It goes without saying that the technology of machinery maintenance in 2021 has advanced to the point where human assessment is on the verge of being superseded by constantly self-evolving artificial intelligence.
The days when human-derived errors are the reasons why significant machine break down are slowly decreasing. As companies push to minimize costs, and maximize gains, the shift towards integrated computing systems is apparent more than ever.
Traditional maintenance approaches such as corrective maintenance, time-based maintenance, performance monitoring, and mechanical health-based maintenance are appreciated for what they did — pinpointing errors in circumstances ahead of potential breakdowns of mechanical parts which leads to preventing loss on countless occasions.
However, many of these methods are seen by field experts as inefficient, and less effective when it comes to decision-making and cost saving when compared to the implementation of software-based systems.
The question is why is analytics-based maintenance the way forward for businesses which operate heavy-unit machines? Like many things in life and business, there is no simple answer.
Let’s consider time — one of the most crucial factors in business. Time-based maintenance is understood as a type of maintenance check where engineers are scheduled to perform thorough reviews of machine parts and units on a periodic basis, depending on the scale of the operation. However disciplined this approach sounds, it does not take into account the potential errors which may occur in between each maintenance check in such unforeseeable circumstances. One tiny unplanned shutdown could lead to unwelcoming consequences.
While time may be partially intangible, spare parts and inventory are not. Stocking spare parts to prepare ahead of a potential deterioration of mechanical pieces is an apt precaution but the cost of storing and keeping that spare unit or units in well-suited condition may set a company back by a significant amount — proving that this approach, too, isn’t the most cost-conscious by today’s standards.
Corrective maintenance is a much avoided remedial response by companies as it is considered as reaction rather than proaction. In a case of an unplanned shutdown, being reactive may be too late. The waiting game of identifying the root problem, investigating the incident and having the business operation stalled due to the lack of spare parts is a risk many businesses are not prepared for.
Again, there is no right answer but a preference that may suit one’s business scheme. By largely relying on programmed algorithms, raw data constantly fed from sensors, and computational calculations, errors can be foreseen or predicted ahead of a potential crisis.
This allows companies the advantage to get head start by having engineers investigate possible reasons for the deviation of data output, as mapped out across multiple charts, graphs, and key indicators which artificial intelligence is capable of reporting. By affording this time gap, assessment could be made whether to purchase spare parts immediately or not — thus, saving cost and time on this end.
On the contrary, the setup of this software-based maintenance may seem like an overwhelming investment due to the top-of-the-line technology, complex programming sequences, the requirement for experts of niche data science to mount the software setup and break down information for in-depth analysis. However, if the possibility of avoiding a great and costly risk can justify the implementation of predictive analytics-based maintenance, this could be a wise choice to consider for many businesses.
A case study of an incident at a ethane separation plant which engineers of The Soothsayer has resolved using predictive algorithm-based maintenance.
From past to present, we take a look at how engineers assessed risks based on different models of maintenance from corrective to predictive methods.