A feedback control system for laser forming

Mechatronics VoL 7. No, 5. p p 429 441, 1997 1q97 Elsevier Science I.td All rights reserved Primed in Great Britain 0957 -4158'97 $17.00 + 0 IXI

Pergamon Pll: S0957-4158(97)000 i 4-7

A FEEDBACK C O N T R O L S Y S T E M FOR LASER FORMING GARETH THOMSON

and M A R K P R I D H A M

Department of Applied Physics, Electronicand Mechanical Engineering,Universityof Dundee, Dundee DDI 4HN, U.K. (Received 15 January 1997; ret~ised 17 March 1997; accepted 2 April 1997) Abstract--Laser forming is a relatively new technique which is beginning to find applications in a number of areas. The technique involves passing a high powered laser beam over the surface of a material to induce thermal deformation. If the technique is to find widespread commercial application then control of the process will be crucial. The case is made for using feedback control as opposed to a knowledge based system to counter the difficulties arising from the range and unpredictability of the process variables. The development of a basic feedback system is described and the results for forming operations with the system implemented are presented. These indicate the success of the system in producing deformations of predetermined magnitude. Recommendations for future developments and refinements are proposed. ,~ 1997 Elsevier Science Ltd 1. INTRODUCTION Laser cutting has always been the dominant use of high powered lasers in material processing applications [i, 2]. In this role a laser coupled to a CNC control system offers industry a tool to produce high quality sheet components in quantities far below those for which conventional stamping techniques would be economical [3-5]. The laser needs no hard tooling to produce components; to change from one design to another simply requires a change of program. Lead times and down-times are therefore greatly reduced and with no high fixed tooling costs, prototype, one-off and small batch production is viable. However, many components require not only stamping from sheet but also subsequent forming into three dimensional forms. The advantage of laser cutting can therefore be somewhat undermined if a component cut from sheet in this way has associated with it the long lead times, high capital costs and the inflexibility of expensive forming tools or a reliance on the skills and time of craftsmen using basic tools to finish the component.

2. LASER FORMING Laser forming is a technique which could address this problem and also have applications in many other areas such as microfabrication and component repair or adjustment. It is a technique which uses thermal energy from a high powered laser to deform metallic corn429

43(I

(i. TH()MSONand M PRII)HAM

ponents [6 12]. This can be achieved through a number of basic mechanisms, principally melting and buckling. In both cases a defocused laser beam is tracked over the surface of the material, the focus of the beam typically being some It) 20 mm above the surface of the sheet. At the University of l)undec a 1000 W ('arbon Dioxide laser is used. though this is not contirluously on full pow.'er but pulsed tit 400 H;/with a resultant average energy density on the surface of the sheet of lypically 6 ( ) W m m : . If this density, is allowed m increase signiticanl[.',., for example by bringing thc beam into sharper focus with respect to the component surface, then there is a danger the beam `,~ill cut the material. With inch laser forming, the laser passes over the component such that it melts a narrow track on the surface. The depth of this track ,,,.ill typically be 13 lhat of the component sheet thickness with a track breadth of a similar order of magnitude. As the track cools after melting it contracts, drawing [hc tipper faces of the sheet together. This results in a permancnt bend of typically 2 degrees on a single pass in 1 mm thick mild or stainless steel sheet. Laser forming by: the buckling mechanism is somewhat more complex. In this case the material directly subjected to the laser beam attempts Io expai~d xery rapidly as the beam passes over it. The relatively cool material either side o|" the track is therefore put into compression and if the parameters are chosen correct[`, then a small degree of permanent Iocalised buckling will occur. By repeating either process the degree of l'orming can be increased. So. lor exarnple, to produce a 90degree fold the laser would be tracked over the sarnc point as many times as is required Alternatively.. curved geornctries can bc produced by' offsetting the tracks slightly to produce a series of parallel passes. The greater the offset the more gentle the curvature induced will be. l.aser forming, in somc respects,, i~, I,ir more SellSili\ e h) process parameters than cutting. When a sheet is cut by a laser it is essentially either cut or 11oi cut. though there will of course bc `,ariation in cut quality dependirig on the specific parameters set. With laser forming a specific &,.qrec of deformation is required. There are basically.' two approaches to tr\ and develop a s.vstem which will accurately and repeatably produce spccitic degrees of laser induced detk)rmation w.ithout direct human interaction. These are the predictive approach and the feedback a p p r o a c h

