Sir Nigel Gresley - The L.N.E.R.’s First C.M.E.

Most of streamlining steamlocomotives with windtunnels was playing scientific for other peoples money.
They could not simulate interaction of ground and rotating of 2meter big drivers with twenty spokes each where the upper parts moved 400 km an hour perpendiculair against very turbulent air.
Try that with aircraft propeller.
The 05 002 and Mallard was very different looking and jury is still out who ran fastest.
The form chosen can not have been results of scientific devellopment with hundreds of tests and changes.

 

Hi Hermod. Sorry but you couldn't be more wrong. Frederick Johanssen's archive on his work under Professor Dalby at the National Physics Laboratory includes the work on the LNER and LMS streamlining proposals and their approaches are highly scientific. With respect, if you've not seen this archive (and I would encourage you to seek out Johannsen's work) then you can't say it was unscientific.

I go into this at length in my book and I am confident that both the 05 and A4 classes are "correct answers" for the streamlining of locomotives. The A4 represents a greater development which is the streamlining of an entire train, and in Gresley's case, a wider project which was probably the earliest full systems development of high speed rail (but war inteverned).

Having studied Mallard's dynamo-meter roll at length and now waiting on peer review for some work I have done, I feel confident in saying that I believe Mallard was indeed fastest.

 

May I suggest: " developments in other countries" and also cross fertilisation with other fields - especially aeronautics.

Also I do think the penultimate sentence "My premise is that such ideas were "in the aether...." mentions something important: there is a period when something is a clearly generally seen as a possible major step this is recurrent - coal rather than coke burning in locomotives, how to get a good filament in a light bulb, for the last fifty years anything to do with electrical batteries - while there is no overwhelming standard solution people keep trying. While they try it is not easy to be clear what idea has contributed to what further string of ideas let alone whose ideas - what would be vital clues to us now put immediately in the waste paper basket as working notes.
Holcroft and others all contributed to the idea being known of and not a nonsense to consider - eventually the two to one levers were never improved on as a conjugated system to drive the valves for three sets of cylinders from two sets of valve gear.

Finally because of the paragraph above there is a real problem of two people coming up with the same solution quite independently at the same time - the famous example is the miners safety lamp: George Stephenson began to produce experimental lamps first, Sir Humphrey Davy began his experiments second but was believed in his sucess first, however, Stephenson's lamp was used in a mine
before Davy's and in the end the best features of both lamps were combined. Two people simultaneously separately inventing the same
thing can easily get very fraught, especially if a lot of money seems to hang on priority for a patent, but it does really and recurrently happen.

 

We know that the model was modified in the wind tunnel: the model was made in "plasticine" - pliable clay - originally the model must have been faired in behind the chimney in some way and the effect of the smoke from the chimney was being simulated which clung to the top of the boiler apparently obscuring the view from the cab, then when - accidently - a depression was made behind the chimney the exhaust rose clear.

And the full-size arrangement of the front end streamlining on the A4s bore this out in practice.
There must, presumably, have been separate tests with an unstreamlined model to compare with the streamlined model to establish the benefits of streamlining.

 

Hi Hirn. The development models for the A4 Pacific were wooden test ends.

The model with the depression for the chimney was where plasticine/clay was applied to modify the model's shape.

 

Perhaps you should also consider the PRR T1s? In his book 'American Steam Locomotives Design and Development 1880-1960, Bill Withuhn discusses interviews with two employees of the Franklin Railway Supply Company. Withuhn's interview of Delano and Kirchhof, independently, offered the same story without prompting from the interviewer. Namely, that a "spy" - a Franklin staffer was sent to ride trains anonymously for a month on the Fort Wayne Division and to clock their actual speed. This was in connection with an investigation into broken valves. There were some instances of speeds up to 140mph. He presented his logs and watch to Kirchhof for verification. Once or twice a week in the recorded month, when a train was ten cars or less in length and running behind schedule, the engineer had made up time by exceeding 125mph. Twice that month, with short trains of six or seven cars, speed had reached 135 to 142mph, as clocked over several miles. Careful inspection of logged entries and watch, as well as the consistency of successive time intervals between mileposts on all the timed runs, attested to the veracity of the logs.

 

Folks, folks, be sensible.

The science of aerodynamics was in its infancy in the 1920s and 30s. To argue that research at that time was quite inadequate by modern standards is true, but quite besides the point. The only question should be whether they did a reasonable job by the standards of the time, which is surely true.

