Evaluation of measurement uncertainties for five-hole probes based on Monte Carlo method
-
摘要:
为对五孔探针的测量精度进行评定,发展了基于统计学的蒙特卡罗法(MCM)评定五孔探针测量不确定度的流程和方法,并对比了最大误差限法和不确定度传播律法(GUM),同时研究了抽样数
M 对MCM评定结果的影响。该方法适用于无数学表达式的测量模型中,能考虑模型非线性的影响,概率密度函数(PDF)能更科学地表征输入量的分布,而不局限于正态分布。为了验证该方法,将五孔探针在校准风洞中进行了校准和验证。结果显示,对于风洞实验某测量点的静压,MCM提供的95%最短包含区间较GUM区间长度小11.1%,该差值占标准偏差的33.3%,表明利用MCM评估测量不确定度能引入复杂数据处理过程中非线性的影响,相比GUM评定方法具有一定的优势。将MCM应用于叶栅栅后流场五孔探针测量结果不确定度评定,发现各参数的不确定度在整个测量截面内分布情况与误差分布类似,在叶顶泄漏涡区域内测量不确定度较大,能避免相对静压无法表示的困境,同时能去除粗大误差的影响。MCM本身的统计误差可通过适当增加M 来解决,需要综合考虑计算机性能和时间成本。Abstract:To evaluate the measurement accuracy of five-hole probes, the method of evaluating measurement uncertainties based on statistical Monte Carlo method (MCM) was developed and compared with the maximum error limit method and the guide to the expression of uncertainty method (GUM). At the same time, the influence of sampling numbers
M on the evaluation result of MCM was studied. The method was applicable to measurement models without mathematical expressions, and the influence of nonlinearity of the model can be considered. The probability density function (PDF) can more scientifically characterize the distribution of the input quantity, instead of being limited to the normal distribution. To verify the method, a five-hole probes was calibrated and used in a standard wind tunnel. The results showed that for the static pressure, the shortest 95% coverage interval provided by MCM was 11.1% smaller than the GUM. The difference accounted for 33.3% of the standard deviation. Furthermore, using MCM, the non-linear influence in data processing procedure can be considered compared with GUM. The MCM was applied to evaluate the measurement results of the flow field behind the cascade using the five-hole probes. The results showed that the uncertainty distribution of each parameter was similar to the error distribution in the whole measured cross section. The uncertainty was larger in blade tip leakage vortex area. However, it can avoid the trouble that the relative static pressure can not be expressed, and the influence of gross error can be removed. The statistical error of MCM itself can be solved by appropriately increasingM while considering the computer performance and time cost.-
Key words:
- five-hole probes /
- accuracy /
- measurement uncertainties /
- Monte Carlo method /
- cascade
-
表 1 五孔探针校准范围及测量点
Table 1. Calibration range and measurement points of five-hole probes
参数 校准点 测量点 马赫数Ma 0.2 0.1 偏转角α/(°) ±24 ±24 俯仰角β/(°) −28~16 −28,−27,−21.5,
−19, −17,−13偏转角测量站 13 17 俯仰角测量站 12 6 间隔/(°) 4 表 2 叶栅参数
Table 2. Cascade parameters
参数 数值 雷诺数/105 2.81 弦长/mm 126.8 叶高/mm 120 稠度 1.5 攻角/(°) 0 (叶尖间隙/叶高)/% 1 安装角/(°) 30 表 3 五孔探针测量误差分布
Table 3. Measurement error distribution of five-hole probes
参数 极限
误差(3σ)68%置信
概率(σ)95%置信
概率(2σ)偏转角α/(°) 2.30 0.76 1.53 俯仰角β/(°) 2.43 0.81 1.62 总压pt/kPa 0.031 0.010 0.021 静压ps/kPa 0.040 0.013 0.026 表 4 五孔探针测量风洞流场不确定度分量
Table 4. Measurement uncertainties of wind tunnel flow field by five-hole probes
输入量Xi PDF分布 不确定度/ 区间上下限 概率分布 αmou, βmou/(°) 矩形(定值) |a|=b=1 R(a, b) αtrav, βtrav/(°) 矩形 |a|=b=0.05 R(a, b) 测量pj /Pa 矩形 |a|=b=3.11 R(a, b) pref /Pa 0 表 5 各方法计算β的结果对比 (u (y) = 0.06)
Table 5. Comparison of β results of each calculation method (u (y) = 0.06)
(°) 特征量 MCM GUM 误差分布 估计值 −27.29 −27.29 −26.64 95%最短
包含区间[−27.41,−27.18] [−27.41,−27.17] [−28.26,−25.02] 区间长度 0.23 0.24 3.24 表 6 各方法计算ps的结果对比 (u (y)=0.06)
Table 6. Comparison of ps results of each calculation method (u (y)=0.06)
kPa 特征量 MCM GUM 误差分布 估计值 99.880 99.880 99.879 95%最短
包含区间[99.875,99.884] [99.874,99.884] [99.853,99.905] 区间长度 0.009 0.010 0.052 表 7 各参数95%包含概率对应的误差分布
Table 7. Error distributions of parameters at 95% probability
