Ultrasonic testing of spot-welded joints on coated steel sheets and optimization of welding parameters

Transcription

1 Ultrasonic testing of spot-welded joints on coated steel sheets and optimization of welding parameters Richard Kaminski 1. Introduction The requirement for an improved corrosion prevention in today s automotive production has led to increased fabrication of galvanized steel sheets. This also applies in particular to the passenger car type Mondeo currently produced by Ford at Genk (Belgium). The nondestructive ultrasonic testing of spot welds has been extremely successfully applied by Ford all over Europe, and also in Taiwan, for many years now. It was at first only used as a supplement to the classical hammer-and-chisel test method. However, the ultrasonic test method has meantime proved to be indispensable, especially in connection with galvanized steel sheets, and in that case mainly in optimizing the parameters of welding machines for series production. 2. Problems with spot welds on galvanized steel sheets The zinc deposit has considerable effects on resistance spot welds: 1. Resistance welding requires a certain initial resistance value in order for the avalanche effect to take place. This is based on the fact that, along with the rising temperature of material its resistance increases, which on the other hand accelerates heating, so that Fig. 1: Welding range with three curves for heat build-up in the material with a different initial resistance 1

2 very soon the necessary temperature is reached for the local fusion of material and consequently for nugget formation (1400 C). However, the zinc deposit reduces the initial resistance resulting in a flatter heat build-up curve than is required for a good fusion. Fig. 1 shows the quality of the spot weld as a function of the heat-input rate curve: the curves A and B show good or adequate initial resistance values, the curve C represents an initial resistance which is too low. In the latter case, which applies to galvanized sheets, the curve may rise due to the increased welding current to such an extent that it already reaches the welding range at fairly short welding times. The upper range is limited by the spatter limit and the lower range by a too small resulting nugget diameter. However, higher welding currents lead to higher temperatures on the contact faces of the electrodes and, as a result of this, to a lower copper hardness. This, on the other hand, leads to a faster production of unacceptably small nuggets and, in borderline cases, even to cold welds. In addition, a softer copper material causes more frequent sticking of new electrodes being used making them hollow more quickly and enlarging their contact faces. 2. In the case of galvanized sheets, the burning of zinc deposit has an additional effect on the surface condition of the electrodes. A perfect resistance welding requires a constant current flow through the electrodes to the farthest possible degree. This is not only affected by the higher welding current, as already mentioned, but also most of all by the relatively low melting temperature of the zinc deposit: an electrolytical zinc coating melts at approx. 420 C. This means that as the sheet metal is heated to 1400 C, the zinc deposit burns at an early stage of the process and leads to a metal pickup on the copper electrodes, combined with an increased electrical resistance at these points. This phenomenon of metal pickup is quite irregular and generally at its strongest in the middle of the electrodes or in the hollow area. The changes in the contact resistance caused by this lead to a varying current flow which may result in irregular weld nuggets or even cause too small nugget diameters. - It has been noticed in series production that the metal pickup with subsequent hollowing of the electrodes is at its strongest on stationary welding machines because welding is always made at one and the same angle. It all looks better with robot welding devices: every welding position is individually programmed in this connection, the result being different gun positions per weld spot and thus different welding angles. This leads to less metal pickup and hollowing of the electrodes. 3. With advancing metal pickup and hollowing, the risk of ring welds becomes more likely, Fig. 2: Micrograph of a ring weld please see Figs. 2 and 3. Such ring welds are difficult to detect with sheet thicknesses below 1 mm using the classical hammer-andchisel method. In view of the subsequent paint coating, the sheets must not be battered too much during the test. The test force is therefore sometimes too low to destroy the ring weld. Additional zinc bonding often occurs in ring welds, i.e. at the outer nugget edge and within the ring weld, leading to an increased mechanical resistance which will prevent a definite and unambiguous flaw detection even by means of destructive testing. The faith in the reliability of the hammer-and-chisel method has considerably suffered due to these bonding effects and decisively lost significance in comparison with the ultrasonic testing. 4. Another consequence of the hollowed electrodes is the shrinkage cavitation. As the zinc deposit is burning, gas pockets or blowholes develop under high pressure and expand as soon as the sheet metal melts at the weld spot. Due to the hollow shape of the electrodes, the ring weld is formed at the outer edge and grows towards the in- 2

