NEW RESEARCH IN POWERLINE BUSHFIRE SAFETY: VEGETATION CONDUCTION IGNITION Dr Tony Marxsen Monday 1 June 2015

Videos of some vegetation ignition tests Test 16 typical still air ph-ph Melanoxylon.avi Test 54 ph-ph Eu. Baxteri.avi Test 58 ph-e Eu Baxteri.avi Test 28 ph-ph A. Melanoxylon.avi 1 June 2015Marxsen Consulting Pty Ltd 2

What happens when a tree branch touches a powerline The four phases of ignition: Phase 1: development of contact between the branch and the wire Phase 2: expulsion of water from the branch Phase 3: charring/flame progressively extending down the branch Phase 4: flashover (if it’s a ‘branch across wires’ fault) These can all be seen on the videos of the tests carried out at Springvale in February and March 2015 in a major powerline bushfire safety research project. In this presentation: Why we did the research How we did it What we found 1 June 2015Marxsen Consulting Pty Ltd 3

Why we did the research Black Saturday fires February 2009 – Victoria’s toll: 173 dead 4,000 injured, many severely Thousands of houses, tens of thousands of animals lost $4.4 billion economic loss The response: Royal Commission 2010 Powerline Bushfire Safety Taskforce 2011 Powerline Bushfire Safety Program 2012 – to present Research so far: 2011 metal-metal arcs near dry grass 2013 auto-reclose settings for metal-metal arcs near dry grass 2014 REFCL Trial at Frankston – metal-soil arcs through dry grass 2015 Vegetation Conduction ignition test program at Springvale 2015 REFCL technologies trials at Kilmore (underway) 1 June 2015Marxsen Consulting Pty Ltd 4

The vegetation conduction ignition test facility: concept 1 June 2015Marxsen Consulting Pty Ltd 5

The test facility: location 1 June 2015Marxsen Consulting Pty Ltd 6

The test facility: 22kV substations SV and SVW 1 June 2015Marxsen Consulting Pty Ltd 7

The test facility: site before start of work 1 June 2015Marxsen Consulting Pty Ltd 8

The test facility: earth grid, rock surface and HV poles 1 June 2015Marxsen Consulting Pty Ltd 9

The test facility: HV supply and HV resistors 1 June 2015Marxsen Consulting Pty Ltd 10

The test facility: HV resistors internals 1 June 2015Marxsen Consulting Pty Ltd 11

The test facility: test rig and sample containers 1 June 2015Marxsen Consulting Pty Ltd 12

The vegetation conduction ignition test rig 1 June 2015Marxsen Consulting Pty Ltd 13

The test rig for ‘wire into vegetation’ tests 1 June 2015Marxsen Consulting Pty Ltd 14

The vegetation conduction ignition test facility 1 June 2015Marxsen Consulting Pty Ltd 15

The vegetation conduction ignition test facility 1 June 2015Marxsen Consulting Pty Ltd 16

The vegetation conduction ignition test facility 1 June 2015Marxsen Consulting Pty Ltd 17

The test design principles: 1. Realism – replicate real powerline faults 2. Worst case fire risk conditions – Code Red day 3. Direct comparison of options – fire risk quantification 1 June 2015Marxsen Consulting Pty Ltd 18

The test program: 1038 tests in 33 days of testing ‘Branch touching wire’ tests ‘Wire into bush/grass’ tests ‘Branch across wires’ tests Sample procurement was a constant challenge 11 upper storey species tested 4 bush species tested 2 grass species tested Data base of test records assembled Fault signatures collected Analysis: fire probability by fault current limit and by species 1 June 2015Marxsen Consulting Pty Ltd 19

First learning: vegetation conduction ignition occurs in four distinct phases 1 June 2015Marxsen Consulting Pty Ltd 20

Test 426: Test 426 Eu. Viminalis Manna Gum ph-e Test 426 Infrared 1 June 2015Marxsen Consulting Pty Ltd 21

Phase 1: development of branch/wire contact 1 June 2015Marxsen Consulting Pty Ltd 22 Each initial contact point produces a collar of plasma around the branch – increased current entry area means increased current flow

Phase 2: expulsion of water 1 June 2015Marxsen Consulting Pty Ltd 23 Occasionally, an extremely noisy process – audible across the test site.

Phase 3: arcs in the flame 1 June 2015Marxsen Consulting Pty Ltd 24 Flame is a good conductor. Electric arcs short-circuit part of the branch, causing current to fluctuate.

Phase 4: flashover Flame is a good conductor of electricity – when it bridges the two wires, there is a direct phase-to-phase short-circuit which trips the feeder circuit breaker on over-current protection. 1 June 2015Marxsen Consulting Pty Ltd 25

Melbourne 2014/15 – a cool summer 1 June 2015Marxsen Consulting Pty Ltd 26 Only one day > 35 C; only five days > 30 C Little evidence of seasonal conditioning of test samples and no Code Red day samples.

