Since the copper plate was very thin, and copper is highly conductive, it was evident that the temperature difference through the thickness of the copper plate was negligible. Effect of Orientation on Thermocouple #1

After examining figure 3-20, it was evident that thermocouple #1 experienced small temperature changes as the orientation was varied. It has already been established that there is a significant temperature difference between thermocouple #1 and thermocouples #2, #3 and #4. It would therefore be misleading to decide whether a change in orientation would have an impact on the overall heat transfer by referring to the behavior of thermocouple #1 only. After examining figure 3-21, it was evident that thermocouple #1 followed the same pattern in its cooling cycle. Although orientations were varied, it showed little influence on the behavior of thermocouple #1. Summary of Variation in Heating Assembly Orientation Although a difference was noted in the heating cycles of thermocouples #2, and #3 where orientation #4 deviated from the rest of the orientations, however the effect of this difference is considered to be small. It was also evident that as the

orientation changed, there was no change in the behavior of the cooling cycles. It can therefore be concluded that the convection coefficient of air did not change significantly as the orientation of the heating assembly was altered. Simulation Phase Moving on to the simulation phase, a CFD model of the heating assembly was constructed using Gambit. In order to investigate the relative effects of these parameters, different values were used in each of the simulation runs so that the relevance and importance of each parameter could be assessed. Simulation results were later compared with experimental results. This will aid in determining the accuracy of the simulations that were constructed. A total of 11 simulation runs were performed. In those simulations, theparameters under investigation previously mentioned were changed in order to study the effects that they may have on heat transfer. Table 3-3 summarizes 8 simulations in which tape thickness and convection coefficients of air (hair) were varied. Table 3-4 summarizes the last 3 simulations in which the conductivity of the heating element was varied.

For the simulations outlined in table 3-3, the heating element conductivity was set to 0.216 W/m K and copper conductivity was set to 387.6 W/m K. The conductivity of the layer of combined Kapton & 3M double sided tape was set to 0.14 W/m K. As for the simulations outlined in table 3-4, the hair used was 15 W/m2 K, the tape thickness used was 0.07mm. The conductivity of copper used was 387.6 W/m K.

Change in Convection Coefficient

The parameter which is most difficult to predict is the convection coefficient of air. Since the convection coefficient is not a constant value and varies with changing temperature, it usually determined empirically. In reference to Al-Arabis and ElRiedys work [8], experiments were performed over a horizontal flat plate with similar dimensions to the copper plate used in this project. The convection coefficients obtained varied roughly between 6 W/m2 K and 9 W/m2K. The convection coefficients used in the simulations varied from 5 to 25 W/m2 K. In the following figures, each thermocouple was graphed separately. Since it has been previously concluded that the orientation had no effect on the heating cycles of the heating assembly in the experimental phase, Figures 3- 29 3-32 contain one curve from the experimental phase representing the average heating cycle for each thermocouple. This curve will be compared to the curves obtained from the simulation phase where the convection coefficients are changed. This will help indicate an approximate convection coefficient of air during the experimental phase

After examining figure 3-29, a significant difference was evident between the experimental curve and the 5 simulation curves. As discussed previously, the heating element is connected to the copper plate from underneath hence heat travels mainly through the heating element and to the copper plate rather than dissipating through the air by convection. It was concluded that it is quite possible that the convection coefficient of air around the heating element is different from the convection coefficient of air around the remainder of the plate since the heating element is the only component that is attached to another solid medium where conduction would be much more dominant than convection. It was therefore concluded that the convection coefficient around the heating element is lower than those used in the simulations. Summary of Convection Coefficient Determination

The purpose of running simulations with different convection coefficients was to try and match a possible convection coefficient of air surrounding the heating assembly. According to the findings of this section, the convection coefficient around the heating element seems to be very small whereas the convection coefficient around the copper plate seemed to match a value of 10 W/m2 K. Change in Tape Thicknesses (Kapton & 3M combined) In order to evaluate the effect of change in tape thickness, 3 simulations were run. Each of the different tapes was measured using a vernier caliper. The tapes were placed on top of each other. The Kapton tape is a single sided tape which was attached to the heating element. The 3M double-sided tape was used to attach the heating element to the copper plate. The two tape layers were simulated as one layer. An assumption of perfect contact between the two pieces of tape was made since the thickness of both tapes together is very minute; therefore contact resistance was neglected. Thicknesses used in the charts are in millimeters. Thermocouple #1: Effect of Tape Thickness A significant difference was witnessed at thermocouple #1 as the tape thickness was changed. This result however may be misleading. As discussed previously, there is a significant temperature difference between the top of the heater and the bottom where it is in contact with the copper plate. In no way does thermocouple #1 reflect the overall impact the tape thickness has on heat transfer. An increase in tape thickness may have impeded heat from being transferred into the copper plate due to an increase in thermal resistance; however it is hard to determine the effect that the tape thickness has on the overall heat transfer by relying on the results of thermocouple #1 simply because thermocouple #1 is not an indication of the heat flux flowing through to the copper. The heat flux flowing to the copper plate is not known yet. Thermocouple #1 can only give an indication of the amount of heat present within the heating element.

