You are on page 1of 32
Armin Scholl Data of Assembly Line Balancing Problems Nummer 16/93 November 1993 Letzte Anderurg: September 1997 inschaft + Prof Dr. H, Stadler ssysteme + Prof Dr. HJ. Petzold De, Amin Scholl ‘Tectnische Universitit Damstadt Insti for Barlebawinschailehre Hocachulrae I, D- 64289 Damstact ‘Tel, 6151/1645, Fax: 05151 165504 EMail: schol@bwlsh-darmstat. de Data of Assembly Line Balancing Problems ‘Armia Scholl ‘Technische Hochschule Darmstd, lin fOr Bticbwirschafetre ochsculsuae 1, D= 61259 Darmsta, Geemaay In this pape, we present benchmark dita sts for Simple Assembly Line Balancing Problems (GALBP) which may be used totes and compare solution procedures. The (genzal) SALBP nsits of assgning tasks to work stations of en assembly line sich thatthe sum file tes [S minimized. The number m of sation as well as the prouction rac, or equlaleay, the ‘yee time have to be determined simultaneously. Additonal, peeedence constants ‘tween thc tasks have t0 be. considered. Two rericed versions ofthe prblem, fen ‘considered in the literature, are SALBP-I and SALBP-2, SALBP-1 is to minimize the number ‘of sation fora given cycle ie whereas SALBP-2 ie fo minimize the cyte tn for a ven ‘umber of stations. 1. Introduction to SALBP ‘The simple assembly line balancing problem (SALBP) arises in the mass production of « single product ona paced assembly line. The production proces is subdivided into m tasks (divisible elements of work) whose operation times t, (=I) ae assumed t0 be integral constants. Due to technological restritions, precedence constraints partially specifying the sequeace of tasks have tobe considered. These constraints canbe represent~ ed by a precedence graph containing. nodes for all tasks with operations times as node weights and acs (i,j) whenever a tak i as to be completed before a task jcan be started The assembly line consists of a sequence af m work stations which are connzcted by a con- ‘eyor belt. Each station repeatedly performs a subset ofthe tasks on consecutive product units moving aloag the line at constant speed, The fixed time interval avaiable for oper ing a unit in each station is called cyee time . The production rate ofthe sssembly line is sven by 1/e ‘The general typeof the simple assembly line balancing problem (SALBP-G fr short) can be stated as follows (see, eg, Baybars (1986) or Domscke etal. (1953, ch, 4). Each task has to be assigned to one sation of the assembly line in such a way that ao precedence constraint is violated. Equivalently, a partition ofthe set ofall tasks into disjoint sets S, (station loads) with k=1,.t0 8 t0 be found. For each are (i,j) of the precedence graph the relation b For the precedence graphs Arews 1, Arcus 2, Kilbrd, Sawyer, and Tonge, « data set with 50 instances results (See Table 3). If m* is currently not available for an instance, the best known values of a lower and an upper bound on m* are given (separated by a hyphen). ‘Tonge 1 2st wT BS rss a ie GSE mo ‘Avcust 