張泉+馬小威+張林鋒
摘 要:巖土及填料的熱物性參數(shù)可直接影響地埋管換熱器的性能設(shè)計(jì).針對(duì)現(xiàn)場(chǎng)熱響應(yīng)測(cè)試中填料熱物性參數(shù)需要預(yù)先測(cè)量,而無(wú)法通過(guò)計(jì)算直接評(píng)估的難題,在無(wú)限長(zhǎng)線熱源模型及管壁熱阻修正復(fù)合介質(zhì)線熱源模型的基礎(chǔ)上,應(yīng)用遺傳算法對(duì)巖土及填料的熱物性參數(shù)同時(shí)進(jìn)行評(píng)估,并結(jié)合熱響應(yīng)測(cè)試實(shí)驗(yàn),驗(yàn)證了該方法的準(zhǔn)確性.鉆孔熱阻、巖土導(dǎo)熱系數(shù)和填料導(dǎo)熱系數(shù)的評(píng)估值與實(shí)測(cè)值的相對(duì)誤差分別為3.47%,1.42%和4.93%.2種模型計(jì)算所得流體平均溫度與實(shí)測(cè)值的均方根誤差分別為0.050 5 ℃和0.172 ℃.研究結(jié)果對(duì)地埋管換熱器的設(shè)計(jì)具有重要參考價(jià)值.
關(guān)鍵詞:填料熱物性;遺傳算法;復(fù)合介質(zhì)線熱源;熱響應(yīng)
中圖分類號(hào):TU831.6 文獻(xiàn)標(biāo)志碼:A
文章編號(hào):1674-2974(2017)03-0151-06DOI:10.16339/j.cnki.hdxbzkb.2017.03.019
Abstract:The parameters for thermal properties of ground and grout have directly effect on the performance of ground heat exchanger (GHE). For the problem that grout thermal properties of in-situ thermal response test (TRT) need to be measured in advance and cannot be evaluated directly by calculation, this paper applied genetic algorithm to evaluate the parameters of ground and grout thermal properties simultaneously on the basis of infinite line source (ILS) model and composite media line source (CMLS) model with wall thermal resistance correction. The method was validated by in situ TRT. Compared with the measured values, the calculation relative error of borehole thermal resistance, ground thermal conductivity and diffusivity was 3.47%, 1.42% and 4.93%, respectively. The root mean square error (RMSE) of the calculated average temperature by these two models and the measured value is 0.0505 and 0.172 ℃, respectively. The results provide important reference for GHE design.
Key words:grout thermal properties;genetic algorithms;composite media line source;thermal response test
巖土及填料的熱物性參數(shù)是地埋管換熱器設(shè)計(jì)的關(guān)鍵參數(shù),準(zhǔn)確地評(píng)估巖土及填料的熱物性參數(shù)可以解決地埋管換熱器設(shè)計(jì)不當(dāng)?shù)膯?wèn)題.目前,巖土熱物性參數(shù)主要采用斜率法和參數(shù)估計(jì)法進(jìn)行評(píng)估計(jì)算.相比于斜率法,參數(shù)估計(jì)法可有效降低測(cè)試中熱流不恒定對(duì)評(píng)估參數(shù)值的影響,其評(píng)估精度較高[1].因此,在ASHRAE手冊(cè)中,推薦采用參數(shù)估計(jì)法進(jìn)行巖土熱物性參數(shù)的評(píng)估[2].
參數(shù)估計(jì)法評(píng)估巖土熱物性參數(shù)是一個(gè)典型的逆?zhèn)鳠釂?wèn)題.該方法以巖土熱物性參數(shù)為決策變量,以各時(shí)刻流體平均溫度的計(jì)算值和實(shí)測(cè)值的平方和誤差(SSE)或均方根誤差(RMSE)為目標(biāo)函數(shù),通過(guò)最小化尋優(yōu)算法求解巖土熱物性參數(shù).但目前所采用的參數(shù)估計(jì)法存在以下2個(gè)方面的問(wèn)題:1) 主要運(yùn)用單純形法、模式搜索法等算法尋優(yōu)求解[3],其收斂速度較慢,并受迭代初值的影響,導(dǎo)致評(píng)估結(jié)果可能存在較大誤差.2) 參數(shù)估計(jì)法評(píng)估巖土熱物性參數(shù)時(shí),常以鉆孔壁溫度為計(jì)算耦合點(diǎn),忽略了鉆孔內(nèi)填料的熱物性,無(wú)法采用參數(shù)估計(jì)模型直接評(píng)估填料熱物性參數(shù).
