第五章 鎂合金板材之熱間擠製製程之開發研究
6.6 結論
1. 鎂合金隨著鋁合金的含量愈高，其擠製負荷愈高，因此 AZ80 棒材 之擠製負荷最高，依序為 AZ61 與 AZ31。
2. 棒材在相同條件擠製加工條件下，其擠製負荷隨著擠製比的增加而 提高。
3. 棒材在相同條件擠製加工條件下，擠製速度越低則擠製負荷越高，
其擠製速度 1 mm/s 之擠製負荷最高，其次為 3 mm/s、5 mm/s。
4. 在相同條件擠製條加工件下，AZ80 棒材之抗拉強度大幅高於 AZ61 與 AZ31 棒材約 30~35 MPa。
5. 在相同條件擠製加工條件下，擠製速度為 5 mm/s 之棒材抗拉強度 最高，其次為 1 mm/s、3 mm/s，其擠製速度對棒材抗拉強度會有影 響。
6. AZ31 鎂合金之熱間擠製加工，當材料加熱溫度達 380℃時，其抗 拉強度大幅地降低。
7. 抗拉強度隨材料加熱溫度的降低而提高，因此棒材欲要求極佳強度
的機械性質，應選擇材料可成形溫度之範圍的最低溫度進行擠製加
工，才能大幅提升棒材之機械性質。
8. 擠製加工所得之棒材具有晶粒細化之效果，可提升鎂合金的機械性 質，且晶粒大小會隨材料加熱溫度而改變，其材料加熱溫度越低，
則晶粒尺寸越細小。
第七章 結論
本計劃為三年之整合型計劃，其研究為鎂合金管材、板材與棒材 之熱間擠加工成形性及擠製後成品之機械抗拉分析探討，其所得之成 果如下：
1. 管材經壓平試驗，依據規範之要求均能通過標準，其 AZ31、AZ61 與 AZ80 管材破裂負荷分別可達 2416.20、2820.38 與 4104.42N 以 上。
2. AZ31、AZ61 與 AZ80 管材之抗拉強度之平均值分佈分別在 261.08
～296.20、263.97～299.44 與 327.36～336.24 MPa 之間。
3. 運用田口法之品質特性分析，可分別尋求出 AZ31、AZ61 與 AZ80 管材抗拉強度之最佳製程參數組合。
4. AZ61 板材之抗拉強度均比 AZ31 板材高，且模具半角之角度愈 大，所擠出之板材的抗拉強度亦較高。
5. 在高擠製比 35.9 之鎂合金板材進行熱間擠製加工，必須使用變速 法，才能擠製健全之板材。
6. 在相同擠製加工條件下，其 AZ61 板材之變速時機點均比 AZ31 板材高；隨著材料加熱溫度的提高其變速時機點愈低；模具半角 之角度愈小其變速時機點愈高。
7. 變速時機點之 Fuzzy prediction system 之預測誤差百分比都在 6.2
％以內，其誤差均在可接受範圍內。透過預測 model 將可以掌握 板材在熱間擠製加工時的變速時機點。
8. 棒材之抗拉強度隨著材料加熱溫度之降低而提高。
9. 棒材之抗拉強度隨著擠製比的提高而增加。
10. AZ31 鎂合金棒材之熱間擠製加工，當材料加熱溫度達 380℃時，
其抗拉強度大幅地降低。
11. AZ80 棒材之抗拉強度高於 AZ61 與 AZ31 棒材之抗拉強度約
30~35MPa。
參考文獻
01. R. F. Decker, “The Renaissance in Magnesium,” Advanced Mater &
Proc, Vol.154, 1998, pp.31-35.
02. T. Aida, H. Hatta and C. Ramesh, “Workability and Mechanical Properties of Lighter than Water Mg
2Li Alloy,” Proc. of the 3rd Inter.
Magnesium Confer., Manchester, 1996, pp.143-153.
03. N. Ogawa, M. Shiomi and K. Osakada, “Forming Limit of Magnesium Alloy at Elevated Temperatures for Precision Forging,”
International Journal of Machine Tools & Manufacture, Vol.42, 2002, pp.607–614.
04. 楊智超，“鎂合金材料特性及新製程開發，”工業材料雜誌 152 期，
72-80 頁，民國 88 年。
05. 林昇立，“塑性加工學”，新科技書局，民國 81 年。
06. 陳振華，”鎂合金”，化學工業出版社，民國 93 年。
07. B﹒L﹒Mordike and T﹒Ebert﹐”Magnesium Properties Application Potential” ﹐ Materials Science and Engineering, Vol.302, 2001, pp.37-45.
