USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability

Transcription

1 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Dr Marc ANTONI

2 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Summary : 1. Why does Asset management need Modelisation? 2. Modelisation for Infrastructure management after 3. Modelisation for Infrastructure management before 4. Conclusions & perspectives

3 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Summary: 1. Why does Asset management need Modelisation? 2. Modelisation for Infrastructure management after 3. Modelisation for Infrastructure management before 4. Conclusions & perspectives

4 Why does Asset management need Modelisation? Asset management : looking for the optimal balance Traffic fees Maintenance costs Target performance of the network Line capacity, speed, acceptable unavailability rate, safety rate, track availability Levels of maintenance and renewal State of Network Performance, level of deterioration, evolution, safety level

5 Why does Asset management need Modelisation? Asset management : Purpose à Spend less to obtain more. 1 not spent for the same result = 1 of net profit Each maintenance decision has a cost: A poorly-managed asset = unavailability risk = costs A postponed expense will increase risk exposure è cost The major difficulty is optimising the balance between the two risks è how to predict future maintenance needs in relation with the usage and the environment in the real context of use

6 Why does Asset management need Modelisation? Asset management : what does it consist of? à To create consistency between: - the decisions made by head-office (strategy and target performances) - the operational management of assets A structured approach of: The knowledge and the management of life cycles The global long term costs (deterioration laws) Calculated risks (economic risk value assessment) Risk/cost scenarios Justifying decisions and their return on investment

7 Why does Asset management need Modelisation? Asset management: in practice Obtain reliable deterioration models and survival models for all infrastructure components Formalise the survival model for unreplaced components by fixing the failure rate λ(t) in coherence with several covariates influencing the life span of the components Formalise the existing economic link between maintenance costs and the volume of infrastructure renewal Reduce costs by optimisation of the maintenance/renewal ratio, which is a major issue for infrastructure asset managers.

8 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Summary : 1. Why does Asset management need Modelisation? 2. Modelisation for Infrastructure management after 3. Modelisation for Infrastructure management before 4. Conclusions & perspectives

9 Modelisation for Infrastructure Management after Three steps (example for track) : 3 Tools for LCC calculation at the national or route levels, including environmental effects, track possession and unavailability costs 2 Tools for the estimation of maintenance needs of the track (with different renewal strategies) 1 Work of the deterioration and failure laws of each the track components

10 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (French Mass Transit line) Failure laws of rails : - lifespan of the rails on a ballasted Mass transit line is about 600MT with 6% cumulative renewed rails length - the parameters of these laws are sensitive to the track type : ballasted or ballast less track MT

11 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Failure laws of rails : - lifespan of the rails on a ballasted HSL is about 400MT with 3% of cumulative defects, 700MT with 6% - the parameters of these laws are sensitive to track topology and aggressiveness of the rolling stock After The failure 30 years rate of can operations grow more and quickly 650MT if with the V300, rolling more stock than has ¾ an of important rails of LN1 rate are of original. The slippage last rails (20% are for to some be replaced materials) in 2018 MT

12 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Failure laws of aluminothermy welding: - lifespan of a weld on ballasted HSL is about 400MT with 3% of cumulative defects [even without preventive grinding] Taux cumulé de défaillance - H(t) 0,05 0,045 0,04 0,035 0,03 0,025 0,02 0,015 0,01 0, TBC (Mt) H(t) empirique toutes LN H(t) estimée toutes LN MT

13 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Failure laws of manganese or movable frogs: - lifespan of these components is longer on wooden sleepers then on concrete ones - the parameters of these laws are sensitive to the aggressiveness of the rolling stock 0,3 0,25 0,2 0,15 0,1 0,05 0 Cœur à Pointe Fixe LGV Cœur Pointe Fixe Béton Cœur Pointe Fixe Bois MT MT

14 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Failure laws of manganese or movable frogs: - lifespan of these components is longer on wooden sleepers then on concrete ones - the parameters of these laws are sensitive to the aggressiveness of the rolling stock 0,3 0,25 0,2 0,15 0,1 0,05 0 Cœur à Pointe Mobile LGV Cœur Pointe Mobile Béton Cœur Pointe Mobile Bois MT MT

