Nozzle design and coolant placement can improve the quality of ground workpieces
Grinding is a thermally dominated process. If done incorrectly, it can lead to surface damage to the work material, and unsatisfactory process economics due to inadequate removal rates and/or excessive wheel wear. Power consumed by the process flows into the wheel, work, chip, and coolant. The energy that enters the workpiece must be removed quickly to prevent high local temperatures and phase transformations from developing, and to prevent high residual temperatures after the wheel has passed by. Phase transformations are often responsible for tensile residual stresses, white layer formation, reduced fatigue life, cobalt leaching, and surface and subsurface cracking. Cooling is achieved by the application of a cooling and lubricating fluid, as well as by selecting process parameters that reduce heat generation.
Regardless of how important optimum coolant application is known to be, during walkabouts around recent machine tool shows in Chicago and Hannover most of the grinding machines that I saw were fitted with either bendable plastic nozzles or open metal-tube nozzles. In many cases, these nozzles are afterthoughts that allow highly sophisticated grinding machines to produce chips without requiring a great deal of investment, thus helping the OEM or distributor sell a machine. However, the manufacturer and end-user will not realize the best economics, or achieve the best part quality, by employing these nozzles. In essence, a machine costing $500,000, fitted with a $75,000 filtration system and a $10,000 CBN wheel, is being compromised by a $100 nozzle system.
Although the function of the grinding coolant is primarily to cool the process, it also has to cool the grinding wheel (very important with resin-bonded diamond wheels), allow the lubricants in the coolant to do their job (especially important with single-layer superabrasive wheels), flush the chips from the machine and workpiece area, and clean the wheel (especially when grinding ductile materials while using water-based coolant). The following benefits can often be realized by improved coolant application:
- Reduced dressing frequency, due to less loading with work material and reduced abrasive grain wear.
- Thermal damage of the workpiece material is reduced, allowing higher removal rates.
- More of the applied flow rate will be effective, allowing the overall applied flow rate to be reduced.
- Reduction in entrained air (foaming), misting, and vapor problems.
- Reduced disturbance of the jet from the air barrier surrounding the wheel.
- More robust and stiffer setup.
- No need to reduce wheelspeed or soften wheel grade to alleviate burn.
Over the last 16 years, I have championed coherent-jet coolant nozzles, and advocated a philosophy of how to use them most effectively. The pressure, flow rate, temperature, and direction of the jet all influence the fluid’s cooling ability. Pressure controls the velocity of the fluid; the flow rate and temperature control the rate of heat transfer into the fluid. The direction of the flow allows the fluid to remove the air-barrier that travels with the wheel.
Many researchers have studied the role of grinding fluids in preventing thermal damage to the workpiece over the past 30 years. One researcher found that when the workpiece surface temperature is below the boiling point of the fluid, a liquid phase will be present in the grinding arc. As the temperature of the surface increases, the liquid will undergo nucleate boiling, which can enhance convective cooling. As the fluid temperature increases still further, a vapor film will form that blankets the heated workpiece surface, suppressing further heat transfer. This film-boiling threshold temperature is around 130°C for a water-based fluid. The study defined the critical power flux at “burnout” to be approximately 30 W/mm2in the contact area between an aluminium oxide grinding wheel and the workpiece during continuous-dressing creep-feed grinding (CDCF).
Another researcher obtained values greater than 200 W/m2 for High Efficiency Deep Grinding (HEDG) with CBN grinding wheels, in oil coolant, at 10 times the material removal rate observed in the CDCF study.
Other research showed that specific grinding energy decreased to a minimum of around 5 gpm per inch (0.75 L/min/mm) of grinding width with electroplated CBN grinding of 52100 steel, much less than CDCF would use, and that additional flow rate made little difference. Investigation of the effect of flow rate and nozzle aperture on residual stresses revealed lower tensile stresses with the smallest aperture, due to higher jet speed produced. For grinding 52100 steel with vitrified CBN wheels, 25 gpm per inch (3.7 L/min/mm) was adequate.
Much work has been published on jet nozzles, shoe nozzles, and chambers, with regard to their overall grinding performance. Shoe nozzles have been shown to be extremely effective, but in practice are difficult to keep sealed against the wheel surface as the wheel diameter reduces with wear (especially true with CDCF grinding). This is one of the reasons why jet nozzles are so popular for research and on the shop floor. These nozzles are often designed to produce a coherent jet that allows them to be placed at a convenient distance from the grinding arc.
Several experimental tools have been developed to evaluate the effect of nozzle application parameters on the transport of fluid through the grinding arc. Researchers have measured the pressure in the grinding arc (because the pressure determines the fluid boiling point and the degree of penetration of fluid into the wheel structure). They have found that a coherent-jet nozzle with jet velocity matched to the wheel peripheral speed provides the best cooling of a simulated grinding process.
