Thursday, May 4, 2017

Technical Appendix: Apollonian Days

I was going to go with "Apollonian Days of Future Past," but too wordy. I'm still going to keep the image of Old Wolverine getting barbecued, even though I'd need to write an essay in this block to explain why. Source
I'm referring to the Armstrong Whitworth Apollo. Two weeks ago, I thought, "Well, before I make a joke about the way that its original name (the Avon) poaches the Rolls Royce "River" theme, I should find out when the Avon [1947/8, as it happens] appeared --and, for that matter, when the themed naming schemes of postwar British engines were finalised." It turns out that naming a plane "the Avon" was perfectly fine in 1947; and, in any event, the plane's name was changed to the "Apollo" well before Armstrong Whitworth slunk away in shame from the "turboprop feeder airliner" market in 1951, leaving this mess on Farnborough airfield for someone else to clean up.

By RuthAS - Own work, CC BY 3.0,
Before I disappear up my own ass, here's my argument and conclusions:

Given that the United Kingdom had bombed out the only viable competition [which is a story about the American engineering industry that hasn't been well told], there was room for its aviation industry to take advantage of first mover advantage and get an effective monopoly on each of the three main types of commercial gas-turbine engines: the turbojet, turboprop, and turbofan. When I started bouncing around after the full story of the Armstrong Whitworth Apollo, I thought I knew the results of this fleeting advantage. It was thrown away in the case of the turbojet,  seized with impressive sales results in the case of the turboprop, and irrelevant in the case of the turbofan, which came after the British lead had evaporated.

This goes to show how little I know about the subject. In my defence, this is a "technical appendix," that is, a hypothesis about an aspect of the history of technology that surprised me as much as I hoped it will you, and which we should keep an eye on as the postblogging series continues.

The inquiry turned up a planned military transport version of the Vickers Valiant, and a proposed commercial variant. The variant, the Vickers VC7, would have been powered by the first viable turbofan engine, the Rolls-Royce Conway. This would have made it the first turbofan commercial airliner by at least a year or two, but it was cancelled in 1955. Questions were asked in Parliament, and the received opinion discovers a "political decision" entangled in the earlier developments. Maybe it's the last moment when the British aviation industry could have been saved. Maybe not.

What can be said is that this brings the British aviation industry's grade down from a gentleman's D (pity marks for the Comet), to a clear fail. It's to be sent down to find a job in the City, where it can laugh at those poor, mucky manufacturers up at least through to Brexit.

So: Is there a profound lesson here about technology policy? Brendan Flynn thinks so. Arguing with someone named "Professor Geels," who has a sociology-of-science explanation about research sites and networks, Flynn proposes that the missing ingredient is state funding. Can Rachel Summers send Kitty Pryde's mind back to the 1980s to put everything right with the one crucial bit of information that a bit more Keynsianism is needed?

 Is this an accurate picture? At this point, who knows?  I'll state it affirmatively, but with mental reservations. I'm really just laying out a programme of research.
Mentalites: Gadgets

Not only is this image scraped from Greg Goebel's posting here, I am also leaning on his research and reporting his conclusions. I feel so sleazy. That said, I do have a slightly different angle. This is Bristol's concept for a "100 ton bomber," as proposed by the Air Ministry in 1943. It bears certain similarities to the Convair B-36, not least of which being that, had it been built, it would have turned out to have grown far past optimal weight and become a near white elephant. The 100 ton bomber led to the Bristol Brabazon, the Brabazon led to the Bristol Britannia, and the political need to support Britannia production in Belfast led to a political decision to cancel the VC7. Now throw in one more neat little connection: The Britannia's early teething problems were due to its engine being designed as a reverse air flow type.

Ice built up in the "U" of the air inlet of the Bristol Proteus, causing all sorts of problems with the prototypes. Now look at the 100-ton. It has pusher props, necessitating reverse air flow into (turboprop) engines. So that's where this doomed "improvement" came from!

A Proteus.

Probably not, actually. Reverse flow is also a popular gimmick of the day, and this story is way too neat. I want instead to take a slightly different angle and single out the way in which pusher engines keep popping up on "gadgety" designs. The 100-Ton Bomber also has the 4x20mm cannon turrets that popped up on the B.1/39 ideal bomber, and, no doubt, the fire control computer that was to go with them. The B-29's computer pops up often enough in contemporary literature for us to know that it was a big thing in the day. Like many pioneering IT applications, it never actually had to work, so, you know, "win" right there. The Proteus, by way of contrast, may have gone on to a decades-long career as a power plant, but failed when it was most needed. The gadgety kind of gadget that's called upon to work in the real world often does that, too.. 