3m 21" 38S O06 ase aT oe 1g HO IS GDI ES now fron? | SBS 27 OIG 26ST 25" HY wT Tie 2 — 7320 20-2176 20" RSS TBAT. IB S00 tomer te 0M TTT 1D ‘Taatz 3, Data Set of Hoffmann, ‘Additionally, Hoffmann checked the precedence graphs Tonge forall cycle times between 156 ‘and 1746 (23 to 3 stations) and Sawyer for all cycle times between 25 and 161 (14 to 3 stations). Only few instances of both graphs (in particular those which are contained in the above table) are hard instances in terms of computation time requirements. 43 New Data Set for SALBP-1 We propose a data set with 168 instances which includes precedence graphs not yet cnnsidered in other data sets. Cycle times are determined by the method of Hotimann (see Section 4.2). In contrast to the data sets in Section 4.1 and 4.2, there are problem instances with the dife- rence m*~m,,,, being larger than 1. In those case, this difference is specified following the "sign, Name © om 6 m 6 m 6 mt cmt Romtez | lt 6 8 m8 PG Fy a Boxey a Or ow oe oD a7 OF Let wd ut sm toe Fam was me unter isle tele ee ro estes 6: or at Hahn 200s 887 HOS aT eT Wanete | $4 31 56 ew? wr Tt pe as romeo se reiz alam ee a al eet oe ee m4 Wee-Mg | 28 6% 29 GM 0 ao ts ogists Bam Gs as 35 os 37 gant BOD % Om 4 wR MD Sgr % so 3 ae soo? 2 I 56 SO Lie i ee a ara fis) a i 07 6 DS aM te Bo mw mw tT me Ore te mo eo om ow rie 150 Musbero | 176 257 us 2H tsk st tk ato me mis we it 38 ir at Oe soos me als Babes! | 3 4 4 3 mM SHO as ee Se 7 Banolé? | 4 Sl 85 SOSI TT 8 & uso km a a re us 360 Ss MOB BoM TT 1 26 © 1m 2s Scholl 1396 S0-S1 m2 50" 1452 AB 14S ATR ASIS. 46-47 156 46 ISBT aa 168D ADS 1D AID ia 401787 30-40 183438 RES BT 198536 11 35 20 332 a3 mm 3 tm 2 Meh 2580 7686 mm 2 “Tanus 4 Data St of Schall 5. Benchmark Data Sets for SALBP-2 In the literature, there is no benchmark data set available for SALBP-2. We present two new ata sets consisting of the precedence graphs of Table 1. For each graph, a range {=, i] of 9 station numbers is defined resulting in up to Hi-m.+1 different instances. Generally, the ranges are restricted by m3 and ¢ min {th,/ty,] +2 [n/2]]}. Additionally, some ‘easy instances are eliminated and the number of instanoes per precedence graph is limited to 28 in order to prevent one problem structur dominating the whole data set. Therefore, in some cases only a subrange of [mi] is used, ‘5.1 Data Set 1 for Type 2 ‘The first data set for SALBP-2 includes all graphs with n 230 of the data sets for SALBP-1 in Sections 4.1 and 4.2. Additionally, it contains the graphs Buxey, Gunther, Luts 1, and Lutz 2 resulting ina total of 128 problem instances. For each graph, the diferent instances are grouped into five columns. For each value of m, the minimum eyce time e* is specified. Ife" isnot yet known, an interval of possible values is provided (for the upper bounds only the last digits which differ from the lewer bound valves ‘are noted). Instances for which c* is larger than the theoretical minimum cycle time Cy max [tg / Ts