Optimizing Gas Quenching Technology Through Modeling of Heat Transfer
Low-pressure carburizing (LPC) with high pressure gas quenching (HPGQ) offers an environmentally friendly alternative to standard atmosphere carburizing associated to oil quenching. Gas quenching eliminates the need for part washing and oil residue processing steps, allows more automation of the process and offers process control opportunities due to easily changing quenching intensity through a change in gas pressure or velocity.
The current growth of HPGQ technology has been made possible through the improvement of vacuum heat treatment installations , which today have productivity comparable to atmosphere carburizing installations. Part of this improvement is due to the quenching process itself; that is, quenching pressures have increased from 5 to 6 bar in the early 1990s  to 20 bar today, and HPGQ now is typically performed in a separate cold chamber for improved cooling capacity .
Despite these developments, cooling rates needed for some types of loads to obtain the heat treat specification requirements are achieved only with oil quenching . Quenching with helium offers a possible route to increasing cooling rates, but some experimental results infer that helium has equivalent heat transfer performance compared to nitrogen .
This article proposes an optimization of cooling rates for high pressure gas quenching based on a study of the heat transfer modes involved in gas quenching and a one-dimensional (1-D) gas quenching model. The optimization proposed has been validated by experimental tests.
Modeling cooling rates of quenched parts
A simplified heat-transfer model was developed based on the theoretical elements of the HPGQ process (see sidebar), the goal of which is having the capability to quickly evaluate cooling rates inside parts having simple geometries, taking into account convection inside the fluid, conduction inside the steel part, and conducto-convective heat transfer at the interface. This model has been presented previously . Figure 1 shows a representation of a modeled cylinder with the different modes of heat transfer.
Assumptions used to simplify the situation to be modeled include:
- Steel parts are long cylinders (length is at least five times the diameter)
- Gas flow is one-dimensional along the cylinder axis
- Heat exchange between the gas and part is conducto-convective heat transfer
- Conductive heat transfer in the cylinder is only in the radial direction
- Radiation is not taken into account
Modeling results and experimental validation
Temperature evolution was computed for a 30-mm (~1.2-in.) diameter cylinder using nitrogen, helium and an optimized CO2-He mixture. The results shown in Fig. 2 allow the estimation of the efficiency of the heat transfer improvement performed on the steel part in terms of cooling time. The cooling time is given relative to the cooling time obtained for quenching under nitrogen. Cooling time between 850 and 500°C (1560 and 930°F) can be reduced by 30% using an optimized gas mixture of 60% CO2 and 40% helium.
Experimental validation tests for the study were performed on an ECM quenching installation (including LPC) located at ECM Service in Fontaine, France, which allows up to 20 bar HPGQ in a separate cold chamber. The installation has the same specifications as standard ECM LPC-HPGQ equipment.
The main criterion used to compare the efficiency of various quenching solutions is the core hardness of the steel parts, which depends on the cooling rate actually reached at this location. Some quenching trials were characterized directly by the cooling rate measured inside the parts. Some authors  use the average cooling rate reached between 800 and 500°C as an efficiency criterion, which represents the critical temperature range where undesired phase transformations may occur. Core hardness is determined by measuring the temperature inside the parts using thermocouples linked to a heat-insulated data recording device attached to the load, which is independent of the treated steel grade.
Tests consisted of comparing quenching gas efficiency on furnace loads with embedded temperature measurement devices to determine cooling rates on actual loads of varying geometries and to compare experimental results with oil quenching. Figure 3 shows the load of 28 mm diameter, 16MnCr5 cylinders in the first series of tests. The use of cylinders allows independence to any specific geometry in conditions that are close to those that can be modeled by the equations which have been put into the simplified model presented above. Figure 4 shows the cooling rates for the test parts. Cooling rates were calculated for nitrogen and CO2-He mixtures with various proportions of CO2 and helium.
Quenching pressure was 20 bar for each test. Fig. 4 shows the dispersion of the cooling rates values over the load; the dispersion range was the same for all tests.
Experimental results presented in Fig. 4 confirm the existence of an optimized CO2-He gas mixture (corresponding to the 60% CO2 mixture) as inferred by theory. Cooling rates are 30% better compared with average rates achieved using nitrogen (i.e., 15.7°C/s).
Confirming experimental results
Furnace loads containing two types of industrial parts (massive engine parts and transmission ring gears) were used to verify the improved cooling rates achieved in experimental work, which took into account the influence of parts geometry and steel grade. Cooling rate was the criterion used to determine quenching efficiency by measuring the temperature at the location used for core hardness control.
Average cooling rates measured for both types of loads are shown in Fig. 5 for nitrogen, helium and CO2-He mixture quenching at 20 bar. The results confirm the results from experimental tests performed on cylinder parts; i.e., CO2-He improves cooling rates by more than 30% compared with nitrogen for both types of parts. Note that the improvement brought by helium seems to depend on the part geometry. For the most massive part, helium and nitrogen show similar cooling rates.
The first two steps of the validation tests confirmed the improvement of heat transfer efficiency shown by the theoretical and modeling approach. The last step of the validation tests compared the performance of high pressure gas quenching with oil quenching, which sets the standard for high cooling rates.
