AI technology to identify problems with onboard machinery

In The News

The interview was published first in Digital Ship magazine.

Interview with Will Roberts, CEO of ioCurrents based in Seattle, discussing how companies identify problems with onboard machinery using predictive analytics

Seattle-based software company ioCurrents, has developed digital tools which help shipping companies identify problems with their machinery.

Their proprietary platform, MarineInsight, does this by gathering large sets of sensor data, which can be analyzed to identify ‘normal operation’. If data patterns change, the system can provide useful early warning of a problem.

MarineInsight measures and assesses data from across an entire vessel: “From every sensor that we can find on board the vessel that has an electric signal, we can also add more and then put them together.”

In the case of one fishing client, MarineInsight was used to determine when generators should or should not be switched on. Alerts are now sent to the captain and shore-side personnel when the generators are running under recommended- ed loads so they can be switched off.

After installation, the generators’ fuel consumption was reduced by 40 percent, and periods between maintenance checks increased by 60 percent, resulting in the client saving $70,000 (USD) per year.

In another case study, a client’s main engine overheated during a port call. The issue was undetected by the crew or the engine control module but was spotted in MarineInsight’s data analysis. The system sent the crew an alert, enabling them to take action to avoid an engine failure. Later data analysis, including comparison with historical data, found that the ship was docked incorrectly, putting back pressure on the exhaust.

A third client, Western Towboats, requested MarineInsight to identify reasons for it suffering 3-4 engine failures a year. The client identified the immediate cause as a high-pressure diesel pump injecting metal into the engine but did not know why it happened or how to prevent it.

After investigation, MarineInsight identified other elements of sensor data that could predict pump failure before it happened. The formula was converted into a continuously running alert tool, predicting when a pump was likely to fail so the crew could take corrective action.

Since the alerts were first set up in 2015, Western Towboats has not experienced any further pump failures, saving the company upwards of $75,000 (USD) per incident.

During an interview with Will Roberts, the Chief Executive Officer at ioCurrents, Digital Ship explored the system provided by the company.

How the system works

MarineInsight’s data analytics platform uses AI and machine learning models. MarineInsight utilizes IoT devices installed on vessels to perform real-time analytics, generating analyzed and compressed data.

This data is transmitted to a secure cloud and accessed by ioCurrents’ shoreside

teams. These shoreside teams can effectively harness “an almost infinite amount of data coming off of the equipment.”

This data can then be used to “look into the future and see when something might not be operating as efficiently, or might be deteriorating, or there might be something we need to troubleshoot”.

Prior to hardware installation, ioCurrents collaborates with teams onboard the vessels to establish Key Performance Indicators (KPIs) covering areas such as fuel optimization, failure prediction, and enhanced maintenance.

The IoT system requires less than 24 hours for installation, without the need for drydock, and all configurations are handled remotely. Shoreside teams gain live access to the MarineInsight system shortly after installation, ensuring seamless integration with operations.

Once the system is installed, MarineInsight can alert teams on vessels to potential hazards by “taking the signals and sending them to shore, we can set new limits so that we can direct shore side and the crew to alerts that they might not have seen in their alarm panels.”

Using the system

Potential warning threats are communicated with the crew through “text, an email, or phone call from our teams to the vessel operator.” These messages are sent with a hyperlink allowing crew onboard vessels to dive straight into the ship’s data. This means that the crew “can look at the trend and can look at how that’s interacting with other ship operations at the time.”

“The hyperlink will take you to the data, and the data that you’re seeing is displayed in the same format that you’d be used to seeing on board the boat, so it’s very easy and intuitive for the end customer to use.” Roberts stated that the system is designed to avoid becoming a “burden on normal operations.”

Crew members and shore-side personnel receive guidance on navigating the system’s interface, allowing them to make informed decisions during operations.

Currently, Roberts views the system as one that guides the experts onboard a vessel toward an issue. The system “guides users to a pretty specific area, and then we give operators all the surrounding parameters that are happening at that same exact time. If it’s the temperature of a gas, then we’re also relaying the pressure of the gas and the temperature of the cooling water”.

