This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Nickel superalloy turbine blades are highly engineered components. Industrial gas turbine blades required the blades with longer time creep rupture properties and good hot-corrosion resistance.  To survive in harsh and extremely high temperature environment, some of the turbine blades undergo heat treatment after the casting process to achieve optimal balance of its mechanical properties. Gas quenching becoming the common heat treatment process used because of its advantages in comparison to the conventional heat treatment process such as oil and polymer quench. The advantages of gas quenching include the controllable cooling parameters of the furnace. The flexibility of controlling the direction of flow, furnace pressure and the mass flow rate of the gas make gas quenching much more favourable compared to the conventional methods.  Furthermore, the surface quality of the components undergo gas quenching will be retained after the treatment process.  Hence, the industry which required the good surface finish of the component will be beneficial from gas quenching, especially turbine blades where the surface roughness of turbine blades affects the flow of fluid. Gas quenching is more environmentally friendly compared to conventional quench methods as there is no harmful gases are used or produced during the heat treatment process. 
Heat treatment involves heating and rapid cooling or quenching of the components. The heating up process required to increase the components from room temperature to the set temperature through radiation from the heating elements. The time period for the components undergo heat treatment to reach the set temperature uniformly is called thermal hysteresis time.  After the thermal hysteresis time, heat is provided continuously during the 'hanging time' to allow the microstructures of the component develop to austenite which is also called gamma phase.  In this case, the nickel-base alloy in the turbine blade undergo the phase transformation from body-centred cubic (BCC) to face-centred cubic (FCC) configuration of gamma phase. 
Gas quenching is controlled rapid cooling of heat-treated component.  The desired mechanical properties of the turbine blade are obtained by controlling the quenching process. The flow of gas during quenching is very complex due to the placement of the components along with the other devices that are used to control and perform the heat treatment process such as process control thermocouples, heating elements, and the basket to contain the turbine blades. The complex flow path of quench gas leads to the complex heat transfer from the turbine blades to the gas. The furnace must be designed in a way to deliver uniform distribution of flow to all the quenched components. Plastic deformation may occur during quenching due to the different in thickness along the cross section of turbine blade, see figure 1. The thinner part of the turbine blade will be cooled faster compared to the thicker part of the blade. Matters will experience contraction when they are cooled and expansion when they are heated up. The expansion and contraction will occurred in different part of the turbine blade due to the temperature differences. Hence, this could lead to plastic deformation in the along the turbine blade.
Figure 1: Photo of turbine blade from aero-engine 
Figure 2: Thermal, metallurgy, and mechanical couplings in heat treatment. 
Turbine BladesThe couplings chart in shown on figure 2 can be used to predict the residual stress and distortions. According to the chart, temperature distributions and temperature gradients in the quenched components generated thermal stresses and phase transformation during quenching. Therefore, it is necessary to investigate the temperature distributions in the turbine blades. Experimental methods for determine the temperature gradients of the turbine blades is done on the fastest and the slowest cooling rate during quenching. The down side of the experimental methods is that the stress developments are not able to be accessed during the quenching process. Hence, a simple model to describe the quenching process is done on the root block of turbine blade. According to experienced engineers in Doncasters Group Ltd, they believed that the cracking happened during the first five minutes of the quenching process. Therefore, the analysis is focused on the first twenty five minutes of the quenching.
There are a few experiments had been designed and performed in order to achieve the aim. To start with, the cooling rate in the root of the turbine blades at different positions in the furnace is investigated. During the heat treatment process, the turbine blades are arranged in array in the furnace, shown in figure 3. Hence, an experiment is conducted to determinate the cooling rate of every root block placed in the furnace. However, due to the limitation of the maximum of twelve thermocouples that is able to support by the furnace, the thermocouples are placed as shown in the figure 3(b).
Figure 3(a): Side view of the furnace
Figure 3(b): Top view of the furnace with the placement of thermocouples. (The 'x' denotes the placement of thermocouples.)
