The Coral Fringe

Indian contrasts - 1

2010-08-26T11:28:00.000-07:00

Just got back from a two week trip to India. I saw the following mind boggling juxtaposition of scenes on same day in different locations in Chennai:

According to cardekho.com, the Porsche Cayenne in the first photo costs more than 75 million Rupees. I have no idea what a cycle rickshaw costs but its safe to say that its a lot lower. Cycle rickshaws have been plying the streets of Indian cities mostly unchanged since the 1940s.

In terms of conspicuous consumption and sheer unsuitability to Indian conditions, its hard to beat the Porsche Cayenne - despite its incredible bulk, its just a four seater. Its enormous width would make it just about impossible to maneuver in most of the narrow streets of Chennai. Even the low-end Cayenne is grossly overpowered for most Indian roads. Finally, given the chaotic traffic, its only a matter of time before it gets dinged by some uninsured rattletrap...

Sensors for model railroads - 2 - reed switches

2010-01-20T06:34:00.001-08:00

Reed switches are a very simple and effective way to implement point sensors (see my previous post that explains my sensor taxonomy). I have experimented with several kinds of reed switch and generally find them to be reliable, robust and easy to install. Reed switches are installed between the rails while small but powerful rare-earth magnets are mounted on the locomotives. When a magnet equipped locomotive passes over the reed switch, the magnet causes the reeds to make the contact. Since a reed switch requires no separate power, you can directly connect the switch to a sensor encoder or some other electronic gadget. Sensor encoders are devices that generate messages for a computer based on sensor activity. I will discuss them in detail later in this series.

The above photo illustrates two different kinds of reed sensors that have I tried on my layout. Click on the photo for a bigger version.

On the left is a traditional inexpensive reed switch which I bought from All Electronics. I found that it was necessary to mount this sensor longitudinally, that is to say, along the length of the track. I drilled two holes and ran wires up to the reed switch and soldered them to the leads of the reed switch. The sensitivity of this kind of reed switch does not permit any other method of installation. I did try to mount this reed switch parallel to the ties (its actually small enough to permit this) but found that it just did not work.

On the right side of the photo, a different kind of reed switch called a cylindrical reed sensor can be seen peeping out from between the ties. This little gizmo, which is made by a German company called Meder Electronic, is really just a reed switch but is packaged in a way so as to permit vertical mounting. It is available from Mouser Electronics.

Now turning to the topic of magnets: I bought a few small disc shaped neodymium magnets from a small company called KJ Magnetics and mounted them on my locomotive as shown in the photo above in a very simple way - I simply stuck them to the steel screw that holds the knuckle coupler in place! This particular locomotive is a Proto 2000 U30B but the same technique also works fine on Athearn and Bachmann locomotives. The only requirement is the presence of a steel screw for the coupler. I also have a Trix German locomotive that uses European style couplers which do not have such a mounting point. For this locomotive, I had no choice but to super-glue the magnet to the chassis.

Now, although this may not sound like a secure way to mount the magnets, it actually works very well in practice. You have to play around with neodymium magnets to appreciate how strong they are. I have had a few derailments on locomotives equipped with magnets mounted this way but the magnets have never been displaced.

Both kinds of reed switches work well but there are some tradeoffs to consider in making a choice:

Traditional reed switches: The main advantage of traditional reed switches is that they are very inexpensive - only 50 cents apiece. However, they have two drawbacks. Firstly, they must be installed in a rather obviously visible manner. Secondly, due to a quirk in the physics of magnetic fields, a traditional reed switch actually delivers two pulses when a train passes over it. This phenomenon is explained the following illustration taken from this document from the Meder Electronic web site.

Cylindrical reed sensors: This kind of reed switch has the advantages of being far more sensitive than conventional reed switches and is also very small. In fact, its barely noticeable when installed. Furthermore, the small size makes it very easy to install. All you need to do is to drill a single 1/8" diameter hole. The reed sensor fits snugly in such a hole and needs no glue to hold it in place. Even though I am a rather ham-handed carpenter, I found it easy to drill a hole between the ties on code 83 HO track. The only drawback to reed sensors is that they are considerably more expensive. The particular model that I bought (MK20/1-B-100W) is between five and six dollars depending on quantities which is almost 10 times the cost of ordinary reed switches.

After weighing the merits of the two kinds of reed switches, I decided to go for the cylindrical reed sensor mainly for reasons of ease of installation. I find that I am able to install a sensor in just ten minutes.

