AAIB Bulletin: 10/2007

the engnes due to the ambent nose around them. The arcraft rolled to wngs level but then contnued to roll to its right pausing briefly at a bank angle of approximately 30º before the rght wng and nose appeared to drop and the arcraft dsappeared behnd some trees. Recorded information The aircraft was not fitted with a Flight Data or Cockpit Voce recorder, and was not requred to be so equpped. National Air Traffic Services, the provider of en‑route air traffic control services throughout the UK, provided recorded radar data for both the outbound flight to Teesside and the return flight to Denham. This data ncluded both alttude and poston. From the recorded radar data, the ground track of the aircraft and the vertical profile of the outbound and return flights were established. The ground track distance for the outbound flight was 196 nm and the return ground distance flown was 184 nm. This was a total increase of 24 nm over the planned distance of 178 nm. The flight profiles were plotted and used to estimate the outbound and return flight fuel consumption. Figure 1 Figure 2 5 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 Survivability The plot and passengers were all secured n ther seats by restrant harnesses. The plot and front seat passenger had lap and dagonal, upper body restrants. The rear cabn passengers had lap restrants only. The arcraft had passed through the trees before strkng the ground n a level atttude wth vrtually no forward speed. All those on board suffered serous njures and were ncapactated, experenced dfferent levels of conscousness and were unable to ext the wreckage. The rear cabn door on the left sde had burst open durng the mpact. There was no fire. The accdent was wtnessed and reported by a member of the publc usng hs moble telephone. He was promptly on the scene and provded detaled nformaton to the polce control room operator. The call was logged at 80 hrs. The polce ntated ther Major Incdent procedure and the first police officer was on the scene at 1817 hrs. The Denham Airfield staff, who had also seen the accident, immediately deployed the Airfield Rescue and Fre Fghtng Servce. Followng some difficulty in locating the scene, they supported the police and paramedcs n renderng assstance to the njured. The county Fre and Ambulance Servces arrved and, followng stablsaton and treatment by paramedcs, the first casualty was extracted at 1858 hrs, departing for hosptal at 905 hrs. The last casualty was removed by ambulance from the scene at 95 hrs and all the casualtes were taken to hosptal. Training On 4 August 2006, the day before the accdent, the pilot completed his Licence Proficiency Check (LPC) and Instrument Meteorologcal Condtons (IMC) revaldaton test. The person conductng the LPC was an experenced nstructor/examner who had carred out the plot’s ntal converson on to the type and perodcal flight checks since he acquired the aircraft. The flight test comprised of simulated instrument flying, visual circuits and upper ar work wth both engnes operatng and sngle engne asymmetrc handlng. The plot demonstrated a satisfactory level of flying and passed the LPC. For the LPC, the arcraft departed Denham at 206 hrs and landed back there at 1340 hrs giving a total flight tme of  hour and 34 mnutes. The start and tax to and from the runway was estmated to take approxmately 10 minutes. Prior to the flight the aircraft was refuelled to the half full ndcaton on both fuel gauges gvng a total fuel of 465 lbs. No weght and balance calculatons were recorded but the examner recalled that followng the flight both fuel tank gauges indicated slightly more than one quarter full, whch would have been at least 9.3 uSG (6 lbs) per sde, or 232 lbs total. Fuel used during the training flight would have been 233 lbs, gvng a fuel consumpton rate of 48 lbs per hour ncludng start and tax. Following the flight check, both candidate and examner seated themselves n the rear of the cabn. The examner asked the plot to explan how he would carry out the engine fire drill and the fuel crossfeed drll. The examner stressed the need not to trust to memory for crossfeed procedures because, n hs experence fuel crossfeed labellng was frequently ambguous. The plot correctly covered the memory items of the fire drill but stated that he would consult the arcraft checklst for the fuel crossfeed operaton. Neither pilot nor examiner could find the crossfeed drll n the checklst and therefore consulted the fuel system descrpton n the arcraft’s POH. From the fuel system dagram and the system descrpton text, they concluded that to crossfeed fuel from the left 6 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 tank to the rght engne, two actons were requred. The rght engne rotary fuel selector should be turned to the CROSSFEED (yellow sector) poston and the crossfeed emergency shutoff control should be pulled out to open the crossfeed fuel lne. The use of the crossfeed emergency shutoff control s not clearly explaned n the fuel system