Summary of Tape Thickness Effect Although it seemed apparent at first glance of figure 3-36 that the tape thickness had a significant effect on heat transfer, after looking through plots of temperature against time for the different thermocouples, it was obvious that the change in thickness had negligible effects on the heat being transferred to the copper plate. As the tape thickness was increased, less heat was able to pass through to the copper plate. The effect of the difference in the heat flux as the tape thickness was varied is more evident in the heating element than in the copper plate. This is due to the fact that volumetric heat capacity of copper is 1.5 times greater heating element. The effect of a small change in energy input is more evident in the heating element than it is in the copper. Change in Thermal Conductivity of Heating Element No significant change was noted after varying the thermal conductivity of the heating element by as much as 100 times. A temperature difference of approximately 2 degrees was noted.

The results obtained at every thermocouple were in agreement; hence it was concluded that the conductivity of the heating element has negligible effects on the overall heat transfer from the heating element to the copper plate. Integrally Heated Conductive Tooling Model The second stage of this project involved what was referred to as the Integrally Heated Conductive Tooling (IHCT). As mentioned previously, the IHCT was developed by the NRC IAR as an out-of-autoclave molding process for polymer composites. The IHCT model that was used in this work consisted of a silicone rubber heating element similar to the one used in the copper plate heating assembly, a layer of combined 3M double sided and Kapton tape, a 296 mm 296

mm 1.15 mm copper plate, a layer of heat resistant silicone adhesive, and a 296 mm 296 mm 5.68 mm carbon fiber epoxy composite (AS4/3501-6) plate. The heating element was bonded on the top face of the copper plate by 3M double sided and Kapton tape. The bottom side of the copper plate was attached to the composite plate by a layer of heat resistant silicone adhesive. The thickness of the silicone adhesive was not known since the technique in which it is applied makes it difficult to measure. Figure 4-1 illustrates the physical model of the IHCT model. This stage was also divided into an experimental phase as well as a simulation phase. A total of 4 thermocouples were used to record temperature readings at different locations. The thermocouples were attached to a computer where LabVIEW was used to import temperature readings. In the simulation phase, Gambit was used to construct the geometric model and mesh of the IHCT model. The Gambit mesh was exported to Fluent for heat transfer analysis. A constant heat flux of 3968254 W/m3 was applied to the heating element simulating the 25W power input. During the simulation phase, 6 simulations were performed in which the effect of silicone thickness and the thermal conductivity of the composite plate in the longitudinal direction were evaluated. Experimental Phase In the experimental phase, the IHCT model was setup in the same orientations that were used in the copper plate heating assembly. Figure 4-2 illustrates the 4 orientations. As explained previously, the motive of orienting the IHCT model is to determine if there is a specific orientation in which heat losses due to convection are minimal.

Procedure Since the composite plate used in the IHCT is 5.68 mm thick, 600 seconds was allocated to the heating cycle to ensure that enough heat would reach the

composite plate for analysis. After 600 seconds of heating, the IHCT model was allowed to cool to room temperature. A total of 8 experimental runs were performed, where 2 trials were conducted for each orientation. In the experimental phase the IHCT model was held by a clamp which in turn was attached to a tripod. Locations of the 4 thermocouples are illustrated in figure 4-3.

Results Horizontal plates

By inspecting the heating cycle of orientation #1, thermocouple #1 heated up most followed by thermocouple #2. After approximately 100 seconds, thermocouples mounted on the composite plate started to heat up at a faster rate.