Moisture content during the test program 1 June 2015Marxsen Consulting Pty Ltd 27

Sample moisture content 1 June 2015Marxsen Consulting Pty Ltd 28

Average moisture content by species 1 June 2015Marxsen Consulting Pty Ltd 29

Average conductivity by species 1 June 2015Marxsen Consulting Pty Ltd 30

Moisture content and conductivity – the transverse profile 1 June 2015Marxsen Consulting Pty Ltd 31

Ignition occurs in the outer layers 1 June 2015Marxsen Consulting Pty Ltd 32 The phloem and cambium layers conduct the current, the core wood is unaffected.

How bushfires start: conceptual model The conceptual model: First ignition occurs at height (typically 8-10 metres) Embers drop to the ground Dry grass on the ground ignites Wind builds the grass fire into a bushfire The fire risk criteria for branch/wire faults: embers must fall to floor of rig glowing: If small, T > 350 C If large, T > 250 C Wire into bush/grass faults: fire risk is a judgement call based on evidence of multiple factors. 1 June 2015Marxsen Consulting Pty Ltd 33

What makes a species ‘worst case’? Not simply ‘ignite-ability’ Not simply ‘time to ignite’ The key question is: regardless of how long it takes, will fire risk be produced before powerline protection systems can remove the voltage? The method adopted was: Tests were stopped at a current limit: 0.5/1.0/2.0/4.0 amps Fire risk production was monitored The worst case species was that which produced the highest fire risk at a particular current limit. 1 June 2015Marxsen Consulting Pty Ltd 34

‘branch touching wire’ fault tests 1 June 2015Marxsen Consulting Pty Ltd 35

‘branch touching wire’ results 1 June 2015Marxsen Consulting Pty Ltd 36

‘branch touching wire’ tests – variation by species 1 June 2015Marxsen Consulting Pty Ltd 37

‘branch touching wire’ – species ranking 1 June 2015Marxsen Consulting Pty Ltd 38 Species Average fire probability 95% CI min/max Salix sp. (Willow) F. Angustifolia (Desert Ash) A. Mearnsii (Black Wattle) Pinus Radiata (Radiata Pine) Eu. Baxteri/Obliqua (Stringybark) Eu. Viminalis (Manna Gum) A. Melanoxylon (Blackwood) C. Glaucophyllus (Cotoneaster) A. Pycnantha (Golden Wattle) P. Undulatum (Native Daphne) Allocasuarina Verticillata (Drooping Sheoak) S. Molle (Peppercorn)

Factors in species variation: slow current growth 1 June 2015Marxsen Consulting Pty Ltd 39

Factors in species variation: bark type 1 June 2015Marxsen Consulting Pty Ltd 40

So what does a worst case ‘branching touching wire’ fire look like? Test 307 Salix ph-e Test 307 Salix ph-e Infrared Is there hope of detecting and responding to a 0.5 amps fault? REFCL Trial test NER in service REFCL Trial Test REFCL in service 1 June 2015Marxsen Consulting Pty Ltd 41

‘wire into vegetation’ results 1 June 2015Marxsen Consulting Pty Ltd 42 ClassSpeciesTestsFires Bush Rubus FruticosusBlackberry219 Ulex EuropeausGorse126 Kunzea EricoidesBurgan128 Banksia MarginataSilver Banksia60 Grass Lomandra LongifoliaSpiny-headed Mat-rush113 Poa LabillardieriCommon Tussock-grass93 Bush tests: Higher fire risk than ‘branch touching wire’ Detection and response to 0.5 amp fault cuts fire risk by 80% Grass tests: If moist, immediate flashover If dry, no current and no fire

‘branch across wires’ results Current develops four times faster than ‘branch touching wire’ tests due to 73% higher voltage Current path is same as customer load current path, so fault is not detectable until flashover Tests that went to flashover had 55% fire probability, but this is optimistic as real flashovers have much higher current 1 June 2015Marxsen Consulting Pty Ltd 43

‘branch across wires’ results 1 June 2015Marxsen Consulting Pty Ltd 44 Fault detection and response would have to be fast (<5s) for reduction in fire risk

Fault signatures 1 June 2015Marxsen Consulting Pty Ltd 45 Fault

Fault detection: hope for the future 1 June 2015Marxsen Consulting Pty Ltd 46

Next steps Discussions with suppliers on fault signature analysis – the only likely avenue for detection of ‘branch across wires’ faults and SWER earth faults Tests at Kilmore on worst case species with REFCL network protection – to address the 85% of overhead conductor that is non-SWER. 1 June 2015Marxsen Consulting Pty Ltd 47