Podnik

Federation of transport cooperatives of the Republic of Fluid

Design note · Southern operating cooperative with the Sederna Carriage Factory

A relief driver kept a clock in his pocket and reported himself for half a second. This note gives the drawings, the mass ladder and the braking work that answer him, set against the Japanese commuter practice he was trained in. It is a working paper of the drafting group and states no decision of the federation.

Drawing 1

The R8/A keeps the 56.20 m of the R8 over three cars and three door pairs a side. The Japanese reference is a three-car formation of 20 m cars with four door pairs a side, which is the arrangement the driver knew from Osaka.

Side elevations of the R8/A and of a three-car Japanese commuter reference at the same scale The R8/A measures 56.20 metres over three cars with three door pairs a side. The reference formation measures 60.00 metres over three cars of 20 metres with four door pairs a side. SED R8/A · 56.20 m · three cars 56.20 m over couplers Japanese commuter reference · 60.00 m · three cars 60.00 m over couplers, 20.00 m a car 10 m
Scale 1:200 at the drawn size. Door positions are indicative and the pantograph is omitted.

Drawing 2

The body is narrowed by 150 millimetres so the unit stays inside the gauge on the 110 m curves of the southern route at the approach speeds this proposal implies. The reference runs on 1,067 mm track, which we do not have.

End elevations of the R8/A and of the Japanese commuter reference The R8/A is 2.80 metres wide and 3.70 metres high over a track gauge of 1,435 millimetres. The reference is 2.95 metres wide and 3.62 metres high over a track gauge of 1,067 millimetres. SED R8/A 2.80 m floor 1.10 m gauge 1,435 mm 3.70 m Commuter reference 2.95 m floor 1.13 m gauge 1,067 mm 3.62 m
Scale 1:60 at the drawn size. Roof equipment and the coupler are omitted.

Drawing 3

Twelve tonnes come out of the shell. An aluminium extrusion carries the load at 84.0 tonnes empty against 96.0 tonnes for the steel R8, and the loaded axle falls from 21.6 to 18.4 tonnes. The reference sits heavier as a formation and far lighter on the axle, because it spreads 99 tonnes over twelve axles on lighter track.

Empty weight and loaded axle weight compared Empty weight: R8 96 tonnes, R8/A 84 tonnes, reference 99 tonnes. Loaded axle weight: R8 21.6 tonnes, R8/A 18.4 tonnes, reference 11.5 tonnes. Empty weight, tonnes SED R8 · 96.0 SED R8/A · 84.0 Reference · 99.0 Loaded axle weight, tonnes SED R8 · 21.6 SED R8/A · 18.4 Reference · 11.5
Bars share a scale within each pair. Axle figures are for a full car in the peak count.

Drawing 4

It is where most of the delay can be reduced. The R8 brakes softly and blends the regeneration, so the driver commits early and corrects by feel. The R8/A computes the curve for the measured load and the measured rail, follows it, and releases into the target stop. The reference is drawn at the deceleration a pattern-controlled commuter set works to.

  • SED R8 in service, 0.90 m/s²
  • SED R8/A corrected, 1.10 m/s²
  • Commuter reference, 1.25 m/s²
Speed against distance to the stopping mark over the last two hundred metres Three braking curves. The R8 in service decelerates at 0.90 metres per second squared, the R8/A at 1.10, and the commuter reference at 1.25. All three reach zero at the mark. 200 150 100 50 0 0 20 40 60 80 metres to the mark km/h the mark
Curves are v equal to the square root of twice the deceleration by the remaining distance. Jerk limiting at the release is omitted.

Drawing 5

The driver logged twenty-four stops on the southern route through one week in June, measured from the front of the cab to the end of the platform less one metre, which is the rule he set himself. The band is the tolerance specified for the R8/A, and it is the figure Japanese automatic stopping equipment is tuned to.

Stopping positions recorded on the R8 and specified for the R8/A Twenty-four recorded stops on the R8 spread from 1.9 metres short to 2.9 metres beyond the mark. The R8/A is specified to stop within 0.35 metres either side of the target. the mark SED R8, recorded SED R8/A, specified -3 -2 -1 0 +1 +2 +3 metres from the mark, beyond to the right
Band drawn at 0.35 m either side. Recorded stops are from the driver's own log and are not official benchmarks.