Sensors for model railroads - 1

2010-01-15T02:18:00.000-08:00

I had originally planned to start writing about the construction of my layout but I am going to jump ahead a little and talk about sensors, a subject that has been the focus of much of my recent work on my layout. I'll get back to layout construction in subsequent posts.

The basic purpose of a sensor, in the context of model railroading, is to detect whether or not there is a train on a particular piece of track. Now, there are some new kinds of sensors that can additionally identify which specific train. However, they have not gained widespread popularity at the current time. In any case, as I will explain later in this series of postings, its rarely necessary to be able to identify individual trains. You can generally get by just fine with a simple yes/no status from the sensor. When a train is detected by a sensor, there are two ways to use the information generated by the sensor:

Send a notification message to a computer: There are a number of model railroad specific programs that can handle such notifications including JMRI, Railroad & Co. Train Controller, RocRail and Win-Digipet.
Directly control some electronic gadget such as signals, grade crossing gates, lights, sound effects. In addition, some DCC turnout controllers can be configured to directly select routes based on sensor notifications.

The main model railroad applications for sensors are listed below:

CTC (Centralized Traffic Control) installations require sensors to help the dispatcher (or line controller) determine whether a particular block of track is occupied by a train. Track circuits are a commonly used kind of sensor on prototype railroads that implement CTC. A number of model railroaders have implemented CTC on their layouts in a very sophisticated manner.
Computer control of the layout: If you wish to have a computer running trains on your layout, you will need sensors to allow the computer to keep track of the location of trains accurately. In fact, you will need a lot of them. My friend Dale Schultz has described this approach to automation very nicely nicely on his web site. As I have indicated previously on this blog, my own interests lie squarely in this direction.

Control automatic block signals: Many prototype railroads use automatic block signal systems on sections with heavy traffic. Among the model manufacturers Atlas offers a line of automatic block signals.
Specific layout function control: Even those model railroaders who prefer manual of their trains, often use sensors for a number of specific functions such as:

Simplifying the operation of staging areas, especially hidden ones. It is useful for operators to be able to quickly determine which staging tracks are occupied at the moment.
Operating grade crossings: Trigger flashing of lights, wig wags or crossing arms when a train approaches a grade (i.e. level) crossing.

Types of sensors

Broadly speaking, there are two kinds of sensors (note that this is my own terminology. I have not seen this used anywhere else but I think it serves a useful purpose so please bear with me):

Point sensors: This class of sensors are used to detect the presence of a train at a particular point in the layout. If you are interested in layout automation, the use of point sensors is, more or less, mandatory since the computer needs to know exactly where a train has reached. However, they can also be used for some of the other applications listed above. A number of technologies have been used to implement point sensors including infrared, reed switches, mechanically trigged switches and (recently) RFID.
Span sensors: This class of sensors are used to detect the presence of a train anywhere on a section of track. Span sensors are a natural fit for CTC and automatic block signals. The main technology for implementing span sensors is by detecting current drawn by the electric motors in locomotives and from special resistor wheel sets in other rolling stock. In the case of three rail systems like Märklin, it is also possible to take advantage of the fact that wheels are electrically conductive and create a span sensor by electrically isolating one of the outer rails in a block.

I will present the pros and cons for the various sensor types and their technologies in subsequent posts.

How I designed my layout - an overview

2010-01-01T09:47:00.000-08:00

Lets cut to the happy ending :-) Here is the final design that I came up with:

But how did I get here? In the course of some ten odd postings on this blog, I have described the process I followed in coming up with this design. It therefore seems like a good moment to provide an overall narrative that ties it all together. Some background material first:

My model railroading background:I have been a model railroader for a long time. I describe some of my past activities in this hobby in this posting.

Capturing complexity: Some thoughts about why I enjoy the hobby of model railroading - its all about selectively capturing the complexity of the real world in miniature.

Operating styles: An analysis of the distinct styles of model train layout operation that I have encountered. I find it interesting that operating styles vary considerably between the US, the UK and Germany.

Inspirations for my train layout: My interest in railroads both model and prototype arises from long train journeys taken while growing up in India.

I started work on designing my current layout in earnest in October 2007 after I visited my friend Dale Schultz and saw his layout in person.

Dale has built a truly innovative layout that featuring a very high level of craftsmanship and full computer control. It is remarkable that Dale has written his control software entirely by himself from scratch. Motivated by what I had seen, I designed my layout in a matter of just six weeks. I kept a diary of my progress which was the basis for the next set of articles:

An account of the constraints that guided the design.

Lessons learned from previous layouts - that have been incorporated into the current layout.