為克服參數(shù)估計(jì)法的上述缺點(diǎn),本文結(jié)合地埋管無(wú)限長(zhǎng)線源模型和管壁熱阻修正的復(fù)合介質(zhì)線熱源模型,利用遺傳算法同時(shí)評(píng)估鉆孔熱阻、巖土導(dǎo)熱系數(shù)和熱擴(kuò)散系數(shù)以及填料導(dǎo)熱系數(shù)和熱擴(kuò)散系數(shù)等5個(gè)參數(shù),并結(jié)合熱響應(yīng)測(cè)試實(shí)驗(yàn),驗(yàn)證了該方法的準(zhǔn)確性.
1 地埋管傳熱模型
1.1 無(wú)線長(zhǎng)線熱源模型
Kelvin[4]提出了豎直地埋管鉆孔外一維無(wú)限長(zhǎng)線熱源傳熱模型,該模型通過(guò)鉆孔壁溫度耦合鉆孔內(nèi)穩(wěn)態(tài)導(dǎo)熱和鉆孔外非穩(wěn)態(tài)導(dǎo)熱.當(dāng)傳熱時(shí)間大于10 h時(shí),通過(guò)該簡(jiǎn)化模型可求得流體平均溫度為[5-6]:
1.2 復(fù)合介質(zhì)線熱源模型
在無(wú)限長(zhǎng)線熱源模型的基礎(chǔ)上,Li等[7]提出了一種全時(shí)間范圍內(nèi)的地埋管溫度響應(yīng)G函數(shù)模型.該模型將U型管管壁作為溫度耦合點(diǎn),以U型管外壁面A和B兩點(diǎn)的平均溫度代表U型管管壁的平均溫度,如圖1所示.
復(fù)合介質(zhì)線熱源模型由于考慮了填料熱容和鉆孔埋管幾何尺度對(duì)換熱性能的影響,在一定時(shí)間范圍內(nèi),可較準(zhǔn)確地計(jì)算流體平均溫度.
2 基于遺傳算法的熱物性參數(shù)評(píng)估
遺傳算法(Genetic Algorithm)是一種模仿自然界生物進(jìn)化,并帶有隨機(jī)性的全局搜索方法.該算法由可能潛在解集的種群開(kāi)始計(jì)算,通過(guò)優(yōu)勝劣汰的準(zhǔn)則產(chǎn)生適應(yīng)度較高的解集(種群),在每一代進(jìn)化過(guò)程中,選擇種群中適應(yīng)度水平較高的個(gè)體進(jìn)行交叉重組以及變異,并產(chǎn)生子種群,由此通過(guò)一定代數(shù)的進(jìn)化即可得到最優(yōu)解的個(gè)體[10].與傳統(tǒng)優(yōu)化算法相比,遺傳算法具有較好的全局收斂性、計(jì)算速度快、不受目標(biāo)函數(shù)的約束、可并行計(jì)算等諸多優(yōu)點(diǎn),在工程計(jì)算中得到了廣泛應(yīng)用.
本文遺傳算法采用謝菲爾德大學(xué)所開(kāi)發(fā)的Matlab遺傳算法工具箱(gatbx)[11].求解中,設(shè)定鉆孔熱阻Rb,巖土導(dǎo)熱系數(shù)λs,巖土熱擴(kuò)散系數(shù)as,填料導(dǎo)熱系數(shù)λb和填料熱擴(kuò)散系數(shù)ab等5個(gè)參數(shù)為決策變量.以式(1)和式(6)計(jì)算所得的流體平均溫度與實(shí)測(cè)值的綜合均方根誤差為目標(biāo)函數(shù)[12],如式(7)所示.該目標(biāo)函數(shù)最小時(shí)所對(duì)應(yīng)的決策變量即為最終評(píng)估結(jié)果.
