The Aerodynamics of the V-bombers - Part 3

28 September 2022

image-13

Dr Steve Liddle CEng FRAeS, is a Vulcan to the Sky Trustee and Principal Aerodynamicist at Aston Martin Formula One Team. The articles here are republished from Steve’s occasional series on ‘the aerodynamics of the V-bombers’, that he writes on the Vulcan to the Sky Trust LinkedIn page.

Rocket Javelin

When talking about the development of the V-bombers, it is well worth looking at the wider ecosystem of contemporary aerospace engineering and particularly other efforts to build viable transonic combat aircraft. We started this series by comparing the integrated airframe and propulsion layout of the Vulcan with that of the Hawker-Siddeley group’s other transonic delta combat aircraft, the Gloster Javelin all-weather fighter. It’s fair to say that, at this separation in time from the events, the reputation of the latter does not compete well with that of the former. The first article postulated an aerodynamic reason for the relative lack of success of the Javelin, but in reality it was just one of many possibilities, related as much to delays in the programme as to the engineering itself. 

The wonderful Flight magazine archive contains a telling snippet from the issue of 8th July, 1955, revealing some of the pressure that Hawker-Siddeley (HS) were under at the time to make the RAF’s new fighter – the UK’s defence against nuclear attack in anything other than the daylight and good weather conditions that the majority of the fighter force was suitable for – work effectively. Under the title ‘Javelin Rumpus’, HS produced an eight point rebuttal to allegations in a Sunday Express article, concluding with, “There is no similarity between the state of the Javelin and the [Vickers-Supermarine] Swift story, and any attempt to create it is misleading and irresponsible.”

However, let us not be churlish about this. In the late nineteen-forties, men and women living on rationed food left their homes in bomb damaged streets, travelling to war-weary factories on equally tired bicycles or steam-hauled trains. When they got there, they engaged intellect and slide rules to produce world-beating technology. They knew it had to be done; they knew what a war felt like and there was a new enemy. Even if peace prevailed, their economy was on its knees and exporting these incredible products was a way of balancing the books. Not all of these aircraft worked as intended, but such is life at the cutting edge.

The aerodynamic engineering effort discussed so far has focussed on the requirement for high speed wind tunnels to adequately characterise and allow development of these pioneering transonic aircraft. Wind tunnels allow a controlled environment to be created, in which repeatable tests can be performed. The models used can be made sufficiently stiff and strong that the engineers can be confident that they are measuring aerodynamic loads from a myriad of internal sensors, rather than the effect of deflections of a flexible structure. They were expensive, complicated and of themselves approaching the limit of known technology.

1:72 scale high speed WT model of Gloster Javelin Mk.1 (Sutton and Stanbrooke, 1965)

Wind tunnels were certainly not the only way to gather data, however. One could choose alternatively to strap a large instrumented model on to a big rocket and stand well back. This is exactly what happened during development of the Javelin, as related in another remarkable report now available freely online (Greenwood, 1964). And you thought aerodynamicists have no sense of humour…

The Javelin was an advanced design at conception and used some of the aerodynamic technology previously discussed. The basic planform of its delta wing was not dissimilar to that of the baseline Vulcan design and it used the RAE 101 transonic aerofoil, a more modern approach than the relatively conservative modified NACA aerofoil of the Vulcan’s wing. We know though that the bomber benefited hugely from the section and planform modifications proposed by the RAE engineers, which the swept root and tip of the maximum thickness line and the extended, dropped leading edge. The Javelin wing was not modified from its vanilla design to anything like the same extent. This technology level seems consistent with published performance data and anecdotal statements from those associated with operating it: the Javelin would have struggled to catch a high-speed, high-altitude contemporary bomber like the Vulcan. This led to a development project that became known as the ‘Thin Wing Javelin’

Thin wing Javelin contemporary model, adapted from Buttler (2018). Although not explicitly stated, this appears to be a wind tunnel model and hence clearly an engineering proposal rather than simply a ‘brochure’ configuration.

Aviation historian Tony Buttler briefly described the Thin Wing Javelin and its fate in a paper for the online, free to access Journal of Aeronautical History published by the Royal Aeronautical Society. Warning: once you discover this resource, you will be lost for some time! He relates [Buttler, 2018] that a modified, thinner wing was the major change considered, with the rest of the aircraft remaining similar originally. Unsurprisingly, the development became more ambitious and ultimately little of the original was left, however performance was still likely to have been, “just supersonic in level flight at altitude.” Buttler states that the Minister of Supply Reginald Maudling (responsible for the provision of military aircraft) became aware that HS Group was building a much more advanced and potent fighter aircraft, the Avro Canada CF-105 Arrow, in December 1955. The Thin Wing Javelin was obviously out of date and was therefore cancelled in June 1956; Hansard states an expenditure of £2.3M on the project (HC Deb 17 June 1963). Was that a lot? Certainly not compared to the £22M which the same source records as having been spent on the Supermarine Swift fighter, for much the same outcome.

