Rail road switches enable trains to be guided from one track to another, and rail road switches heaters are used to avoid the formation of snow and ice during the cold season in order to guarantee their correct functioning. Managing the energy consumption of these devices is important in order to reduce the costs and minimise the environmental impact. While doing so, it is important to guarantee the reliability of the system. In this work we analyse reliability and energy consumption indicators for a system of (remotely controlled) rail road switch heaters by developing and solving stochastic models based on the Stochastic Activity Networks (SAN) formalism. An on-off policy is considered for heating the switches, with parametric thresholds representing the temperatures activating/deactivating the heating. Initial investigations are carried on to understand the impact of different thresholds on the indicators under analysis (probability of failure and energy consumption).

1. Stochastic Model-Based Analysis of Energy Consumption
in a Rail Road Switch Heating System
Davide Basile Silvano Chiaradonna Felicita Di Giandomenico
Stefania Gnesi Franco Mazzanti
Istituto di Scienza e Tecnologia dell'Informazione A. Faedo,
Consiglio Nazionale delle Ricerche, ISTI-CNR, Pisa, Italy
SERENE 2015, 08/09/2015
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 1 / 18

2. Overline of the presentation
1 Motivations and Objectives
2 The System Under Analysis
3 Stochastic Model-Based Analysis
4 Description of The Model
5 Analysis Results
6 Conclusion
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 2 / 18

3. Motivations and Objectives
Motivations and Objectives
Nowadays there is a great attention towards cautious usage of energy
sources, to save both in nancial terms and in environmental impact;
While optmizing the energy consumption, other non functional
properties must be guaranteed: reliability, availability, safety;
A study devoted to the energy saving in the context of a railway
station is presented;
Stochastic-model based analysis exploited to measure the energy
saving and the reliability of the system under analysis.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 3 / 18

4. The System Under Analysis
The System Under Analysis (1/2)
Rail road switches enabling trains to be guided from one track to
another:
Reliability of the railway system depends on them.
Snow and ice can block them. Heaters are used to kept the
temperature above freezing.
Reliability and energy consumption of rail road switch heaters:
an overhead of energy consumption can lead to a blackout,
an parsimonious policy of energy saving can cause a failure.
Minimum and maximum temperature thresholds policy, to guarantee a
satisfactory trade-o between energy consumption and reliability.
Stochastic model-based approach, modular, parametric and extensible.
Stochastic Activity Network (SAN) to model the system,
Möbius tool to perform experiments.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 4 / 18

5. The System Under Analysis
The System Under Analysis (2/2)
Tubular at heaters along the rail road track, induction heating;
Sensors to read the temperatures of the air and of the rail road;
simulated by a physic engine to perform experiments.
The policy to activate/deactivate the heating on two threshold
temperatures:
warning threshold: if lower, the heating system needs to be activated;
working threshold: if higher, the heating system can be turned o.
Location and the weather conditions inuence the actual temperature;
Network of heaters, Powerline, Central unit to manage the maximum
amount of power that can be delivered to the system and priorities.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 5 / 18

6. Stochastic Model-Based Analysis
Stochastic Model-Based Analysis
Useful to support development in all phase of software life cycle, for
validation, analysis of dependability, performance;
SAN: a generalization of Stochastic Petri Nets;
composed of places,activities,arcs,input gates,output gates;
activities are instantaneous or timed (stochastic temporal distribution
of time), with associated probabilistic cases;
dierent policies of activation/reactivation of an activity;
activities can use C++ code.
Möbius: a tool for modelling complex systems in a modular way, with
dierent analytical and simulative solvers;
used for evaluate the energy consumption and the probability of failure.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 6 / 18

7. The System Under Analysis
Description of The Model (1/3)
Five atomic models: stochastic selector of weather prole and locality,
selector of unique ID, main model, queue shared among the heaters;
Parameters: temperature thresholds, maximum amount of heaters
that can be turned on at the same time, table with average
temperatures and localities.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 7 / 18

8. Description of The Model
Description of The Model (2/3) Heater Module
mc ∂T
∂t = −uA(T − Tenv ) + ˙Q
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 8 / 18

9. Description of The Model
Description of The Model (3/3) Composed Model
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 9 / 18

10. Analysis Results
Analysis Results: Measure of Interests (1/8)
CE(t, l): the mean energy consumed by a heater in the interval
[t, t + l] (6:00 pm to 6:00 am);
PFAIL(t, l): the mean probability that a switch fails at time t + l,
given that at time t is not failed;
Simulations with Möbius tool with 10000 batches, we analyse 4 pairs
of thresholds more and more wider and distant from the temperature
of failure;
We have analysed the measure of interests trend at the varying of the
available energy and the gap between the temperature thresholds.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 10 / 18

11. Analysis Results
Analysis Results: Probability of Failure (2/8)
1 - 4 4 - 5 3 - 7 4 - 50
0
0.2
0.4
0.6 0.27
4.89·10−2
0.23
0.61
2.4·10−2
0
2.5·10−5
0.37
1.35·10−3
0
0
0.16
(warning threshold-working threshold)
PFAIL(t,l)
NHMax = 1 NHMax = 2 NHMax = 3
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 11 / 18

12. Analysis Results
Analysis Results: Energy consumption (3/8)
1 - 4 4 - 5 3 - 7 4 - 50
0
2
4
6
8
10 1.25
2.57
2.19
2.65
1.68
3.15
3.36
5.18
1.7
3.18
3.61
7.6
1.7
3.18
3.73
9.91
(warning threshold-working threshold)
CE(t,l)
NHMax = 1 NHMax = 2 NHMax = 3 NHMax = 4
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 12 / 18

13. Analysis Results
Analysis Results (4/8)
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 13 / 18

14. Analysis Results
Analysis Results (5/8)
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 14 / 18

15. Analysis Results
Analysis Results (6/8)
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 15 / 18

16. Analysis Results
Analysis Results (7/8)
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 16 / 18

17. Analysis Results
Analysis Results (8/8)
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 17 / 18

18. Conclusion
Conclusion
Model-based analysis for a rail road switch heating system;
SAN and Möbius tool to evaluate energy consumption and probability
of failure at varying of temperature thresholds and available energy;
Physical model for calculating the temperature of the rail road track,
weather data taken from http://www.meteoam.it/;
Results: small gap between thresholds that are reasonably distant
from freezing temperature are better;
Future works: FIFO queue with priority to implement a smart logic
for activating/deactivating the heaters, apply our technique to an
existing railway station.
D.B. et al. (ISTI CNR) Stochastich Model Based Analysis SERENE 2015 18 / 18