I investigate the thermal behavior of spiral bevel gears during precision grinding because the quality of gear cutting and the final tooth surface depend strongly on the temperature field generated in the contact zone. In my approach, I combine gear geometry, gear cutting kinematics, abrasive contact mechanics, heat partitioning, and finite element thermal analysis. The main objective is to predict the transient temperature field over the whole grinding process and to identify the influence of process parameters. I treat gear cutting as the common foundation of both generating grinding and form grinding, and I use the same geometric framework to describe the wheel-workpiece engagement.
I begin with the assumption that the grinding wheel and the workpiece form a conjugate pair through a virtual generating gear. This idea is central to gear cutting because the wheel represents one tooth of the virtual generating gear, and the relative motion between the wheel and the workpiece reproduces the cutting motion. For generating grinding, the contact between the wheel and the tooth surface is a line contact at each instant. For form grinding, the wheel profile matches the tooth space, and the contact is distributed over a larger area. In both cases, the heat generated by gear cutting is concentrated in a small region and then conducts into the workpiece, chip, wheel, and coolant.
| Symbol | Meaning | Typical unit |
|---|---|---|
| \(T\) | Temperature | °C or K |
| \(t\) | Time | s |
| \(\mathbf{R}_{br}\) | Position vector of a tooth surface point | mm |
| \(q_r\) | Cradle angle | rad |
| \(\theta\) | Blade phase angle | rad |
| \(b_{tr}\) | Distance along the blade from the tip | mm |
| \(K_x,K_y\) | Normal curvatures of the fitted surface | 1/mm |
| \(G_{xy}\) | Geodesic torsion of the fitted surface | 1/mm |
| \(l_c\) | Contact arc length | mm |
| \(F_n,F_t\) | Normal and tangential grinding forces | N |
| \(q\) | Heat flux density | W/mm² |
| \(R_w\) | Heat partition ratio to workpiece | — |
| \(v_s\) | Wheel linear speed | m/s |
| \(v_w\) | Workpiece feed speed | mm/min or m/s |
For gear cutting by generating grinding, I define the velocity relation at any instant. The grinding speed is obtained from the vector sum of the wheel speed and the relative feed motion. If the feed motion is small compared with the wheel speed, the resultant grinding speed is close to the wheel speed, but its direction changes along the instantaneous contact line. I express the velocity triangle as
$$ \mathbf{V}_m=\mathbf{V}_s+\mathbf{V}_{gx}, $$
where \(\mathbf{V}_m\) is the resultant grinding speed, \(\mathbf{V}_s\) is the wheel linear speed, and \(\mathbf{V}_{gx}\) is the feed speed. For form grinding with an eccentric spindle mechanism, the reciprocating speed is written as
$$ V_{wi}=V_{fw}\sin\alpha_t, $$
$$ V_{fw}=\frac{2\pi n_w e}{60}, $$
where \(n_w\) is the rotational speed of the eccentric mechanism, \(e\) is the eccentric radius, and \(\alpha_t\) is the instantaneous rotation angle. These kinematic relations are the first step in my gear cutting thermal model because they determine the local sliding velocity and the local heat generation rate.
