| 摘要: | 隨著全球工業水準上升，人類所需求的電量越來越多，馬達被要求提供的能量也越來越大，因而馬達的體積以及功率也朝向巨大化發展。相對所產生的廢熱也越積越多，無法順利排出，這將減少馬達壽命以及可靠性。目前採用的冷卻方式可分為空氣冷卻、氫氣冷卻以及水冷卻。
為了冷卻馬達損耗所產生的熱能，吾人利用馬達熱交換器的排列方式、數量的多寡、擺放的位置將馬達散熱的效能作逐一的測試，以及導風區隔板的擺放方式。然而實驗上要做馬達元件更替的動作實在太耗費時間，更何況有時候只是一些概念上的構思，無法保證是最佳的設計。
因此利用數值計算軟體上的便利性，即為更改元件尺寸的參數和位置便能得到所需要的結果，至於計算得到的結果是不是真的可以當作最後的結果，則要比對數值模型原型與實驗值的誤差，一但誤差在大家公認可接受的範圍內，此馬達就認可為有執行的空間，數值模型便可拿來執行之後的更改動作，大大減少時間和成本。
本文配合東元電機產學合作計畫，一共分析五台空氣冷卻馬達，研究最適合馬達之網格配置與一致性的參數設定。最後發現自動生成的網格表現在簡單的外型都比較接近實驗值，一旦複雜外型以及大氣進入方式不同時則沒有辦法模擬得到準確的答案。將手動生成網格用在馬達上也是不能保證有最好的結果，雖然第二台有效的解決不準確的問題，之後第四以及第五台卻沒有預期的好。在k-e紊流模式下，差分法則使用MARS與LUD比使用UD會更接近實驗值。最佳導風區的導流板設計為無中央分隔導流板。其中使用ICEM -CFD作為前處理器，使用泛用型商用計算軟體STAR-CD作為求解器，後處理器為Tecplot。
Along with the global industrial level rising, there is more and more mankind''s electric consumption of the demand, different kinds of electric motor is required the energy offered becomes greater; therefore the volume and power of the motor move towards and melt to develop enormously. Since the used heat relatively produced is accumulated increasingly, unable to discharge fast enough; it will reduce life-span of the motor and dependability. Therefore, the studies of motor cooling become extremely important.
The motor cooling method can be divided into the air cooling, hydrogen cooling, and water cooling. For air cooling, we utilize the variation way, for example, the design of motor heat exchanger efficiency include the motor heat dissipation, and lead the way of putting of the wind district baffle. But the experiment study is too time consume and unable to provide optimize design.
Using the CFD simulation approach, it is parameter and position of altering the size of the component that can get the results needed , whether can really be regarded as the last result as for calculating the result received, than to the model prototype and error, experiment of value, once error generally acknowledge acceptable range in everybody, this motor approves that in order to there is space carried out, this model can bring change movements after carrying out, reduce the time and cost greatly.
In this study we analyzed five TEAAC motors and study the most suitable for the motor and dispose the parameter with consistency to establish. Finally, we find that auto mesh much closer to experiment in simple cases, but not suitable in complex cases. Manual mesh can solve these problems. In k-ε model, we find the MARS and LUD differential schemes are much accuracy than lower differential scheme (UD). The best up-cover design is taking apart the middle diversion device. Using ICEM-CFD as the pre-processor and commercial CFD software STAR-CD as solver, and the post-processor is Tecplot. |
