Updated: Dec 14, 2020
As more and more companies are going "paperless", clean, usable data is quickly becoming a company's most valuable asset. Everything from preventive maintenance to supply-chain efficiency now relies on accurate data that can be accessed and analyzed rapidly. As part of their overall charter, EAM Masters is dedicated to helping their clients increase confidence when making data-driven decisions. Recently, EAM Masters staff had an open interview with the company's founder, Jay West, and Bob Michel, an expert in equipment reliability and data cleansing. In the interview they answered questions and explained why it is imperative that modern companies have clean data.
Q: We have often heard the expression “garbage in garbage out”. How does that expression often correlate to a company’s Master Data and what are the consequences?
Bob Michel: The garbage-in-garbage-out refers to technical data. If we look at it from the point of reliability, when we need to assess or repair an asset, typically these days, we look at a computer. If the initial input data or documents is not accurate (trash in), it takes longer, it's harder to fix, and it's harder to get the correct parts (trash out). You'd be surprised at the trash that's in computer systems.
We were recently at a site where we went to build some Bills of Material (BOMs) and they asked us to look at their pumps first. We were looking at approximately 50 pumps and at least ten of them had been replaced over the years. The staff never went back and updated the BOMs and never analyzed the need or obsolescence of the spare parts for the old pumps. So, you can imagine technicians going to work on this pump and getting parts for an old model and trying to figure out what the correct parts should be.
In a critical situation where technicians are trying to assess or repair an asset, involved staff need to be able to rely on the information (data and documents) they view. The data and documents must be clean and accurate. Technicians, engineers, operators and warehouse staff need to know what is physically out in the plant, what is designed to be there, and that the data and documents being viewed accurately reflect the design of the physical asset. This is referred to as the basic principles of Configuration Management.
Q: What is involved in a data cleansing & validation?
BM: It goes back to the principles of Configuration Management where maintenance and engineering need to have documents and data that accurately match what is physically in the field. It starts with the engineering design documents that are translated into systems and assets (Asset Hierarchy). The Asset information is then broken down into Asset Management information, including BOMs, spare part, setpoint data, obsolescence plans and Work Management activities (maintenance/test). This information (again, documents and data) needs to be accessible to the maintenance, engineering and procurement departments. To support updates and modifications to equipment in the field, a good Change Management process needs to be documented and followed. Today, in the digital world, we need editable, readable data. We need to be able to compile and compare data in a timely manner. Consider the principles of Quality Control and/or Quality Assurance. Data verification and validation (V&V) needs to start during construction and startup. Data cleansing and validation is the process of identifying and correcting inaccurate data, as well as following Change Management processes to initiate updates to documents and data when configuration changes are made in the field.
Q: Companies that are already in operation have data in some type of computer system that may not be considered good data. Where does somebody like that start?
Jay West: If a company is not sure about the quality of their data or they have an initiative to clean it up, I would suggest they start with a Data Quality Assessment. There is a basic list of Master Data Sets, and each one of them is important. They need to validate whatever is in their current system against what's really out there, either in the warehouse or out on the plant floor.
They don't have to review all of the data, but I would suggest they pick a random sample of 5% or 10% in different parts of the plant and assess the extent to which that data is complete and correct; that will set the stage. Some data will have a higher priority than others.
BM: We start with a critical equipment list and start cleansing that data, and then we work our way down through the Asset Hierarchy based on Criticality Ratings. If equipment hasn’t been assessed and assigned a criticality rating, we determine critical equipment based on discussions with plant personnel.
JW: You want to start with your assets. What is the equipment out in the plant? From there, it's the equipment, bill of materials, or the parts list. Do we have these parts in the warehouse? Then, you've got your job plans for all the repairs. Do we have a job plan for all of the expected repairs for this asset? There's a number of different data sets that all have to be correct for the maintenance system to work properly and for people to trust it to make decisions and to do the things they need to do efficiently in the maintenance department.
Q: A lot of companies are going through or planning data-driven maintenance initiatives. How can a data cleansing program prepare and support those types of initiatives?
BM: The companies that want to go digital, what they want to do is get rid of paper. They don't want to print work orders; they don't want to print a bunch of supporting documents like drawings and set-point data for their technicians when they go to the field. They want everybody to have a handheld and be able to perform their work. To do that, all of the data needs to be accurate and accessible. Your data needs to be electronic somewhere instead of looking it up in the manual. The digitization needs to be accurate, and it needs to be kept current through your Change Management processes. When configurations change, you need that process to initiate updates to data and documents, but only when it’s, physically modified in the field.
Q: Many companies are moving towards predictive and conditioned-based maintenance. How will a data cleansing and data development plan support implementation of those strategies?
JW: The ultimate mission of maintenance is to operate perfectly without any problems until the next planned outage. In other words, no surprises, no unexpected failures, and no downtime.