08. 范光堯，”機械成形技術於鎂合金材料的應用概況”，工業材料雜 誌 162 期，139-144 頁，民國 89 年。
09. B.W.Niebel and A. B.Draper 原著，李春輝譯，”金屬材料學”，科技 圖書股份有限公司，民國 74 年。
10. http://www.matweb.com/。
11. 呂戊辰，”鎂及其合金的表面處理工學”，傳勝出版社，民國 91 年。
12. B.W.Niebel and A. B.Draper 原著，李春輝譯，”金屬材料學”，科技 圖書股份有限公司，民國 74 年。
13. 蔡幸甫，”鎂合金成型產業現況及發展趨勢”，工業材料雜誌 211
期，82-85 頁，民國 93 年。
14. 莊錦川，”具備輕量化潛力的擠型鎂合金”，工業材料雜誌 180 期，
134-136 頁，民國 90 年。
15. 林景扶，”鎂合金安全作業”，工業材料雜誌 152 期，110-112 頁，
民國 88 年。
16. Friedrich, S. Schumann, “Research for a “new age of magnesium”in the automotive industry”, Journal of Materials Processing Technology, v117, 2001, pp.276-281.
17. Louis Braun, “Casting of Magnesium Alloys on Hot and Cold Chamber Die casting Machines”, International Magnesium Conference in Taipei ,R.O.C , pp.153-169.
18. Pekguleryuz, O. Mihriban, “Development of creep resistant magnesium diecasting alloys”, Materials Science forum, v350, 2000 , Proceedings of the 1
stNagaoka International Workshop on Magnesium Platform Science and Technology 2000 , Jul 27 – Jul 29 2000 , Nagaoka, Jpn, pp.131-140.
19. Anon, “Recycling must be carefully controlled”, Foundry Trade Journal, v173, n3551, Feb, 1999, pp.18-21.
20. G. Hanko, “Recycling automotive magnesium scrap”, JOM, v54, n2, Feb, 2002, pp.51-54.
21. S.H. Hsiang and J.L. Kuo, “An investigation on the hot extrusion process of magnesium alloy sheet”, Journal of Materials Processing Technology, v140, 2003, pp.6-12.
22. 金美敬， 『台灣自行車產業鎂合金材料應用探討』，工業材料 159 期，民國 89 年 3 月，91-96 頁。
23. N. Ogawa, M. Shiomi, K. Osakada, “Forming limit of magnesium
alloy at elevated temperatures for precision forging”, International
24. E. Doege, K. Dröder, “Sheet metal forming of magnesium wrought alloys – formability and process technology”, Journal of Materials Processing Technology, v115, 2001, pp.14-19.
25. H. Takuda, T. Yoshii, N. Hatta, “Finite-element analysis of formability of a magnesium-based alloy AZ31 sheet”, Journal of Materials Processing Technology v89-90, 1999, pp.135-140.
26. 『鎂合金之成形技術之應用』 ，模具技術資訊第 55 期，民國 88 年，
10 頁。
27. 馬寧元， 『鎂合金表面處理簡介』 ，鍛造第九卷第一期，民國 89 年 3 月，37-49 頁。
28. 楊金瑞、葉信宏，“鎂合金表面處理簡介”，工業材料 152 期，民 國 88 年 8 月，106-109 頁。
29. 陳明佑，“利用模糊目標規劃法求解多田口式多品質特性最佳化問 題”，民國 91 年。
30. 田口玄一，陳耀茂譯，“田口實驗計劃法”，滄海書局，民國 86 年。
31. Genichi Taguchi (Yuin Wu, technical editor for the English edition), Taguchi Methods / Design of Experiments,Dearborn MI / ASI Press, Tokyo.
32. Alan Wu, “Robust Design Using Taguchi Methods”, Workshop Manual, American Supplier Institute (ASI) , Version 3.0,(2001).
33. William Y. Fowlkes and Clyde M. Creveling, “Engineering Methods for Robust Product Design”, Addison Wesley(1998).
34. 蘇朝墩，“品質工程”，中華民國品質管理學會，民國 91 年。
35. 鄭燕琴，“田口品質工程技術理論與實務”，中華民國品質管理學 會，民國 82 年。
36. Taguchi and Genichi,“Introduction to quality engineering：designing
quality into products and processes, The Organization, Tokyo（1986）.
37. Huei-Huang Lee, “Taguchi methods： principles and practices of quality design”, Gau Lih Book CO.,LTD., Taiwan（2000）.
38. L.A. Zadeh, Fuzzy sets, Information and Control, Vol.8 (1965) 338-353.
39. M. C. Su and H. T. Chang, Machine learning：neural networks, fuzzy
systems and genetic algorithms, No.2, Chuan Hwa Book CO., LTD.,
Taiwan, 2006.
