RAM Modeling with Fred Schenkelberg

We’re pleased to have Fred Schenkelberg back with us. He’s a reliability engineering and management consultant with a background in product development. He’s also been a manufacturing engineer and shift supervisor working in a factory and helped found AccendoReliabilty.com. Fred will give us more insight into RAM Modeling.

What is RAM Modeling?

It can be described as reliability and maintenance, or it could be reliability, availability, and maintainability, among other descriptions. However, they all mean the same thing. When modeling a system from a reliability standpoint, RAM is a way to say this is how something works or doesn’t work from a performance view. Then how do you gather the data to simulate and calculate its performance? Models are used to answer several questions, allowing the possibility of doing ‘what if’ scenarios. They’re especially useful with complex systems as they help justify making various changes by enabling simulations of such effects. You can also use the model to check if your preventative maintenance program is working.

Where to find data for the model

Without good data, the model will not achieve the standards you had in mind. As you begin, keep in mind that the model will always be wrong at the start. That’s the nature of modeling. However, you can improve it till it’s good enough for what you’re trying to do.

Creating a model relies on the purpose you’d like that model to serve. It needs to help:

1. Answer some questions
2. Inform some decisions

Once you create a reasonable RAM model, it finds all other kinds of questions it can address. You then get the investment to improve on it more. It’s a major decision tool.

A RAM model can also get represented as a reliability block diagram, which is a very common way of modeling. You can add distribution data to it, making it a very powerful simulation tool. However, there are times modeling is more complex, meaning your system can run in a degraded mode.

RAM modeling is a spectrum of tools to address different questions, ranging from simple back of the envelope estimates all the way up to Petri Nets, and such. It can go all over the map, depending on what you’re trying to achieve.

Which stage of the life cycle does RAM modeling apply?

A RAM model is most effective in the design phase. During this phase, there’s minimal information, yet we can:

• Do FMEAs
• Have modeling or sims
• Have experience from similar equipment

Sometimes, it’s much more than a guess. But it allows us to at least get a sense of where we’re having trouble or not. From there, we can get better data. A lot of design development programs use various reliability models. It’s just a question of whether you’re going to meet your objective. So, in case a forecasted failure occurs, will the machine still be able to deliver the expected output. It helps answer questions like:

• How do you make maintenance easier?
• How do you make the system more robust so that it doesn’t fail as often?

These types of questions are in the development process to meet your throughput objective. Yes, you don’t have a lot of data. However, you still need to answer the questions. What do you need to improve so that you’re more likely to hit your objectives? It could be that the data’s there but a bit sketchy, forcing you to guess. But where it becomes critically important, such as in million-dollar decision type questions, you’ll need to go get the data. Set up experiments to sort it out that way.

When you’re looking to create a reliability model for your system, you need to start where you are and start building it out to solve the problems you currently have in front of you. Even if you’ve got a warehouse full of data, it may not be appropriate to populate your model for the type of questions you’re going to have. You rarely get the data sets that you need for the question at hand. That’s why you ask the developer to work on it so that you can get a handle on it.

Modeling is a tool to help you and your organization make decisions. So, the model has to fit the ability to inform those decisions. It’s unlikely you’ll do it once and never have a repeat of the process. You may create a master model that you can use several times for different kinds of questions in different scenarios. But often, you might do a system model, then have to do different models for different pieces of equipment. It’s a constantly evolving skill set that we have to use to create simulations and models to inform the questions addressed.

What makes the biggest difference in being successful with the RAM?

It’s understanding what you’re trying to solve for. You’ll never have perfect data. You’ll make a bunch of assumptions along the way. If you don’t have sufficient data in the model to help get an answer, you have to solve that problem. But it’s in the context of how well do you need to know this.

If you’re early on in the concept phase, it’s basically figuring out what the difference is in running it in series or parallel and what’s the risk. In series, if one part fails, the whole system goes down. In parallel, some complexity gets added to it. So, there’s a higher chance of some failures but you’re more resilient because you have the parallel structure. So, a simple model will allow you to compare those two and just use the same failure rate for your equipment as a starting point for it.

The actual failure rate in that case, since you’re doing a comparison, isn’t critical. You may want to explore what the range of likely repair rates for that equipment is. But by being a comparison, it allows you to look at structures and design decisions early in the concept phase. Then as you’re getting to where you have data and you’re looking at, for instance, a line optimization question, having better data from your factory’s maintenance program is more informative to do an optimization type of thing on the floor.

Where to learn more about RAM modeling

Accendo Reliability has got the basics about series and parallel, k-out-of-n. Chris Jackson is doing a third on a series of webinars covering the Monte Carlo Markov modeling. It describes how to model a system without a lot of ambiguity, and to do it efficiently. You can also send us a note to get assistance with what you’re doing.

Eruditio Links:

• Eruditio
• HP Reliability
• James Kovacevic’s LinkedIn
• Reliability Report

Fred Schenkelberg Links:

• Fred Schenkelberg Linkedin
• Accendo Reliability
• Speaking of Reliability Podcast
• Past Fred Schenkelberg Episodes

244 – RAM Modeling with Fred SchenkelbergJames Kovacevic

share

• Social:
• Link:
• Embed:

Rooted In Reliability podcast is a proud member of Reliability.fm network. We encourage you to please rate and review this podcast on iTunes and Stitcher. It ensures the podcast stays relevant and is easy to find by like-minded professionals. It is only with your ratings and reviews that the Rooted In Reliability podcast can continue to grow. Thank you for providing the small but critical support for the Rooted In Reliability podcast!