Table of Contents Session: Aerospace

Detection of Thermal Damage IN Steel Components After Grinding Using the Magnetic Barkhausen Noise Method A.S. Wojtas*, L. Suominen; B.A Shaw, J.T. Evans; Design Unit / Materials Division, The University of Newcastle, UK. *Corresponding Author Contact: Stresstech - AST, Schwalbach, Germany ; ASWOJTAS@compuserve.com

TABLE OF CONTENTS

Summary Introduction Magnetic Barkhausen Noise Method Conventional Inspection Methods Instrumentation Applications Conclusion Acknowledgements References

1. Summary

Bearings, gears, shafts and many other surface hardened components in modern automotive and aerospace industries form a challenge to design and production engineers. They require special attention in choosing the correct parameters for the heat treatment as well as for the subsequent machining processes. The latter, if carried out inaccurately, may reduce the surface hardness and diminish the favourable compressive surface stresses after surface hardening. Accurate and continuous control of machining processes such as grinding or hard turning is essential in today's production control. An inspection method based on the analysis of the magnetic Barkhausen noise (BNA) is capable of 100% in-line control of many types of machine components. Proposed by a Finnish - American company, this method has been shown to be more sensitive than the conventional acid etch inspection. It is also superior to the alternative method using eddy current. The sensitivity of the Barkhausen noise method to minute changes in the level of surface residual stress allows detection of grinding damage at its onset. This paper discusses the possibilities of the industrial application of the BNA method and illustrates it with two practical cases.

2. Introduction

Grinding is used in the manufacture of high accuracy components to achieve the required tolerance and compared with other machining processes, grinding requires a very large energy input per unit volume of material removed. The majority of this energy is converted to heat, which is concentrated in the surface layers of the material, within the grinding zone and as a result of this, a rapid rise in the localised temperature within the surface can occur. The actual rise in temperature depends on a range of factors, including the type of coolant, method of coolant supply, type of grinding wheel and the speed and depth of cut of the wheel. The heat generated in the process and plastic deformation in the surface layer of the part will produce a considerable amount of residual mechanical stress. Heat treatment and e.g. shot peening induce compressive residual stresses in the substrate surface. Presence of a compressive layer is desired to achieve optimal fatigue strength and wear resistance. Grinding and turning are known to often reduce those compressive stresses and sometimes even to turn them into tensile thus representing a risk for early material failure in critical applications such as bearings, axles, shafts, gears etc. The residual stress distribution generated by grinding under ideal conditions i.e. with no heat involved will leave a residual compressive stress layer (Figure 1).

During grinding of hardened and tempered steel samples, the production rate, i.e. the stock removal rate, is limited by the increasing risk of thermal damage to the component. The severity of such a damage, also known as grinding burn, will depend on the temperature to which the work piece surface was heated.

A temperature increase to below ca. 500° C will usually result in a partial relaxation of the earlier mentioned compressive stresses. Though no noticeable change in microstructure will take place in this case, residual stress measurement by e.g. X-ray diffraction will reveal a decrease in the surface stress. In some critical applications e.g. in the aerospace industry this may be a reason to reject such a part. A temperature increase above the original tempering temperature, usually near 600° C, will result in a B class thermal damage, also called a temper burn. (Figure 2.) Resulting overtempering of the material will have as an effect a tangible decrease in the surface hardness as well as development of residual tensile stress. This type of a surface defect can be usually detected using both residual stress and microhardness measurement as well as through a chemical inspection i.e. by etching the surface in Nital etchant (a 4% solution of nitric acid in alcohol).

A further temperature increase, well into the austenite temperature range, above 720° C and subsequent immediate quenching by the coolant used in the grinding process, will cause a D class thermal damage, also called a rehardening burn. This type of surface defect is characterised by an unacceptably hard and brittle layer of untempered martensite. Spots of a D class burn are always surrounded by areas of B class burn. A residual stress profile across a rehardening burn is complex and exhibits both compressive as well as highly tensile stress zones. Since the level of the residual surface stresses after grinding has a detrimental effect on the fatigue strength it is important that any of the above-described defects are detected before the component is placed in service. The method used in the tests described, advanced and promoted by a Finnish - American group, Stresstech AST, employs Barkhausen noise Analysis to evaluate the integrity of the ground steel surfaces.