3. PREDICI'IVE-OPEN I,OOP A P P R O A C H The predictive approach uses either previous empirical results, theoretical methods or a combination of both In try to determine the values of the various process parameters which should be used to producc a certain dcgrcc of laser Ik'wming. Trials would be carried out over a range of materials and thicknesses, varying control parameters and measuring the corresponding degrees of delormation. These would then be stored as look-up tables within a computcrised database. When a ncv,,.job was entered, the appropriate table would be consulted and process parameters recalled. If no exact match was found in the tables then interpolation could be used to bridge gaps. Theoretical work could also be used to improve interpolation methods. For example, the bending stiffness of a sheet is generally proportional to the tube of its thickness rather than simple proportionality. hence non-linear interpolation would be used when the specific thickness required fell between table nodes.

Feedback control system for laser forming

431

This approach would then instruct the user to use a certain set of parameters: focal position, traverse rate, number of passes, offset between passes etc., to achieve the target deformation. From these simple tests it was hoped that the knowledge base could be expanded, using basic statics theory to include the bending moments needed to induce these target deformations. This would provide a correlation between the process parameters and the bending moments they induced. Using finite element analysis the moments required to produce more complex geometries were calculated and it was the intention that the process parameters required to produce these shapes would be extracted from the knowledge base and used to manufacture realistic components. The predictive approach, however, is highly dependent on the overall repeatability of the process. This unfortunately is not particularly good as far as laser forming is considered. Some initial work was carried out to assess the viability of the predictive system. For example, a test was carried out using nominally identical 10 mm broad strips of I mm thick mild steel. The first strip was laser formed by passing it beneath the beam 10 times and the resultant deformation was noted. The other specimens were then laser formed at 30 minute intervals using identical process parameters. In a repeatable system the deformation of the specimens would be identical; however, in this case there was a spread of results of around + 15% of the mean value. There are a large number of parameters involved in the process over which the user has little influence, but which will alter the degree of forming. These can broadly be grouped into those associated with the laser, those with the material and those with the CNC system used to position the material beneath the laser beam.

3.1. Laser

Most lasers exhibit a degree of variability in their output [13-15]. Lasers are a complex balance of optical, electrical, gas and coolant circuits. If any of these have a tendency to fluctuate then the laser performance is also likely to vary. Optical components gradually accumulate dust and dirt. Regular cleaning can minimise this problem but will not eliminate it. This can alter the beam shape and power. If too much debris collects on an optic it will begin to thermally deform and this will induce large variations in the beam shape as it is reflected by or passes through the distorted optic. The laser beam itself is generated within what is called the optical cavity. Within this cavity exists a mixture of g a s e s - i n the case of a CO2 laser these are nitrogen, helium and CO2 itself, all in specific proportions. Small lasers tend to have these gases sealed in at the factory, while larger devices use bottled gas which is partially recirculated. In either case gas can leak and impurities can enter through imperfect seals in the system, resulting in an incorrect gas mix and a degradation of performance. With the sealed lasers the degradation is slow and these devices may be able to operate for up to 20,000 hours without significant drop-off in performance [16]. With the larger lasers performance would drop off quickly without the ability to pump in a fresh batch of correctly adjusted gas as required. Most lasers do this automatically, however this results in power fluctuation rather than degradation. Similarly, if the cooling system has some variability associated with it, even for examplc the mains water temperature or pressure altering, then this will result in optical or electrical components running slightly hotter or cooler than normal, again altering the output laser beam.

