RVII Testing on Tenaga Power System

Dino Lelic, Rahul Anilkumar, Boza Avramovic, Tony Jiang, Damir Novosel Quanta Technology LLC

Nik Sofizan B Nik Yusuf , Sheikh Kamar Sheikh Abdullah, Muhammad Tarmizi Azmi , Mohd Khairun, Nizam Mohd Sarmin Tenaga Nasional Berhad NASPI International Synchrophasor Symposium Atlanta, Georgia Quanta Technology, LLC 4020 Westchase Blvd. March 22-24, 2016 Raleigh, NC 27607, USA www.quanta-technology.com Confidential & Proprietary | Copyright © 2015 Confidential & Proprietary | Copyright © 2016 RVII Foundation Fundamentally, local voltage instability detector

■ Just like a relay, real time computation that uses only local information (no network model needed) ■ Unlike a relay , able to avoid point #1 (figure a & b), and includes point #2

(a) (b) System impedance, vs load impedance 1 0.8 System X 0.7 Load X 0.98

0.6 0.96

0.5 0.94 0.4 0.92 0.3

0.9 0.2

0.88 0.1

0.86 0 0 50 100 150 0 50 100 150

■ Bus RVII ■ Area RVII ■ Corridor RVII • Leverages Bus RVII and analytical computation of the corridor to improve quality of Equivalent parameters

■ Three stage approach to meet objectives

■ Real Time voltage instability DETECTION ■ WHAT IF analysis: • Extrapolation − Margin to Instability − Margin to Operating boundary

• Predictions − Margin to Instability − Margin to Operating boundary

■ Extrapolation • Constant power factor load increase until boundaries reached • System equivalent params kept fixed ■ Prediction • Additional information used (hybrid of RT model-less, and much slower 5000 Operating Point model-based) Operation Limit Collapse Limit − To define more accurately load change 0 path -5000 − To update equivalent parameters as a

function of load increase -10000 − Plot reactive power margin -15000 ■ RVII implements Extrapolation

-20000 0 5000 10000 15000

■ Three main power utilities in Malaysia: TNB, SESB & SEB

PERLIS LANGKAWI Chuping Kangar

Kuah TTPC Kota Bharu

Kota Setar Alor Setar

KEDAH Gurun Tanah Merah Bedong Sg. Petani PERGAU GELUGOR PRAI Georgetown Butterworth PULAU PINANG Bukit Tengah TEMENGOR Kuala Terengganu Kulim BERSIA Junjung Bukit Tambun KENERING KELANTAN SG PIAH UPPER SG PIAH LOWER Kuala Berang CHENDEROH KENYIR Gua Musang Kuala Kangsar TERENGGANU Taiping PERAK PAKA Ipoh

Papan YTL Batu Gajah SEGARI Kampar JOR Ayer Tawar Telok Kalong Seri Iskandar Kuala Lipis Dungun Lumut WOH JANAMANJUNG ODAK Jerantut Teluk Intan High Kuantan Load PAHANG Kuala Kubu Baru Mentakab DemandBukit Tarek SELANGOR Kg Awah Kuala Selangor Bentong Temerloh

KL (N) KLWILAYAH (E) KAPAR Shah Alam PERSEKUTUAN KL (S) SERDANG CONNAUGHT BRIDGE Hicom G Muadzam Shah

Banting NEGERI SEMBILAN GENTING SANYEN Salak Tinggi Kuala Pilah Seremban Gemas PD POWER Segamat

TJPS Mersing POWERTEK MELAKA Melaka PAHLAWAN Melaka JOHOR MELAKA Yong Peng (N) Kluang Muar

Batu Pahat

Skudai YTL Pontian Kechil Johor Bahru Pasir Gudang Gelang Patah PASIR GUDANG

■ Quanta Technology(QT) and Tenaga Nasional Berhad (TNB) worked together on integrating RVII functionality with their OpalRT simulator, which uses a complete transient stability model of TNB bulk power system. ■ Testing was extensive, using a combination of • Time domain transient simulations (PSS/E) in a preliminary tuning and testing stage • Software in the loop simulations (OpalRT) ■ Tests clearly indicate the capability of RVII to correctly distinguish voltage stable from unstable cases • Referring to Slide #2, it correctly characterizes points #1 and points #2 • For example, RVII was able to detect proximity to voltage instability when relatively high voltages would have masked the actual proximity to voltage collapse if viewed only through SCADA data (or under-voltage relays). ■ Currently implementations of RVII algorithm in TNB’s embedder controllers are being carried out for Hardware in the loop testing.

