Вы находитесь на странице: 1из 8
SESMA Captain Jack Diamond “Knowledge is of two kinds. We know a subject ourselves, ‘or we know where we can find information upon it." Dr. Samuel Johnson D. scrvson pes not pono aphorism, but most contained a strong element of truth. The trouble is, he did not feel the need to elaborate on how to extract @ kemel of information from the huge quantities of data which are available inthe modem world. British Airways Flight Data Recording Department analyses about 250,000 flights per year, with each producing about § megabytes of data. This amounts to about 1250 billion alpha-numeric characters of data on the systems and flight conditions of our airerat ‘The data from the cassettes is read by the SESMA (Special Event Search & Master Analysis) programme, where itis checked against limits set ‘out in the SESMA events listing. These events are concentrated around the take-off and landing stages of the flight and they test whether a particular ‘measure, such as bank angle is greater than a chosen limit. The SESMA programme prints out a trace, or graphical record, of a short period around the event but no flight recorded data is stored on ‘computer. This makes good sense as most of the data recorded on the cassettes is routine and it would be pointless to waste computer memory storing it When a SESMA eventis identified and confirmed an entry is made into a database called SESBASE, a variant of the very successful BASIS group of programmes developed and marketed by British Airways. Each of these entries could be considered ‘as a nugget of useful information culled from the vast quantities of data on the cassettes. ‘The SESMA programme will, by its nature, only store information on the one in two hundred flights which incur SESMA events. This gives a very jaundiced view of what isin general a thoroughly professional and sate operation because it stores no information cn the flights. which are well flown. Sadly, we have not devised a SESMA event which identifies @ superb flying technique so those of you guilty of excellence will have to obtain your reward in Heaven. ‘The amount of data stored in SESBASE amounts to about one megabyte per annum and there is no doubt that itis a very important measure of the flying standards of the British Airways fleet. However, we are at liberty to ask the question: Can we find a better, or complementary, measure? ‘The methodology by which the SESMA computer programme identifies an exceedence offers a possibilty. The maximum vertical acceleration measured during the landing is compared with an event limit and itis recorded in SESBASE when an exceadence occurs. However, ifthe maximum value recorded for every flight is saved in a ‘computer file, it wil provide a database which puts the occasional exceedences in context, We can Use such a database to find the average value of the landing acceleration forall fights and use graphics and statistics to show the spread around the mean. -000- ‘We have developed a programme called MAXVALS which gathers data on about 80 citferent parameters, on each flight. The stored data cannot be traced to a particular flight and so confidentiality is preserved. Its great use is in producing statistics on large ‘numbers of fights and using them to produce trends over time. It can also show comparisons between fleets and between diferent airfields. The rather prosaic name was chosen because the stored value is the highest recorded value of a particular parameter such as airspeed, pitch or bank angle. Each parameteris checked by SESMA and the storage of the values for each fight allows us to know the spread of normal flying of which the SESMA events are an extreme example. A histogram to show the spread of maximum vertical accelerations for 12,682 landings. Frequency Landing vertical acen. for the 734, 1995 Figure 1 “The SESMA limit for vertical acceleration on landing is 1.89 and the table ‘above shows that 33 out of 12,682 landings were over this value, with an average of 1.359 -000- Maxvals stores data on the following: For take-off: ‘Maximum instantaneous pitch rate Average pitch rate Pitch attitude on take-off Speed relative to V2 at take-off ‘Speed relative to V2 at 30 feet ‘Time to 1000 eet Height of first flap reduction. For all of flight regimes: ‘Maximum bank angles CAS relative to Vimo ‘Mach number relative to Mmo ‘Maximum vertical acceleration Maximum speeds for each flap setting, For approach and landing: ‘Speed relative to Vref at 90 seconds to landing ‘Speed relative to Vref at 1000 feet Speed relative to Vref at 500 feet Spoad relative to Vref at 30 feet radio ‘Speed relative to Vref at touchdown Maximum