Greenwood’s report was published in 1964, but this is a collected version of an internal report dated November 1958, with the work obviously taking place sometime prior to this. The free flight model tests were divided in two strands. The first used a baseline Javelin fuselage and nacelle centre section and tested four sets of wings. The figure extracted from the report and reproduced below illustrates these.

Illustration of the four wing planforms tested, showing section and maximum thickness distributions (Greenwood, 1964)

Wing 1 is the baseline Javelin wing, referred to in the report as the ‘Mk.1 Javelin wing’, although confusingly not actually the wing that would be fitted to the Javelin FAW.1 when it entered service. This was of constant RAE 101 10% thick section. Wing 2 was a baseline, constant section ‘thin wing’, similar but with a thickness of 7% and intended to allow postponement of drag rise to an increased Mach number. Wings 3 and 4 are much more interesting but illustrate the scope of the challenge. They were designed to exploit the structural advantages of tapering thickness, which should have meant that they were lighter for a given stiffness.

Both were designed aerodynamically to what was referred to as ‘Haines’ concepts’, which we have met before and were implemented in the Vulcan wing prior to the prototypes being built. A.B. (Barry) Haines demonstrated that by tailoring the chordwise position of the maximum thickness of the wing sections across the span, then the aerodynamic sweep of the isobars could be displaced from the geometric sweep of the wing itself. This was important at the root and tip in particular, where the isobars would otherwise unsweep themselves as they passed through the fuselage or into the freestream flow respectively. In this case, the geometry lofted through appropriate sections, including the baseline RAE 101 with maximum thickness at 31% chord, to the RAE 104 which placed it at 41% chord. Towards the root, a section with the max thickness moved far forward to his 12% chord completed the effect.

However, the thickness variation was not only of position, but of magnitude. Wing 4 was designed to be optimised aerodynamically while exploiting the structural benefit of tapering the thickness outboard. Wing 3 was based on this optimum design, but made more pragmatic allowance for actual internal stowage requirements. As such, while root thickness was 10-11%, both tapered down to 5% at the tip. 

Qualitatively, the strategy that would be embodied on the Vulcan B2 wing is evident here. Both used relatively benign symmetrical inboard sections of moderately high thickness for a transonic aircraft, at 10%. Towards the tip, both blend to the RAE 104 aerofoil, with its rearwards displaced maximum thickness serving to further sweep the isobars. On the Vulcan, this was further refined by the curved tip leading edge, providing locally increased sweep but still retaining a finite chord and hence sufficient taper ratio to avoid tip stall. The inboard section had the thickness moved forward in the spanwise region of the intakes, which is clearly visible. The design of these wings was of course roughly contemporary. It is also interesting to note that this tapered thickness approach was part of the basis of the crescent wing used on the Victor, although the implementation was of course very different.

Cdo (Zero lift drag coefficient) v Mach number for the four wings investigated in free flight tests (Greenwood, 1964)

What is remarkable and illustrated in the figure above, is just how much better all three of these developments were in terms of delaying drag rise. Cdo on the y-axis is the zero lift drag, i.e. not related to the tip vortices and hence independent of load. This is roughly constant with speed until wave drag starts rear its head at high subsonic Mach number, the problem that low thickness was specifically aiming to address. The largely impractical constant 7% thick wing can essentially be discounted, but at least wing 3 would have been an implementable option for a viable aircraft. It can be seen that not only is the low speed drag much lower, but that the onset of drag rise is postponed from about M=0.88 to M=0.96.

An outcome of the test results related in the report was the statement, ‘For all practical purposes the mean thickness (other things being equal) is an adequate guide to the transonic performance of the particular configurations tested.’ This is of course of great relevance to an assessment of the high subsonic Mach number performance of the Vulcan, which especially in B2 form used similar strategies of chordwise and spanwise thickness distribution. 

While such a gain would have been immensely valuable in pushing out the performance boundary of a firmly subsonic bomber like the Vulcan, it is also clear that the advantage to be conferred on a fighter aircraft which aimed to be supersonic was not significant enough to justify such vast engineering effort.

There is a second reason why these geometries may not have been sufficiently advantageous. By the time the Javelin entered service, problems with tip stall identified very early in the flight test programme had led to the necessary adoption of a new outboard wing for stability and control reasons, rather than pure performance. From 57.5% semispan, the wing leading edge sweep was reduced to 41.5 degrees, resulting in an increase in tip chord. At the tip, the thickness was reduced to 7%, the leading edge radius significantly sharpened and the maximum thickness moved back to 51% chord. This region was then defined by smoothly lofting between the basic inboard section. Consequently, the production Javelin has taken a very modest step towards becoming ‘thin wing’ and sweeping back the tip isobars; its drag divergence Mach number was presumably already better than the baseline case considered for the free flight tests. This planform is illustrated below in a figure extracted from another contemporary report on Javelin wind tunnel tests (Kirby and Spence, 1958)

Javelin wing planform for early production aircraft (representing Javelin FAW.1) and showing reduced sweep outboard and long chord tip adopted from third prototype onwards (Kirby and Spence, 1958)

While the tests above were conducted using wing models attached to a bespoke fuselage containing the rocket motor, a second strand of tests was undertaken in which a complete aircraft model was mounted on a separate booster, as shown in the article header. This looked at adopting a form of the area rule, which was being found (at roughly the time of the initial tests) to reduce transonic drag. This required the rate of change of cross-sectional area of the complete aircraft with respect to the direction of flight to be minimised; in other words there should be no abrupt discontinuities. The tests all used the favoured Wing 3, i.e. the best overall practical solution.