3 side. This means that the gas pocket is forced inward and finally forms the shrink hole or piping within the ring and mostly in the middle of the total thickness because this is where the pressure equilibrium is during the liquid phase, ref. Fig. 4. Shrink holes mostly develop on thicker plates because longer welding times leave the gas pockets more time to grow. Robot welding devices are also of advantage with regard to shrinkage cavitation since the effects described do not count too much with alternating welding angles. 3. Principle of ultrasonic testing of spot welds Fig. 5 shows a weld spot with an ultrasonic probe positioned on it and transmitting sound pulses into the weld metal, as well as the echo sequence generated on the screen display of the ultrasonic instrument. Let us start by assuming that the weld spot was flawfree. In addition, only one sound pulse is viewed at first. This sound pulse is transmitted from the probe into the weld spot and partially reflected from the interface between the probe and weld spot. This reflection appears as interface echo at sound entry (1st indication to the farthest left) on the screen display of the ultrasonic instrument. Fig. 3: Ring weld after carrying out the hammer-and-chisel test Fig. 4: Micrograph of a joint between three sheets with shrink hole The continuous part of the pulse enters the weld spot and is only reflected from its rear boundary, provided there is no flaw. This reflection is displayed as 1st backwall echo to the right of the interface echo. The sound pulse can run several times back and forth between the front and rear end of the weld spot, and delivers a part of the sound pulse to the probe every time it hits the front end. This ever decreasing part of sound pulse is displayed as 2nd, 3rd, 4th backwall echo at the same intervals on the screen. In this connection, the interval between the individual backwall echoes corresponds to twice the material thickness (round trip within the material). If there is a flaw in the weld spot, e.g. in the form of a gas pocket, a part of the sound pulse corresponding to the size of this flaw is additionally reflected from it. As the flaw is situated between the front and rear end of the weld spot, the corresponding flaw echoes also occur between the backwall echoes. In the case of major weld flaws, the flaw echoes are higher and possibly only recognizable by the fact that the intervals between them are shorter than those of the backwall echoes. 3

4 higher percentages of certainty can probably be attained, however, at present not at a reasonable technical and practical expenditure. The best results were achieved using the probes type G 20 MN... and the ultrasonic flaw detector type USIP Flaw evaluation criteria with ultrasonic testing Fig. 5: Probe on weld spot with echo sequence If the echo sequence of the 1st sound pulse has died out, the probe transmits the next sound pulse into the weld spot. The described procedure is repeated. A continuous graphical representation is obtained by synchronising the displays on the screen with the individual sound pulses. 4. The special spot weld probes from Krautkrämer Ever since 1988, Ford has been using special ultrasonic probes for testing spot welds. These 15MHz probes were developed by Krautkrämer in cooperation with Ford for spot weld testing on uncoated steel sheets and have proven to be most efficient in manual testing in combination with high-resolution ultrasonic instruments. The attempt to use these probes for spot weld testing on zinccoated steel sheets as well has led to difficulties insofar as small nugget diameters and bondings or cold welds could often not be detected as reliably as is usually the case with uncoated metal plates or sheets. These difficulties were clearly provoked by the zinc bondings that often exist with zinc-coated plates or sheets and impair the reflection behavior of flaws. This gave Krautkrämer reason to develop new 20MHz probes early in 1993 and to make them available to Ford for correlation tests. These probes having higher frequencies are characterized by a very high resolution, not least due to the very short sound pulses. The extremely critical tests at Ford revealed that intensively trained inspectors could detect the flaws mentioned, even on zinc-coated plates or sheets with 93% certainty using these new probes and state-of-the-art, high-resolution ultrasonic flaw detectors. Even Coated steel sheets having variable thicknesses are used in car body construction, and two or several sheets having the same or a different thickness are joined with each other by spot welds. The minimum weld nugget diameter is specified as follows: it should be at least four times the root of the thinner sheet s thickness, please refer to Fig. 1. At Ford (Europe), the minimum weld nugget diameters are divided into groups, e.g and 5.6 mm. According to these default data, there are probes whose sound beam diameters correspond to these minimum diameters of the weld nuggets. The test is always carried out from the thinner sheet side. The inspector must have precise information about the thickness and number of sheets welded together. Three basic features of echo representation on the display screen of the ultrasonic instrument are used for the detection of flaws in the spot welds and for the determination of flaw type: 5.1 The amplitude decrease of the backwall echo sequence 5.2 The occurrence of flaw echoes 5.3 The amplitude and position of flaw echoes 4

