by Divina Paredes

Cobus Nel of Transpower: Accelerating transformation and innovation for the digital era

Apr 17, 2019
APIsArtificial IntelligenceBig Data

Experimentation and iteration is a core part of problem solvingCobus Nel, Transpower

“We are a large critical infrastructure company, which is typically risk averse, and this is not necessarily a bad thing,” says Cobus Nel, GM information and services technology at Transpower.

Indeed, Nel has shown that working in a traditional, asset-heavy organisation is no deterrent for him and the IST team in harnessing emerging and disruptive technologies to tackle some of their biggest operational challenges.

An important facet of their work is being open to partnering with non-traditional firms, such as startups,in deploying these new technologies.

This is all in support of helping Transpower “keep the energy flowing” across New Zealand.

Transpower, Nel explains, is the national transmission grid owner and operator. They manage 11,000 kilometres of transmission lines and 174 substations.

“We stand together to keep the lights on for New Zealand – planning, building, and operating the system that powers our country and lives 24 hours a day, seven days a week,” says Nel, who spoke at the CIO50 event in Wellington.

Their industry is heavily impacted by changes not only in technology, but also environmentally.

“Climate change is the defining issue of our time,” Nel declares.

“We rely on access to energy to power our way of life, but traditional forms of energy are compromising our climate, and the world we will leave to the future generation,” he says.

To realise a more sustainable energy future, he says New Zealand will need to embrace rapidly emerging energy technologies.

Transpower forecasts that electricity use across New Zealand will double by 2050.

No caption

“But we need to supply this electricity from somewhere, so se need a lot more renewable generation”, he says.

This is likely to come from wind and solar (combined with batteries), with some more geothermal in the mix. This, however, does not mean all of these will be centrally generated.

Nel says IST has a critical role in supporting Transpower in its goal of realising a more secure and sustainable energy future.

“It’s important that we don’t lose sight of the ‘Energy Trilemma’ when we look at our future.”

This means balancing the needs of affordability, reliability, and sustainability.

He also reveals some of the challenges they face.

These include security of their assets, driving efficiency to ensure competitiveness, better asset management practices to build and maintain their assets to a suitable level and usable life, and in general, gain better insight from the data they collect.

As he notes, many of these problems are now becoming too complex to solve using traditional software engineering methods.

No caption

Working with startups

According to Nel, accurate interpretation of data is ever more crucial for making asset management or operational decisions.

This is where Transpower’swork with startupsis gaining traction.

For example, it is working with a Christchurch-based firm Isogonal to develop risk-based decision-making models usingmachine learning.

Transpower’s work order management system contains multiple descriptions entered manually by their service providers for equipment defects.

These work order descriptions are data rich, but they are unstructured and problematic for performing meaningful and reliable analysis.

“With the large volume of open work orders, manually processing and interpreting them is not practical,” Nel explains.

No caption

To solve the types of large scale, highly complex, and connected problems we are dealing with these days require working collaborative with our customers, stakeholders, and domain experts from different industriesCobus Nel, Transpower

Recent advances in machine learning methods for extracting information from natural language have enabled them to automatically interpret work order descriptions, categorising them systematically and consistently.

This involved cleaning up the descriptions, which might contain spelling mistakes, abbreviations, and a wide range of industry-specific vocabulary used to describe defects.

“We then constructed a set of hierarchical machine learning models to map work order descriptions to an asset and defect ontology created in parallel with the modelling process,” Nel says.

The asset and defect ontologies (an ontology is a set of categories along with a description of their properties and the relationships between them) describe the asset at risk.

Once the ontologies were established, they assigned each asset and defect pair with a likelihood of failure for five different failure consequence dimensions – service performance, direct cost, public safety, worker safety, and environmental impact.

Machine learning was also used to predict the estimated cost for rectifying each defect, using historical data.

The likelihood of failure and failure consequence gave a relative risk value for each work order, which along with the estimated cost, was mapped to an overall work order priority. This enabled them to do risk-based prioritisation of work orders.

Transpower also works with a company called Intela in Wellington to explore using AI for inspecting transmission lines.

It is currently difficult and expensive to assess the condition of Transpower’s high-voltage conductors because it relies on visual inspections, either from a helicopter or by capturing images which are then studied by experts.

Nel explains it can take them two to five days to assess 2000 images. “The job is like watching a clock tick.”

Intela is assisting in developing a prototype that analyses images using machine learning to detect defects. The system’s intent is to rapidly identify, classify, and report the identified defects, thereby reducing the time and cost of inspecting transmission lines, while increasing accuracy of reporting.

“Better and timely information would let us manage these assets better,” Nel states.

Transpower is using both human and machine learning analysis to start applying these prototypes sooner, while continuing to develop new functionality. Ultimately, they expect machine learning systems to be cheaper, faster, and more detailed.

He also talks about a different partnership on emerging technologies, this time with Massey University.

Transpower has two substation robots: Wall-E and Evie, which provide real-time situational awareness of their remote sites.

The robots are semi-autonomous and can navigate around an outdoor substation to points of interest, allowing engineers or operators to respond in a safe way to faults or undertake inspections.

No caption

An area they are working on is fostering innovation across the group.

“Our people have really great ideas, but we need a governance or clear framework on how to assess then filter which ideas to progress,” says Nel.

Transpower is working with Creative HQ to build in-house capability in the innovation space.

According to Nel, the programme aims to help the organisation focus their innovative energy in a disciplined way to solve big problems and upskill their people at the same time.

They also want to demonstrate to the staff thatinnovation is a discipline, and identify people within the business who are well suited to working in this new manner.

“The linear ‘expert’ model for solving problems no longer works in the 21st century,” he stresses, on the driver for this change.

“To solve the types of large scale, highly complex, and connected problems we are dealing with these days require working collaborative with our customers, stakeholders, and domain experts from different industries.”

He shares that the key themes they are instilling across the team and with the wider group are:

  • Innovation is a distinct skill and capability (rather than a by-product of general business/administration skills);

  • Experimentation and iteration is a core part of problem solving;

  • Working in multi-disciplinary, cross-functional teams – fully focused on one initiative at a time is key;

  • Innovation and collaborative work requires discipline – proven methodologies, formats, data/evidence focus and time boxing.

He ends with a key message on innovating in a ‘volatile, uncertain, complex, and ambiguous (VUCA) world’:

“‘Failure’ is not the opposite of success. It is part of success.”

No caption

Get the latest on digital transformation: Sign up for CIO newsletters for regular updates on CIO news, career tips, views and events. Follow CIO New Zealand on Twitter:@cio_nz

Send news tips and comments to @divinap