出席國際學術會議心得報告
心得報告(一)：8 th ESDA2006
向四海 教授 會議名稱：第八屆工程系統設計與分析之國際研討會
8
thInternational Conference on Engineering Systems Design and Analysis
發表論文題目：Study on the Mechanical Properties of AZ31 Magnesium Alloy Products Under Hot Extrusion Process
本 人 於 民 國 九 十 五 年 七 月 三 日 至 七 月 七 日 至 義 大 利 杜 林 (Turin,Torino)參加由 ASME、COREP 及 Politecnico Di Torino 等單位 所 共 同 主 辦 之 第 八 屆 工 程 系 統 設 計 與 分 析 之 國 際 研 討 會 (8
thInternational Conference on Engineering Systems Design and Analysis, ESDA2006). 此次之國際研討會共有 458 篇 論 文發表，總共分成 Advanced Energy Systems，Advanced Materials，Aerospace，Automation and Robotics，Automotive Systems，Bioengineering and Biomedical Technology，Dynamic Systems and Control，Fatigue and Fracture，Fluids Engineering，Heat Transfer，Manufacturing，Nanotechnology，Noise Control and Acoustics，System Engineering，Technology and Society，
Tribology 及 Symposium on Design and Analysis of Advanced Structures
等十九個主題，幾乎涵蓋了機械之全部領域。本人投稿之論文屬於 Manufacturing 主題中之 Plastic Deformation 方面，此一主題除了本人 投稿之塑性變形之外尚有 Machining and Welding、Nontraditional Manufacturing Techniques 、 Product/Process Development 及 Manufacturing Systems, Tolerancing and Measuring 等 。 在 Plastic Deformation 之發表會場其 Section chairman 為義大利籍之女性教 授，參與發表及討論之成員有來自伊朗、西班牙、日本、義大利、新 加坡及我國等之學者專家，彼此交換研究心得，除了瞭解其他國家之 學者在相關領域之研究外，也能藉由他人之詢問修正目前之研究方向 與內容，可說是獲益良多。參加研討會之成員來自世界各地，至少有 三十以上之國家之學者專家參與此一國際研討會。研討會之晚宴於七 月六日晚上在杜林之古城堡 Castello di Pavone 舉行，離研討會之會場 尚須坐一個多小時之巴士。晚宴前所有與會人員在城堡內之小廣場聚 集，除了表揚此次研討會主辦人員之辛勞外，ASME 及主辦單位決議 下次之 ESDA2008 之承辦國家為以色列，希望各位屆時也能多投稿。
晚宴開始上菜時間已經是晚上十時左右，吃完飯回到旅館已經是凌晨 一點半，首次領會到歐洲國家之宴會是如此的累人。此次為本人首次 以國科會之補助出國參與國際研討會，與本校工程學院於 2003 年所
th
相比後覺得由於發表之論文篇數甚多，對參與人員之照顧無法周全
外，會場之導引與標示也有改善之空間。
心得報告(二)：7 th APCMP2007
向四海 教授 會議名稱：第七屆亞太材料製程研討會
7
thAsia Pacific Conference on Materials Processing
發表論文題目：Investigation of the Influence of Process Parameters on Hot Extrusion of Magnesium Alloy Tubes
本人於民國九十五年十二月三日至十二月七日赴新加坡參加由 新加坡大學及南洋科技大學等單位所共同主辦之第七屆亞太材料製 程 研 討 會 (7
thAsia Pacific Conference on Materials Processing, 7
thAPCMP).此次研討會之主題在於對各種材料(包含金屬、陶瓷、高分 子、複合材料、半導體及生醫材料)之加工製程(熔煉、半固態成形、
粉末冶金成形、金屬成形、精密切削、微米及奈米加工等)之研究成 果發表及心得交換，共有 100 多篇論文發表。分成 4 個場所總共有 21 個 Sessions 發表，並有 4 個 Keynote speech。第一場的 Keynote speaker 是來自愛爾蘭的 Professor M.S.J Hashmi，他的演講題目為
“International Research Directions in Materials and Manufacturing
Processes＂，Professor Hsahmi 同時也是 Journal of Materials Processing
Technology 之主編；第二場的 Keynote speaker 是來自香港理工大學的
Optical Microstructures and Freeform Optics＂； 第三場的 Keynote speaker 是來自美國喬治亞理工學院的 Professor Steven Y. Liang 他的 演講題目為 “Micro and Nano Scale Mechanical Machining＂； 第四 場的 Keynote speaker 是來自香港中文大學的 Professor Michael Y.
