Narrative:

We received a TCAS RA but there was no loss of separation. This report is submitted to highlight an example of inadequate performance of the TCAS system for entry in the NASA ASRS database. The event that I am reporting is not that there was an near midair collision or even an RA; but the poor performance of the TCAS algorithms that could have led to an near midair collision. We were in the descent for arrival at ZZZ. I think we were still talking to center. We were told to maintain 11000 ft and to expect lower crossing traffic. The traffic was a B737 climbing out on a reciprocal course. They were told to maintain 10000 ft and expect higher after they passed us. We picked them up visually at about 7 miles while still in our company's fuel-saving descent profile (3000 FPM). We knew the other aircraft intended to level off below our assigned altitude. I do not believe they had called us in sight at that time. About 13000 ft I backed off the descent rate to smooth out the level off in accordance with company procedures. As the autopilot achieved the newly-commanded descent rate; the TCAS commanded a descent RA in excess of 3000 FPM. I decided that the safest course of action was to comply with our clearance and complete our level off at 11000 ft rather than stuff the nose and descend through the oncoming aircraft's flight path. To do so would have ensured a loss of separation and might have injured our flight attendant who was up doing compliance checks for landing. I looked back outside and saw that the B737 had leveled off below us and started an evasive turn away. I presume that the other aircraft had a climb RA but decided to ignore it as well; but; not knowing what I would do; made a turn to get out of my way just in case. In the case of the aircraft which was already climbing; complying with a climb RA might well have put it in a nose-high unusual attitude with decaying airspeed and required maximum available (engine-damaging) thrust setting. We each leveled off at our assigned altitudes and there was no loss of separation. Flight crew recognized that further descent as commanded by the RA would have led to loss of separation. The 'evasive action' taken under 'captain's emergency authority' was to disregard the RA and comply with ATC altitude assignment. The flight crew had situational awareness beyond the trajectory predictions of the TCAS. We knew that the target aircraft knew it had to level off below us and intended to do so. We had already begun our level off above the target aircraft's assigned altitude. The human factor is noted as a positive that avoided an incident. Less experience might have led to thoughtless adherence to an inappropriate RA and an near midair collision. The crews of both aircraft disregarded their respective RA's and leveled off at their assigned altitudes rather than comply with RA's that would have created a loss of separation. We made this decision based on situational awareness and common sense. The TCAS has neither of these advantages. It could not anticipate that both aircraft planned to level off 1000 ft apart. What concerns me most is the apparent latency of the TCAS processing. It saw the high climb rate of the other aircraft and our high descent rate and projected a conflict. That is fine; but it did not give the RA command until we had already slowed our descent rate. It seemed to calculate its projection based on a snapshot of vertical speeds without considering the rate of change of vertical speeds. The RA also came a few seconds after our high descent rate had been swallowed so the conflict geometry no longer applied when the RA was given. I am also concerned about the uncertainty of predicting the actions of the other aircraft. Each leveled off as assigned which was the best course of action and prevented a loss of separation. If this had happened in IMC; both aircraft would probably have complied with the RA's. That would have caused a loss of separation; but it would probably be only a near midair. The worst case would have beenif one aircraft had complied and the other did not. That would have placed the complying aircraft squarely in the path of the non-complying aircraft. This is an issue of TCAS algorithm programming. I don't have the expertise to comment on that; but it seems that it needs to process faster and take rates of vertical speed change; pitch rate; autopilot pitch rate programming and changing autopilot commands into account. This is not the first time that I have seen an RA that was inappropriate or came too late. It seems that RA's can't be counted upon; but telling crews to disregard the TCAS is problematic as well. Perhaps TCAS training should mention that it is acceptable to disregard an RA if the crew has sight of the intruding aircraft and the RA seems inappropriate.callback conversation with reporter revealed the following information: the reporter stated he flies an EMB145 for an air carrier. He has had one other traffic conflict with a TCAS anomaly -- he observed an aircraft in his one O'clock position moving right to left; and after the aircraft reached his eleven O'clock position and was flying away from any conflict zone; he received a TA; which he ignored. He expressed concern that TCAS system latency is excessive; and combined with possibly flawed algorithms TCAS can generate inaccurate TA's and RA's.

Google
 

Original NASA ASRS Text

Title: AN EMB145 CAPT DISREGARDED A TCAS RA COMMAND BECAUSE IN HIS JUDGEMENT IT WOULD HAVE PUT HIM IN CONFLICT WITH THE OPPOSITE DIRECTION AIRCRAFT.