The experiment is begun by drilling a hole into the centre of root block of the turbine blade to accommodate the thermocouples. In this way, the temperatures in the root blocks are able to measure correctly. The thermocouples placed in the root are fixed to be 84 mm of depth and the depth of the insert is set to be constant for all turbine blades. The reason of setting the fix depth into the hole is to set the thermocouple inserts depth to be controlled. Hence, the factor is not causing any error to the readings. The turbine blades with the thermocouples is then loaded to the cage and placed in the middle of the furnace. The serial number of the thermocouples is recorded for and the sockets are connected. The door is closed and the furnace is run for heating and quenching process using the recipe recommended for the turbine blade. The data is sampled once every second using the Euroterm Datalogger for the furnace along the heat treatment process. Once the process is done, the data is downloaded from the company database for analysis and the experiment is repeated to validate the results before proceeding to the next experiment.
Once the data from experiment 1 and experiment 2 being analyzed, the cooling rate of the turbine blade root block varies with the positions in the furnace is able to be determined and calculated. The next experiment is done to find the temperature distributions within the turbine blades at the location with fastest and slowest cooling rate. Two freshly-scrapped turbine blades are used for instrumenting. Holes and grooves are made and labeled in the specific places in the part of the blade as shown in figure (4).
Figure 4(a): Placement of thermocouples in position with slowest cooling rate.
Figure 4(b): Placement of thermocouples in position with fastest cooling rate
A thin nickel alloy plate is welded to location 4, 5, 6, 4a, 5a and 6a to form a holder for the thermocouple. The hole in location 1 and 1a is drilled through from the side of the turbine blade. During quenching process, the momentum of the gas on the thermocouple may causes the vibrations and changes in place of the thermocouple. This may cause in the experiment failure and errors in the data recorded. The purpose of the holder is to hold the thermocouples in place in good thermal contact with the turbine blade as well as to isolate the thermocouple from the direct contact with the gas that is pumped into the furnace during quenching process. The blades are placed in the position where the root block has the fastest and slowest cooling rate respectively according to the results from experiment 1. The thermocouples are made sure to be in close contact with the blade to avoid incorrect measurements. The fully loaded cage is then placed in the middle of the furnace. The serial number of the each thermocouple is recorded and the sockets are connected. After the door is closed, the furnace is run for heating and quenching using the same recipe in the first experiment. Similar to the first experiment, the data is sampled once every second using the Eurotherm Datalogger for the furnace and the data is downloaded from the company database after the process for analysis.
For the full heat treatment process and accurate simulation, it is necessary to model the full furnace with all the turbine blades and devices to describe both the heating process and the quenching process.  However, due to the time constrain and very powerful computer is needed to simulate the both heating and quenching process, a simple simulation is modeled. Since the heat transfer from the root block during quenching is mainly due to convective heat transfer, a computational fluid dynamic (CFD) analysis is able to use to simulate the heat transfer from the surface of the root block. A k-epsilon turbulent flow and heat transfer model of the turbine blade root block during the quenching process is made to predict the temperature distribution over the root block of turbine blade by using Fluent software. 
To begin with, the root block model is imported to Gambit software. Gambit is the software to model and mesh the CFD model which is compatible with Fluent software. A simple furnace block is constructed around the root block. Because of the side symmetry of the furnace and the root block, half of the model is used. The root block is meshed with finer tetrahedral mesh due to its complex geometry and to obtain accurate results during simulation. The furnace is meshed with coarser tetrahedral mesh where the flow in simple mesh is less important compared to the temperature distribution in the root block. Furthermore, the gas is defined to flow from the bottom of the furnace to the top which is similar to the setting for the first and second experiment. The meshed model is exported to Fluent software for numerical simulation for CFD analysis. The boundary conditions set up for the simulation could be referred to table (1).