区域 β/(°) α/(°) pt/Pa Δpt/% ps/Pa 泄漏涡区 2 2 2.4 4 2.8 主流区 0.2 0.2 1.8 0.8 0.4 表 8 五孔探针测量叶栅栅后流场不确定度分量
Table 8. Measurement uncertainties of flow field behind the cascade by five-hole probes
输入量Xi PDF分布 不确定度/
区间上下限概率分布 αmou,βmou/(°) 矩形(定值) |a|=b=1 R(a, b) αtrav,βtrav/(°) 矩形 |a|=b=0.05 R(a, b) pj /Pa 矩形 |a|=b=2.05 R(a, b) pref /Pa 0 -
[1] MA Hongwei,ZHANG Jun,ZHANG Jinghui,et al. Experimental study of effects of grooved tip clearances on the flow field in a compressor cascade passage[J]. Journal of Turbomachinery, 2012, 134(5):051012.1-051012.12 [2] 林洪桦. 测量误差与不确定度评估[M]. 北京:机械工业出版社,2010 [3] 马宏伟,贺象,单晓明,等. 一种测量跨声速多级压气机转子出口二维流场的方法[J]. 推进技术,2013,34(6): 754-759.MA Hongwei,HE Xiang,SHAN Xiaoming,et al. A method of measuring 2D flow-field at rotor exits of transonic multistage compressor[J]. Journal of Propulsion Technology,2013,34(6): 754-759. (in Chinese) [4] 李井洋,马宏伟,贺象. 楔顶双孔探针测量跨声多级压气机转子出口三维动态流场的方法[J]. 航空动力学报,2012,27(10): 2262-2268.LI Jingyang,MA Hongwei,HE Xiang. Method of measuring 3-D unsteady flow at exits of transonic compressor rotor passages using a two-hole tip-wedge pressure probe[J]. Journal of Aerospace Power,2012,27(10): 2262-2268. (in Chinese) [5] 马宏伟,蒋浩康. 速度梯度、近壁效应和Re数对压力探针测量误差的影响[J]. 航空动力学报,1995,10(4): 337-342.MA Hongwei,JIANG Haokang. Influence of velocity gradient,near-wall effect and Reynolds number on pressure probe measurement error[J]. Journal of Aerospace Power,1995,10(4): 337-342. (in Chinese) [6] GRIMSHAW S D,TAYLOR J V. Fast settling millimetre-scale five-hole probeS[R]. ⅩⅩⅢ Biannual Symposium on Measuring Techniques in Turbomachinery,2016. [7] DÍAZ K M A,ORO J M F,MARIGORTA E B. Extended angular range of a three-hole cobra pressure probe for incompressible flow[J]. Journal of Fluids Engineering,2008,130(10): 1135-1150. [8] REICHERT B A,WENDT B J. A new algorithm for five-hole probe calibration,data reduction,and uncertainty analysis[R]. NASA TM-106458,1994. [9] ANGELES H M,GONZÁLEZ A G. Evaluation of measurement uncertainty in analytical assays by means of Monte-Carlo simulation[J]. Talanta,2004,64(2): 415-422. doi: 10.1016/j.talanta.2004.03.011 [10] KREINOVICH V ,KOSHELEVA O ,POWNUK A ,et al. How to take into account model inaccuracy when estimating the uncertainty of the result of data processing[R]. International Mechanical Engineering Congress and Exposition,IMECE2015-50339,2015. [11] ZAKHAROV I P,VODOTYKA S V. Application of Monte Carlo simulation for the evaluation of measurements uncertainty[J]. Metrology and Measurement Systems,2008,15(1): 117-123. [12] 刘智敏. 用MC仿真计算不确定度[J]. 中国计量学院学报,2005,16(1): 1-7.LIU Minzhi. Calculating uncertainty by MC simulation[J]. Journal of China Jiliang University,2005,16(1): 1-7. (in Chinese) [13] BIPM,IFCC,ISO,et al. Evaluation of measurement data:supplement 1 to the “Guide to the expression of uncertainty in measurement” propagation of distributions using a Monte Carlo method:JCGM 101:2008[S]. Geneva:International Organization for Standardization,2008:10-20. [14] 中国计量大学,中国计量科学研究院,浙江省计量科学研究院. 测量不确定度评定和表示 补充文件1:基于蒙特卡洛方法的分布传播:GB/T 27419-2018[S]. 北京:中国标准出版社,2018:22-26. [15] 北京理工大学,中国计量科学研究院,国家质检总局计量司. 用蒙特卡洛法评定测量不确定度:JJF1059.2-2012[S]. 北京:中国质检出版社,2013:1-11. [16] 周桃庚. 用蒙特卡洛法评定测量不确定度[M]. 北京:中国质检出版社,2013. [17] 陈怀艳. 基于蒙特卡罗法的测量不确定度评定[J]. 电子测量与仪器学报,2011,25(4): 301-308. doi: 10.3724/SP.J.1187.2011.00301CHEN Huaiyan. Evaluation of measurement uncertainty based on Monte Carlo method[J]. Journal of Electronic Measurement and Instrumentation,2011,25(4): 301-308. (in Chinese) doi: 10.3724/SP.J.1187.2011.00301 [18] CROWDER S V,MOYER R D. A two-stage Monte Carlo approach to the expression of uncertainty with non-linear measurement equation and small sample size[J]. Metrologia,2006,43:34- 41 -