5 Fig. 6: Echo sequence unwelded sheet Fig. 7: echo sequence from spot weld 5.1 The amplitude decrease of the backwall echo sequence The galvanized sheet material used in car body construction has a mostly fine-grained, ferritic structure. This leads to a low sound attenuation in the sheet and develops a long echo sequence that dies out late. Fig. 6 shows a corresponding example of the echo sequence from a 2mm thick single sheet. This long echo sequence is a characteristic feature of unwelded material, or also of a cold weld. If the material is fused, a more coarse-grained, pearlitic structure is obtained. The sound attenuation is increased by this and the length of the echo sequence is clearly reduced. In comparison with Fig. 6, Fig. 7 shows the echo sequence from a 2mm thick weld nugget. The length of the echo sequence is proportional to the weld nugget penetration or the fusion. The length of the echo sequence is likewise slightly influenced by the surface condition. The deformation caused by the welding electrodes leads to inferior reflection conditions compared to those of a perfectly flat sheet. These deformations must therefore not exceed a certain limit for a clear and definite evaluation of the echo sequence. The sound path of the echo sequence observed on the display screen can be used as a measure for the fusion. This is calculated from the product of twice the total sheet thickness and the number of echoes to be observed. The result for the unwelded material, Fig. 6, is : 2 x 2 mm x 16 echoes o 64 mm sound path. Compared with the welded material, Fig. 7, the result is: 2 x 2 mm x 6 echoes o 24 mm sound path. 5.2 The occurrence of flaw echoes As already mentioned, the sound pulse travels several times to and fro in the structure with a good spot weld until the amplitude has died out due to the sound attenuation. The distance between the backwall echoes in this connection corresponds to the time of flight of twice the total wall thickness. If there is a flaw within the sound beam range, then a part of the sound is reflected from it, and a flaw or intermediate echo is produced, ref. Fig. 8. The position of the flaw echo gives information about the depth of the flaw. With Fig. 8, a shrink hole or piping in the middle of the joint, the flaw echo is positioned accordingly in the middle between the backwall echoes. With Figs. 9 and 10, a too small nugget diameter with varying sheet thickness (0.7 and 2.0 mm), the flaw echo A is already situated close behind the interface echo corresponding to the short sound path with probe coupling to the thinner sheet. In the present case, even a 2nd flaw echo B is generated by the multiple reflection from the flaw. 5.3 Amplitude and position of the flaw echo The height of the flaw echo depends on the size of the reflector and on its reflection behavior. It should be mentioned with regard to the size of a flaw that only the surface portions that are oriented 5

6 vertically to the direction of sound propagation determine the echo height. Due to its voluminous shape, a shrink hole will not always completely reflect all the sound toward the probe, and for this reason the flaw echo generated does not always show the true flaw size. Fig. 8: Echo sequence with mid piping/shrink hole Fig. 9: Micrograph of a small nugget diameter The amplitude of the flaw echo consequently depends on the sound pressure reflected from the flaw. If this is less than 50% of the total sound pressure in the flaw level, then the flaw echo is smaller than the corresponding backwall echo positioned to the right of the flaw echo. If it is over 50%, the flaw echo is larger. If the ratio is 50% to 50%, then the flaw echoes have the same height as the backwall echoes. If the inspector does not pay attention in this situation, the interpretation will go wrong because the impression is given here that there is an echo sequence from only one sheet thickness. With galvanized sheets, the flaw types small nugget diameter and cold weld are of special significance due to the previously Fig. 10: Echo sequence of too small nugget diameters with varying sheet thickness ( mm) Fig. 12: Echo sequence with Napoleon s hat 6