The numbers

Quantity SED R8 in service SED R8/A Commuter reference
Formation Three cars Three cars Three cars of 20 m
Length over couplers 56.20 m 56.20 m 60.00 m
Width 2.95 m 2.80 m 2.95 m
Height to the roof 3.70 m 3.70 m 3.62 m
Floor height 1.10 m 1.10 m 1.13 m
Track gauge 1,435 mm 1,435 mm 1,067 mm
Body Steel shell Aluminium extrusion, welded Stainless steel
Empty weight 96.0 t 84.0 t 99.0 t
Axle weight, loaded 21.6 t 18.4 t 11.5 t
Doors a side, each car 3 pairs 3 pairs 4 pairs
Power 25 kV 50 Hz AC 25 kV 50 Hz AC, 220 kWh reserve 1,500 V DC
Service deceleration 0.90 m/s² 1.10 m/s² 1.25 m/s²
Stopping accuracy Not specified 0.35 m at the cab 0.35 m under automatic control
Expected life 40 years 40 years Replacement at 20 to 25 years
Build account 250 t CO2e 268 t CO2e Not published
Running account 14 g CO2e per vehicle-km 12 g CO2e per vehicle-km Not published

Reference figures are the published dimensions of current Japanese commuter stock and the tolerance quoted for automatic stopping equipment. They are held here for comparison and no vehicle of that kind is offered to the Republic.

The cost of the aluminium

The build account rises and the running account falls. An aluminium shell costs 18 tonnes of CO2e more to make than the steel one it replaces, and returns it through 2 grams a vehicle-kilometre over a forty year life. The group records that a vehicle bought for punctuality alone would fail the test, and that this one passes it on the energy.

Repair had to survive the change. A welded extrusion is harder to open than a bolted steel shell, so the sections are drawn to be cut and replaced whole at the half-life rebuild, and the factory holds the jigs. Nothing in the design leaves a depot mechanic unable to work on it.

The correction does not directly assist the driver. The system follows the computed curve over the last two hundred metres and releases at any touch of the brake handle instead of keeping it. The group considered a full automatic stop of the Japanese kind and rejected it, since it needs marked platforms, gates and a signalling system that regulates a train to the second, none of which exists here and none of which is proposed.

Fifteen centimetres of width cost seats. The narrowed body takes the seated count from 138 to 132 and the total from 306 to 292. On a route running eight-minute headways at the peak the group accepted the loss. On a busier route it would refuse it, and the R8 remains the standard vehicle everywhere else.

Drawing 6

A southern route platform is 70 m of unpainted concrete with a name board, a shelter and no gates. The target stop puts the front of the cab one metre inside the far end, which brings all three cars alongside and every door onto the surface. The lower panel repeats that point at an expanded scale, with the same twenty-four recorded stops and the tolerance the R8/A is specified to meet.

Plan of a southern route platform with the unit at the target stop, and a detail of the stopping positions Panel one: a 70 metre platform 4.5 metres deep, with a body 2.80 metres wide standing over a track of 1,435 millimetres, the cab front one metre inside the far end. Panel two, at an expanded scale: twenty-four recorded stops on the R8 spread from 1.9 metres short to 2.9 metres beyond the target, against the specified tolerance of 0.35 metres either side. 1. Platform of 70 m, no gates and no painted marks shelter and name board track 1,435 mm, body 2.80 m over it, doors marked on the platform side end of platform target stop: cab front 1 m inside the end 2. Detail at the target stop, expanded scale -3 -2 -1 0 +1 +2 +3 metres from the target stop, beyond it to the right recorded stops, R8 in service: 24 of them, total spread 4.8 m specified tolerance, R8/A: 0.35 m either side of the target
Panel one at one scale in both directions. Panel two at about eight times that scale, so the tolerance can be read.

Drawing 7

Everything in the loop is measurement the train already takes. The line record is the published survey of gradient and curve for the route, held as a file in the register. The load comes off the air suspension pressure, which is how the brake already knows the weight of the car. Adhesion is estimated from the slip the wheels report under the first application. The solver returns one braking curve and the blend follows it.

Block diagram of the stopping correction Line record, odometer and doppler, suspension load and adhesion estimate feed a curve solver. The solver sets a brake demand which the blended brake applies to the train. Position and speed return to the solver. The driver handle overrides the demand at any point. Line recordpublished survey file Odometerwheel pulses, doppler Loadsuspension pressure Adhesionslip, first application Curve solverone curve, recomputedevery 100 ms Brake demanda deceleration,never a brake step Blendelectric first,friction after Driver handlereleases the loop measured position and speed return
Nothing in the loop reads a wayside beacon, since the southern route has none.

Drawing 8

The system arms itself when the line record says a stop is 400 m ahead and the driver has made no application. It follows the curve, releases into the target stop and reports the error it finished with. A touch of the handle at any point drops it to standby for the rest of the approach, and it stays there until the train has left the platform.

State diagram of the stopping correction States are standby, armed, holding, releasing and stopped. The driver handle sends any state to standby. Standby is left only after the train departs. ARMED HOLDING RELEASING STOPPED STANDBY 400 m curve met last 6 m handle touched, from any state error reported, then standby until departure
The handle has priority over every transition. No state can be entered against it.