After investigation of the cooling cycle of thermocouple #1 it was evident that after approximately 170 seconds, all thermocouples with the exception of thermocouple #4 reached the same temperature. At about 200 seconds, all thermocouples including thermocouple #4 reached an equal temperature value and resumed cooling at a uniform rate to room temperature. This is due to the fact that the conductivity of copper is higher than that of the composite. Heat has the tendency to spread at a faster rate in copper than it would in the composite. Thermocouple #3 would have showed the same behavior however it did was not exposed to enough heat to portray this effect. The plate is tested for a Horizontal orientation with Heating Element Underneath. The heating cycle of orientation #2 was similar to that of orientation #1 with little variance. After examining both horizontal orientations, a similar behavior between orientations was witnessed in both the heating and cooling cycles. By examining the cooling curves of orientation 2, similar to the case of the copper plate heating assembly, conduction was more dominant than convection in the IHCT model. Since conduction in the thickness of the composite plate is low compared to that of the copper plate, it took thermocouple #4 sometime before reaching an equal temperature to that of thermocouples #1 and #2. Thermocouple #4 reached higher temperatures than thermocouple #2 note that thermocouple #2 is located on the copper plate where thermocouple #4 is located on the composite plate. From the behavior that was witnessed, it can be concluded that although the composite material has lower conductivity in the thickness, the composite heats up and cools down more uniformly than the copper plate. Behaviors of thermocouples #2 and #4, this will be discussed in the final chapter.

Vertical plates

In analyzing figure 4-11, a similar trend to orientation #3 was witnessed. Thermocouple #4 experienced a higher heat flux about 30 seconds through the heating cycle which consequently caused thermocouple #4 to heat to a higher temperature value than thermocouple #2 by the end of the heating cycle. After observing figure 4-12, thermocouples #1, #2, and #3 reach the same temperature at about 100 seconds. Thermocouples #1 and #2 then cooled further to a temperature lower than that of thermocouples #3. After approximately 250 seconds through the cooling cycle, the curves of thermocouples #3 and #4 start to merge together and cool uniformly with thermocouples #1 and #2. The same behavior seen in the previous 3 heating cycles was also witnessed in vertical heater orientation. Thermocouple #4 experienced a higher heat flux after approximately 30 seconds. At approximately 100 seconds, the temperature of thermocouple #4 exceeded that of thermocouple #2. It was noticeable that the cooling curves of thermocouples #1 and #2 reached an equal temperature to that of thermocouple #3. At approximately 270 seconds, the cooling curves of all 4 thermocouples merged to cool uniformly. A common trend that was witnessed in the heating plots of orientations #3 and #4 was that for the first 100 seconds, thermocouple #2 was at higher temperature values than thermocouple #4. After the 100 seconds mark, it was noticeable that temperature values of thermocouple #4 exceeded those of thermocouple #2. After analyzing the cooling curves, of vertical heater orientation, it was noticed that thermocouples #4 and #2 cooled much slower than thermocouple #2 meaning that the composite plate cooled at a slower rate than the copper. Since the conductivity of the composite is very low, the heat contained within the composite is not distributed throughout the plate as it would in copper. Cooling within the composite

is therefore influenced mostly by convection. The advantage of using a composite plate can be seen here. Since the conductivity of the composite in the longitudinal direction is much lower than that of the lateral direction, when cooling, heat would naturally dissipate in the lateral direction at a faster rate than it would in the longitudinal direction. This is favorable since the ultimate goal is to achieve uniform temperature in the lateral plane of the composite. Effect of Orientation:

it is evident in figure 4-16 that the orientation of the plate had negligible effects on thermocouple #1. A small variation is witnessed for orientation #3; however this may be explained by experimental noise that is unavoidable. The cooling cycle that is witnessed at thermocouple #1 is unchanged as the orientation of the IHCT is varied. In inspecting the heating cycle of thermocouple #2, the variation in temperature witnessed is 2 degrees. A similarity between orientations #2 and #4 is evident however the curves of orientations #1 and #3 are different. Although there may not be much of a difference in temperature variation, the curves appear to be diverging from each other as they are heated. In orientation #3, it took approximately 500 seconds to reach 305.5 K whereas it took orientations #2 and #4 approximately 600 seconds, and it would probably take orientation #1 between 150 200 seconds more than orientation #3. This time difference is considered to be of significance. the cooling cycles of all orientations seem to be similar throughout, although each curve starts at a slightly different temperature. Although a small difference is witnessed in the heating curves of thermocouple #3, this difference however is not as significant as the one observed in orientation #2. Although it appears that heat transfer is different in the composite plate than it is in the copper, the difference between orientations #1, #2, and #4 is insignificant.

It can be noticed in figure 4-21 that the curves of the heating cycle are similar to in all orientations. Little difference was noted among heating curves at thermocouple #4. n examining thermocouple #4s cooling curves; the cooling behavior at thermocouple #4 appears to be similar in all 4 orientations. Little differences were noted among orientations and could therefore be considered insignificant. In inspecting the variation of orientation in the IHCT assembly, unlike the copper plate heating assembly, it was not apparent that orientation had insignificant impact. Thermocouples #2 and #3 showed some differences in the heating curves whereas thermocouples #1 and #4 showed little differences. Simulation Phase A CFD model was also constructed for the IHCT model using Gambit and was exported into Fluent for analysis. A total of 5 simulations were performed for the IHCT model where 3 different silicone thicknesses were used in the first 3 simulations. The conductivity of the composite through the thickness (kz) was varied for the latter 2 simulations. Table 4-2 summarizes the simulation map followed.

