The Start-Up Drama of a Steam Locomotives

Penichette docked at a mill on the Mayenne River, 2001
Penichette docked at a mill on the Mayenne River, 2001

Starting is a big deal for a steam locomotive.  Unlike the electric motors in diesel-electric locomotives, which deliver maximum power at start-up, steam locomotives are wimpy at startup and gain power with speed. This is the result of a fixed ratio between the pistons and the wheels: the pistons move too slowly at start-up to use the available steam power (you can’t “down shift”). There are four ways to help the locomotive start: increase the coefficient of friction, decrease the weight being towed, maximize the leverage of the rods that crank the wheels, and temporarily increase the number of driving wheels.

To move a train, the locomotive’s drivers must push back against the rails without slipping. The “coefficient of friction” is the percentage of the weight that can be converted into horizontal force before the wheel slips, about 25% for steel wheels on steel rails. One hundred tons on the driving wheels translates into twenty-five tons of pulling force.

So the first step to beef up the starting force is to increase friction by sanding the rails. For this purpose, steam locomotives carry sand in one or two large domes atop the boiler, kept dry by boiler heat. Tubes from the sand domes, each with a valve, curve around the side of the boiler and then around the perimeter of each driver, terminating just above the rail.

Sand pipe
Sand pipe

The second trick, if you are hauling uncomplaining freight, is to reduce the load you are pulling. This is possible because in American trains  lack the “buffers” used in European trains to keep the couplings taut, allowiwng some slack in the couplings between cars. So you set the brakes at the back of the train, and back down until the couplings are compressed together. Then you take off like a bat out of hell, pulling first one car, then two, gathering more and more cars, until the last car, always the unfortunate caboose, became part of the chain only when the train was already moving at 5 or 10 mph.  Serious injury was the fate of  conductors caught unaware, and broken “drawbars” that connect the coupling to the car were common. The sound of a starting freight train was memorable, as the clank of slack couplings coming together ran like a zipper down the train, a kind of rolling thunder.

With a passenger train this approach can’t be used. It was a matter of pride for the engineer not to “spill the soup” when starting. Instead, you stretch out each coupling so all the cars move at once, with no start-up jerk. You are stuck with the whole load.

The third move is back up a bit until the rods that turn the wheels are at the angle at which they exert the most leverage.

The fourth move was available on a few locomotives, including the SP’s 4-8-4’s. These were equipped with “boosters”, a compact steam engine set between the wheels of the trailing truck, that added the weight on that truck to contribute to the tractive force at low speeds. They added almost 25% tractive force while emitting furious sideways snorts of steam, in detailed counterpoint to the much larger main drivers. It made a glorious show for the enjoyment of a track-side teenager.

So the engineer adjusts the position of the wheels, sands the rails, turns on the booster, applies steam very carefully, and hopes for the best.  A youth at track-side hopes for the worst: the drivers lose their grip, and the locomotive rapidly chuffs and clanks in place, like a trained horse.

The gradually accelerating tempo of the exhaust blasts has been imitated many times in music. Villa-Lobos’ The Little Train of the Caipira is a delightful piece that vividly captures the sound and motion of a train ride (listen and watch the wonderful graphics at https://www.youtube.com/watch?v=1rRFDnTEu6g ). Prokofiev intentionally or unintentionally captures the essence of a starting locomotive in the second movement of his great 5th Symphony. You can hear it at https://www.youtube.com/watch?v=-e0c4GRx4So : the start-up begins at 5:06, but I encourage you to listen to the whole movement – it is a thrilling performance (learn more about this remarkable youth orchestra at the Wikipedia entry for “El Sistema”). Train enthusiast Arthur Honegger’s 1923 composition Pacific 2-3-1  is best heard at https://www.youtube.com/watch?v=T9u7_WAkAPw. I also found a wonderful 10-minute 1949 film that captures the excitement of the steam locomotive, using Honegger’s score. The clearest video on YouTube is at https://www.youtube.com/watch?v=rKRCJhLU7rs  (it is the better for being without sound).

Why I Went to Architecture School

I drew a lot when I was a kid, mostly sequences depicting incredible explosions, or cars each with more exaggerated features than the last. I didn’t have good drawing materials, just bond paper and pencils, so the drawings had no depth – they were outlets for my distressed imagination rather than productions for display.

Then I got into model railroading, and spent all my spare time designing the layout, putting the layout together in our tiny cellar, or making rolling stock, but I never finished the layout – I dreamed its completion, just as I dreamed my own completion, my own empowerment. I was never finished.

My uncle Bud, one of father’s half-brothers, was a contractor, and he designed and built suburban houses in Southern California were we lived at the time. I watched his and others’ houses go up, utterly fascinated by the wood framing. I remember climbing onto the roof of one of his houses under construction and sticking a meat thermometer under the black shingles – it read 180 degrees, a datum that I found useful many years later.

