3 Ways to do Reliability Allocation #4

This is our fourth article about the 3 ways to do reliability allocation. The first three articles describe the reliability design cycle and how reliability allocation is a part of this cycle. Our last article focused on what options you, as a design team leader, have at your disposal when you realize that your components aren’t on track to meet their allocated goals (or doing something!) Now we talk about working out what these allocated goals are. And how to get them. And don’t worry – it is much easier than you probably think!

 

If you want to learn more about a straightforward approach to reliability allocation – read this!

Recap

I make no apologies for making you read three articles before we talk about how we get our reliability goals allocated to specific components. And they are all about allocation factors. Remember in article 2, we talked about the single equation you need to know to calculate individual reliability goals to be allocated to specific components.

$$R_{i(DG)}=\sqrt[\frac{\sum{a_{i’}}}{a_i}]{R_{Sys(DG)}}$$

where $$R_{i(DG)}$$ is the allocated design reliability goal for the ith component, $$R_{Sys(DG)}$$ is the system reliability design goal, and $$a_i$$ is what we call the ‘allocation factor’ for the ith component.

Allocated reliability goals are inputs to your role as a design leader. And understanding how these goals are assets you have at your disposal makes your job immeasurably easier.

So let’s have a look at the ways you get these allocation factors and component reliability goals.

Way #1 – Based on complexity

So lets go back to our allocation factors. The first step in assigning allocation factors to each component typically starts the ‘most complex’ component. And giving it an allocation factor of ’10.’

The next step is then giving allocation factors to all other components relative to the complexity of the ‘most complex’ component.

For example, a component with an allocation factor of 8 means we believe it is 80 percent as complex as the most complex component.

And once you have done this, you sum the allocation factors for our reliability allocation equation. The sum of all allocation factors is 49 for our medical device example above. This means that we can calculate the allocated reliability goal for our most complex component with an allocation factor of 10.

$$R_{5(DG)} = \sqrt[\sum{\frac{a_{i’}}{a_5}}]{R_{Sys(DG)}} = \sqrt[\frac{49}{10}]{97.8\%} = 99.55 \%$$

remembering that $$R_{5(DG)} $$ is the allocated reliabiltiy design goal for our fifth and most complex component, and $$R_{Sys(DG)}$$ is the system reliability goal.

Let’s compare this to a much less complex component, say one that has an allocation factor of 3.

$$R_{3(DG)} = \sqrt[\sum{\frac{a_{i’}}{a_3}}]{R_{Sys(DG)}} = \sqrt[\frac{49}{3}]{97.8\%} = 99.86 \%$$

You can see that the less complex component has a higher or harder reliability goal to achieve. And you need to get used to high reliability goals – goals that are close to 100 %. Sometimes it is easier to look at it in terms of unreliability. Our most complex component can afford to have an unreliability of 0.45 percent, while the least complex component can afford to have an unreliability of 0.14 percent.

And if we follow this approach for all components, we get allocated reliability goals that support the system reliability goal.

Way #2 – Based on complexity and risk

Complexity is one thing. But sometimes we are more concerned about different types of failures. Some failures are ‘worse’ than others. Meaning we want them to occur less often.

Sometimes this goes all the way down to our components. Let’s introduce a ‘severity scale.’ And on this severity scale, a component whose failure results in the ‘most severe’ type of failure has a ‘severity factor’ of 2. All other components are also given severity factors. We can then adjust our original allocation factors that are based on complexity by multiplying them with our severity factors.

So the sum of our adjusted allocation factors is now 59. And we substitute the adjusted allocation factors into the equation we used above.

Way #3 – Anyway you want!

Why? Because your initial allocated reliability goals are guides only. They are not nearly as important as following the reliability design cycle.

You can create any formula you want for allocation factors. Things that matter to the customer are always important and relevant. An approach known as the Aeronautical Radio Incorporated (ARINC) Approach creates allocation factors based on historic reliability performance levels for that category of component. And you can even start with equally allocated reliability goals – which will simply just start the conversation among your design teams to something more reasonable (… sticking with equal allocated reliability goals is not recommended at all!)

The Advisory Group on Reliability of Electronic Equipment (AGREE) Approach includes weighting factors that try to incorporate the likelihood of failure if a particular component fails.

This is sometimes used for electronic products where component failure is not easily and directly linked to system failure. Perhaps failure will only become apparent in very few user input scenarios. Ideally, you will replace this approach with a system reliability model, but the AGREE Approach may be useful for certain applications for initial reliability allocation goals.

A very sophisticated approach is provided by several software packages and service providers that optimize system cost against allocated reliability goals. This involves creating a ‘reliability-cost curve’ or a discrete set of component design options linked to costs.

The approach is problematic. You need to know every conceivable design permutation, and how much each design decision would cost. And pretty much every outcome of every design crisis you will face in the future. Good luck working this out at the start of your development process! And usually, you spend way too much time trying to find numbers, reliabilities, and costs that you forget this is for guidance only.

Another approach (which is really only included here for completeness as it is found in some textbooks) is the expert judgment approach. This simply means experts come up with allocated goals.

The many problems with this include the fact that pure expert judgment on its own will almost certainly fail to achieve system reliability goals. Which means they need to be adjusted accordingly. And the best way of doing this is with allocation factors. So use allocation factors from the start!

So how far ‘down’ do I allocate?

Some products or systems can be very complicated. They may have hundreds, thousands or hundreds of thousands of components. So should you allocate reliability goals for each and every one of them? Should you go down to a single supplier? A PCB? An assembly?

As a rule, you should allocate as far down as what makes sense from a human accountability perspective. This means you typically have a single person accountable for each allocated reliability goal. They may decide to allocate reliability goals to their subordinate designers if necessary – but leave this decision to them.

You also shouldn’t spread single allocated reliability goals across disciplines. Your electrical and hydraulic teams should not be working toward the same allocated reliability goal.

And at the end of the day – allocate as far down as what makes sense you as the design team leader.

So what can go wrong?

Focusing too much on allocating reliability goals perfectly from the start. It is worth re-emphasizing this point.

Many organizations make finding their initial allocated reliability goals a suffocatingly detailed activity. This means that everyone feels so invested in these goals that they don’t understand or feel the need to update them. And because there are substantial uncertainties at the start of any design process, these goals are almost certainly too hard or not hard enough for different components.

But we are locked in now! The initial effort took 50 people two weeks and seventeen spreadsheets to get our first set of allocated reliability goals. And they are now set in stone!

But our reliability demonstration test regimes are based on these allocated goals. Many of which are not feasible. So a lot of components fail these unfeasible reliability demonstration tests.

And then everyone points out that these initial allocated reliability goals were largely uncertain anyway. And we convince ourselves that those components that performed very well on reliability demonstration tests will make up for those that didn’t.

And a lot of people in suits nod in agreement at the end of the production process because they need to get the product out the door to the customer or market.

So the whole process was meaningless as it didn’t really provide design guidance. And no one really knows (or by this stage cares) what the actual reliability of the product was.

So what is left?

Our fifth and final article covers some variations to reliability allocation we have described so far. We talk about allocating Mean Time Beween Failure (MTBF) – albeit begrudgingly. We also talk about availability allocation and maintainability allocation. And just like MTBF allocation – you are ill-advised to use traditional maintainability allocation approaches. Instead – use availability allocation!

We also talk about how to allocate multiple system reliability goals. And reliability goals for different levels of failure. Perhaps there is a critical reliability requirement to take care of the really bad failures we don’t want to happen.

So if you want to finish off your understanding of reliability allocation … there is one more article to go!