Wang 他的演講題目為＂Level Set Method for Design of Functionally
Gradient Materials and Structures＂。本人投稿之論文被安排在 3 C
Session 第 一 位 發 表 ， 主 持 人 為 日 本 Oyama National College of
Technology 的 Professor H.Watari，發表及討論時間為二十分鐘，此一
Session 除了本人投稿之有關鎂合金管材之熱間擠製之製程參數探討
之外，尚有鎂合金之 Twin Roll Casting、Ultrasonic Vibration-Assisted
Tapping、Powder Injection Moulding 等論文發表。參與發表及討論之
成員有來自中國大陸、日本、英國、新加坡及我國等之學者專家，彼
此交換研究心得，除了瞭解其他國家之學者在相關領域之研究外，也
能藉由他人之詢問修正目前之研究方向與內容，可說是獲益良多。參
加研討會之成員來自世界各地，至少有十個以上之國家之學者專家參
與此一國際研討會。研討會之晚宴於十二月五日晚上舉行，並聽取
Journal of Materials Processing Technology 之前主編 Professor Frank Travis
之近況報告。除了表揚此次研討會主辦人員之辛勞外，主辦單位決議
下次之 APCMP 之承辦單位為中國之廣州大學，希望各位屆時也能多
投稿。大會也安排十二月六日參觀位於新加坡南洋科技大學校區內之
Singapore Institute of Manufacturing Technology(SIMTech)，此一機構與
我國之工研院相似，以政府之部分補助進行研究發展之工作。此次為
本人第二次以國科會之補助出國參與國際研討會，此次之研討會與本
校工程學院於 2003 年所主辦之 6thAPCMP(6
thAsia Pacific Conference
on Materials Processing) 之規模相比，在 2003 年本校主辦時之參與人
數較多，對與會人員之照顧也較好。
心得報告(三)：10 th AMPT2008
向四海 教授 會議名稱：第十屆先進材料與加工技術國際研討會
The 10th International Conference on Advances in Materials and Processing Technologies
發表論文題目：
(Ⅰ) Application of Fuzzy Theory to Predict Deformation Behaviors of Magnesium Alloy Sheets under Hot Extrusion
(Ⅱ) A study of improving warped characteristics during flange forging
本人於民國九十六年十月七日至十月十二日赴韓國大田(Daejeon) 參加由韓國高等科技學院(KAIST)所主辦之第十屆先進材料與加工技 術 國 際 研 討 會 (The 10th International Conference on Advances in Materials and Processing Technologies, 10th AMPT)。此次研討會之主 題 分 為 Deformation Processing 、 Forming in Melt or Near Melt Condition、Powder Forming、Laser Processing、Processing of New and Advanced Materials、Surface Engineering與Nano Technology等七個領 域，共有208篇研究成果論文發表。
研討會之會場分成4個會場總共分為36個Sessions發表，並有2場
Keynote speech。第一場的 Keynote speaker 是來自韓國POSCO的 Dr.
O. J. Kwon，他的演講題目為“Steel－what we have to know about additionally”；第二場的Keynote speaker 是來自韓國現代汽車公司 (Hyundai Motor Comp.)的Dr. M. S. Kwon，Dr. Kwon的演講題目為
“Hyundai-Kia Motor, Towards future global leadership”。
本 人 所 發 表 2 篇 之 論 文 領 域 皆 為 Deformation Processing 中 之
Extrusion Processes 與 Forging Processes，其兩篇論文名稱分別為
Application of Fuzzy Theory to Predict Deformation Behaviors of
Magnesium Alloy Sheets under Hot Extrusion 與 A study of improving
warped characteristics during flange forging ， 其 論 文 編 號 分 別 為
AMPT-098與AMPT-099。第一篇論文(AMPT-099)被安排在A 1 Session
第 三 位 發 表 ， 此 一 Session 本 應 由 德 國 University of
Erlangen-Nuremberg 的 Professor U. Engel與本人共同主持，但由於
Professor U. Engel缺席且臨時代理的韓國Professor J.R. Cho教授也未
到場，故全場均由本人當主持人。此一Session除了本人投稿之有關鋁
合金凸緣鍛造加工之變形特性探討之外，尚有鍛造模具破裂之預測及
防止破裂之研究、鈦合金在等溫與非等溫鍛造之研究、運用有限元素
分 析 法 進 行 鍛 造 相 關 加 工 之 研 究 等 論 文 發 表 。 另 外 第 二 篇 論 文
(AMPT-098)被安排在A 5 Session第三位發表，主持人由德國 Institute
of Forming Technology and Lightweight Construction的Dr. M. Schikorra