tyye} 86 marked with '*", Ifthe diference eq. is larger than ts value is added after the '*' sign. eer ee er) ee a ate oor ise re ee es (Gamer |e) ee ees ee Ge ee 0 ele oe noe 2 4 an eS conde ne i 3 @ 8 @ Ww % nN uu |e mere 9 ee 0 ea ted ee es ee a tse yn ost Dp % » 3 2 4 %» B B DB a od » %» b mm F @ ® Sewer |7 47 a ee ee ee nom wk Teoge [3 1170 a a 3 39 eer ee mh eh Bom lS BSG aT Be i B16 see 12318116) Avast | 325036 wo $s? 12a «786410 8 95a se" 107580" © TOBe7O1 az ggrzrtm 3 SR64n0 Ser 1S SIONS 16 aa astee? gira foe 2) Mere 3a ars? | 3503 sr 3am 786 3 18800 Ie, 10 1500-41 13673 2 3 11570 sore7 1S 10035-36°° 16 seliniar 17 s854-61"7 1% §576-77%0 19 TMD-Bwr? 20 T5026 TIBL-DORD ZB RIDSI 2 6SsL- mare 25 ams-10s 25 SESE «27 5889 “Tanus 5. Fit DataSet for Samra 10 5.2 Data Set 2 for Type 2 ‘The second data set contains graphs with n 253 tasks and a total of 174 instances, Hain | 30 arernsh43errm sos 6 m7 gga 8 orm = Imre io ITs ims [le We ee asic ean aa) eck, 8 etre) 0 i ie ee ore Bom of mS Sta Bor mm a Bo om we 8 a te to eae BR 9 Set we |3 0 4 SS ms aegis ieee 0s ei 0 ie leisy ae a eG ee le ie ee h ur 8 wee om rt a nm 2 fe B 6 6 6 3 oe TEST a toes |3 0 se # ak Ss teh 8 or 8 wes BD Sse? ere to oo ee aM Mother] 3 4034S aE? ais isis (8 reas east col ceaett ete e woit 12) yet 3 mee SST se Bm 9 me me Toe Bes ms II ere se eee A ee iy, em 7 8 38 wt |8 OS 5 SHC Sa i Bore [27 1781S moa (52 30 ts Sa 3 os ®t te 2 om 8 8 & 97 & 3% 6 ee) he ee ea aes si a Seon 25 27 6 2580 7 2580 hess 0» mm oS mT 2 um BoM OD 5519135 tsss30 0338 M390 © e741 16m 2 ee 8S Ie MISH % oor 46 ism 47 Ms SG Lae so iseas Sl 138552186 “Tanus 6 Second Data Set for SaLaP-2, n 6. Benchmark Data Set for SALBP-G. We consider SALBP-G which is to minimize the sum of idle times m-c ~t,., and give a data set with 256 instances containing the precedence graphs of Table 1. Each instance is charac- terized by the name of the graph and a range [m, fi] of possible values of m. Optimal soli tions are described by the parameters idle*, m, and e* denoting the minimal total idle time, the corresponding number of stations and the cycle time, respectively. If idle* is not yet ‘known, an interval is given If several optimal combinations of m and c exist, one is arbitrari- ly chosen. Mans [3 5 1 3 0 45 6 5 7 Bomn | 30509038 45 0 5s 4 Pm ip oon oe Aes ae 0 5 7 8 $ 8 67 Ww 6 8 wien | 37 2G 4702 4 2 Seay ee 1G) Manor | 35 2 oo ee Mice | 3 99 9 3 38 49 09 5 a Cs To 8 Rosters | 30 13 ae so 1 6 4 ws 5 0 ts 0 fee 52 100) 0s 2g oe ose 5 es oe meas a 3 ‘at ial 8 0 8 8 9 mo 9 6 Buy SO soe 4 1 os 6 Ho 6 Sete a) Oe ee 8 al 10 ise 10, ob 2 B® flere ee) eo ay eo ee 8 5B 1 8 6 6B 4 8 at 6 7 5 7 wt 5 1 90 9 ie ie 8 os ge eo ate 1 Ro ee su sun 6 1 B66 6 7 us 7086 so wm 8 es 1 1m 191536 Come oe ier oe ai 5 3B 2 $F 6 Bb 3 9 6 8 om 6 Bm o Bo 50 eee 6 ae ta ise ea! itis oo. 3 4 os is 5 0 6 OD 3 ° 8 @ ot ee sy el 10 ne Hahn Be a eae ar Set ee ese ral 6 8 mH 6 UM i eer 2 Name 2 8 et mt ot 2 det ome et Wonede | 3 30 0 3 Oe ol se 0s eo