Drawing 9

Position error is largest far out and collapses as the train closes on the mark, because the wheel count has less distance left in which to drift and the slip is measured rather than assumed. The line record is the one term that stays flat, since it is the resolution the survey was taken at. Reaching below it would mean surveying the route again at a finer step.

Position uncertainty at three points in the approach At 200 metres the terms total 1.77 metres, at 50 metres 0.88 metres, and at the mark 0.40 metres. The line record term stays at 0.20 metres throughout. 0.0 0.5 1.0 1.5 2.0200 m out 1.77 m50 m out 0.88 mAt the mark 0.40 m metres of uncertainty Odometer drift Wheel slip Line record Brake response
Terms are added as they arrive at the solver. The specified 0.35 m sits inside the final bar because the last term is measured during the stop itself.

Drawing 10

This is one recorded approach on the test track with water laid down deliberately, taken as the worst case for the specification. The demand stays flat while the measured deceleration falls under the wet. The lower panel gives what that costs in position, in centimetres, on a scale of its own. The solver recovers it over the remaining hundred metres by asking for slightly more than the curve requires, and the error is zero at the target stop.

  • demanded deceleration
  • measured deceleration
  • position error, lower panel
Demanded and measured deceleration over the last two hundred metres, with the position error below The demand stays at 1.10 metres per second squared. The measured value falls to about 0.99 in the wet patch between 140 and 100 metres, and the position error reaches about 38 centimetres. The solver recovers it over the remaining hundred metres and the error is zero at the target stop. 1. Deceleration, demanded and measured, m/s2 200 150 100 50 0 0.90 1.00 1.10 1.20 1.30 water laid on, 140 m to 100 m 2. Position error against the target stop, centimetres 200 150 100 50 0 0 10 20 30 40 worst error 38 cm metres to the target stop, zero at the right
One approach from the Sederna test track, 4 June. Sampled at two metres and drawn without smoothing.

Drawing 11

Braking distance from 70 km/h under the three conditions the southern route actually offers. The R8 loses most in leaf fall because the driver has to commit early and cannot know what the rail will be like. The corrected unit measures the slip and pays for the wet in distance rather than in accuracy, which is the whole trade.

Braking distance from 70 kilometres an hour under three rail conditions Dry rail: R8 210 metres, R8/A 172 metres. Wet rail: R8 246 metres, R8/A 198 metres. Leaf fall: R8 312 metres, R8/A 244 metres. Dry R8 · 210 m R8/A · 172 m Wet R8 · 246 m R8/A · 198 m Leaf fall R8 · 312 m R8/A · 244 m
Distances include the reaction and build-up of the blend. Sanding is not assumed on any of the three.

Nothing here is proprietary

Automatic braking makes a decision affecting passengers, so the deciding component is published. The solver, the estimator and the line record format are held in the federation repository, the survey of every route is a file any person may download from vcs.podnik.rof/rolling/surveys, and a depot that changes anything writes the change into its own register. A cooperative may run the unit with the correction switched out entirely.

git clone https://vcs.podnik.rof/rolling/r8a-approach
cd r8a-approach
./configure --with-line-record --without-telemetry
make && make check
make record ROUTE=southern    # rebuilds the survey file from the register
Piece What it is Where it is Terms
Curve solver The braking curve and the release Federation repository, vcs.podnik.rof/rolling/r8a-approach Published, modifiable, no signature required to run
Estimator Position, speed, load and adhesion Federation repository, vcs.podnik.rof/rolling/r8a-approach Published, modifiable
Line record Gradient, curve and stop positions at 0.20 m Public register of the route Free of charge, without account
Record format Plain text, one row a chain Annex to this note Published, no licence sought
Stop log The error every stop finished with Published nightly by the depot Free of charge, no driver named
Brake blend Electric first, friction after Sederna Carriage Factory Drawings publicly released on request

A deviation from the reference build is recorded in the register of the operating cooperative, in the manner Organic Law 44 requires of a home server.

Drawing 12

The group has the shell and the estimator. The braking work is done on the test track and undone on the route, which is the part that takes a year. Nothing is promised for a date, and the entry in the catalogue counts one unit because one unit is what exists.

Programme of work Proposal sent, shell drawn, estimator written and test track running are complete. Route trials, the assembly decision and series build are outstanding. Proposal to the federation Shell and gauge drawn Estimator written Test track running Route trials Assembly decision, then series solid is done, dashed is not
The first unit enters service with the correction as fitted. Every later change goes through the same publication.

Standing of this note. A working paper of the drafting group, published because everything is published while it is being drafted. The proposal itself sits with the assembly of the federation. Nothing here binds a cooperative, a commune or a factory, and the figures for the reference vehicle are taken from published sources abroad.