Activity on Arrow (AOA) network model to enable mold productivity optimization Lim Ming Siong Infineon Technologies (Malaysia) Sdn. Bhd. Free Trade Zone, Batu Berendam, 7535 Melaka MingSiong.Lim@infineon.com Abstract This paper is to discuss an engineering optimization method to fully utilize the mold equipment and deliver its maximum productivity yet no compromising on both product and equipment quality. A network model, which using activity on arrow (AOA), is constructed for the activities in the entire molding process starting with loading till offloading of the magazine. The network becomes complicated when 2 mold presses are in production mode simultaneously where only a universal toolset, such as er and degator is required to accommodate both presses concurrently. In this case, slack times often happen where certain activities need to wait for the activities to complete before it starts. This non-productive waiting time is one of the seven wastes in lean manufacturing concept. A dummy is used to indicate as a precedence activity of another activity in this network model. By considering all the activities time of the entire network, the critical path can be identified. Critical path illustrate focus area of the entire mold process which reflecting the opportunity for productivity improvement. By tabulating down the time for each activity in the critical path, a decision can be made when the opportunity is identified. The model can be further expanded to 3 presses model where the critical path and the slack time vary accordingly. A simple program are developed with entry data of the mold equipment capability and process parameter to ease the calculation. The put of the program will present the critical path and expected crashing area 1. Introduction Mold is one of the common bottle neck processes for the molded package manufacturing line. Unlike other assembly processes such as die bond and wire bond which processing products in unit form; mold able to provide higher put due to its ability to process products in batch form. As such, the ratio of numbers of mold equipment to other assemblyprocess equipment is relatively low. However, the mold equipment cost investment is extremely higher as compare to other equipments. A higher floorspace needed due to the requirement of mold system design also a major consideration for any capacity expansion. Productivity improvement appeared to be a better choice for investment avoidance and achieving favourable equipment cost of ownership. However, due to modular design of the mold equipment, the mold productivity varies according to products configuration on the mold equipment. Product configuration include the process time (value-added time) and machine time (operation value-added time). Different package design and platform such as mold compound material, package thickness, leadframe material, size, density and mold system design would heavily influence the mold product configuration. These differences make the bottle neck of the mold system keep on changing according to mold configuration. The corresponding effect leads to difficulty to identify the opportunity for productivity improvement and establishing the optimum mold process and setting maximize the throughput. Situation become more critical for low volume high variety manufacturing model. Due to highly volatile market demand especially on smart mobile phone industry, the volume demand fluctuates tremendously from product to product and from time to time. An investment on mold equipment might ended up having severe impact on capital cost if the volume demand deviate significantly from the forecast. An optimum mold configuration is needed to cater such complex manufacturing model. Conventional method using trial and error to identify the opportunity is very time and resource consuming. A fast and agile method such as using a network model to detect the opportunity is needed. This will maximize productivity improvement and minize resources needed. Network model is normally used for project scheduling and controlling. It is a very effective method to manage complex project and to ensure the project success within the timeline. There are many types of network diagrams but the classical ones are the activity on arrow (AOA) and the activity on node (AON). PERT (project or program evaluation research technique) is the best-known AOA-type diagram and is the historical basis of all network diagramming. It was developed in the 195s for the Department of the Navy to aid in monitoring and tracking the very large, very complex Polaris program [1]. The approach of using AOA network model in mold system is one of the innovative approach for mold productivity improvement. 