descrpton. Immedately above the red coloured crossfeed emergency shutoff control, wrtten n whte letters on a black background, s the followng nstructon: ‘PULL-EMER FUEL X-FEED SHUT OFF’ In the ‘Emergency Procedures’ secton of the POH, the ‘Engne Fre n Flght’ and ‘Landng Gear Malfuncton’ procedures clearly state the purpose and operaton of the crossfeed emergency shutoff control. For example, n the ‘Engne Fre n Flght’ non-memory tems and n three of the landng gear abnormal procedures, the followng acton s requred: ‘Emergency Crossfeed Shutoff - - PULL TO CLOSE’ The examner and canddate read the text above the shutoff control but dd not lnk the ‘Fre Drll’ non-memory acton shown above. They had no reason to consult the landng gear malfuncton procedures when dscussng the crossfeed ssue and therefore placed an ncorrect nterpretaton on the nformaton contaned n the fuel system dagram. An addtonal lmtaton was relevant when usng the fuel system crossfeed controls. The crossfeed fuel lne pckup n the tank was above the lowest pont of the tank. In order to prevent the plot attemptng to crossfeed when the fuel level was lower than the pckup, a mnmum fuel level and phase of flight was imposed. This was stated n the fuel system descrpton as follows: ‘If single-tank operation is being used when fuel levels are low, the fuel quantity in the tank in use should not be allowed to drop below 60 pounds prior to re-establishing normal single-engine per tank operation; this will avoid the possibility of dual engine stoppage due to fuel starvation.’ A note was also included to emphasise the phase of flight when crossfeedng fuel should not be used: ‘The fuel selector valve handles must be turned to the NORMAL FLIGHT, L. TANK, T.O./LDG (green sector)position for the left engine and the NORMAL FLIGHT, R. TANK, T.O./LDG (green sector) position for the right engine for takeoff, landing and all normal operations. Crossfeeding is limited to level flight only.’ The nformaton avalable to the plot contaned n the arcraft’s POH regardng crossfeedng can be summarsed as: 1. Only crossfeed during level flight and not durng takeoff and landng. 2. Ensure that crossfeedng s stopped before the fuel quantty n the tank beng used drops below 60 pounds (0 uS gals). 3. The crossfeed emergency shutoff control s pulled to close the valves, not open them, and s not operated when crossfeedng. Weather An aftercast provided by the Met Office gave the synoptic stuaton at 200 hrs on the 5 August 2006. It showed a rdge of hgh pressure extendng across the Brtsh Isles from the south-west wth a weak warm front lyng north to south across the country. A lght north to north-west 7 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 wnd covered the route. By 800 hrs there was lttle change n the general condtons and the weather was good for the flight to and from Durham Tees Valley. There was a possblty of slght ran from a strato-cumulus cloud layer manly near the Teessde area but the weather was manly dry throughout the route. The vsblty was 20 to 30 km wth a Mean Sea Level pressure of 020 hPa. In the Denham area at 200 hrs, the cloud was manly shallow cumulus base 3,500 to 4,000 ft wth small amounts of strato-cumulus and crrus above. The strato-cumulus layer ncreased to full cover around the East Mdlands/Lncolnshre area, base 4,000 to 6,000 ft. For the return journey, extensve strato-cumulus covered the route from Teesside to the Cranfield area with the base around 3,500 to 5,000 ft. From Cranfield southwards t appears to have mproved, wth just small amounts of cumulus. The table below sets out the actual wnds for the alttudes gven whch were recorded from the Nottngham radosonde ascent for mdday on 5 August 2006. It s also a good gude to the wnds later n the afternoon for the return journey and throughout the route. (Table ) Heght AGL Wnd speed and drecton 2,000 ft 300º/05 kt 5,000 ft 330º/05-0 kt 0,000 ft 020º/20-25 kt Table 1 Fuel planning Artcle 52 (e), ‘Pre-flight action by commander of aircraft’ of the Ar Navgaton Order (ANO) places the followng requrement on the commander: ‘In the case of a flying machine or airship, that sufficient fuel, oil and engine coolant (if required) are carried for the intended flight, and a safe margin has been allowed for contingencies.’ The CAA produces Safety Sense Leaflets covering many aspects of aviation. Safety Sense Leaflet number 1e ‘Good Airmanship’ contans a secton on fuel plannng and offers the followng advce to prvate plots: ‘Fuel planning • Always plan to land by the time the tank(s) are down to the greater of ¼ tank or 45 minutes cruise flight, but don’t rely solely on gauge(s) which may be unreliable. Remember headwinds may be stronger than forecast and frequent use of carb heat will reduce range. • Understand the operation and limitations of the fuel system, gauges, pumps, mixture control, unusable fuel etc and remember to lean the mixture if it is permitted. • Don’t assume you can achieve the Handbook/ Manual fuel consumption. As a rule of thumb, due to service and wear, expect to use 20% more fuel than the ‘book’ figures.’ From hs evdence to the nvestgaton, the power settngs generally used by the plot of G-PTWB n the cruse were 23 nches Manfold Ar Pressure (MAP) and 2,300 propeller rpm on both engnes. From the performance secton of the POH, ths equates to approxmately 67% power or 43.5 lbs per hour cruse fuel consumpton (2.4 lbs per mnute). The POH states that a normal rate clmb at 5,50 lbs All up Weght (AuW) to 8,000 ft takes approxmately 0 mnutes and uses about 33 lbs of fuel (3.3 lbs per mnute). Descent 8 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 from 8,000 ft takes 0 mnutes and the fuel requred s gven as 2 lbs, gvng a consumpton of 2. lbs per mnute. Applyng these consumpton rates to the vertcal profile of the radar data indicated that the fuel used on the first sector was 220 lbs and on the return sector was 86 lbs. To ths must be added 25 lbs for the start, tax and takeoff at Denham and Durham Tees Valley, gvng an addtonal total of 50 lbs. Based on ths calculaton, the total fuel consumption for the ‘round trip’ flight was approxmately 456 lbs. From prevous experence the plot had derved a plannng figure of 100 litres per hour. This was based on 80 litres per hour (26 lbs) consumpton, wth an addtonal 20 ltres (32 lbs) for contngency or the equvalent of a total 58 lbs per hour. From hs experence ths provded adequate fuel for the flight he undertook with a reserve whch, f not requred, would stll be avalable on landng. If payload permtted he would also take addtonal fuel depending on the weather or nature of the flight being carred out. He had not prevously experenced any difficulties with a shortage of fuel. The plot used a plannng arspeed of 60 kt whch, gven the lght wnds at hs crusng level, he used as a groundspeed for calculatng the tme to cover the 78 nm track dstance from Denham to Teessde. Ths gave a flight time of 66 minutes at the 158 lbs per hour rate, requiring 174 lbs for the flight up and 174 lbs for the return flight. To this he added 25 lbs for each sector for start, tax and clmb and one hour reserve gvng a total fuel requred of 556 lbs. The POH contans comprehensve tables, graphs and examples coverng fuel consumpton for all phases of flight in order for a pilot to establish the fuel required for a specific flight. In the introduction to the Performance secton, the followng statement s made: ‘It should be noted that the performance information presented in the range and endurance profile charts allows 45 minutes reserve fuel at the specified cruise power. Fuel flow data for cruise is based on the recommended lean mixture setting. Some indeterminate variables such as mixture leaning technique, fuel metering characteristics, engine and propeller condition and air turbulence may account for variations of 10% or more in range and endurance. Therefore, it is important to utilise all available information to estimate the fuel required for the particular flight.’ In the performance section, specific ‘Fuel and Time Requred’ graphs were provded for 50%, 60% and 70% power. The graphs permt the plot to calculate the fuel required for a specific distance, wind conditions, alttude and power settng. Ths ncludes the fuel used for engne start, tax, takeoff, normal clmb, descent and 45 mnutes reserve. By enterng the 50% power graph wth a dstance of 78 nm and nl wnd, a fuel requred of 265 lbs s obtaned. By addng the 0% contngency from the note above, a fuel requred of 29.5 lbs s obtaned. The manufacturers were provded wth tme versus altitude data for the flights and asked to calculate the fuel used durng the round trp ncludng ground taxng. They concluded that, based on the Cruse Performance chart, with a flight time in radar contact of 2 hours and 27 mnutes the arcraft used 377 lbs of fuel. Addng 25 lbs of fuel for start, tax and clmb at Denham and Teessde gave a total consumpton of 427 lbs for the round trip flight. If the Fuel and Tme Requred chart was used and the 45 minute reserve of 104 lbs subtracted the figure ncreased to 455 lbs. The dfference was accounted 9 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 for by the Fuel and Tme Requred chart ncludng an allowance for start, tax, clmb and descent, whlst the cruse chart does not. The AAIB calculaton was based on the mnute/lb burn rates set out above wth 25 lb start, tax and takeoff allowance at Denham and Teessde, and produced a figure of 444 lbs based on the performance at a maximum AuW of 5,50 lbs. From the dfferent methods of calculatng the POH fuel consumpton, the ‘round trp’ fuel consumpton was estmated at between 427 lbs and 456 lbs. Weight and balance No wrtten record of the weght and balance calculatons carred out by the plot was avalable to the nvestgaton. The weghts of the plot and passengers are ther actual weghts at the tme of the accdent, subsequently provded to the nvestgaton. The calculaton set out below s based on the examner’s recollecton of the fuel remanng on board the arcraft followng the tranng