al i 3 862 0 BoB 0 no 6 is 2 15 108 nets oe eo ee 9» 3% RD mo 3 » 2n 7 mo oy 2» 6 OB ‘Tooge a 4B 0 sm ol oe aa eae ea ee a Mee el ge a Misae te: asta) 6 7 2% Oe 6 3 ow DB & 19 16 yr) ey a ee Weemag [3 38 13 4B 1 4 oe eet 6 8 1 6 2% eet 0 ee ae oh ial nn ieee) a 6) BR 1 ob jm 5 6 9% i D5 6 6 B56 oe Po oe of 2% 9 2 % 8 Im 3% 6 26 a % 8 os 2 2 2 2M os 2 Ow 35 35 8 i Avast | 3 3 4 ou 1 4 ieee 5 5 6 2 ob 6 RED 7 7 ee al 2 98 2 1102 er 8 8 no Be. wD un 1B Mom ae 14 aad 15 5 is 83S SIO 16 0 9 sane ate a a ete see ce a 7 3 o 8 5 0 8 0 2 ty) ese elas os) 15 5 ile 8 » B® 8 » 1 20 20 m % 9 wD a a a 3 is 5 2 % eee i ai is 30 3 a 36 a 24 8 4 2 2 Hy 4% 8 6 % 1 Late 3 3 3 4B 0 4 an 3 5 6B 6 0 16 3 7 89 2 9 Be " 5 now 8 ou 2 5 ey a ey 8 6 BB ROG re » i Name 2 idee mt 2 8 ide om ot ews? | 3 27 3 50133 4m 0 9 emt 43 4 mao |s s 1 5 sooo 6 7 6 25067 see ee inn 0 16 nu Bes | 10 27 be A363 no 1 ues | 16 9 ise BIS te re B16 1S 1036 3 82 19 Te 2 186 % SO 2 a9 207 2 8D 3B pn as nou 2 3s 27 sSas-tts7 265856, Bar 3 = a. in fH 4 tS bo 1 4 nes a a 7 1 7 Bas 60 S68 1 6 2 40 i re) 4 Gog Par 3 f A ee us bold 2 é 1 7 o 5 9 an 5 toon Bs B of 2 38 B ss ee a wo 6 OS 0 3 6 «mR Dm 4 % 1B ma gon Pe py me se ist i ase ts Bo is me Ru ob Hs “6 Boo 6] tt ne 7 ® ees ee a im ie. a ol 62 6s 8 9 “8 Mw “9 (3 & # es 7 4 6 8 8 & B a Sebo aa a a so re 67 207 5m 5 9 THO os 2 om NOs 0 om aes a Nae oe te on sf ges % 7 9 6 435 % % $5 1B 3 0 S$ SaRs 2% nD B a ak in a se) ms 2 Homer mo 2% % same Se go 2.0 58 5 oan 2 48 8 eas tig |e) fat ies es ee Po pd) 2 5 mos 3 1@T Bs ae 4% Gear SoH St isa 7 8 458 7 aM 50 St 4549550 “ References Aros, AL (1960 A aay ofa compte med of segecngasetly ln penis. PAD. dissertation, University of California, Berkely. 2 : Baler, KR (1974): lazoduction to sequracng and scedog, Wiley, New York Barhold, JJ. (1993): Balmcing twosided asembly lines: a ease study. Intemational Joural of ‘Prodiction Research 31, 247 » 2461 . Baybars, L. (1986), A survey of exact algorithms for the simple assembly line balancing problem. “Management Science 32, 909-932. Bhatachajee, TK. and S. Sahu (190): Complexity of single model assembly line blancing problems, “Eaghocrng Casts and Production Beonmics 18, 203-214 Bowman, EH, (1960); Assembly tine balancing by linear programming, Operations Research 8 385 - 39, Bacey, GM. (1974 Assembly line balancing with mutiple stations. Management Science 20, 1010 - 1 Dar El, EM. (1973): MALE - a beursc technique for balancing large single-maelasembly ines. ATE. ‘Transactions 5, 33-356 Domchie, W:: A. Scholl and 8. VoB (1993): Protktcasplanung ~ Ablauforgaistrische Aspe ‘Springer, Bett. RE; GD. Jolwson and RS. Peterson (1983): Curatly practiced formulations fer the assembly line blanceprobem. Jounal of Operations Management 3, 209-21. Haln. R. (972): Proubsonsplanung bei Linicfenigung. de Gre, Bedi. Hsia, H, (1968): heuristic method for balancing asembly lines. The Westem Electric Engineer 128, 9-16 Hofinane, TR. (1850): Assesbly ine balancing: a st of