I heard somewhere (but now I cannot recall exactly) that according to Boeing, almost half of their airplane manufacturing cost goes into software testing and validation.
Does anyone know if it is true? If not, then what is the correct percentage?
Any statement and/or document to read would be really appreciated.
Best regards,
I believe that any organization involved in any software development spends a significant part of their SOFTWARE budget on testing (including validation of some data resources). However, the software development budget for a manufacturing organization is a small portion of the overall budget. The term "manufacturing cost" is too vague. It consists of many components that are interrelated but at the same time could be quite independent. Are we talking here about the budget for programming CNC machine tools? It is also called "programming" but it is not software development even though it includes testing and validation for sure. Is the cost of using and supporting various CAD systems also included into the manufacturing cost? Since so many processes today involve using computers including communications over the smart phones, e-mail, web conferencing and so on, we need to accurately define what is and what is not "manufacturing cost" before answering this question
@David, Thank you so much for your reply. Really appreciated.
What I wanted to know is that how much does a critical mission related projects/product such as airplane spend on the testing and validation of the software (as compared to the whole cost of the product). Is it safe to assume that they spend more than half of the total cost on the software testing or will it be an overstatement?
I read it somewhere but now I can't recall where it was that is why I needed to see if someone else has any idea about this.
I am writing a paper about autonomous cars and in the software testing and validation challenge of the autonomous car, I need to make a point that software testing and validation for a system like autonomous car is not going to be traditional and easy.
Well, the software for the autonomous cars could be a significant part of the overall design of those cars, but manufacturing is still a different process. There is also a huge difference between manufacturing and developing software. The software can be replicated for the cost that is negligible comparing to the cost of development. Making the hardware: motors, drive train, body is quite opposite in terms of cost. The design of parts is significantly cheaper than making thousands and even millions of instances of those parts. In other words, making copies of software (manufacturing) is very inexpensive comparing the cost of its development. Making copies of mechanical parts is much more expensive than designing them. Also, comparing manufacturing of cars and planes is not a good idea IMHO, since these two industries are very much different in many respects and thus the cost distribution is significantly different. Making cars is a good example of the "mass production". Making planes - especially big ones - is a much more unique process. Building ships will be an example of yet another industry that is different from another two. Building small private homes is WAY different from, say, building a contemporary skyscraper. And, so on and so forth...
@Rasheed: Thank you, I just wanted to share with you my 50+ years engineering experience of working in various manufacturing industries in quite different capacities. I worked in shipbuilding, optical, mechanical, electronics, CNC manufacturing, robotics, industrial automation, engineering and construction. Even though there are some high level similarities in work process organization and project controls, the differences are still huge and they seriously affect cost analysis.
Hi Rasheed, i would look at some of the work done by Eric Honor on the costs of Systems Engineering, there is one paper that breaks down the proportion of costs across the different SW development lifecycle stages.
BTW there are lots of good points made in this thread about cost of V&V compared to Manufaturing, but the main concept that i deal with (that Automotive is just starting to get its head around) is Model Based Systems Engineering with simulation. This left shifts the work done and focus on the upfront analysis of the design with testing at an early stage that reduces the testing and rework at the end. The majority of the cost of testing is actually due to having to rework the SW (late in the development cycle) as it does not work due to incorrect requirements or implementation. It is much more costly to do it later in the lifecycle than it is upfront.
These techniques are now commonplace in Aerospace and the next wave is Automotive.
I agree with your ideas, these techniques are i think just the beginning, Autonomous SW development needs a very different approach than is currently practiced in the automotive domain.
All the best Graham
- Badges
- Science topic
- Similar topics
- Correction