A factorial data rate and dwell time experiment in the National Transonic Facility

AIAA 2000-0828A FACTORIAL DATA RATE AND DWELL TIME EXPERIMENTIN THE NATIONAL TRANSONIC FACILITYRichard DeLoach*NASA Langley Research CenterHampton, VA 23681-0001AbstractThis report is an introductory tutorial on theapplication of formal experilnent design methods to windtunnel testing, for the benefit of aeronautical engineerswith little formal experiment design training. It alsodescribes the results of a study to determine whetherincreases in the sample rate and dwell time of theNational Transonic Facility data system wotdd result insignificant changes in force and moment data. Increasesin sample rate from I0 samples per second to 50 samplesper second were examined, as were changes in dwell timefrom one second per data point to two seconds. Thesechanges were examined for a representative aircraftmodel in a range of tunnel operating conditions definedby angles of attack fi'om 0° to 3.8 °, total pressure from15.0 psi to 24.1 psi, and Mach numbers from 0.52 to 0.82.No statistically significant effect was associated with thechange in sample rate. The change in dwell time fromone second to two seconds affected axial forcemeasurements, and to a lesser degree normal forcemeasurements. This dwell effect comprises a"rectification error" caused by incomplete cancellationof the positive and negative elements of certain lowfrequency dynamic components that are not rejected bythe one-Hz low-pass filters of the data system. These lowfrequency effects may be due to tunnel circuitphenomena and other sources. The magnitude of thedwell effect depends on dynamic pressure, with angle ofattack and Mach number influencing the strength of thisdependence. An analysis is presented which suggeststhat the magnitude of the rectification error depends onthe ratio of measurement dwell time to the period of thelow-frequency dynamics, as well as the amplitude of thedynamics. The essential conclusion of this analysis isthat extending the dwell time (or, equivalently,replicating short-dwell data points) reduces therectification error.IntroductionThere is considerable emphasis on minimizing dataacquisition time in the National Transonic Facility (NTF)* Senior Research ScientistCopyright © 2000 by the American Institute ofAeronautics and Astronautics, Inc. No copyright isasserted by the United States under Title 17, U. S. Code.The U. S. Government has a royalty-free license toexercise all rights tinder the copyright claimed herein forGovernment Purposes. All other rights are reserved bythe copyright holder.at NASA Langley Research Center (LaRC) because ofthe relatively high cost of cryogenic operations requiredto achieve flight Reynolds numbers. The NTF datasystem cun'ently uses a sample rate of 10 samples persecond, for a one-second dwell time per data point. Theresulting 10 frames of data are averaged to produce asingle point for each channel of data. The selection ofsample rate and dwell time was based in part on a study ofelectrical noise reduction techniques performed forNASA in 1988 by Wyle Laboratories._ The one-seconddwell and 10-samples/second rate were identified as themost economical combination that would produceacceptable data quality in the presence of line noise andother sources of variation in the instrumentation and datasystem that existed at that time. Because numerouschanges have been made in the data system andinstrumentation in the years since this study, NTFpersonnel decided to reevaluate the sample rate anddwell time settings. Research customers of the NTF havealso requested higher sample rates and longer dwelltimes. The potential benefits, if any, of such proposedchanges in sample rate and dwell time have not been aseasy to quantify as the costs, which for cryogenicoperations are known to be high. So it has not been clearwhether the contemplated changes in data system samplerate and dwell time wouId represent an attractive tradeoffbetween costs and benefits. The experiment described inthis paper was conducted to address this question byexamining the effects on forces and moments ofincreasing the sample rate by a factor of five and thedwell time by a factor of two.There is little time available in the NTF schedule todevote to process experimentation because of thedemands of research applications which comprise thefacility's raison d'&re. This circumstance is not uniqueto NTF, but is common in many industrial applicafi6nswhere facility operating time is precious and operatingcost is high. Formally designed experiments areespecially useful in such situations. They employ highlyefficient experimental techniques capable of providingreliable answers from relatively small data sets. For thisreason, and as part of a larger eflbrt at Langley ResearchCenter to evaluate the utility of formally designedexperiments in wind tunnel research,-" NTF personnelrequested that the LaRC Experimental TestingTechnology Division (ETTD) design a relativelycompact experiment to examine the rate/dwellalternatives under consideration. The objective of theexperiment was to evaluate these combinations of samplerate and dwell time over a representative range ofttmneloperating conditions, while minimizing the NTFschedule interruption required for such an evaluation.This paper describes this effort, and provides sometutorial information on an evolving approach to wind1American Institute of Aeronautics and Astronautics.