Similarly to a good extent the quest for who influenced who is, I suggest, not only futile, but also feels more like fannishness. Many railway design shops of the time could be strikingly insular, but this tends not to be true of the greats, who would take inspiration from anywhere, and this was a strength, not a weakness. Chapelon and Gresley were both highly competent men who doubtless kept a good eye on what each other were doing, not to mention other top design schools like say Wagner's in Germany and the US builders. To say A definitively influenced B and not the other way round is really rather pointless.

There's a nice parallel in the field of racing sailboats, which I'm familiar with. We can trace the origin of one particular innovation back to a post race conversation in a club bar in Sydney. A made an observation about boat performance, and B said something on the lines of "In that case why don't we", after which at least two of those present decided they were going to try exactly that. The first to actually race a boat with the innovation and complete the detail design to make it work is usually given the credit, and according to *nearly* every version of the story that was neither A nor B! Communications in the 1930s weren't that slow, and while it might take a few days to disseminate an idea rather than a few hours with email, it's still quite trivial compared to the time needed to actually convert the idea to metal.

 

1. The Air Resitance of Passenger Trains. F.C.Johansen paper given to the I.Mech.Eng December 1936.
   
   QUOTE “ After the meagre information available in 1931 on the air resistance of trains the LMS advisory
   Committee for Scientific Research initiated the experiments on which this present paper is largely based.
   The investigation was carried out with models in a 7 foot wind tunnel at the National Physical Laboratory
   on the common behalf of the LMS and LNER Railways. These two companies together with the Southern
   Railway to whom the results were communicated, defrayed the cost of the work “ END QUOTE
   
   The Paper attracted a lot of discussion, ( 48 pages in total ) the following quote is interesting ?
   
   QUOTE “OVS Bulleid said he was quite unable to understand how a theory as to what would happen
   to a full-size train under working conditions could be built up from results from small models,
   obtained under such conditions. “:END QUOTE
   
   My personal observation is that the quadratic equation that Johansen derived from this work for
   resistance gives values between 10-15% higher than later on the road derived data.
   
   In the USA in the late 1920s early 1930s similar tests were undertaken. ( Alco at NYU and the
   Burlington route at MIT). Regarding loco streamlining there was a c. 60% reduction in air resistance
2. at 60mph. This represented only a small resistance of total train resistance ( both the consist
3. and the mechanical and frictional resistances of the loco. )

1. The first streamlined locos in the USA were I think on the New York Central in December 1934
2. (rebuild of existing loco) and the first streamlined locos built new in April 1935 for the
3. Hiawatha. I think streamlining in the USA had more to do with post the 1925 Paris Exposition
4. ( and the subsequent rise of art deco etc, ) than savings in energy.
5. 
   Similarly I understand that Stanier disassociated himself from the Streamlined Coronations.
6. 
   (ps please excuse the numbering, not sure where it has come from )
7. 
   Michael Rowe

 

Hi Michael, I would be interested in your working out for Johansen's data, because I have the full LNER and LMS reports and my observation is that the principle of the wind tunnel tests was to recognise that the models in isolation were not going to represent the full size models, only trends towards more aerodynamic shapes. In particular, we should note that his work on the "ideal train" and the modelling that came out of it shows principles that we have adopted in the modern era for streamlined designs.

Re your points:

I quite agree, which makes the emergence of the W1 in 1929 all the more significant, given its final outcome was a genuinely streamlined locomotive.

Which is bizarre, arguably in terms of technology it is the LMS' crowning glory in steam traction (the obvious development of their diesel units 10000 and 10001 not withstanding).

 

Apologies for my shorthand. It was the streamlined casing Stanier I understand considered superfluous.
( as arguably did the later LMS authorities )

Using Johansen R = 4 + 0.025V + 0.00166Vsquared

at 20mph R in lb/ton = 5.2, at 40 - 7.7, at 70 - 13.9, at 90 - 19.8

figures derived by BR for bogie passenger stock ( for 7.5mph wind at 45 degrees. )
at 20 - 4.1, at 40 - 6.5, at 70 - 11.6, at 90 - 16.3

for very nearly still air BR at 70mph 10.5 lbs/ ton

The significant influence of wind is often overlooked

Michael Rowe

ps Were Prof. Dolby’s wind tunnel tests concerned with ‘streamlining’ or
ensuring the exhaust was lifted clear? I have never thought the “Hush Hush”
front end deserved the epithet ‘streamlined ’ ( as per the Coronations, A4s
and 60700 ). I assume the casing was designed to cover the tri-drum boiler
and the curved ‘risers’ and ‘downcomers’

pp ss Hermod is IMHO correct. Models in tunnels cannot predict locomotive
resistance. Johansen stated ‘train resistance’ and his formula for R relates
to the consist. It is DBHP that earns revenue.

