Hey everyone, in this video, we are going to be discussing robust design. Now, this is one of my favorite concepts because we get to take a look at what non-ideal conditions can occur while using a product or service, but then also where you set your parameters to be like, it doesn't need to work under these conditions because that situation is very unlikely to happen. What is robust design? A robust product or process is one that performs as intended under non-ideal conditions, such as the manufacturing process. It can occur in the manufacturing processes, but also in the service delivery and how you apply things. When you start to offer a service using scripts that can bring people back on track and reengage the customer, and following up between different components in your business will help you to be able to bring everything back on track and things won't fall through the cracks. An example from a surface delivery standpoint, is let's just say that you are a CPA and you're offering accounting services to a company, and the company doesn't have a whole lot of paperwork, or they might have an extra set of books, what do you do in those situations? Or if you're a CPA and you need to help them with a financial plan to show to the bank for a business loan and you say, okay, so you're showing that you only made 300,000 in revenue in cash, but then your telling me you made four million, where did you pay taxes on that? That would be a non-ideal condition, is that somebody is shady. What do you do in that situation? How do you bring it back on track? You might have a script saying, until you can justify all this information and show that you collected this revenue, we're not going to be able to apply for a $200 million loan or whatever you want. Not that those numbers line up at all, but just an example. If they say no, then you know they'll shut it down. One of the reasons why we look at non-ideal conditions is so that you can know when to shut stuff down. Another example as well from a robust design standpoint with hardware or physical object is, is a user using it under ideal conditions or are they using it in a not very intelligent way? Do you want to be able to harden your tool so that it will work for people under those ridiculous situations? Prompt example, I have a parrot, an African gray parrot, and he is always pulling keys off my keyboard. When I bought the warrant key for the keyboard, I was like, "Will this cover key removal in parrots?" Yeah, sure. I kept bringing back my keyboard. From a robust design standpoint, if you're the warrant key provider, you might say, okay, we've replaced enough keyboards because of this parrot, or these parrots, we're going to include this now in the verbiage that the warrant key will not cover people who let their parrots destroy their keyboards. From a hardware perspective, you might say, okay, we don't want the keys to be too easy to remove, but ultimately, it would be too cost-prohibitive to make sure that we can have a keyboard that doesn't allow for parrots to destroy it. However, when you start to look at creating a product or service that will appeal through a certain market, you might say, okay, now suddenly parrots are extremely popular, and we can create a product that will appeal to this market and meet their needs, so we can harden the features so that the keys won't be removed, and that's how we will expand into new markets. It all varies. This is why it's important to take a look at the innovation curve, trends, and tying demographics, psychographics, how people are spending money because new opportunities show up all the time, so even if it's not parrots, when Harry Potter came out, a lot of people wanted owls. I don't know if owls destroy keyboards, but that's a perfect example of how to look at your specifications, whether the needs change and how you set those set points. It's conceptually simple. For a given performance, there may be many combinations of parameters that will yield a desired result, and that can come down to which materials you use. You may initially have used a carbon fiber for a certain product, but then you find using a composite made from hand-pulled bamboo, that's a highly sustainable crop, that will provide the same protection and specifications. There are some really interesting work going on in University of Maryland in material sciences related to this sort of thing. Go tips. Robust design exploits nonlinear relationships to identify set points for the product performance is less sensitive to variations. We'll go through this. When we take a look at the relationship first in response to factor A, you can see it's very linear. With response to factor B, you see a huge jump in how the performance is impacted and how the noise and variations will impact the performance. A does not affect robustness, but B does. Choosing B1 minimizes effect of variation in factor B on overall performance. With the design of experiments, approach to robust design is based on the design of experiments. This all comes back to what I've been rambling about the entire time. That is, it's really important that you collect data from the beginning. Take care of the data that you collect. Don't just set links and say, oh, I'll get back to that. No, you set your spreadsheets early. Remember it's sort of like a crop. You plant all these seeds at once and you bring it all up together, make sure it's integrated. That way, when you start to do your design of experiments, you've got enough data to pull everything together and reconcile these different values. With the design of experiments when you switching at different materials and specifications and looking at the probability of certain outcomes occurring, data is going to be your biggest friend and especially spreadsheets. It will also be able to help you reconcile the manufacturing process as well, and time that it goes into developing the company as a whole. If you're doing service delivery, how long does something take to provide a certain service? Which applications and software and technologies can you use to support you in your ambitions to provide this product or service? Remember, robust design could be, this software is going to be a lot less expensive, but it will be able to provide a similar or comparable sort of value and the customer is not going to really know the difference. Or in another case, it might need to be a specific software that will integrate with their systems, so hypothesis testing, again, it integrates with this, yes, no. Look at the tools that your customer is using, and it comes back to looking at your independent and dependent variables and way you can make modifications and create contingency plans. Design of experiments is not a substitute for technical knowledge. When you take a look at the robust design process through the design of experiments, ID control factors, noise factors, and performance metrics. Take a look at, say, for example, environmental factors impacting the performance of a certain tool. It could be they will work in environments with 50 percent precipitation, or it could work in a category one hurricane, but not in a five. A lot of those again come down to quantitative with like wind speeds and all this sort of fun stuff. Develop the experimental plan, start in spreadsheets and conduct your analysis. Robust design, it's designing your company to perform the same even if conditions vary, the outcome will be predictable. What noise is involved and set the parameters as well because you don't want to just collect market surveys from some crazy woman with a parrot who's going to say, "Well, that keyboard is not good because my parrot can destroy it." You've got to know which data you accept in, where the outliers are and not what, all that sort of stuff. But the warranty company should probably update their plan to exclude parrots. Good job. Then design of experiments. Switch out the different variables and see how it will impact at as well and where you can make some modifications with the materials. Remember, spreadsheets are going to be your friend in this. Start very early on because this can be a really intense process if you leave it till the last minute. Start early, bring it all up together, and then you can start doing the DOE with a lot mores, a lot less headaches. Thanks.
