Originally published in "Engineering Designer", the magazine of the Institution of Engineering Designers.
For ten years, from 1970, I worked for British Rail on the Advanced Passenger Train at the Railway Technical Centre in Derby. The programme had begun in the early 1960s with a research project to speed up 4-wheel freight wagons. Part of the legacy of the Beeching cutbacks of the 1950s was the consolidation of “fast” and “slow” lines into a single track over much of the system. The result was that freight trains, particularly the ubiquitous 4-wheel wagons of that time, limited to 72 km/h (45 mph), were getting in the way of the push to speed up passenger services. The speed limit was imposed because at around 90 km/h (55 mph) the 4-wheel suspension began to “hunt”, oscillating from side to side with the wheel flanges striking sparks from the rails. This instability often led to derailment.
A small group of researchers were recruited from the aircraft industry, where they had worked on a similar instability of light aircraft nose wheels. They began by looking at the existing theories of wheel on rail stability. If you take a pair of rigidly connected conical wheels (a wheelset) and run them down parallel rails, they will tend to correct any misalignment or displacement by steering towards and across the centre line before steering back again from the other side. (You can try this with a pair of plastic cups glued together running down two long rulers.) At normal running speeds, friction in the suspension removes the energy of these oscillations leaving the wheelset running stably on straight or gently curved track. The wheel flanges do not normally make contact and are only there as a safety measure and for any sudden changes at points and crossings. However, beyond a critical speed, the oscillation energy exceeds that which can be absorbed by the suspension and hunting begins.
To see this from the beginning, the new research team visited a workshop to see wheelsets being produced. Wheels are cast individually and press fitted to the axles. The final process was to put the complete wheelset in a lathe, where both treads were turned to a conical taper (usually 1 in 20). The finished wheelset was lifted from the lathe, placed on some rails and rolled away. They went to examine the newly turned treads and found they each had a shallow groove down the centre. The weight of just the wheelset acting on the point contacts between wheel and rail had deformed the steel surface. A number of wheelsets were given identifying marks and followed as they were put under wagons and into service.
The groove deepened and spread across the tread and, after only some 100 km (60 miles), became a stable and consistent slightly hollow shape. The steel surface had become work-hardened by the rolling process. The tread shape and the wheel diameters were carefully measured as mileage built up. After the initial deformation, no further diameter or shape changes were measurable as the distance travelled built up to 16 000 km (10 000 miles). At this point, in accordance with standard railway practice, the hollow tread was inspected against a 1 in 20 taper and pronounced “worn”. The wheelsets were taken away to be re-turned to their original taper. Clearly the inspection was not measuring real wear but only the results of the original deformation. All trains run on “worn” wheels and the model of coned wheel steering was inappropriate.
The research team designed a tread with a double cone, so that it was slightly convex. Wheels machined to this shape were run under test trains and their treads deformed into a less hollow work-hardened shape, which was carefully measured to use as the basis of new dynamic equations for the analysis of railway suspensions. After several prototypes were tested, a new suspension for 4-wheel wagons, with leaf springs and hydraulic dampers, was developed and given extensive testing. Instrumented wagons were attached to the rear of passenger trains running at speeds up to 145 km/h (90 mph), with no instability. Some test wagons were run without their wheels being re-turned and reached 260 000 km (165 000 miles), before there was any measurable wear. This work has led to a new faster fleet of freight vehicles.
Early in the research it was noted that if 4-wheel freight wagons could be made to run faster, then so could 4-wheel bogies under passenger trains, and the Advanced Passenger Train was born. Analysis showed that a train, with appropriate suspension characteristics, should be able to run safely at least 50% faster on straight track and 40% faster on curves than current passenger rolling stock for any give speed restriction without any track modification. Since track curves were not banked, or “canted”, for these speeds, it would be necessary to tilt the trains for passenger comfort. Existing research papers indicated that the tilt should be enough to completely compensate for this “cant deficiency”. With a top speed of 250 kph (155 mph), this would require a maximum tilt of 9°. Since the train had to stay within the restrictive BR loading gauge, the amount of tilt constrained the maximum cross-section of the train.