I model the tooth surface of the spiral bevel gear from the virtual generating gear and the wheel geometry. The blade surface in the cutter coordinate system can be described by a tangent vector, a normal vector, and a position vector. I write the initial blade vectors as
$$ \mathbf{t}_r^{(o)}=\begin{bmatrix}0\\ \sin\alpha_{br}\\ \cos\alpha_{br}\end{bmatrix}, $$
$$ \mathbf{n}_r^{(o)}=\begin{bmatrix}0\\ \cos\alpha_{br}\\ -\sin\alpha_{br}\end{bmatrix}, $$
$$ \mathbf{R}_{cr}^{(o)}=\begin{bmatrix}0\\0\\-r_{cr}\end{bmatrix}, $$
where \(\alpha_{br}\) is the blade pressure angle and \(r_{cr}\) is the cutter tip radius. After rotating the cutter coordinate frame by the two setup angles, I obtain the transformed vectors
$$ \mathbf{t}_r^{(c)}=M_j(j_1)M_i(i_1)\mathbf{t}_r^{(o)}, $$
$$ \mathbf{n}_r^{(c)}=M_j(j_1)M_i(i_1)\mathbf{n}_r^{(o)}, $$
$$ \mathbf{R}_{cr}^{(c)}=M_j(j_1)M_i(i_1)\mathbf{R}_{cr}^{(o)}. $$
I then rotate these vectors around the generating surface axis to obtain the tangent, normal, and tip position of any axial section. The resulting vectors in the machine coordinate system are
$$ \mathbf{t}_r=M(q_r,\theta)\mathbf{t}_r^{(c)}, $$
$$ \mathbf{n}_r=M(q_r,\theta)\mathbf{n}_r^{(c)}, $$
$$ \mathbf{R}_{cr}=M(q_r,\theta)\mathbf{R}_{cr}^{(c)}. $$
The tooth surface point is finally expressed as
$$ \mathbf{R}_{br}=\mathbf{D}_{rr}+\mathbf{A}_{cr}+b_{tr}\mathbf{t}_r, $$
where \(\mathbf{D}_{rr}\) is the offset vector from the gear axis to the machine origin, and \(\mathbf{A}_{cr}\) is the cutter tip position after rotation. For generating grinding, the parameter \(b_{tr}\) is determined by the meshing condition and the roll ratio. For form grinding, \(b_{tr}\) is treated as an independent parameter, and the tooth surface is obtained directly from the wheel profile. This distinction is essential in gear cutting because the two methods produce different contact zone shapes and different temperature histories.
Because the tooth surface is a free-form surface, I discretize it before thermal analysis. I first determine the theoretical tooth surface boundary in the axial section. I then shrink the boundary slightly to avoid divergence near the edges. The shrink distances are chosen proportionally so that the retained area remains as large as possible while the numerical scheme remains stable. I divide the tooth height and tooth length into a grid, compute the coordinates of each grid point, and fit the resulting point cloud with spline surfaces. This discretized surface becomes the geometric basis for my gear cutting contact calculations.
For generating grinding, I determine the instantaneous contact line by searching the tooth surface with fixed cradle angle and varying blade phase angle. A point is accepted only if it lies inside the theoretical tooth surface boundary. I use four boundary functions to test whether the point is inside the projected tooth surface. If the point satisfies
$$ R>f_1(L),\quad R<f_2(l),\quad r="" rf_4(L), $$
then it belongs to the valid tooth surface. This filtering step is important in gear cutting simulation because invalid points outside the tooth surface can cause large errors in contact length and heat flux.
The local contact between the wheel and the workpiece is approximated by two fitted surfaces. At any point on the instantaneous contact line, I expand the surface height function in a local coordinate system and retain terms up to the second order. The fitted surface is
$$ z_1=\frac{1}{2}K_x x^2+G_{xy}xy+\frac{1}{2}K_y y^2. $$
I compute the two orthogonal normal curvatures and the geodesic torsion from the relative motion parameters. Using Euler’s formula, the normal curvature in an arbitrary direction is
$$ K_n=K_1\cos^2\theta+K_2\sin^2\theta+2G\sin\theta\cos\theta, $$
and the geodesic torsion is
$$ G_n=(K_2-K_1)\sin\theta\cos\theta+G(\cos^2\theta-\sin^2\theta). $$
These quantities define the shape of the contact removal region. For a given grinding depth \(h_z\), the intersection of the fitted surface with the plane \(z=h_z\) gives the boundary of the removal area. The area is
$$ S_g=\frac{1}{2}\int_L\left(x\,dy-y\,dx\right). $$
I transform this boundary curve into a plane perpendicular to the resultant grinding speed. The transformed curve represents the effective profile of the workpiece material removed by the wheel. I then shift the curve downward by the grinding depth and define two functions \(g_1(y)\) and \(g_2(y)\). The contact arc length is
$$ l_g=g_2(0)-g_1(0). $$
For form grinding, the contact situation is different. The wheel profile contacts the whole tooth surface, and the eccentric spindle repeatedly shifts the wheel between the concave and convex flanks. The reciprocating motion helps the coolant enter the contact zone and helps the chips escape. I express the incremental grinding depth for the concave flank as
$$ z_{ai}=\frac{Z z_t}{h_c}\cos\alpha_a, $$
and for the convex flank as
$$ z_{bi}=\frac{Z z_t}{h_c}\cos\alpha_b, $$
where \(Z\) is the total grinding depth, \(z_t\) is the feed per unit time, \(h_c\) is the total feed length, and \(\alpha_a,\alpha_b\) are the outer and inner blade angles. These equations allow me to describe the time-varying material removal in gear cutting by form grinding.