Some explorations of the folded dog-bone configuration. I discovered that it just would not work in the space available.

Some additional single track designs that I explored - it was becoming a wild goose chase at this point. Nothing seemed to be working.

Discovering double track: I finally figured out that single track designs would not work for me. Once I realized this, I was able to make rapid progress towards the final design.

An analysis of the final design and why I think it works for me.

Having reached a design that I was happy with, I started construction of the layout in November 2007 - and its still under way as I write this in in January 2010. Here is a peek at the layout as it currently stands:

In the next few months, I will be describing the construction and control systems - stay tuned!

Designing my train layout - 6 - Why it works for me

2009-12-27T07:16:00.000-08:00

In my last posting, I explained the process that led to my final design. In this posting I'll try to explain why the design works for me.

The first aspect of this layout that I feel happy about is that of staging. Let me digress a bit and show you an old Marklin layout that I built back in 2003:

This layout was also essentially an oval but the key difference compared to the current design is that I had placed some bare staging tracks at the back of the layout separated by a view blocker (1/4" plywood). The intent was that viewers would stand at the front of the layout and watch trains stop at or pass through the station. While operating this layout, I realized that I could have made a second station instead of the bare staging tracks.

In the current layout, one possible style of operation would be to simulate time-table driven movements through a station. In this mode of operation, each station can act as the staging area for the other station.

In the diagram shown above, each green arrow represents a train. As you can see, it is possible to have as many as eight trains staged around station 1 in order to run a timetable sequence of trains passing through station 2. Of course, in a similar manner, I can use the station 2 as staging for running traffic through station 1. A word about the term "timetable": Unlike the US where most trains on freight railroads are extras, in India, Most trains are actually listed on the timetable.

In addition to timetable driven operations, I also plan to use the layout as two bidirectionally signaled single lines. This style of operation is often used in places where traffic density is higher than can be supported by dedicated double lines. This happens near big cities and on hill sections.

Finally, I have made provisions for four storage sidings for trains not used in the main sequence of operation. Since I tend to be very careful in my rolling stock purchases, I don't thing I will exceed this relatively modest amount of storage space for a while.

Since there is a considerable amount of available space in the center of the layout, I could add storage sidings there in the future if I need the space.

So this concludes the postings on the design of the layout. Next, I will cover the construction as well as some tips and tricks that I have learned in the process.

Designing my train layout - 5 - double track

2009-12-19T13:12:00.000-08:00

While leafing through the Armstrong track planning book, I came across this diagram that changed my thinking drastically. The thing that caught my eye was this text from one of the captions: "In model practice, where mainline length is severely limited, two and three track railroading may actually surpass single-track operation in variety"."

Most of the trunk lines in India are double track (with the Chennai-Mumbai section being the only important route with a lot of single track). Upon reflection, I realized that what I wanted was a lot of parallel action. Furthermore, the station configurations could be made more realistic. I then decided to give double track a try. I started by taking a fresh look at the "water wings" configuration from a double track perspective:

The results were surprisingly good. I was able to fit two stations that each had a single passing loop shared between the two tracks. I have, in fact, seen this layout in some small stations on double line sections in India:

Although promising, there were two issues:

Lack of visual separation between the stations. The two stations are physically close together even though there is a nice long run between them. The only obvious solution was an oval. Now, I had initially rejected ovals because I though they looked boring. However, it was clear that a good old oval is what would work best for me.

I came to the conclusion that I wanted the stations to have different configurations. Specifically, I wanted a conventional double station as shown in the next diagram. Although configuration B would be more prototypical and offer greater operational flexibility, I was willing to settle for configuration A. The point is - it looks like a regular double track station.

At this point, I showed the last design to my friend Dale Schultz who has built a large Marklin layout. Like myself, Dale is interested in using computers to automate train operations. He is a very sharp thinker whose judgement I greatly respect. He told me I was on the right track but suggested that I reduce the number of sharp curves. The next design I came up with was this:

This was now getting interesting. Notice that the lower station remains curved. This was intentional - I had come to feel that at least one of the stations ought to be curved for aesthetic reasons. The two problems were that the passing sidings in the upper station were too short and there was only one cross over and it was too far removed from the nearest station. The next step was to integrate the crossover with the upper station, thus:

This was definitely an improvement. There was a pleasing curve to the upper station and both stations had passing sidings of acceptable length. Dale suggested investigating curved turnouts. I looked through the Walthers catalog and found that the Shinohara #7 curved turnout would be perfect for my needs. This turnout has radii of 24 and 2.8.5 for the diverging and normal routes. This is the next design I did incorporating curved turnouts for the crossovers.