3. Magnetic Barkhausen Noise Method

The BN method is based on inductively detecting a noise like signal generated in ferromagnetic materials subjected to alternating external magnetic field. Magnetic domain walls forced to move back and forth by the exciting magnetic field, will pin to crystal lattice defects and dislocations only to jump further once forced by sufficient excessive energy of the magnetising field. Each such jump will generate a pulse in a coil placed near the surface of the sample. Pulses generated in a finite volume of material close to the coil will add to a noiselike signal called Magnetic Barkhausen Noise. The intensity of this signal, BNA, is further used to characterise the integrity of the stress state and the microstructure of the ground surface.

The effective voltage (RMS) of the generated signal, also referred to as MP (Magnetoelastic Parameter), is a strong function of the microstructure and the residual stress state of a material. This makes the Barkhausen noise technique ideal for detecting grinding abuse since the onset of grinding damage will result in a decrease in hardness and a more tensile residual stress which will both result in an increase in the Barkhausen noise level.

Both residual stress level and the hardness of the specimen will affect the BNA. In materials with positive magnetic anisotropy (iron, most steels, cobalt) the intensity of the signal will raise with increasing severity of a grinding damage. Presence of residual stresses will affect the BNA in a way shown in the figure 3. A sample under high compressive residual stress will generate a low intensity of the Barkhausen noise (low MP). With the stress condition changing from high compressive to less compressive, then slightly tensile and highly tensile, the measured MP will show a continuous increase. Although towards the elasticity limits the curve will tend to deflect, the main part of it will directly and proportionally reflect on the level of the residual stress in the sample.

Microstructure, i.e. the hardness of the sample, will similarly affect the measured MP level. A hard sample will generate a low intensity of Barkhausen noise (low MP). With decreasing hardness the MP will continuously increase as shown in figure 4. Intentionally, the graph is shown as a mirror image i.e. with the hardness data in a reversed order - left to right. It can be seen that the MP curve is a function of both the material characteristics described earlier, i.e. residual stress and hardness, have a similar increasing trend. Thus, it confirms the experimental data showing that a component yielding low MP values has high hardness and shows compressive residual stresses in the surface. Should the MP level increase it will indicate a decrease in hardness or an unfavourable alteration of the surface stress, as shown in figure 5. Examples of two practical applications of this property of the BN method will be shown in section 6.

4.Conventional Inspection Methods

At this point an overview of the common inspection methods is offered to compare their advantages and disadvantages.
Table 1. Methods for detection of grinding abuse.

Feature	Nital Etch	Eddy current	X-ray diffr.	BNA
Non-destructive	No	Yes	Yes	Yes
Standardised	Yes	Yes	Yes	Yes
Use of Chemicals	Yes	No	No	No
Automated	No	Yes	Yes	Yes
Objective	No	Yes	Yes	Yes
Quantitative	No	Yes	Yes	Yes
Reliable	No	Yes	Yes	Yes
Evaluation Through Coatings	No	Yes	No	Yes
Danger of Hydrogen Embrittlement	Yes	No	No	No
Influenced by Both Stress and Microstructure	No	No	Yes	Yes
Fast	No	Yes	No	Yes
Can Easily Examine Large Areas	Yes	Yes	No	Yes

The chemical inspection method mentioned earlier was until recently the most commonly used due to its relatively inexpensive set-up(*).
(*) Recently introduced environmental regulations restrict the use of the acid etch method making it very expensive in both the initial investment in the installation as well as in the acquisition and disposal of the hazardous waste.

Although this method may be practical for low-volume inspection, it can not be used for 100 percent inspection of high volume production. It is time consuming, can not be automated and suffers from subjective determination of the surface quality. Inspected surfaces, attacked by acid etch, have to be additionally treated and often require further heat treatment to avoid hydrogen embrittlement.