Table 1: Boundary conditions used in CFD analysis.
Wall of the furnace
Heat flux = 0
Initial Temperature = 1232 °C
Velocity inlet = evenly distributed 1 m/s
Pressure Outlet = 3 x 105 Pa
The convective heat transfer between the root block and the flow gas can be calculated by using Newton's law of cooling,
where Q is the heat flux, h is the heat transfer coefficient, A is the area of the part, and is the temperature difference between the surface of the component and the flow gas.  From the equation, the heat flux is proportional to the area of part times the difference between the surface and the gas temperature.
However, heat transfer coefficient between irregular surfaces of the root block with the quenching gas is almost impossible to determine directly.  In Fluent software, transient convection problem is able to overcome the problem. Hence, the simulation of the heat transfer through convection is able to make with coupled and transient solver in Fluent software. Four points are placed on the similar places to the holes of the root block in the second experiment where the temperature gradient of the turbine blades is investigated. The temperature distribution and the temperature of the points in the root block for every time steps are recorded for the first 300 seconds of the quenching process. Further 1200 seconds is simulated to obtain the temperature of the points placed in the root block for comparison with the results from second experiment.
Table 2: Table of cooling rate with respect to the position in the furnace in experiment 1 and 2. The position of the turbine blades is corresponding to figure (3b). Cooling rate for the first 300 s is calculated for both experiments and the mean absolute error of the cooling rate is calculated. Similarly for cooling rate for 300s to 1500s.
Cooling rate for first 300s (°C/s)
Mean absolute error (%)
Cooling rate for 300s to 1500s (°C/s)
Mean absolute error (%)
Cooling rate for first 300s (°C/s)
Mean absolute error (%)
Cooling rate for 300s to 1500s (° C/s)
Mean absolute error (%)
Cooling rate for first 300s (°C/s)
Mean absolute error (%)
Cooling rate for 300s to 1500s (°C/s)
Mean absolute error (%)
Figure 5: Graph of cooling rate vs time for experiment 1 respective to the position in the furnace. Refer table (1) for the position of the turbine blades in the furnace. The pressure is embedded into the graph and labeled where the scale for the pressure at Y axis is 1104 Pa. Mean temperature of all 12 turbine blades in different position is plotted in the graph with the scale 110 °C.
Figure 6: Graph of cooling rate vs time for experiment 3. The legends and the labels on the graph represent the position of thermocouples at turbine blades (See figure 4(a)). The position of the turbine blade is place at position A shown in table (2).
Figure 7: Graph of cooling rate vs time for experiment 3. The legends and the labels on the graph represent the position of thermocouples at turbine blades (See figure 4(b)). The position of the turbine blade is place at position K shown in table (2).
Figure 8: The combination graph of temperature vs time for experiment 3 with the comparison with simulation results. The quench target is drawn on the graph where the turbine blade had to be quenched to 1080 °C at the first 300 s and to 540 °C after 1500s quenching process. The numbers of the points in simulation is corresponding to the points at the blade, example 1 (Simulation) is corresponding to 1 and 1a on the blades.
Figure 9(a): The crack rate based on the heat treatments cycle runs that contained cracks in the process. The size of the circle represents the crack rate which the colour represent the cooling rate in the first 300s based on the result from experiment 1. The black colour of circle represents the crack without known cooling rate and hollow circle are the position without the crack.
Figure 9(b): The crack rate based on the heat treatments cycle runs that contained cracks in the process. The size of the circle represents the crack rate which the colour represent the cooling rate in 300s to 1200s based on the result from experiment 1. The black colour of circle represents the crack without known cooling rate and hollow circle are the position without the crack.
