How many people will be running your plant?
Process manufacturers have to deal with basic head count challenges. The answer may be fewer people than you think, but how do you find the right number?
The question of how many people are necessary to run a process plant is not a new one. Producers in competitive industries have been trying to figure out how low they can go since the earliest days of automation and probably before. It has taken on new significance in recent years as companies are dealing with the great shift change as baby boomers retire and millennial workers take their places.
Naturally, the universal answer applies: "It depends." Some of the relevant factors include the nature of the plant itself, its level of automation, and the types of people that make up the immediate workforce.
The conventional wisdom says that qualified engineers and operators who are happy to work in a typical process plant environment, say an oil refinery for the sake of argument, are going to be fewer in coming years. There’s no doubt that people will be available, but the ones who have relevant education and training will be scarce. Companies may be able to outsource some of that need (see sidebar), but having the right people on-site, 24/7, will be a challenge.
To answer the question, or at least fill in some of the qualifications, Control Engineering turned to two industry experts who spend a lot of time thinking about such topics. Stan DeVries is senior director, solutions architecture, and Peter G. Martin, PhD, is vice president, business value consulting, both with Schneider Electric. Control Engineering contributor, Peter Welander, asked the questions.
CE: So, how many people does it take to run a plant?
DeVries: Some of the customers I’m dealing with now are investigating that question for several reasons. It’s more than simply cost. People have been talking about the "great crew change" for a long time. But this is the first time in my experience that we have hard data from a customer, driven by their human resources managers, saying that when the aging workforce leaves, the new mix of people, whether they’re hired or contracted, won’t be the same number. It will be far fewer, and they won’t be the same kind of people who will take on the same ways of doing work. That’s not just knowledge going out the door, it’s a whole culture change. Everybody’s dynamic differs, but they’re saying that it takes so many years for their wave to completely go, and it will take them the same number of years for them to get ready to do whatever they think is the right thing to do. It’s not just people leaving. It’s not just a productivity issue, but asking a deeper question: What can the workers of tomorrow do to run a plant safely and effectively?
Now, for petroleum refiners, one of the KPIs that Solomon Associates has people subscribe to is a personnel performance index. You take the number of people you have and compare it to some sort of business measure, unfair as it might be, that you can benchmark against others. That’s dangerous with gross benchmarks, but it’s an attempt to establish a productivity measure and determine how that’s shifting from year to year.
CE: So regardless of the relevant measures, we have to find new ways to do things to replace the people who won’t be available?
DeVries: Yes. It extends to the maintenance of the technology. So maybe you say, "I have the answer-I’m going to throw more technology at the plant." You have to be careful if you’re assuming that it takes a given number of highly paid bodies to do that, whether you’re paying somebody else or your own people. All that factors into the cost and productivity.
Martin: I just completed a study as part of a talk I was giving on the differences of the characteristics between the baby-boom generation and the millennial generation, because to a large degree, it’s not going to be Generation X people that will be replacing the baby boomers, it’s the millennials. Some of those generational characteristics support Stan’s points.
Baby boomers had the tendency to learn things from the view of wanting to know everything down to the last nut and bolt. When we learned computer science, we learned how a computer worked-how an arithmetic logic unit worked, binary and-gates, or-gates-it’s just the way we are. Therefore when baby boomers are running plants, the run them from the perspective of the way they learn. If everything goes wrong, the baby boomers know how the plant runs, they know it down to the last valve, and very often, they can bring upset conditions into a steady state because of they way they learn.
Millennials don’t learn that way. They don’t learn how the tool works down to the nut and bolt, but they know how to use the tool better than the baby boomers. Millennials are really good at technology and very comfortable using technology, but they don’t have the deep need that baby boomers do to understand every nuance. Baby boomers panic when they think that there won’t be anybody left who understands how to change the rivets on a distillation column. The good news is that the baby boomers came before the millennials, so a lot of the base knowledge that the baby boomers developed has been embedded into the technology.
When you think about it, the technology base that the millennials are walking into may just be perfect for the characteristics that they’re bringing to the table. They’re very good at collaborating across organizational structures, they don’t need to know everything about everything, they’re very connected, and they’re oriented to immediate gratification so the feedback from a real-time engine fits right into the millennial mind-set. A lot of the tools that baby boomers have been developing for the last 35 years are really better suited for millennials to use than baby boomers. Millennials have the perfect mind-set to pick up on what the baby boomers put in place. That might mean, going forward, that you can do things in plants with far fewer people, to the same degree of efficiency and proficiency, because of the background work that’s been done for the last 35 years.
DeVries: I’ll give you an example of one of the questions that people are re-asking: "How many control loops per operator?" The answer is always, "That depends." You have to ask who the operator is, and what people and work processes are supporting that operator. People don’t think through the problem. They assume that nothing else changes and that maybe there’s some mystical new kind of displays and alarms. That’s a wrong and dangerous way to think about it. You can’t assume that more loops can be handled by a single operator if you think they’re going to be doing the same kind of work and be the same kind of people.