I felt that I was closing on the final design. The last tweak to the design was to confine all the curved turnouts to one side of the layout - mainly for aesthetic reasons.

And that did the trick. I was now fully satisfied with this design - The full process had taken about seven weeks but it was time well spent. A 3D view is shown below that includes the benchwork. I am deliberately not using the space in the center of the layout for the time being. I may eventually come up with some use for the space but it will be left vacant for the time being.

In my next posting, I will explain the operational possibilities of this layout.

Designing my train layout - 4 - more single track designs

2009-12-17T05:53:00.000-08:00

With the "folded dogbone" eliminated from consideration, I took a look at some alternatives, starting with the "water wing" configuration. The following is an initial sketch to explore this configuration and it seemed promising:

After some thought, I was able to come up with the following design. As you can see, I was actually able to fit three stations into this design. I should note that the third station (on the straight section at the top of the layout) was really more in the nature of a what-if exploration. I was still thinking in terms of two stations, the main ones being on the ends of the two lobes. The two main stations are well separated and the passing loops are surprisingly long.

The problem with this design was that the stations could simply not accommodate more than one passing loop due to the curves involved. also, there were no good locations for storage sidings of reasonable length. My other concern was the S-curves at the entrance to the passing loops. In hindsight, I realize that I ought to have looked into the possibilities of using curved turnouts. In any case, this configuration just did not look right to me.

At this point, I decided to relax some of my requirements and explore more exotic configurations. I present a few of these below

None of them seemed to be very appealing. At this point, I decided to take a few days off from 3rd Planit and re-read John Armstrong's classic treatise on layout design.

As I shall explain in my next posting, this break turned out to be a good idea. and I was able to break my mental log jam.

Designing my train layout - 3 - Folded dog bone

2009-12-10T05:03:00.000-08:00

Building on my experience with my previous layout, I decided to start by investigating single track main line designs. Basically, what I wanted was not much more than this schematic:

In other words, an oval with two stations. I wanted a reasonable amount of separation between the two stations. I wanted the train to have at least a short run between the stations. This schematic may seem very simple but it should be noted that complex behavior can arise from deceptively simple systems. Consider for example the difference between a simple and double pendulum. While a simple pendulum can be fully described by equations for simple harmonic motion, the tip of a double pendulum can move chaotically. In an analogous manner, even with just two stations, it is possible to create complex patterns of train movements. Having decided on a two station configuration, the next step was to see how long a main line I could fit within the space available available.

I prefer to do my design work directly on the computer using the 3rd PlanIt layout CAD software from trackplanning.com. I have been using this package for many years, initially using to design sectional Märklin layouts. I think that 3rd PlanIt is a very powerful package and there is indeed quite a bit of a learning curve. However, once you get the hang of it, 3rd PlanIt just seems to get out of the way and lets you explore designs very quickly.

The first general shape I investigated was a folded dog-bone. I have seen this shape used in many track plans and tried to get it to fit in the space available.

The nice thing about this shape is that there are good locations for locating the stations allowing decent separation between them. However, the main problem is that the curves at the end of the "bulbs" necessarily have to be very tight. Nothing more than 20 inch radius curves can fit in the space available. another problem is that access to the interior of the layout would be problematic.

In order to use curves of minimum 24 inches radius, I discovered that there were two options, neither of which were attractive.

1. Overlap the bulbs: This allows two large radius bulbs to be accommodated. However, as previously mentioned, I was not keen on a multi-layer layout. A second problem is that by having the bulbs on different levels, the main line and the stations must necessarily be on an incline.

2. Widen the layout: Widening the layout allowed the use of 24 inch radius curves but the increased width unacceptably cut into walkway around the layout. One more problem that I discovered was that the station passing loops could not be made very long. I don't really want very long trains but this seemed a bit ridiculous - barely 41 inches for the station on the top.

Clearly a folded dogbone was not going to work for me. In my next posting, I will describe some exotic variants of the folded dogbone that I tried.

Designing my train layout - 2

2009-11-25T06:03:00.001-08:00

One more digression before I begin describing the design process for my current layout. Every layout that I have built so far (five in all) has incorporated lessons learned from previous layouts and the current one is no exception. A schematic diagram of my old layout is included below.

When thinking about layouts from an operational viewpoint, I find it useful to think in terms of schematic diagrams (which are similar to electronic circuit diagrams) rather than actual track diagrams.A video of this layout in operation under computer control is included below:

The focus of this layout was a station on a single line (somewhere in Germany near the Swiss border since all the rolling stock was either Swiss or German!). There were two storage yards and trains would proceed back and forth between the yards through the station. From an operational view point, this allowed me to model meets and overtakes.