flight, that is, approximately ¼ full. The addition of 70 ltres per sde on the mornng of the accdent has been added to that amount. (Table 2) usng the plot’s recollecton of the tanks beng between 26 and 30 uSG per sde before refuellng at Denham, for the lower figure an additional 80 lbs should be added to the total fuel weght. At 30 uSG per sde, an addtonal 28 lbs should be added to the 453 lbs shown n Table 2. The Maxmum permtted TakeOff Weght (MTOW) for the arcraft was 5,50 lbs. The arcraft CG envelope at 3,300 lb was from the forward lmt at 46.5 n to the aft lmt of 57.2 n aft of the CG datum. The forward lmt s constant to 3,800 lb and then reduces n a lnear fashon to 5.2 n at the MTOW of 5,50 lb. The aft CG lmt remans constant at 57.2 n up to the MTOW. Item Weight (lbs) Arm (in) Moment Arcraft basc weght 3,696 559,083 Plot 9 38 26,358 Front passenger 2 38 5,456 Mddle seat passengers (2) 476 78 84,728 Rear seat passengers (2) 353 26 76,248 Cargo 25 250 6,250 Fuel 453 73,000 Departure Denham 5,306 158.5 841,123 Flght fuel burn *220 Landing Teesside 5,086 158.37 805,489 Flght fuel burn *86 At impact 4,900 158.4 776,223 *AAIB calculated leg consumpton, no ncluson of 25 lbs for tax and takeoff. Table 2 20 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 From the weghts provded and the estmate of fuel carred and consumed, the arcraft was operated ntally 56 lb above the MTOW durng the departure from Denham. Ths would ncrease to 284 lbs f the hgher fuel quantty was carred. The CG was calculated ntally at 58.5 n aft of the CG datum reducng to 58.4 n aft of the datum as fuel was consumed. Ths was beyond the aft CG lmt for the aircraft throughout the flight. When loadng the arcraft, the plot had placed the heaver passengers and baggage at the rear. By re-seatng the heaver passengers at the front and lghter passengers at the rear, as well as placng the baggage n the forward baggage hold, the CG could have been brought forward of the aft lmt. The arcraft could also have been operated wthn the MTOW of 5,50 lbs, f fuel for the outbound flight only had been carried, as set out below, although t would have been necessary to refuel for the return flight. (Table 3) Medical After the accdent, the plot was admtted to hosptal and a sample of hs blood was taken for hosptal purposes. Durng the course of the day, the plot had been seen to consume alcoholc beverage and analyss of the blood by the hosptal ndcated the presence of alcohol. The amount detected was not consdered to be a major contrbutory factor n the accdent but the exact effect on the plot’s performance could not be establshed. Performance The arcraft was observed by ground wtnesses n a left turn wth an angle of bank of 30º to 40º before rollng through the wngs level atttude to approxmately 30º rght bank. At ths pont the rght wng dropped. The stall speeds with 10º of flap set with an aft CG and the angle of bank flown are reproduced below, showing both indicated and calbrated arspeeds (kIAS and kCAS). (Table 4) Item Weight (lbs) Arm (in) Moment Basc weght 3,696 559,083 Plot 9 38 26,358 Front passenger 245 38 33,80 Mddle passengers 48 78 74,404 Rear passengers 280 26 60,480 Forward baggage bay 25 82 2,050 Fuel 295* 48,800 Weight and CG at Takeoff 5,150 156.31 804,985 Weight and CG at Landing 4,950 155.72 * The fuel required of 295 lbs would have been sufficient to operate the aircraft on the sector to Durham Tees Valley wth 45 mnutes reserves and 0% contngency, usng 50% power settngs. Table 3 2 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 A note states that: ‘Altitude loss during an engine inoperative stall recovery may be 300 feet with a pitch below the horizon of 30°.’ As an ndcaton of the arcraft’s performance wth one engne noperatve at 4,800 lbs at sea level, the rate of clmb at 97 kt (Vy ) wth the faled engne propeller feathered, landing gear and flap retracted and maximum power set on the operatng engne s 270 ft per mn. The followng decrements must be subtracted from that rate of clmb to calculate the arcraft clmb/descent performance. (Table 5) Configuration Decrement Landng gear extended -350 ft/mn Flaps extended 0º -50 ft/mn Flaps extended fully -450 ft/mn Inoperatve propeller wndmllng -250 ft/mn Table 5 With landing gear lowered, flap set to 10° and the right propeller wndmllng, a net rate of descent of 380 ft/mn would result. If power was not avalable from the left engne, the drag from both propellers wndmllng and the aircraft configuration would have resulted in a rapid loss of arspeed had a postve nose-down atttude not been adopted. Significant Aircraft Features The arcraft type s equpped wth two ntegral fuel tanks. These are postoned n the outer wngs and are formed by the upper and lower skns and the front and rear wng spars. They are bounded at ther nboard ends by closure rbs, approxmately co-ncdent wth the outboard sdes of the engne nacelles, and extend outboard from there to statons close to the wng tps. The fillers are at the outboard ends of the tanks and since the wing