萊氏現象之自動化量測與聲響分析

Automatic measurement and acoustic analysis of Leidenfrost phenomenon

曾固威(Ku-Wei Tseng)

指導教授 ： 黃振康

國立臺灣大學/生物資源暨農學院/生物產業機電工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.02390

摘要

本研究使用時域以及頻域的訊號處理方式分析萊氏實驗液滴碰觸高溫平板的聲響，利用聲響訊號在時域以及頻域上的特徵判斷平板是否已到達萊氏溫度。以往進行萊氏實驗相當耗時且耗費大量人力，當人疲勞時實驗的誤差就會增加，故本實驗第一步是將實驗自動化。系統設備可分為四個部份，分為即時影像辨識、滴定控制、溫度控制以及聲響資訊擷取，使用LabVIEW軟體撰寫程式將四個部份連結在一起搭配硬體實現自動化。在系統的穩定度上使用肉眼與影像辨識做比較，結果顯示肉眼與影像辨識相差最多為3.8秒，並能判斷出相同的萊氏溫度；本實驗加熱表面為不鏽鋼板，選用的流體為水、乙二醇/水(重量百分濃度0%、5%、50%、70%)、異丙醇/水 (莫耳分率0、0.01、0.02)以及正庚烷/乙醇 (莫耳分率0、0.01、0.02)。取得實驗聲響資訊後，以六種訊號處理方式辨識萊氏溫度，分別為最大音量法、總音量法、時間長度法、短時傅立葉轉換法、小波轉換法以及傅立葉轉換法，若判斷結果介於正確萊氏溫度±6oC則判斷成功。汽泡爆炸現象為本研究聲響來源，汽泡爆炸屬於非穩態之聲響，聲響頻帶能量分佈廣且無主要頻帶，因此時域之分析方法對萊氏溫度點之判斷能力較差，其中最有效的方法為第一碰撞時間法，辨識成功率為58%；而頻域的方法較能有效的判斷萊氏溫度點，最有效的方法為小波轉換法，辨識成功率為72%。

關鍵字

萊氏現象、自動化、訊號處理、頻譜分析

並列摘要

The purpose of this study is to use the signal processing method of time domain and frequency domain analyzing the acoustic of droplet touching high temperature surface, and use the characteristic of the acoustic signal of time domain and frequency domain to determine if reach Leidenfrost temperature or not. In the past, Leidenfrost experiment usually costs very long time and intensive human power. When a person is fatigue, there will be more errors with experiment data. Therefore, the first step is to make the experiment automatic. We can separate the system equipment into 4 parts, real-time image identification, micro-pump control, surface temperature control and acoustic information recording. By using LabVIEW software, we wrote a program to link these 4 sections to achieve automation. As for the stability of the system, we compared human eye and image identification, the maximum difference is 3.8 seconds to read the same Leidenfrost temperature. We used stainless steel panel as heating surface, and the fluid we used are water, water/ethylene glycol(mass percentage 0%, 5%, 50%, 70%), water/IPA(mole fraction 0, 0.01, 0.02), ethanol /n-heptane (mole fraction 0, 0.01, 0.02). We obtained the acoustic information of the experiment via the automation equipment, and indentified Leidenfrost temperature trough four signal processing methods (the largest volume method, the total volume method, the first collision time method, as well as wavelet transform method). If the result is in between ±6oC of accurate Leidenfrost temperature, we considered it’s a successful identification. The source of the sound in this experiment is from the bubble explosive, therefore, the time domain methods fail to successfully indentify the Leidenfrost temperature. The most successful method of time domain is the first collision time method; it successfully indentified 19 out of 50 sets of experiments. The methods of frequency domain can identify Leidenfrost temperature more effectively, wavelet transform method successfully indentified 36 out of 50 sets of experiments.

並列關鍵字

Leidenfrost phenomenon, automation, signal processing, spectrum analsis

參考文獻

Baumeist, K.J. and F.F. Simon. 1973. Leidenfrost temperature - its correlation for liquid-metals, cryogens, hydrocarbons, and water. Journal of Heat Transfer-Transactions of the ASME. 95(2): 166-173.

Bayazit, M. and H. Aksoy. 2001. Using wavelets for data generation. Journal of Applied Statistics. 28(2): 157 - 166.

Berenson, P.J. 1961. Transition boiling heat transfer from a horizontal surface. J. Heat Transfer. 83.

Bernardin, J.D. and I. Mudawar. 1999. The Leidenfrost point: Experimental study and assessment of existing models. Journal of Heat Transfer-Transactions of the ASME. 121(4): 894-903.

Bernardin, J.D. and I. Mudawar. 2002. A cavity activation and bubble growth model of the Leidenfrost point. Journal of Heat Transfer-Transactions of the ASME. 124(5): 864-874.