Silicone Thickness Variation As mentioned previously, the silicone layer is what bonds the copper plate onto the back face of the composite tooling. Since the thickness of the silicone layer is hard to determine, a variation in the thickness would help determine whether it is an important parameter to consider or not. Note that all thicknesses in the legends of the plots are represented by t and are measured in mm.

Thermocouple #2

After inspecting the results of thermocouple #2, it appeared that as the thickness of the silicone was increased, more heat reached thermocouple #2. This result is logical and was expected. Since thermocouple #2 is located on top of the copper plate, as the silicone thickness was increased, it impeded the heat from traveling to the composite plate, as a result more heat was provided to the copper plate. The magnitude of the change witnessed due to increase in silicone thickness however is negligible. The CFD model revealed that a change in silicone thickness will have a small effect on heat transfer. As was witnessed in figure 4-25, thermocouple #2 did show a slight increase in heat flux as the silicone thickness varied. For thermocouples #3 and #4, they showed that although there was a change in heat flux, the effect that the variation in silicone thickness had is negligible. The second set of simulations was carried out by varying the through conductivity or the kz of the composite plate. Measurement of composite conductivity is a challenging and difficult task. Values of composite conductivity in table 4-1 were obtained from the work of Hind using a dedicated apparatus called Thasys and Thisys. As expected, no change was witnessed in the behavior at thermocouple #1. Since thermocouple #1 is located far away from the area under inspection, no changes have occurred. This is another indication that the simulations are working. Figure 429 revealed a slight increase at thermocouple #2 (top of the copper plate) in heat flux as the conductivity of the composite was varied in the z-direction. The reason behind this behavior is that less heat was transferred to the composite as conductivity of the composite in the thickness was decreased. That excess heat was therefore trapped in the copper plate causing a rise in temperatures at thermocouple #2. The net effect of this however is very small.

Since thermocouple #3 only heated up by 7 degrees, it is difficult to notice a variation. Figure 4-30 however indicates a slight increase in heat flux in the composite plate as the longitudinal conductivity of the composite was increased, which was expected. The effect of varying kz in the composite plate was the most obvious at thermocouple #4. Similar to the observation made for thermocouple #3, as the conductivity of the composite was increased, a higher heat flux was noticed.

Discussion
Validation
In the physical model of the IHCT, the method in which the silicone is applied produces a non-uniform silicone thickness. Thickness variation was taken into consideration when performing simulation runs however a non-uniform thickness of the silicone and possible contact resistance between the silicone and the plates was not accounted for. This did cause some dissimilarity between experimental and simulation results.

By looking at figure 5-1, it is noticeable that the rate at which the simulation model is evolving is faster than the experimental curve. This discrepancy may have been

caused by imperfect contact between the copper and composite plate. Experimental noise is also not accounted for when running simulations. The convection coefficient of air has proven to be an extremely important parameter when developing a simulation model. As was noticed, the convection coefficient of the surrounding medium can drastically change simulation results. One important consideration when considering convection coefficients is that it is a function of temperature difference between the object being heated and the surrounding medium. Previous results revealed that the convection coefficient of air around the heating element was different from the convection coefficient obtained for the copper plate. It is acknowledged that the simulation models developed are of good accuracy and can be used for further study in the future.

Effect of Kapton & 3M Double Sided Tape Thickness


Thermocouple #1 revealed a significant change in temperature distribution as the tape thickness was increased. As the tape thickness was increased, less heat was able to pass through to the copper plate; however this difference in the heat flux at large has a greater effect on the heating element than on the copper plate. This is due to the difference in the volumetric heat capacity as well as the size of each of the copper and the heating element. Given a volume of silicone rubber, it would require 1.5 times more energy to heat the same volume if it were made up of copper. The volume of the copper plate is already 7 times greater than that of the heating element; consequently an increase in energy would be much more noticeable in the heating element than it would in the copper plate. It can therefore be concluded that tape thickness has negligible effect on heat transfer optimization.