與日本電通大學(University of Electro-Communications)的Professor M.
Murata共同主持，此一Session除了本人投稿之有關鎂合金板材擠製加 工之研究之外，其他尚有探討鋁合金擠製之微結構組織與晶粒尺寸分 布之論文、在管中加入肋板之新擠製加工方法之論文、參數的選擇對 於對稱擠製加工之溫度分佈探討等論文，論文發表及討論時間總計為 二十分鐘，參與發表及討論之成員有來自德國、中國大陸、韓國、日 本、伊朗、奈及利亞、沙烏地阿拉伯、波蘭、澳大利亞及我國等之學 者專家，彼此交換研究心得，除了瞭解其他國家之學者在相關領域之 研究外，也能藉由他人之詢問修正目前之研究方向與內容，可說是獲 益良多。參加本次研討會之成員來自世界各地，至少有二十個以上之 國家之學者專家參與此一國際研討會。
研討會之晚宴於十月九日晚上舉行，並安排韓國傳統表演，且由 研討會主席Professor Y. T. Im 表揚參與此次研討會籌備及主辦人員 之辛勞外，並由AMPT之組織委員會頒發對加工技術領域之年度貢獻 獎，本屆之得獎人為本次研討會主席Professor Y. T. Im, 同時介紹下次 2008年AMPT之研討會決定在 Bahrain (巴林)舉行，主辦之主席為 Professor B.S Yilbas，希望各位屆時也能多投稿。
大會也安排十月十日參觀POSCO(浦項鋼鐵)與Hyundai Motor
Comp.(現代汽車)兩家工廠，研討會參加人員可擇一參加。本人選擇
參觀POSCO鋼鐵，此公司與我國之中國鋼鐵公司相同於1970年代籌 建，目前POSCO之每年粗鋼總產量已達3000萬噸，其產量為中鋼之3 倍，其主要原因為POSCO也有造船廠可消費其產能，而且韓國之幾 家大汽車廠如現代、大宇、起亞等也是POSCO的大客戶，故其產能 才有辦法在30多年間有如此快速之成長。參觀中有解說人員之講解及 影片介紹有關POSCO的歷史與未來的目標，並參觀其在廠區內之鋼 鐵博物館及至現場參觀線材之軋延生產線。由於浦項鋼鐵之地理位置 離研討會之開會地點大田(Daejeon)有三小時之路程，參觀浦項鋼鐵之 後回到大田已將近下午七點，因此原本欲拜訪韓國高等科技學院 (KAIST)之行程也就無法成行，為此行中稍感遺憾之事。
此次為本人第三次以國科會之補助出國參與國際研討會，同時是 第一次拜訪韓國，感覺韓國之物價也是相當高，與出發前之想像不 同。此外韓國在各方面之發展也是有目共睹，是台灣的競爭之對手，
也有部份產業已超越台灣如鋼鐵、汽車與家電產業，有值得借鏡之
處。此次研討會主辦單位對於與會人員之照顧與安排工廠參觀等是值
得讚賞。
投稿期刊論文(一)
S. H. Hsiang and Y.W. Lin, (2007, 10) “Investigation of the influence of process
parameters on hot extrusion of magnesium alloy tubes”, Journal of MaterialsProcessing Technology, Vol.192-193, pp.292-299. (SCI, EI) (NSC94-2212-E011-020)
投稿期刊論文(二)
S. H. Hsiang and Y.W. Lin, (2008, 05) “Application of Fuzzy Theory to Predict
Deformation Behaviors of Magnesium Alloy Sheets under Hot Extrusion”, Journal ofMaterials Processing Technology, Vol.201, No.1-3, pp.138-144. (SCI, EI)
(NSC95-2221-E011-104)投稿期刊論文(三)
向四海、林益瑋、李倉誠 (2008,01)，「AZ 鎂合金板材之熱間擠製加工之研究」，
社團法人台灣鎂合金協會九十七年度會員大會暨論文發表會論文集
，151-156 頁。
出席國際會議研究心得報告及發表論文
報告人：向 四 海
服務機構及職稱：國立台灣科技大學機械系教授
會議時間地點：2007 年 10 月 7-11 日於韓國
會議名稱：第十屆先進材料與加工技術國際研討會
The 10th International Conference on Advances inMaterials and Processing Technologies
發表論文題目：(二篇)
Application of Fuzzy Theory to Predict Deformation Behaviors of Magnesium Alloy Sheets under Hot Extrusion
A study of improving warped characteristics during flange forging
中華民國九十七年五月二十一日
本人於民國九十六年十月七日至十月十二日赴韓國大田(Daejeon)參加由韓 國高等科技學院(KAIST)所主辦之第十屆先進材料與加工技術國際研討會(The 10th International Conference on Advances in Materials and Processing Technologies, 10th AMPT)。此次研討會之主題分為Deformation Processing、Forming in Melt or Near Melt Condition、Powder Forming、Laser Processing、Processing of New and Advanced Materials、Surface Engineering與Nano Technology等七個領域，共有208 篇研究成果論文發表。
研討會之會場分成4個會場總共分為36個Sessions發表，並有2場 Keynote speech。第一場的 Keynote speaker 是來自韓國POSCO的 Dr. O. J. Kwon，他的 演講題目為“Steel－what we have to know about additionally”；第二場的Keynote speaker 是來自韓國現代汽車公司(Hyundai Motor Comp.)的Dr. M. S. Kwon，Dr.