15 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Failure laws of switch half switch set: - lifespan of these components is longer on wood sleepers then on concrete ones - the parameters of these laws are sensitive to the aggressiveness of the rolling stock and the hardness of the track 0,3 0,25 0,2 0,15 0,1 0,05 0 Demi Aiguillage LGV Demi aiguillage Beton Demi aiguillage Bois MT

16 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Geometry degradation laws: - lifespan of the ballast, without sand-gravel mix bitumen or PAD, is ~25years on HSL (>300) - this lifespan will be much higher with sand-gravel mix bitumen and/or PAD - maintenance needs follow Cochet-Maumy laws Coûts de maintien de la géométrie (yc surveillance) sur voie ballastée LGV Voie Gr2 Sans Meulage Voie Gr2 Avec Meulage Préventif N 5 Im( N) = k 0,8 δ a + b 2 1 MT

17 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) The nature of the under layer has a significant influence on track lifespan and HSL geometry specific Cochet-Maumy parameters Frequencies [Hz] Influence bitumen mix at 320km/h with a distance between axles of 3m Signatures des structures géotechniques? Energy measured [Log (m/s 2 /Hz)] Energy measured [m/s 2 /Hz] The test section test with Bituminous mix makes it possible to obtain a reduction of a factor 2 of the annual percentage tamp length

18 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Some under sleeper PAD have an influence on track lifespan and HSL geometry specific Cochet-Maumy parameters Maintenance interventions

19 Modelisation for Infrastructure Management after Step 1: Lifespan of the components (ballasted HSL) Some under sleeper PAD have an influence on track lifespan and HSL geometry specific Cochet-Maumy parameters Maintenance interventions Average of longitudinal levelling

20 Modelisation for Infrastructure Management after Step 2: Estimation of maintenance needs (ballasted HSL) Tools for estimation of track maintenance needs (EBM) : Principe / ballasted track: 1 Cyclical or programmed operations: Fixed charges determined by the standards for track surveillance, programmed maintenance, structure 2 Preventive conditioned maintenance: - Levelling maintenance charges: Interventions conditioned by the information coming from track surveillance. Probabilistic estimation of the intervention needs for a specific route, for a UIC group of routes - Asset replacement charges: Interventions conditioned by asset defect detection Probabilistic estimation of the failure laws of each asset

21 Modelisation for Infrastructure Management after Step 2: Estimation of maintenance needs (ballasted HSL) Example of estimation of maintenance needs for the French network Switches & Crossing UIC 1 to 6 Normal Track UIC 1 to 6 Millions Extrapolation non modélisable Millions Scénario SsR Entr. mode 1 Scénario 1 Entr. sc. 1 Scénario 2 Entr. sc. 2 Scénario Entr. sc. 3 3 Scénario Entr. sc. 4 4 Entr. sc. 5 Scénario 5 Entr. sc. 6 Scénario 6 Entr. mode 3 Scénario RaD Entr. Ss Renouv Scénario de base Variante 1 (+50%2-4) Variante 2 (+50%5-6) Variante 3 (+100%5-6) Variante 4 (base prolongé) Entr. Renouv à date (<1000km/an)

22 Modelisation for Infrastructure Management after Step 2: Analyse of the maintenance needs (ballasted HSL) Example HSL1 : evolution of the standard deviation and the annual tamping effort Km (40-60) Time ( ) The parameter are feted for each track kilometer

23 Modelisation for Infrastructure Management after Step 2: Analyse of the maintenance needs (ballasted HSL) Example HSL1 : evolution of cumulative tamping effort and the ballast renewal Km (40-60) Time ( ) The parameter are feted for each track kilometer