After defining the contact geometry, I calculate the grinding force. I use a single-grain plane grinding force model and then integrate or sum the grain forces over the contact zone. The total normal force and tangential force are divided into chip formation, plowing, and sliding components:
$$ F_n=F_{nch}+F_{npl}+F_{nsl}, $$
$$ F_t=F_{tch}+F_{tpl}+F_{tsl}. $$
The chip formation components are written as
$$ F_{nch}=\frac{u_{ch}v_w a’}{v_s}, $$
$$ F_{tch}=k_{ch}F_{nch}, $$
where \(u_{ch}\) is the specific chip formation energy, \(v_w\) is the feed speed, \(a’\) is the effective grinding depth, \(v_s\) is the wheel speed, and \(k_{ch}\) is the chip formation coefficient. The plowing components are
$$ F_{npl}=k_{pl}F_{tpl}, $$
and the sliding components are
$$ F_{nsl}=p_c A_g b l_k, $$
$$ F_{tsl}=\mu F_{nsl}, $$
where \(p_c\) is the average contact pressure, \(A_g\) is the wear flat area, \(b\) is the contact width, \(l_k\) is the sliding contact length, and \(\mu\) is the friction coefficient. Therefore, the unit-width normal and tangential forces become
$$ F_n’=\frac{u_{ch}v_w a’}{v_s}+k_{pl}F_{tpl}’+p_c A_g b l_k, $$
$$ F_t’=k_{ch}\frac{u_{ch}v_w a’}{v_s}+F_{tpl}’+\mu p_c A_g b l_k. $$
For generating grinding, I integrate these unit-width forces along the instantaneous contact line:
$$ F_{gn}=\int_L F_n’\,ds, $$
$$ F_{gt}=\int_L F_t’\,ds. $$
For form grinding, I count the active grains on the inner and outer flanks. If the total grain numbers are \(N_1\) and \(N_2\), the active numbers are
$$ N_a=N_1\frac{\alpha}{2\pi}, $$
$$ N_b=N_2\frac{\alpha}{2\pi}, $$
where \(\alpha\) is the engagement angle. The inner flank force and the outer flank force are
$$ F_{an}=N_a F_n’,\quad F_{at}=N_a F_t’, $$
$$ F_{bn}=N_b F_n’,\quad F_{bt}=N_b F_t’. $$
I then combine the normal and tangential components to obtain the resultant force on each flank.
| Grinding method | Contact type | Force calculation | Dominant thermal feature |
|---|---|---|---|
| Generating grinding | Line contact along instantaneous contact line | Curve integral of unit-width force | Local hot spot near the root region |
| Form grinding | Area contact on the tooth flank | Sum of active grain forces | More uniform temperature over the flank |
Heat generation in gear cutting is closely related to the grinding force and the grinding speed. I assume that nearly all mechanical work is converted into heat, while a small fraction is stored as strain energy, kinetic energy of chips, and new surface energy. The total heat flux entering the contact zone is calculated from the grinding force and the contact area. For generating grinding, I write the heat flow as
$$ Q_1=\int_L\frac{(F_{gn}+F_{gt})v_s}{S l_z}\,ds, $$
where \(S\) is the contact area and \(l_z\) is the effective width in the depth direction. For form grinding, I use
$$ Q_2=\frac{F_{fa,b}v_s}{N l\pi}, $$
where \(F_{fa,b}\) denotes the force on the active flank, \(N\) is the number of active grains, and \(l\) is the single-grain contact length. These heat flow expressions are the input to my thermal model.