I guess the moral of the story is, for now, that you need to keep your science fiction writers in check?

Mentalites, Gadgets, Engines

This is a Lockheed Constellation. I'm not sure that it's one of the Connies that used a Wright R-3350 Duplex Cyclone Turbo-Compound engine, but that particular gadget, which saved the Connie and the DC-7, is the hard core about which the next bit of discussion is going to wind.

I'd discuss the turbo-compound engine at greater length were it not that regular readers already have some sense of the gadget from its frequent recent advocacy by British internal-engine design doyens Rod Banks and Harry Ricardo, in these pages. If you've avoided the details, the Wiki page is pretty good; if you're not inclined to follow up on that, it's basically an elaboration on the old turbocharger. You put a turbine (specifically, a new kind of turbine called the "lay-down," or Power Recovery Turbine) in the exhaust stream of an internal combustion engine and extract more energy from it. It's a great deal more thermodynamically efficient than a jet engine, as Banks and Ricardo keep reminding us. In the Wright engine, it also gives an extra turn of speed and range, which is how it came to save the Constellation and DC-7. The Brits, on the other hand, want to use turbo-compounding on an aircraft diesel engine, already a dodgy proposition, to create a very long range engine suitable for a maritime reconnaissance aircraft. 

It's hard to sufficiently emphasise what a goofy idea this is. Starting down the road of designing a new engine in 1947 is basically a bet on the turbojet revolution failing. This isn't an impossible position to take in 1947. I'll direct you again to the link through to the Wiki article on the DC-7. Douglas started designing the -7 with the jet age clearly in sight, and the plane appeared after the Comet, and was specifically aimed at a specification (coast-to-coast in under 8 hours) that might have been written for the Comet. As it happens, even the fuel efficiency argument was scuppered by the appearance of the turbofan, which effectively addressed the economy gap between jets and large piston engines. More importantly, long distance flying in piston-engined planes was a dreadfully draining experience due to noise and vibration, and a diesel improved with additional parts hanging in the exhaust flow would just make things worse, Britain's main postwar long range maritime reconnaissance aircraft had a terrible safety record and it is not that hard to see that coming. The 8 hour requirement, after all, was based on the CAA's requirement that commercial flight crews not spend more than 8 hours in the air, consequent to coast-to-coast flights' tendency to discover the Appalachians and San Francisco Bay.

Although, in multiple levels of irony, the Argus, a highly successful RCAF maritime reconnaissance aircraft, was a Britannia variant with Wright Turbo Compound engines.

That, however, is central to my point. The Napier Nomad was hardly the London engineering community's first gadgety intervention in the engine-design process. Way back in the early 30s, they had considered the problem of developing a really hot interceptor, and concluded for various impeccable reasons that the next generation engine would have to get rid of the old-fashioned poppet valve in favour of the sleeve valve.

Close up detail of a Bristol Centaurus Sleeve valve. By Kitplane01 - Own work, CC BY-SA 4.0,
 Roy Fedden, of Bristol Engines, took them seriously, poor dear. While everyone else in the world (except London engineering firm, Napier) pursued incremental improvements of the poppet valve, he launched into a research programme to create a metal that would meet the necessary hardness and heat resistance requirements while matching the expansion rate of the aluminum of the cylinder. Heroically, Bristol Engines succeeded, yielding an engine with minor advantages that, in the way of drawbacks, needed to have the sleeve valve disassembled and reground on specialised machine tools in the same circumstances where the ground crew servicing a Wright engine would just replace a valve.

Because Fedden had launched the company in the sleeve valve path, when it came to designing the Britannia in the late 40s, the company had no real, conventional choices. The British industry, and the Navy, was livid that there was no British equivalent to the gigantic late-war American radials, and it would be a hard, long slog to create a "double Centaurus." A twin Centaurus installation was partr of the Brabazon proposal, but was a bit science-fictiony even for the era. It wasn't clear that the Brabazon I would fail when the Britannia proposal was being put together, but it remained the case that Bristol was eager to put the Centaurus into the plane, and get a bit of momentum going. Or there was the alternative of the Nomad. Or, it could take a flier on the new"turboprop" technology. For lack of anything better, the Britannia ended up being a turboprop airliner.

Part Marks for Sympathy: de Havilland Comet

Ah, sympathy grades. How else would I have passed "Topics in Quantum Mechanics?" This is a grounded DC-6, repurposed as a diner.