24 Modelisation for Infrastructure Management after Step 3: LCC calculations (ballasted and unballasted HSL) For Asset Management, it is essential to be able to evaluate comparatively, and in each context Ballasted and Slab track: Expected lifespan 50 years ~ 60 years Technology UIC60, Concrete sleepers Max Bögl, Rheda and UIC60 Maintenance Grinding, Grinding, Drainage, Tamping, Geometry corrections of Ballast & Rail renewal (25y) Rails renewal / 15years Track initial investment 0,5 M /km 1 to 2 M /km Track renewal 0,7 to 1,4 M /km twice Signalling investment TC (BR detection) AC without BR detection => more grinding for safety Speed limit (announced) >320 if d=3 m >350 km/h Speed limit in service 320 km/h (SNCF) 250 (DB) 310 (China, Korea)

25 Modelisation for Infrastructure Management after Step 3: LCC calculations (ballasted and unballasted HSL) HSL ballasted track (UIC group3) Σ Invest / T E/T I/T E+I/T Σ Maintenance / T HSL300 - UIC - current track without switches

26 Modelisation for Infrastructure Management after Step 3: LCC calculations (ballasted and unballasted HSL) HSL ballasted track (UIC group3) E/T I/T E+I/T Co HSL300 - UIC - current To track without switches

27 Modelisation for Infrastructure Management after Step 3: LCC calculations (ballasted and unballasted HSL) Slab track Σ Invest / T E/T I/T E+I/T Σ Maintenance / T HSL300 - UIC3 - current track without switches

28 Modelisation for Infrastructure Management after Step 3: LCC calculations (ballasted and unballasted HSL) Slab track E/T I/T E+I/T Co HSL300 - UIC3 - current track To without switches

29 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Summary : 1. Why does Asset management need Modelisation? 2. Modelisation for Infrastructure management after 3. Modelisation for Infrastructure management before 4. Conclusions & perspectives

30 Modelisation for Infrastructure management before Thanks to its experience of component and sub-system behaviour, SNCF-Infrastructure can: à specify and optimise new components to facilitate maintenance, taking into account usage, environment, specific quality targets à optimise the dimension of spare parts and the corresponding maintenance organisation. The following examples come from ballasted track: à choice of failure laws, à track definition with and without PAD

31 Modelisation for Infrastructure management before Modeling methods: renewal density for successive replaced components (for example fastening or rail welding) Without system ageing The renewal density gives the replacements due to failure at time t : h( t) = [(1 F( t n= 1 where * denotes the convolution. ))'] n The integral of this function gives the number of expected replacements before time t. h(tr) 0,10 0,09 0,08 0,07 0,06 0,05 0,04 0,03 0,02 0,01 0,00 0,00 0,08 0,16 0,24 0,32 0,40 0,48 0,56 0,64 0,72 0,80 0,88 0,96 1,04 1,12 1,20 1,28 1,36 Reduced time tr 1,44 1,52 1,60 1,68 1,76 1,84 1,92 Eta 25 Beta 3 Eta 25 Beta 6 Eta 25 Beta 9 Eta 25 Beta 12

32 Modelisation for Infrastructure management before Modeling methods: renewal density for successive replaced components With system aging We can include the ageing of the system (or effects of repairs) by using a factor K η n = η 0 K at the n th replacement. This translates the fact that even a new component has a reduced lifetime if it is introduced into an ageing system. h(tr) 0,10 0,09 0,08 0,07 0,06 0,05 0,04 0,03 0,02 0,01 0,00 0,00 0,08 0,16 0,24 0,32 0,40 0,48 0,56 0,64 0,72 0,80 0,88 0,96 1,04 1,12 1,20 1,28 Reduced time tr 1,36 1,44 1,52 1,60 1,68 1,76 1,84 1,92 K = 0,75 Eta 25 Beta 3 Eta 25 Beta 6 Eta 25 Beta 9 Eta 25 Beta 12

33 Modelisation for Infrastructure management before Modeling methods: renewal density for successive replaced components Maintenance expenses: Y(t) = c i (t) + c u n h(t) c i : current costs c u : replacement costs for one component n: number of components h(t): renewal density Expected global maintenance expenses (including renewal) per year: C( T)/ T With X: renewal costs = [ X + T 1 t+ 1 t= 0 t Y( t)]/ T E(T) Co To with K=1 Som(X)/T Som(3;Y)/T Som(6;Y)/T Som(9;Y)/T Som(X)/T + Som(1,7;Y)/T Som(X)/T + Som(3,4;Y)/T Som(X)/T + Som(5,1;Y)/T