7 mentioned zinc solderings at the outer nugget edge or within ring welds. As is well known, zinc melts at a temperature of 420 C. It can be clearly recognized in Fig. 11 that the molten zinc has mainly accumulated outside the electrode area in the form of a bonding ring. However, there is also a very thin zinc bonding within the ring weld. These bondings can show different qualities, viz.: strong, medium or light. They can be distinguished from each other using ultrasonics. In the case of a strong bonding the flow resistance for the sound pulse is very low with a correspondingly small reflection at this point. This results in only few and small flaw echoes. With decreasing bonding quality the flow resistance becomes greater and the reflection at that point correspondingly larger. A characteristic feature of the bondings is their appearance in the echo display. Ultrasonic users call this Napoleon s hat, please see Fig. 12. The bell-type shape of the echo sequence can be explained as follows: the small reflection from the partly transmissible bonding causes an echo sequence with low sound pressure in the upper sheet as well as an echo sequence with high sound pressure in the total thickness. Every time the echo sequence with high sound pressure from the total thickness passes through again, a small portion is again reflected from the partly transmissible layer. This overlaps the echo sequence of lower amplitude. The result is at first an increase in the multiple echo sequence until the amplitude again drops due to the influence of sound attenuation in the material. Fig. 11: Zinc bonding after sheet separation A strong bonding has a very low reflection coefficient. For this reason, the flaw echoes are small, and the maximum of Napoleon s hat does not appear until later in the echo sequence from the total thickness. Compared to this, a light bonding has a somewhat higher reflection coefficient so that the flaw echoes are consequently larger and the maximum of Napoleon s hat appears early on in the total echo sequence. When sheets having identical thicknesses are welded together, the Napoleon s hat appearing is not only limited to the flaw echoes. It can also occur in the total echo sequence, please refer to Fig. 12, Backwall echoes 1 to Structural examination and optimization of welding parameters The prototype construction of Mondeo was the first time the ultrasonic inspection system was applied at Ford in order to inspect the structure of spot welds as well as to correct and optimize the welding parameters by means of analyses. The welding range is changed to higher current values due to the low interface resistance with galvanized sheets. However, these values are limited by the occurrence of spatter. Nevertheless, the fact that this limit is reached, i.e. spot welds with copper residues and/or welding stress cracking, does not guarantee that adequate fusion has been achieved with galvanized material. The extension of the welding range could be achieved by longer welding times. However, as welding time is the most expensive welding parameter due to the loss of cycle time, it is set as short as possible. The following quality features of spot welds were mainly inspected using ultrasonic test methods: 6. 1 Adequate nugget diameter with high fusion, item A in Fig Adequate nugget diameter with poor fusion, item B in Fig Too small nugget diameter, item C in Fig Spot weld with shrink holes or gas pockets 6.5 Bonding or cold weld 6.6 Burnt weld spot 7

8 6.1 Adequate nugget diameter with high fusion One characteristic feature of a spot weld with high fusion is the number of backwall echoes which should generally not be more than about 7 echoes. The indentations caused by the welding electrodes should be within the specification. Larger indentation depths lead to sound pressure losses due to sound scattering on the surfaces of the spot welds and mostly require a higher instrument gain. In principle, flaw echoes must not occur. As sound attenuation is mainly due to the welding structure, the damping value is more or less constant with high fusion: a few more echoes are therefore generated with thinner sheets than with thicker sheets. For example, Fig. 13 shows the echo sequence from a two-sheet joint (1.75 mm and 1.38 mm) with high fusion, Fig. 14 shows correspondingly the echo sequence from a three-sheet joint (1.75 mm and 1.38 mm and 1 mm). The resulting sound paths are more or less the same for both joints: ( ) mm x 2 x 5 echoes o 31.3 mm ( ) mm x 2 x 4 echoes o 33.0 mm With the same fusion, 10 backwall echoes should be expected for a joint between two 0.8 mm thick sheets: ( ) mm x 2 x 10 echoes o 32 mm) It can be said in general that spot welds with sound paths below 40 to 45 mm and without flaw echoes require no correction of the welding parameters. However, the electrode maintenance during production (tip dressing) is still necessary. 6.2 Adequate nugget diameter with poor fusion Fig. 15 shows the echo sequence from a two-sheet weld (1.75 mm and 1.38 mm) with poor fusion. The sound wave is only slightly dampened when passing through the spot weld so that an echo sequence of more than 13 echoes is displayed. The sound path measures ( ) mm x 2 x 13 echoes o 81.4 mm Due to the long sound path, this spot weld should be classified as having poor fusion. As there are no flaw echoes between the backwall echo sequence, the nugget diameter is adequate, ref. item B in Fig. 14: Echo sequence from a three-sheet joint Fig. 13: Echo sequence from a two-sheet joint Fig. 15: Echo sequence with poor fusion 8