式中:Tf,ILS為無(wú)限長(zhǎng)線熱源模型計(jì)算的各時(shí)刻流體平均溫度,℃;Tf,CMLS為復(fù)合介質(zhì)線熱源模型計(jì)算的各時(shí)刻流體平均溫度,℃;Tf,EX為實(shí)測(cè)各時(shí)刻流體平均溫度,℃,計(jì)算中取各時(shí)刻U型管流體進(jìn)出口溫度的算術(shù)平均值;n為用于無(wú)限長(zhǎng)線熱源模型計(jì)算的測(cè)量數(shù)據(jù)組數(shù);m為用于復(fù)合介質(zhì)線熱源模型計(jì)算的測(cè)量數(shù)據(jù)組數(shù).
在算法優(yōu)化過(guò)程中,幾個(gè)關(guān)鍵遺傳操作設(shè)置如下:
1)編碼:為提高算法效率和求解精度,本文采用實(shí)數(shù)型編碼的方式對(duì)種群進(jìn)行編碼.
2)適應(yīng)度函數(shù):目標(biāo)函數(shù)至適應(yīng)度函數(shù)的轉(zhuǎn)換,采用非線性排序算法且選擇壓差為2[11].
3)交叉:采用隨機(jī)遍歷抽樣(SUS)選擇父體進(jìn)行交叉運(yùn)算.由于采用實(shí)數(shù)型編碼,交叉概率的取值不影響離散重組產(chǎn)生子代個(gè)體的表征,故將其設(shè)為1[11].
4)變異:為防止算法局部收斂,變異運(yùn)算采用自適應(yīng)算法[13].
5)遷移:在子種群遷移中,將子種群定義成網(wǎng)狀結(jié)構(gòu),并根據(jù)其適應(yīng)度值進(jìn)行各子種群間的個(gè)體遷移[14].
算法中各控制參數(shù)設(shè)置及優(yōu)化流程分別如表1和圖2所示.
3 算法應(yīng)用及分析
3.1 沙箱熱響應(yīng)測(cè)試
Beier等[15]在實(shí)驗(yàn)室中搭建了一個(gè)較大的沙箱(sandbox)進(jìn)行熱響應(yīng)測(cè)試實(shí)驗(yàn).沙箱周圍采用木板結(jié)構(gòu)固定,里面填充潮濕沙土,其長(zhǎng)和寬分別為18.32 m和1.83 m.中心處設(shè)置內(nèi)徑為12.6 cm的鋁管構(gòu)建鉆孔壁,其內(nèi)水平安裝單U型地埋管換熱器,并用質(zhì)量分?jǐn)?shù)為20%的泥漿膨潤(rùn)土進(jìn)行回填.
熱響應(yīng)測(cè)試實(shí)驗(yàn)中利用2臺(tái)電加熱器對(duì)流體進(jìn)行加熱,加熱器輸入功率不確定性為±1%;熱敏電阻測(cè)量U型管進(jìn)出口流體溫度,其測(cè)量不確定性為±0.03 ℃;流量計(jì)測(cè)量流體體積流量,其測(cè)量不確定性為±5%.總測(cè)試時(shí)長(zhǎng)為48 h,其主要測(cè)試參數(shù)見(jiàn)表2.
3.2 目標(biāo)函數(shù)進(jìn)化
圖3給出了綜合均方根誤差(RSME)最小值及平均值隨遺傳代數(shù)的變化.可以看出,目標(biāo)函數(shù)(綜合均方根誤差)在40代以后變化較緩,此時(shí)計(jì)算收斂,最小值和平均值分別為0.111和0.118 ℃.