At that time, the Japanese and French TGV solution, of putting trains with a high power to weight ratio on a purpose built track, was not seen as viable in the UK, since no route carried enough traffic to warrant the costs. The option of running trains faster with minimal track modification raised the possibility of increasing speed and capacity on all routes. The current discussion of HS2 uses similar arguments to reach the opposite conclusion.
A short experimental train, APT-E, was designed utilizing gas-turbine engines then being developed by British Leyland (BL) for trucks and buses. Ten were used as generator sets, eight for traction and two for other power services in the two end power cars. Standard traction motors powered the end bogies with fully articulated trailer suspensions under each pair of vehicle ends. The train was articulated to reduce weight by reducing the number of bogies. In addition, a 2-car dummy train with the same frame structure as the power cars and similar but unpowered suspensions was used for towing trials.
Initial tests of the powered train were delayed and disrupted by union action, but the 2-car towed tests on the Old Dalby test track south of Nottingham quickly revealed some problems.
Rate sensitive dampers, new to the railways had to be used to give a stiff reaction to the dynamic vibrations, but allow larger slow movement when entering curves or passing over points and crossings. The suspension connections relied on the stiffness of the body structures to react the vibration loadings.
However, there were vibration modes within the bodies and, because of the fully articulated configuration, even whole train vibrations, which were excited by the suspension frequencies. The powered train was modified to test these findings when it did run (breaking the UK rail speed record). A new set of parameters were evolved for the next stage, pre-production prototypes to be used in service. By this time BL had withdrawn from gas-turbine work and the APT-Ps were designed as electric trains for the West Coast Mainline from London to Glasgow.
To avoid the interactions between the suspension frequencies and the train vibrations, the prototypes were semi-articulated, with more flexibility between cars. In addition, all vehicles were required to have much stiffer body structures. It was also decided that the train would be the first on British Rail to meet the full international (UIC) strength standards. A new form of structure using aluminium extrusions was developed for the trailer cars. The extrusions included features such as air ducts and seat rails allowing the vehicle shells to be produced more quickly and cheaply than the steel built vehicles then common.
A new weld preparation and clamp point was developed as part of the extrusions, which enabled multiple sections to be welded simultaneously producing a complete roof or floor section in one operation. This was later used extensively by overseas manufacturers but not by BREL, then the main UK rail vehicle builders.
New hydro-kinetic brake systems were developed to reduce the speed to a level where normal friction tread brakes could take over. We were required to stop from full speed in the same distance as a current train from 145 kph (90 mph). Our new brakes stopped the APT smoothly in 70% of that distance. At that time, in normal service the then new French TGVs simply cut power and coasted from top speed to a level where they could use friction brakes without excessive wear. If they had had to use their friction brakes for an emergency stop from full speed, the train would have been taken out of service for a complete brake pad replacement. Today’s TGVs have high performance brakes, which can be used at full speed.
The relatively light aluminium APT trailer cars had the potential to achieve the required strength and stiffness without much difficulty. The more heavily loaded power cars were far less uniform in cross-section making an extruded structure less appropriate and steel became the material of choice. I managed the team designing the power car structures and worked with BREL to co-ordinate the build of six cars for three prototype trains. The maximum permitted load on the track is 17 tonnes per wheelset, which limited the total vehicle weight to 68 tonnes. With more than 30 tonnes of equipment to carry and two 12 tonne bogies, the power car shell could not exceed 13½ tonnes. Since this included the heavy structure round the couplers and other strong points, the main body shell, over 20m long, had to weigh less than 8½ tonnes. It needed to meet the UIC end load test requirements of 2000 kN compression and 1500 kN tension at coupler level and various smaller loads at other positions. To meet the suspension requirements, it also had to have minimum body vibration frequencies of 12.5 Hz laterally and 10 Hz vertically, far stiffer than any similar power car or locomotive then in service.
Each power car had four body-mounted, 1 MW, traction motors, each driving a wheelset via a gearbox on the body, a large Cardan shaft through the floor, a second gearbox on the bogie frame and a Quill drive. (A Quill drive is a tube around the axle connecting a gearbox mounted on the bogie frame to the back of one wheel via flexible mounts. This minimizes the wheelset mass.) Since the bodies were to tilt to match the trailer cars, and the pantographs had to stay horizontal to pick up the overhead power, it was necessary to make space for an “anti-tilt” mechanism through the body using long rods connected to the bogies. In addition, large air intakes and outlets for cooling air, doors and access panels all left the structural shell full of holes and space constraints posing challenging structural design problems.
