ERTMS/ETCS braking curves Robert Dijkman EEIG ERTMS Users Group 15-17 December 2004 UIC ERTMS Conference Rome 1
Why are braking curves an issue? SRS 222: 1. No feasible input defined to enable the driver to enter brake characteristics in trains with changing composition. 2. Brake model (SRS chapter 3.13) defined only in a global way. Implementation is supplier dependent. 3. Harmonised safety margin for the braking curves is not defined. 15-17 December 2004 UIC ERTMS Conference Rome 2
Who are the players? 1. ERRI/UIC B126.15: brake experts. 2. EEIG: ERTMS experts. 3. Interested railways (SBB, OEBB, BV): specific operational problems. 4. UNISIG: ERTMS specifications. 15-17 December 2004 UIC ERTMS Conference Rome 3
Finding the right balance performance safety driver ergonomics 15-17 December 2004 UIC ERTMS Conference Rome 4
Braking curve overview Variable composition ETCS curves 2 Fixed composition - brake percentage - brake type: P/G Driver Data Entry - guaranteed deceleration - service deceleration - guaranteed brake delay time measured: - max deceleration - service deceleration - brake delay time Onboard ETCS conversion model: -deceleration - brake delay time 1 Safety margin 3 Pre-programmed in onboard ETCS 15-17 December 2004 UIC ERTMS Conference Rome 5
Conversion model ERRI model versus reality example 1: passenger train 2,5 Deutsche Bahn AG - Passenger train with disc brakes (pneumatic brake accelerators off) in R 2,0 deceleration [m/s²] 1,5 1,0 Nominal speeds 200 km/h 180 km/h 160 km/h 140 km/h 120 km/h 100 km/h 0,5 Models for 135 % ERRI model calibrated model 0,0 0 50 100 150 200 250 velocity [km/h] 15-17 December 2004 UIC ERTMS Conference Rome 6
Conversion model (2) ERRI model versus reality example 2: freight train 2,5 Deutsche Bahn AG - Goods train with cast iron brake blocks (Minden 2004) in P 2,0 deceleration [m/s²] 1,5 1,0 Nominal speeds 120 km/h 0,5 100 km/h Models for 101 % ERRI model possible model 0,0 0 20 40 60 80 100 120 140 velocity [km/h] 15-17 December 2004 UIC ERTMS Conference Rome 7
Conversion model (3) Options for improvement: 1. Delete the hidden safety margin. The output deceleration will be put at the average measured value instead of a conservative value. All neccessary safety margin will be concentrated in one part of the whole braking curve model structure. The conversion model remains a one size fits all solution. The output deceleration is a single value independent from the train speed, because it is based on a single input: the brake percentage. 2. Make the model brake type dependent. The output can be a more accurate representation of the deceleration of the specific brake type. This however requires the brake type to be an input to the model. Is it realistic to expect the driver to input more detailed characteristics of the actual brake system of the whole train? 15-17 December 2004 UIC ERTMS Conference Rome 8
ETCS brake model deceleration intervention audible warning permitted speed indication pre-indication EOA 15-17 December 2004 UIC ERTMS Conference Rome 9
Impact on approach of the EOA A flat braking curve will force the train to approach the EOA with low speed, resulting in a long time to stop. steep curve short stopping time good performance flat curve long stopping time bad performance EOA Platform 15-17 December 2004 UIC ERTMS Conference Rome 10
Impact on level 1 lines model not fully defined permitted curve deceleration curve Where to put the balise? Distant signal EOA Main signal 15-17 December 2004 UIC ERTMS Conference Rome 11
ETCS brake model Actions: 1. Decrease the time between the different curves to the absolute minimum that is acceptable from ergonomic point of view. This is done by the railways in the context of EEIG and B126.15 with the help of an SNCF simulator in Lille and a DB simulator in Munich. Tests with these simulators are ongoing. 2. Define the ETCS braking curve model in sufficient detail to guarantee a predictable braking distance of a given train, independent from the supplier of the onboard ETCS equipment. This is done in close co-operation between railways (co-ordinated by EEIG) and UNISIG. 15-17 December 2004 UIC ERTMS Conference Rome 12
Safety margins Output conversion model or Measured deceleration A basic A guaranteed = K v * K r * A basic T guaranteed = K t * T basic A guaranteed EOA K v K r (=1) - speed dependent factor for reduction of basic deceleration (=1) - train length dependent factor for reduction of basic deceleration K t (=1) - factor for prolongation of brake development time A max - maximum deceleration when reduced adhesion is active 15-17 December 2004 UIC ERTMS Conference Rome 13
Safety margins (2) The facts: Harmonised safety margin work ongoing in B126.15. Based on actual safety margins used in existing systems. Results not available in time for SRS300. Relation between performance requirements and safety requirements can vary between the different railways, specifically in the case of an existing line with level 1 overlay. The short term solution for SRS300: Implement the safety margins as a national value which can be sent to the train from trackside. This is a temporary solution, to be superseded by the harmonised safety margin as soon as it becomes available. 15-17 December 2004 UIC ERTMS Conference Rome 14
Conclusions The current pragmatic approach gives the best possible mix of: safety performance driver ergonomics flexibility for the railways to implement ERTMS in the way that best suits their needs 15-17 December 2004 UIC ERTMS Conference Rome 15