Process manufacturing plants have distributed control systems (DCSs) that are connected to plant floor sensors that monitor pressure, heat, temperature, etc. These sensors are used to regulate processes. For example, flow measurements can be transmitted to the DCS; when the measurement reaches a certain point, the controller instructs a valve to open or close. These sensors can also generate alerts if measurements fall outside desired parameters. These alerts are a key part of supply chain risk management at the plant level.
However, too many alerts can cause problems.
ARC’s Dick Hill recently wrote a detailed case study on how Monsanto’s Soda Springs plant in Idaho addressed this issue. I’ll briefly summarize some of the key points, but ARC clients are encouraged to read the article itself.
Monsanto’s Soda Springs plant in Idaho produces elemental phosphorus. More than seven years ago, the plant began a journey to analyze and fix its nuisance alarm problems. The plant traced these problems to too many alarms with little or no defined response. While overloading operators, the nuisance alarms also hid some real problems.
As Monsanto’s plant in Soda Springs, Idaho came to realize, being able to create alarms freely in a distributed control system is not necessarily a no cost feature. What happened is typical in many DCS implementations; a large number of measurements were alarmed, yet many of these alarms had no defined response. This made it too easy for operators to ignore important alarms that may have come in with the nuisance ones. As a result, they simply suppressed many alarms.
Prior to the digital era of distributed control systems, alarm policy generally required justifying the not-insignificant cost associated with establishing each additional hard-wired alarm. Once DCSs appeared on the scene however, alarms could be added with minimal effort and at no additional cost. This resulted in far too many alarms being added to control systems, often without a clear purpose or, in many cases, without a defined response. Alarms with no defined response tend to overwhelm operators to the point where they ignore even the important ones.
Monsanto put a project team in place to address this issue.
The Monsanto Soda Springs project team approached the problem by first assessing the plant’s most active alarms. Once they got into it, team members found the problem was even more severe then they initially thought. Not only were there far too many alarms, but in many cases, the operators had no idea how they were supposed to respond to them. Not surprisingly, over time, the operators became desensitized, often ignoring even important alarms.
The team used metrics to help address the issue. The team focused on these metrics:
- Average Alarm Rate: the average number of alarms per hour per day;
- Operator Alarm Loading: the number of 10-minute periods per day when the operators experienced 10 or more alarms;
- Standing Alarms: alarms lasting longer than 24 hours;
- Alarms per console point count;
- Alarms per plant process area.
The project team took this data and implemented a Six Sigma continuous improvement process to tackle the problem.
In reviewing the alarm situation during the control phase of the DMAIC (define, measure, analyze, improve, and control) process, however, the team discovered that there is entropy in everything. A large number of alarms had crept back into the system without always showing up on the alarm activity analyses. This is because many alarms had been suppressed. Soda Springs then established a quarterly review of the alarm situation in manufacturing with suppressed alarms becoming a major addition to the effort. Since then, the plant has implemented a number of methods to keep the benefits of the initial effort evergreen. These include smart alarming to produce a single alarm for a single event rather than individual alarms for each data point associated with that event.
The best practices established at Soda Springs were than disseminated to other plants.
Today, the operators at Monsanto’s Soda Springs plant are much happier. The benefits from the ongoing alarm management project go far beyond reducing operator loading and nuisance alarms, allowing the operators to think more clearly about what actions they need to take to head off problems and improve the operation of their plant.
I found this article interesting because, just as with DCS, modern software architectures make it much easier for companies to generate alarms in a variety of different supply chain execution and visibility systems. The key lessons that Monsanto addressed at the plant level also apply to other supply chain processes.