Machine building the smart way

Let me start of by saying that I am not a traditionally trained mechanical engineer, I have studied several subjects that are the same as what mechanical engineers study but I don’t have a degree as a one. I think this might be an advantage when it comes to building machines, cause even if I have studied physics and mathematics and solid mechanics, I have never been told how to build or design a machine

I have however spent a lot of time building and repairing machines that others have designed and time and again I noticed how things were looked over, or missed, or just ignored because it wasn’t convenient.

When I was put in a position were I had to design stuff myself I realized that it’s not quite as simple as I had thought previously. But even so it is still not as difficult as many designers make it out to be.

The foundation

I have come up with this short list, I’m sure there are more of them floating around out there. But I prefer the process of figuring out stuff like this for myself. I find that my ideas are generally more matched to my current situation than any cookie-cutter philosophies you can find online.

Here is what the list looks like right now. Remember that it is a living document and it changes over time to fit the current reality.

  • What is the purpose of the machine?
  • What size should the machine have?
  • Design for off the shelf components.
  • Design the machine with serviceability in mind.
  • Minimum Viable product.
  • Don’t redesign something for the sake of redesigning.
  • If you are a product based company, don’t sell prototypes.
  • Keep track of the statistics.
  • Knowing a CAD package does not an engineer make.

I will go into my reasoning behind each of these statements/questions and explain why I think these rules are important to keep in mind when building machines or managing a project.

What is the purpose of the machine

We are taught that the customer is always right. Ask a customer what they want they will invariably tell you that they want a single solution for everything.

This is not possible and it’s not really what they want either. On the outset it might look like a good idea but if you dig deeper you will find that they only really want one thing. When that one thing is taken care of the other problems tend to sort them self out.

I am a proponent of the 5 why:s to figure out exactly what the root cause of a particular subject is.

This works both for figuring out a design specification and for finding the root cause of a problem. All you have to do is ask why, 5 times. When you have an answer, ask why again but this time the question is based on the first answer.

  • “I burned myself”,
  • “Why did you burn yourself?”
  • “I touched a newly welded beam in the workshop”
  • “Why did you touch the beam?”
  • “I took support from it to tie my shoe”
  • “Why were you able to take support from a newly welded beam?”

This was only 3 whys:s but you get the idea, keep asking until you are sure of what they actually want instead of building what they tell you up front.

What size should the machine have

This is a really basic thing, and yet it is one that is missed more often than you might think. The last time i saw it was at my current company when a customer had ordered a press with a working envelope of 1.2×4 metres. What they got was a press with a working envelope of 1200×3970 millimetres, 30 mm shorter than what they had ordered.

The engineers of my company arrived at this size by scaling another machine we had just built and somewhere along the way it was decided that it was close enough with a tray that was 30 mm shorter. The embarrassing thing was that it was the customer that discovered the discrepancy during the factory acceptance test

How I think it should be done

This could have been avoided by building the machine from the inside out. Start with the working area of the machine and let the outside of the machine take the shape and size that it needs to have.

And in the case of the press, if our engineers had scaled it with the long side of the press tray as reference instead of the short side, I can guarantee you that the customer would not have minded if the tray was 1209 mm wide instead of 1200. But they tore up heaven and earth about it being 30 mm to short.

Design for of the shelf components

During the winter 2013/2014 i was working at a shipyard in Norway, welding aluminium and stainless steel.

We were prefabbing housing and kitchen units that later were shipped out to an oil-rig to be installed while it was at sea. Let me start of by saying that the drawings were horrendous, when we questioned why they were so bad it turned out that the lowest bidder had won the job and they had outsourced all the engineering and design to India.

Apparently they don’t have access to stainless square hollow section or L-profiles or H beams in India because according to the schematics we were to cut every single piece from plate and weld it into beams and L-profiles and square hollow sections. If you have ever welded stainless you know how much it expands and warps when you weld it, now imagine welding a 6 metre long 150x450mm U-beam from 16 and 10mm stainless plate. It would be kind saying that it looked like a banana afterwards.

I’m sure that mechanical engineering students are told that it is more economical to use off the shelf components but it seems like it is included in most projects as an afterthought. As if the machine is already complete when they start looking at what components already exist and can be included in the project, and if they don’t exist they spec out components that have to be rebuilt or manufactured from scratch.