2. Methodology- single press configuration The pre-requisites to use AOA network for this mold productivity improvement are:- 1) The time including the tolerances for each activity is reliable. These data can be found through specification, experience, qualification and actual time study validation. An in-depth knowledge on specific process or machine time is needed to tabulate each activity time. 2) There are no close-loop control configured on molding system to compensate the slack time. There are total 9 steps required to optimize the mold productivity. Figure 1 show the overall flow of the 9 steps. 37 th International Electronic Manufacturing Technology Conference, 216
Figure 1: 9 steps using AOA network model Step 1, 2 and 3 is to define the activities and the respective time from start till end of the molding. A typical activity table is illustrated in Figure 2. The optimistic and pessimistic time is referenced to the possible min and max that activity might fluctuate due to process variation. These time shall be considered process window or specification that will not cause any deviation on product quality and machine performance. If possible, group the activities on process and machine time. The reason of grouping is to visuallize the percentage of value-added and operation value-added time. This also help to clearly visualize whether the improvement shall be done on process or machine which typically is handled by different parties or personals. Optimistic Pessimistic 2 8 Yes in 18 15 21 Yes Leadframe 3 27 33 Yes Loading 33 3 36 Yes 1 5 15 12 18 Yes Curing 1 8 12 Yes Unloading 7 4 1 11 17 Yes Cooling 1 5 15 4 1 Yes 14 11 17 Yes Magazine 5 2 8 Figure 2: Time taken for each activity After each activity time is tabulated. Continue step 4, 5 and 6. A typical example is shown in Figure 3 for 1 press mold configuration while the corresponding put is shown in Figure 4. The solid line arrow represent activity while the dotted line arrow represent dummy activity. A dummy activity is used as a precedence before the next activity can start. The green coloured box represent the slack time which means we have idling time of this activity while waiting for the activity before that to complete. The box with any colour means we do not have any slack time,. From the network, continue step 7 to identify the critical path. Critical path is the path for these activities with any slack time. These activities time is directly impacted the overall cycle time. All the activities in critical path are considered opportunity for improvement. If improvement done on the time for any activity in the critical path, the slack time for other activity will be reduced. This will fully utilize the whole mold equipment. For step 8, start with identifying the opportunity on AOA network. If we pareto the activity time with the tolerance included, we can see that cooling time is the higher opportunity since it has the highest tolerance. However, cooling time is not the activity in critical path. Improvement on cooling time has less impact to the overall cycle time as shown in Figure 5. Improvement on cure time however, show higher productivity improvement as illustrated in Figure 6. The improvement may continue till step 9 by keep on optimizing until the productivity is maximized. One thing need to take note changes of the activity time might lead to certain set back. The corresponding verification needed to be performed to ensure the set back is well controlled. 3. Methodology-Multiple presses configuration For volatile market demand scenario, mold configuration might change in a very short time. The scenario of conversion to multiple presses during high demand season and back to single press during low demand season might happen. For this manufacturing model, the improvement as demonstrated in section 2 might not help. To visualize the effect, the AOA model need to further expand to multiple presses using step 4, 5 and 6. A typical example is shown in Figure 7 which is using the improved activity time from section 2. The red coloured box indicate the insufficient time to complete the activity. The existing AOA model cannot be used for this activity time setting. Therefore, the optimization need to be continued by identifying the new critical path. Continue setp 7,8 and 9 to optimize the productivity for 2 presses configuration. Figure 8 show the example of improvement on cure time and cooling time. This improvement is also reflected in single press configuration. Similarly when the improved setting loaded into 3 presses configuration, the AOA model show that the improvement cannot work as illustrated in Figure 9. Same approach is applied to optimize the activity time. Figure 1 show the overall improvement after improvement on loading, inmold, curing, cooling and unloading time on 3 presses configuration. Significant improvement also can 37 th International Electronic Manufacturing Technology Conference, 216