cballenging problems. Intenational Journal of ‘Production Research 28, 1807-1815. Hoffman, TR. (1982): EUREKA: a bysid sytem for assembly line blancng. Management Science 38, 94, Gur oelson J.R (1956): A computing procedure for line balancing problem. Managemen: Scence 2, 261 ~ Hi. Jocschte, 6. (1964): Branching and Bounding: Eine aligemeine Methode rr Liming tembnstcacher ‘Probleme. Ablauf- und Planungsforchung 5,133 155. ofnson, RY. (1988): Optimally balancing large assembly lines with "FABLE". Management Seicace 34, "240-253, Kap, RM, (1972): Reducbilty among combinsocal problems, Ia: Miller, RE. and JW, Thatcher (eds): Complexity of computer cmpatton, Plenum Press, New York, 85-103. Kilridge, MD. and L. Wester (1962): A review of analytical qstems of line balancing. Options Research 10, 626 - 638 Lag, L. (1974: Abtaken von Mootaglinin, Krauskept, Mainz ‘Mansoor, EM. (1964): Asembly line balancing ~ an improvement co the ranked positional weight ‘ectnigu. Journal of Industial Engineering 15, 73-77 and 322-323 Masor, 4A. (1970): An experimental invesignion and comparative evaluation of production line Tulancing techniques, Management Seence 16, 728-746. Merens, P, (1967): FlieBbandabsiounung mit dem Verfahren der begenaiza Enuneruion auch Mller- ‘Mesbach Ablauf-und Planngsforcimg 8, £29 433. Mitchell J, (1957). compattional procedure for balancing zoned assembly lines. Research Report ‘694601-1-83, Westinghouse Research Laboratories, Pitsburg 1s Matberjee, SKC and SK. Basu (1968) An application of bewistic method of asembly line balancing in ‘an Ian industry. Proc. asm, Mech, Eng 78/11, 27 = 292 Rosenberg, O. ond H. Ziegler (1992: A comparison of hes algorithms for cxt-rinted sembly line ‘balancing Zeitschrift fr Operations Research 36, 477-435 Sowyer, LILF. (1970: Line balancing. The Machinery Publishing, Brighton Talbot, FB: LH. Paterson and W.V. Gebviein (1985: A compare evaluation of teristic line balan ‘cing tchnigus. Management Science 32,430 438, Tonge, FM, (1960: Summary ofa heuristic line balancing procedure, Management Science 7, 21-29. Tonge, FM. (1961: A heuristic progam for asembly line balancing. Prentice Hall Inc, Englewood ‘tits, ‘Warnecke, J (1971: Anwendung mathematscher Methoden bel der Leltungubstimmang ven Moaage- Tinie, CIRP Annals 20, 9-100, Wee, TS. and MJ, Magazine (1961): An efficient branch and bound algvthm for an assembly line ‘balaelog problem = part I: maximize the prodstion rate. Working Pape NO. 151, University of ‘Wateron, Wateroo, 19 Bartholdi 2: 148 tasks, Bartholdi (1993), modified operation times It difers from Bartholdi 1 only with respect to the operation times. In order to complicate the problem, the operation times are computed by t, mod 100 (the last two digs of the original times are used). ‘Bowman: 8 tasks, Bowman (1960) 2 Gunther: 35 tasks, assembly of an auto engine cradle, Gunther et al. (1983) pp. 47) a Heskiaofl: 28 tasks, reconditioning of telephone sets, Heskiao (1968), modified as inthe data set