Narrative: WE RECEIVED A TCAS RA BUT THERE WAS NO LOSS OF SEPARATION. THIS REPORT IS SUBMITTED TO HIGHLIGHT AN EXAMPLE OF INADEQUATE PERFORMANCE OF THE TCAS SYSTEM FOR ENTRY IN THE NASA ASRS DATABASE. THE EVENT THAT I AM REPORTING IS NOT THAT THERE WAS AN NMAC OR EVEN AN RA; BUT THE POOR PERFORMANCE OF THE TCAS ALGORITHMS THAT COULD HAVE LED TO AN NMAC. WE WERE IN THE DESCENT FOR ARRIVAL AT ZZZ. I THINK WE WERE STILL TALKING TO CENTER. WE WERE TOLD TO MAINTAIN 11000 FT AND TO EXPECT LOWER CROSSING TRAFFIC. THE TRAFFIC WAS A B737 CLIMBING OUT ON A RECIPROCAL COURSE. THEY WERE TOLD TO MAINTAIN 10000 FT AND EXPECT HIGHER AFTER THEY PASSED US. WE PICKED THEM UP VISUALLY AT ABOUT 7 MILES WHILE STILL IN OUR COMPANY'S FUEL-SAVING DESCENT PROFILE (3000 FPM). WE KNEW THE OTHER AIRCRAFT INTENDED TO LEVEL OFF BELOW OUR ASSIGNED ALTITUDE. I DO NOT BELIEVE THEY HAD CALLED US IN SIGHT AT THAT TIME. ABOUT 13000 FT I BACKED OFF THE DESCENT RATE TO SMOOTH OUT THE LEVEL OFF IN ACCORDANCE WITH COMPANY PROCEDURES. AS THE AUTOPILOT ACHIEVED THE NEWLY-COMMANDED DESCENT RATE; THE TCAS COMMANDED A DESCENT RA IN EXCESS OF 3000 FPM. I DECIDED THAT THE SAFEST COURSE OF ACTION WAS TO COMPLY WITH OUR CLEARANCE AND COMPLETE OUR LEVEL OFF AT 11000 FT RATHER THAN STUFF THE NOSE AND DESCEND THROUGH THE ONCOMING AIRCRAFT'S FLIGHT PATH. TO DO SO WOULD HAVE ENSURED A LOSS OF SEPARATION AND MIGHT HAVE INJURED OUR FLIGHT ATTENDANT WHO WAS UP DOING COMPLIANCE CHECKS FOR LANDING. I LOOKED BACK OUTSIDE AND SAW THAT THE B737 HAD LEVELED OFF BELOW US AND STARTED AN EVASIVE TURN AWAY. I PRESUME THAT THE OTHER AIRCRAFT HAD A CLIMB RA BUT DECIDED TO IGNORE IT AS WELL; BUT; NOT KNOWING WHAT I WOULD DO; MADE A TURN TO GET OUT OF MY WAY JUST IN CASE. IN THE CASE OF THE AIRCRAFT WHICH WAS ALREADY CLIMBING; COMPLYING WITH A CLIMB RA MIGHT WELL HAVE PUT IT IN A NOSE-HIGH UNUSUAL ATTITUDE WITH DECAYING AIRSPEED AND REQUIRED MAXIMUM AVAILABLE (ENGINE-DAMAGING) THRUST SETTING. WE EACH LEVELED OFF AT OUR ASSIGNED ALTITUDES AND THERE WAS NO LOSS OF SEPARATION. FLIGHT CREW RECOGNIZED THAT FURTHER DESCENT AS COMMANDED BY THE RA WOULD HAVE LED TO LOSS OF SEPARATION. THE 'EVASIVE ACTION' TAKEN UNDER 'CAPTAIN'S EMERGENCY AUTHORITY' WAS TO DISREGARD THE RA AND COMPLY WITH ATC ALTITUDE ASSIGNMENT. THE FLIGHT CREW HAD SITUATIONAL AWARENESS BEYOND THE TRAJECTORY PREDICTIONS OF THE TCAS. WE KNEW THAT THE TARGET AIRCRAFT KNEW IT HAD TO LEVEL OFF BELOW US AND INTENDED TO DO SO. WE HAD ALREADY BEGUN OUR LEVEL OFF ABOVE THE TARGET AIRCRAFT'S ASSIGNED ALTITUDE. THE HUMAN FACTOR IS NOTED AS A POSITIVE THAT AVOIDED AN INCIDENT. LESS EXPERIENCE MIGHT HAVE LED TO THOUGHTLESS ADHERENCE TO AN INAPPROPRIATE RA AND AN NMAC. THE CREWS OF BOTH AIRCRAFT DISREGARDED THEIR RESPECTIVE RA'S AND LEVELED OFF AT THEIR