Reduce costs and downtime for asset maintenance
A perennial goal of every manufacturer is to improve efficiency by increasing asset uptime, optimizing maintenance and minimizing cost. Equipment downtime can be extremely costly, making it critical for maintenance to be proactive rather than reactive. Optimization must happen on all aspects of the production process – individual units, entire assembly lines and other assets – to maximize Overall Equipment Effectiveness (OEE). For example, keeping a production process running at optimum levels means scheduling maintenance on each individual asset at non-disruptive times. Ensuring that the right service technicians are available, with the right parts, after every component has been used for the longest possible lifespan. This without causing detrimental effect to the machine or the product being produced. Each aspect of the maintenance process must be carefully coordinated to ensure the operation continues to run as efficiently as possible.
There are three main elements to consider when working to optimize asset efficiency and maintenance:
- The machines themselves,
- The inputs and raw materials being put in the machines and
- The way the machines and components are being operated by humans.
To correctly identify opportunities for optimization across the three elements, we recommend following a four-part approach:
- First, understand how the production process is currently structured and functioning.
- Second, model and analyze the data to identify trends.
- Third, act – eliminate inefficiencies or anomalies based on the data analysis.
- Fourth, monitor the actions taken to evaluate results – did OEE improve as expected? Unplanned downtime reduced? If not, why not (go back to step 1). Quality Improvement with DMAIC: Define, Measure, Analysis, Improve, Control
Step 1: Understand the baseline situation
Gaining a full understanding of overall operations environment is essential before targeted action can be taken. Dramatic reductions in the cost of digital sensors and ubiquitous access to the internet has given manufacturers the ability to collect data from a wide variety of assets that can provide essential data. Monitoring sensor data from all vital assets such as pumps, compressors, valves, hydraulics and generators provides an in-depth picture of how individual parts of the operation are functioning. It’s also important to track how individual equipment operators may be affecting the health and performance of equipment. Is part of the production process functioning more efficiently with some staff than others? This may call for further investigation and possibly more training for some employees.
From that baseline, you’ll likely discover that there are additional data points you need – making the next round of sensors easy to choose. For example, if the sensors built into one part of the equipment don’t show anomalies but the machine is still experiencing frequent problems, it may be time to add sensors to related aspects of the operation. All the while, it’s essential to collect all these sensor inputs into a single repository where it is processed, analyzed and shared. Some critical sources/domains to bring together include MES, SCADA, operator logs, error/failure data, recent maintenance reports, energy consumption and waste, among other metrics. Once all the data is in place, it’s time to start analyzing the trends that are affecting the operation.
Step 2: Model and analyze trends
Modeling and analyzing trends in the manufacturing environment are essential to understanding how asset maintenance can be improved. Processes can be modeled for theoretical optimal performance by taking optimum machine capabilities, factoring in line constraints, productivity and component constraints. Historical data gives the real-life picture and the in between is the opportunity for improvement. With enough data and analysis, maintenance can be carried out only when it’s needed but still early enough to avoid a break down; avoiding unnecessary downtime and reducing costs of parts replaced too early. New capabilities in cognitive analytics employ big data computations and machine learning to help inform this type of modeling.
Machines that repeatedly fail during a shift, might indicate that operators need additional training that will increase output while reducing energy consumption and waste. Consistent equipment failures might also mean machines have been incorrectly positioned or are suffering from design failures. Examining data for anomalies and deviations from expectations, for example in energy use, can reveal important details about potential problems in the operation.
Finding these patterns in the data can also reveal problems with the equipment itself. Data patterns may reveal subtleties like a product being made out-of-tolerance due to a calibration issue. If a machine is consistently failing but the issue seems to be different every time, it might be time to find replacement equipment that can function better under on-site conditions. For example, equipment that malfunctions during hot and humid days may be causing disruption but it may not be readily obvious that environmental conditions are to blame without data analytics. With robust analysis, a variety of factors can be considered to reduce costs and downtime.
Step 3: Act
Once sufficient data is available and initial analysis has been conducted, the next step is to make operational changes. In many cases, leveraging preventive maintenance would be an essential first step. Other changes may include adjusting the maintenance schedule so that the staff with the most relevant experience are scheduled to service the right machines (schedule by expertise), and ordering processes to ensure that the right parts are always on hand. Recognizing patterns is critical to acting. If a particular engineer has resolved a specific or complex problem in the past, it makes sense to schedule them to do the same repair again, or at least ensuring stand by. If another facility has experienced the same recurring problem, it’s important to make sure that siloed knowledge is available to all technicians. Often, this information about repair history and experience might live in various systems, but is not always available alongside machine failure and current repair data. With adequate data and insights, it’s possible to build in automation to new areas of operation and improve overall efficiency.
Forward looking companies are developing and introducing “digital twin” scenarios in which both the physical and virtual (digital) world run in parallel. As sensors monitor the status of a physical components, the digital twin makes it easy for engineers to determine what is underperforming and what is overperforming. This parity makes it much easier to see problems coming down the line and act quickly to avoid large scale issues. It also helps product designers get clear insight into deviations between ‘as is’ and ‘should be’ to incorporate in to future designs.
Step 4: Monitor and evaluate
Every action you take that affects the manufacturing process will have an expected reaction. It’s critical to carefully monitor results closely to see if the changes you make are having the expected impact. Metrics that show little or no improvement, or even move in the wrong direction, must be investigated further and additional action taken. Once the data shows improvement in the right areas, additional opportunities for improvement will emerge, leading to a virtuous cycle of continued analysis, modeling, action and improvement. Gathering and closely monitoring data before, during and after each point in the optimization process is critical to understanding the opportunities for improvement and whether those changes have had the expected impact.
Reduced costs and downtime for asset maintenance are best achieved through end-to-end optimized operation that uses the latest digital technologies, such as internet of things, intelligent modeling and advanced analytics.
Microsoft has worked with hundreds of manufacturers to uncover inefficiencies in asset operation and maintenance programs and implemented lasting changes that reduced costs and downtime. Our teams have extensive experience in industry and verticals and understand the unique needs and considerations of each. Partnering with Microsoft can accelerate your digital transformation and cost reduction.
| Learn More |