of Talbot etal. (the are (1,2) was originally contained) eckson: 11 tasks, Jackson (1956) Facschke: 9 task, Jacichke (1964) 2 Kilbrid: 45 tasks, Kilbridge and Wester (1962) Lutz 2: 89 tasks, modification of Lutz 3 with operation times :=j mod 10, tasks with f,=0 get the value t,= 10 Lutz 3: 89 tasks, assembly of a refrigerator, Lutz (1974, p. 18-119), precedence constraints as Lute 2 ‘Task times: m ‘Mansoor: 11 tasks, Mansoor (1964) Mertens: 7 tasks, Mertens (1967) ‘Malkherje: 94 tasks, assembly of an electronic consumer good in an Indian manufsctury, ‘Mukherjee and Basu (1964) ‘Task times: Precedence constraints: pairs ij for task i preceding task j 12 13 la 1s is in 18 19 vio air 32 gil alg SSB 318 ate ou) i) Te 2 a8) au 1088 the Bs 1433 wl wea rr 1038 1439s 675 rE 2078 no ss 2138 2140 nn ms BM mas 2437 2568 Bm BM (AS 2536 ne Ws HSS 86 pe ee ee vrs as) az 3085S TH Ro SSH TE Ce ee ee a ay Ge 4a TS ass aes Har Bs Bm ONS SBE SAT SB 377 Serl 5796S] Sw GLTT en GT OM 6m 66 6676 OM | RD (7 Oe OT 0 ia ea 74RD 753076277} 7RBD UBL wag $283 «38483858486 SRG RGBT | BTRB RA. $39 «e203 82H Sawyer: 30 tasks, Sawyer (1970, p. 38), modified asin the data set of Talbot et al. (the ares (1,11) and (6,10) were originally contained) ‘Scholl: 297 tasks, assembly of an engine ‘Task times: 101 mt 2 re ta ret ist m 1a 11 2 21 ma Bi 2a 231 261 ™m a ‘Precedence constraints: pair i,j for task {preceding task 12 438 414 610 150 110 36 198 ot 23 456 az TAL 83 95 16 1933, 1386, aa 43 4207 ug B8asa8 101 356 Baeeve! BeaBERee8eezass si6 a 10 a 130 150 184 10 BsSeeBsae¥s e38 21 130 32 aan 17818), 18h 186 11116 u72s7 ia3n4s 130 18153 150,161, 135165 168176 war 7981 18187 e498 104108 108,120 115,20 129,41 134,36 14a 146156 154.166 reer e574 175180 181183, 81197 188195 195203 200202 205207 21213 218219 6228 2128 238240 243245 24820 255261 26120 rez 215230 233293 791293 105110 nang e122 121128 70174 176181 B14 181295 189195 195205 212m 206009 212316 219200 2330 232036 28s 43246 ost 255262 2220 zat 276281, 1o6112 g20 Hn 123329 125130 waa 136139 waLAsi 168159 156.166 16467 wre W781 181185 12.190 ‘94195 19527 2204 207210 sais 28 Warnecke: 58 tasks, assembly of a miniature camera, Warnecke (1971), operation times randomly generated, because they are not reported inthe reference SScuIFTEN ZUR QUANTITATIVEN BETRIEBSWIRTSCHAFTSUEHRE ‘Technische Hochschule Darmstadt 195 1195 Swater, Hartmut: Hirarchische Prodltionsplanang. Mai 1995, ersclenen in: Kem Wo; Hl Seiroder und J. Weber (Hg): Handnérterbuch der Produtionsvirtchaft 2 Aufl, Schfer~ Poescheh, Stiga, Sp. 631-652, 2/95, Und, Hans Sagem und Amin Scholl rzcuging von Testaten fr Reihenfolggbleme bel Valan- ‘entering ni 1995 3795 Sonderge, Late und Stfan Vol: A Star-shaped Diverication Approach in Tabu Seach Jn 1995, erschienen ine Osman, LH, und J.P. Kelly (Hie. Mta-Hewrisies: Theory & Applications, Kroner, Boston, 1996, 8 489- S02 4195 Stull, Armin, Robert Klin und Wolfgang Dome: Patten Based Voestalary Building for Effectively Sequencing Mixed Model Asembly Lines, August 1995. 