has significant dihedral, the tanks can contain a large proporton of ther capacty before any fuel can be seen via the filler orifices. The fuel pck-up ponts are postoned at the forward and aft ends of manfolds sted at the extreme nboard ends of each tank. Each pck up pont s postoned close to the plane of the lower wng skn and s closed by an individual float valve. Thus, when fuel is present at the pck-up pont, the valve admts t to the manfold, but when t s absent, closure of the valve prevents ar from flowing into the manifold. The POH states that each tank has a maxmum capacty of 77.5 uSG, whlst the total unusable fuel s quoted as 2.0 uSG. The fuel system supples the engnes va fuel selector valves postoned n the wngs, just nboard of the tanks. These are controlled, va sldng cables wthn conduts, by means of handles mounted on a console between the two front seats, just above the cabin floor. The relevant tank contents gauges are to be found above Angle of Bank 0° 0° 30° 30° 45° 45° 60° 60° Weght (lb) kIAS kCAS kIAS kCAS kIAS kCAS kIAS kCAS 550 57 62 6 67 68 74 8 88 4650 53 59 57 63 63 70 75 83 Table 4 22 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 the fuel selector handles. Each of these s annotated wth whte markngs on a black background at 0, 30, 50, and 70 uSG levels. These numercal ndcatons are postoned below a horzontal whte lne. Above the lne, fuel quanttes are annotatons n lbs. Those graduatons ndcated are at the 00, 200, 300 and 400 lbs levels. The secton whch appears to fall between the empty and 0 uSG graduatons on each gauge s coloured yellow and whte, whlst a narrow red lne graduaton s postoned approxmately at the empty poston. Each selector can be turned to the ‘ OFF ’, ‘ ON ’ or ‘ CROSSFEED ’ poston. Wth the selector set to the ‘ ON ’ poston, the relevant engne s suppled by the tank on the same sde of the arcraft. When the ‘ CROSSFEED ’ settng s selected, the engne on the same sde as that selector receves fuel from the tank on the other sde of the arcraft, va crossfeed ppes whch pass beneath the cabn. To prevent leakage of fuel, should one or both crossfeed ppes become damaged, crossfeed shutoff valves are provided. These are fitted close to the tanks. They ensure that only the fuel volume wthn the ppes, and no fuel from ether tank, can be lost through any crossfeed ppe leakage once the shutoff valves are moved to the ‘ OFF ’ poston. Both shutoff valves are operated va cables wthn conduts from a snge T- handle below the fuel selector valve console. If the handle s pulled when both fuel selectors are n the ‘ ON ’ poston, engne operaton s not affected. If, however, t s pulled when a fuel selector s set to ‘ CROSSFEED ’, the supply to the engne on the sde of the selector wth that settng wll be nterrupted and the engne wll not contnue to operate. The cross-feed shut off valve control handle s panted red, sgnfyng ts emergency control status. The crossfeed ppes have open pck-up ponts postoned on the inboard closure ribs of the fuel tanks, significantly above the plane of the lower wng skns. Ar can thus be drawn nto the crossfeed system and thereby nterrupt the fuel supply to the engne selected to crossfeed, f the fuel n the tank n queston s below the level of the orifice of the crossfeed pick‑up. The fuel dvder unts on the engnes each ncorporate a sprng-loaded valve. Ths shuts off the fuel supply to the njectors postvely when the fuel pressure to the relevant dvder drops below a threshold. Loss of fuel supply to an engne fuel/ar control unt thus results n closure of the valve and engne stoppage. A volume of fuel, however, remans n the engne fuel system upstream of the flow divider, following such engine fuel starvation. Figure 3 Fuel system controls and gauges Fuel gauge Fuel gauge 23 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 Figure 4 Fuel system dagram 24 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 The accident site The arcraft came to rest n a wngs-level atttude on an upward slopng surface n dense woodland, at a pont having no significant ground vegetation. The slope formed the upper part of a ralway embankment. Examnaton of the damaged trees revealed that the arcraft had struck and demolished one, but had inflicted little damage to adjacent trees. It had come to rest whle movng laterally to the rght, as ndcated by the vertcal trunk of a small tree whch had penetrated the wng tp and travelled nboard for approxmately ½ metre. A substantal branch had passed vertcally between the elevator and the horzontal stablzer. Both engne nacelles were deformed nto a pronounced ‘hoggng’ (e down at the extremtes) shape. Extensive damage had been inflicted to the nose of the aircraft forward of the windscreen although no significant longtudnal compresson damage was evdent. The fuselage was reduced n depth and the tal unt, complete wth fuselage talcone, was separated from the aft end of the fuselage. The seatback of one of the rear row of forward facng seats had collapsed backwards. Both propellers were n the normal