Some of the problems (and lessons learned are describe below):

Since the layout featured only a single station, there was really only one viewing position. After a while, this became a bit boring since watching the storage loops was not much fun. This led to the requirement that the next layout would have at least two stations.
It became very tedious to sort out problems in the lower level storage yard since access was so limited. This in turn was due to the fact that I did not have the room for a helix to connect the levels. I therefore decided to avoid multiple levels in the next layout.
Unused freight yard: I created a freight yard for the station because I wanted to combine some switching action along with mainline train movement. The intention was that freight trains would pick up and drop off cars at the station. It was hard enough to get reliable automatic operations on the main line so this little freight yard was never utilized. Automating a layout is enough of a challenge and in any case, I did not really enjoy switching very much anyway.
Inadequate storage for unused trains: This is a perennial problem with railway modelers. There was simply no place to add any extra storage sidings on the old layout. I therefore decided to incorporate additional storage sidings in the next layout.
Automating a layout requires a large number of sensors since the computer needs a good understanding of the location of a train. There were not enough sensors on the old layout and the sensors were not well located. These problems arose since I had not really designed my layout with automation in mind. It became clear that sensor locations have to be considered before designing the layout.

All of these lessons have been incorporated in the current layout to various degrees.

Designing my train layout - 1

2009-11-20T06:42:00.000-08:00

As a child I grew up with model trains. I was given my first train set when I was just two years old. Later, at the age of eight, my parents bought me a great Triang-Hornby train set when passing through England. Modeling activities stopped in school and college. While attending graduate school at Syracuse University, I bought a small loop of N-gauge track and used to run a train around my computer, much to the amusement of fellow students.

When I resumed active railway modeling in 1992, I started with Märklin, a well known German model train company. One of my best friends had Märklin train set and I always wanted those trains. Photos of my old Märklin layouts can be seen here and here. For a variety of reasons (chronicled in more detail in this article on my web site), I decided on a new direction.

I finally dismantled my Märklin layout towards the end of 2007 and cleared the train area in my basement for a new layout. As such, I had a clean sheet of paper to work with. My layout design was constrained by a few factors:

Space: My old layout was just 4 feet by 10 feet. I felt that there was some scope for expansion. After careful measurement and negotiations with my wife, I determined that the actual space available for the layout was was 14 feet by 7 feet. However, there is a furnace close to the layout and I need to ensure that there is enough room around the furnace. Hence, I lost a triangular a nibble from this space.
Now, 98 square feet is not much compared to the large spaces available to the typical American modeler. However, I felt that even this amount of space would be a bit of a challenge. Larger layouts are more costly both to build and to maintain. Its all too easy to bite off more than you can chew in this hobby! The shape of the available space is shown below. The available space is the area in white with thick borders. The diagonal section on the left side of the diagram is caused by the need to leave enough room around the furnace. Another issue is that this has to be a free-standing layout since I need access to things stored around the room.
Operating style: This played a very important role in shaping the design of the layout. The one clear requirement in my mind for this layout is that it must support fully automatic computerized operation. Consequently two sub-requirements became clear:
a. A continuous run is absolutely essential. A point-to-point layout is simply not suitable for automated operation given the need to switch locomotives around their trains.
b. There need not be any provision for local switching in industries and freight yards as is common in most American layouts.
Carpentry skills: I am not much of a carpenter - I just don't have a very steady hand and am a bit impatient. As a consequence, I wanted the benchwork to be as simple as possible. A major consequence of this constraint is that I decided to go in for a single level layout. Apart from being much easier to build, its much easier to debug computer control programs if the trains are visible at all times.
Scale is HO (1:87.1): This is because I have a lot of European outline rolling stock in HO that I would like to run on this layout. Also, a large variety of HO scale equipment is available from manufacturers around the world.
Minimum radius for the mainline must be at least 24 inches: This would allow the operation of most modern equipment although even this radius is too small for longer cars like the Amtrak Superliner car which is 85 feet long for example (nearly a foot in HO scale). There will be a fair bit of overhang for the longer cars on 24 inch curves but at least, they don't derail.

After spending a week or so considering these constraints, I updated my copy of 3rd PlanIt layout design software and got to work. In the next posting in this series, I'll describe the actual design process.

