Pietro Manni, P.Eng. PQ Logic Corporation Harvesting Energy from Regenerative Breaking Operations of DC Powered Train Systems. Pietro Manni, P.Eng. PQ Logic Corporation

PQ Logic Corp. Brief Introduction Founded in 1996 (Toronto, Canada). Specialised in Power Quality, Power System Analysis & Technical Training. Provided Consulting & Training Services in 5 Continents. Our client profile: Utilities, Medical, Industrial, Data Centres, Military, Governments, etc. PSL Value-Add Distributors since 2009.

Study Overview Subway system in a major city in North America. Subway system powered at 600 VDC. Trains equipped with Regenerative Breaking capabilities, currently not used. Breaking energy is basically being dissipated as heat. Customer wishes to estimate the technical and economic viability of harvesting the energy and storing it into battery banks.

Objective of the Study The objective of this project is to study the power usage patterns at the subway’s DC Rectifier Stations in order to estimate the amount of energy that can be potentially recovered from the trains’ energy regeneration capabilities.

Constraints Due to the large extent of the subway line and equipment access limitations, the initial study is based on a portion of the line consisting of: Sub-Station 1: 2 x 3MW rectifiers. Sub-Station 2: 3 x 3MW rectifiers. Sub-Station 3: 2 x 2MW rectifiers. # of Subway Stations: 7 Total one way distance covered by the section under study: 3.1 miles (~5.0 km). Total system calculations are (for the time being) extrapolated from this section of the analysis. Further analysis will be required to fine tune the model.

How is this analysis done? Mathematical model capable of running a DC Power Flow analysis for every operating scenario of the system. Not Easy! Modeling needed: Rectifiers (fault levels, etc.) Supply Cables Contact Rails Running Rails Loads (Train Load Current Profiles for Start, Run, Braking operations) – Acceleration Patterns. Average speed between stations. Travel time between stations.

What is the PQube’s Role? The load characteristics of the trains are well known, but this is not enough for the analysis: Total VFD Load: 3.75 MW Auxiliary Load: 300 kW Since the objective is to determine the available energy for harvesting, we need to know how the trains behave in real time. In other words we need to account for train loading conditions, running schedules, traffic, acceleration patterns, etc.

What is the PQube’s Role? PQubes to be installed on trains to record voltage and current consumption/generation patterns for a total of 7 days. These patterns are used to verify the dynamic behavior of the loads (trains) used in the computer model.

Distance Patterns Between Stations FROM TO TRAVEL TIME (s) STOP TIME (s) TOTAL TIME (s) STATION 1 STATION 2 45 15 60 STATION 3 90 105 STATION 4 STATION 5 STATION 6 STATION 7 TOTALS   405 495 Notice that there are 2 basic time patterns: 60s and 105s travel time.

Train Load Current Patterns 60 Second Run

Train Load Current Patterns 105 Second Run

Identification of Operating Patterns A total of 7 different train movement patterns were identified along the trajectory between Station 1 and Station 7 (and vice-versa). Each pattern has a distinctive load profile that characterizes the 495 second time-span between these boundary stations. The patterns were converted into time-current profiles and were assigned a scenario number for the analysis.

Operating Patterns Example: Pattern #7 Westbound TIME STATION 7 STATION 6 STATION 5 STATION 4 STATION 3 STATION 2 7:50 S   R 7:51 7:52 7:53 7:54 7:55 7:56 7:57 7:58 R: Train Running Through Station S: Train Starting

Operating Patterns Example: Pattern #7 Eastbound TIME STATION 1 STATION 2 STATION 3 STATION 4 STATION 5 STATION 6 7:50   S 7:51 R 7:52 7:53 7:54 7:55 7:56 7:57 7:58 R: Train Running Through Station S: Train Starting

Operating Patterns Total Frequency Weekday Frequency (times/day) Weekend Frequency (times/day) 1 35 104 2 32 -- 3 12 4 13 5 6 7 36

Network Model

Simulation Results Example: Pattern #7

Computer Model & DC Power Flow Run computer model for the audience here.

Energy Calculations (per simulated segment) ENERGY (kWh) per 495s pattern STATION 1 STATION 2 STATION 3 Ec Er Scenario 1 62.13 -15.14 150.22 -38.68 93.11 -25.03 Scenario 2 101.80 -15.44 216.35 -26.58 174.47 -24.38 Scenario 3 96.56 -22.88 283.19 -64.62 234.61 -60.37 Scenario 4 100.68 -19.82 264.90 -46.09 211.12 -37.94 Scenario 5 137.88 -19.03 345.64 -41.50 290.89 -61.19 Scenario 6 75.46 -19.22 203.92 -44.16 184.70 -44.58 Scenario 7 104.71 -4.88 290.78 -29.08 219.76 -40.79 Ec: Energy Consumed Er: Energy Regenerated

Energy Calculations (from computer model) Station 1 Station 2 Station 3 Ec Er kWh/Day (Weekdays) 12,322.33 -1,866.04 31,201.84 -4,929.65 23,818.41 -4,687.59 kWh/Day (Weekends) 6,461.52 -1,574.56 15,622.88 -4,022.72 9,683.44 -2,603.12 MWh/Year 3,847.10 -648.75 9,659.41 -1,698.65 7,124.74 -1,479.59 Ec: Energy Consumed Er: Energy Regenerated

Comments Traditional power flow simulations rely on known specific load values and equipment specifications. This approach is enough for worst case scenarios used in system design and snap shot evaluations, but not enough for accurate performance estimations. A better approach is to use the time series based modeling of loads and system components as described in this presentation.

Comments Recommendations were made to permanently install PQubes at rectifier stations to verify and fine tune the model as well as to keep track of the energy consumed and harvested in real time. Thank You! Questions?