How do you use predictive maintenance to optimize your maintenance schedule?

The merest and very basic first step to get started to implement a maintenance program and define a scope for your installations is to rely on manufacturer's maintenance guidelines. Then afterwards, sensors, AI and other tools will further help , down the road, to optimize maintenance costs and equipment downtime.

Predictive maintenance are affective to detect random failures that may happen on the assets. Before implementing the predictive maintenance strategy, one should analyze how the predictive tools/technology can help mitigate potential failure during operational, especially for those critical asset. From there, we can measure the benefits of predictive maintenance by how much and is it worth to do compared to the investment required.

The fundamental goal of the predictive maintenace is to find or detect failures early on. A sudden breakdown while in operation or on demand could have disastrous consequences. PDM measures ensure that failures are detected early in the development stage. Replacement and repair plans discussed in advance, including safely shutting down the entire system if necessary, based on the findings and inspection results. MTTF, MTBF, MTTR and number of sudden breakdown are great KPI's to start with.

While defining goals and scope, we have to find critical assets based on production, safety, environment. By adopting predictive maintenance strategies for those critical assets, initial program can be kick started.

Defining objectives in PdM should align closely with your company's asset management strategy plan. Typically, these objectives focus on reducing planned and unplanned downtime, minimizing equipment failures, especially in critical machinery, and enhancing overall plant productivity. Here are three examples of practical objectives for a predictive maintenance program: -Reduce unplanned downtime by 20% within the next year through improved monitoring and maintenance of critical equipment. -Increase machinery availability by 15% over the next 18 months by using advanced predictive analytics. -Achieve a 30% reduction in maintenance costs for high-failure-potential assets within two years by implementing condition-based monitoring systems.

I find that visualising data is usually best. I have created several spreadsheets and databases that bring in a lot of data from other sources and prioritise or slimline in order to highlight key areas of failure or to visualise live faults across multiple assets in order to best prioritise maintenance and manpower

From what I've seen, effective predictive maintenance hinges on well-defined procedures. These should cover everything from selecting the right technologies and setting inspection frequencies to developing inspection routes and training analysts. Post-deployment, it's important to establish protocols for reporting, triggering maintenance orders, maintenance repair feedback, equipment setup verification, diagnostic case documentation, and savings calculation.

You can also use visual inspection to assess the condition of the assets. Any change in the performance, condition of the asset based on vibration, bearing temperature, motor current, efficiencies needs to be considered.

Data is becoming a given... there is no shortage and storage is cheap. But the question comes to, "What data do we really need to make decisions with?" Best to understand the decisions/answers you need before trying to capture and filter through stacks and stacks of data.

The big assumption here is that the foundational data is being updated, from which a trigger would be generated. Certainly, meter values and the like can be manually updated by humans as a result of inspections, or via telemetry from the equipment itself. Either way, if the data is not flowing or updated on a useful frequency, then the setting of triggers is not of much use. You also need to be made aware when/if the data is not flowing and the equipment is unknowingly melting down.

Even with advanced monitoring systems equipped with machine learning, the reality on the ground often requires a human analyst to fine-tune diagnostics and recommend appropriate corrective maintenance actions. Alarms and rules are set to initiate work orders for maintenance, but human judgment remains essential in this process.

A preditiva iniciamos com os limites definidos pelo fabricante, avaliando as curvas e realizando manuten??es utilizando estes critérios. Porém criando uma rotina de preditiva eficaz podemos a partir daí utilizar o máximo do equipamento sem paradas nos equipamentos, esse é um dos conceitos do RCM.

?El mantenimiento predictivo va más allá de nuestros sentidos? — nos muestra las condiciones ocultas de los equipos para anticiparnos a las fallas antes que ocurran. Nos permite evitar: ? Costos no planificados. ? Y garantizar la continuidad de la producción. Nos da un panorama para tomar decisiones acertadas, pero hay que definir los umbrales para establecer zonas de alerta y de reacción. Análisis de vibraciones, monitoreo de temperatura, análisis de aceite, pruebas de ultrasonido y hasta ensayos no destructivos.

If/when implementing PdM, scheduling is where things can literally fall apart. Defining all the metrics and triggers is great however, if you don’t have buy-in from all levels of the organization to actually take the downtime when reaching said triggers, all the technology, pre-scoping, and data is useless. When PdM indicates it’s time to perform maintenance, then it’s time to schedule and perform maintenance. All too often the information is there, but the pressure to continue running or the thought that we have more time pushes the schedule. Don’t let this step and the equipment fall apart. Take the downtime and continue optimizing the triggers.

In real-world applications, predictive maintenance drives the creation of work orders for repairs or corrective maintenance tasks based on ongoing diagnostics. It's vital for maintenance analysts to understand that the recommendations they write in reports will directly translate into work orders sent to the maintenance department. Moreover, there should be a process allowing the maintenance department to accept or reject these orders, ensuring quality control over the recommendations.

Keep in mind that planning and scheduling are two different things. Planning is establishing what is going to be done to what, by who and with what. Scheduling is bringing the plan into a specific time window for execution.

Documentation is the most important for maintenance, it can be referred in future if any deviation, changes are to be done. The history of the equipment maintenance should be easily accessible for better decision making.

A challenge here is to make the "documentation" very easy for the folks in the field. No one likes to "do the paperwork." Mobile applications, drop-down lists, and a host of other software "tricks," can greatly reduce this burden on your field staff and also increase the quality of the data captured. Of which, of course, is more useful for decision making.

For thorough documentation, maintenance analysts need to feed into a failure log and asset condition registry integrated into the CMMS. It's also important to document diagnostic cases, log the savings each case generates, and compile failure statistics to identify recurring issues and conduct root cause analysis.

Documentation needs to be simple but the tools you use need to be robust, especially with so many companies looking to move to paperless working. Plenty of times I've done work in the field that went smoothly but been hampered by overcomplicated paperwork or applications or devices that are unsuited for the work required or recording the tasks performed. It's very important that the right hardware is matched with the right software for the jobs being performed.

Annually, it's advisable to gather statistics from the failure records, diagnostic cases, and total savings generated by predictive maintenance. Continuously monitor compliance with predictive maintenance objectives and attempt to quantify these achievements in economic or financial terms.

The key metrics to measure your return on investment are time and money. It’s crucial to track how much time and money you’re investing in the predictive maintenance program and how much of both you’re getting back. Predictive maintenance saves time by optimizing preventive maintenance (PM) cycles and saves money by catching failures before they occur. Addressing potential issues early allows for corrective work, which is far less costly than repairs after a failure. To keep things simple, focus on these two metrics. Using a solution that includes them in its dashboard will help you monitor and evaluate the true performance.

Predictive maintenance is not just about technology but about creating a symbiotic relationship between technology and human expertise.

El mantenimiento predictivo utiliza datos y análisis avanzados para predecir cuándo se necesitará mantenimiento en equipos y maquinaria, permitiendo intervenciones precisas justo a tiempo. Esto optimiza el programa de mantenimiento al reducir los tiempos de inactividad no planificados, minimizar los costos de reparación y maximizar la vida útil de los activos.