Optimizing Energy with Predictive Operations White Paper
Optimizing Energy with Predictive Operations White Paper
Optimizing Energy with Predictive Operations
Sheila Kester, VP of Operations at TwinThread
When we talk about energy concerning your operations, what do we mean? In simple terms, we are talking about the processes that are reliant on water, air, gas, electric, and / or steam - also referred to as WAGES.
It goes without saying that these processes consume energy. But how much? When energy consumption is left unmonitored, it can represent a fairly substantial piece of your operational costs. Therefore, energy consumption must be given visibility and managed accurately. To truly optimize your operations and bring operational costs down, you need energy management at scale.
The very best method for turning energy consumption into energy efficiency is by digitizing your data related to energy. Digitizing your energy data will unleash your engineers to identify and mitigate inefficiencies, so you aren’t unnecessarily saturating your operations with undo energy output.
Honing in on these energy inefficiencies and tackling unnecessary energy consumption will contribute to greater global citizenship and significantly increased profits. In other words, optimizing energy for your operations will greatly enhance your business as a whole.
Today’s Energy Management
Unfortunately, the state of operations, as they typically occur today, leaves many engineers in the blind (when it comes to energy consumption). Generally, energy consumption is measured at the plant or, at best, the production line level which causes problems for your engineers. Additionally, there is a lack of energy consumption information that is SKU-specific (product-run specific). Due to this, engineers are unable to see energy consumption at a granular level and are therefore unable to make meaningful adaptations that result in greater energy efficiency.
Another issue with today’s energy management is there is little or no information about process-specific methodologies and their impact on energy usage. Without operator behaviors being factored in, it makes things a lot more complicated when it comes to trying to establish a baseline for energy consumption.
Many organizations allocate resources based on consumption metrics - however, how reliable is that metric if energy is only being measured from a bird’s-eye-view? With sparse information to go on, there is no doubt that bad data is bound to surface. And, there is almost certainly bad data that is influencing decision-making.
This current state of affairs, described above, is certainly better than nothing. But do we really want to accept that? There’s a lot left to be desired. And, fortunately, with relative ease - there’s a lot more that can be accomplished and optimized.
What’s Been Missing?
For your engineers to feel they have strong visibility into the energy consumption of your operations, they must have access to data reports at the machine level. The plant level and line level reports are not enough because they don’t drill down far enough to see where energy waste is occurring.
A large factor in energy waste is within the machines themselves. Temperature and vibration excesses and deviations within machines are where energy consumption compounds over time. So, you need to know the operational details of each machine to understand where there are opportunities to improve energy efficiency.
As an example, let’s assume your family vehicle typically consumes a tank of gas per week. However, last week, it only took half that time to consume the entire tank. What happened? Well, to know that, we need to know all the factors. Such as, who was driving the vehicle? What are their driving habits? How far did they drive? How fast did they drive? How was the vehicle running? What’s the condition of the vehicle? Was it particularly warm or cold outside last week?
All of these factors will paint a picture of how the gas was consumed quicker and will allow us to predict whether a similar event will occur in the future. The same goes for your machines. The details matter, no matter how seemingly insignificant.
You need machine-specific data. This means having sensors or meters that can collect data and monitor specific conditions. Many production / manufacturing organizations already have these in place. The key is to exploit that critical information by connecting those sensors to tag-based data or to feed it directly into troughs that evaluate based on production data, energy data (across all facets), or cost data.
A “production data” trough typically relies on an MES system and evaluates what is running / what is being made (SKU or product based), how frequently the production process is subjected to a planned stoppage, how often the process must be stopped unexpectedly, whether there are swings in external variables (temperature, humidity), and production volume. In other words, what good is it to know you consumed a certain amount of energy one day and a lot more the next if you don’t know the relative output of the production line?
An “energy data” trough pays close attention to the changes in consumption across all assets (identifying top performers). It takes into account any type of energy (WAGES or other). It also tracks variation in consumption from production line to production line, or SKU to SKU.
A “cost data” trough allows you to more accurately attach a cost to production processes and the energy consumption associated with that process.
All of this data and evaluation will allow you to uncover patterns that would otherwise remain hidden. Questions like, “when are we most efficient energy-wise?”, or “when are we least efficient?”, begin to become more clear. Answering these questions will allow your operation to begin recognizing patterns and taking preventive actions before events culminate and lead to excessive energy costs or, worst-case, asset failures.
This data also allows you to standardize energy output to units of consumption. The first step in creating greater efficiency is establishing a baseline. Once that is done, you can optimize production processes to match performance expectations.
Best Practices in Solving – A Phased Approach
When looking at your operational performance history, you need to ask yourself, what trends do you see over time? It’s critical to success to pull your subject-matter experts in when evaluating any data. They’ll connect dots quicker and drive action faster.
Your experts should be able to easily identify top causations for energy fluctuation or excessive consumption. Once these root causes are identified, it’s time to drill in and validate whether the assumptions have merit. You can begin by visualizing these findings and serve them up in a manner that benefits the experts evaluating the insights (via digital twins). This typically takes days to weeks with an action period of days to shifts.
Then, evaluate what can be done - whether it’s changing processes or executing alerts (or another approach). Keep in mind, this is a cyclical, always iterative approach. You may find that a piece of equipment needs to be replaced because it has been running out of spec for too long or has simply run its course. However, in the future, by staying close to the indicators (derived from the data), you’ll be able to see this coming and transition more effectively.
You’ll then need to execute predictive analytics at the operator level. Train digital models to interpret and identify top causes of energy consumption (or anomalies) in advance (via alerts).
Then, with this insight, you’ll direct operators to correct “outside ideal” issues (in real-time), and identify best practices (related to operator behaviors) and iterate. This process takes weeks to months and the action period is immediate (literally seconds).
From here, you’ll identify top performers and propagate. By comparing like production runs or assets, and by developing a centerline, you’ll be able to accurately tune like elements. As stated above, predictive operations are cyclical. There is no endpoint, only never-ending improvement. This step (because it’s ongoing) takes months and the action period varies (depending on the scope of the challenge).
What You Need
First and foremost, you need the right team. A collection of subject-matter experts. This team usually consists of Process Engineers, Energy Leaders, and Domain Experts. Once you’ve got the right set of people ready to be unleashed, you need to digitize your energy consumption (ensuring collection via a sensor or meter). To do this, you need an IoT Data Store that collects information from multiple repositories (Historians, MES Systems, PLCs, and even Excel spreadsheets). You also need a tool that can aggregate, visualize, interrogate, operationalize, and then scale the insights you find.
And, all of this needs to be presented in a form that the team you’ve brought together can easily understand and action (so it has to understand the different panes of glass your subject-matter experts are looking through).
These needs lend themselves to a Predictive Operations Platform that values speed, puts experts first, gives your organization visibility where they most need it, and is backed by a provider that’s experienced in your environment. With all these pieces in place, you enter into a perpetual state of operational iteration and improvement. This results in a globally conscious brand image and significant savings on your bottom line.
This resource has been an exploration of how a Predictive Operations Platform can be used to scale your energy efficiency across your operations. Doing this will lead you to greater global citizenship and significantly increased profits.
A Predictive Operations Platform will allow your subject-matter experts and engineers to have strong visibility into the energy consumption of your operations, standardize energy output to units of consumption, and identify top causations of energy fluctuations or excessive consumptions.
If you’d like to learn more about TwinThread’s Predictive Operations Platform, and how easy it makes accessing your energy data and insights, get started here.