Kwon的演講題目為 “Hyundai-Kia Motor, Towards future global leadership”。
本 人 所 發 表 2 篇 之 論 文 領 域 皆 為 Deformation Processing 中 之 Extrusion Processes 與 Forging Processes，其兩篇論文名稱分別為Application of Fuzzy Theory to Predict Deformation Behaviors of Magnesium Alloy Sheets under Hot Extrusion 與 A study of improving warped characteristics during flange forging，其 論文編號分別為AMPT-098與AMPT-099。第一篇論文(AMPT-099)被安排在A 1 Session第三位發表，此一Session本應由德國 University of Erlangen-Nuremberg 的 Professor U. Engel與本人共同主持，但由於Professor U. Engel缺席且臨時代理 的韓國Professor J.R. Cho教授也未到場，故全場均由本人當主持人。此一Session 除了本人投稿之有關鋁合金凸緣鍛造加工之變形特性探討之外，尚有鍛造模具破 裂之預測及防止破裂之研究、鈦合金在等溫與非等溫鍛造之研究、運用有限元素 分析法進行鍛造相關加工之研究等論文發表。另外第二篇論文(AMPT-098)被安 排在A 5 Session第三位發表，主持人由德國 Institute of Forming Technology and Lightweight Construction 的 Dr. M. Schikorra 與 日 本 電 通 大 學 (University of Electro-Communications)的Professor M. Murata共同主持，此一Session除了本人投 稿之有關鎂合金板材擠製加工之研究之外，其他尚有探討鋁合金擠製之微結構組 Professor Y. T. Im, 同時介紹下次2008年AMPT之研討會決定在 Bahrain (巴林)舉
行，主辦之主席為 Professor B.S Yilbas，希望各位屆時也能多投稿。
大會也安排十月十日參觀POSCO(浦項鋼鐵)與Hyundai Motor Comp.(現代汽 車)兩家工廠，研討會參加人員可擇一參加。本人選擇參觀POSCO鋼鐵，此公司 與我國之中國鋼鐵公司相同於1970年代籌建，目前POSCO之每年粗鋼總產量已達 3000萬噸，其產量為中鋼之3倍，其主要原因為POSCO也有造船廠可消費其產能，
而且韓國之幾家大汽車廠如現代、大宇、起亞等也是POSCO的大客戶，故其產能 才有辦法在30多年間有如此快速之成長。參觀中有解說人員之講解及影片介紹有 關POSCO的歷史與未來的目標，並參觀其在廠區內之鋼鐵博物館及至現場參觀線 材之軋延生產線。由於浦項鋼鐵之地理位置離研討會之開會地點大田(Daejeon) 有三小時之路程，參觀浦項鋼鐵之後回到大田已將近下午七點，因此原本欲拜訪 韓國高等科技學院(KAIST)之行程也就無法成行，為此行中稍感遺憾之事。
此次為本人第三次以國科會之補助出國參與國際研討會，同時是第一次拜訪 韓國，感覺韓國之物價也是相當高，與出發前之想像不同。此外韓國在各方面之 發展也是有目共睹，是台灣的競爭之對手，也有部份產業已超越台灣如鋼鐵、汽 車與家電產業，有值得借鏡之處。此次研討會主辦單位對於與會人員之照顧與安 排工廠參觀等是值得讚賞。
APPLICATION OF FUZZY THEORY TO PREDICT DEFORMATION BEHAVIORS OF MAGNESIUM ALLOY SHEETS UNDER HOT EXTRUSION
S.H. Hsiang and Y.W. Lin
Department of Mechanical Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, ROC; email: [email protected]
ABSTRACT
The study predicts forming behaviors and tensile strength of the magnesium alloy sheet under extrusion in converging dies using fuzzy theory. The variable speed method was used to extrude perfect magnesium alloy sheets with a high extrusion ratio.
However, the selection of adjustment timing would affects the success of sheet forming.
Therefore, if fuzzy theory can be applied to predict the adjustment timing, then the experimental effectiveness of the trial and error method for adjustment timing will definitely increase. The billets used in this study were AZ31 and AZ61 magnesium alloys, and the two corresponding sets of converging dies had semi-die angles of 20°
and 40°, respectively. The process parameters are set using orthogonal arrays, which are used in the experiments. The experimental results are then used to establish the predictive model based on fuzzy theory. The variable speed timing and the tensile strength of the final product are then predicted at semi-die angles of 20°, 30° and 40°, and at heating temperatures of the billet between 320~360 ℃. Further experiments are performed to verify the accuracy of the approach.
KEYWORDS: Fuzzy prediction system, Magnesium alloy sheet, Extrusion, Variable
speed method1. INTRODUCTION
The demand for light, power-saving and recyclable products has increased in recent years. Magnesium alloys have become alternatives to steel, aluminum and plastic in some products. Magnesium alloy is 35 % and 77 % less dense than aluminum alloy and steel, respectively. Therefore, the use of magnesium alloy has gradually increased. Die-casting remains the main technique of magnesium alloy forming and fabrication. Zhang et al. [1] investigated the mechanical properties and microstructures of ZA104 and AZ91 magnesium alloys in die-casting products. Lee et al. [2] studied the effect of the parameters of the AM50 magnesium alloy process on porosity distribution during high-pressure die-casting. Huang et al. [3] observed the microstructure of an AM60B test rod manufactured by die-casting, and found that the porosity is least around the runner, and worst in the central part. If metal forming techniques, such as forging, rolling and extrusion, can be applied to form and fabricate magnesium alloy, then the scope of applications of magnesium alloy products can be expanded, and the yield will be increased. Chen and Huang [4] studied the formability of AZ31 magnesium alloy sheet under coining process. At room temperature, the formability of AZ31 is poor, but it improves as the temperature increases. The highest forming temperature is below 400 ℃. Hsiang and Kuo [5] investigated the properties of the finished products of AZ31 and AZ61 magnesium alloy thin sheets after hot
extrusion and the effects of fabrication conditions. A theoretical predictive model that could be used in the selection of forming and fabrication conditions would improve forming and fabrication. Hsiang and Kuo [6] employed an artificial neural network to predict the formability of magnesium alloy sheets during hot extrusion processing. Jung and Im [7] utilized fuzzy control theory to predict tension variations during hot rolling.