432

(i. TH()MSON and M. PRIDHAM

3.2. ,llal~'rial

r h e material being formed ,,,,ill be subject to degrees of variability. Not all I mm standard stock mild steels for example will have exactl), the same specification, with respect to both composition and manuthcturing process. Sligh! variations in composition, particularly cltrbon contenl, can have a marked effect on properties. Higher carbon levels increase the slrength whilst effecting the rneltmg range and lowering the linal melting point. Metal forming processes such as rolling can be highly intluential in tailoring properties, particularlv when the final dcformalion is carried out belo,a the recrystallisation temperature. This is as a result of diclating the grain (crystal) size and shape its well its producing preferential orientations or texlurc in the structure ,.~,hich can lead to anisotropic properties. II is well know.n that this ~ariability in properties can manit?st itself in cold rolled sheet which is subsequently deep drav,.n. While furthcr \~,ork necds Io be carried out with respect to this aspect, it is feasible thal material variation will exert some influence on how easih. the material ,.,,ill laser l\~rm. The nature of the material's S,ulface significantly ahel'S ]lo~A. ;.1 laser beam interacts with the shecl surface. A small spot of"oxidc will drasticall) improve the absorption of the beam energy due to its less reflectivenat t,re while a highl.~,polished surface can have the oppositc cl:['ect. In addition, if any grease or oil Is presenl on the surface then this can protect the sheet from the beam. A small variation in lhickncs~, signilicanll.~ altcr, the sheet rigidity' and so laser li)rming ~vould be olher lhan predictcd. 3.3. ( ' . V ( ~rwcm

l h c C'N(' system including the .X } table hm, goud ;.lc(2urac} alld repeatability [I 7J. The exact degree of repeatability in this t.',pc of equipment will var) from table to table but m comparison with the x.ariabilitv associated v~.ith the laser and the material it is unlikely to bc a signiticant factor. Cunyulalivelv these xariabilitic~, in the system ,nc ~,Cl3. significant. Problems ~aith variability in the laser could be reduced bv improved servicing, though the le'~.el required would bc extremely lime constnning, impractical and costly.. Similarly. the material could be controlled to very tight specilications which ~,,ould again result in increased purchase and prcparalion costs. As a rcsull it w:as decided that a kiser forming system based entirely on predictive and theoretical work would bc impractical. "lhis same conclusion has been reached elsewhere hv researchers faced with similar ~ariabilities in the lield of hlscr cutting and welding, resulting m uneven processing under nornmalh.' unilorm conditions. As a restih a number of techniques based on feedback of the culling process have been arid continue to bc explored to try to ensure optimum processing at till times I I 8 20]. Therefore. a better approach to achieve precision laser forming would be to introduce on-line measurement to continually monilor the process and control key parameters rather than rely entirely on predictive information. 4. FEEl)BACK A P P R O A C H

"lhe feedback approach relies on sensor~, to monitor the degree of deformation, compare this to a target value and adjust the process parameters accordingly.

Feedback control system for laser forming

433

The exact nature of such a system will vary depending on the specific design of the component. If the formed geometry is of prime concern then displacement transducers would be fitted to key points around the component. In other cases however, the geometry as such may be of secondary importance. In addition to component manufacture, the laser forming process may be used, for example, as a joining technique. In these circumstances two components would initially have a small degree of clearance when fitted together but an interference fit could be produced by inducing distortion around the interface using laser forming. In this example it is the interference stresses which are critical and so strain gauge monitoring of the induced strain in the components may be the best approach to use. Similarly, if some form of capacitive device is produced by laser forming the position and orientation of one plate in relation to another, then the size of the gap between the plates in itself is likely to be of less importance than the resultant capacitance. Hence a capacitance test circuit, rather than one based on displacement monitoring, would be used to provide feedback and so control the process. To optimise the forming process it was clear that a higher level of control than simply switching offthe laser when the target deformation was achieved would be required. In the early stages of a forming operation a fast fold rate would be required to rapidly reach thc approximate target and so minimise the overall processing time required. This fold rate would then need to be reduced as the target was more closely approached in order to avoid the risk of overshooting. The rate of forming depends on a number of factors. Given a specific material and thickness of component the user is able to change the fold rate by altering the amount of energy entering the fold line per pass. In general the greater the energy the faster the fold rate. Thus to control the fold rate, the amount of energy entering the material must be controlled. To increase the amount of energy entering the fold per mm of traverse there are essentially two main approaches: either the average power of the beam can be increased or the traverse feedrate can be reduced. To alter the average power of the laser the pulse duty ratio is changed. In other words the laser is switched on for a greater or lesser part of each pulse cycle. This can be achieved by using a signal interface fitted to the laser power supply controller. A system to do this was used in an earlier project concerned with laser cutting [21] which varied the laser pulsing as a cut was performed. While this system in principle worked correctly it was very prone to interference due to high voltage switching in the laser's main power supply. This interference would produce spurious pulse settings and so the method suffered from a general lack of reliability. The laser used in the current project is of the same type and so would also suffer from these shortcomings. However, these problems are associated with the particular laser type used and should not be allowed to detract from the rationale of the technique. The alternative method was to alter the traverse feed of the laser. The faster the pass, the less energy entering the material per pass and so the slower the fold rate. This would be relatively easy to implement and would have the benefit that the signal wires would pass from PC to the X Y table controller and so could be kept remote from the high voltages associated with the laser itself. A third method which could have been used involved altering the position of the beam focus and hence the spot size of the laser as seen by the component surface. This is slightly different from the other two methods in that the energy input per unit length of track length

