top of page
Search
Writer's pictureAvinash Manerahimatpurkar

How to Implement Predictive Maintenance

Updated: Jun 30, 2020

When you discuss ideas about Predictive Maintenance (PdM), often you hear

comments like "Oh its too complicated", or "Its only for elite companies like Bosch, ABB and BMW!" or "Its just hype, no real benefit!"


Well, what these critics do not realize that, Predictive Maintenance is no more alien. It has already made inroads into standard operating procedures of manufacturers all over the world. Companies are already taking advantage of the value it brings.


Understandably, in these times of Pandemic and Economic Recessions, question comes -- "How expensive is that!"?


Well, answer to this question, depends on - "How you want to do it?"


You choose BIG System Integrator, with bunch of its consultants start talking BIG data, Artificial intelligence and big jargon - then you are already on the high expense path without tangible benefits!"


OR You choose BIG Bang approach, decide to do Predictive Maintenance for every device you see on your floor, sure, its recipe for expenditure without value add!


You can actually start Predictive Maintenance Pilot with the budget of Rs 20,000/-, get everything in that - Sensor Devices, Analytics Software, Deployment Service; taste the business benefits. Then you are ready to do incremental investments and scale to more devices!"


"That's good Price!, Next - What is way for Simplified, Lean Implementation of Predictive Maintenance?" Yes that's right question! You are already thinking about it and this blogs provides guidance on step by step approach to achieve this.


Before starting the implementation, define objectives of PdM, scope, expected benefits and budget required. Get buy-in from the management and approval for required funding.



Step 1 : Decide Pilot based Approach



Kick off the Project. Decide to run a Pilot with only 1 or 2 critical devices. Its always good to start small, implement PdM methodology for 1-2 devices, learn the methodology, fine tune it and then go for larger Implementation.






Step 2: Identify right assets for Pilot Project

Predictive Maintenance should not be implemented for ANY device. Some devices may not be Production Critical, for some cost of PdM may be justified and you may want to continue with regular preventive maintenance procedures for them.


Following are the good candidates

  • Assets that are Production Critical. Their downtime halts production.

  • Assets that have frequent failure history.

  • Assets that have high cost of repair or take long time to repair


Step 3 : Establish Machine Data

  • Collect Machine Manufacturer information such as Model Number, Make, Default Configurations

  • Collect Machine Manuals.

  • Collect History of down times, Failure Codes

  • Take help of Maintenance staff. They will have knowledge about the Machine History, type and frequency of failures


Step 4 - Put together Pre-requisites - Sensor Device, Gateway, Analytics Software

This step involves putting together all pre-requisites. These include right Sensor devices, Gateway Device, Analytics Software. Sensor device is responsible to collect all the data, Gateway device is responsible to send the data to Server. Analytics Software that runs on Server.

It is advisable to go for cloud deployment during the Pilot phase, as the cost and timeline of procurement of the on-premise hardware is not justified for Pilot use cases. During final deployments, it’s a choice your IT department makes to go for On-Premise or Cloud based on organizational policies.


Step 5 - Implement Asset monitoring and begin data collection

  • Configure Sensor Device, Gateway device. Perform Deployment of Analytics Software on Server.

  • Start the data collection. The Sensor device collects the data and send it to Server via Gateway Device.

  • Collect the data for 2 to 4 weeks. During this period, note down any failures of the system. This initial period is the learning period for the device.


Step 6 -Apply Machine Learning to Collected data

Once the data is collected, System will use Machine learning algorithms to derive Patterns out of it.These learnings become basis to Predict the failure conditions.









Step 7 - Analyse Failure Predictions. Perform necessary Maintenance activities

Now that Analytics Software has analysed data and predicted failures, its time to analyse those Predictions. Based on the Predictions, take necessary Maintenance actions.



Step 8 - Prepare Project Report with RoI

Its time to closure of the Pilot Project. Prepare project report, findings. You will have learnings while moving from Step 1 to 7. Document those learnings, so that when you move to scaling, your journey is much smoother. Calculate RoI. Present all these artifacts to your management to get approval for expansion of the project to more assets.



Its important to understand that, the long term success of Predictive Maintenance depends on continuous data monitoring. All your devices and Systems need to be up and running all the time. The System will alert the maintenance team continuously about the predicted failure conditions. Monitoring those, addressing those alerts should be part of Standard Operating Procedure.


The Predictive Maintenance Program give 15 to 25% percent reduction in equipment downtime, 15 to 20% percent savings in the cost and extend the lifetime of aging assets by 20%.


In the today's competitive world, its utmost important to take advantage of technology advances, especially when highest technology has been brought down to affordable bracket. With well defined and methodical way, this is certainly achievable goal.


Resonating Mindz (www.resonatingmindz.com), an Industrial IoT focused company, provides end to end Solution that includes Sensor Devices, Analytics Software, Deployment Service and Project execution. For Industrial Motor Monitoring and Predictive Maintenance, please reach out to marketing@resonatingmindz.com

255 views1 comment

1 Comment


Vivek Mane
Vivek Mane
Dec 12, 2021

Very effectively & crisply decoded "Predictive Maintenance" phobia by the author..!

Like
bottom of page