I took mechanical drawing in high school and loved it. I loved the wonderful ruling pens, I loved the precision, and I loved the geometry. We drew other views of objects for which some views were given – a great aid to spatial visualization (which may be learned and not innate – why don’t they teach mechanical drawing today?)

My brother Bob and brother-in law John ran a blueprint company for a couple of years and I helped out, trimming blueprints. In those days, the early 50’s, blueprints were really blue. A roll of heavy paper coated with light-sensitive dye was kept in the dark under the machine. You pulled the paper across a table and up through a series of rollers. Once the machine was started, you laid the tracing paper original drawings one after the other onto the moving paper, being careful to align them correctly and avoid wrinkles or folds.

In their trip through the machine’s rollers, the blueprint paper and drawings first moved under a brilliant light source, which exposed the paper except where the pencil or ink links on the tracing paper blocked the light. The paper then ran up vertically, and you had to grab the originals as they peeled off the blueprint paper (while simultaneously feeding in the new originals – it required some skill). Up and over, the paper then ran through a bath of developing fluid that activated the dye, turning the exposed paper a beautiful Prussian blue.

The wet paper traveled over some burners that dried it (and shrunk the image – hence the dictum “never measure a blueprint”). All this paper was under tension, and if it got off track, it would wrinkle up dramatically, and you had to cut the paper and re-feed it. Finally, the trimmer used long scissors to trim the final prints as they came out of the machine. I searched the web and could find only one image of blueprint machine, from a patent application. This one has many more rollers than the ones I worked with, and doesn’t have a table to trim the prints.

Blueprint machine diagram from patent application
Blueprint machine diagram from patent application

The new “diazo” process was just coming into favor. Diazo paper was coated with a yellowish dye that was actuated by intense ammonia fumes. In this process the paper didn’t shrink, plus you got a black on white image (actually more purple on yellowish-white). Not long after this, the diazo process replaced blueprinting, but the old name stuck. The diazo machine was trade-named “Ozalid”, which is diazo backward with an added “L”. The ammonia came in big 10-gallon glass bottles. Once my brother-in-law dropped a full bottle and we all had to run out of the shop before we burned our lungs. Both diazo and blueprints faded when exposed to light for any period of time.

When I trimmed blueprints and diazos in the blueprint shop, I got to look at the plans for new houses that we were printing. I would take cast-off prints of them home to study, and then draw up my own floor plans. All these houses were one-story ranch houses without basements, with hipped roofs, one where the roof planes slope in all directions, like a tent. You can design a hipped roof over any plan, no matter what its shape, so you never had to think about anything but the plan layout. That’s where the idea of architects “drawing up the plan” came from. Few people were sophisticated enough to ask an architect to “draw up the spaces.”

So that’s why I went to architecture school at Berkeley: to learn how to draw up the plans for houses. Nothing more enlightened than that. My folks somehow managed to pay for college, which was affordable in 1953. The $1,200 cost per year for tuition, room and board was partly offset by $200 scholarships, available to anyone who had a B average. I had a scholarship every semester, one I remember getting because I was from Nebraska. In later years when I had my own apartment and car, my annual outlay including plane fare to and from home for Christmas and summer vacation, was around $2,500.

My first job as an architectural drafter (draftsman in those days) was 65 cents an hour, so $2,500 was not chump change, especially for my not so rich parents. It’s an interesting exercise to compare the 65 cent hourly wage and $1,200 annual cost with contemporary numbers. Let’s say a beginning drafter in an architectural office makes $15 an hour ($31,000 a year). The cost of a year in college varies, but I read on the web that Berkeley is in the $30,000 range. So that’s a year’s pay. A year’s pay at 65 cents an hour is $1,352. If my numbers are correct, today’s college tuition compared with earning power hasn’t changed that much, at least at Berkeley. It has become much more selective, however. In the 50’s, you got in if you had a good grade-point average and were a California resident.

Needless to say, architectural education does not consist solely of “drawing up the plan.” But that is another story.

I'm still drawing up plans for houses, typically in traditional styles
I’m still drawing up plans for houses, typically in traditional styles

MBL Lectures

Ritterhof courtyard, Weingut Fitz-Ritter, Bad Dürkheim, Pfalz (founded 1785), watercolor and ink 1984
Ritterhof courtyard, Weingut Fitz-Ritter, Bad Dürkheim, Pfalz (founded 1785), watercolor and ink 1984

We summer in Woods Hole (when our house is not rented, which is most of the summer) and occasionally can go to one of the Friday night public science lectures at the Marine Biological Laboratory, the world-famous research facility in Woods Hole. It is a magnificent privilege to hear and see world-class scientists give beautiful slide lectures on fascinating, cutting-edge science.