The eddy current method is an alternative to acid etch technique and solves some of its shortcomings. However, similar to Nital etch, it will reveal a grinding burn only if a metallurgical transformation has taken place. Insensitive to the residual stresses, it can be used to indicate softening of the surface layer due to thermal damage but will not reveal critical tensile stresses, which would lead to pitting or cracking. Both the Nital etch and eddy current techniques, are routine methods for finding burns, but can be misleading if the burns have been 'removed' by a final grind since some damage can still remain even though the etch does not reveal this.

X-ray residual stress measurement is the most accurate and quantitative method but it is rather slow and expensive. Also, the assessment of in-depth stress profiles is possible only by successive removal of the top layer e.g. by electropolishing. As such, the method is not non-destructive and can be quite time consuming.

Finally, there is the BN method, which is strongly gaining ground as a reliable, fast and practical method capable of testing complex shapes and high production volumes. The BN method has already been endorsed by the American Society of Automotive Engineers and the Federal Administration of Aviation. In a relatively short period of time since its endorsement many companies in the automotive and aerospace industry have introduced it and many have already included it in their specifications and quality assurance standards. Its sensitivity to both stresses and hardness, combined with its significant penetration depth, enables it to detect grinding burns that may have been partially 'removed' in finish grinding and also allows measurements through coatings, e.g. chromium plating.

5. Instrumentation

Measurements were made using analysers developed and manufactured by Stresstech AST A laboratory unit Micro-Scan 500 and its industrial version Rollscan 200 were used for all BN measurements. A block diagram of a Rollscan is shown in figure 6. In its default mode the magnetising signal has a sinusoidal shape and a frequency of 125 Hz. The BN signal, picked-up by the sensing coil has a frequency spectrum stretching from the magnetising frequency to several MHz. While the Microscan 500 is capable of processing the full spectrum of the signal, the Rollscan 200 unit is equipped with a band filter that will pass only the frequency range from 70 to 200 kHz. This part of the BN spectrum has been found to contain the most information pertaining to the presence of thermal defects in the surface of the ground component. The relatively high frequency of magnetising and a signal-processing algorithm allow measurement in real time. Larger surfaces and complex shapes can be scanned using a variety of sensors. XRD diffraction measurements were made using the X2001 unit - an X-ray diffraction stress analyser developed and built by Stresstech AST.

6. Applications

1. Case 1. A large industrial double helical gear for use in a power generation plant had suffered thermal damage during the first stage of grinding. Nital inspection revealed the presence of severe grinding burns on 10 adjacent right hand helix teeth out of a total of 119. After finish grinding the etch test was repeated and did not show any traces of thermal damage any more. For verification a BN inspection was carried out using a Rollscan device. All the teeth, on both left and right hand flanks, were inspected, at three points (outside, centre and apex) across the flank. The data obtained have revealed the presence of suspected high tensile residual stresses.

Their location, i.e. the numbers of the damaged teeth, was also compared with the prior Nital inspection report and was found to be the same. The MP data (Barkhausen Noise Number) obtained from both left and right flanks are shown in figures 7 and 8 respectively. On the right hand helix, tooth number 25 and the neighbouring ones yielded MP values more than double the average level measured on the remaining teeth. The large change in BN numbers and the consistency in the other measurements allowed the potential damage to be easily identified and corrected for through remedial action.

2. Case 2.
It is believed that changes in the residual stress profile caused by grinding damage are responsible for many gear failures accompanied by both micro and macro-pitting (ref. 3.)

Sample gears, intentionally micro-pitted, were manufactured specifically to investigate this problem. A consistent depth of 5 - 10 µm of micro-pitting, across the full face width of the gear, was produced on each test gear. Gears with both light and severe levels of grinding damage were produced after shot peening. In terms of the surface temper etch process, the light burn corresponds with Class B damage and the severe burn Class D. This was also identified by measuring the Barkhausen Noise level. Values for the maximum residual stress are shown in Table 3.

Burning was achieved by slowing down the stroke rate of the grinding wheel compared to normal grinding conditions. A wheel rate of 100 strokes per minute, with a 100µm cut and coolant on was used to produce Class B burns. Class D burns were produced by turning the coolant off and increasing the depth of cut to 200µm.

