Beyond automation: Humans as process controllers
CE: It will have to be some sort of iterative search optimization. There’s not going to be a neat tenth-order differential equation that relates inputs to outputs, which is the fundamental premise of virtually all regulatory control.
The other issue when you start to look at multiple-objective control and optimization is historically, linear and non-linear programming with multiple constraint functions requires a lot of computing power to process things and come up with optimal settings. As the time gets shorter and shorter, how do you get that amount of computing power in a cost-effective way? If you think about it, there are some really fun challenges. I think the golden age of the control engineer is almost upon us.
CE: I think the golden age will be when we can start to use some of these new techniques to control our processes, instead of following tradition and returning to PID as we have for the last 60 years.
The thing that I worry about, and I’m speaking in large generalities here, is if it truly is a cascade control structure, and we cascade profitability to efficiency, we have to remember the old concept that the secondary loop of a cascade controller has to be four times faster than the primary loop. As the business variables get faster and faster, we may find PID control of the process becoming a constraint. The process could become chaotic if the business variables move faster than the periods of the loops on the plant floor. That will be a fascinating thing. If that starts happening, we’ll have no choice but to look at other methods of control.
CE: That’s a natural extrapolation, but thinking in terms of costs and business variables changing faster than flow or pressure blows the mind.
Put on a ticker for commodity prices some time. When you see the price of copper changing every 32 seconds, can you control a temperature loop much faster than that? So ask yourself, why don’t we have a control problem right now, because some of these variables have gotten themselves to the speed where they’re actually breaking into the classical temperature and level domain, if not the flow and pressure domain. The issue ends up being that instead of worrying about the problem, we’re ignoring it and we buy our raw stock in bulk so the inventory gives us the buffering. We just don’t worry about it.
Where it will all blow up, I believe, is when some business realizes that it can have a huge competitive advantage by playing the price of its raw materials effectively. When that one company does it, everybody else is going to say, “Wait a second, in order for us to survive in this environment, we have to think differently.” I don’t think we’re far, time-wise, from that happening when I see sites like Momentive Chemical in Deer Park. No on-site inventory combined with real-time acquisition and distribution of product via pipeline. It fascinates me. Huge amounts of money are at stake.
CE: This sounds interesting, but as a practical matter, how much leeway will, or can an operator have to change what’s happening? If the plant is in a sold-out state and the objective is to create as much product as possible, when does it become practical to start trying to fine-tune the parameters?
We have to look at the dynamics of the process itself, but I think the decision criteria could get complicated and it may be impractical if the dynamics of the physical process are slower than the business process. That could get into some very difficult things.
The issue is this: We’ve been working in a world of sold-out everything. The primary philosophy of business has been to make as much as you can in a given period of time, and life as we know it is good. That’s what everybody has been doing. The best example is the power industry. For years, the way the power industry worked was they had central station, coal-fired power plants that made as much power as they could 24/7. Pricing was regulated so they always made a profit, everything was good, and nobody could imagine anything different.
Then what happened? They deregulated and the grid opened up. That means all of a sudden there were, by law, other power suppliers, such as windmills, photoelectric, and co-gen plants, all these sources of power, which if they wanted to send power onto the open grid, they had to be paid money, based on the current value of electricity on the grid. You went from a situation where you had a single producer producing for a series of users, to where you had multiple producers producing for multiple users, and the supply side is going up and down at a higher rate than anybody had seen it go up and down. You might have a co-gen operation at your plant, and for the next three hours you don’t need its output yourselves in the plant, so you put it out on the grid. Now the supply of electricity goes up and people have to adjust. That’s what the smart grid is all about. We’re in a brand new dynamic environment, which is also fascinating because you can’t store power. So with no inventory buffer, what’s going on? You’ve got real-time strategy—we actually had to go to a new structure of the generators, the producers, the grid managers, and the consumers. All of a sudden you have a very dynamic environment. So what did the energy producers have to do? They built new types of plants. Combined cycle plants that could start up and shut down quickly, because if there was a demand on the grid, and the big energy companies couldn’t provide enough power to meet the demand, they were penalized, so they had to have those combined cycle plants start up. Those plants produce energy at much higher cost than the coal-fired plants, so all of a sudden you see these big swings in price. In response to that, even at the home level, we’re going to see things like thermostats that read not just the temperature, but the current price of electricity. You’ll put wet clothes in the dryer in the morning, and it will turn itself on when it sees that the price of electricity is low enough.
The power industry is an extreme, but it’s where all industrial operations are going. Trucks, trains, and material in pipelines don’t travel as fast as electricity, but the dynamics of these systems are going to drive a true reconstruction of how we manufacture. I can picture in the future, where instead of building mega-scale Texas City refineries, we start building smaller refineries that can be more agile, just like the combined cycle plants in the power industries. Hydrocarbon companies won’t run a plant at a time; they’re going to be running their asset set across the entire value chain because if the price of crude is too high right now, so that I make gasoline at a loss, I should slow down production, even if I’m in a sold-out condition. If you can’t play that game, you’re going to lose money during these difficult periods. They are already hitting the power industry, and they’re going to hit the rest of our industries. The ones that can play that game are going to be profitable, and the other guys are going to be out of business. I believe this is where it’s going.
The day of the mega-scale coal-fired power plant is behind us. The day of the mega-scale refinery may be behind us. Maybe our manufacturing philosophy has to change. Maybe we have to go to more localized micro-manufacturing in order to match the dynamics of the market place. Those are the kinds of things that we’re going to have to start thinking about. As an industry, we came up in the era where everything was sold-out, all you had to do was make as much as you can, and margins were huge. That time is behind us. Now, margins aren’t huge and everything isn’t sold out. Some times of the day we’re making money, and other times of the day we’re losing money. The world has changed, and it’s going to require a different set of strategies. I think at least initially, some of these strategies will be to use the people in these production operations in a more effective way. At least initially, we’re going to see some manual controls involved in profitability, with the safety of the operation, and with the environmental integrity. As we learn more and more, those manual controls are going to have to be replaced with automatic controls, from a purely practical point of view, because they’re going to become too fast for humans to deal with.
Here’s another indication of how things are changing: When you think about the chemical industry, the average tenure of a CEO in a chemical company is somewhere between 13 and 20 months. That’s a frightening concept. That means that these people have 13 to 20 months to prove themselves. So what do they do? Everything is designed for short-term gains, and it’s driving them nuts. It’s all going to catch up. Sooner or later, the group that’s looking beyond that and looking for the root cause is going to become more profitable, and those people that are playing the short-term gain game are going to be left by the side of the road.
I don’t believe there are many instances on the plant floor that can be brought under automatic control where we should go back to manual control. We should be trying to control as much as possible in an efficient, automatic way, while at the same time involving people more and more in the operation because of the business dynamic.
Edited by Peter Welander, firstname.lastname@example.org.
- The number of variables that have to be considered in a control strategy is growing in a way that includes more business-related elements.
- Control strategies that are adequate for running the process may not be capable of controlling new variables.
- Human controllers may need to fill the gap between needs and capabilities.