For oil quenching, the temperature measurement setup used previously could not be used; no temperature measurement apparatus was used in the loads of massive parts. The loads were vacuum carburized and quenched in an ECM installation with oil, 20 bar nitrogen and 20 bar CO2-He. The criterion used to determine quench efficiency, and how the heat transfer and cooling rate improvement translates into hardness improvement, was final core hardness.
Figure 6 shows the core hardness of parts in three different levels of stacking in the load. For oil quenching, the lower level of the load is quenched first and therefore reaches higher hardnesses. For gas quenching, the upper level is quenched faster since quenching gas is injected from the top. The CO2-He gas mixture results in an average hardness increase of 30 HV compared with nitrogen quenching, and more importantly, it produces the same core hardness level as oil quenching.
High-pressure gas quenching using an optimized gas composition opens the way to achieve quenching performance equivalent to that of oil quenching. However, these tests show that parameters such as the type of parts that are treated may have a strong influence on the actual effect of changing the quenching gas compositions. Further tests will be conducted for various working conditions to determine optimized heat transfer.
SIDEBAR: Theoretical study of the high pressure gas quenching process
Heat transfer mechanisms
High pressure gas quenching consists of cooling the load through direct heat exchange with a pressurized gas. Figure A shows four different heat transfer phenomena involved in gas quenching:
1) Heat conduction inside steel parts. Heat extracted at the part surface is transferred from its core through diffusion inside the solid, involving no physical displacement. Conduction of heat in solids is described by Fourier's law : φcond=-λs·grad(T) where jcond is the surface conducto-convective heat flux and -λs is heat conductivity of steel part. The conductive heat flux at each location depends on part geometry and thermophysical properties and on boundary conditions (heat flux at the part surface related to other heat transfer modes. Heat conduction is responsible for the delay in cooling of the core compared with the surface.
2) Forced convection. Heat convection is related to the displacement of matter; it can only occur in the gas in gas quenching, and is described as forced convection due to the high velocity gas directed into the chamber. Convective heat transfer allows the transfer of heat from the load to the heat exchangers inside the quenching installation. The main way to improve convective heat transfer for gas quenching is to increase gas velocity. This requires higher pump speed, which is limited by currently available motor power and impeller rotation speed.
3) Radiation. Hot steel parts to be quenched emit infrared radiation, which can be absorbed by colder surrounding elements (such as quenching cell walls), thus contributing to cooling the load. A part located within the load is not cooled by radiation because it is surrounded by hot parts. Quenching gases including nitrogen, helium and argon are transparent to infrared radiation, and do not exchange heat by radiation. Better use of radiative heat transfer could contribute to higher gas quenching cooling rates.
4) Convection coupled with conduction (conductive-convective heat transfer) at the part surface. Because of the weak gas velocity close to the part surface in the boundary layer, the importance of conductive heat transfer in the gas cannot be neglected . Therefore, both conduction and convection of heat at the part surface occur in a coupled way. In most cases where radiation can be neglected, conducto-convective heat transfer is the only means for cooling the part, and the corresponding heat flux is the main parameter to improve to achieve faster cooling rates.
Conducto-convective heat transfer at the interface between a solid and a fluid is a well-studied heat transfer mode. The corresponding heat flux is usually modeled as a function of the difference between the interface temperature and the average fluid temperature.
How to increase cooling rates
The main way to increase cooling rates in the quenched parts is to make the conducto-convective heat transfer more efficient, which can be achieved by increasing the conducto-convective heat transfer coefficient (h) given by: h=K·Reα·λ. Equation 2 shows that one way to improve h is to increase the Reynolds number (Re), which is dimensionless number indicating type of flow; i.e., the higher the Reynolds number, the more turbulent is the flow. Higher turbulence can be achieved with higher gas velocities, which as mentioned previously is limited by available motor power and impeller speed. The Reynolds number also can be increased by using higher molecular weight gases and higher gas pressure. Another way to improve h is to use a gas having higher heat conductivity (l). In Eq. 2, the heat transfer at the part surface results from coupled convection (through Re) and conduction (through l) of heat inside the gas.
Use of radiative heat transfer
Using an infrared-semitransparent gas instead of a transparent gas is a supplementary way to improve heat transfer at the surface of quenched parts. Radiation from the parts to the gas affects all parts within the load as opposed to radiation from the parts to the furnace walls. To increase heat transfer at the surface of the quenched parts for faster cooling, it is necessary to optimize and balance the quenching gas density, heat conductivity and infrared absorption properties.
Figure B shows the Reynolds number, heat conductivity and Real product corresponding to a standard configuration for nitrogen and helium (commonly used quenching gases) and CO2 as an infrared absorbing gas. Generally, lighter gases such as helium produce a lower Reynolds number (lower density) and higher heat conductivity compared with gases having a higher Reynolds number such as CO2. Therefore, Re and l should be balanced to optimize h. Mixing a high-conductivity gas such as helium with a dense gas such as CO2 allows balancing these properties to optimize the heat transfer coefficient. Figure B shows balanced Reynolds number, heat conductivity and Real for the best CO2-He mixture (i.e., one containing 60% CO2).
Except for density and heat capacity, the physical properties of gas mixtures cannot be determined as the sum of each individual gas properties, weighted by the mass fraction of each gas. Extensive research has been conducted to determine the physical properties of a gas mixture. Some authors have proposed simple relationships with low (<2%) error [9, 10 , 6]. On the basis of the theoretical approach and Fig. B, the use of the CO2-He mixture increases the heat fluxes by almost 80%.