Original simulations have been developed and validated by Laser Doppler Anemometry methods for study of flow in the wheel peripheral boundary layer by Liverpool JM University (UK). This is the first work to investigate flow phenomenon in each of the tangential, axial, and radial directions. Simulations have been developed to analyse the effect of scrapers on boundary flows.
For silicon nitride grinding with resin-bonded diamond wheels, the Technical University of Berlin (Germany) found that only 0.4% of the 25 gpm per inch (3.7 L/min/mm) applied was effective. Using side-shields to separate the flow going through the grinding zone from the coolant going around the grinding zone, and dividing by the applied flow rate, the University of Massachusetts measured an effective flow rate of 26–55% for creepfeed grinding with high-porosity aluminium oxide wheels. One researcher quoted an effective flow rate of 3% in high efficiency deep grinding (HEDG) with electroplated CBN grinding of steel. Another achieved 9.6% with similar wheels, grinding nickel-based alloys. Research is currently being conducted in this area at the University of Bremen (Germany) and Liverpool JM University (UK).
Many researchers have demonstrated that the boundary layer of air that surrounds the grinding wheel, traveling at the same speed as the periphery of the wheel, can disrupt the flow of coolant into the grinding zone. The depth of the air barrier depends on the grit size and porosity of the wheel, and is difficult to estimate. The energy associated with the air barrier should not be underestimated. Many researchers agree that matching the wheelspeed with the jet speed can be an effective procedure for dealing with the air barrier.
Recent work using high-speed photography revealed that low-pressure flood coolant was easily deflected by the air barrier, especially when the nozzle jet was aimed directly into the “V” of the grinding zone. The air was shown to remain with the wheel and enter the grinding zone, and the coolant was reflected back. In the same experiment, the nozzle jet was aimed at the wheel a few millimeters ahead of the grinding zone, and the air was peeled off. Coolant attached to the wheel, and was transported into the grinding zone, evidenced by a “rooster tail” spray surrounding the wheel.
The air barrier can also be overcome by using rigid scrapers fitted close to the wheel surface, to reduce the thickness of the air. These units must be adjusted every time the wheel is dressed, and are therefore more convenient to use with nondressed, electroplated superabrasive wheels. With shoe nozzles, air can either be driven into the wheel surface or leakage from the back of the shoe, against the direction of the grinding wheel, can peel off the air.
Jet nozzles are the simplest type of nozzle for grinding. Bendable plastic nozzles and open-metal pipe nozzles fall into this category. The jet coming from the end of these nozzles can be very dispersed, leading to air entrainment. Other disadvantages include:
- Lack of stiffness when subjected to the reaction force of the jet.
- The nozzle must be placed very close to the grinding zone to offset the dispersed jet, risking damage by the wheel and the fixture.
- The nozzle is at the compression fitting holding the pipe, and once formed the jet is constrained by the tube until it exits
- The tubes are often of small diameter so they can be bendable, and the velocity of the coolant down the tube far exceeds the critical velocity for turbulence (20 fps or 6.1 m/sec).
Shoe nozzles are very effective when grinding with superabrasive wheels, when the diameter of the wheel changes little and good sealing of the shoe to the wheel can be maintained. One design uses a replaceable tip that is ground by the wheel to form a good seal. Megasonic nozzles show promise because they use very low fluid flow rates for medium stock-removal rates. Ultrasonic excitation of the fluid gives it greater energy, allowing it to travel a greater distance at much lower pressures and flow rates than coherent jet nozzles would require.
Coherent-jet nozzles use a precisely machined internal profile to produce the desired jet quality. Formed (usually crushed-tube) nozzles often include a stabilizer (planar) section after the transition from the input tube diameter to the output dimension (round or rectangular), to minimize break-up of the jet once it exits. Wedge nozzles are often fabricated from welded metal sheet, and taper down from the inlet pipe diameter to the desired exit aperture. With this design, the exiting jet can disperse at the same angle as the wedge, reducing cooling effectiveness.
An example of a coherent jet nozzle geometry is the Rouse profile. This design requires a contraction ratio (inlet to exit diameter ratio) of at least 2:1 to be coherent. A definition of acceptable coherency has been given previously as being a dispersion of 2–3x the exit diam, at a distance of 300 mm from the nozzle. In other words, a 6-mm diam jet would open out to 12–18 mm, at a distance of 300 mm from the nozzle end. This is often bettered by Rouse-profile nozzles.
Coherent-jet nozzles give better grinding performance because they concentrate the coolant into the grinding zone, allowing the pores of the wheel to be fully filled, and thereby minimize the applied flow rate. A dispersed jet pumps air and coolant into the grinding wheel, displacing some of the fluid with air. In some cases the dispersion is so high that much of the coolant hits the workpiece and deflects off. A coherent jet also concentrates its kinetic energy to break through the air barrier and wet the wheel.