This study investigated the AZ31 and AZ61 magnesium alloy sheets by hot extrusion using converging dies. The process parameters are set using an orthogonal array. Various experiments are performed. The experimental data, including speed adjustment timing, tensile strength of the finished product, and others are considered to construct the fuzzy theory-based prediction system model. Based on this model, this study predicts the speed adjustment timing and tensile strength as the billet heating temperature of the two materials is varied between 320 ℃ and 360 ℃. Finally, confirmatory experiments demonstrate the accuracy of the model.
2. EXPERIMENTAL SCHEME
In this study, six process parameters are considered. The variable speed method is adopted to extrude perfect magnesium alloy sheets with a high extrusion ratio.
Meanwhile, the speed adjustment timing has to be found in the process of extrusion experiment. Speed adjustment timing data are adopted to establish a FPS model (model 1) of speed adjustment timing and perform the tensile test of the extruded sheet. The obtained tensile strength data are adopted to establish a FPS model (model 2) of tensile strength. These two models are expected to predict the speed adjustment timing and tensile strength. Finally, confirmatory experiments demonstrate the accuracy of the models, which is compared with that of Artificial Neural Networks. Fig. 1 presents the experimental procedure.
Furnace Billet
AZ31 or AZ61
Extrusion machine
Die
Success
Fail Selection of process parameters
Plan the L9 orthogonal array
Extrusion experiment
Tensile strength of sheet
Prediction model 2 Establish the fuzzy prediction model of the
tensile strength and Prediction
Obtaining the tensile strength
Confirmation Experiment Prediction model 1
Establish the fuzzy prediction model of the timing of speed adjustment
and Prediction
Obtaining the timing of speed adjustment Find out the timing of adjusting extrusion speed
Fig. 2 schematically depicts the hot extrusion process. The extrusion ingot is ψ80 mm ×
length 100 mm. After extrusion, the size of sheet is 2 mm × 70 mm. The extrusion ratio is 35.9.2.1 Variable Speed Method
The variables that determine the success of hot extrusion process are material, extrusion ratio, type of die, billet heating temperature, initial extrusion speed, container temperature and lubricant. When the fixed-speed method is adopted to extrude the magnesium alloy sheet at high extrusion ratio, the extrusion load reaches the limiting load of the extrusion machine. The limiting load of 600 tons is maintained for 16 second, forcing the magnesium alloy material to be maintained in a state of friction at high temperature and high pressure. Therefore, serious defects or cracking are formed in the extruded sheet. The photograph of the failed sample is shown in the lower part of Fig. 2. The speed is immediately reduced when the maximum extrusion load.
The time for which the maximum extrusion load is maintained can thus be shortened.
Accordingly, perfect sheets can be extruded. The photograph of the successful sample is also shown in Fig. 2. Fig. 3 plots the extrusion load curves obtained using the fixed speed and variable speed methods [5, 6].
2.2 Setting the Experimental Parameters
The parameters of the process for the experiments in this study are set using orthogonal array (OA). Therefore, before planning, the trial and error method is firstly used to set the control factors in the extrusion experiment, to understand the levels of the control factors. Table 1 presents the parameters of the experimental process. The converging die used for AZ31 and AZ61 materials and at semi-die angles of 20° and 40° is the outer OA. The inner OA is L9 OA. The four control factors are (A) billet heating temperature, (B) initial extrusion speed, (C) container temperature, and (D) lubricant. Each control factor contains three levels. The final extrusion speed is set to 1 mm/s.
3. FUZZY PREDICTION SYSTEMS (FPS)
Prof. Zadeh proposed fuzzy set in 1965 [8]. Fuzzy theory handles uncertainty and imprecision. A membership function is used to describe the degree of correlation between systems. It has been extensively applied in controlling projects, decision-making systems and the domains of inference and prediction, etc.
FPS has four components-the fuzzifier, the fuzzy rule base, the fuzzy inference engine and the defuzzifier. The function of the fuzzifier is to transform the crisp input into suitable linguistic fuzzy information. The fuzzy rule base stores the required rules and knowledge for solving related problems. Comprising fuzzy rules of the “If-Then” form, a fuzzy rule base describes the input-output relationship of the system. The fuzzy inference engine is the core of a fuzzy system. Approximate reasoning or fuzzy inference is used to simulate the thinking and decision-making modes of human beings to solve problems. The defuzzifier transforms the fuzzy information that is inferred from the fuzzy inference engine to a crisp output [9].
The fuzzy rule base of the fuzzy inference engine herein is based on the Mamdani fuzzy rule. Each of linguistic rule is presented in “If-Then” form, as in Eq. 1.
R
j：If x1 is A1j and x2 is A2j and … and xn is Anj then y is Bj (1) Variable speed methodFig. 3. Curve of load during extrusion process.
Timing for adjustment speed
denotes the linguistic fuzzy variable of output y, and xi (i：1,2,3,…,n) and y are the jth input and output variables, respectively. The related membership functions are expressed μAij(xi) and
μ
Bij(y), respectively. Triangular and trapezoidal membership functions are used herein.The max-min operation is adopted to compute the fuzzy inference engine using Mamdani implication method. The defuzzifier determines the center of gravity to transform fuzzy information to a crisp output y*, given by the following Eq.
