1、_SAE Technical Standards Board Rules provide that: “This report is published by SAE to advance the state of technical and engineering sciences. The use of this report is entirely voluntary, and its applicability and suitability for any particular use, including any patent infringement arising theref
2、rom, is the sole responsibility of the user.” SAE reviews each technical report at least every five years at which time it may be reaffirmed, revised, or cancelled. SAE invites your written comments and suggestions. Copyright 2011 SAE International All rights reserved. No part of this publication ma
3、y be reproduced, stored in a retrieval system or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of SAE. TO PLACE A DOCUMENT ORDER: Tel: 877-606-7323 (inside USA and Canada) Tel: +1 724-776-4970 (outside US
4、A) Fax: 724-776-0790 Email: CustomerServicesae.org SAE WEB ADDRESS: http:/www.sae.orgSAE values your input. To provide feedback on this Technical Report, please visit http:/www.sae.org/technical/standards/ARP5758AEROSPACERECOMMENDEDPRACTICEARP5758 Issued 2011-03Trend Analysis for Maintaining Correla
5、tion of Gas Turbine Engine Test Cells RATIONALEThe FAA has issued Advisory Circular, AC43-207, on December 26, 2002 that recommends a 7 year re-correlation, trending or periodic checks. The FAA, AC43-207 bases their recommendation on the SAE Aerospace Recommended Practice (ARP) 741 rev. B (or latest
6、 revision).This document describes a recommended practice and procedure for the trending of parameters to maintain the test cell correlation status. Trending is performed to monitor test cells for changes that can affect engine performance or the data acquired from engine tests. Over time the qualit
7、y of the acceptance test data for an engine can be affected by the test cell environment. This could lead to unanticipated and expensive results of either accepting a poor performing engine or rejecting a good performing engine. The use of trending the data for engine acceptance can avoid such costl
8、y errors. TABLE OF CONTENTS RATIONALE 1 1. SCOPE 21.1 General . 21.2 Benefits . 21.3 Limitations . 22. APPLICABLE DOCUMENTS 32.1 SAE International Publications 32.2 Other References 33. DEFINITIONS . 34. TEST CELL TRENDING . 44.1 General . 44.2 Factors Affecting Performance Measurement of Gas Turbin
9、e Engines . 64.3 Parameters to Trend . 74.4 Parameter Trend Requirements . 75. EXAMPLE . 85.1 Typical Trend 85.2 Trend Shifts . 9Copyright SAE International Provided by IHS under license with SAENot for ResaleNo reproduction or networking permitted without license from IHS-,-,-SAE ARP5758 Page 2 of
10、91. SCOPE 1.1 General This document describes a recommended practice and procedure for the trending of parameters to maintain the test cell correlation status. Trending is performed to monitor test cells for changes that can affect engine performance or the data acquired from engine tests.1.2 Benefi
11、ts This recommended practice will benefit the original equipment manufacturer (OEM), commercial users, repair stations, and military depots as well as intermediate level maintenance activities. Specific cases in which the information contained herein will be beneficial are: a. As a recommended and l
12、ess expensive method to maintain test cell correlation status. b. As a method for maintaining correlation of test cell data between engine, airframe and third party overhaul centers supporting:1. Commercial requirements 2. Military requirements 3. As an early indicator for changes in performance suc
13、h as: (a). Performance changes driven by the test cell (b). Performance changes driven by the engine (c). Performance changes driven by the measurement system By trending engine and test cell parameters it can be determined whether changes have occurred to the test stand, instrumentation, or engine
14、that will affect the performance data collected in that specific test stand. The trended data allows the test cell owner to be confident that the data taken from tests in the test cell is valid and accurate. If a change tothe test cell, instrumentation, or the engine has occurred, trending allows th
15、e change to be identified early. If the test cell is subjected to abnormal test conditions (i.e. stalls, surges, blade-outs, engine failure) either intentional orotherwise which can damage instrumentation or the test facility hardware, then data trending will show a shift in the trend indicating a r
16、esulting effect. Trending will also capture instrumentation drift.After a test facility is correlated and the correlation factor has been determined, trending is an effective way to maintain confidence in the correlation. The trending of the data needs to commence immediately following the correlati
17、on test. Therefore the test cell does not need to be re-correlated as often. Indeed, referring to the reference documents, trending is a recommended alternative to a physical re-correlation of the test cell. If trending is performed on a consistent basis then re-correlation need only be performed wh
18、en deemed necessary based on the trended data, engineering judgment, or if modifications have been made to the test cell.The trend data can be taken a step further and the operator can compare the trend data not only to the correlation factors, but they can also use the trend data to monitor the sho
19、p performance, engine build standards and overall fleet health for that engine model.1.3 Limitations This recommended practice is meant as a general guide to trending; therefore specific practices and details may be instituted by the OEM or government agencies. Although this document describes the p
20、ractice of trending gas turbine engine parameters, these trending practices can be used for trending any type of measurement. Copyright SAE International Provided by IHS under license with SAENot for ResaleNo reproduction or networking permitted without license from IHS-,-,-SAE ARP5758 Page 3 of 92.