The assumption made before the experiment stated that the temperature of the blades at the symmetrical position would have the same heating and cooling rate. Referring to figure 3 and table (2), the arrangement of the turbine blades is in array and the features of the furnace are side symmetry. The inlet and the outlet are at the bottom and the top of the furnace respectively. This makes the furnace is able to be assumed as side symmetrical where turbine blade A is corresponding to turbine blade D, turbine blade F is corresponding to turbine blade E and the turbine blade I is corresponding to turbine blade L. In the assumption, the cooling rate of corresponding position would have the same cooling rate. The reason of the assumption made is because the furnace supported up to maximum of twelve thermocouples. Uneven gas flow is likely to be happened in the furnace during the quenching process when there are variations of cooling rate in the turbine blades of every location. Turbine blade at position K is cooled fastest compared to the blades at the other positions. The variation can be seen too in the symmetrical positioned turbine blades where turbine blade at position L is cooled faster compared to the turbine blade at position I in experiment 1. It is obvious too for the turbine blade in position A and D. The cooling rate is differed for the symmetrical positioned turbine blades. The reason could be due to the shape of the turbine blades which has the concave and convex shape at one of their side. The view factor is used to account for the effect of orientation on radiation heat transfer between surfaces. Technically speaking, Fii = 0 for plane and convex surfaces while Fii has a value for concave surface.  Therefore, there is no radiation which able to leave the convex surface and hit back later. However, it is different for concave surface. The radiation left the surface able to hit back the concave surface. Therefore, the turbine blades which received the direct radiation from the heating element with concave surface will have higher heating rate and slower cooling rate. Besides, the variation of the temperature could be affected by the placement of the basket. The basket could be place off centred in the furnace.
The first experiment is repeated with five second data sampling to validate the results from the first experiment. Based on table (2), the cooling rate is varied for the turbine blade in the same position for the both experiment. It could be seen that the cooling rate at position K has faster cooling in experiment 2 compared to experiment 1 in the first 300 s during quenching. However, value of the cooling rate seems to be closer from 300 s to 1500 s for both of the experiment. The trend is the same for turbine blade in position I too. This could be due to turbulence that occurred when quenching gas is pumped to fill the furnace and to achieve the furnace pressure of 3 bars. This can be related to the fluctuation of the cooling rate at the first 150 s (see figure 5). it can be seen that the fluctuation of the cooling rate is related to the pressure of the furnace. The fluctuation of the pressure can be at the beginning of the quenching process can be refined to increase uniformly or in the smaller stepping size to avoid the fluctuation of cooling rate.
As the pressure of the furnace increase, collision rate of the gas atoms increases. When the gas is being compressed, the collision rate within the gas with the turbine blade is higher. Hence, the cooling rate would be higher as the pressure increase. The cooling rate of the turbine blades is at their maximum when the pressure of the furnace reached 3 bars. However, there is a lagging time in when the furnace reached 3 bars and the time that reaches the maximum cooling rate of turbine blades. The reason is the thermal conductivity of the alloy. The thermocouples are inserted into the middle of turbine blades root block and the temperature of the internal root block is being measured. When the quench gas started to extract the heat from the surface of the turbine blades the heat is being conducted from the middle of the turbine blade to the surface. Hence, temperature gradient is existed in the cross section of turbine blades.
Figure 9: Diagram of cooling rate and ranking at 300s. The darker the colour in the circle, the higher the cooling rate. The size of the circles represents the ranking of the cooling rate. The biggest circle ranks the first in the fastest cooling rate.
Refer to figure 9, the turbine blades placed at the front row and the middle row have higher cooling rate compared to the back row due to the design of the inlet and the baffles of the inlet and the outlet. This is due to the inlet that is designed inclined towards the front part of the furnace with several baffles to reflect the gas to the back of the furnace, as shown in figure 3(a). The quench gas is more focused on the turbine blade at the front row and at the middle of the furnace. The baffles that placed at the outlet of the furnace reflected a part of the gas flow to the back of the furnace and some of the flow is redirected to the outlet and to heat exchanger. The design of the baffles could cause the recirculation and turbulence of gas flow. Besides, one of the reasons that the back row of turbine blades had slower cooling rate is the size of the baffle that reflected the gas to the back of the furnace. The size of the baffle could be increased to reflect more quench gas to the back of the furnace.