On occasion, when the lecture is particularly clear and my brain is fresh, I can go home and write down the gist of the lecture. So since I haven’t posted anything for almost a month, I dug up one of these writeups.

MBL LECTURE 19 JULY 2013

The lecturer, a famous Chinese neuroscientist Mu-Ming Poo, around 70, spoke on neural plasticity.

He started with a much-cited quote by the famous neuroscientist Donald Hebb, to the effect that neurons that fire together wire together, forming more or less permanent circuits – i.e. memories. This is a qualitative statement; Poo and his colleagues sought to explore it quantitatively.

He asked how close together in time the two firings had to be in order to create a memory. He found that the window generally was 40 milliseconds wide, with exceptions in certain animals. He also asked whether the sequence mattered (i.e. what happened if the second neuron fired before the first one), and he found that when the sequence was reversed, exactly the opposite effect occurred: the two neurons became less likely to fire together than previously.

Background: neurons collect inputs from “dendrites”, sum them in complicated ways, and if the incoming stimulus is sufficient, they fire. The electrical signal runs rapidly down the cell’s axon (at about 45 mph), which is the long “wire” that carries the electrical charge from the cell to the other cells to which it is connected, when the cell fires. During development, each neuron seeks out and finds the neurons in the part of the brain to which it “should” be connected. Retinal cells connect via intermediate links to cells in the visual cortex, which in humans is located at the back of the brain (“cortex” is the thin grey-colored coating of neurons on the outer surface of the convoluted brain). Back to the lecture.

To understand why the 40 millisecond window is adaptive, imagine a row of adjacent retinal cells. Each retinal cell connects with a large number of cortical cells in the visual cortex that are adjacent to one another. Call the retinal cells “A” and the cortical cells “B.” This means that each B cell receives inputs from a lot of A cells. So the image created by one A cell is spread out and blurred in the visual cortex.

The goal is to create a sharply focused “map” on the visual cortex that matches the image falling on the retina. To accomplish this, the brain needs to prune away connections between A cells and distant B cells, and strengthen connections between A cells and B cells that are close together.

Imagine a moving spot falling on the retina and hitting one A cell. The A cell will fire and cause all the B cells to which it is connected to fire. Since the B cells fire together within the 40 millisecond window, with the A cell firing first, the connections are strengthened. Now the spot moves to the next A cell (it is moving fast). Again the A cell sets off all the B cells to which it is connected. But some of these will already have fired during the previous 40 millisecond window mentioned above. In those cases, the B cell has fired before the A cell, which weakens the connection. Over many occurrences, this process sharpens the map in the visual cortex.

In the second example of neural plasticity, he explored how mature neurons in a frog’s brain form short term memory by training a string of neurons to fire in sequence. Remarkably, there are instruments that can probe individual neurons in a living animal brain, as well as a brain in a petri dish (in vitro – glass – as opposed to in vivo – life). First the investigators associated neurons in the retina with the corresponding neurons in the visual cortex. Then they passed a moving spot over the retinal cells and noted that the cortical cells lit up one after the other.

After doing this many times, training the cells, they then stimulated just the first retinal cell, which caused the string of cells in the visual cortex to fire one after the other. The neurons had learned that the spot moves on this particular track  (this is short term memory, lasting only about 10 minutes). When they stimulated the last cell in the sequence, nothing happened. They then stimulated the cortical cells directly, and the same thing happened, showing that it was the cortical neurons that learned and not the retinal cells.

In a third demonstration, they found that cells in a zebra-fish could remember the timing between sequential stimuli. This became evident because if they stimulated the cells five times or more, the cells fired one more time after the stimulus was removed at exactly the same interval as the initial sequence. This occurred at intervals up to about 10 seconds. The larger the number of sequential stimuli the more firings occurred after the stimuli stopped, but only up to 3 repetitions. He showed a movie in which the stimuli caused the tail of the fish to twitch to the side (an escape behavior), and sure enough, after the stimuli ceased, the tail twitched twice at exactly the same interval as the stimuli.

Finally, it was believed that only humans and some apes could recognize themselves in a mirror and that monkeys could not. He experimented with Rhesus monkeys. If you paint a spot on the monkey’s face (or even shine a light at the spot so he doesn’t feel anything) he ignores it when looking in the mirror, showing that he is not aware that the image is of himself.

So Poo did a clever thing: he applied the spot in a way that irritated the same location on the monkey’s face, which caused the monkey to reach up and touch the spot. By doing this many times, he trained the monkey to associate the two spots and thereby become aware that the image in the mirror was himself. Once they learned this (2 out of 3 could do so) they took advantage of their new skill by examining parts of themselves that they couldn’t see (their bottoms). It was hilarious to see the contortions they went into in order to inspect their nether regions.

Ritterhof courtyard, contemporary photo
Ritterhof courtyard, contemporary photo