Chris and Don's Fireside Chat...

Chris Lyden and Don Clark have been around. They've worked for the other guys. Now they are part of the bull-moose brain trust that Invensys has put together to help them survive the coming decade. Some of the others are Peter Martin and Hesh Kagan. All old automation guys who've been there, done it, and given the smelly old tee shirt to goodwill long ago. At a special press briefing tonight, Chris and Don put forth their vision for the future of automation. I disagreed with them once and once only. Otherwise, it was clear that either they've been reading Control or that the three of us have been reading the same tea leaves. "This is kind of a fireside chat," Clark began. "Yeah, so turn the lights down," Lyden cracked back. Lyden began by saying that they'd just finished the strategic planning for the year, and he was going to London tomorrow to present it to the Invensys Board. "Now that most of our financial problems are behind us," he said, "we can take a real look at how we see the landscape changing." "This really should be called 'What Keeps Us Up at Night'," Lyden finished. There is a huge change coming to automation in all of the industries we serve and practice in, they began. It is going to radically change the way we run our plants, and they way we think about automation and the place automation has in the enterprise. Clark began with the Purdue Automation Model, which was created in the 1970s and 1980s under the direction of Ted Williams, a member of the Process Automation Hall of Fame. It is a five layer model from the field device layer to the enterprise layer. Remember this...it comes back around later. In 1966, a typical refinery had maybe 5000 points of I/O and a ratio of about 1.5:1. In other words, just about every output had an input associated with it. After the introduction of the DCS, refineries went quickly to 20K points, with a 4:1 ratio, all the way to 50K points, with a 10:1 ratio, and now, today, the average refinery has well over 150,000 points of I/O, with a 50:1 ratio of I to O. We've been leveraging the I/O all this time. Lyden sees what he calls an "autocatalytic effect" coming...the more the I/O the more things can be done with it, which will drive more I/O... He claimed that what is coming is a bifurcation. Instead of the traditional five layer model, field sensors, basic automation, advanced process control and optimization, plant intelligence and enterprise, there will be a collapse of the layers into three, maybe two layers...field devices, ubiquitous control. Lost cost sensors, coming soon to a popstand near you, will proliferate and facilitate application growth, fueled by I/O growth. After paying homage to John Boyd's OODA loop (go look him up here, Lyden and Clark pointed out that there are a heck of a lot more things going on in a plant than single loop control. "Wouldn't it be neat to be able to do a query across all your plants to see if, for example, Goulds pumps performed better than Flowserve, or vice versa?" Cross collaboration will grow. Plants may decentralize. New Human Interface metaphors will emerge that will facilitate cross disciplinary collaboration. Instead of the HMI of today, we may look at a real-time 3-D and 4-D plant, and be able to zone in on whatever is of interest. Data mining tools will facilitate this. Lyden pointed out, again as have I, that 50% plus of all technical workers in manufacturing will retire in the next five to seven years...leaving with all that institutional knowledge. Lyden thinks, and here is one of the areas in which I disagree with him, that the remote heap big expert will be able to run multiple plants remotely and make up for the lack of local talent at each plant. He, of course, is looking at this from a supplier's point of view who is interested in providing those remote experts...and Invensys is having just as much trouble recruiting people who know what they are doing as the rest of us are. So where are these heap big experts coming from? Lyden and Clark believe that wireless will make sensors cheaper and ubiquitous. I agree. Fiber backplanes will make sensor networks fast enough to handle thousands of sensors in real time. Together, sensing will be ubiquitous and inexpensive. Now what do we do with the data??? What's going to happen is that companies will bifurcate into sensor companies and application companies, and what is necessary is a framework for the absolute requirement of interoperability between applications. These applications aren't going to be the traditional DCS applications, either. Networked control will be ubiquitous, like everybody has a watch. It will still be necessary, and it won't be the "Ginzu knives of process control," either, as Don Clark pointed out. But there will be applications running in the framework that are in background, sometimes in deep background. Clark and Lyden gave the example of a pump that comes with a VFD, that is entirely already instrumented wirelessly, to put out signals for temperature, vibration, differential pressure across the pump, speed, flow, and all the other variables of interest, and an application that monitors all this stuff, and only intrudes into the decision space when something is necessary: an alarm because a variable has gotten out of spec, that sort of thing. More and more, devices are going to come with full kit, rather than it being necessary for the plant instrument people, or a system integrator, to add the necessary sensors. This is a compelling vision-- one shared by me, and by lots of the other automation vendors and automation professionals around the world. Lyden and Clark's vision complements the vision shared by Yokogawa at the Tech Fair last week-- where very high speed networks and small, MEMS based sensors, coupled with decision support software, make it possible to go from running the plant itself, to running to profitability instead. This is the ultimate message from Chris Lyden and Don Clark-- we will finally be able to concentrate on running the business, rather than running the plant. Lyden says this will happen within three to four years. I disagree. I think he is wildly underestimating the incredible risk aversion and conservative nature of the process automation space. He says that maybe I'm right, but that it is already happening in the "non-boom" industries-- pharma, biotech, fine chem, pulp and paper, etc., and to prove it, he noted that Invensys has nearly $240 million worth of InFusion business in the pipeline since introduction in May-- "not bad for six months," he said. It is clear that this vision, which I share, will happen. The technology will happen. The question is whether we can change the cultures of manufacturing plants to be able to do it. It may happen. I told Chris about the latest from his very close friend and former co-worker, Michael Saucier at Transpara...that the Visual KPI system has jumped out of the plant operations layer into the business layer at Keyspan Energy with the spot buyers and sellers of energy using it to give them a heads' up whenever they will have more or less power to broker. It may be that applications like that will cause the tipping function Chris is expecting to happen closer to his time frame than mine.