∑
4. FPS FOR ADJUSTMENT TIMING IN VARIABLE SPEED METHOD (MODEL 1)
The main factors affecting the speed adjustment timing in the variable speed method are the four process parameters - material type, semi-die angle, billet heating temperature and initial extrusion speed. Extrusion experiments are carried out according to the planned OA that is presented in Table 1. In each experiment, the speed adjustment timing for slowing down the speed is found. Therefore, in the FPS model of speed adjustment timing, the input variables are material type, semi-die angle, billet heating temperature and initial extrusion speed, and the output variable is the speed adjustment timing.Table 1 Experimental setting of factors and levels
Controllable factors（Outer OA） Material type and converging die 1
2 3 4
AZ31 and semi-die angle 20°
AZ31 and semi-die angle 40°
AZ61 and semi-die angle 20°
AZ61 and semi-die angle 40°
Controllable factors
（Inner OA）
Billet heating temperature (℃)
Initial extrusion speed (mm/sec)
Container
temperature (℃) Lubricant 1
4.1 Construction of FPS of Speed Adjustment Timing
Based on the speed adjustment timing obtained from the extrusion processing using the planned OA, shown in Table 1, this study uses Matlab software to construct the FPS model of speed adjustment timing. This model is a system of four input
variables and one output variable. For the input variable, material type (x1) is divided into two language items: x1 = {M1(AZ31), M2(AZ61)}. The other three input variables, semi-die angle (x2), billet heating temperature (x3) and initial extrusion speed (x4), are divided into three language items, {S (small), M (medium), L (large)}. For the output variable, the speed adjustment timing (yadj) is divided into 13 language items:
y
adj ＝{A1, A2, … …, A13 }. Thirty-six fuzzy rules are established from the experiments, and they are as follows.Fig. 4. An example of the I/O mapping of the FPS of
R
1：If x1 is M1and x2 is Sand x3 is Sand x4 is S then yadj is A10R
36… ：If x1 is M2and x2 is Land x3 is Land x4 is L then yadj is A4The max-min operation is inferred using the Mamdani implication method. The defuzzifier uses the center of gravity to determine the crisp speed adjustment timing. Fig. 4 presents one of the I/O mappings between input variables and output variables of the FPS of the adjustment speed timing.
4.2 Prediction of Speed Adjustment Timing for Different Semi-Die Angles
The AZ31 and AZ61 sheets using the FPS of speed adjustment timing was performed at semi-die angles of 20° and 40°, billet heating temperatures from 320 to 360 ℃ in increments of 5 ℃, and initial extrusion speeds of 2, 3 and 4 mm/s. The prediction results are shown in Figs. 5 and 6.
Figs. 5 and 6 show that speed adjustment timing with the semi-die angle and the initial
extrusion speeds of the AZ31 and AZ61 sheets. For same process parameters, the adjustment speed timing of the AZ61 sheet exceeds that of the AZ31 sheet. At the given billet heating temperature, the speed adjustment timing at a semi-die angle of 20° exceeds that at 40°. The speed adjustment timing is highest when the initial extrusion speed is 2 mm/s, and decreases as the billet heating temperature increases. When the AZ31 sheet is placed in the two dies and the billet heating temperature exceeds 350 ℃, the speed adjustment timing at initial extrusion speeds of 2 mm/s and 3 mm/s are the same.The parameters of the confirmatory experiment we set to billet heating temperatures of 320
℃, 340 ℃and 360 ℃, the temperature of container of 350 °C, and the use of lubricant MoS2, yielding the results plotted in Figs. 5(a) and 6(b). The results of the confirmatory experiment are approximately the same as the analytical results of FPS. The predictive model constructed in this study can accurately predict the speed adjustment timing.
4.3 Speed Adjustment Timing at Semi-Die Angle of 30°
FPS is used to predict the speed adjustment timing of the extrusion process of AZ31 and AZ61 when the semi-die angle is 30°. Fig. 7 plots the results. The figure shows that the properties of all speed adjustment timing curves are the same as those of the curves when the semi-die angles of 20° and 40°. However, no related rule applies at a semi-die angle of 30° in the fuzzy rule base. An experiment was performed to confirm the prediction. The parameters of the confirmatory experiment are set such that the billet heating temperatures are 320 ℃, 340 ℃ and 360 ℃, the temperature of container is 350 ℃ and the lubricant is MoS2, yielding the results
FPS is used to predict the speed adjustment timing of the extrusion process of AZ31 and AZ61 when the semi-die angle is 30°. Fig. 7 plots the results. The figure shows that the properties of all speed adjustment timing curves are the same as those of the curves when the semi-die angles of 20° and 40°. However, no related rule applies at a semi-die angle of 30° in the fuzzy rule base. An experiment was performed to confirm the prediction. The parameters of the confirmatory experiment are set such that the billet heating temperatures are 320 ℃, 340 ℃ and 360 ℃, the temperature of container is 350 ℃ and the lubricant is MoS2, yielding the results