26 Jun 2018

With Enough Data, Houston-based SensoLeak Is Poised To Eliminate Pipeline Accidents

Written by Jim Pierobon

Artificial intelligence and machine learning are poised to improve how leaks in U.S. gas, oil and other liquid pipelines can be detected long before they become catastrophic accidents.

As America’s network of aging pipelines continues to grow and new pipelines are under construction to transport fracked gas to markets, a 32-year-old entrepreneur and immigrant from Israel – Shoshi Kaganovsky – is scaling up SensoLeak in Houston in a bid to become the go-to leak detection software solution to avoid even relatively minor incidents.

Currently, leaks are detected primarily, if not entirely, by devices called “pigs” that crawl through pipelines looking for safety threats. They are faulted by some industry professionals for not preventing incidents but also generating too many false alarms. Kaganovski claims the SensoLeak software can reduce false alarms to “virtually zero.”

The list of major pipeline accidents grew as recently as June 7 when a leak in TransCanada’s Leach Xpress pipeline in Marshall County, West Virginia triggered an explosion. That pipeline is less than a year old and joins a list of accidents that beg for an industry solution despite a reasonably good track record.

Any gas or liquid pipeline in the world conceivably could deploy SensoLeak’s technology. CREDIT: SensoLeak

In hailing the start of the Leach Xpress pipeline’s operations, a spokesman for TransCanada said it was built as a “truly a best-in-class pipeline.” The spokesman said TransCanada looked forward to “many years of safe, reliable and efficient operation.” TransCanada said it expected to resume operations next month (July).

Data: Risks of Pipeline Accidents Continue to Grow

Among the data suggesting the risks of pipeline accidents are growing is a recent analysis of federal data published by the Associated Press. It found the annual number of “significant” accidents on oil and petroleum pipelines shot up 60 percent from 2009 by 2015. See a list of pipeline accidents since then, below.

That got Kaganovsky, a technology investor and startup consultant residing in Haifu, Israel at the time, sensing a compelling need for an algorithm to detect pipeline leaks. She had been developing software to detect needed helicopter engine repairs but refocused her work. She began developing the pipeline algorithm using “MATLAB” software on Linux computers and shifted to Python open source software to write the computer code.

SensoLeak is a customized program for every pipeline network. The software stays on company computers and works with supervisory control and data acquisition (SCADA) systems. None of SensoLeak’s hardware is involved.

What Other Industries Will Embrace SensoLeak?

Kaganovski’s target markets include any business that uses sensors and collects lots of data to monitor fluid flow and rotating equipment. That includes wind turbines.

Beyond her 26 paying clients, whether one or more major pipeline operators deploys SensoLeak’s software may not be known until 2019 or later. The American Petroleum Institute has what it labels a “robust” system in its “Recommended Practice 1175” designed “to minimize the size and consequences of leak events.” But to detect them in advance? That is the gap SensoLeak purports to plug for oil and gas pipelines.

Kaganovsky has conducted pilot programs with three “super majors” each of which is covered by a non-disclosure agreement. One of the pilot programs successfully completed a proof-of-concept, she said. Meanwhile, she has opened the first round of seeking investors for SensoLeak. She is simultaneously pitching her algorithm to several of the large pipeline owner / operators in the U.S. such as BP, Chevron and Shell. Calls or emails to those and several other operators for their assessments of Sensoleak were not returned.

Her Professional ‘DNA’

Kaganovsky was born in Russia in 1985. After her father – then the Minister of the Economy there — died while her mother was pregnant with her, the family moved to Israel when she was 5. She said was educated in a private school there and grew up to serve in the Israeli army in Lebanon and Syria, ultimately leading a team of about 50 soldiers along Israel’s Northern Border. She said she has five college degrees, including double-majors from the University of Haifa.

Shoshi Kaganovsky at her desk in Houston’s energy corridor CREDIT: SensoLeak

Judy Robertson, an attorney in Houston represents oil and gas pipeline companies but is not affiliated with SensoLeak calls it a possible “game-changer.”

Robertson said two other firms have laid claim to their own detection solutions but neither has secured anywhere near as much interest throughout the industry at this stage of SensoLeak’s introduction.

“I’ve been in the room where a lot of these solutions have been pitched. I’ve never seen anything like it,” Robertson said.

It addition to possibly “revolutionizing” pipeline risk mitigation, “it could give the public and governing bodies more comfort about the safety of pipelines,” Robertson added.

Why no such approach until now?  “Gains in artificial intelligence are now making discoveries such as Sensoleak possible,” Robertson said. But, she added, it’s taken someone of Kaganovsky’s creativity, drive and problem-solving instincts to propose a solution.

“I don’t think she realizes how different she is,” Robertson said.

Here’s how Kaganovsky explains her professional DNA to her 26 employees:  “Be your own hero. Don’t wait for doors to be opened. Rather  . . . collect the keys and open the doors to your own paths.”

Costly Pipeline Explosions Since 2015: A Sampling

Part of the 9,000-mile Texas Eastern system in Westmoreland County, Pennsylvania that ships natural gas from the Gulf Coast to the Northeast exploded in May 2016. It had reportedly been inspected in 2005 and 2012.

Massive flames and clouds of thick black smoke rose over central Alabama following a deadly explosion at the Colonial gasoline pipeline in November 2016.

The Belle Fourche Pipeline in North Dakota operated by Energy Transfer Partners reportedly spilled 176,000 gallons of crude oil into the Ash Coulee Creek in December 2016.

1 Comment to With Enough Data, Houston-based SensoLeak Is Poised To Eliminate Pipeline Accidents

  1. July 15, 2018 at 10:50 am | Permalink

    Understanding that this upstart’s focus is on pipeline leak detection, what are the prospects (if any) for applications in petroleum refinery operations, such as detection of anomalies in
    rotating equipment (e.g., FCC wet gas compressor)?

Leave a Reply