I assumed from the premise that this had to be in Florida, but, no, Coventry.  The old guy has that "How do I even begin to explain?" look going on.
This is the best I could do for the windows of the De Havilland Comet airliner:

Reading the ad from right to left, the far frame shows a model thinking deep thoughts, presumably about being away from her family and how scary London/New York are likely to be (big ups for the educational outcomes of Italian reading language courses!) on a de Havilland Comet 4. It's the closest I can come to the idea that the Comet had large windows, although, to complicate things, it is the revised window design, far closer to the DC-6 than the original Comet, with its square windows. The point here is that we make a big fuss of the fact that the Comet had much bigger windows than you would expect from the first aircraft to fly commercially at 40,000ft.

The reason for the fuss is that the square corners of the windows led to structural discontinuities that led to fatigue fracturing within unexpectedly low lifetimes, leading to the mid-air breakup of three Comet airliners, including two that broke up in clear air in 1954. The type was recalled, rebuilt with thicker skin as well as oval windows, and re-entered service. It was a financial fiasco, forced BOAC to buy  too late to capture the magic of the "new Elizabethan Age" that seemed to be dawning on 2 May 1952.
A section of a Comet I used for testing at Farnborough and released to the de Havilland Heritate Museum. Posted by user Norwich (imputed c. 2013) at

Looking back, we might do better to see the Comet's failure as inevitable, than as a consequence of a single, tragic flaw. Donald Douglas' bet against the jet was hardly unique. Prehistories of the jet turbine in England emphasise A. A. Griffith's effort at RAE to develop an axial turbine suitable for powering a turboprop. That seemed like the right way to go to pretty much everyone interested in the subject, and the appearance of Frank Whittle's centrifugal turbine, powering a straight jet engine, blindsided many other people besides Griffith; but Whittle was trying to power a skyrocket, not a liner. What might be worse, his solution was radically different from the turbocompounded two-stoke diesel sleeve valve monstrosity favoured by the London engineering community.

Once Whittle's formula was accepted, it was inevitable that other, established engine designers would get into the business. In general, the axial turbines favoured by Griffith were too complicated for small design and consultancy offices, but Frank Halford, the designer behind several Napier engines and the small, civil plants that de Havilland built as a sideline, was willing to take a swing at a centrifugal engine. The result was a series of simple, early jet engines: the Goblin and Ghost. The Ghost, in, particular, was mature and ready for some new applications by 1952.

Meanwhile, there was de Havilland. Taking its own course in the late 1920s, when the Air Staff set its face definitively against the use of wood in aircraft structures, de Havilland had persisted with lumber, in both its old fashioned, timber form, and the new-fangled plywood. Plywood being only as good as the epoxy that glued the layers, it also took up with new materials guru Nick De Bruyne [pdf], The upshot was a series of famous planes, with the war winning Mosquito bookended by the the Comet Racer and the Vampire. With this much experience in cutting-edge, high speed aircraft made possible by aerodynamic refinement and novel structural materials, Throw the Ghost into an airliner built that way, and de Havilland could visualise a paying jet airliner, refuting all the critics who thought that a jetliner would be too expensive and short-ranged.

As it was, the 1952 Comet wasn't quite a worldbeater, but with the Ghost replaced with the Rolls Royce Avon, an engine originally developed for the Mosquito-replacing Canberra, a "stretched" Comet would have a transatlantic range. With the success of the original Comet, orders were starting to come in when-- I guess the point is, here's the Avon, and the Canberra.

So pretty. And so good for the balance of trade! By Photo: SAC A K Benson/MOD, OGL,
Full Marks in a Hard Currency Era: The Viscount

It will not have escaped regular readers that the 1940s and 50s were an age of dollar shortages. The world simply wanted to buy more American stuff than America wanted to buy of the world's. (Before Yank readers get too chuffed, a lot of that had to do with good old agricultural exports and POL.) So a really successful British aircraft is a big deal. It's an import substitute, and that matters even if it replaces old DC-3s, since the spare parts come from America; and if it generates North American exports, it's golden. The Canberra, pictured above, was an almost unthinkable sale of a British military aircraft to an American air force. As the Wiki article makes clear, in spite of the aircraft's huge margin of superiority over American rivals, this could not have happened were it not for the Korean Emergency; but, given the sale, further penetrations of the American market became inevitable, as Wright undertook to license produce the Armstrong Siddeley Sapphire (the follow-on to the Avon). Wright's sales team would become an agency charged with scaring up dollar-denominated invisible exports. Of course, we're talking about the Vickers Viscount, here, and in spite of being a highly economical buy with handsome sales, it had somewhat less American penetration.