BM: We used to just tear select equipment apart and rebuild them every year. Instead of doing that, now we want to start predicting when they are going to fail and only do maintenance while it's in the P-F curve, that is, when Predictive Maintenance (PdM) and/or Condition Based Maintenance (CBM) identifies a potential failure is in progress. We need to be able to predict failure, and that's what a good PM Program (including PdM and CBM) is all about.
JW: What we're doing is detecting failures in progress early enough so that we can plan and schedule an outage that has minimal interruption to production. When Nowlan and Heap did their study 40-50 years ago, what they discovered is that for 90% of all failure modes, given that its functioning properly today, you cannot predict when something will fail. It's only when you detect something abnormal that you can start to predict when it's going to have a problem.
BM: All of this reliability stuff was started by the U.S. Navy and by NASA. Think of it this way, when we start up a nuclear power plant, we fully expect it to run ~550 days without failing. In fact, we are surprised and disappointed if we have to bring down the reactor before the next scheduled refueling. Now, why can't someone run a chicken plant until their next scheduled outing if we can do that. Think about the space shuttle. What if we accepted failure and it failed every time it went up? It's not acceptable; failure's not acceptable in a lot of industries.
So, you plan your outages, and your equipment needs to get you from this scheduled outage to the next scheduled outage. To me, every plant should be able to run until their next outage. If you can't, your outage frequencies are wrong, or you need to do a design change.
JW: And that's particularly true for 24/7 operations. Some manufacturing or other industries may not run on the weekends or something, and, in that case, you can have a little bit different maintenance philosophy. But even if you are down in the evening, and you do all of your maintenance detection then, if you don't find the problem or have good data, it will still be there in the morning.
Q: So, if I want good data, what do I have to do to clean up my bad data?
JW: Step one is to do a Data Quality Assessment. EAM Masters has a formal process for doing these assessments. Make sure you know the extent to which you have a problem and the magnitude of it. The next step would be to understand the business value of improving your data. Then, because data cleanup projects tend to be labor intensive, long duration, and a lot of work, you should prioritize. The way to prioritize should be by asset criticality. March down the path of what data is most important, and let's get that right first. Then, you need to bring people in who know what they're doing–who know how to quickly and efficiently review, validate, and correct data and get it into the maintenance system.
BM: They need a good data team. The problems that we've seen are when they hire companies without a lot of experience. They come in with a big name and in two years they never finish the project. We can come in and do it in four months. That happens a lot, and I'm not sure what the solution to that is.
JW: Talk to us first! Let us come in and give you a sample of what we can do.
BM: Typically, the best strategy is for us to come in, do the critical equipment, and train their people so that they can attempt to do the rest if they like. That's the best course for most companies.
JW: It's cost efficient and effective. It gets it done and done right. Then, they can also have ownership of it. We are trying to teach them to fish.
Q: How have the changes in part specification shaped the way that parts are now catalogued? And how does that affect how companies should keep track of their parts?
BM: I'll give you an example that we teach with: bearings. It used to be with bearings that you had to get the number of the bearing—remember, Jay? You had to find the bearing number before you could order a bearing from the manufacturer.
These days, they've standardized bearings around the world. So, if you have a bearing
number, most are the same number at all of the different manufacturers. You don't have to go through all these cross-references. You can now also order bearings by measurements. Those have also been standardized. Now, there's no reason to stock all these different brands of bearings in your warehouse. Bulbs, belts, and fuses have also been similarly standardized. You've got to stay on top of this standardization, so you don't overstock your warehouse with duplicate items based just on an original manufacturer.
Q: What is asset data taxonomy, and why is it important to develop one?
JW: You want to standardize your naming conventions so that everyone looking at data, regardless of what it's on or where it is, will be able to understand it. The purpose of a taxonomy document is to document what the data format should be and how to properly structure the data. It needs to be agreed upon by everyone because we need to have a common format across multiple plants or multiple sites. Even if data is correct, if a human interprets it incorrectly, they can still get the wrong part or work on the wrong asset.
Q: So, what are the major data items in eAM that need evaluated to ensure a company is getting efficient and effective use out the application.
JW: Asset Groups, Assets, Rebuildable Items, Rebuildable Assets, Asset Group Bills of Material, Activities, Activity Routes, PM Schedules, did I miss any?
These could be thousands or tens of thousands of records depending on the size of the operation, and it all has to be correct. Some of that data is more important than others, and that importance is the criticality of the asset. How important is that asset to the overall operation?
Q: For companies that are going to Maintenance Cloud from eAM, is that a good time for them to look at the quality of their data?
JW: Yes, but people need to understand that when they go to Maintenance Cloud, there's no push button program that will upgrade them; they have to reload all of the master data to set up and configure Maintenance Cloud. So, if you're making the transition, for heaven's sake, if you're going to reload all of this data, make sure it's correct. Do that data cleanup as a part of that transition project.