In manufacturing, transportation, energy, or any asset-intensive field, downtime and shrinking MTBF (mean time between failures) are the enemies, topped only by outright system failure. All three greatly increase overhead. And so does crisis maintenance.
One large US automotive manufacturer reported that for its maintenance staff of 15,000 to 18,000, “85 percent to 90 percent [of their maintenance work] is crisis work.”
A 19:1 ratio of planned to unplanned maintenance is commonly considered to be world-class by manufacturers. But this ratio raises the question: even if the plant is operating at 19:1, is all that planned maintenance really necessary?
In equipment-heavy fields, operations traditionally works to prevent machine failures and predict equipment replacement or maintenance to keep costs down. Accurate prediction is the key, but it’s also the toughest part of the job.
Today, replacing reactive maintenance with predictive maintenance (PdM) is a way to reduce the cost and disruption of planned maintenance without increasing the unplanned downtime and the phenomenal cost of operations shutdowns. Although the majority of the data for predictive maintenance originates from operations technology, IT can help with predictive maintenance as well by providing predictive analytics and business data.
In the simplest terms, the purpose of predictive maintenance systems is to shift maintenance practices from reactive and preventative to predictive in order to reduce unnecessary planned maintenance, unexpected downtime and lost productivity
What is predictive maintenance?
Predictive maintenance (PdM) is the analysis of equipment sensor data to predict equipment failures and increase uptime while minimizing costs. Any industry that operates machinery — including manufacturing, transportation, energy, building automation, and many others — can benefit from predictive analytics enabled by sensor-generated data. The value of predictive maintenance over other maintenance models is that PdM empowers maintenance and operations decision-makers to “see” when an asset will need intervention well in advance of its failure.
PdM provides the highest possible asset visibility by collecting and analyzing various types of data. Here are some of its capabilities:
- Identifying key predictors and determining the likelihood of outcomes
- Optimizing decision-making by systematically analyzing measurable real-time and historical data
- Planning, budgeting and scheduling maintenance repairs and replacements
- Ensuring proper spare parts inventory
Introduction
Sensors have proliferated the shop floor for decades but in most cases have only been used for command and control — a huge opportunity for analytics and insight is being missed. This sensor data has been limited to only operations teams and, rarely, if ever combined with IT data. Predictive Maintenance is predicated on the principle of bridging the gap between IT and OT.
However, there are a couple of considerations when looking at how to get started with Predictive Maintenance. It can be very easy to get sucked into all the equipment that could be rigged up with sensors and start collecting data. In this sense, IoT and Predictive Maintenance have followed a similar path to Big Data – collect as much as you can with the hopes that it will someday be useful.
Dell recommends starting the investigation from a business perspective instead. Investigate equipment types of largest maintenance expense, biggest impact to overall downtime and most sensor data availability. Trying to focus on these parameters can help companies find opportunities that will yield short-term gains while establishing a foundation for long-term benefits.
Once the best place to start is identified, understanding how to get the data from OT and IT systems is required and in some cases additional sensors will also be necessary. In many instances the best place to start is with historian systems that may already be collecting and storing sensor data. Combining this historical data with IT data can be a quick win in the Predictive Maintenance space.
However, in lots of cases additional sensor data will be required. In these situations, many customers are concerned about making sure they are using the “right” standards and thus projects are stalled in a fruitless search for a standard. The IoT industry is several years from having definitive standards but IoT Gateways, like those supplied by Dell eliminate the concern of developing to the wrong protocol. These gateways are specifically designed to be a universal translator between the operations world and the IT world. In some instances, new sensors will need to be installed in order to make sure that the right data is collected. In this case customers will need to find the right sensor from the right provider and get the whole stack to work together.
The complexity of building a system ranging from sensor hardware to analytics software has been a major barrier to IoT adoption. This hurdle has been recognized at each level of the stack. Dell, SAP, and ifm have all faced this challenge as well as the realization that none of us could cover the whole stack on our own. This collaboration is a proposal for building a complete architecture for Predictive Maintenance from bestin-class providers that all interoperate.
Objective
The Internet of Things offers companies the ability to aggregate existing data sources, gain visibility into new data, and identify patterns through analytics to make better business decisions.
According to a recent survey report conducted by the Aberdeen Group, “best-in-class” companies are increasingly utilizing IoT to implement predictive maintenance models that address their top operational challenges in order to improve their use of assets. Some advanced solutions even include automated work orders in coordination with ERP systems.
Results show that such predictive maintenance practices:
- Reduce unplanned downtime to 3.5% – The amount of unscheduled downtime against total availability
- Improve Overall Equipment Effectiveness (OEE) to 89% – Availability x Performance x Quality = OEE
- Reduce maintenance costs by 13% YoY – Total maintenance costs including time and personnel
- Increase return on assets (RoA) by 24% – Profit earned from equipment
Implementing
Check out the next two articles in this series.
Six steps to using the IoT to deliver maintenance efficiency
Predictive maintenance sensors and implementation
Read more on the benefits of Digital Transformation here.