The reason to lean on the Canberra in a discussion of the Viscount is the engine angle. The Viscount had a Rolls Royce engine, and succeeded, while the Apollo had an Armstrong Siddeley engine, and failed. It seems that the ultimate reason that the Apollo was abandoned is that it took too long for it to get into the air, due to delays in developing its Mamba engine. This is odd, in that, up to this point, the two companies had pretty much parallel developmental schemes. Both companies built an initial turboprop engine based on mating axial stages to centrifugal, resulting in very powerful, but problematic engines. Rolls Royce quickly abandoned the Clyde, while the Python was put into one of the more troubled postwar British military aircraft, but, if anything, the sad story of the Wyvern implies that A. S. put little more effort into the Python than RR did into the Clyde. Without leaning into an excessively deterministic line about Rolls Royce just being a more functional organisation than Armstrong Siddeley, the main difference between the two companies is that in 1948, Armstrong Siddeley inherited Metrovick's line of jet engines: above all, the world-beating Sapphire.

So the tentative story I'm going with is that while Rolls Royce was flying Viscount prototypes for BOAC bigwigs and generating the change orders that eventually scaled the aircraft up from 30 seats to 48, the Apollo was stuck on the ground for lack of working Mambas. By the time the Mamba emerged from the shed, the Apollo was hopelessly behind the Viscount. And, anyway, the Korean War was on, and all hands were on deck for the Sapphire and whatever Armstrong Whitworth was working on. (A.S. and A.W. both being divisions of Hawker Siddeley, but with a residual special relationship from when they were two divisions of the  group that was previously named something else. "Armstrong" being a signal of truly convoluted corporate histories.)

And so the Viscount emerged, unique, proved the worth of the turboprop, for which see the Canadian engineering report already linked to, and here it is, cut and pasted, in case the linked website goes down someday:
The version of the Viscount that TCA is receiving has a permissible take-off weight of 60,000 lbs., with a landing weight of 54,000 lbs., and a zero fuel weight of 49,000 lbs. It is powered with four Rolls Royce Type R.Da3 ‘Dart’ engines with a take-off rating of 1,400 SHP and modified to permit cruise power associated with a take-off power of 1,550SHP. The actual TCA engines as received will not have the particular modification to permit a take-off at 1,550 SHP because the present 1,400 is more than ample for the 60,000lbs involved.
The latest information available indicates that TCA's Type 724D ‘Viscount’ has a block speed 25 MPH faster than the Convair, across practically the whole band of ranges from 300 miles to 1,100 miles.
The optimum cruising altitude of the Viscount remains at 15,000ft for ranges up to 600 miles: it then rises sharply and is 22,000ft for ranges above 700 miles.
Although the Viscount consumes considerably more fuel than the Convair, the cost of fuel burned, assuming JP-4 for the Viscount in Canada, is almost identical for both airplanes over the entire band of ranges from 300 to 1,100 miles.
Mr. Zevely mentions that 85% of all airline trips are 300 miles or less: we do not doubt these figures but merely wish to point out that this does not apply to TCA’s particular operations where almost all of the flight legs that would be flown by this airplane would be at least 400 miles.
According to TCA’s cost estimates, the direct operating cost of the Convair would be .05c per seat mile cheaper than the Viscount, assuming both airplanes have the same depreciation period, and making numerous other assumptions to cover the fact that Viscount costs are of the most part estimated rather than based on experience, like the Convair costs.
Both airplanes can be equipped with 48 passenger seats. Assuming this configuration in order to put the airplanes on a comparable basis, the break-even passenger load will be from 1 to 2 passengers less for the Convair than the Viscount. Again it must be remembered that TCA has made very conservative assumptions in order to cover contingencies with the Viscount because it is a new airplane. It is our opinion that within two years after the airplane comes into service with TCA, the direct operating cost of the Viscount will be less than it would be for the Convair and the break-even points similarly equal or in favour of the Viscount.
We can sum up the direct operating costs situation for the two airplanes by saying that for TCA’s inter-city services, it is expected that the Viscount will cost a little more than the Convair the first year of operation: the two airplanes would be about equal the second year: and thereafter the Viscount will be expected to operate at a lower cost.  
VC7: Failing Grade