 

They can not perfectly replicate a locomotive's air resistance, but I imagine they can be useful in providing base data, where replicating the reality requires tremendous amounts of resource and space. It must be useful to use small scale static models in physical wind tunnels, because it is a practice that is perpetuated to this day.

Even with the availability of 21st century CFD software it is still usually quicker and often cheaper to put something physical in a wind tunnel and look at what is actually happening. My experience of vehicle design is that the effects of your rotating wheels is often left out until your very final versions of the test. Even today, wind tunnels with such capability are few and far between and very expensive to hire. You can simulate it with CFD, but you're putting your computation time up to days for every iteration and dropping another few £KK on the additional software license.

What the technology we have now available now has done is sped up the processing of your buck. It's much easier to drop a foam block in a CNC machine than it is to have someone carve it out by hand into wood. We still go through a lot of modelling clay before we commit anything to virtual steel.

On a separate note, is it worth iterating the difference between 'streamlining' and aerodynamic design again? Streamlining, especially in the context of the early 20th century, can be managed successfully by anyone with a basic eye for aesthetics. The metric of success is purely visual. Designing something aerodynamically requires the application of scientific method. Even if you don't get a successful result, you should at least have the data to prove that it has been computated.

 

But the aerospace industry makes great use of models in wind tunnels. Much can be learnt from them even if the real proof of the pudding is when the real thing actually flies.

 

For this non-mathematician, would it be possible to clarify this equation and the implications for the value of streamlining.

 

Essentially it is saying that train resistance has three components:

• A fixed one, ie one that is independent of speed
• A component that is proportional to speed (ie if you double the speed, you double the resistance
• A component proportional to the square of the speed (ie if you double the speed, you quadruple the resistance)

Typically rolling resistance goes with speed (ie all the friction in bearings) and air resistance with the square of the speed.

The effect is that broadly wind resistance (and therefore efforts to reduce it) is largely irrelevant at low speeds, but becomes increasingly important in the whole picture as the speed increases.

To put it in context, at 10mph, the train resistance is 4.4lb/ton, and 90% of that comes from the fixed component; only about 4% is the squared component (ie primarily wind resistance)

At 50mph the train resistance has risen to 9.4lb/ton (ie more than doubled) and of that, 44% comes from the square component.

At 100mph, the resistance is 23.1lb /ton and of that 72% comes from the squared component.

All assuming the figures in the equation are correct of course; empirically measured train resistances seem to have been a bit lower than those numbers suggest.

Tom

 

Thank you. That leaves a question about the role of streamlining in this.

 

The questions I’d have would be “(1) whole train is probably important, not just frontal area and (2) mostly irrelevant at the speeds that were typical in steam days except in a few cases.

On start to stop journey times, high acceleration / braking gains you more time than high maximum speed, especially where stops are frequent.

Tom

 

From Bert Spencer's paper in 1947 regarding the benefits of streamlining:



And a graphical version:



I am investigating the possibility of re-running the experiments done in the 1930s using modern tech, as part of my PhD.

 

It might also be worth saying that in addition to your maximum speed, a more aerodynamic train useless less power to maintain the same speed than an ordinary one would. In the context of aerodynamic trains at this time we're only really talking about long distance, non-stop services. Even at 50mph you can to make a noticeable dent to that 44%.

The design of the trains suggests Gresley was thinking about the whole picture. Powerful locomotives, hauling short trains, with articulated coaches (which lowers your rolling resistance), incorporating features (smooth panels, near flush fitted windows, chassis valances...) from front to back to lower your aerodynamic drag. Even a unique design of rear coach to finish it off. Taking out mass and rolling resistance means that your same loco can get those gains in acceleration and braking and be more adapt at climbing hills. All of that brings down your total journey time and gets you more performance out of your fixed power unit.

 

Yes, the W1 was streamlined. The casing was multi-finctional and was aerodynamically tested and modified by way of the wind tunnel experiments. Draughting for the smoke deflection, draughting for the boiler, and to be aerodynamically better. The multi-functional front end of the W1 is incredibly clever.

Bert Spencer disagreed, and his paper from 1947 was based on a mixture of modelling and real world data collected.

 

Indeed - the impact of gradients is really stark, particularly when you get past about 1 in 100, so cutting overall train weight has a big benefit there (also on acceleration and braking of course).

I wonder what the impacts were (on fuel used, availability and track occupancy) of needing to turn round the streamlined trains at the end of each journey. Arriving at King’s Cross, say, how many ECS miles did you need to do to get the Beaver tail to the back of the train? Or did they do it by shunting and turning just one carriage on a turntable?

Tom