be seen if the same setting loaded into single and double presses configuration. 4. Discussion The AOA network model able to demonstrate few informations which is faster and clearer compare to conventional trial and error approach. AOA network model offer a pictorial chart for user to visuallize the opportunity by illustrating:- a) Slack time for each activity b) Critical path which is the bottleneck of the entire complex molding process c) Changes on slack time, network and productivity during optimisation Similar approach can be used in very complex manufacturing model. This can avoid unneccessary investment or resources consumption when market demand change later. For example, a development engineer can qualify a molding process parameter using optimized 3 mold presses configuration at the beginning stage. Manufacturing engineer can improve productivity by identifying opportunities present on existing complex manufacturing model. 5. Conclusions AOA network model demonstrate a fast and agile method for productivity optimization. However, the importance of the pre-requisites shall not be neglected. Optimizing the network model with considering the impact of the changes on activity time might lead to other side effects. To ensure the pre-requisites are fulfilled, the determination of tolerance time shall cover the variation factors which include material properties variation and machine condition. Also, the machine design that influence activity time such as built-in closed-loop control will affect the network model. Acknowledgments The author would like to thank to the management for their support and opportunity to publish this paper. Appreciation also goes to colleagues and supporting staffs for all the development and evaluation work. References 1. James C Taylor and PMP, Project scheduling and cost control-planning, monitoring and controlling the baseline J. Ross Publishing, pp. 62-67, 28. 37 th International Electronic Manufacturing Technology Conference, 216
Mag in 158 344 344 5 5 5 18 23 23 3 53 53 33 86 86 1 96 96 15 111 111 1 211 211 7 218 218 14 232 232 1 332 49 7 339 683 14 353 697 79 172 179 53 184 18 71 22 3 11 232 33 265 265 1 275 275 15 29 29 1 39 39 7 397 397 14 411 411 1 511 59 7 518 69 14 53 2 711 262 262 262 Mag 11 363 18 119 381 3 149 411 33 444 444 1 454 454 15 469 469 1 569 569 7 576 576 14 59 59 1 69 69 7 69 7 697 14 711 711 5 716 716 Figure 3: AOA network model for 1 press configuration Item Preset data 1 press Number ofstrips 14 Number ofshots 7 Shot in press1 7 Shot in press2 Total processing time 1432 UPH 35.2 Unit cycle time 1432 Figure 4: Expected productivity Mag in 258 344 344 5 5 5 18 23 23 3 53 53 33 86 86 1 96 96 15 111 111 1 211 211 7 218 218 14 232 232 5 282 54 7 289 633 14 33 647 129 172 179 53 184 18 71 22 3 11 232 33 265 265 1 275 275 15 29 29 1 39 39 7 397 397 14 411 411 5 461 59 7 468 64 14 482 661 262 262 262 Mag 11 363 18 119 381 3 149 411 33 444 444 1 454 454 15 469 469 1 569 569 7 576 576 14 59 59 5 64 64 7 647 647 14 661 661 5 666 666 Yes in 18 Yes Leadframe 3 Yes Loading 33 Yes 1 Yes Curing 1 Yes Unloading 7 Yes Cooling 5 Yes 14 Yes Magazine 5 Figure 5: Improvement on cooling time Item Preset data 1 press Number of strips 14 Number of shots 7 Shot in press1 7 Shot in press2 Total processingtime 1382 UPH 36.47 Unit cycle time 1382 Mag in 118 34 34 5 5 5 18 23 2 3 3 53 53 33 86 86 1 96 96 15 111 111 8 19 1 191 7 198 198 14 212 212 1 312 43 7 31 9 623 14 33 3 637 111 111 111 59 152 159 53 164 18 71 182 3 11 212 33 245 245 1 255 255 15 27 27 8 35 35 7 357 357 14 371 371 1 47 1 53 7 47 8 63 14 492 651 222 222 222 Mag 11 323 18 119 341 3 149 371 33 44 44 1 41 4 414 15 429 429 8 59 59 7 516 516 14 53 53 1 63 63 7 637 637 14 651 651 5 656 656 Yes in 18 Yes Leadframe 3 Yes Loading 33 Yes 1 Yes Curing 8 Yes Unloading 7 Yes Cooling 1 Yes 14 Yes Magazine 5 Figure 6: Improvement on curing time Item Preset data 1 press Number ofstrips 14 Number ofshots 7 Shot in press1 7 Shot in press2 Total processing time 1292 UPH 39.1 Unit cycle time 1292 37 th International Electronic Manufacturing Technology Conference, 216