I use a moving heat source model to represent the thermal load. The classical moving line source solution is
$$ T(y,z)=\frac{q}{\pi k}\int_0^{l_c}\exp\left(-\frac{v(y-\xi)}{2\alpha}\right)K_0\left(\frac{v\sqrt{(y-\xi)^2+z^2}}{2\alpha}\right)d\xi, $$
where \(q\) is the heat flux density, \(k\) is the thermal conductivity, \(\alpha\) is the thermal diffusivity, \(v\) is the source speed, and \(K_0\) is the modified Bessel function of the second kind of order zero. I adopt a triangular heat source distribution because the undeformed chip thickness varies from zero at the entry to a maximum near the middle and then decreases toward the exit. In my gear cutting model, the triangular distribution gives a more realistic temperature gradient than a uniform rectangular distribution.
Heat partitioning is another key part of my model. The heat generated in gear cutting is distributed among the workpiece, the wheel, the chips, and the coolant. I write the energy balance as
$$ Q_t=q_w+q_s+q_f+q_{ch}. $$
If each part is represented by an equivalent heat transfer coefficient and the maximum contact temperature is \(T_{\max}\), then
$$ Q_t=h_wT_{\max}+h_sT_{\max}+h_fT_{\max}+h_{ch}T_{\max}. $$
The workpiece heat transfer coefficient is approximated as
$$ h_w=\frac{\sqrt{k_w\rho_w c_w v_w}}{C\sqrt{l_g}}, $$
where \(k_w\), \(\rho_w\), and \(c_w\) are the thermal conductivity, density, and specific heat of the workpiece, \(v_w\) is the feed speed, \(C\) is a surface temperature coefficient, and \(l_g\) is the contact length. The energy carried away by chips is estimated from the specific grinding energy:
$$ e_{ch}=p_w c_w T_{mp}, $$
where \(T_{mp}\) is the melting point or a critical softening temperature. The chip heat flux is
$$ q_{ch}=\frac{z v_s}{S}p_w c_w T_{mp}. $$
For the wheel-coolant composite, I use the partition ratio
$$ R_w=\left[1+\frac{v_s(k\rho c)_s}{v_w(k\rho c)_w}\right]^{-1}, $$
where the subscript \(s\) denotes the wheel-coolant composite. For wet grinding, the heat flux entering the workpiece and the coolant is
$$ q_{wf}=(q_w+q_f)=(Q_t-q_{ch})R_w. $$
For dry grinding, the heat flux entering the workpiece is
$$ q_w=R_wQ_t. $$
These partition expressions allow me to estimate how much energy remains in the tooth surface and how much is removed by the coolant. In my gear cutting analysis, wet grinding significantly reduces the maximum tooth surface temperature because the coolant removes a large portion of the heat during each pass.