How I think it should be done

Instead I believe that you should look at what already exists at the earliest stages of the concept design. Start the project of by investigating what pieces already exist and design the machine around those components and this applies to any type of design work. And for goodness sake, use standardized fasteners everywhere.

Design the machine with serviceability in mind

We know the service-life of certain components, we now that most machines will be taken apart and refurbished once every year or so. And if we don’t know the exact lifetime of a pump or contactor, there are damn good ways to get an estimate of it using tools like vibration sensors and heat cameras and when it starts to wear down we can replace it preemptively and save on valuable production time and avoid the costs of standstills.

Despite this, most systems i have seen have been designed completely ignoring this.

Cooling systems are built with the pumps in the middle and the pipes looking like a birds-nest around them. Filters that can only be replaced after a mechanical rape of half the system. You need to have both an angle grinder and a welder so you can rebuild your tools on the fly to even be able to access the fasteners, and you can forget tightening them to the correct torque.

How I think it should be done

When you place a screw anywhere, you have to be able to reach it with a tool. When you place a weld somewhere, the welder has to be able to see and reach it. Gaskets and O-rings need to be replaced at regular intervals preferably before there is a leak. And most important, every moving component will wear down and need to be replaced sooner or later.

Minimum Viable Product

I think there are several different business philosophies that call this a couple of different things but i really like this expression. This ties in with the first heading, what is the purpose of the machine.

Whatever it is you are designing, whatever it is you are building make sure you strip away everything that is not necessary for a functioning product.

The more features you have, the more potential problems you are risking. On top of that, every customer I have ever met would much rather have a machine that never breaks down than one with extra “nice to have” features.

In this situation it is really important to have a well defined design specification, because if you don’t, how do you know where to draw the line? How do you know when you have reached your goals?

So look at what is absolutely necessary to perform the desired action, discard everything else and be very restrictive with how many features you re-implement after that.

Don’t redesign something for the sake of redesigning

Ten years ago, before I started working there my current company had to downsize quite drastically, we’re talking almost half the workforce. After this they had hard times for a couple of years and a bit more than a year before I started they started rehiring and today we are almost as many as before the downsizing.

This had 2 extreme downsides. First is that a lot of competence was lost, people that had worked for 20-30 years were given early retirements without any chance to train their replacements or make sure that their knowledge wasn’t lost. This is almost a no brainer when it comes to downsizing, competence will be lost one way or another, what was not expected was the next problem.

When they finally started rehiring people one of the departments that grew the most was the engineering department, it grew to more than twice its former size. A lot of the engineers were young and hungry, looking for a challenge. All of them wanted to make their mark so to say, and i am sad to say that this is still ongoing.

My employers legacy

We are an old manufacturer of machines, we were founded in the late 50:s, but if you look at all the machines we built for the last 10 years, you would be hard pressed to find 2 that are the same. The same problems are solved in new ways over and over again because each engineer and each project team wanted to do it their own way instead of looking at what already worked.

This has led to a situation were we have a fleet of machines that are close to un-maintainable, we don’t know which problems have been solved on which machine and what problems are waiting to pop-up. We have for the last 2-3 years sold machines which we know have legacy issues which are waiting to be fixed because the engineering department are more focused on coming up with new designs and solutions than they are ironing out all the problems that were introduced with the previous redesign.

How I think it should be done

Changes in design and engineering should have a clear and definite reason for being implemented.

  • It doesn’t work – The design needs to be changed.
  • The customers demand it – The design needs to be changed.
  • The CE/Law requirements have changed – The design needs to be changed.
  • Our sub supplier have changed their production – The design needs to be changed.
  • Several components have been discontinued – The design needs to be changed.
  • One of our engineers wants to see if it will work – It absolutely, positively, does NOT need to be changed.

If you are a product based company, don’t sell prototypes

No machine was perfect the first time it was built, problems were discovered and fixed several times before it was ready to be sold to the first customer. If you are in the manufacturing industry you have trial series and pre-production runs to fix 90% of the problems before the first unit is ever sold.

