February 21, 2008

CMSS Welding 101

The joys of working on a long term project come not only from creating a working product you are proud of, but from being able to analyze its faults. In this manner you not only get to correct your mistakes, but you also get to reconnect with your prior work long after you thought life had moved on. For last year's Sprint Configuration this is no different.

During the competition last year a few of the welds on the Motor Mount broke loose. In the spur of the moment some quick fixes were made in order to continue competing, and life went on. This year the propulsion group has revisited this event in the hope of learning what went wrong.

This post serves to detail to the rest of the group, to our alumni, and to the general public some of the reasons behind the weld failures. In so doing, this post will also cover some basic welding topics as well as what we plan on doing to move forward. All of this on our quest to gain a greater understanding of our design and to reconnect with a creation that many people gave countless hours to produce. Enjoy.

The Sprint Configuration: A Postmortem

Upon disassembling the mount and inspecting the welded pieces, we had some immediate questions. Notably, why did these welds fail? Was it our design? Did we not predict the amount of stress correctly? Was our anticipation of the dynamic load factors during the shalom event too liberal? What about the quality of the welds?

To answer these questions we turned not only to our theoretical background, but to people such as Jim Dillinger, the Mechanical Engineering Shop Foreman, Robert Reid, an Adjunct Professor and structural analysis consultant, and Bernard Carter, the Drama Scene Shop Foreman and a certified welding instructor.

But before we get into our analysis of the situation, I'd like to present a few pictures so that you can view the situation through our eyes:

A view of one of the pieces which first broke during the competition

Another view of that same piece

Crack propagation across a welded joint

Now that you have a better idea of what we were looking at, we'll hear what our experts had to say, and we'll follow it up by a brief review of welding concepts so that readers can better understand the advice we received.

Jim Dillinger: MechE Shop Foreman

The Skinny:
Definitely not enough penetration on most of the welds. One of the pieces fell off the mount right into his hand. Fatigue had greatly weakened whatever small bonds had previously existed. Jim noted that while the welds looked nice, most of them were just layers of what he called "bubble gum". By this he meant that the weld bead existed, but only as a thin layer on top of the surface, without any real hold on the underlying material. This is demonstrated in the picture below:
What this means is that all of the force that would usually be distributed across all of the tube's area, was now distributed over a much smaller area. As a consequence, the stress increased causing the weld to fail.
Jim noted that given the minimal penetration of our weld joints, the subsequent grinding down of the weld beads on some of the upper surfaces removed even more material, reducing this joined area even more. He also noted that the edges of the tubing were only welded on the top and the bottom, rather than all around like they should have been:

His Suggestions: Luckily, Jim thinks the welds on the parts have not rendered it irreparable, which means we should not have to re-manufacture those complex parts. He recommends better preparation for the joints, as well as ensuring we get sufficient penetration. Also, he cautioned that under no circumstances should we grind down the weld beads, as they are a structural advantage to the weld.

Robert Reid: Professor and Practicing Engineer

The Skinny: My informal discussion with Professor Reid centered around the structural implications of welds. I wanted to know if it was at all possible to better account for welds in our design process, as well as how to account for them in our structural analysis. We reviewed the mount, and it broke apart even further in Reid's hands. He reiterated several concepts that I had previously learned from him in Intermediate Stress Analysis, although now they had some real physical meaning:

#1. Connections should always be stronger than the members they are connecting. This seems to make intuitive sense, but deserves to be expanded upon. If the connection can only support 50 pounds, then it won't matter whether or not your tubing supports 60 or 100 pounds.
A less obvious conclusion is that the primary mode of failure is different for connections and members. When our aluminum tubing "fails" we mean that it yields, deforming to a point from which it cannot fully return. This typically means minor permanent deflections or extensions, but with no real large scale change in the system. However, when a connection fails you suddenly lose all contact and force transmittance between two members. This has a sudden, large scale impact on the system. This same type of failure was seen during the competition, and produced drastic results.

#2. Welds as stress concentrations. When the hot weld pool is no longer being fed heat by the welding arc, it starts to cool. Over the course of a few seconds it transforms from a molten hot liquid into a solidified structural state. This rapid transformation is not uniform throughout the joint, and creates so called "residual stresses" in the material as a result of the thermal gradients that previously existed. As a result of this, the joint is already pre-stressed and can be thought of not only as a stress concentration, but also as a more brittle material due to its rapid cooling. Bottom line: you should try to "normalize" the weld by heating up the surrounding material around the joint, thus reducing the relative thermal gradient.

#3. Importance of Dynamic Load Factor. For any physical system, its dynamics can play an extremely important role in its structural integrity. We tend to rely on FEA to validate our designs, but that is only as good as our knowledge of our system. For something as complicated a our boat we have to be very cognizant of the role that the operation of the boat will have on the mount. The concept of a "Dynamic Load Factor" plays itself out here, where something like taking a tight turn in the boat or a sudden stop or start might effect the conditions and loads on the mount.

His Suggestions: Repairing the welds will help. It is hard to tell, but the most likely cause of failure was poor weld quality. This doesn't mean we shouldn't take a close look at our design, but rather that the welds exacerbated the problem. When repairing the welds, make sure to use similar materials, as dissimilar materials can increase the chance of erosion over time. He recommended getting a professional welder to do the welding, and not an engineer or a student (but where is the fun in that, right?)

Bernard Carter: Welding Instructor and Drama Scene Shop Foreman

The Skinny: Ben was able to shed light on significantly more details regarding what went wrong. Aside from the penetration issues discussed earlier, Ben pointed out certain aspects of the welding process that he noticed from the weld itself. The picture below highlights an area of the weld whose crystalline structure differs from it's surroundings. Ben maintains that this is the result of the weld pool being too hot during the welding procedure.
Part of the reason that the weld pool could have been too hot is that the joint was not properly prepared for the type of penetration we needed. As a result the procedure used additional heat to try and dig deeper into the weld, ultimately causing this weakened more crystalline weld bead seen above.

His Suggestions: Proper joint preparation is key for creating decent penetration. According to Ben, for 1/8" thick tubing, you want to bevel each side by 45 degrees, leaving about 1/32" of flat face remaining. For the original joints the preparation left as much as 1/16" or more of the flat face. To clarify, compare the two images below, one which shows the previous year's tubing, and one which shows the 1/32" flat face recommended by Ben.
The difference in expected penetration is not hard to see.

Moving Forward: Practice, Practice, Practice

Given these suggestions, we now are given the chance to fix the situation. I have already begun practicing on pieces using the MechE shop's welder, and am incredibly grateful that Jim has given me the opportunity to practice there. I have prepped several pieces in order to practice laying the beads but also the necessary precautions for ensuring decent strength.
Hopefully, with enough practice we can validate the welds using strength testing before finally cleaning up and repairing the spring configuration.

Stay tuned for updates and apologies for the obnoxiously long post, but I hope it was enlightening to those of you are interested.

- Mark

1 comment:

Anonymous said...

Dont talk about my friend like that