5195 Amberg, Anita; Wolfgang Domuchke und Stan Vo: Capectted Minimum Spaning Tees: Algo- rithms Using lteligent Search, August 1995, ershienn in: Combinatorial Optimition: Theory ‘and Protce 1, 1996 8. 9-39. 6195 Stranger, Sosame: Zum Begriff des Metamodll. September 1995, erschlewen als: W1-Depot- Beitrag 963, Winschafsinformatk 39/5, 1996 8. 545. 7195 Domschie, Wolfgang: Robert Klein und Armin Schall: Aniipative Leisungsbstimmung bei ‘modererVariantenficbfeigang. Oktober 1985, erschlenen in: Zeitschrift fir Betritwischaft 66 (1906), Het 12 $.1465- 1491, 18/95 Sinvinger, Saanne: Fine lardlnliatshezopene Erweiterung der Entity Relationship Modelienng. Older 1995. 9195 Siadir, Hartmat: Reforulatons of the Shortest Route Model foe Dynamic Mi Level Capactted Letting. September 1995 1996 1196 Dems, Wolfgang und Gabvicla Kelsi: Location and Layout Panning - A Survey. lms 1996, cerscheint in: OR Spelaram. 2/96 Strabrnger, Susan: Konzept metamodelbasieer Methodenergleshe. Febru 1996, 3796 Schell, Armin; Robert Klein und Chvisian Sirgen: BISON: A. Fast Hybrid Frooedure for Exaetly Salvng the One-Dimensional Bin Packing Problem, Mire 1996 erschlenen in: Computers & Opera tions Research 24 8, 627-648. 4796 Domschkc, Wolfgang; Robert Kicia und Armin Scholl: Tabu Search - Eine itsligente LAsungstre= tegie fr komplexe Optimerungprobleme, April 1996, erschlenen In: WISE 25 (1996), Heft 12S. 605-610 5196 Stadler, Hartmut: On te Equivalence of LP Bounds Provided by the Shortest ous and the Simple Plant Location Mode! Formulation fe Dynamic Mult-Rem Multi-Level Catitated Lot-Sizing. Mai 1956, (496 Domschkc, Wolfgang; Robert Klein wad Armin Scholl: Tebu Scare - Durch Vetbte zum scnelen Erfolg. August 1996, erschienen als: Tlssche Tobus ~ Tabu Search: Dursh Verbte scneller ‘optiiere. et - Magazin fr Computerteinik, Heft 12 (1996S. 326-382 7196 Stale, Hartt: Optimal Dimensions for Automated Strage/Retioval Systems. uli 1996, cerscheit in: Proceedings of 196 International Maeral Handling Research Coioguiam. 1997 197 Scholl, Armic; Cabvicla Krispin; Robert Klein uad Wollfng Domichte: ranch and Round Optimieren auf Blumen: je beschrinker,desto besser Sante 1997, erchet te = Magan fr Conputencchi, 2/97 Schall, Arm; Wolfgang Damschis und Robert Klin: Logistik: Aufgaten der Tournplnog. Februar 1997, erachein in ISU, 3/97 Schall, Armin; Wolfgang Damschke und Robert Kien: Lops: Methoen der Toueaplanang Februar 1987, ercheint in WISU. 4/97 Kicin, Robert md Armin Scholl: Computing Lower Bounds by Destictie Improvement = an Applicaton to Resouree-Consrained Projet Scheduling. Marz 1997 5197 Schall, Armin und Robert Kicin: Balancing Assembly Lines Effectively ~ a Computation Comparison. Apel 1997, (197 Schall, Armin und Robert Klin: Improving the Efficiency of JIT Assembly Lines the U-Line ‘Balancing Problem. Mai 1997 ‘IST Sebell, Armin und Robert Kien: lnvestonspoyrammplaning tei Obelix GnbHt & Co, KO, Jul 197,

You might also like