434

G. TItOMSON and M. PRII)HAM

is not altered from pass to pass though its distribution will be as the focus and so the spot size are wiried The more focused the beam, the smaller the spot size. In other words, the energy' from the laser is directly interacting with a smaller area o f material, creating very rapid temperature changes beneath the beam. A large temperature differential between the track heated by the laser and the cool surround results and so a fast fold rate is produced. Conversely a slow fold rate can be produced by using a more defocused beam which delivers the energy' more difl'usely. In this case the temperature differentials are smaller and so a slower fold rate is achieved. Adjusting the focus to vary the fold rate could however tend to damage the c o m p o n e n t surface excessively if the focus is too sharp, since conditions become close to those associated with laser c u t t i n g From the above considerations it was decided that the best method to use for the particular hardware available in this case was to use feedrate as the control parameter. Adjusting the feedrate could be achieved easily and reliably during processing time and w.ithout the need to resort to the design and construction of the interfaces needed with both the pulse and focus adjustmenl methods.

5. [e]XPERIMENTAL S Y S T E M 1"o illustrate the fcasibilit}, of feedback control of laser forming a test rig was c o m m i s s i o n e d The simplest case to develop the principles involved would be to use linear displacement as the control variable. A test rig was therefore constructed as shown in Fig. I. This was designed to control the deflection o f test pieces o f length 80 .[00 mm. breadth

i "~,l

Loser Nozzte B~onk

486 PC

ANO~O XY CNC Controller

LVD/

(DfgS) PCL-B18L

I -

Anolo~ue

in

I0 v

Jl7

-

i

TO XY Oigitol Out /

In

[----"~IILVDT / P C L O - e t 15 t"

/*-o,I.oo=,

Toble

sig. ,n 1o loser shutter

I_~.l

a

eloy

Eorth switchi~;J to Anorod 8 lines moxlmum Return dlgitol signols to PC 2 lines octive

Fig. I. Diagrmn sho~.ing the various components within the laser fornmlg control systcnl and ho,a the,, are connected to each other.

Feedback control system for laser forming

435

Fig. 2. Close view of the component being formed showing also the position of the laser nozzle and the location of the displacement sensor at the tip of the component. 10-25 mm and thickness 0.5-1.5 mm. A maximum deflection of 10 m m could be measured with this particular arrangement. The specimen is located in a jig and held in place using a toggle clamp. The displacement is measured using an L V D T with a contact probe, attached to the plunger and positioned so as to sit on the tip of the test component. As the component deforms then so the probe is lifted and the displacement can be recorded from the L V D T (see Fig. 2). The motion of the X Y table, and so the component, is controlled by the Anorad controller. This uses C N C coding to manoeuvre the component while it also offers the possibility of using "M-codes" to communicate with other external devices. The M-codes are simple digital signal lines- in other words inputs or outputs associated with these are either " o n " or "off". This prevents the direct use of the Anorad itself as a device to read in the analogue signal from the LVDT. A PC was therefore incorporated into the system. When fitted with an interface card this device was able to read in the signal from the LVDT, process the results and then send out signals to adjust process parameters accordingly. Figure 3 shows a screen shot from the control program. The user is prompted to enter various details regarding the specimen to be deformed. The facility to log the deformation after each pass and send this to a file is also available. The three buttons toward the bottom of the screen are used to stop the process, start it and clear data for a re-run. The circle in the centre is simply a graphical device to indicate how close to completion a particular task is. In addition to the PC control program, the Anorad XY controller also had to have software specially written for the task. Figure 4 shows the forming control system flow diagram. Essentially, after each pass is executed the PC is triggered, by a signal from the Anorad, to read the position of the LVDT. This value is then converted into a displacement using an appropriate zero offset and calibration factor. In this case L0 is the zero offset and C is the L V D T sensitivity calibration factor (in V/ram). The displacement value is compared to the lower target limit.