34 Modelisation for Infrastructure management before Modeling methods: renewal density for successive replaced components Maintenance expenses: Y(t) = c i (t) + c u n h(t) c i : current costs c u : replacement costs for one component n: number of components h(t): renewal density Expected global maintenance expenses (including renewal) per year: C( T)/ T With X: renewal costs = [ X + T 1 t+ 1 t= 0 t Y( t)]/ T E(T) Co To with K=0,75 Som(X)/T Som(3;Y)/T Som(6;Y)/T Som(9;Y)/T Som(X)/T + Som(1,7;Y)/T Som(X)/T + Som(3,4;Y)/T Som(X)/T + Som(5,1;Y)/T

35 Modelisation for Infrastructure management before Modeling methods: renewal density for successive replaced components Design choices could have a huge impact on a maintenance strategy and on the chances of reaching the right quality level (availability, security, safety ) with the economic target value The terms of the requirements have to be chosen taking into consideration the context of use and the economic and organizational targets which are not known by suppliers

36 Modelisation for Infrastructure management before Track type choice in regard of the economic target If we choose the Ballasted track for it s: - Cheaper investment - Easiness for the maintenance operations - Compatibility with signaling systems We have still to make an important choice: without or with PAD Under sleeper PAD allows : - Reduction of the tamping frequency (average 1/5) - Increase of the sleeper lifespan (above 10%) - Increase of the ballast lifespan (above 100% for HSL) - To have high hopes that the renewal of the ballast and the sleepers can be done at the same time

37 Modelisation for Infrastructure Management before Track type choice in regard of the economic target Example HSL5 : evolution of the standard deviation and the annual tamping effort V320 Without PAD With PAD Km Time ( ) The parameter are feted on the with PAD test zone

38 Modelisation for Infrastructure Management before LCC calculations BT without PAD vs. BT with PAD Expected rail lifespan 25 to 40 years 25 to 50 years at MT at 600MT or more Expected ballast lifespan 25 years 50 years Expected sleeper lifespan years years Initial investment Tamping frequency 1 1/5 Rolling stock TGV TGV Broken Rail Detection Yes Yes

39 Modelisation for Infrastructure Management before LCC calculations Case 1 : Ballasted track without PAD (seen before)

40 Modelisation for Infrastructure Management before LCC calculations Case 2 : Ballasted track with PAD (with safety-conscious parameters)

41 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Summary : 1. Why does Asset management need Modelisation? 2. Modelisation for Infrastructure management after 3. Modelisation for Infrastructure management before 4. Conclusions & perspectives

42 Conclusions & perspectives This approach allows a good modelling of the intuitively perceived phenomena: - regeneration vs. maintenance - a system cannot last indefinitely The calibration of the model was based on accessible real data. The method is very general. This presentation shows that the application range is very wide. All replaceable infrastructure equipment can be used for such a study. 42

43 Conclusions & perspectives This approach can be applied to particular sub populations: high speed lines, lines with dedicated traffic, environmental variables This approach can predict failure trends. The method estimates the trend of future maintenance expenses. It can evaluate in an objective way SNCF maintenance strategies and can be used for asset management 43

44 Conclusions & perspectives The battle for maintenance is won or lost at the system definition & design stage. This is essential to consider the industrial balance of the trio made up of Maintenance costs Network Performances Quality. A technical solution doesn't, on its own, produce the best answer.

45 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Thanks for your attention Contacts : SNCF Infrastructure Dr. Marc ANTONI 18 rue de Dunkerque PARIS (+33) marc.antoni@sncf.fr Any questions?

46 USING LCC MODELLING TO DETERMINE THE BEST TRACK CHOICE : Minimising Life-Cycle Costs Whilst Maintaining Reliability Thanks for your attention Contacts : SNCF Infrastructure Dr. Marc ANTONI 18 rue de Dunkerque PARIS (+33) marc.antoni@sncf.fr Any questions?