| Material or medium | Thermal conductivity \(k\) (W/m·°C) | Density \(\rho\) (kg/m³) | Specific heat \(c\) (J/kg·°C) | Thermal contact coefficient (J/m²·s⁰·⁵·°C) |
|---|---|---|---|---|
| Water-based coolant | 0.68 | 1000 | 4180 | 1686 |
| 20CrMnTi workpiece | 40 | 7900 | 460 | 12057 |
| Al₂O₃ wheel | 35 | 3980 | 765 | 10300 |
I solve the transient temperature field using the finite element method. The governing heat conduction equation is
$$ \rho c\frac{\partial T}{\partial t}=\frac{\partial}{\partial x}\left(k_x\frac{\partial T}{\partial x}\right)+\frac{\partial}{\partial y}\left(k_y\frac{\partial T}{\partial y}\right)+\frac{\partial}{\partial z}\left(k_z\frac{\partial T}{\partial z}\right)+q_v, $$
where \(q_v\) is the volumetric heat source. The boundary conditions are classified as follows. The first kind specifies the surface temperature:
$$ T=T_1. $$
The second kind specifies the heat flux:
$$ k_x\frac{\partial T}{\partial x}n_x+k_y\frac{\partial T}{\partial y}n_y+k_z\frac{\partial T}{\partial z}n_z=q. $$
The third kind specifies convection:
$$ k_x\frac{\partial T}{\partial x}n_x+k_y\frac{\partial T}{\partial y}n_y+k_z\frac{\partial T}{\partial z}n_z=\alpha(T_f-T), $$
where \(\alpha\) is the convection coefficient and \(T_f\) is the fluid temperature. In my model, the grinding contact zone uses the second boundary condition, while the remaining tooth surfaces use the third boundary condition. For wet grinding, the contact zone also has a convection condition, so I create a duplicate surface to apply the heat flux and the convection separately.
After discretizing the gear body into finite elements, I write the nodal temperature vector as \(\mathbf{T}\). The transient finite element equation is
$$ \mathbf{C}\dot{\mathbf{T}}+\mathbf{K}\mathbf{T}=\mathbf{P}, $$
where \(\mathbf{C}\) is the heat capacity matrix, \(\mathbf{K}\) is the thermal conductivity matrix, and \(\mathbf{P}\) is the thermal load vector. The element matrices are
$$ C_{ij}^e=\int_\Omega \rho c N_iN_j\,d\Omega, $$
$$ K_{ij}^e=\int_\Omega\left(k_x\frac{\partial N_i}{\partial x}\frac{\partial N_j}{\partial x}+k_y\frac{\partial N_i}{\partial y}\frac{\partial N_j}{\partial y}+k_z\frac{\partial N_i}{\partial z}\frac{\partial N_j}{\partial z}\right)d\Omega, $$
$$ P_{qi}^e=\int_{\Gamma_2}qN_i\,d\Gamma, $$
$$ P_{Hi}^e=\int_{\Gamma_3}\alpha T_fN_i\,d\Gamma. $$
I divide the whole gear cutting process into load steps. Each load step represents a short interval of grinding. At each step, I apply the heat flux to the current contact region, apply convection to the other surfaces, solve the transient problem, and use the result as the initial condition for the next step. This procedure allows me to track the temperature history over the entire grinding cycle.
| Parameter | Large gear | Small gear |
|---|---|---|
| Number of teeth | 41 | 11 |
| Module | 4.024 mm | 4.024 mm |
| Shaft angle | 90° | 90° |
| Offset | 23.00 mm | 23.00 mm |
| Face width | 26.00 mm | 29.78 mm |
| Hand | Left | Right |
| Pressure angle sum | 38° | 38° |
| Grinding method | Form grinding | Generating grinding |
I use a workpiece material whose thermal properties vary with temperature. The variation is included in my finite element analysis because the thermal conductivity decreases and the specific heat increases as the temperature rises. This temperature dependence improves the accuracy of the predicted temperature field in gear cutting. The material property table used in my calculation is summarized below.
| Temperature | Thermal conductivity (W/m·°C) | Specific heat (J/kg·°C) |
|---|---|---|
| Room temperature | — | — |
| 200 °C | 41.4 | 569 |
| 300 °C | 40.5 | 611 |
| 400 °C | 38.6 | 657 |
| 500 °C | 36.2 | 712 |
For generating grinding, I set the wheel speed, feed speed, and grinding depth according to a typical finishing schedule. The heat flux is applied along the instantaneous contact line, and the convection coefficient is applied to the remaining surfaces. For form grinding, I set a slower feed speed and a larger depth per pass because the wheel profile contacts the whole flank. I also include the effect of the eccentric spindle motion, which alternately exposes the concave and convex flanks to the coolant. The grinding parameters used in my calculation are listed in the following tables.