43~,

(i THOMSON and M. PRII)HAM .

.

.

.

.

Jll

--

--

~

'

.

.

.

.

.

.

.

.

.

.

.

.

Fig. 3. Screen shot from the laser forming control program. The user has entered process parameters and is about to start the l~rming process. If it is above this level then the forming process is complete, though a check is made to ensure that the upper target limit has not been breached. If the displacement value is below the target then further passes will be required A current reading close to the target will demand a slow forming rate to avoid risking overshooting the target, while a large discrepancy will allow for faster l~)rming The pC" theretk)re uses basic rules to select an appropriate forming rate and triggers the A n o r a d controller using one o f up to eight lines. each potentially providing a ditferent IZ~rm rate. The A n o r a d program meanwhile has been held in a loop waiting for the signal from the P C This signal then allows the A n o r a d to break out of the loop and access a particular subroutine to instruct the X 1' table to perform a forming pass at a particular leedrate. Once this is completed the A n o r a d informs the P(" that the pass is complete, returns to its main loop and the whole process is then repeated as required. Figure 5 shows results o f t'orming tests for strip specimens deformed to I 6ram target minimum deflections in I mm increments. It is noticeable that, for example, only six passes were required to reach a target delk)rmation o1"3 mm where as 19 passes were needed when a deformation o f 2 mm was specilied. The control system itself can account for some o f these apparent irregularities. For example, consider the case whereby the threshold tk)r using process parameters to produce the slowest tk)ld occurs when deformation is less than or equal to 0.5 mm from the target minimum. After a series of passes a c o m p o n e n t may have been say, (.).6 mm from the target and then 0. I mm from the target after the next pass, thereby triggering the slow fold rate settings. This c o m p o n e n t is very close to the target deformation and so will need very few passes at the slowest fold rate tot completion. A

Feedback control system for laser forming

437

PC Controller I

Readi n ~ dets/ls : I l~n/wmm d ~ Train Maxinmm~ Tnwx R e a d i n ' ~ " from LVDT : L0 [

~ L V D T : LI I

Anorad XY Controller

[~

Deflection: D=(LI -LO)* C Wlz~ C is a eali~tion factor

-

1

IsE< 0.1ramtlz~ UiSgerline 1 Is0.1 < E < 0.25ramthen U/gi~ line ~ Is 0.25 < E < 0.75 mm then tri88er line El,selriggef line 4

(say...) ,r I~0

Formed to withla

Fig, 4. Forming control program flow diagram. second component may however be recorded as being 0.5mm from the target after a number of passes have been performed, thus only just triggering the 0.5 mm threshold. This component would therefore require more passes at the slow fold rate to complete the process, resulting in a longer processing time. The results obtained, however, also indicate the need for some form of on-line process control. The slope of the curves is a measure of the fold rate. With a repeatable system, using the same parameters should give the same fold rate and so the same gradient of graph. Comparing the initial slopes of the plots for the components designed to give 5 and 6 m m target minimum deformations respectively, it can be seen that the gradients are markedly different even though the parameters used would have been identical. This is due to the process variabilities discussed earlier. The process variability is further highlighted in Fig. 6 which shows the results of forming

438

G. THOMSON and M. PRIDHAM

I - 6 mm Target Folds

E

~4 0 03

T

@ D2

-

*

1

*

0

1

•

o

o

o

•

•

•

•

o

o

°

m

•L - I ......... 2 3 4 5 6

lmm Ta~et •

o~

4ram T a ~ e t

-~ . . . . . . . . . . . . . . 7 8 9 10 11 12 13 14 15 16 17 18 19 NO. o f Passes e