當(dāng)目標(biāo)函數(shù)最小時(shí),2個(gè)模型分別計(jì)算所得的均方根誤差遺傳代數(shù)的變化如圖4所示.當(dāng)算法逐漸收斂之后,復(fù)合介質(zhì)線熱源模型計(jì)算所得均方根誤差比無(wú)限長(zhǎng)線熱源模型大,收斂之后其差異平均值為0.124 ℃.這是因?yàn)椴捎脧?fù)合介質(zhì)模型計(jì)算流體平均溫度時(shí),未舍去熱響應(yīng)測(cè)試前10 h的溫度數(shù)據(jù),其計(jì)算時(shí)長(zhǎng)即為熱響應(yīng)測(cè)試總時(shí)長(zhǎng).由于傳熱前期鉆孔內(nèi)非穩(wěn)態(tài)傳熱對(duì)流體溫度的影響較大,根據(jù)復(fù)合介質(zhì)模型計(jì)算的流體平均溫度和實(shí)際值的溫度差也相應(yīng)較大.
3.3 參數(shù)評(píng)估結(jié)果及分析
基于地埋管無(wú)限長(zhǎng)線源模型和管壁熱阻修正的復(fù)合介質(zhì)線熱源模型,分別采用模式搜索法(Hooke-Jeeves算法)、單純形法和遺傳算法進(jìn)行巖土及填料熱物性參數(shù)評(píng)估,結(jié)果見(jiàn)表3.表中鉆孔熱阻的實(shí)測(cè)值是通過(guò)單獨(dú)測(cè)量鉆孔壁平均溫度計(jì)算所得.模式搜索法由初始值開(kāi)始交替實(shí)施軸向搜索和模式搜索,直至找到最優(yōu)值.單純形算法則是基于線性規(guī)劃問(wèn)題的可行域多面凸集原理,通過(guò)查找該凸集的某頂點(diǎn)得出最優(yōu)值.可以發(fā)現(xiàn),遺傳算法計(jì)算所得的綜合均方根誤差最小,為0.111 ℃,且鉆孔熱阻和導(dǎo)熱系數(shù)的評(píng)估相對(duì)誤差均小于5%.
圖5所示為綜合均方根誤差最小時(shí)所對(duì)應(yīng)的流體平均溫度隨時(shí)間的變化曲線.由圖可知,模型計(jì)算流體平均溫度與實(shí)測(cè)值均吻合較好,所得流體平均溫度與實(shí)測(cè)值的均方根誤差分別為0.0505和0.172 ℃.同時(shí)可見(jiàn),復(fù)合介質(zhì)模型計(jì)算流體平均溫度與實(shí)測(cè)值的差異主要集中在前10 h以內(nèi).
4 結(jié) 論
本文采用管壁熱阻修正的復(fù)合介質(zhì)線熱源模型與無(wú)限長(zhǎng)線熱源模型并聯(lián)求解流體平均溫度,并以流體平均溫度的綜合均方根誤差為目標(biāo)函數(shù),采用自適應(yīng)變異型多種群遺傳算法尋優(yōu)求解了鉆孔熱阻、巖土導(dǎo)熱系數(shù)及熱擴(kuò)散系數(shù)、填料導(dǎo)熱系數(shù)及熱擴(kuò)散系數(shù)等5個(gè)熱物性參數(shù).與傳統(tǒng)參數(shù)估計(jì)法相比,通過(guò)管壁熱阻修正復(fù)合介質(zhì)線熱源模型可較準(zhǔn)確地計(jì)算流體平均溫度,同時(shí)可評(píng)估填料熱物性參數(shù).在沙箱熱響應(yīng)測(cè)試中應(yīng)用遺傳算法進(jìn)行巖土及填料熱物性參數(shù)評(píng)估,通過(guò)計(jì)算可以得出以下結(jié)論:
1)尋優(yōu)過(guò)程中,遺傳進(jìn)化到40代左右時(shí)計(jì)算收斂,所得最終均方根誤差的最小值為0.111 ℃.
2)與其他尋優(yōu)算法相比,應(yīng)用遺傳算法評(píng)估巖土及填料的熱物性參數(shù),精度較高,鉆孔熱阻、巖土導(dǎo)熱系數(shù)、填料導(dǎo)熱系數(shù)的相對(duì)誤差分別為3.47%,1.42%和4.93%.