■ Prior to RT Deployment, RVII List Branches Identify Volt-weak deployed in test environment (for flows and volts) buses For an RVII location

• Comprehensive TNB transient PSS/E case PSS/E MATLAB/LINUX MONO MATLAB COM stability model used (.raw, PSS/E Dyn. Results ENVIRONMENT INTERFACE .dyr,.idev − PSSE – based fies)

• Establish RVII locations, Python Scripts/ RVII DLL/.P suitable inputs, and verify Response files ACCC tools, accuracy CON files ■ At TNB OpalRT-based “deployment” List of • scenarios (TNB − Entire TNB network approved) − Physical and Virtual PMUs • Next Step – in progress: Input: PSSE case; output plots of − Implementation in TNBs Thevenin and “Load” Impedance embedded controller − Hardware in the loop setup

Slide 9 Confidential & Proprietary | Copyright © 2016 Example 1: Interface @ base Power Transfer • Impact of outage at Zero Transfer Level on critical TNB interface • This references post - outage condition • Low voltages in the vicinity of the study zone, but • Clear separation between Thevenin and Load Impedances • The conclusion: system voltage stable • Fig. below shows system operating further away (wider margin) from operational or collapse limits.

1000 Operating Point Operation Limit 500 Collapse Limit

Note: in this case 0

voltage was relatively -500 low -1000

-1500

-2000

-2500

-3000

-3500

-4000 0 500 1000 1500 2000 2500 3000 3500 4000

• Impact of outage at Interface Transfer limit. • Path TTC limited by transient stability problem. • RVII successfully detects instability • System is operating with close to zero margins, at the boundary of instability limits.

RVII detects instability System operating at limits

1000 Operating Point Operation Limit 500 Collapse Limit

-500

-1000

Note: in this case -1500

voltage was -2000 relatively high -2500

-3000 0 500 1000 1500 2000 2500

Slide 11 Confidential & Proprietary | Copyright © 2016 Under-voltage Relay Example Double Contingency to Load Pocket ■ Drastic action may be taken on voltages of 0.95 p.u to prevent instability. ■ RVII would show, however, that system is far from collapse (verified by PSS/E time domain simulations).

System impedance, vs load impedance 1.8 System X 1.6 Load X

1.4

1.2

0.8

0.6

0.4

0.2

0 0 20 40 60 80 100 120

UVLS Triggers Very insignificant change in Thevenin Impedance

■ RVII successfully identifies separation between system and load impedances, indicating a stable system condition. ■ Absent RVII, drastic under-voltage action might be taken at voltages of 0.95 pu to prevent instability.

System impedance, vs load impedance 0.8 1 System X 0.7 Load X 0.98

0.6 0.96

0.5 0.94

0.4 0.92 0.3 0.9 0.2

0.88 0.1

0.86 0 50 100 150 0 0 50 100 150

UVLS Triggers Insignificant change in Thevenin Impedance

■ RVII successfully identifies separation between system and load impedances, indicating a stable system condition even in scenarios with loss of source supply to radial load pocket. ■ Also verified through simulation (PSS/E), is the impact of additional dynamic compensation triggered within the load pocket that moves the system away from instability.

System impedance, vs load impedance 0.8 1.02 System X 0.7 Load X 1.01

0.6 1

0.5 0.99 0.4

0.98 0.3

0.2 0.97

0.1 0.96

0 0 20 40 60 80 100 120 0.95 0 20 40 60 80 100 120

■ Model-free algorithm, like RVII, requires extensive validation via simulation ahead of deployment ■ A Real-Time testbed, like the one available at TNB, or a batch simulation testbed as used in house is crucial to validate results ■ RVII in a pure model-free implementation is suitable for instability Detection and Extrapolation (both important when close to instability); Prediction (which is important when far from instability) requires a hybrid approach ■ Work being pursued in direction of contingency- based RVII -- showing promising results. ■ Several applications leveraging benefits of RVII results, such as RVII-triggered load shedding, are currently being validated.

Quanta Technology, LLC 4020 Westchase Blvd Raleigh NC, 27607 www.quanta-technology.com