Improving safety continues as a concern in manufacturing companies. Automation has helped alleviate some problems, but also allowed companies to push the boundaries of the safety envelope. Over the years, the industry’s safety record has improved overall, but when things go wrong, the consequences are often more severe. It's time to improve manufacturing safety with analytics.
Safety and environmental regulations are becoming more stringent and media scrutiny is increasing. This adds to the pressure of maintaining production, protecting physical assets, reducing insurance and other costs, and - in some cases - avoiding jail time. The tools industry has used up to now have served us well, but it’s time to take it further.
Analytic technologies offer new ways to augment the current technology, improve procedures, and better support operators. Predictive analytics promise to reduce risk by changing variables like reaction times and likelihood of occurrence. The question is, are companies taking advantage of the available improvements?
Understanding the Layers of Protection to Improve Manufacturing Safety with Analytics
Layers of Protection Analysis (LOPA), a risk assessment method, provides a way to evaluate the risk of hazard scenarios and compare that risk to a company’s risk tolerance criteria. This allows companies to decide if existing safeguards are adequate or additional safeguards are needed. How a company addresses the risk is not part of the analysis.
Process designers use a variety of protection layers, or safeguards, to provide defense-in-depth against catastrophic accidents. They consist of devices, systems, and/or actions that can prevent a scenario from proceeding to an unwanted consequence or, if necessary, mitigate the consequence.
Ideally, these protection layers should be independent so any one will perform its function regardless of the action or failure of any other protection layer or the initiating event. When they meet this criterion, they are called independent protection layers (IPL). Not all safeguards meet the requirements to be classified as an IPL, although all IPLs are safeguards.
The layers of protection consist of a labyrinth of automation and mechanical constructs. The desired result is to capture or mitigate an event at the inner most layer possible, because the consequences increase as the other layers of protection are called upon. As useful as these protection layers may be, one observes that they are largely reactive, rather than preemptive or predictive. The only protection layer with any level of prediction is alarms and manual controls, but these rely on human intervention to catch the event and prevent the consequence. Humans, unfortunately, are somewhat unreliable.
Improve Manufacturing Safety with Analytics and Prediction
The risk calculations use frequency of an event, frequency of the consequence, and probability of failure on demand for each layer of protection. Changing any of these three, changes the risk. An argument could be made that predictive and prescriptive analytics provides the opportunity to change all three. Analytics can help quickly identify areas of improvement or risk. This allows the owner-operator to reduce the possibility of failure on demand.
Analytics have already proven valuable for equipment maintenance. Earlier warnings have led to timelier and targeted repairs and better use of existing labor. This has led to fewer unscheduled outages and helps address 30 percent of causes of abnormal situations. It also reduces the likelihood of a failure on demand and decreases the frequency of an event and/or severity of consequence.
Analytics can also help improve alarming and manual control. Predictive analytics provide earlier warnings. Cases have been documented in which analytics have helped detect deviations several weeks out. This gives operators plenty of time to discuss appropriate actions and implement the correct remedies. Analytics also provide smarter alarming, helping operators to focus on the true problem by reducing distractions. Smarter, predictive alarming can provide critical time to plan a correction and avoid operator errors. This leads to more consistent operations, fewer unscheduled outages, and improved operating procedures.
Although we’ve made significant progress in improving process safety, studies show that there is still room for further improvement. Analytics offer another “lever” that owner-operators can use to reduce risk and improve safety. In many cases, this would require little or no capital investment and offer big dividends to the bottom line.