3)通過(guò)評(píng)估所得巖土及填料熱物性參數(shù)計(jì)算流體平均溫度,與實(shí)測(cè)值相比,根據(jù)線熱源模型和復(fù)合模型計(jì)算所得流體平均溫度的均方根誤差分別為0.050 5 ℃和0.172 ℃.
參考文獻(xiàn)
[1] SPITLER J D,GEHLIN S E A. Thermal response testing for ground source geat pump systems-an historical review[J]. Renewable and Sustainable Energy Reviews,2015, 50:1125-1137.
[2] ASHRAE. ASHRAE handbook: HVAC applications [M]. Atlanta,GA: American Society of Heating,Refrigerating and Air-Conditioning Engineers,Inc, 2011:34.13-34.17.
[3] ZHANG C,GUO Z, LIU Y, et al. A review on thermal response test of ground-coupled heat pump systems[J]. Renewable and Sustainable Energy Reviews,2014,40(8):851-867.
[4] KELVIN T W. Mathematical and physical papers II[M]. London:Cambridge University Press,1884:41-60.
[5] 刁乃仁,方肇洪.地埋管地源熱泵技術(shù)[M].北京:高等教育出版社,2006:47-50.
DIAO Nairen,F(xiàn)ANG Zhaohong.Ground-coupled heat pump technology [M]. Beijing:Higher Education Press,2006:47-50.(In Chinese)
[6] ZHANG L F,ZHANG Q,HUANG G S,et al. A P(t)-linear average method to estimate the thermal parameters of the borehole heat exchangers for in situ thermal response test[J]. Applied Energy,2014,131(7):211-221.
[7] LI M,LI P,CHAN V,et al. Full-scale temperature response function (G-function) for heat transfer by borehole ground heat exchangers (GHEs) from sub-hour to decades [J]. Applied Energy,2014,136:197-205.
[8] LI M, LAI A C K. Analytical model for short-time responses of ground heat exchangers with U-shaped tubes: Model development and validation [J]. Applied Energy,2013,104:510-516.
[9] ZHANG L F,ZHANG Q,HUANG G S. A transient quasi-3D entire time scale line source model for the fluid and ground temperature prediction of vertical ground heat exchangers[J]. Applied Energy,2016,170:65-75.
[10]張泉,杜亞星,張林峰,等. 基于遺傳算法的單U地源熱泵鉆孔內(nèi)熱阻研究[J]. 湖南大學(xué)學(xué)報(bào):自然科學(xué)版,2014,41(9):100-105.
ZHANG Quan,DU Yaxing,ZHANG Linfeng,et al. A study on the heat resistance of single U pipe in the ground source heat pump system based on genetic algorithm[J].Journal of Hunan University: Natural Sciences, 2014,41(9):100-105. (In Chinese)
[11]ANDREW C,PETER F,HARTMUT P,et al. Genetic algorithm TOOLBOX for use with MATLAB,users guide v1.2[R]. Sheffield:University of Sheffield,2003:9-36.
[12]FUJII H,OKUBO H,NISHI K,et al. An improved thermal response test for U-tube ground heat exchanger based on optical fiber thermometers [J]. Geothermics,2009,38(4):399-406.
[13]SRINIVAS M,PATNALK L M. Adaptive probabilities of crossover and mutation in genetic algorithms[J]. IEEE Transactions on Systems,Man and Cybernetics,1994,24(4):656-667.
[14]雷英杰.MATLAB遺傳算法工具箱及應(yīng)用[M].西安:西安電子科技大學(xué)出版社,2005:68-69.
LEI Yingjie. Genetic algorithm toolbox and application[M].Xian:Xidian University Press,2005:68-69. (In Chinese)
[15]BEIER R A,SMITH M D,SPITLER J D. Reference data sets for vertical borehole ground heat exchanger models and thermal response test analysis [J]. Geothermics,2011,40(1):79-85.