ETS: Sensory networks
The two BP guys who work in a small technology organization working for the CTO are delivering this keynote. This organization is separate from IT, which as centralized operations but is otherwise decentralized. One described their job as being "shit disturbers," because their job is all about change.
There role includes:
- Technology Transfer
- External Relations and Events - "bluechalk" events - inform 40-50 executives out of around 400 about a technology topic that's applicable to them over two days. The cost runs $50-100K, including travel and expenses for externals. No vendors.
- Technology Integration
- Game Changers - $150 million or more impact
BP gets the worst return in this space on open supply chain stuff, like the Walmart initiatives. They get a much better return on closed supply chain, like LPG cylinder tracking. Asset condition management is a good use of read/write RFID tags. It eliminated the problem of folks photocopying the bar codes from their route, and then sitting at their desk and scanning them. Remote tank monitoring has improved the efficiency of supply chain by a third, because you know when you have to send a truck instead of guessing. They also now track over 16,000 rail cars - where they are and their status. Finally, the use a wireless mote mesh network for rotating equipment vibration monitoring.
The next evolution of this pushes decision making to the edge. Motes on their 16,000 miles of aging pipeline infrastructure provide geo-fencing - situational awareness at the site. Alerts are only sent upstream when certain threshold events occur.
Looking ahead, they are working on the CoBIs "smart item" initiative. The focus is on safe storage of chemicals - what quantities, proximity to reagents, etc. Motes on these storage barrels should know what they contain and what their storage requirements are and then talk to one another to ensure that they are in compliance and provide alerts and durations when they are not.
They see sensory networks as similar to the Internet as a major driver of change in the years ahead. Combining handhelds with sensor networks is very powerful. They provided a great example of how these were used during a plant maintenance shutdown, what they referred to as a "turnaround."
Looking ahead, they are trying to move from reactive to proactive. One thing they are trying to do is to move predictive maintenance to the end device. Various BU's are actively engaged in the following, although they are not as far along:
- advanced predictive maintenance - tied to risk
- ship movement optimization - when to you divert?
- well relationship model - can you understand how different wells in the same field effect one another?
This talk has me pondering our organization...