Correction of Aircraft Flux Valve Based Heading for Two-Dimensional Winds Aloft Calculations Using Weather Model Comparisons

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 11:45 AM
Correction of Aircraft Flux Valve Based Heading for Two-Dimensional Winds Aloft Calculations Using Weather Model Comparisons
2B (Washington State Convention Center)
Daniel J. Mulally, AirDat, LLC., Lakewood, CO; and A. K. Anderson
Manuscript (1.4 MB)

Among the atmospheric measurements reported by AirDat's TAMDAR in-situ aircraft sensor are two-dimensional winds aloft (speed and direction). The accuracy of the aircraft heading is very important for this calculation and even an error of as little as one degree can degrade the accuracy of the wind calculation. Flux valve (magnetic) based systems are often prone to errors; we've found that hard iron effects are likely the main cause. Airline operations can tolerate errors up to several degrees and even though the systems are periodically calibrated, errors of four degrees or more are not uncommon making the resulting TAMDAR wind calculation unusable.

The effect of poor heading systems can easily be seen by considering the wind performance of the TAMDAR equipped Chautauqua ERJ-145 fleet. There are two types of heading systems in the fleet: approximately one half of the fleet has the Honeywell AH-800 AHRS (Altitude Heading Reference System) and the others have the AH-900 AHRS. The AH-800 heading is based on a flux-valve system whereas the AH-900 is based on a much more accurate ring laser gyro system. The latter system does not use the earth's magnetic field. When wind performance (RMS magnitude vector wind difference from various weather models) is compared for the two systems, the wind error for the AH-800 heading source is typically almost twice as bad as that of the AH-900.

The method developed by the authors characterizes the heading errors as a function of heading; this produces a magnetic deviation curve unique to each airplane. A lookup table based on this characterization is then uploaded to the TAMDAR unit and used to pre-correct the heading used in the wind calculation. The curve is derived from a long term analysis (several months) of comparisons of the TAMDAR reported winds and weather model predictions. The models used are the NCEP NAM, GSD Rapid Refresh and AirDat's RTFDDA. Observations from AMDAR planes are also used to verify the trend. It's important to note that the wind results themselves are not adjusted; only the heading data going into the wind calculation is adjusted.

The Mesaba fleet of SAAB-340s use flux valves as the basis for the heading. The fleet at the time of this study (30 October 2010) consisted of 33 SAABs. Eighteen of these were delivering winds of poor quality. A magnetic deviation correction table was uploaded to the TAMDARs on each of these poorly performing planes. After this upload, the winds aloft performance compared to the combined model data (for the models listed above) on all eighteen had improved to the point where they were considered good.

Another beneficial effect was to reduce the SAAB descent wind errors so that they are almost as good as the ascent winds. Previously descent winds were much worse than ascent winds. The reason for this improvement is not clear and it is being investigated.

Preliminary results from applying the method to the six worst performing flux-valve based ERJ-145s are encouraging. All six showed dramatic improvement; and the wind quality category has been changed from bad to good. Although winds are still not as good as those planes using the AH-900, these six ERJs perform about as well as the Mesaba SAAB fleet after its performance was improved with this method.

The variability of the inclination and strength of the earths magnetic field is a strong function of magnetic latitude; thus, it is expected that a particular plane's lookup table works over a limited range of magnetic latitudes. Results indicate that planes flying over a region as large as CONUS can still show significant improvement from the method.