operatng range and the lower two blades of both were embedded n soft sol. Nether propeller exhbted any evdence of rotaton at mpact. A number of tree boughs were found to have been chopped n an orentaton approxmately perpendcular to the branch axes. It was known, however, that sawng equpment had been used to cut away tmber to gan access to the forward end of the cabn durng rescue operations. This created significant quantities of cut tmber of smlar appearance to tree boughs havng suffered blade strkes from fast rotatng propellers. The arcraft had the landng gear extended and one stage of flap (10º) was set. On enterng the arcraft cabn t could be seen that the rght fuel selector was n the crossfeed poston whlst the left selector was n the normal tank to engne poston. The crossfeed shutoff control was n the shutoff poston. After ntal examnaton, the arcraft was dragged approxmately four metres forward on to level ground, usng strops attached to the man landng gears, n order to ensure there was no danger of t sldng down the embankment and descendng through trees on to the adjacent ralway track. It was subsequently noted that the nteror of the rght tank at ts nboard end could be seen through a hole n the upper wng surface. No fuel was present. The lower surfaces of the tank appeared to be undamaged so t was postulated that the tank may have been empty at mpact. When a quantty of water was poured into the tank filler, however, a rupture was identified where the lower edge of the rear spar had deformed close to the nboard end of the tank. A hole delberately created n the top skn of the left tank revealed that t was also empty and ntroducton of water revealed a correspondingly positioned rupture to that identified in the rght tank. Samples of the water ntroduced nto the tanks were then recovered n a transparent beaker and examned. Only a scarcely detectable layer of hydrocarbon appeared to be present on the surface of the water from each tank. It was reported that rescue of the occupants ntally requred access to both sdes of the arcraft from the rear, nvolvng rescuers passng behnd the ponts of tank rupture. Wth the arcraft on a steep slope ths requred personnel to pass below the ponts from whch any fuel present would have draned mmedately after the mpact. None of the personnel on the scene mmedately after the mpact reported seeng or smellng any fuel or notcng any dampness of the otherwse very dry sol. The absence of surface vegetaton precluded the examnaton for dscolouraton whch often reveals the presence of Avgas resdue. 25 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 Detailed examination The arcraft was cut nto a number of sectons for removal from the woods before beng transported to the AAIB headquarters where a detaled examnaton was carred out. Pror to separaton of structural elements, all piping requiring cutting was crushed flat using special equpment, thus sealng the ends aganst loss of fuel or ngress of other substances. The crushed areas were then cut at md length, preservng, as far as possble, the sealng effect of the crushng on both sdes of the cut. Durng subsequent examnaton, the settngs of the two fuel selector valves and the crossfeed shutoff valves were establshed by determnng the presence or absence of flow resulting from application of air pressure to various fuel lnes followng cuttng away of the crushed sectons. It was thereby establshed that all four valves were set to the same poston as ther cockpt selectors ndcated. Both crossfeed ppes were found to contan fuel. The powerplants were removed from the firewalls and examned n the presence of the AAIB and a specalst provded by the engne manufacturer. All the engne fuel system components were rg tested n accordance with their manufacturer’s specifications. All were found to contan varyng amounts of fuel and to functon correctly, wth the excepton of one varable fuel valve mounted co‑axially with its throttle butterfly. This valve exhbted a small volume leak. Examnaton of the local area revealed no evdence of dscolouraton from pre-accdent leakage n ths area, however. Examnaton of the seals on the shaft n the regon of the leakage dd not reveal any excessve wear, deteroraton or damage. The possblty that slght bendng of the shaft had occurred durng the mpact resultng n reduced performance of the seal could not be ruled out. Strp examnaton of both engne carcases revealed no evdence of pre-crash falure. Discussion From the evdence at the accdent ste, t could be deduced that the arcraft had struck the top of a tree n an approxmately erect atttude wth very lttle forward speed but significant vertical speed. The restrcton of major damage to one tree n a wood of closely spaced trees all of smlar heght was a partcularly postve ndcaton of ths. It was further concluded that the presence of the tree had reduced the final descent rate. The ground impact force on the man landng gear appeared, however, to have been sufficient to produce the deformation of the nacelles and contributed to the