ASSIGNED ALTITUDES RATHER THAN COMPLY WITH RA'S THAT WOULD HAVE CREATED A LOSS OF SEPARATION. WE MADE THIS DECISION BASED ON SITUATIONAL AWARENESS AND COMMON SENSE. THE TCAS HAS NEITHER OF THESE ADVANTAGES. IT COULD NOT ANTICIPATE THAT BOTH AIRCRAFT PLANNED TO LEVEL OFF 1000 FT APART. WHAT CONCERNS ME MOST IS THE APPARENT LATENCY OF THE TCAS PROCESSING. IT SAW THE HIGH CLIMB RATE OF THE OTHER AIRCRAFT AND OUR HIGH DESCENT RATE AND PROJECTED A CONFLICT. THAT IS FINE; BUT IT DID NOT GIVE THE RA COMMAND UNTIL WE HAD ALREADY SLOWED OUR DESCENT RATE. IT SEEMED TO CALCULATE ITS PROJECTION BASED ON A SNAPSHOT OF VERTICAL SPEEDS WITHOUT CONSIDERING THE RATE OF CHANGE OF VERTICAL SPEEDS. THE RA ALSO CAME A FEW SECONDS AFTER OUR HIGH DESCENT RATE HAD BEEN SWALLOWED SO THE CONFLICT GEOMETRY NO LONGER APPLIED WHEN THE RA WAS GIVEN. I AM ALSO CONCERNED ABOUT THE UNCERTAINTY OF PREDICTING THE ACTIONS OF THE OTHER AIRCRAFT. EACH LEVELED OFF AS ASSIGNED WHICH WAS THE BEST COURSE OF ACTION AND PREVENTED A LOSS OF SEPARATION. IF THIS HAD HAPPENED IN IMC; BOTH AIRCRAFT WOULD PROBABLY HAVE COMPLIED WITH THE RA'S. THAT WOULD HAVE CAUSED A LOSS OF SEPARATION; BUT IT WOULD PROBABLY BE ONLY A NEAR MIDAIR. THE WORST CASE WOULD HAVE BEENIF ONE AIRCRAFT HAD COMPLIED AND THE OTHER DID NOT. THAT WOULD HAVE PLACED THE COMPLYING AIRCRAFT SQUARELY IN THE PATH OF THE NON-COMPLYING AIRCRAFT. THIS IS AN ISSUE OF TCAS ALGORITHM PROGRAMMING. I DON'T HAVE THE EXPERTISE TO COMMENT ON THAT; BUT IT SEEMS THAT IT NEEDS TO PROCESS FASTER AND TAKE RATES OF VERTICAL SPEED CHANGE; PITCH RATE; AUTOPILOT PITCH RATE PROGRAMMING AND CHANGING AUTOPILOT COMMANDS INTO ACCOUNT. THIS IS NOT THE FIRST TIME THAT I HAVE SEEN AN RA THAT WAS INAPPROPRIATE OR CAME TOO LATE. IT SEEMS THAT RA'S CAN'T BE COUNTED UPON; BUT TELLING CREWS TO DISREGARD THE TCAS IS PROBLEMATIC AS WELL. PERHAPS TCAS TRAINING SHOULD MENTION THAT IT IS ACCEPTABLE TO DISREGARD AN RA IF THE CREW HAS SIGHT OF THE INTRUDING AIRCRAFT AND THE RA SEEMS INAPPROPRIATE.CALLBACK CONVERSATION WITH RPTR REVEALED THE FOLLOWING INFO: THE REPORTER STATED HE FLIES AN EMB145 FOR AN ACR. HE HAS HAD ONE OTHER TRAFFIC CONFLICT WITH A TCAS ANOMALY -- HE OBSERVED AN AIRCRAFT IN HIS ONE O'CLOCK POSITION MOVING RIGHT TO LEFT; AND AFTER THE AIRCRAFT REACHED HIS ELEVEN O'CLOCK POSITION AND WAS FLYING AWAY FROM ANY CONFLICT ZONE; HE RECEIVED A TA; WHICH HE IGNORED. HE EXPRESSED CONCERN THAT TCAS SYSTEM LATENCY IS EXCESSIVE; AND COMBINED WITH POSSIBLY FLAWED ALGORITHMS TCAS CAN GENERATE INACCURATE TA'S AND RA'S.

Data retrieved from NASA's ASRS site as of January 2009 and automatically converted to unabbreviated mixed upper/lowercase text. This report is for informational purposes with no guarantee of accuracy. See NASA's ASRS site for official report.