2mm T a ~

3mm T a ~ e t

~ ~rnm T a ~

•

6mm T a ~ e t

Fig. 5. Results of six lorming tests, with targets ranging l'r(,m l to Oturn. tcsts carried out under nominally identical conditions tor thrce strips which wcre tormed to a target minimum ot'4 mm. This illustrates the need for a feedback based control systcm rather than a predictive approach, since whilc two of" the tests had similar fold pattcrns thc third did not. A predictive a p p r o a c h using data from the tirst two tests would have resulted in the deformation o f the third specimcn falling short o f that requircd. Whilc using a feedback system reduces the influencc o f the variabilities mentioned earlicr, the process is still subject to these fluctuations. For example, as previously discussed, when the control dctects that the specimen is close to the target deformation it increases traversc rate in order to slow down thc rate o f forming. However. the actual rate of fold for the

D e f l e c t i o n v. P a s s e s 4ram Target

~'5 E

o

•

o

O O3 (D ~D e",2

•

•

•

I

o

,-

;

0 I-U 0

qp

r,

-~l! . . . . . . ~ 1 2 3

• 4

e

~ 5

: . . . . . . 6 7 8 9 10 No. o f Pintoes

Test1

•

Test2 •

. . . . . . 11 12 13 14

15

Test3

Fig. 6. Results of" three forming tests each with a target deformation o1"4ram. illustrating the ability of the control to overcome process variability.

Feedback control system for laser forming

439

Overshoot

0.8

~0.6 v

)

~

0.4

)

•

1

•

0.2

•

3

•

.

.

.

1

.

.

.

.

2

.

.

.

.

o

.

.

.

3 4 Tl.~Bet (ram) •

Test I •

Test 2 •

•

T

6

6

Test 3

Fig. 7. Plot of degree of overshoot past target minimum deformation for a series of 18 forming tests. These consisted of three tests at each of six different target deflections. given traverse speed will be a function of both the controlled parameters and the uncontrolled variables. It is these uncontrolled variables which make it difficult to minimise the degree of overshoot beyond the target deflection. Figure 7 shows the degree of overshoot for three nominally identical series of forming trials including those featured in Fig. 5. This indicates a tolerance of 0.4mm above the target minimum deformations for these tests. Further work will therefore be required to minimise these errors.

6. C O N C L U S I O N S This work has shown that laser forming, when coupled to a suitable control program, can produce satisfactory results. However, further work will be required before a commercial system can be developed. At present the control system will produce laser formed components to the aforementioned tolerances. Many industrial applications will however require a greater degree of precison, A number of strategies to overcome this problem could be adopted. A cautious adjustment to the process would involve using higher feedrates than previously when in the final approach to a target. This will, in general, bring down the rate of fold regardless of whether the uncontrolled variables have induced a slow or fast fold. This method would be straightforward to implement but by using a blanket approach, excessively long process times could result in certain cases. For example, if the uncontrolled variables were such that a slow fold rate was natural, then adjusting the feedrate to also give a slow fold would result in much longer processing times than necessary. Reducing some of the variability in the system would tend to make the steps per pass more uniform for given control conditions and as a result allow for tighter tolerances. The inability to tighten these variables adequately was given as the main reason for rejecting the predictive approach. In the predictive approach a +__20% variability on each pass, say,

440

G. THOMSON and M. PRIDHAM

is accumulated to give a similarly unacceptable tolerancc on the final delbrmation. Using feedback, however, the del%rmation is assessed after each pass, hence variability is only critical on the final pass of a forming operation and so is of much smaller absolute magnitude, e.g. _+20% of the linal pass only. Reducing variability even slightly would therefore directly tighten tolerances, hence extra care in ensuring relative uniformity of the material to be formed and improved laser maintenance would result in improvements. This is in contrast to the massive levels of variable tightening which would be required to allow a predictive approach to operate successfully A method which could perhaps prove very successful would involve a form of adaptive control. By logging the deik)rmation of a particular component it could be determined how quickly the specimen is deforming. In other words, rate as well as displacement are the feedback variables. A specimen with a fast fold rate would require more caution close to the target, while a component forming excessively slowly could have its control variable (the feedrate) adjusted to boost this. Ideally this system could incorporate fuzzy logic and artificial intelligence techniques to alter and improve the control rules as more and more specimens are formed. Laser forming holds promise in a number of areas where precise deformation of metallic components is required. The conventional sheet metal area is one possible end user, however greater areas of exploitation may lie elsewhere. Possible future uses lic in the production and adjustment of sensors, medical applications, specialist areas where non-contact forming is of use, and also particularly in the production of microcomponents. Controlling the process adequately will however determine the success of the technique in any given application.