9 Fig. 1. For this reason, spot welds producing this echo pattern should be evaluated as functionally perfect. However, practice has shown that these spot welds quickly overlap spot welds having too small nugget diameters or even cold welds in series production. The lower limit of welding range is reached with item B in Fig. 1. This kind of echo sequences should therefore be avoided when putting new welding systems into operation, and the adjustment parameters should be corrected. The most important causes of poor fusion are listed below. In practice, a combination of these errors also often occurs: Software No optimum stepper function; Insufficient welding time; Insufficient welding current; Hold interval is too long (escape of residual heat). Hardware Electrode pickup and/or contact faces are too large; Insufficient electrode cooling; Insufficient cooling of primary or secondary circuit; Shunted electrodes; Electrode pressure is too high; Welding with copper back-up bar. Material Varying thickness of zinc deposit; Change of surface coating (a change of surface coating calls for readjustment of welding parameters!); Shunt due to zinc deposit. 6.3 Too small weld nugget diameter If the actual nugget diameter is smaller than the required one, ref. item C in Fig. 1, then reflections from nugget edge cause flaw echoes between the backwall echo sequence, please see Figs. 9 and 10. The fusion may vary in this connection. The causes, even combination of causes, are the same as already mentioned with poor fusion. The following additional causes are possible: Software Spatter or material ejection due to too high welding current or too long welding time; Too short preheat pressure time. Hardware Contact faces of electrodes are too small; Electrodes are out of alignment in the longitudinal axis; Electrodes are too domed; Breaking of welding cable. Material Bad fit of parts; Sheet metal surface is contaminated. 6.4 Spot weld with shrink holes The formation of shrink holes or gas pockets with black sheets is mostly caused by contamination of the sheet surface, e.g. with oil. With galvanized sheets, the shrinkage cavitation has strongly increased in comparison with ungalvanized sheets. An increae of shrinkage cavitation is observed in this regard e.g. with electrolytically galvanized sheets more than with galvannealed material whose deposit s zinc-iron (ZnFe) alloy has a higher fusing temperature. The conditions for shrinkage cavitation with galvanized sheets are described in section 2. The following may be given as main reasons for the shrinkage cavitation, possibly also a combination of these causes: Software Welding current is too high; Preheat pressure time is too short; Welding time is too long (especially with thicker sheets). Hardware Electrodes are hollowed (ring weld); Contact face of electrodes is too large; Electrode pressure is too low. Material Varying zinc deposit thickness; Sheet surface is contaminated. 6.5 Bondings or cold weld Spot welds with a somewhat longer backwall echo sequence and a small flaw echo sequence, even if they don t occur until later in the form of a Napoleon s hat, deserve special attention. As already mentioned in section 2, the hammer-and-chisel method of testing does not always provide reliable information with sheet thicknesses of less than 1 mm, especially with galvanized sheets due to the strong zinc bonding and the occurrence of ring welds. Ring welds are mostly present in cases where destructive testing and ultrasonic testing give opposite evaluation results. A micrographic examination is advisable in these cases. Bondings already occur with sound paths of 55 mm of the backwall echo sequence! For two sheets having thicknesses of 1 mm and 1.25 mm and being welded together, this would mean: 55 mm sound path o 2 x ( ) mm x 12 echoes, i.e. a spot weld with 12 backwall echoes and small flaw echoes may already be suspected as bonding. This should especially 9