rate of descent below 2000 feet Minimum N1 below 500 feet ‘Maximum positive and negative glideslope deviations between 500 feet and 150 feet radio Pitch attitude at touchdown ‘Maximum vertical acceleration at touchdown Time from 200 kts CAS to touchdown Height above runway at which gear and landing flep were extended. A comparison between the CAS relative to Vref at 30 feet radio for 737-400 and A320 Frequency. Landing CAS at 30R for the 734 versus the A320 CAS rolaivo to Vet at SOfoet ako Figure 2 shows that the autothrottle system on the A320, which is engaged until touchdown is on average more accurate in getting the aircraft over the threshold at a speed close to Vref than is the ‘manual throttle system on the 737-400. However, there is a greater spread of speeds away from the mean on the A820. This is probably caused by the autothrottle selecting a speed which takes account of the potential windshear expected near the runway. The pilot, when handling the throttles is much less likely to accept a speed below planned threshold speed, as can be shown by the very few 734 approaches with a speed below Vref. Itis not possible to draw very firm conclusions from the relative speeds on the A320 because the FMS determines and displays the target threshold speed, whereas the Vref assumed by the SESMA programme is calculated from the basic values in the flight manual. The proportion of fights with a threshold speed of above +20 kts is roughly the same on both types, Figure 2 about 1 in 280 sectors, although many of these higher speeds are likely to happen on days with strong headwinds and would not signify any potential for over-runs. ‘The information gathered from the two aircraft types does pose a question about the advantages in upgrading the 797-400 autothrottie system so that it could be used while the airoraft is flown by hand. This is probably more complicated than it sounds because the A320 flight control laws are designed to protect the pilot from any adverse trim changes with thrust, On the 737-400 the undersiung engines can have a marked affect on pitch trim and could destabilise the descent path ifthey were active, Under control of the autothrottle system, near the ground. Would the potential savings in tyre and brake wear be greater than the cost of developing the autothrottle? ‘A comparison between the CAS relative to Vref at 1000 feet for the 747-400 and the 757E4 Frequency nots relative fo Viet th i tidal CAS rel. at 1000ft for the 744 versus the 75704 Figure 3 Figure 3 shows that at 1000 feet radio the 744 is flown with a speed distribution tightly peaked around the mean of Vref +8.9 kts. The 757 has a mean of Vref +15.6 kts and a distribution with a pronounced tail in the direction of excess speed. In particuler the 757 has 32 out of 1561 flights with a speed of greater than Vref +40 kis at 1000 feet radio, whereas the 747-400 has only 2 out of 3316 flights with the same excess. The reasons for this disparity are not easy to explain but its only because of the MAXVALS program that the question can be posed. ‘The following suggestions are made with due humility: 1. AN ATC requirement to maintain 160 kts to 4 miles is an incrementally higher excess on the 757. 2. The 757 is more slippery and itis more difficult tolose speed than the 747-400. 3. The wording of the stabilised approach criteria varies between fleets. 4, The ATC environment in which the 757 habitually flies is more likely to request a slightly higher speed untll a late stage on the approach. -000- Statistics on landings of the 737-400 fleet ‘The table and graph on the following pages are made by extracting data from MAXVALS for roughly 412,000 landings on the combined 737-400 fleets. The stored information is based on event SOA in ‘SESMA which measures the depth of landing beyond the glide-slope aerial. ‘The airfields are atranged in order of the average depth of landing, with the shortest at the top. The measurement of the landing depth has potential random errors in that it requires a number of flight data parameters to be measured at the same instant during the landing. Although the values in the column for longest landing cannot be guaranteed, the averages and standard deviations should be more accurate, Table produced from the MAXVALS programme for landings on the 737-400 fleet [Aipors [Number of [average ont Standard |longest |e perentie [average ot exc Neraela lcings [pastes aerate | deviation —_[anding ret t 30 feet iw 202 112] 256 20 xo] 125 at ER %1 2 zm 70 692 5.09 1.07 FAO a 303 3) 73 ai 727 1.385 Bua 16] 312 za] 1170 i071 638 | EDI Ey 958 aoa] 1280 696 7.8 1.0 [a 25 502 267] 1194 687 72 1a MPL 2 506 45] 68 7113 792 1.3 