The image below shows one of the development options (Model C), with an undernose fairing and – most notably – highly swept intakes replacing the baseline Javelin’s straight pitot inlets. Clearly, the latter change removes a major discontinuity that may have been a significant player in this aspect of the actual aircraft’s performance. The lower image shows a detail view of the rear fuselage of Model D, which featured large fairings aiming to compensate for the sudden end of the wing’s contribution to cross sectional area. It should be (and was, in the report) noted that a blended delta is inherently suited to the area rule, but that the wing trailing edge is one of the regions where action such as this would be required.

Complete aircraft Models C (top) and D (bottom, rear fuselage only) used in free flight tests of area ruled designs (Greenwood, 1964)

Ultimately, the results of the tests showed little effect subsonically, although all options showed a considerably better trend supersonically. The zero lift drag coefficient is again plotted against Mach number in the extracted figure below. Two results of particular significance from our perspective of the V-bomber designs are the variations in intake sweep. Both of these show considerably better performance in terms of drag at both subsonic and supersonic Mach number. Although the Vulcan design was far too early to consider the area rule, in effect Avro lucked in with their application of the swept, slotted inlet to sweep the inboard isobars as per Haines’s theory, whereas Gloster’s use of the pitot inlets had the opposite effect. Once again, the perils of being the pioneer!

A further point drawn by the report was that the subsonic drag increase (not related to wave drag) for the Model D rear fuselage modifications was considerable, despite their positive effect supersonically. This was surmised to be due to the increase in base area from the additional volume; a properly designed solution could have addressed this.

This brief look at the Thin Wing Javelin through the prism of freely available contemporary technical data has highlighted that the changes required to produce any worthwhile performance delta would have amounted to a new aircraft. After initially testing a completely new wing, it became obvious that any improvement in transonic performance would require application of the area rule, and that in practice would have meant a new fuselage, as the effort with Model D illustrated. Given that the manufacturer would need to start again and that the basic aircraft itself was proving difficult, the decision not to proceed with the Thin Wing Javelin can be seen to be entirely logical. In spite of any legend that may have grown to show this as one of the great lost opportunities of the Britain’s 1950’s aerospace programme, we can use real technical data to show that was unlikely to have been the case.

On the other hand, it has given us some more background on just how advanced the eventual application of some of these strategies during Vulcan development was. The Avro bomber clearly benefitted from the RAE’s theories on thickness distribution with span, may have lucked in with transonic wave drag delay due its swept intakes, while undoubtedly was more efficient due to its thin, swept and curved tip outboard wing in B2 form. On the Javelin, there was also no choice but to add any rear fuselage volume increase for area rule purposes outside of the engine nozzles, for obvious reasons. Had a rear fuselage fairing been advantageous in terms of transonic drag for the Vulcan, then the rear fuselage could be extended as a streamlined body with the baseline area included, minimising any non-wave drag losses. As it happened, Avro did need somewhere to put an enhanced electronic warfare system for the Vulcan. Did the new, much larger tail cone serve two purposes, giving wave drag benefit for free? I don’t know, but perhaps you do.

If it had been easy, everyone would have done it.

References

Buttler, Tony; The 1957 Defence White Paper, Cancelled Projects; The Journal of Aeronautical History, Royal Aeronautical Society, paper 2018/03, 2018. https://www.aerosociety.com/media/8134/the-1957-defence-white-paper-cancelled-projects.pdf

G.H. Greenwood; Free-flight model drag measurements on a transonic fighter (gloster javelin); ARC/CP-678, 1964. ( http://naca.central.cranfield.ac.uk/reports/arc/cp/0678.pdf )

HC Deb (17 June 1963) vol. 679, col. 7. Available at: http://bit.ly/2QZPyuK (Accessed: 28 November 2019).

Kirby, D.A. and Spence, A; Low-speed-tunnel model tests the flow structure behind a delta wing aircraft and 40 deg swept wing aircraft at high incidences; ARC/R&M 3078, MoS April 1955 ( http://naca.central.cranfield.ac.uk/reports/arc/rm/3078.pdf)

Sutton, E.P and Stanbrook, A; A Wind-Tunnel investigation of the directional and longitudinal stability of the Javelin aircraft at transonic speeds, including comparison with flight test results; ARC R&M 3403, December 1959 (http://naca.central.cranfield.ac.uk/reports/arc/rm/3403.pdf)