21、 APPLICABLE DOCUMENTS The following publications form a part of this document to the extent specified herein. The latest issue of SAE publications shall apply. The applicable issue of the other publications shall be the issue in effect on the date of the purchase order. In the event of conflict betw
22、een the text of this document and references cited herein, the text of this document takes precedence. Nothing in this document, however, supersedes applicable laws and regulations unless a specific exemption has been obtained. 2.1 SAE International Publications Available from SAE International, 400
23、 Commonwealth Drive, Warrendale, PA 15096-0001, Tel: 877-606-7323 (inside USA and Canada) or 724-776-4970 (outside USA), www.sae.org.ARP5435 APU Gas Turbine Engine Test Cell Correlation ARP741 Turbofan and Turbojet Gas Turbine Engine Test Cell Correlation ARP4755 Turboprop/Turboshaft Gas Turbine Eng
24、ine Test Cell Correlation 2.2 Other References Downing Ph.D., Douglas, and Jeff Clark, Ph.D. “Statistics The Easy Way.” Barrons Educational Series, Inc., 1989. Wheeler, Donald J.; and Chambers, David S. “Understanding Statistical Process Control”; SPC Press, Inc.; Second Edition, 1992. Rolls Royce P
25、LC Publication Reference: PPR 1474; Issue No.1; March 2007. Federal Aviation Authority (FAA) Advisory Circular, AC43-207 issued on December 26, 2002. 3. DEFINITIONS a. CALIBRATION: The comparison of a particular instrument or system with a standard of known accuracy b. CORRELATION: The comparison of
26、 engine performance parameters measured on a common engine tested in two test facilities, where one facility is the reference c. CORRELATION FACTOR: A multiplier used where appropriate to adjust for the difference in performance between the customer facility and a reference facility, also known as a
27、 “correction factor” or a “facility modifier”. d. ENGINE DRESS KIT: Typically consists of aerodynamic hardware, accessories, and test instrumentation required to permit operation of the engine in the test cell. e. INDOOR TEST CELL: A facility for the testing of gas turbine engines in an enclosed env
28、ironment. f. OUTDOOR TEST STAND: An open air facility, without any enclosure, for testing gas turbine engines g. TEST FACILITY (TEST CELL): An area in which a gas turbine engine is operated to determine its performance and other information as required by a given test. h. TRENDING: The statistical p
29、ractice of recording and plotting parameters over time. i. Outlier: Data sample that is outside of predetermined tolerance band. Copyright SAE International Provided by IHS under license with SAENot for ResaleNo reproduction or networking permitted without license from IHS-,-,-SAE ARP5758 Page 4 of
30、94. TEST CELL TRENDING 4.1 General When Gas Turbine engines are tested in a test cell, the characteristics of the test cell have a significant effect on the engine performance data. Engine performance data, within the same engine family, follow normal variations (natural process variation) within st
31、atistically acceptable limits. Such variations are caused by engine-to-engine differences, and variations in other factors (e.g. humidity, fuel, ambient conditions, etc.) which are analytically corrected for, but carry a certain amount of uncertainty within the correction. Changes in the test cell c
32、haracteristics can be detected by a trend-analysis of the engine performance data for pass-off parameters, and detecting points that fall outside the statistical limits of normal variation, or outliers. It may be noted that an outlier may also be caused by a change in the engine production, or in th
33、e post-overhaul assembly process, or the inadvertent installation of defective parts. Hence, the observation of an outlier in the data plots may not necessarily indicate a change in the test cell characteristics. It is recommended that a list of parameters be prepared for trending before starting th
34、e process. The OEM, or other correlation agency, and the production/overhaul shop should have a formal agreement on the list. Typically the parameters are the ones used for production or post overhaul pass-off testing. Trending is the statistical process of recording and plotting a parameter against
