Platform Engineering

Also known as Product Family Engineering, Product Line Engineering, or Domain Engineering

If you look up Platform Engineering in Google and Wikipedia you find Product Family Engineering and an explanation of how this technique was created by the Software Engineering Institute and is used in software engineering and the term Domain Engineering was coined by James Neighbors in his 1980 thesis.

There is no mention of the fact that this methodology has been used for over 100 years in mechanical engineering. (Typical of the current generation’s blind focus on software!). Although to be fair, if you read James Neighbors’ thesis you will find that he does refer to Eli Whitney’s demonstration of “interchangeable parts” for firearms for the United States military in 1798, although being American he fails to say that this idea was first demonstrated 20 years before by Honoré Blanc in France.

The basic idea is that instead of creating everything from scratch with every new design, with a bit of thought and planning, one could re-use components and assemblies, or systems and sub-systems and thereby come up with a new product more efficiently.

There are actually two interconnected ideas here.

  • Manufacturing from standard interchangeable parts.

This is what Honoré Blanc and Eli Whitney did and it is the same principle that Henry Ford used in setting up his production line. In this case the idea is to minimise variation by building one design that is always the same.

  • Designing a Family of Products

Here the idea is to design and then make a number of different products from common building blocks. The goal is to maximise the variation in terms of number of different products, whilst minimising complexity by minimising the number of different components in the family.

Various people have derived indices to evaluate how well this has been achieved.

3 Types of Product Families

  • Inter-generational

This is using as much as possible of a ‘Mark 1’ version, when developing a ‘Mark 2’. This was the idea used by GM back in the ’20s with their annual model change.

  • Different Versions

This is perhaps what most people mean or think of when they talk about a product family. Different products from the same maker all based on the same ‘platform’, e.g. different sizes, different capabilities etc.

  • Different Brands

This is perhaps the most cynical version. This is where essentially the same design is marketed as different brands (e.g. low end, medium and high end). It is difficult to pull off this trick successfully, and can leave the manufacturer open to accusations of ‘badge engineering’ – see below.

Offshoots of this are joint developments by 2 or 3 companies to produce a common platform that each then adapts to its own products, different manufacturers all using a platform developed by a third party (e.g. a computer operating system), and selling a platform or a whole model to another manufacturer (usually in a less developed market), for them to put a new badge on it when it is being pensioned off by the original manufacturer.

The basic idea (and the terminology, I believe) goes back to the early days of the automotive industry where the car manufacturer would provide a running chassis, i.e. frame, suspension, steering, engine, transmission, axle, fuel system, cooling system and brakes (a platform) that could have a number of different bodies fitted to it. In some cases different engines would be fitted into a common frame. Thus a wide range of products could be offered for minimal extra investment.

This method of construction enabled Alfred Sloan at GM to bring in the idea of the annual model change in the 1920s, because it was relatively quick and easy to just engineer a new body (perhaps I should say ‘design and make’, because I don’t think bodies were ‘engineered’ in those days) on top of an existing platform.

Pros & Cons of Product Families

The obvious advantage is that the manufacturer gets more bang for his buck. By re-using components he avoids re-engineering and re-tooling costs and gets economies of scale.

The disadvantages are that:-

  1. This is fine if the base platform is a good one, but if it is costly or old technology, then its re-use is locking in something bad rather than something good.
  2. Some or all of the product variants may be sub-optimal for performance or cost or both, because they are saddled with a platform and architecture that is not optimised for that variant. To take the car example again. The chassis frame has to be man enough for the heaviest vehicle, which can mean that it is over-engineered (i.e. over weight and too costly) for the lightest variant. Also the whole concept of a separate chassis and body is structurally less efficient that an integrated chassis and body, so the vehicle ends up being heavy and less rigid.
  3. The customers may not see a product family as a good idea. They see insufficient differentiation between models and start to see through it as just a cost saving ploy (a ‘con’) by the manufacturer, rather than a series of good products for the customer. The motor industry provides the classic example of this from the 60’s and 70’s, when BMC (subsequently British Leyland and no longer in business) were heavily criticised for ‘badge engineering’. i.e. making products that were marketed as being different, but were almost 100% the same, apart from the badge and a radiator grille. GM in the USA suffered similar criticism (and lawsuits!) with regards to engines as well as vehicles. This quote from Gavin Green of Car magazine shows what the motoring press thinks of badge engineering, “Nonsense! Badge engineering is always bad. It deceives the public, an insult to their discernment and intelligence. And it buggers the brand.”

Example of Badge Engineering – Fortune Magazine & GM A-body cars from 1983

Badge Engineering

 

4. Often a target for commonality (parts re-use) is established for a family of products at the start of the programme in the ‘big cigar talk’ meeting, but then when one gets into the detail design this proves difficult to achieve, because individual design decisions gradually move things away from that target. Thus the products can end up falling between two stools, not common enough to get full advantage of the potential economies, but not individual enough to optimise performance.

For further discussion on aspects of platform engineering follow these links:

Product Complexity

System Architecture

Mass Customisation

So there you have it. If you are think the idea of platform engineering would be good in your situation, but you want to avoid the pitfalls, then please contact us to arrange a discussion.

PS If you are in the rail industry and you think a platform is something you stand on while waiting for a train, you are in the wrong place!