| Parameter | Generating grinding | Form grinding |
|---|---|---|
| Wheel linear speed | 31.4 m/s | 31.4 m/s |
| Feed speed | 1000 mm/min | 20 mm/min |
| Grinding depth | 0.05 mm | 0.1 mm |
| Coolant | Water-based fluid | Water-based fluid |
When I examine the transient temperature field, I observe that the heat-affected zone is not symmetric. In generating grinding, the hot spot tends to appear near the root region of the tooth surface. This occurs because the contact line is closer to the root at certain stages, and the heat cannot dissipate quickly from the narrow root area. Even though the interference volume near the root is not the largest, the local geometry limits heat conduction into the bulk material. As a result, heat accumulates there, and the maximum temperature is higher than in other regions. In gear cutting by generating grinding, this root-near region is therefore the most likely location for burn defects.
In form grinding, the temperature distribution is more uniform over the tooth surface. The wheel contacts a larger area, and the eccentric spindle motion helps the coolant reach both flanks. The maximum temperature is lower than in dry generating grinding under comparable conditions, but the temperature remains high over a wider area. The convex flank and the concave flank show slightly different temperature histories because their pressure angles and local contact lengths differ. The concave flank often displays two local maxima near its edges, while the convex flank is more evenly heated.
I use a burn criterion to identify the risk zone. For the material considered in my study, the critical temperature is approximately 600 °C. Regions above this value are treated as potential burn zones. In generating grinding, the burn zone is an irregular elliptical region near the root. In form grinding, the burn zone is less localized, and dry grinding can cause the entire flank to approach or exceed the critical value. These predictions are consistent with general experimental observations of gear cutting burn patterns.
| Grinding method | Cooling | Hot spot location | Burn risk |
|---|---|---|---|
| Generating | Dry | Near the root of the tooth | High, localized |
| Generating | Wet | Near the root of the tooth | Moderate, reduced area |
| Form | Dry | Whole flank, especially concave edges | High, distributed |
| Form | Wet | Whole flank, less severe | Low to moderate |
I also study the influence of grinding depth. For generating grinding, the maximum temperature increases nonlinearly with grinding depth. When the depth increases, the contact area grows, the material removal rate rises, and the grinding force increases. The heat flux therefore increases, and the temperature rises. In my calculation, the temperature growth becomes steeper when the grinding depth exceeds a certain value. This indicates that the fitted surface contact model has a limited range of validity. For dry grinding, a depth above approximately 0.1 mm may cause the maximum temperature to exceed the burn threshold. For wet grinding, the threshold is delayed, but a depth above approximately 0.13 mm still presents a risk. For this reason, I recommend a rough-then-finish strategy. A rough pass may use a depth near 0.05 mm, while finish passes should use 0.01–0.03 mm.
| Grinding depth | Generating dry maximum temperature | Generating wet maximum temperature | Recommendation |
|---|---|---|---|
| Small | Low | Lower | Safe for finishing |
| Medium | Moderate | Moderate | Use with coolant and reduced depth |
| Large | High | High | Avoid or split into multiple passes |
For form grinding, the maximum temperature also increases with grinding depth, but the increase is slower when the depth is small. The whole flank participates in the contact, so the local pressure does not rise as sharply as in generating grinding. However, when the depth becomes large, the chip space and coolant access are reduced, and heat accumulation becomes more severe. I therefore recommend that form grinding should also use a rough-then-finish sequence. A rough depth near 0.03 mm and a finish depth of 0.01–0.02 mm are reasonable in my model.
The wheel linear speed has a different effect. In generating grinding, I find that increasing the wheel speed slightly reduces the maximum temperature. This happens because the grinding speed increases, the average undeformed chip thickness decreases, and the specific grinding energy decreases. The number of active grains per unit time increases, but the force per grain decreases. The friction coefficient also tends to decrease as the relative sliding speed increases. As a result, less heat is generated per unit volume of removed material, and the maximum temperature decreases. The improvement becomes smaller at very high speeds.