10 be taken into account with thre e - sheet joints. A weld nugget between 2 sheets dampens the echo sequence and can thus larg e l y conceal the influence of the bonding toward the third sheet on the echo sequence. For this reason, a three-sheet welding should always (if possible) be checked from both sides. In practice, it is mostly the combination of several influences that p roduces bondings. Attention should there f o re be paid to an optimum use of the welding range when putting new welding systems into operation. New elect rodes and maximum fusion re s u l t in a higher product stability during series production! The most important causes for this error are : S o f t w a re No optimum stepper function; I n s u fficient welding curre n t ; I n s u fficient welding time; Too long hold interval; Changeover to multiple impulse w e l d i n g. and the spot welds are afterward s checked using the hammer- a n d - chisel test method or the teardown test method. The welding c u r rent is set up to and just below the spatter limit. Burning of spot weld surface and sticking of elect rodes result in short electro d e lives. As already mentioned reg a rding this, burnt surfaces do not indicate that good fusion has been achieved. If the sound path of the backwall echoes is greater than 45 mm, then the welding time is too short with too high current. An example of a poor fusion in spite of burnt surfaces is shown for a spot weld with zinc penetration by Fig. 16 and for a spot weld with copper penetration by Fig. 17. H a rd w a re E l e c t rode contact face is too larg e, shows pickup or annular hollowi n g ; I n s u fficient cooling of electro d e s and/or of the primary or secondary c i rc u i t ; E l e c t rode pre s s u re is too high; B reaking of welding cable; S h u n t ; Welding with copper back-up bar and the thinnest sheet at the bott o m. Fig. 16: Micrograph, crack with zinc penetration M a t e r i a l Change of surface coating; Varying thickness of zinc deposit. 6.6 Burnt weld spots The welding parameters are traditionally determined by experience, Fig. 17: Micrograph, copper penetration 10

11 The main causes of burnt weld spots are: Software Welding current is too high; Welding time is too long (with very short echo sequence); Welding time is too short (with longer echo sequence); Preheat pressure time is too short. Hardware Electrode pressure is too low; E l e c t rode contact face is too small. Material Sheet surface is contaminated. 7. Final remarks Ever since 1985, the ultrasonic method has been used at Ford for the quality assurance of spot welds. At the beginning, it was only used as a complementary method to the hammer-and-chisel test method and the tear-down test method. In series production, the ultrasonic method has the advantage that inspection can be made quickly, repeatedly and without producing any test scrap. Its use is of advantage on outer skin parts that remain visible after painting, as well as at hard-to-getat points for hammer and chisel. In the prototype construction of Mondeo at Genk ultrasonic testing has proved to be extremely valuable for setting up new welding systems and for the use of galvanized sheets. The ultrasonic spot weld analysis enabled to optimize the welding systems within the shortest period of time. This made it possible to reduce the production of test scrap to a quite considerable degree when setting up new welding systems. The ultrasonic method is at present not only used for manual testing within the quality assurance, but also for optimizing the welding parameters. The striving for a maximum fusion in the spot weld enables to achieve a higher stability of the welding process and maximum utilization of the welding range. This results in better assurance of product quality in series production. By favour of the Ford company Krautkrämer GmbH & Co., P.O. Box 1363, D Hürth 11

12 Krautkrämer GmbH & Co. USA Great Britain France Italy Robert-Bosch-Str. 3 Krautkramer Branson Buehler Krautkramer Ltd. Krautkramer France Branson Ultrasuoni S.P.A. D Hürth (Efferen) P.O.Box 350 University of Warwick ZAC Sans Souci Div. Krautkramer Italiana P.O.Box 1363 Lewistown Science Park 68, chemin des Ormeaux Via dei Lavoratori 25 D Hürth PA GB-Coventry CV4 7HS F Limonest I Cinisello Balsamo # # # # Milano Fax Fax Fax Fax # Tx Tx TX Fax Tx