Martin: We believe that some of the KPIs, such as productivity and others, are valid for keeping your fingers on the pulse of things, but the truth is, the objectives that these producers and manufacturers have is to maximize the profitability from their operation. The fact of the matter is that it should not be reducing headcount. It should not be productivity in the traditional sense. It should be asking what is the exact mix of technology and talent that I need to optimize the profitability of what I’m doing. In some operations, the number of people may go up because they’re more craft-oriented operations and they don’t have the degree of scientific experience, working in specialty chemicals and industries like that. Other operations are very scientifically driven and may need fewer operators per output and unit of profit.
The danger is blindly following a couple of KPIs that can max out, while the overall profitability of the organization declines. There’s no easy answer to this question. There’s a lot of opportunity, but the opportunity should be strictly aligned with the business measures relevant to the business in question, and the objective should be to maximize the profitability of that operation.
CE: So profitability is the ultimate goal? How do we relate that to headcount?
DeVries: We’re working with two large oil and chemical companies, and they aren’t looking at a number so much as a trend. They’re looking at the last 10 years, the last two years, of some measures of people, number of people, cost of people, compared with utilization of those areas and process availability. They aren’t looking to see if a number is high or low, but they want to see a correlation. If they spent more money on people, was there more availability and utilization? Those are pretty gross measures, but they’re trying to consider people and technology.
This is really shaking up some of the internal departments. They’re saying, "Look, you are asking for investment, and we have approved, for you to add technology, top to bottom-everything from earpiece systems to instrumentation and everything in between. But what are we getting when these trends seem to be going the wrong way?" Those companies aren’t opposed to spending money, but is the approach and strategy wrong for technology that’s supposed to help people? It’s an old problem, but certain executives are turning their gaze to this aspect of their operation.
Martin: Generally, people are better than machines at doing things that require reason. I lived through the 1980s when people started talking about lights-out manufacturing-let’s eliminate all the people in the operation. The truth of the matter is that most companies that did that created less profitable and less safe environments to the point where that became a very short-lived fad in industry. When profits are squeezed, the easy answer is to cut costs. We have a lot of people, so lets cut people. I’m not recommending that you over staff, but I’m recommending that when we move ahead with some of these trends, that we do it with a better understanding of the business impact. You don’t just say that you’re going to cut 30% of your workforce and see if you sink or swim. That type of thing doesn’t work.
CE: Are there any useful measures?
DeVries: Managers are looking again at the people productivity issue, not just from a cost factor. They’re trying to come up with a useful measure of productivity in anticipation of the great crew change. Previous attempts-number of control loops per operator or however you want to measure it-they know are grossly insufficient. Everybody is scrambling trying to come up with a better measure. The other part to it is, assuming you can justify a leaner operation, not just the operators but the people who support them, you have to watch out on the other side of the strategy, not to depend too much on the cost of people with the right skills to maintain the technology that was, in a sense, replacing people. So how do you measure, appropriately, what that productivity is, and how far can you go, replacing or augmenting people, because there is a cost of people in your company and in suppliers, to care and feed that technology.
CE: So if companies don’t recognize all the costs of technology, how do we know it helps?
Martin: It’s not the technology that’s the issue. There are a lot of things that happen in manufacturing and production operations that we don’t understand or can’t predict that are going to happen up front. We can’t sit down as humans and say, "This is exactly what’s going to happen, and therefore this is the exact response we need." There is too much unknown. The reality is that production plants like oil refineries and chemical plants are non-linear dynamical systems in the sense of chaos theory. Without controls, they behave in a very chaotic fashion. Even with controls there is a considerable amount that is unknown. We just don’t know the fullness of the operation, the environment in which it runs, the quality of the products, and the quality of the energy. We don’t know enough of that to precisely predict what’s going to happen.
Could the technology deal with situations if we could program that knowledge into them? Yes, I don’t think the issue is that; it’s the same issue we ran into in the 1980s when we were trying lights-out manufacturing: we cannot define, in a predictive fashion, everything that’s going to happen so we can determine what the correct move is. Machines can’t respond unless they can do that. Humans have an uncanny capability of looking at an undefined situation, something that may not have happened in the past, and reason it out, including a reasonable response. It’s not a matter of getting the latest and greatest technology, although that’s not a bad idea; the idea is to be able to completely analyze the environment that you’re in, so that it’s completely predictive. If you can do that, then machines are perfect. But I’ll tell you something, we’re not even close.
DeVries: It’s one thing for a machine to recognize a valuable pattern. Not necessarily just to identify when something goes wrong, but when something is going well. But the constant complaint that I hear from customers is that the system told us something true and at the right time but we couldn’t act on it because it didn’t give us actionable information, so we had to ignore it. That’s not a reason not to use technology, but you have to understand the limitations.
CE: So you can’t program your system for what you can’t imagine happening. But after all these years, all those millions of operating hours in all the refineries around the world, shouldn’t we know all the things that could go wrong in a refinery environment?