Roberts stated how this approach allows for “vessel experts to get to the solution incredibly quickly”.

AI and predictive analytics

By assimilating huge amounts of data emanating from onboard vessel equipment, MarineInsight uses predictive analytics to predict potential equipment failures.

The analysis not only predicts potential equipment failures but also optimizes vessel operations by discerning patterns of normal and irregular behavior. Roberts expressed the positive impact of such instances, stating, “We’ve been able to catch a number of issues before they actually created an actual incident. A couple of them have been pretty big for our customers and make them very loyal.

The system is also vessel specific, even if a fleet owner has “90% the same equipment in their fleet, our solutions and our guid- ance remain very vessel specific”. This is due to the differences and nuances between the same type of equipment on different boats, which AI helps highlight.

Roberts believes the vessel-specific approach suits vessel owners. Expressing how “right now customers want granular data about how to make their vessel better.” The system assesses the optimal level of performance for systems onboard vessels, resulting in finely tuned optimized levels of performance.

Looking ahead, Roberts stated, “We might have a half-dozen models running on every ship that we’re looking at right now, and we’re letting our machine learning figure out which model is the best.” The reason is that “the more data we get and the more vessels we have, the smarter our AI will be.”

Cargo monitoring and tank alerts

By delving into sensor data from cargo tanks, the system provides insights into fuel usage, product delivery, and contractual compliance. MarineInsight assists in optimizing vessel operations based on real-time data, accounting for the vessel’s specific load.

The system works through “sensors within tanks which provide feedback to shoreside or to the vessel.” The system can, for example, “optimize speed based on the draft; if we know how much product is on board the vessel, what the draft is, we can see if the equipment has to push harder or not. We can help optimize vessel operations based on very specific criteria and what the vessel load out is at that time.”

Connectivity at sea

The first level of data analytics is done onboard the vessel, meaning the system does not rely on a constant internet connection.

The predictive nature of MarineInsight’s analytics further mitigates the urgency of real-time data transmission, offering a balanced approach to data accessibility.

The system avoids connectivity issues by ” storing the raw data on board the vessel, performing first-level analytics, and then sending it off in small batches continuously.”

Roberts stated despite not being able to “tell you about an issue (within the minute), if I can’t get that information off, we keep that on board and can transmit it the next time it’s possible. We are also working to increase the time between letting the vessel know of a potential issue and when that issue will become a real problem.

A custom approach

Given the diversity of vessels, Roberts believes MarineInsight is distinguished by its user adaptability. Rather than a broad-brush approach, the system offers maintenance optimization solutions to each specific vessel.

With much of the market focus on generalized solutions, MarineInsight appears to offer a customized alternative. Regarding vessel-specific issues, Roberts stated, “We see that there are differences and nuances between the same type of equipment on a different boat.

Continuing, Roberts stated, “Right now, our customers want specifics. They want to know what’s happening on their very specific vessel, and that’s what we do.”

Despite the vessel specific approach of ioCurrents, Roberts conceded a future aim for the system is to compare data against that of other vessels as the company “gets a mas- sive amount of data and we’re [ioCurrents] getting smarter about generalities”.

But for now, Roberts maintained, “it’s best for us to look at that exact vessel to provide very specific information to our customers about their operations.”

Roberts claimed that MarineInsight systems can extend service life by “25-50% over the [vessels] lifetime” and help deliver “double-digit fuel reduction.”

“The capital cost of replacing equipment to get whatever the newer fuels are is incredibly expensive. Whereas there is a tremendous amount to be gained by optimizing your present operations”.

Given that “major equipment failures are putting a huge stress on the environment”, the system’s ability to prevent the failure and replacement of systems has huge envi- ronmental potential.

In addition to preventing system failures, MarineInsight systems can direct users to “which systems need replacing first to provide greater [user] benefit.”