Mag in -82 348 348 5 5 5 18 23 23 3 53 53 33 86 86 1 96 96 15 111 1 11 8 191 191 7 198 198 14 212 212 1 312 23 7 319 667 14 333 681 52 52 52 52 52 52 52 52 52-82 262 262 53 1 5 18 71 123 3 11 153 33 134 186 1 144 1 96 15 159 211 8 239 291 7 246 298 14 26 312 1 412 33 7 419 681 14 433 695 63 63 63-41 217 217 11 164 18 119 182 3 149 212 33 245 2 45 1 255 255 15 27 27 8 35 35 7 357 357 14 371 371 1 471 43 7 478 695 14 492 79 115 115 115 52 52 52 52 52 52-41 149 264 18 167 282 3 197 3 12 33 293 345 1 33 355 15 318 37 8 398 45 7 45 457 14 419 471 1 571 53 7 578 79 14 592 723 126 126 126 86 86 197 323 18 215 3 41 3 245 371 33 44 44 1 414 414 15 429 429 8 59 59 7 516 516 14 53 53 1 63 63 7 637 723 14 651 737 Figure 7: 2 presses mold configuration Mag 178 178 178 52 52 52 52 52 52 245 4 23 18 263 441 3 293 471 33 452 54 1 462 514 15 477 529 8 557 69 7 564 616 14 578 63 1 73 73 7 737 737 14 751 7 51 5 756 756 Mag in 118 298 298 5 5 5 18 23 23 3 53 53 33 86 86 1 96 96 15 111 111 8 191 191 7 198 198 14 212 212 5 262 38 7 269 567 14 283 581 2 2 2 2 2 2 2 2 2 118 262 262 53 55 18 71 73 3 11 13 33 134 136 1 144 146 15 159 161 8 239 241 7 246 248 14 26 262 5 312 43 7 319 581 14 333 595 63 63 63 59 167 167 11 164 18 119 182 3 149 212 33 245 245 1 255 255 15 27 27 8 35 35 7 357 357 14 371 371 5 421 48 7 428 595 14 442 69 65 65 65 2 2 2 2 2 2 59 149 214 18 167 232 3 197 262 33 293 295 1 33 35 15 318 32 8 398 4 7 45 47 14 419 421 5 471 53 7 478 69 14 492 623 126 126 126 36 36 197 323 18 215 341 3 245 371 33 44 44 1 414 414 15 429 429 8 59 59 7 516 516 14 53 53 5 58 58 7 587 623 14 61 637 Mag 128 128 128 2 2 2 2 2 2 245 373 18 263 391 3 293 421 33 452 454 1 462 464 15 477 479 8 557 559 7 564 566 14 578 58 5 63 63 7 637 637 14 651 651 5 656 656 Yes in 18 Yes Leadframe 3 Yes Loading 33 Yes 1 Yes Curing 8 Yes Unloading 7 Yes Cooling 5 Yes 14 Yes Magazine 5 Item Preset data 1 press 2 press Number ofstrips 14 Number ofshots 7 Shot in press1 7 4 Shot in press2 3 Total processing time 1242 765 UPH 4.58 65.88 Unit cycle time 1242 765 Figure 8: 2 presses configuration with improvements 37 th International Electronic Manufacturing Technology Conference, 216
Mag in 18 18 5 5 5 18 23 23 3 53 53 33 86 86 1 96 96 15 111 111 8 191 191 7 198 198 14 212 212 5 262 262 7 269 449 14 283 463 2nd s hot 2 2 2 2 2 2 2 2 2 144 144 53 55 18 71 73 3 11 13 33 134 136 1 144 146 15 159 161 8 239 241 7 246 248 14 26 262 5 312 312 7 319 463 14 333 477 3rd s hot 4 4 4 4 4 4 4 4 4 18 18 11 15 18 119 123 3 149 153 33 182 186 1 192 196 15 27 211 8 287 291 7 294 298 14 38 312 5 362 362 7 369 477 14 383 491 6 6 6-9 -9-9 -9-9 -9 72 72 149 155 18 167 173 3 197 23 33 245 236 1 255 246 15 27 261 8 35 341 7 357 348 14 371 362 5 412 412 7 419 491 14 433 55 8 8 8-7 -7-7 -7-7 -7 36 36 197 25 18 215 223 3 245 253 33 293 286 1 33 296 15 318 311 8 398 391 7 45 398 14 419 412 5 462 462 7 469 55 14 483 519 1 1 1-5 -5-5 -5-5 -5 Mag 245 255 18 263 273 3 293 33 33 341 336 1 351 346 15 366 361 8 446 441 7 453 448 14 467 462 5 512 512 7 519 519 14 533 533 5 538 538 Figure 9: 3 presses configuration Mag in 18 18 5 5 5 18 23 23 3 53 53 3 83 83 5 88 88 15 13 13 8 183 183 4 187 187 14 21 21 5 251 251 7 258 438 14 272 452 2 2 2 2 2 2 2 2 2 144 144 53 55 18 71 73 3 11 13 3 133 5 136 138 15 151 153 8 231 233 4 235 237 14 249 251 5 31 31 7 38 452 14 322 466 4 4 4 4 4 4 4 4 4 18 18 11 15 18 119 123 3 149 153 3 179 183 5 184 188 15 199 23 8 279 283 4 283 287 14 297 31 5 351 351 7 358 466 14 372 48 6 6 6 2 2 2 2 2 2 72 72 149 155 18 167 173 3 197 23 3 231 233 5 236 238 15 251 253 8 331 333 4 335 337 14 349 351 5 41 41 7 48 48 14 422 494 8 8 8 4 4 4 4 4 4 36 36 197 25 18 215 223 3 245 253 3 279 283 5 284 288 15 299 33 8 379 383 4 383 387 14 397 41 5 451 451 7 458 494 14 472 58 1 1 1 6 6 6 6 6 6 Mag 245 255 18 263 273 3 293 33 3 327 333 5 332 338 15 347 353 8 427 433 4 431 437 14 445 451 5 51 51 7 58 58 14 522 522 5 527 527 Yes in 18 Yes Leadframe 3 Yes Loading 3 Yes 5 Yes Curing 8 Yes Unloading 4 Yes Cooling 5 Yes 14 Yes Magazine 5 Item Preset data 1 press 2 press 3 press Number of strips 14 Number of shots 7 Shot in press1 7 4 3 Shot in press2 3 2 2 Total processingtime 1165 721 577 UPH 43.26 69.9 87.35 Unit cycle time 1165 721 577 Figure 1: 3 presses configuration with improvements 37 th International Electronic Manufacturing Technology Conference, 216