The gears were run on the back-to-back test rig using the same procedure to produce micro-pitting. From these tests, it was found that the time required to produce a consistent level of micro-pitting (as previously described) was greatly dependent on the level of grinding damage on the flank (see Table 2).

Table 2 clearly shows that the gears with grinding burns micro-pit much quicker than the gears without grinding burns. There can also be seen to be a difference between the gears with different levels of grinding damage. The Class B gear with moderate damage, produced micro-pitting in just under half the time of the unburned gear. The Class D, severely damaged gear, however, produced a dramatic difference in the rate of micro-pitting and took less than a tenth of the time to micro-pit when compared to the undamaged gear. This effect is not fully understood at present, but must be due to a combination of the change in the residual stress level along with a change in the near-surface microstructure which will result in a softer surface layer with a more tensile residual stress level. It must also be remembered that the undamaged gear was in an as shot peened condition which would have changed the surface texture as well as introducing highly compressive surface residual stress levels. Further work is therefore required to distinguish between the effect of shot peening and the level of grinding damage on the formation of micro-pits, but it is clear that the level of grinding damage in itself is very important.

Table 2 Time to produce 5-10µm of micro-pitting.

	Class A	Class B	Class D
Time (10 6cycles)	8.3	4.1	0.75

Table 3 Comparison of methods for detecting grinding burns.

	Nital Etch	Barkhausen Noise	Residual Stress
			Surface	Sub
No Burns	Class A	20	-700Mpa	-850Mpa
Light burns	Class B	45	200Mpa	650Mpa
Heavy Burns	Class D	70	400Mpa	1050Mpa

7. Conclusions

1. A very good correlation is found between the Barkhausen Noise Level and the degree of thermal damage in ground steel workpieces.
2. Tests made in the laboratory were repeated in a production environment and showed that the test method is suitable for heavy-duty industrial applications.
3. The high sensitivity of the method allows detection of grinding abuse missed by other conventionally used methods of nondestructive inspection.

8. Acknowledgements

BAS would like to gratefully acknowledge the support of the Engineering and Physical Sciences Research Council. The MoD(N) are also thanked for their support and interest in the gearing development work at Newcastle. ASW would like to thank Cornelia for her patience and ample supply of fresh brewed coffee.

References

1. R.M. Fix, K. Tiitto, S. Tiitto, " Automated Control of Camshaft Grinding Process by Barkhausen Noise", Materials Evaluation, Vol. 48, No. 7., pp. 904-908 The American Society for Non-destructive Testing, Inc.
2. L. Suominen, A.S. Wojtas, "A comparative Study of Barkhausen Noise and X-Ray Diffraction Measurements from Ground Ferritic Steel Samples" Conf. Proc., The Fourth European Conference on Residual Stress, June 1996, Cluny, France.
3. M.F. Rose, F.S. Morgan, B.A. Shaw, "The Effect of Grinding Burns on Micro-pitting and Surface Initiated Bending Failure" Department of Mechanical, Materials and Manufacturing Engineering, University of Newcastle upon Tyne, NE1 7RU, U.K.
4. W.P. Ogden, "Detection of Subsurface Tensile Stress in an Aircraft Engine Mainshaft Bearing using Barkhausen Noise" Conf. Proc., Practical Applications of Residual Stress Technology, May 15-17 1991, Indianapolis, USA, pp 161-167
5. Standards BS 7862:1996 and ISO 14104 : 1995 "Gears - Surface Temper Etch Inspection After Grinding"
6. B.A. Shaw, and J.T. Evans, 1996, The Influence of Grinding on the Formation of Residual Stress Distributions in Ground Surfaces, Drives and Controls 1996, Session 4, Kamtech Publishing Ltd., 21-26.
7. M.F. Rose, F.S. Morgan, B.A. Shaw, "The Effect of Grinding Burns on Micro-pitting and Surface Initiated Bending Failure" BGA / GRC Gear Research.

|Top|