Effect of Orientation
Orientation variations were performed for the copper plate assembly as well as the IHCT model. it was determined that the heat transferred from the heating element dissipates to the copper in conduction at a much faster than it would to the atmosphere by convection. The results obtained in the duration of this project contradicted the vertical and horizontal plate examples used to explain the differences between the 4 orientations. The examples used however consider the heated plate completely surrounded by air. In the case of the copper plate heating assembly, the dimensions of the copper plate were considered very small to note any considerable change in convection as the heating assembly was oriented differently. The thickness of the plate was also considered as another factor that caused the change in orientation to have no impact on heat transfer. These findings were in agreement with the work of Al-Arabi & El-Reidi [8]. In the IHCT model however, as the orientation was changed, a small variation was witnessed with the exception of thermocouple #2, and thermocouple #3 which displayed differences in the heating rates. It was noted in the heating cycles of thermocouple #2 that orientation #3 heated up most followed by orientation #4, #2, and finally orientation #1 which heated the least. Similar results were witnessed at thermocouple #3 where orientation #3 heated up most followed by orientations #1, #4, and finally #2. Thermocouple #4 however revealed that orientation of the plate had no impact on the heating cycle as no difference was witnessed between the heating curves.

Further experimentation is required to understand the trend followed by the IHCT tooling as orientation is varied. It is important to understand whether orientation does have an effect on the overall heat transfer from the heating element to the composite part. When manufacturing a complex composite part using the IHCT, it is useful to know whether or not the heating elements used can be oriented vertically or horizontally.

Effect of Conductivity of Heating Element


The conductivity of the heating element was analyzed in section 3.1.1.3 and the findings reveal that negligible changes were seen as conductivity was changed. The heating element was surrounded by air on the top face and was in contact with the copper plate on the bottom face. While is heat transferred in the form of conduction from the bottom of the heating element to the copper plate, heat loss in the form of convection is taking place concurrently on the top face of the heating element. Since heat dissipation is naturally much more rapid by conduction than heat dissipation by natural convection, a temperature gradient is seen in the heating element, where a higher temperature is witnessed on the top face than on the bottom face. When the conductivity of the heating element is varied, a temperature gradient will still be present. It has therefore been concluded that changing the conductivity of the heating element will not optimize heat transfer any further. It could also be concluded that placing heat sensors on heating elements to monitor curing cycles can be very misleading since the temperature at these points are significantly higher than the actual curing temperatures.

Effect of Silicone Thickness Variation


Simulation results revealed that the silicone layer thickness variation did not impede the heat flux by a considerable extent. In the IHCT model however, the silicone is not applied uniformly. This effect was noticed when comparing simulation and experimental results. The composite plate consumed more time in the experiment to achieve the same temperature as the one witnessed in the simulation. This is because the simulation model assumes perfect contact and a uniform silicone layer. Change in silicone thickness may have proved to be a nonfactor however applying the silicone to achieve uniform thickness and better contact between the copper and composite plates has proved to be an important consideration. This is to ensure conduction between the copper and the composite is optimized.

Effect of through Conductivity (kz) Variation in Composite


An unusual effect was witnessed in the experimental and simulation heating curves of the IHCT. Both experimental and simulation results indicated that the thermocouple #4 witnessed a higher heating rate than thermocouple #2. Although copper is 500 times more conductive than the composite plate in its thickness (kz), thermocouple #2 is approximately 15 times more distant from the heating source than the composite. The composite has a lower volumetric heat capacity than copper by a factor of 3; hence it takes 3 times less energy to heat the composite

plate than the copper plate. Other considerations such as heat loss through convection may also have an effect. Further study is required in this area to understand the reasoning behind this behavior. As kz was varied in the simulations, a small change in behavior was noted. As kz decreased, an increase in temperature values was witnessed at thermocouple #2 and vice versa. This is an indication that the IHCT can be optimized by increasing the carbon-fiber volume fraction so that conductivity through the thickness is enhanced. These results are in agreement with Al-Sulaiman, Al-Nassar, and Mokheimer [5].

Conclusion
This project was dedicated to studying the parameters that have an effect on the optimization of heat transfer in the Integrally Heated Conductive Tooling. This method is an out-of-autoclave process to manufacture aerospace grade composite material developed by the National Research Council, Institute of Aerospace Research. Key parameters that were investigated were: convection coefficient of air in the vicinity of the tooling, thickness of bonding tape between the heating element and copper plate, conductivity of heating element, thickness of bonding layer between the copper and composite tool, and finally the through conductivity ( kz ) of the composite. The simulation models developed throughout this work were considered to be relatively accurate and could be used for further study in the future. The convection coefficients of air that were obtained are key values that could be used in simulation models that are more complex. The results attained have aided in understanding key parameters that should be considered when simulating more complex models.