In form grinding, dry grinding shows a decrease and then a plateau as the wheel speed increases. At higher speeds, the contact time becomes shorter, but boundary friction effects appear, and the temperature no longer decreases noticeably. In wet grinding, the temperature continues to decrease with increasing wheel speed because the coolant is transported through the contact zone more effectively and the chips are ejected more quickly. This behavior is important in gear cutting because a higher wheel speed can improve surface quality if the machine and wheel are stable.
| Parameter | Increase in parameter | Effect on maximum temperature | Main reason |
|---|---|---|---|
| Grinding depth | Increase | Temperature increases | Higher removal rate and larger contact area |
| Wheel linear speed | Increase | Temperature decreases or plateaus | Smaller chip thickness and shorter contact time |
| Feed speed in generating grinding | Increase | Temperature decreases slightly | Shorter heat exposure and faster source motion |
| Feed speed in form grinding | Increase | Temperature increases | Larger effective depth per contact |
| Grain size | Too fine or too coarse | Temperature increases | Loading or higher single-grain force |
| Eccentric spindle speed | Increase | Temperature increases | Less time for coolant to remove heat |
I also analyze the feed speed. In generating grinding, a higher feed speed reduces the time during which the heat source acts on a given tooth surface point. The source moves away more quickly, and the coolant has less time to be heated by the same point. Therefore, the maximum temperature decreases slightly, although the temperature peak may occur earlier in the process. This suggests that increasing the feed speed can be beneficial if the machine dynamics and surface finish remain acceptable. However, the feed speed must not be increased without limit, because a very high feed speed can cause impact loading and wheel wear.
In form grinding, the feed speed along the tooth height has the opposite effect. A higher feed speed produces a larger effective depth per contact, so the grinding force and heat flux increase. The coolant also has less time to remove heat from the bottom of the tooth space. As a result, the maximum temperature increases with feed speed. I therefore recommend a moderate feed speed for form grinding, especially when the tooth space is narrow and the coolant access is limited.
The grain size and wheel topography also influence the temperature field. If the grain count per unit area is too high, the wheel becomes dense and may load, which reduces chip clearance and raises the temperature. If the grain count is too low, the individual grains are coarse, the single-grain contact length and width increase, and the grinding force rises. Both extremes are unfavorable. In my gear cutting analysis, a wheel with a moderate grain density and good porosity gives the best balance between material removal and heat dissipation. I also find that a suitable eccentric spindle speed is important. If the eccentric spindle rotates too quickly, the interval between successive contacts becomes too short for the coolant to remove the residual heat, and the temperature rises. A moderate eccentric speed improves coolant access and reduces the risk of burn.
I summarize my main findings as follows. The thermal model that I developed combines gear cutting geometry, contact zone analysis, force prediction, heat partitioning, and finite element computation. It can predict the transient temperature field for both generating grinding and form grinding. The predicted hot spot in generating grinding is near the root, while form grinding produces a more uniform temperature distribution. The burn risk is strongly affected by grinding depth, wheel speed, feed speed, grain size, and eccentric spindle speed. Wet grinding is much more effective than dry grinding in reducing the maximum temperature. A rough-then-finish strategy with controlled depth and adequate coolant is recommended for both methods. These conclusions provide a basis for optimizing gear cutting processes and improving the surface integrity of spiral bevel gears.
For future work, I intend to extend the contact model to larger grinding depths, where the second-order fitted surface may no longer be sufficient. I also plan to develop a three-dimensional force model that better represents the actual grain trajectories in gear cutting. A fully analytical thermal solution could complement the finite element method and reduce computation time. Finally, I will perform dedicated experiments to measure the grinding temperature and validate the model under a wider range of gear cutting conditions. Such validation will further strengthen the reliability of the proposed method and support its application in precision gear manufacturing.