There are many reasons why the VC7 might have failed in service. At the same time, I want to point the reader to the fascinating discussion of the genesis of Douglas' DC-8, the pioneering Douglas jetliner that was such a sales failure that the Wiki blames it for the McDonnell merger. Again, quoting the anonymous expert, Donald Douglas expected to be competing on a level playing field with the 707, but was undermined by an early Air Force order for the Boeing plane. Boeing had won a pioneering order for airborne tankers a few years earlier, and learned the hard way that it was almost impossible for the jet B-47 to fly in formation with prop-driven KC-97s. The upshot is that Boeing had the KC-135 jet tanker design in its pocket when the USAF circulated its call for proposals in 1954, and won an 800 aircraft USAF tanker order. With that momentum behind it, it is perhaps not surprising that the Boeing 707 rolled over its rivals. The one obstacle in its way was a technological step forward; and, in fact, a number of 707s were built with Conways before the arrival of the JT3D. Eventually, surplus JT3Ds left over from decommissioned 707s and DC-8s were even fitted to KC-135s.

This isn't the place to dive further into the Vickers Valiant development programme, about which I am sure I will learn a great deal more. It suffices to say that engine and airframe met the developmental targets, and there is no particular reason that the V1000/VC7 couldn't have done the same. For those who wrinkle their noses in the general direction of state-supported military transport/civilian airliner projects, a timely reminder of the 707's origin in the massive KC-135 programme seems apropos.

Towards Conclusions

It may seem as though the Comet, Viscount and VC7 are succcessive choruses of the same sad story. (The Viscount isn't a sad story, but we can append the Vanguard and the flit-through Britannia to make it so.) I do not think that this is the case at all.

First, the story of the Comet is purely one of hubris, and the tacit conclusion that, had it not been for the Comet, it might well have been an even more horrifying engineering failure that reined de Havilland and the engineering community in.

Second, the story of the Viscount isn't anywhere near the exception it seems, once its origins as a commuter "that grew" is taken into account. British feeders had a solid history before and after the Viscount. The Viscount's intrusion into the next category up is a feel good story about technological advantage, though, and here it is an exception that might prove a rule about pure ingenuity not doing much good in the absence of proper funding.

Third, the VC7 story, while shortchanged here, underlines the most important, and well-telegraphed point, which is that transatlantic airliners represent a sector of the economy in which first costs are simply too high to rely on the private sector's capacity to mobilise risk. This is an observation that strikes me as controversial only if one ignores the context of American development. That is, American airliners are held up as counterfactuals of pure private sector virtue. In reality, small wars in Asia and Mutually Assured Destruction stood behind the airliners of the golden age of the American airliner. Nowadays, Boeing struggles to fund its new planes even within the framework of a global duopoly.

Other conclusions can be drawn; but for the sake of this entry, I think it suffices to hang a question mark over the question of just how viable technological innovation is without state intervention --mainly from the military industrial complex, more's the pity, but identifying the problem is the first step to solving it. 


  1. Without leaning into an excessively deterministic line about Rolls Royce just being a more functional organisation than Armstrong Siddeley, the main difference between the two companies is that in 1948, Armstrong Siddeley inherited Metrovick's line of jet engines: above all, the world-beating Sapphire.

    So the tentative story I'm going with is that while Rolls Royce was flying Viscount prototypes for BOAC bigwigs and generating the change orders that eventually scaled the aircraft up from 30 seats to 48, the Apollo was stuck on the ground for lack of working Mambas

    This is precisely how R-R got the Whittle designs working as production articles though - they spent the money to get a lot of them on test in parallel, allowing them both to run up total hours 10 to 20 times faster than Power Jets or Rover, and to implement changes without reworking/breaking the one-true-prototype. They could also put multiple versions in test simultaneously and do, you know, experiments LIKE A SCIENCER.

  2. It's a little hard to sort this out, since the history of the Armstrong Siddeley and Metrovick engines is less well known than that of their Rolls Royce counterparts. The Dart definitely began development in 1946 under a team that had previously worked on the Clyde, and which would go on to the Tyne. Working Darts were available for test flying the Viscount in the summer of 1948.

    The dates for the Mamba are less well known (to a Wikipedia reader; hopefully the information is out there somewhere), but it is known to have had a 500 hour bench test in 1948. However, even in 1949, engine issues delayed Apollo test flying. That sure looks like something had come up.

    Axial turboprops were much more complicated than centrifugal engines. For example, the Mamba had a ten stage compressor and two stage power turbine, while the Rolls Royce Derwent had a single stage compressor and turbine. Adding on all of those additional stages doesn't just add to mechanical complexity directly. It also raises all sorts of questions about lubrication points, bearings, supports -all sorts of details that have to be got right before run tests are possible.

    Since the Sapphire was a clear winner, it would make sense for A.S. to divert drawing office and toolroom staff to it. I'm not saying that that is what happened; but it works as a hypothesis.