Productivity

2009-08-26T05:23:00.000-07:00

A few years ago, I read an interesting little article by Kenneth Rogoff, the well known professor of economics at Harvard, called "Artificial Intelligence and Globalization" in which he says:

But I wonder whether, even within the next few decades, another factor will influence our work lives even more: the exponential rise of applications of artificial intelligence.

At the time I read this, I found it funny. I have worked for many years in AI and related fields. We have an inside joke: "if it works, its not AI". Now I am less sure.

I have recently started watching an interesting show called How Its Made on the Science channel. As it's name suggests, this show is all about manufacturing technologies. I was struck by the high level of automation in many of the manufacturing processes that they present. Clearly the level of productivity must be high in these automated factories.

This then led to consider other examples of highly productive organizations:

1. Craigslist: Serving millions of daily requests with just 30 employees.

2. The UPS Worldport sorting system in Louisville, Kentucky: check out this this amazing video about how they handle vast numbers of packages with just a few employees to keep watch.

3. Google's data centers: Consider the vast number of machines in one of their data centers supervised by a handful of employees.

4. Plentyoffish: This is a very popular match making web site that was created and run (for the most part) by just one person. Apparently, the founder just hired the first employee recently (only after the company had grown amazingly large).

I don't think I am alone in beginning to wonder where this will lead us. I just came across an article in the Economist web site in which he author makes some interesting points. In a nutshell, the author says that jobs are ever more specialized requiring a lot more training. And even when trained, the skills become obsolete ever more quickly.

I would add a couple of points to the Economist blog posting:

As one gets older, retraining gets harder. This is partly because distractions (house children, finances etc) increase as one gets older. Another factor is that many people are resistant to learning things from people younger than themselves.
Another source of concern is that many of the new job opportunities are in the discretionary side of the economy. I think that jobs based on discretionary expenditure are much more vulnerable to downturns.

New ideas for entrepreneurs in India

2009-05-18T08:32:00.000-07:00

I recently came across instructions for building a DIY milling machine on the great Instructables web site. This fascinating web site has detailed instructions for building all kinds of interesting gadgets ranging from solar powered cars to do-it-yourself CNC milling machines. I myself have successfully built an IR sensor for my model train layout based on plans I found on that site.
It occurs to me that with the tremendous reduction in cost of creating sophisiticated machine tools, it would make a great deal of sense for some entrepreneur in India to create a machine shop on the cheap. Most of the cost in creating a physical object nowadays arises when creating a CAD model. However, there are large numbers of recent Indian engineering graduates trained in most popular CAD tools. All it takes is for some enterprising engineer to get things rolling.
Thirty years ago when I was in high school and college, there was a vibrant DIY culture in India. Due to high import duties and "licence raj", it was very hard to get consumer electronics at reasonable prices. To save a few bucks, many people used assemble their own radios and casette players.
In every big city there was a place you go to get electronics components. Lamington Road in Mumbai, Richie Street in Chennai and Lajpatrai market on Chandani chowk in Delhi. I have many happy memories of hours spent in Lajpatrai market haggling with shopkeepers over transistors, resistors and capacitors. We would design and etch our own printed circuit boards - getting heavily stained with ferric chloride in the process.

The inspirations for my train layout

2009-05-14T06:25:00.000-07:00

One of my main inspirations for my model train layout are my memories of train travel in India where I grew up. I have been on many memorable railway journeys in India but the ones that stand out in my mind were in the mid to late 1970s when I often travelled between New Delhi and Chennai (then called Madras) by the Tamil Nadu Express (Click on photo at left to see the original photo taken by by Mick Pope in 1977 on the IRFCA web site).

Newly introduced in 1976, this train had a remarkable average speed of almost 75 Km/h. While this may be low by European or Japanese standards, it was a very big deal 30 years ago in India. At that time, the only trains faster than the Tamil Nadu Express were the Rajdhani Express trains that ran, respectively, between New Delhi and Mumbai and Calcutta. The top speed of the train was 110 km/h which it touched on many sections. My brother and I used to time the trains with the help of a stopwatch and pocket calculator.

Apart from the high speed, there were several other interesting features on this train. The train had five classes of accomodation: air conditioned first class, air conditioned sleeper (one of the first in India), air conditioned chair car (yes, you had sit for 29 hours!), regular first class and regular sleeper. The chair cars had closed circuit TVs on which movies were shown. The train even featured a small on-board book store that sold magazines and novels.

My eventual (very long term!) goal is to model Indian Railways and the TN Express (as of 1977) in particular. Of course, there are no ready-to-run models of Indian trains - which means a lot of scratch building. To be sure, at the moment, I have neither the time nor the skill to scratch build rolling stock. Consequently, to get things started, I will be running American models. However, given that there is a fair bit of American derived running in India, this is not too much of a stretch.

