Enhance Your Machinery Management Program With Decision Support
Many of you with mature machine condition monitoring programs are well down the path on your digital transformation journey. You’ve employed a wide array of connected sensors and have trained your skilled technicians to keep up to date with the latest trends in machine condition monitoring. Some of you have even installed industry-leading condition monitoring software like Bently Nevada’s System 1, which allows for greater autonomy and insights into the health of your entire enterprise - all the way down to the machine level.
But have you fully unlocked the potential of System 1? Much like your latest phone upgrade, there may be features you are unaware of that can add real value to your solution. One of the features that can supercharge System 1 is Decision Support.
As the name implies, Decision Support is designed to help you and your team make the best decision regarding your machine’s health. With any complex operational decision, it is best practice to analyze and sort as much (relevant) data as you can, and then review your decision (before you act on it) with knowledgeable professionals who can offer their insight. Sometimes that secondary professional opinion exists in the form of a well-written and properly applied algorithm.
Decision Support from Bently Nevada is designed to leverage the fullest potential of your System 1 application, through the use of asset-specific software algorithms or “Insights”. These may be textbook analytics or fully-custom Insights you can create yourself. By creating machine-level algorithms with your specific production environment in mind, you can automate analyses while adding a layer of targeted protection with enhanced notifications, alarming events, plotting, automated failure detection, and more. The resulting library of purpose built and custom algorithms can be deployed directly into your System 1 environment to provide insights into your machinery health and operational conditions.
There are numerous impediments to the growth of a truly successful, digitally transformed condition monitoring strategy. In no particular order:
- An overwhelming quantity of data can make it impractical to continuously monitor and analyze assets for known failure modes and anomalies, especially at scale
- You may be referencing a sparsely populated and hard to maintain central knowledge base for tracking and retaining corporate knowledge related to operations and process . It is one thing to record this information and quite another to get value from it.
- You may also be lacking an integrated application to acquire and deploy proven analytics and documented failure modes, as outlined in plant specifications or OEM documentation
- An aging workforce - the fact is that your experienced (and heavily invested) workforce is slowly marching towards retirement, and that valuable machine and plant knowledge could retire with them.
- The sheer complexity of some of the available tools for creating, modifying, and deploying custom algorithms can be a deterrent. You, like many, might be looking for an easier way to achieve this goal.
- Your organization may be challenged by a limited ability to easily correlate analytic results with machine/process conditions for root cause analysis, due in large part to disparate historians and monitoring platforms.
At the same time there is a drive for ever-greater efficiency, better ROI, and maximum productivity from every machinery asset under your watch, and you can see that even the best condition monitoring strategies still have room for improvement.
The “overwhelming quantity of data” made available by modern systems is a good problem to have, but capitalizing on its value is a problem to be solved. With Insightpaks, you can monitor for many possible events, on many machines, in multiple operating conditions. And the fact that you can write your own custom algorithms within Decision Support means that you can “fine-tune” these algorithms to support your specific machinery health insight needs.
System 1’s Decision Support adds real automation to your condition monitoring strategy by providing early detection of mechanical, operational, instrumentation, or process faults and inefficiencies. Decision Support takes high-resolution data from System 1, analyzes this data in with first-principles calculations and logic, and then serves these findings back to System 1 for notification, visualization, and root-cause diagnostics.
You can leverage the flexibility of Decision Support in two basic ways:
- You can apply Bently Nevada’s ready-made algorithms, or “InsightPaks”, which have been engineered to detect a wide variety of failure modes. Our algorithms are designed to operate within the Decision Support platform and can be easily tuned to suit the unique operational conditions of a given machinery asset.
- You can use Decision Support to create and deploy your own custom Insights. Custom Insights allow you to leverage your knowledge of your equipment, processes, and business solutions. These custom algorithms also serve to preserve operational knowledge in a usable format that can be broadly applied and shared across your operations in a repeatable and manageable way.
A wide variety of event and condition data types can be processed within a Decision Support Rule, including vibration, process, and emissions data.
Shaft System: over 46 unique parameters from bearing temperature ratios to amplitude and phase rate of changes (5% in 60 seconds, for example)
Lube Oil: oil cooler status, oil heater status, oil filter status, oil pump issues, high pressure in the oil tank, lube oil viscosity, water content, TAN, TBN, ISO Particle, etc...
Dry Gas Seals: supply pressure and temperature, control issues, seal buffer gas supply issue, seal gas booster filter differential pressure, primary and secondary seal gas issues
Process systems: over 14 trended parameters from pressure ratio calculations and validity checks to mass flow vs. volume flow deltas, which support identification of potential surging and possible liquid carryover
...again, that’s just for compressors! With InsightPaks and Decision Support from System 1, you can check for a wide variety of statuses and events for steam turbines, gas turbines, generators and more.
Decision Support software provides an organized and intuitive work flow for deploying existing InsightPaks, and for building and applying your own customized analytics.