DeVries: Yes we should, but it isn’t about what goes wrong in refineries, it’s about what goes right. We definitely don’t have enough knowledge about that. For example, when the weather changes, it could produce a situation that is more profitable for you to do some specific thing. How do you deal with a weather change? Some of our customers with simpler processes like power generation have figured out how to take advantage of that. But in a plant as complex as a refinery or chemical plant, it’s far more complex. How do you quantify that as part of a "golden run?" That’s a good problem. It’s not like imminent failure of critical equipment.
Martin: Oil refining is probably the most scientifically analyzed process in the history of the world. Oil companies have scientists working day and night studying the dynamics of these systems. I also think that if you analyzed total percentage of profitability per person, you would find that the whole oil production value chain is among the most profitable in the world. Part of that comes from all that scientific knowledge. That said, there are infinite numbers of conditions that could affect the performance, whether good or bad, of an oil refinery. You could have a lightning strike in some part of a facility that’s never happened before in the history of the world. If you’re going to set up an automation system completely without human support, you have to consider all these infinite conditions that could happen.
New technologies have emerged in the last 10 or 15 years that have made many more things predictable, but when you look at the number of permutations of things that could happen, even in a very small refinery, you’re approaching infinite numbers of things. Trying to model that in software is a huge challenge. It’s not that the technology is deficient, we just haven’t learned enough about all this stuff.
CE: But how do we consider these in a plant with a 20+ year old control system? How much can technology change that?
Martin: That’s a fair question, but we go to sites all the time with old control systems, not pneumatics or electronic analog, but at least digital control systems, that are highly underutilized. Even with a control system from the early 1980s, you can do the same things that you can do with one made today. The difference is that today you have all sorts of pre-developed advanced software that allows you to do these things more easily, but the base structure of digital computing hasn’t changed all that much. I’m not convinced that the barrier is the age of the technology, it’s more the mind-set that if we put something in 10 years ago, we’re not going to use it to do any more than what we planned to do 10 years ago. There’s a lot of capacity, even in these old systems, that is completely untapped.
CE: So how do we boil all this down?
DeVries: The current extreme of 250 control loops per operator isn’t actionable for anybody.
Martin: For me, it’s achieving the right mix that maximizes profitability for an operation on an ongoing basis. It’s not all technology and it’s not all people. It should be the objectives of the business that drive the decision.
- Demographic changes are forcing companies to examine staffing needs and personnel availability.
- Many of the working characteristics of millennials are different from those of their baby-boomer predecessors.
- Ultimately, companies have to calculate the value that human beings bring in a specific manufacturing context.
Learn what Kirk Norris from Maverick Technologies has to say about outside services running a plant.
Can outside services take up the slack?
The idea of having contractors work on a plant site to provide some mix of services has been used for many years, but the idea of having virtual people is newer. Maverick Technologies launched a new service about two years ago that aims to provide 24/7/365 service to customers that is designed to go beyond a traditional help desk. Maverick believes that it is the only company offering service of this nature that covers control systems from a variety of vendors. Kirk Norris, senior vice president of strategic manufacturing solutions, has been overseeing its operation since the beginning and offers these observations:
"When a plant has to run 24/7 in the context of a very competitive business, there are two things management should consider if it wants to use something like PlantFloor24. The first is understanding what your core expertise is as a business, and making sure you have people on staff who are experts in that area. But if you’re trying to hire people who are specialists in common technologies or shared balance-of-plant aspects of your operation, that can be costly in this labor market and might be better outsourced.
"The other thing is calculating a measure of responsiveness for when problems come up in your plant. What does an hour of downtime really cost you, and how many issues happen when you are the least prepared to handle them? How many things take three, four, or five hours to resolve, and how many of those could be resolved in less than one hour assuming you had sufficient numbers of qualified people on hand at all times?
"A service like PlantFloor24 can supplement the people you have on-site with people who are knowledgeable about the platforms that you are running. For problems that are particularly critical or complex, we can escalate to a higher level, including sending people to your site for emergency response in a short time.
"After two years of providing the service, we find that customers use PlantFloor24 for application-level support, and we’re working with multiple customers in multiple industries using multiple platforms. Each customer relationship is unique, and we have tailored our offering to reflect each individual situation. We have yet to lose a customer that has started the program, so that tells us that the service is valued. It also helps us differentiate ourselves from other process automation service providers in a competitive industry.
"The ideal situation is when someone uses us as a continuation of a large project. Often the individuals who provide ongoing service have participated in the project so there is familiarity with the process and personalities on both sides. If we don’t have that connection, we provide additional training for our support personnel internally on the specific application and what is involved so our resources can offer appropriate support.
"There have been a few situations where it simply is not practical to offer support from the outside, mostly related to very old control system platforms that can’t support the kind of networking necessary for us to be effective. It’s difficult to try to offer support when communication is limited to only the phone or we can’t adequately train our people on the platform."
For more information, visit:
Original content can be found at Control Engineering.