The scenery, however, will definitely have a strong Indian flavor. I would like to capture the look of the Coramandel coast. My area of interest is near Ennore, an industrial area north of Chennai. The main line to the North passes over a backwater of the sea on a bridge and is quite picturesque. The scenery will mostly consist of palm trees and scrub bushes. There will be a big oil refinery on the layout. All railway structures will have to be scratch built. This should not be too hard since railway structures in India are not too ornate.

Kevin Kelly on complexity

2009-05-08T07:48:00.001-07:00

Since I am on a complexity theme here it seems relevant to point out a new article The Arc of Complexity by Kevin Kelly in his Technium blog, about the things in the real world that we generally consider to be complex:

It is precisely that goldilocks state between predictable repeating crystalline order and messy chaotic randomness that we feel captures real complexity.

A railroad consists of a large number of independently controlled machines by agents (both humans and computers) that interact in seemingly infinite ways seems to meet this definition perfectly. The question about why some people like myself enjoy analyzing complex systems is outside the scope of this discussion :-)

A related topic: In my own work as a software researcher, I often find it useful to distinguish between what I call "good complexity" and "bad complexity".

An example of good complexity is an intricate algorithm like that for, say, maintaining AVL trees. This algorithm has a personal significance since it was the first really complex algorithm that I encountered in my study of computer science. It took me a while to figure out what was going on but ultimately I found it rewarding.

I encounter "bad" complexity all the time in my work in the form of arbitrary requirements that often seem to arise for business or political reasons. The main characteristic of having to deal with these kinds of complexities is that time spent circumventing them is simply not rewarding. My guess is that this is caused by the nagging feeling that the whole problem is artificial and avoidable. A great place to read about these kinds of problems is The Daily WTF and this recent article in particular epitomizes bad complexity.

A different kind of modelling

2009-05-05T07:06:00.000-07:00

Thanks to the Programming Reddit, I came across this fantastic 12 part article by Shamus Young about how he wrote a program to generate a night-time cityscape from first principles (check out the video below). His methodical, iterative approach to creating the visual model is very similar to the approach that I have seen taken by some of the best model railroaders.

This leads me to think that there is something common to all modelers, whether they are using physical materials (wood, plastic and metal) or computer code. There is the same desire to capture aspects of the real world and to show off to an admiring audience.

My favorite model railroading sites tend to be those where a master modeler takes you through the detailed steps in creating a model such as, for example, Craig Bisgeier's Housatonic RR. In my view the similarities between the respective sites is striking.

Operating styles

2009-04-30T05:18:00.000-07:00

There are many ways of capturing operational complexity of a railroad. I have encountered three distinct ways in which "serious" model railroaders operate their layouts:

Realistic freight operations: This approach tends to be standard practice with those who model North American railroads and is nicely described here. The goal is to replicate the way American freight railroads operate on a train layout - the movement of freight cars from one location to another in a plausible manner. For example, moving a coal car from a mine to a power plant.
Typically, this involves the use of waybills to track consignments. A lot of switching (i.e. shunting) is involved. Trains are assembled in yards from freight cars collected from private sidings and loading docks. Road engines are then attached to the assembled train which is then dispatched to the destination. Along the way, the train may drop off or pick up additional freight cars from line-side industries. At the terminal, trains are disassembled and freight cars forwarded to the proper recipients.

Intensive traffic: This goal of this approach, favored by many European modelers, is to simply keep a lot of trains moving on the layout, usually under computer control. This may not sound realistic to American readers but if you have had an opportunity to do some railfanning in Europe, this does actually make sense. It is likely to appeal the most to those who wish to be "railfans" on their own layouts.

In April 1995, while in Nice in southern France for an extended business trip, I spent an enjoyable afternoon railfanning at a small station on the main line between Marseilles and Ventimiglia on the Italian border.

Most of the traffic consisted of passenger trains with some freight trains mixed in along with them. There were both local and long-distance passenger trains including some TGVs. Some trains stopped at the station while others passed through without stopping. This was purely a passenger station but one freight did stop in a loop line to allow a passenger train to overtake it.

Modelers who adopt this approach typically have large hidden staging yards from which trains are dispatched by the computer according to some pattern. Meets and overtakes are staged at stations in the visible part of the layout.

This video of the Märklin layout at the 2009 Dortmund train show is representative of this approach.