flattening of the cabin. Some backward moton durng the mpact sequence was evdent from the backward collapse of one of the seatbacks. Damage inflicted to the right wing tip and to the elevator / horizontal stablser juncton was ndcatve of, respectvely, lateral and vertcal moton through the trees, whlst absence of wing leading edge damage confirmed an absence of significant forward motion. The mpact wth, and subsequent destructon of, the one tree had left no postve evdence as to the ptch and roll atttude at ntal contact. The lack of leadng edge mpact damage and the falure of the arcraft to mpact nose-down between trees tended to confirm the view that tree‑top mpact occurred n an atttude not grossly dfferent from that of normal flight. It indicated significant downward rather than forward moton. The propellers exhbted no evdence of rotaton, although the soft ground and lack of forward speed consttute condtons whch frequently leave no evdence even when significant power is known to have been produced at mpact. There are two possble reasons for the absence of fuel vsble through holes n the upper skns at the nboard ends of both tanks: 26 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 () The tanks were empty at the tme of the accdent, or (2) At the tme of the mpact the remanng tank contents all draned through the ruptured rear spar jonts at the nboard ends of the tanks. The latter event s a possblty snce the arcraft ntally came to rest on a slope n a nose-up atttude causng the ruptures to be postoned close to the lowest ponts n the tanks. The wreckage was only subsequently dragged to a level surface where much of the examnaton took place. It s surprsng, however, that a small resdue of fuel from the extreme low pont of the tanks dd not reman when the tanks were examned. Although some fuel was found n components of the engne-mounted fuel systems and ppe-work, one component was damaged by the mpact and had allowed some leakage to take place. It was thus not possble to compare usefully quanttes of fuel n the two engine systems. It should be noted that the flow dvder unt ncorporates a sprng-loaded shutoff valve so that when ar enters the engne system leadng to a loss of delvery pressure, the valve wll shut off. Ths causes power loss even though a significant volume of fuel remans n the components and ppe-work. The presence of fuel n these areas, therefore, does not necessarly ndcate that fuel was stll beng suppled from ether tank at the tme of the mpact. The use of a cable and condut system for controllng the fuel tank selector and crossfeed shutoff valves makes t unlkely that ether the valves or ther controls moved from ther mmedate pre-mpact postons. Ths s despte the consderable mpact dstorton of the fuselage relatve to the wng structure. The settngs of the valves, as determned from tests using air pressure, confirm that the left fuel valve was in the normal poston, the rght valve was n the crossfeed poston and both crossfeed shutoff valves were n the closed poston. These all corresponded wth ther cockpt selectons as found durng the ste examnaton and ths s presumed to have been the stuaton at the tme of ground mpact. The sgns of a lack of forward moton through the trees, the relatvely ntact, although severely damaged state of the arcraft and the survval of the occupants ndcate relatvely low energy at the tme the arcraft struck the trees. These factors are consstent wth both a low forward speed and low heght at the tme control was lost. Although the engne power at mpact could not be determned, t appears that the mpact was consstent wth a stall rather than the consequences an asymmetrc power nduced control loss durng the approach. No falure or defect wthn the arcraft or ts propulson system was identified. Fuel starvaton was probably the man causal factor of the accdent, although fuel exhauston could not be ruled out. The lack of a record of the arcraft fuel state pror to the departure from Denham or Durham Tees Valley meant accurate departure fuel quanttes could not be establshed. There were two dfferent recollectons of the fuel quantty remanng onboard the arcraft after the tranng carred out on the day before the accdent. The examiner recalled slightly more than ¼ full or 19.5 USG per sde and the plot thought there was 26-30 uSG per sde. Wth the addton of 40 ltres of fuel pror to departure from Denham the quantty onboard was between 453 lbs by the examner’s recollecton, and up to 58 lbs from the plot’s. No precse quantty of fuel consumed on the ‘round trp’ flight could be established but using fuel consumption 27 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 data from the POH, between 427 lbs and 455 lbs was consdered a reasonable estmate. The two gauges, one alongsde the other, whch were observed by the front rght seat passenger, were probably the fuel gauges. The needle on one gauge was just above the red and the other was n the red. Whlst he could not remember exactly at what pont he saw them, from hs descrpton t was just pror to the approach