Rick Jeffress, Director, Business Development, Fike Corporation, explained how by using Video Analytics, operators could minimize fire risk referring to company’s solution during the 2016 SAFETY4SEA Conference & Awards. He stated that video cameras are widely used in shipboard machinery spaces; Fike Video Analytics processes the video from these cameras, mounted in various hazard areas, to intelligently monitor and detect the unique signatures of hazardous conditions. A shipboard camera network combined with Fike Video Analytics creates a highly efficient means for rapidly detecting oil mist, a precursor to a potential fire or explosion. The system can be monitored by shipboard personnel and linked to automation and safety management systems, allowing critical time to take action before disaster occurs.
There are currently rules requiring oil mist detection inside engines, however oil mist also can create an extreme hazard outside of the engine. Fike video analytics monitors video from off the shelf cameras that are installed inside the engine spaces and analyze that video through software for oil mist, flame, reflected flame or smoke in the space.
This video (see screenshot below) shows an example of early oil mist detection, , allowing crew time to turn of fuel supply and activate the water mist system to suppress the explosive mist. There can be an internal fuel leak in the engine, oil mist is created when it hits a hot object creating an oil mist or vapor which is very explosive. When the analytics detect the oil mist, the software sends an alarm signal to the engine control room with live video where the crew members can identify which engine to activate the water mist and shut down the before there is an explosion.
Most of the fires on ships are caused by oil leaks or oil mist in the engine rooms. Oil mist is created by minute leaks in fuel lines, when released under pressure give off a very fine atomized spray; or when fuel hits a hot surface and boils, generating an explosive mist.
The use of video analytics is widespread in manufacturing and security systems and has been around for decades. For example, video analytics can be used in “heat mapping” monitoring a retail facility for customer traffic and dwell times; in facial recognition; in product manufacturing and quality control e.g. for monitoring that caps are on correctly and even in video game technology.
Video analytics uses visual imaging and trained artificial intelligence for fast processing of information, to detect oil mist, fire and smoke very early. The analytics were trained using videos of real fires and false fires. Although megapixel cameras are used, the processing is performed at 640X480 resolution, which is 300K pixels, we analyze 15 frames per second, therefore the analytics are performing 4.5M calculations per second.
Currently, there are many shipboard applications of video analytics as oil mist detection systems. For example, a major cruise line requires all of its ships to monitor engine room cameras with video analytics. We are monitoring approximately 3400 engine cameras on ships throughout world. . Ford uses the system in its engine test cells and we have 600 cameras in a catering facility, as an example, installed as primary detection rather than using spot type smoke detectors. Video analytics is approved as primary smoke detection in many countries and installed in facilities such as museums, NASA clean rooms, manufacturing facilities, chemical and power plants.
Detecting oil mist or smoke is difficult in engine room spaces due to inherent air flow and obstructions in the ceiling that impede smoke from entering conventional smoke detectors. Video analytics monitor and detect oil mist, smoke, flame and reflected flame within the entire camera field of view.
This video shows an engine repair center in Houston (see screenshot below). There is an injector spraying oil mist to determine whether the camera can detect oil mist. You can see the clearly defined pattern of the oil mist. The blue lines represent standard detection while the clear tiles represent higher sensitivity detection. When the analytics artificial intelligence determine that there is an actual event, it turns on the higher sensitivity to hold the mist longer.
Video analytics software learns the cameras’ field of view and monitors for characteristics of oil mist or spray known to cause flame or explosion. Upon detection, an alarm can be signaled in the engine control room or anywhere else on the ship.
This is an overview of a CCTV system onboard a ship. Off the shelf IP cameras are connected to a CCTV Ethernet LAN along with a video management server and video management workstation displaying live video on engine control room monitors. The Fike servers are installed on the LAN and analyze video in parallel to the existing video management system and display a pop-up video of the detection event with analytics overlay representing the oil mist or flame detection. Relays can be configured to be monitored by the ship’s automation system and an alarm light for visual annunciation.
Video analytics are commonly used in the following areas:
- Diesel Generators and Diesel Propulsion Engines
- Oil Fired Boilers
- Incinerator Room
- Purifiers and fuel modules
- Emergency Diesel Generator
- Auxiliary Emergency Diesel Generator
- Mooring Deck
Another example is that of natural gas compressor station flame detection, where we installed the system. The system recorded three hours of video of an event as was detected prior to connection to the fire alarm system. The customer found out, by reviewing the video,that they have fires in the facility that nobody knew about and that operators were not making their rounds on an hourly basis as required.
Video analytics can detect flame from reflected surfaces such as bulkheads, decks or engine room equipment. The analytics do not need to see the fire directly as can detect the flickering reflection. The analytics are monitoring for flickering light and slow changing brightness, affecting the pixels and dynamic flicker. Purple dots represent reflective flame. This is useful in an engine room where there are a lot of obstructions that UV/IR flame detectors cannot “see”.
This is an actual oil mist event (see screenshot below), oil released inside of the crankcase boils when contacting the hot engine, creating an explosive vapor. They were actually able to shut down the engine and discharge water mist before an explosion; oil mist does not just catch fire, it explodes violently. This was early detection; you can see the clear outline of the oil mist.
Smoke and flame detection can also be used in mooring decks and other areas of the ship. When there is a detection,, there will be a pop up of the display with live video. The video analytics’ video management system creates a pop-up in real time and the time, date, type of event and duration are all displayed.
The system also records video 24/7, not only when there are events but also tracks the events. A timeline allows events to be easily located, viewed and video files downloaded.
The views presented hereabove are only those of the author and not necessarily those of SAFETY4SEA and are for information sharing and discussion purposes only.
Rick Jeffress, Director, Business Development, Fike Corporation
Rick Jeffress resides in Houston, Texas and oversees global business and application development of the Fike video analytics detection product line. He has over 24 years of industrial and marine fire protection experience with complex fire alarm and suppression systems’ design and application for marine vessels, petrochemical and power plants and offshore oil and gas projects.
Rick worked closely with a major cruise line technical safety staff, through pilots and onboard testing, to apply Fike video analytics to the marine engine room environment for early detection of atmospheric oil mist, smoke, flame and reflected flame. The technology is now part of safety directive for some of the world’s largest cruise lines.