Timetable driven operations: The modeler uses a timetable to control the operations of trains. The main prerequisite is, of course, a timetable. In America, where most trains are not timetabled, this approach is not likely to find much favor. In my experience, I have seen this style of operation used mostly by British modelers. This is because they usually model small branch line stations usually based on actual locations for which time tables are available. It is possible to stage realistic time table driven operations .

In my next posting, I will describe how operational considerations governed the design process for my own layout.

Capturing complexity

2009-04-25T06:42:00.000-07:00

As a child, I usually built "tail chaser" layouts - small ovals of track with trains running around at top speed :-). This does get boring after a while - nothing to hold your attention. After I built a tail chaser or two, after returning to model railroading, I realzed that something was missing.

After some introspection. I came to realize that my interest arises from my desire to capture some of the complexity of the real world in miniature. Not all complex things are equally fascinating, of course - somehow I don't feel motivated to spend weekends studying the tax code on for example!

What gets me interested are complex machines. Whenever I see a complex machine, I have the desire to take it apart and see what makes it work. Indeed, in my day job as a software researcher, I spend a lot of time trying to build and understand complex systems.

In one sense, a railroad can be viewed as a gloriously complicated machine with millions of moving parts. When I see a picture like the one above (north end of Vriddhachalam junction in India), my mind gets to work right away trying to identify the pieces - the track, the catenary, the locomotives, the freight cars. Having understood the components, the next step is to understand how everything is organized so it all runs smoothly - signalling, communications, operating procedures.

Even within the model railroading community, there are groups of modellers that focus on capturing different aspects of the real world.

Accurate scale models of structures and rolling stock: There is a significant group of modellers who aim to replicate every detail of some part of a railroad. Although there are such modelers everywhere, I find that this approach seems to be dominant in in Britain. Driven by constraints of space, and to a lesser extent money, the British approach is to build small jewel like layouts that capture a prototype down to the last detail. One of the masters of this approach is Chris Nevard. In particular, his layout Catcott Burtle, which is only 5 feet by 18 inches size is a real masterpiece.
Operational complexity: If space permits a larger layout, it becomes possible to capture some of the operational complexity of a real railroad. That is to say, trains moving on the layout with some apparent purpose. This is my main area of interest and much of the design of my layout has been designed with operations in mind. In subsequent postings, I will go into operational issues in greater detail.

My model railroading history

2009-04-22T11:35:00.000-07:00

I have been a model railroader for a long time. I was given my first train set when I was just two years old and have been playing with trains ever since. My parents bought me a big Triang-Hornby OO set in 1970 when we returned to India from a stay in America. It included a large oval of track, several turnouts and a big lot of freight cars and passenger coaches. Not much remains of this set except for a Britannia Pacific and a few coaches.

This train set gave my brother an myself a lot of enjoyment but like many modellers, high school and college prevented me from pursuing the hobby. When I was a graduate student at Syracuse University in the 1980s, I attended a model train show by chance at the State Fair grounds and got hooked again. Of course, as an impoverished graduate student, there was no way to build a proper layout but I did manage to get a small oval of N gauge track that used to run trains around my Sun Sparcstation monitor.

I resumed model railroading in a big way after I started work at IBM in 1992. On my birthday, with my wife's encouragement, I bought a small Märklin starter set and and some additional tracks to go along with it. Why Märklin? I had always coveted Märklin trains as a child - my best friend Jayashankar Srinivasan had a Märklin set. Although my Triang-Hornby set was very nice, I felt that Märklin had a complete system that seemed to offer a lot of exciting operational possibilities. In particular, I was fascinated by the possibility of controlling trains by actual operating signals.

Over the years, I built a number of Märklin layouts, culminating in a medium size C-track layout that I completed in 2006. I used an Uhlenbrock Intellibox to control my trains along with JMRI-based custom software that I wrote myself. A video of my trains is incuded below

By 2007 however, I was becoming increasingly unhappy with Märklin for a number of reasons. In particular, Märklin uses a proprietary non-standard three-rail system - there is a central rail that is used to power the locomotives. Consequently, Märklin is the sole supplier of tracks for this system. The geometries of the track system leave a lot to be desired in my opinion. I created a small switching layout to explore the 2-rail world. This little layout turned out much better than I had expected.

After much thought, I decided to switch to the two-rail system. I then dismantled my Märklin layout and sold off most of the track and rolling stock. I have started building a new layout based on American prototypes. I will be documenting the construction of this layout in this blog.

My personal blog

2009-04-21T05:42:00.000-07:00

This blog will complement my web site and my Twitter page.