35、 time. As the parameter is trended the average is determined and the upper and lower control limits are established, indicating the acceptable band within the natural process variation. The control limits can be determined by a combination of several methods.x Standard Deviation Techniques x Statist
36、ical Process Control (SPC) x OEM defined tolerance bands Engineering judgment must be applied when using any of these methods. Trending must be started immediately following the correlation test. Once a fair amount of data is collected (at least 10 data points for a parameter) trending analysis can
37、be started. However, to be statistically significant and more meaningful, at least 30 points are needed. For a sample size smaller than 10, and in facilities where engines are tested infrequently, an appropriate statistical process control methodology (like Individual Moving Range, Ixmr) may be used
38、. It is recommended that a running frequency histogram be plotted for each parameter. As an example, the following table is a list of the measured/corrected value of a parameter from testing the same engine-type at the same facility on different dates.TABLE 1 - DATA OBTAINED FOR A PARAMETER FROM TES
39、TING 0.951 0.898 0.951 0.902 0.919 0.898 0.901 0.908 0.904 0.912 0.908 0.920 0.951 0.8990.921 0.951 0.925 0.908 0.907 0.908 0.909 0.903 0.916 0.900 0.925 0.918 0.907 0.9140.908 0.898 0.902 0.909 0.951 0.920 0.899 0.909 Please note that for the purpose of this analysis, the unit of the parameter is i
40、rrelevant. There are 36 data points in the above sample. The minimum (min) value of the parameter from the test data in Table 1 is 0.898, while the maximum (max) value is 0.951. The range, R of the entire sample, which is the difference between the max and the min, is 0.055. By definition, R is the
41、smallest value in the set of data subtracted from the largest value. It would be sufficient, in this case, to create between 10 and 15 intervals, each 0.005 units wide, for plotting the frequency histogram. There is no set procedure to determine the number of intervals. One method is to take the squ
42、are root of the total number of data points rounded to the nearest whole number (in this case, this method will yield 6 intervals). A second method is to use judgment and experience with a general guide line of approximately 10 intervals for 30 to 100 data points in the sample. The histogram prepare
43、d from the above data is given in Figure 1 below. Copyright SAE International Provided by IHS under license with SAENot for ResaleNo reproduction or networking permitted without license from IHS-,-,-SAE ARP5758 Page 5 of 9FIGURE 1 - FREQUENCY HISTOGRAM OF THE DATA FROM TABLE 1. Note that the X-axis
44、is the Interval No. and the Y-axis is the number of data points captured in the range of the specific interval (for example, Interval No. 6 has 3 data points). There are 12 intervals in the above histogram and it reveals a peak at the 12thinterval which is centered over the interval 0.950-0.954. Thi
45、s interesting aspect of the data is difficult to be brought out without the aid of the histogram plot. Even without doing any further analysis (like plotting the control charts, etc.), an investigation should be conducted to find the reason for the unexpected peak. The following is an explanation of
46、 the use of standard deviation. The average, xof a set of data for a parameter, x is given by nxxnii 1, (1)where n is the number of data points in the sample. To find the upper and lower control limits the standard deviation (1 ) must be determined. Standard deviation, 1 can be found by using, 1)(12
47、 nxxniiV(2)Once the standard deviation is found it can be used to determine the upper and lower control limits. The control limits are defined by a multiple of the standard deviation from the average. For example, a 21 spread would mean that the distance between the upper and lower control limits is
48、 twice the standard deviation. Therefore, the upper control limit would be one standard deviation above and the lower control limit would be one standard deviation below the average value. This establishes the band around the average for a 21 spread. In statistical process control, the natural process variation is considered to be the 61 spread, or 31 band from the average. For the data to be acceptable, the pre-set process band (11 , 21 , or 31 ) must be within the acceptable band given by the OEM. It may be noted that in some cases, the