to Denham. The red lne ndcates the tank s almost empty and therefore suggests that the fuel n the tank wth the needle n the red was about to run out. The other tank contaned a small amount of fuel. The plot also recalled seeng an mbalance but could not recall the ndcatons. These ndcatons were consstent wth the plot reducng power on the engne on the sde wth the tank wth the lowest fuel contents, and attemptng to crossfeed from the other tank, whch had slghtly more fuel remanng. From the poston of the crossfeed selector and valve, the rght tank was the tank whch contaned least fuel, and was nearly empty. Openng the crossfeed, however, would not draw fuel from the left tank as the level was below its crossfeed fuel pick‑up. If sufficient fuel had been avalable to crossfeed, the effect of pullng the crossfeed emergency shutoff would have been to prevent the rght engne drawng fuel from the left tank. However, ths was not relevant at such a low fuel state snce crossfeedng was not possble. The rght engne therefore ran down and wth the propeller not feathered the arcraft would have yawed to the rght. The nformaton contaned n the POH (Informaton Manual) and the crossfeed labellng was not clearly understood by ether the plot or the LPC examner, and so the followng Safety Recommendaton s made: Safety Recommendation 2007-086 The Federal Avaton Admnstraton should revew the Cessna T303 Crusader Informaton Manual and Checklsts to ensure that clear and unambguous nformaton s provded for the operaton of the fuel crossfeed system. If the left tank fuel was also exhausted, then the left engne would also have run down. If, however, useable fuel remaned n the left tank, t s possble that n the 30º- 40º left bank, wth the arcraft yawng to the rght, the fuel mgrated towards the left wng tp and uncovered or partally uncovered the normal fuel pck-up. Agan, the left engne would suffer a reducton n power, or stop. Wth the left engne not respondng and the rght engne propeller not feathered, arspeed would have decayed rapdly from the 90 kt approach speed. The stall speed s gven as 60 to 65 kCAS dependng on the angle of bank. If the nose was not lowered postvely the arcraft would stall and possbly drop a wng. Ths was the behavour descrbed by the wtnesses on the ground. There was no evdence of fuel on the ground, and none was reported escaping by those first on the scene. Although a small spllage mght not have been obvous, larger amounts should have been evdent from smell and vsble leaks. From ths, t s probable that the fuel on board on departure from Denham was closer to the lower estmate of 453 lbs than the hgher estmate of 58 lbs. The plot had not carred out a full weght and balance calculaton to determne the AuW and balance of the arcraft. Had he done so the lmted amount of fuel that could be carred and the CG poston outsde the permtted envelope should have been apparent. Wth the weght of the arcraft, the plot, passengers and 28 © Crown copyrght 2007 AAIB Bulletin: 10/2007 G-PTWB EW/C2006/08/01 baggage, only 297 lb of fuel could be carred n order to reman below the 5,50 lbs MTOW. By re-arrangng the passenger seatng and baggage, the CG could have been moved forward to wthn the permtted envelope. Wth only 297 lbs of fuel avalable, the arcraft could have operated the Denham/Durham Tees Valley sector wth 45 mnutes reserves and 0% contngency at 50% power. Ths would not have met the Safety Sense leaflet recommendation of 20%. Refuelling at Durham Tees Valley would have been necessary for the return flight. Conclusions The pilot was properly licensed and qualified to conduct the flight. The aircraft was fully serviceable and the weather was suitable for the flight and was not a factor n the accdent. From the evdence provded, the loadng of the arcraft was such that t was operated ntally above the MTOW of 5,150 lbs and throughout the flight the aircraft was operated outsde the aft CG lmt of 57.2 nches aft of datum. Wth the payload beng carred, the arcraft was not capable of safely completing the ‘round trip’ flight and remanng wthn the permtted weght and balance envelope wthout refuellng at Durham Tees Valley. Insufficient fuel was carried for adequate reserves and contingency fuel to complete the flight. The plot had consumed alcoholc beverage durng the day but the effect on hs decson makng and arcraft handlng ablty s not known. Durng the approach, the fuel crossfeed was used, whch was not permtted. The selecton of crossfeed from the left tank to the rght engne was probably the cause of the rght engne runnng down. Ths was due to insufficient fuel contents being available to allow fuel to be drawn from the left tank by the crossfeed pck-up. Pullng the crossfeed emergency shutoff control therefore dd not contrbute to the accdent. The accdent was caused by fuel starvaton of both engnes wth the rght engne ceasng to produce power and the left engne operatng at reduced power or stoppng. Control was then lost when the arspeed decayed and the arcraft stalled, droppng the rght wng.