Often nozzles are crushed to increase the pressure from the pump and, consequently, experience cooler grinding. In essence, they are increasing the velocity of the jet. The potential energy of the coolant in the nozzle is converted to kinetic energy in the form of a jet. While there is some kinetic energy due to the coolant velocity inside the nozzle, this is typically small compared to the potential energy. Bernoulli’s equation is used to determine the pressure-velocity relationship, and considers the density of the fluid. Since kinetic energy is based on the square of the velocity, to double the velocity the pressure must increase by four times. The simplified version of Bernoulli’s equation is:
where v is in m/sec, or
where v is in feet/minute
SG is coolant specific gravity
and v is jet velocity.
Some engineers adopt the “more pressure is better” approach to coolant application. Even when a nozzle pressure of 58 psi (406 kPa) is required to match a wheel speed of 6000 fpm (18,288 m/min), a pressure of over 200 psi (1400 kPa) may be used. Often the engineer is grinding stringy and ductile materials, such as medical stainless, Inconel, and titanium, and they are seeing the benefit of wheel cleaning as much as cooling. However, the general consensus is to use 600–800-psi (2–5.6MPa) coolant at 2 gpm per inch (0.3 L/min/mm) for bonded wheels, and up to 1500 psi (10.5 MPa) for single-layer wheels.
Several researchers have developed flow rate models based on a 4°C increase in temperature of the coolant exiting the machine over that entering the machine. Others use a fixed flow rate of 25 gpm per inch of grinding width. I use a flow rate based upon the grinding power created during the process, because the more aggressive a cycle is, the more coolant is applied. With conventional abrasive wheels, a flow rate of 2 gpm/hp is effective. For superabrasive wheels, a flow rate close to 1 gpm/hp (3.8 L/min/hp) works well. This flow rate must be applied directly into the grinding zone, with additional flow rate for cleaning and flushing nozzles. The machine controller may be used to determine the approximate grinding power, or a phase-corrected power meter can be used. Once the flow rate has been determined and the pressure decided, the nozzle aperture can be designed to give that flow rate.
Nozzle geometry, pressure, flow rate, nozzle aperture, and jet position have been discussed above. When developing the actual hardware, it’s important to consider wheel-wear compensation, adjustability, and easy reconfigurability, if you are grinding different profiles on the same machine. While formed nozzles have been produced using EDM for specific profiles, a combination of round and rectangular nozzles can often yield better results due to better coherency. Formed nozzles try to bend the flow around a complex profile and lead to flow separation at the corners. Pure round and rectangular nozzles do not suffer from this problem, but must be aimed correctly to blend around the profile.
Coherent Jets Can Assist Grinding
The figure on the left side shows a coolant nozzle setup where the original open pipes were very numerous, and in some cases were crushed to increase pressure. The setup included both a plated CBN wheel and a vitrified CBN wheel on the same spindle, grinding a nickel-based alloy. The pump supplying the mineral-oil coolant had a 50-hp (37-kW) motor, and was capable of 500 psi (3500 kPa) at a potential flow rate of 150 gpm (568 L/min). Pressure measured at the vitrified wheel manifold was 90 psi (620 ksi), and at the plated wheel manifold was 300 psi (2100 kPa). The pressure required to match the 12,000 fpm (3658 m/min) wheel speed is 250 psi (1750 kPa).
The combined area of all the nozzles on each zone was estimated to be 136 gpm (515 L/min) for the plated wheel and 106 gpm (401 L/min) for the vitrified CBN wheel. Optimization work primarily sought to reduce the flow rate of the machine to allow more machines to run simultaneously using the undersized central coolant system. The percentage spindle power load for the 30-hp (22.5-kW) grinding motor was typically 40–45% (13.4 hp or 10 kW) for the plated CBN wheel, and 60–70% (20 hp or 15 kW) with the vitrified CBN wheel. These values include the idle running power without grinding.
The revised coolant application setup on the right side was based around coherent-jet nozzle tubes. As can be seen in the photo, one 0.187″ (5-mm) and two 0.125″ (3-mm) diam nozzles were fitted to supply the center and each corner of the plated CBN wheel, and two 0.187″ (5-mm) diam nozzles were fitted to supply the vitrified CBN wheel. An additional nozzle was fitted underneath the vitrified wheel to lubricate it during dressing. Pump speed was adjusted to give 250 psi (1750 kPa) in the manifolds. Calculated flow rate from the nozzles was 31 gpm for the plated wheel and 39 gpm (148 L/min) for the vitrified wheel, a reduction of 71% for the plated wheel and a reduction of 63% for the vitrified wheel. Due to much lower viscous drag associated with the much lower flow rates, the spindle power decreased to 30% (8 hp or 6 kW) on the plated wheel, and 40% (12 hp or 9 kW) on the vitrified wheel, for no change in the cycle parameters. No high-pressure cleaning jets were deemed necessary since wheel loading was not evident, and oil coolant is less susceptible to this happening.