FU “8| 555 seal itt 820 750 1.39 svG 4 556 sa] 1289 980 7.32 7135 NCL. 234 72 si7] 1384 734 739 1.38 NAP 138 573] sor] 2207 992 574 7.43 NOE za 565 4551504] 1197 372 “20 SN 2] 09] 20| 1600] 1074 5087 135 VE a 25 2e3] 1504 1207 355 1. [cP 7 9 2x0) 1000 72 302 1.35 ann 122] a7 ar] 1058 77 137 foo. 294 733 496] 2204 1090 ans 133 KEP, “2 733 263] 1179 203 7.35 12] [ace 17 735 226] 1694 954 a9 17 [cua Ea 738 35) 158i 887 791 1.32 MAN 201 742 as] 1047 112i Zn 1.35 wav 96 748 aa] 1873 1051 5a 1.35 BFS: B17 722 ari] 2055 177 272 1.33 BRU “34 768 aa] eter 109 778 1.35 cor 181 778 400] 1897 099 7.33 735 HEL 195 822 soo] 2128 978 728 1.35 LED) 3 291 oa] 1059 “22 727] 1.39 Le aaea eat soo 2759 975 34 134 as 055 e4 anz|_ 240] 1050 aii 133 BSL a1 e6t 106[ 1972 1239 349 133] fms 23] 80 ss0| 1505 1309 327 1. Lys: 31 202 306] 1816 250] 773 157 [ean 125 299 245] 1820 253] 391 135] Low: zara 606 Ee] 375 350] 135 FRA 210 16 soe] 2472 war a 137] uw a5 28 235] 1322 1598 751 142] aw 204] 29 ue] 2279 1105 922 137 Zan. Fir] 298 405] 1009 1013 89 135| ran 3 975 zx] 1658 721 54 135| roa 310] 78 as] 1577 880 250 1.38 fate 2 288 zi 1499 203 7.09 a] [svo ai 7088 aso] 2074 7124 208 10] wk, 9] “108 250] 1994 25 800 131 va “79 in| saat 998 a7 135 FO) 9 128] ai] __2302 “102 203 133 Landing distance past glidescope aerial (98 percentile Average dist Figure 4 Figure 4 shows that pilots behave in a professional runways but this is in the order of a few knots. Short manner in that they touch down ‘olose to the landing tends to be at the expense of smoothness in numbers" on those short runways. The speed that the aircraft touches down with a higher rate of excess above Vref also tends to be less on shor descent. In general deep landings are only carried out at airfields with long runways. Sometimes this is because the turn-off point for the runway is at the far end (viz. FCO, ZAH and ATH). The column for 98 percentile is 2.5 times the standard deviation, thus when this is added to the average value it gives an estimate of the depth of landing which encompasses 98% of all landings. (Figure 4.) -000- High energy approaches into San Francisco Approach Pos GS deviation for the 744, 1996 Froquoney ‘bats postive davon betwean 600 and 160 rte Figure 5 The typical approach into San Francisco leaves the aircraft about 3000 feet high at about 30 miles to touchdown. The problem is that the actual distance to touchdown is a moveable feast which is determined by the approach controller in his efforts to fit the aircraft into a traffic stream on parallel runways. Ifa high rate of descent is not initiated immediately a quick turn-in is requested, the aircraft is likely to be above the glideslope on tuming onto the extended centretine. Figure § Is based on SESMA event 56B which ‘measures the maximum glideslope deviation between 600 feet and 150 feet radio on the final ‘approach to land. This information has been used by fleet management to approach the San Francisco ATC authorities in order to point out our misgivings at crews being “encouraged” to perform such approaches and it will also be passed on to the crews in the form of a newsletter item. ‘The average for all destinations is 0.6 dots fly down. For SFO the average is higher. Conclusions ‘The MAXVALS program is in its infancy and the above graphs are but a small proportion of the many comparisons which can be made. It can be used to monitor various aspects of fight operations such as the adherence to the correct take-off path and speeds, the monitoring of approaches to land with a special emphasis on achieving a stable approach path below 1000 feet and also keeping a check on whether aircraft are operated outside structural limits, It can be used to improve our understanding of how well we adhere to our Standard Operating Procedures, especially during periods of extensive training We can now check whether the operation into a particular airfield gives rise to any serious safety concern and we have the data to make our case. For instance, we have been able to show how rough runways at certain airfields cause much higher than average vertical accelerations on the undercarriage ‘during take-off and can use this data to persuade the authorities to resurface or reprofile offending runways, Itcan be used by a manager to check on the significance of an incident; for example, it would be possible to determine whether a tailscrape on landing was an isolated incident or a symptom of a tendency to oversfiare on landing. ‘The data is only useful because it presents statistios gamered from a large number of fights and it is not meant as a means of checking on individual fights. 40

Вам также может понравиться