REFERENCES 1. Mayer, A.. European industrial laser market. In lhe Marketplace./br Industrial Laser.s. SPIE volume 1517, Bellingham, US.A.. 1990. pp. 15 27. 2. Kellock, B., YAG versus C(): power struggle..~achinerv and Pro~h,'tion Engineering, 1991, 149(1318), 18 22. 3. Goodwin. B., The battle of the cutters hots tip. 17u' En.qineer. Vol. _7,. -~ ~ 1991 pp. 31 33. 4. Steen, W. M., Laser ,klaterial Proce.ssin.q. Springer, London, 1991. 5. Williams C., CO, industrial laser systems and applications. In Laser Processing in Manu/acturinq, ed. R. ('. ('rarer and P. J. Oakley. Chapman & Hall. London, 1993. pp. 141 162. 6. t:rackiewicz, H., Laser metal forming technology. Proceedings ~/Fahtech International '93, Chicago, 1993, pp. 733 747. 7. Vollertsen, F.. Komel, I. and Kals, R., The laser bending of steel foils for microparts by the buckling mechanism a model. Mmk, lling attd Simulation in Materials Science and Engineering, 1995, 3, 107 I 19. 8. Geiger. M., Synergy of laser material processing and metal forming. Annals q/ the CIRP, 1994, 43(2), 563 570. 9. Geiger, M. and Vollertsen, F., The mechanism of laser forming. Annals (?/ the ('IRP, 1993, 42(I ), 301 -304. 10. Arnet, H. and Vollertsen, t-.. Extending laser bending for the generation of convex shapes. Journal 0] Enqineerin,q Manu/acture. 1995, 209. 433- 44 I.

Feedback control system for laser forming

441

11. Pridham, M. S. and Thomson, G. A., An investigation of laser forming using empirical methods and finite element analysis. Journal ofDesi9 n and Manufacture, 1995, 5, 203-211. 12. Thomson, G, A. and Pridham, M., Laser forming. Manufacturin9 Engineer, 1995, 74(3), 137- 139. 13. Smith, A. L. S., New instruments add precision to the laser lab. Photonics Spectra, 1995, 29(4), 132-136. 14. Schuocker, D., Dynamic phenomena in laser cutting and cut quality. Applied Physics B, 1986, 40, 9-14. 15. Garcia de Vicuna, G. E., Beitialarrangoitia, J, C. and Ghosh S. K., Defects arising from laser machining of materials. In High Power Lasers, ed. A. Niku-Lari and B. L. Mordike. Pergammon Press, Oxford, 1989, pp. 227-249. 16. Bondelie, K., Sealed carbon dioxide lasers achieve new power levels. Laser Focus World, 1996, 32(8), 95--100. 17. Tullar, P., Machine tool evaluation produces results. American Machinist, 1996, 140(4)~ 63-65. 18. Foldvari, E. J., Kluft, W. and Borger, P., Laser beam diagnostics: the key to metalworking success. Photonics Spectra, 1996, 30(9), 94-100. 19. Zheng, H. Y., Brookfieid, D. and Steen, W. M., Kerf scannning system for laser cutting quality control. Lasers in Enyineering, 1991, 1, 37 48. 20. DiPietro, P. and Yao, Y. L., An investigation into characterizing and optimizing laser cutting quality--a review. International Journal of Machine Tools Manufacture, 1994, 34(2), 225-243. 21. Thomson, G. A., Simpson, G., Ritchie, J. M. and Ross, I. E., A methodology for the development of a CO2 laser cutting database. In Proceedinys Sheet Metal 1992, Birmingham, U.K., 1992, pp. 379--386.