Complexity, Performance, and Cost: the changing landscape of MDO

Higher complexity doesn't imply higher cost

You would be forgiven to assume that as a product’s complexity increases over time, so does its cost. However, this seems to be a very industry-specific trend. If we were to compare a few industries, there would be very different trends to their cost vs. complexity curves.

Let’s take automotive, aerospace, and technology. A car you can purchase today is orders of magnitude more complex than the cars of the past, yet the average household is still able to purchase one, and accounting for inflation, the cost of a new car has stayed relatively flat.

In the technology sector, a CPU that you can purchase today is again orders of magnitude more complex than the CPUs of a just a few years ago, yet the cost has actually decreased over time (still somehow the cost of a modern cell phone is now around $1000 – not sure how to explain that one. Maybe they could benefit from employing MDO).

But what if engineers could get their hands on accurate cost estimating tools early on in the design stage? Would that be enough to reverse the increasing cost trends of an industry? Maybe.

Comparing these industries may seem unfair since there are many factors at play so just looking at “complexity” as a broad metric can certainly be misleading. Yet lowering cost when a product is in its digital twin stage is attractive and is becoming a hot trend across the board of all major industries.

MDO, which has been used extensively over the past couple of decades to help engineers meet performance requirements, has historically not relied on any cost estimation tool of a reasonable fidelity among its many disciplines. Even though MDO is ready and able to include and introduce cost analysis as another discipline and as an objective, the question is - should we?

Pros and cons of cost optimization

In automotive, decreasing a vehicle’s mass while using cheaper materials in strategic locations was and still is of utmost importance since vehicle mass is somewhat of a combined objective to be minimized. Using cheaper materials and decreasing a vehicle’s mass helps keep raw material cost low while improving fuel economy among other benefits. I would argue that the use of MDO to accomplish vehicle mass reduction has indirectly translated into a reliable vehicle that will keep you comfortable on the road and also safe in a crash for about the same cost as vehicles of the past that had very little of these benefits.

However, there is always a tradeoff that must be made. As a computer-aided engineering (CAE) community, we should be extremely careful in shifting our objectives. Having mass as a driving objective has not misled us. It is a measure that is definitive, well defined, accurate, and true. Also, the uncertainty can be quantified and accounted for. Cost prediction is none of these.

The most important part of a successful MDO study is defining your objectives, so I would like to express optimism for the future of MDO but also to exercise caution. This is nothing new in CAE. Each of our now established and accepted disciplines from finite element analysis (FEA) to computational fluid dynamics (CFD) all went through a period of scrutiny. Cost analysis should not be any different.

However, the potential upside outweighs the possible downside. Cost prediction analysis is a bridge into an entirely new domain. This means that MDO can now extend beyond just the engineering designers and analysts and into the world of virtual manufacturing and finance.

Maybe we should start to brush up on industrial operations and economics. Being a mechanical engineer by training, thankfully these subjects do seem more interesting to me when placed in the context of MDO.

For more on this topic: WEBINAR RECORDING | Incorporating Manufacturing Cost into Engineering Optimization