After some four years from blank paper to complete vehicle, the first fully fitted power car was put into the laboratory for structural testing. It passed all the UIC load tests and had a lateral frequency of 12.6 Hz and a vertical frequency of 10 Hz.
The trailer car next to the power cars had a small guard’s compartment. BR insisted that we must carry all the equipment carried on current trains. The list included a first-aid box, a stretcher, track circuit clips, and a large felling axe. We asked, “Why the felling axe?” “That’s for chopping into wooden coaches in an accident.” “But there aren’t any wooden coaches anymore!” BR withdrew over 4000 felling axes from trains across the country. I’ve got one in my shed.
The tilt system on each car consisted of a complex hydraulic pack controlling cylinders on the bogies. These had been through a long development process to achieve the rate of tilt needed to react to the train passing through a set of curves at 250 km/h. The sensors measuring the side forces for each vehicle had to be put on the vehicle ahead to give sufficient time to react. As a result the early packs fitted for the first APT-P test runs had been modified several times and become less reliable. Replacement packs to correct the problem were being built when the BR publicity machine decided to have a run for the Press. On the day of the run, free drinks were available in the buffet car. Which car had tilt problems? You guessed it. The publicity was not good. The three trains went into service, with the new tilt packs and no publicity, a few months later. They were in service for four years with few problems. When other traffic allowed, they cut the London-Glasgow journey to 3½ hours.
There were a few passengers who complained of motion sickness. The original assumptions of the need to eliminate side forces were examined again. It was found that the papers suggesting it were all based on one crude subjective survey done at low speed on a Welsh mountain railway. Motion sickness usually occurs when the body senses a motion which doesn’t match the motion deduced from the visual scene. The problem came from the lack of bodily motion sensations, eliminated by the tilt system, coupled with the sight of the horizon moving up and down. After some testing, it was decided to tune the tilt system to eliminate only half of the side forces. This produced sufficient compensation without significantly reducing passenger comfort. If we had known we only needed to tilt 4½°, the train could have had a larger cross-section and many other challenges would have been easier. Interestingly, now the Class 390 “Pendolino” trains have begun to reach their full speed on the faster sections north of Crewe, there have been some complaints of motion sickness.
The APT started as a research project, but there had been growing resistance to it within the BR establishment once it began to produce real trains with superior performance. Work on the full production “service” trains, APT-S, showed that they could be produced for less cost and would use no more power than current trains. Despite the success of the prototypes in service, the railway regions were discouraged from ordering production trains, so eventually the work ground to a halt and the project was cancelled after some 20 years of development.
The APT programme had developed from an investigation of the interaction between wheel and rail. It had produced a new concept of a train able to travel safely much faster than previous trains on existing track. The observations on wheel tread deformation were published in 1970, but over forty years later, the UK railways still re-turn their “worn” wheels unnecessarily, costing many millions of pounds every year. There was significant interest from other countries. If the APT had gone into production, we could have had a major export success and still have a strong train building industry.
Today, the experimental train, and a few prototype cars, are in the care of the National Railway Museum at their Shildon site in county Durham. Some other prototype vehicles form a short train at the Crewe Heritage Centre, just north of Crewe station. Only one small test piece of the production power car structure was ever made to validate a detail design change. I was given a piece of it, when I left BR in 1980. The programme was finally cancelled in 1984.
Lessons to be learned
The main lesson from the APT experience is that major well established industries can come to believe that they know all there is to know about their underlying technology. Any changes will be small and incremental. They stop questioning the assumptions which underlie their work and refuse to believe that major change is possible. The project failed because it could not convince the BR establishment that there was another way. Since then, the superiority of similar work accepted in other countries came to dominate the UK rail industry and eventually wiped out most of our main line rolling stock industry. The APT programme grew out of the ability to design better suspension systems derived from developments in other industries. It used aluminium extrusions for the trailer cars, later adopted elsewhere but not by BR. It showed what was possible, but its findings were simply ignored.