No need to grab a “quick” four years of coding education (or hire outside contractors) just so you can write an algorithm. Analytic Rule building is accomplished within an easy to use, intuitive graphical user interface. From this application, you can simply drag and drop mathematical and logical steps and then connect them to perform the desired operation. The availability of advanced properties allows you the flexibility to enforce data and unit constraints for error-proof analytic deployment, define states, and qualify data for evaluation.
System Integration and Data Mapping represent the often underestimated “last yard” in putting a given analytic solution into actual service. As part of your System 1 solution platform, Decision Support’s analytic results seamlessly integrate with your System 1’s data historian. Measurement properties are inherently understood to smooth the process of mapping measurements to Rule Deployment inputs. Analytic results populate the System 1 Database and are grouped for ease of navigation, visualization, and diagnostics.
Once Rules are deployed, any desired modifications are simple to apply from within Decision Support. No need to rewrite an entire algorithm from the ground up.
We have discussed the value of Decision Support and how it can take your condition monitoring strategies to the next level, how easy it is to work with provided or customized Insights, and a number of use cases you can apply these Insights to. Now we’d like to share two case studies for Decision Support.
The filter at the front end of a Naphtha Reformer has a direct impact on the quality of the final product. Bypassing this filter or choking the process by operating with a clogged filter, both have serious operational consequences, making the filter’s condition a critical part of the reformer’s process.
A review of historic data for Naptha Reformers showed that their filters are typically replaced every 45 to 90 days; however, 35% of the work orders initiated for filter replacements were categorized as “urgent/emergency.” It is much more costly and inefficient to drive a work order labeled “urgent” or “emergency” as compared to a scheduled, planned outage and repair. Work performed under such conditions not only interrupts the maintenance team’s planned activities for the day, it also diverts resources, disrupts schedules, and increases your maintenance and operational costs. Our client saw an opportunity to monitor filter degradation and shift from “unplanned and urgent” to “planned and routine” thus lowering the impact of these filter replacements.
Two pressure sensors on either side of the filter are fed into an analytic in the System 1 Decision Support module. The analytic was written to subtract the two pressure signals to create a “virtual” differential pressure signal. This value is then trended and compared to empirically derived “normal” values. For example, experience had shown that 6 PSI of differential is the point at which maintenance personnel should be alerted to a need to check or replace the filters. This value was then assigned to a “level 1” severity in the system, which would then trigger an automated notification to maintenance planning engineers, alerting them to the need to check the differential pressure twice a day. When the differential pressure reached 9 PSI, the severity is upgraded from 1 to 2 (which is still below the Alert level – level 3) and another notification is sent - this time with the instructions to enter a work order for a planned filter replacement.
The Insight was written so that in the event the differential value reaches 15 PSI, a level 3 alarm is sent to the appropriate personnel. At this level, the work order is upgraded from a standard work order for a planned replacement to an “urgent/emergency” work order that may require a revised timeline for filter replacement. Finally, if the differential value reaches 30 PSI, a level 4 “danger” notification is sent, instructing the operator to immediately open the bypass valve to prevent damage to the unit.
In the two years since this simple - yet very effective - Decision Support Rule has been implemented, it has been invoked ten times, resulting in ten planned, routine filter replacements without the need for a single “urgent/emergency” work order.
Within nine months of commission, a brand new RGC unit needed to be shut down due to high vibration levels. A water wash was performed, and the unit was returned to operation. However, as part of the client’s efforts to manage their assets below the alert level, the team examined historical data from the machine to see if they could identify trends that could help them to be more proactive in the future - planning for regular maintenance outages versus the more costly unplanned ones.
The client discovered that a particularly reliable advance indicator of the need for a water wash was a change in the compressor’s balance condition. Additionally, they noticed that for this particular machine, the inboard bearing’s X-Y probe pair was the most sensitive to balance condition changes from compressor build-up. After consulting with one of Bently Nevada’s machinery diagnostics engineers, the client formulated a customized algorithm that examined orbit shape, vibration levels, and rate-of-change in vibration levels to provide early detection of changes in balance condition, and thus to help schedule future water washes in a proactive manner.
While the off-the-shelf InsightPak for centrifugal compressors is capable of detecting and isolating imbalances, the client’s customized algorithm provided even more sensitive detection as it was able to account for the idiosyncrasies of this particular machine and its changes in balance when compressor build-up occurred. The ability to “personalize” their system in this manner was particularly valuable to our client.
You’ve already taken the time and resources to invest in a sound condition monitoring strategy that employs cutting edge sensors and industry-leading computational and analytical power in form of System 1. Now maximize those efforts with Decision Support and InsightPaks from Bently Nevada to gain greater “Insights” into your machinery health and operational efficiency.
Bently Nevada has a rich, 60-year heritage in helping customers solve industrial maintenance challenges of every type and application. Through research in over 20 countries with more than 400 end-users, we’ve studied our customers’ team dynamics, site processes, and technology suites to determine how Bently Nevada can support their efforts in attaining comprehensive condition monitoring.
If you’d like to learn more about how Decision Support and InsightPaks from Bently Nevada can add real value to your existing condition monitoring strategy, we’re here for you