In the third experiment, the cooling rate for the blade at position K (see table (2)) which has the highest cooling rate in experiment, point 6a (refer to figure 4(b)) has the highest cooling rate. This is predictable as the point is located at the blade which the cross section is the thinnest compared to the other points. However, for the turbine blade positioned at A where the position has the slowest cooling rate in experiment 1, point 6 (refer to figure 4(a)) ranked third for the fastest cooling rate across the same blade. This is most likely due to the baffles where the gas is reflected to the back of the furnace hit the root block first and but not the blade. This results in the turbine blade in position A has the higher cooling rate at point 3 and followed by 2 where the quench gas reflected is being hit directly to the spot. The blade may be cooled faster when the circulated gas in the furnace from the front side of the furnace directed to the back of the furnace.
From both of the graphs in figure 6 and figure 7, the turbine blades at both positions started off with maximum cooling rate at the first 100s when the gas started flowing into the furnace for quenching. However, the turbine blade at position A seems to have gradual or flatter cooling rate and which finally achieve the uniform temperature gradient with point 1 which the point is placed in within the root block of the turbine blade. By looking at the figure 6, the slope of the graph seems to be much steeper compared to the graph in figure 6. In other words, the surface of the turbine blade at position K cooled much faster than the inner body of the root block of the turbine blade. Hence, a larger thermal stress would be generated due to the larger thermal gradient.
Based on figure 8, the simulation trends resembled the graph for turbine blade in position K which has the highest cooling rate in the furnace. The ranking of points from the fastest cooling rate to the slowest cooling rate in simulation has similarity to the experimental result for the blade at position K, which are 2a, 3a, 4a and lastly 1a. However the temperature at 1500 s is not similar for both simulation and experimental results. This is due to the assumption made during the modelling. The velocity of the quench gas at the inlet is being assumed as there is no investigation made on the mass flow rate through the inlet. Besides, the trends of the simulation are seemed to have straight line graph while experimental graphs are started slowly and had fall very quickly then slowed down. The odd in shape of graph could be caused by other parameters that are not considered into the modelling. The radiation between blades is very much affecting the cooling rate of the turbine blades. Furthermore, the turbulence of flow caused by the devices in the furnace as mentioned above is being ignored in simulation. Besides, the gas is model just to flow straight from bottom to the top of the furnace. However in the real furnace, there is a time gap for the gas to fill up the furnace to reach the pressure of 3 bars. Hence, there will be recirculation of gas flow in the furnace. The simulation required further refinement to obtain the same end temperature with the experimental result. With the success model, the temperature of the other parts on the turbine blades could be estimated by using the correct model without having to do experimental in trial and error method. Other than that, the parameters or quench profile could be changed to obtain the best solution to minimize distortion and crack in the furnace.
From figure 9(a) and figure 9(b), it is shown that the cracking is proportional to the cooling rate distribution in the furnace. The trend of the data shows that the cracks happened frequent at the position with higher cooling rate. Further investigation should be done in the distribution of quench gas and the behaviour of flow in the furnace.
The distributions of temperature on the turbine blade affect the thermal stress developed. Therefore, the design of the furnace is important in distributing uniform heat transfer to the quenched body. Modelling of the full furnace including both heating and quenching process is required to visualise the distribution of flow in the furnace and the heat transfer. Besides, finite element analysis (FEA) could be implied to analyse the residual stress development. The furnace could be redesign to distribute the quench gas more uniformly and new recipe or the quench profile can be developed to minimize the distortion and cracking of the quench component.
This project is done under the supervision of Dr. Mark Ward and Dr. Zhu Zhang. The research is sponsored by Doncasters Group Ltd. The information and supports from the company is truly appreciated.