How I think it should be done

If you are a machine manufacturer you might not have the same kind of luxury, the first machine you build of a certain type might already be sold. Because it is rare that the customer will pay for 2 machines simply to let you iron out all the problems with the first one. It is equally rare or you have the funds to build one for yourself before ever selling it. But even so you should strive to have as few issues with the machine as is possible.

Don’t reinvent the wheel, make sure you use solutions you know will work. Use subcontractors you know have the right skill set. Use components that you are intimately familiar with. And if you have the opportunity to build a prototype, make sure you make the most of it and iron out as many issues as possible and update all the documentation before selling it to a customer.

Keep track of the statistics

This has more to do with the management side of machine building but I think it still is a valid point for this list.

There is a considerable resistance among today’s breed of project managers to keep track of all the hours worked on a project and what parts of the project that took the longest.

I have even been in situations where project managers have moved around hours based on their own choosing simply to make the project over-all seem like a success. The only reason I can think of for doing this is that they are afraid that it will reflect badly on them if the project didn’t match the planned schedule.

What they completely fail to realize is that it only looks good in the short run, flash forward to the next project and the same issues will take the same amount of time to fix because no one that was in a position to order a root cause analysis ever saw the real numbers. Even if the management might see that the project over all took more time how should they know of it was manufacturing, or engineering, or a subcontractor, or purchasing or one of the 10+ other divisions within a large company that was responsible for the delay?

This goes hand in hand with keeping track of the statistics of what components are prone to breaking, because they are probably the ones that should be replaced.

How I think it should be done

I am a welder/mechanic from the beginning that re-schooled myself to become first an industrial engineer and then an automation engineer, everything that I have ever worked with are based on reality, hard facts and logic. I believe that management and business should be no different. Management by emotion doesn’t work in any stage of the production process. If you keep the prestige low and gather a lot of data you are in a much better position to succeed, if not this time surely the next.

Knowing a CAD package does not an engineer make.

I was lucky when I started studying mechanics and electrical engineering, several of my teachers were, lets call it traditional. This meant that I was first taught how to do all necessary calculations with a pen and paper. Same with mechanical design, they thought us how to use a ruler and protractor before we were allowed to learn CAD modelling.

All are not so lucky, I recently I had to explain to a civil engineer what the reference H8 meant on a drawing. This is a person that have spent 6 years studying engineering. Apparently he had completely missed the class on tolerances.

I have had to explain to another former colleague how to measure and compare surface roughness, Ra, in machined components. Or when one design team at my company were tasked with updating the mechanical components and another with redesigning the enclosure of a machine. Both teams did their job without any communication with the other team, so you can image how that went when the workshop started assembling it and the service engineers started commissioning the machine.

Apparently he had completely missed the class on tolerances. I had to explain to another former colleague how to measure and compare surface roughness, Ra, in machined components.

The time when one design team at my company were tasked with updating the mechanical components and another with redesigning the enclosure of a machine were also interesting.

Both teams did their job without any communication with the other team, so you can image how that went when the workshop started assembling it and the service engineers started commissioning the machine.

How I think it should be done

Of course there is a take of distance directly after you leave school and enter the workforce, but these were all persons that were closer to 30 than 20. And this is knowledge that should be taught in mechanics 101 or even before.

I could give more examples but suffice to say that time and time again I have seen people stumble on subjects they should be able to recite in their sleep and forget that the machine is bigger than the piece they are working on right now.

More time should be spent understanding the underlying concepts than learning how to use the tools. It is forgotten that if you learn the concepts the tools are easy to use and only an extension of you knowledge.

End notes

I want to once again point out that these are my set of rules and they suit my reality. There is no guarantee that they will give you any guidance in your current situation. As long as they make you think, and perhaps give you a direction to head in, I have achieved what I wanted with this article.

This article is mostly my ramblings and some thoughts I’ve been wanting to get of my chest for a while. Apart from this I have mostly written about programming since I started this website, and that will still be a part of it because it is a subject that interests me and a big chunk of my projects are software based. I will, however, really soon be in a position to do more hands on type projects with both metal/wood-working, designing and making stuff to put it simply.

Stay tuned for that in the near future and until then, do something cool for yourself. Like this 3D printer i designed and built.

Till next time.

Posted in All, How To's, Ramblings.

Leave a Reply

Your email address will not be published. Required fields are marked *