Aluminum Die Casting Product Traceability

[AU_Fan]

Hello Elsmar forums,
I work for a "Small" company in the Southeastern U.S., our HQ is based in Italy. We only have the two locations and the company specializes in Aluminum Die Casting. The company that used to own the US operation is no longer, and was bought by the Italian company a little over a year ago. I started here less than a year ago after leaving a well established corporation to venture into other markets and expand my knowledge of Machining. Since this is such a small company, and has yet to clean up the mess from the prior owner, I am responsible for 45 employees and 38 machines on my two shifts (Second and Third Shifts). Being responsible for so much was not something I had expected, but I've made it my passion to create a smooth operation for my company.
We have made great strides, and have admirable goals, but time has cut us short and there's so much work to do. We have many inherited issues, the biggest of which is our Lot Traceability... and this is where I need all of the help and advice I can get. The system set in place is used to trace the quality of 35 different products for various automotive companies. About half of our products are casted in-house and machined in my side of the plant. The other half of the products we make are from solid aluminum blocks. We have many issues in our casting department, and its constantly improving. The aluminum blocks provide very few issues and are easier to control, but we still have our issues from time to time.
Our lot tracing methods seem very primitive, a literal tag (printed sheet of paper) that physically follows the parts.. or at least that's the idea. My last company spoiled me and used SAP (which after some intense training was an extremely powerful tool). Our current system is prone to various issues, even down to a fan blowing a tag away from the parts. I want to convince my employer to completely restructure our traceability, preferably using a computer system. Theses are my current obstacles:
1. Our company is very frugal, but spending isn't an issue.
2. Our machine and die cast operators are trained to run a machine, we expect nothing more from them other than running the machine and alerting a shift leader if they notice any anomalies. Their near-minimum wage pay reflects the amount of responsibility they are tasked with. So preferably a system that's very user friendly, such as scanning a barcode at the end of each shift.
3. It must be able to support new growth. We have won 14 new contracts in the last 4 months, and new production is starting within a year.
4. Detail is everything. We need to be able to narrow it down to the shift at a minimum. Preferably the time, operator, machine, part #, batch #, quantity.
5. The same system has to be used for all stages of manufacturing: Casting, Trimming, Shot Blasting, Machining, et all.

I am an absolute novice with this side of business, and have always stayed focused on maintaining CNC machines, writing programs and leading projects. Any and all help would be greatly appreciated.

Sincerely, SG.

PS. I'm not sure if I posted this in the right section of the forums. If I incorrectly posted this here, please forgive me.

 

[yodon]

So are you asking for recommendations on methods / systems or are you asking how to 'sell' it to upper management?

 

[AU_Fan]

Part of my job is preparing projects, ordering tooling and making the job better and easier for our employees. Within my scope of operations, selling to management is rather easy. The difficult part about proposing a new lot tracing system is its outside of my scope, it would be easy for them to shrug me off since they wouldn't consider me knowledgeable enough to make those decisions in our current process. So suggestions on approaching them is appreciated.

My biggest concern right now is figuring out where to start, maybe narrowing down a few options and doing some research. Once I learn more about how traceability works I can make a presentation to my plant manager.

 

[yodon]

Ok, that makes sense.

The level of traceability probably depends on the type of industry to which you're supplying. If, say, you're making parts for a medical device and the materials in that part may adversely affect someone (I don't know, maybe silicone for breast implants?) then you'll need traceability back to the source and (if you're the manufacturer of record), out to each device owner (serialized). If you're making something that doesn't matter all that much, maybe no traceability is required at all.

I look at traceability as a risk mitigation. What's your risk of not knowing when they were made, with what material, and where they went?

 

[AU_Fan]

Primarily Dealing with automotive. Hydraulic pump houses, brackets, transmission parts. Everything is casted except for the pump houses. We want to be able to trace a batch of parts all the way down to the 360&380 aluminum ingots we used to make them. Casting, trimming, shot blasting, machining, deburring, washroom, assembly, packaging and shipping.

The biggest issue we are having is customer complaints, and without sufficient information from our lot tracing process, we aren't able to investigate. Some of the issues we believe are untruthful... Such as a customer paying a premium to have the parts ASAP. As soon as they receive the parts they send a complaint. Almost like clockwork. We also need a better system in place to ensure our employees are following the proper protocols set in place. That way if action must be taken we can do so. We are inflexible and very limited as to what we are able to trace right now.

 

[ncwalker]

A casting process is typically:
1) Cast and trim
2) Sometimes some secondary filing
3) Some sort of tumbling
4) Machining
5) Possibly some 100% measuring and/or leak testing.

Let's talk casting tracability. There are two strategies here: cast in a lot code or imprint a lot code during trim.

MOST people just cast in the the lot code. This is done with a date pin or two. The most effective methods are:
1) Julian date. The year is cast into the part either on a pin or with a matrix. Then there is a screw in the die that has the julian date inscribed on it. This screw is changed every day. Each shift pin pricks it to indicate shift.
2) Week number. A screw is put in place with the week number on it and a pie slice for each day of the week. Pricks are put in each pie slice as each shift passes. This has the advantage of requiring the screw be changed weekly instead of daily, but it is oft times harder to read.

The other way it's done is with a pin printer or a laser attached to the trim die or as a sub operation in the cell. The advantage to this method is that you can literally serialize each casting and if you have to, trace this back to the shot profiles that generated a particular part. This can be a requirement for safety critical parts or very expensive parts. But is overkill. The downside is that a foundry is hard on ancillary equipment, to say the least, and when the marking system goes down, the entire cell must stop. Production guys don't like this.

You establish this traceability back to your incoming ingots based on furnace records. But in truth, this is not as important as tracing it back to what your central melt was doing that particular day. Because whatever you have in your ingots, it changes in central melt. Or can.

Once you go to the machining area, it's a different ballgame. Machining is tricky if your machining cycle time is slower than your casting cycle time. It frequently is. In a job that exceeds, say, 40,000 components a year, you typically have multiple machining lines, multiple fixtures. And the game becomes which line machined a particular casting?

This is done very simply by casting a matrix on the part and each machine line, or fixture, you wish to trace marks the matrix in a different location with a drill point.

Next level is adding engraving with a pin printer or laser and again, you get the benefit of serialization.

That's basically how it's done.

 

[AU_Fan]

We utilize some form of time stamp in some of our dyes, I'm not sure about all of the parts. A couple of our customers require it so I know we have it. I feel that it's mostly for looks, as we have never used that to help us solve a problem in the past. Ill look more into this at work and come up with some answers.

On my side of the plant, we machine batch numbers onto our pump houses that matches the numbers from our material supplier. We have a dotting system on a couple of our parts that either indicate the shift they were made on, or the location of the parts within the fixture. The biggest issue so far is we do these things, but I don't feel like we follow through.

Here's an example: One of our vertical mills makes one product and that's it. We never have to set it up for anything else, but we do make the same exact part for two different customers. One is far more strict than the other. We can make 8 parts at a time, and the machine is equipped with a pallet changed. So we dot the part to show what pallet it comes from and the best location within the fixture. This is a tremendous help for many reasons, none of which have helped with traceability. We have a supplier that casts the parts for one of the two customers, and we cast the other. So right there we have two sources for the same part. The parts come in lots, but it's not done by shift and seems entirely random or based on the availability and size of storage containers. So we could end up having multiple lots mixed in with other boxes, or containers that are only half full. These same parts go on to several other processes after they're machines, including leak testing.

So if we are using some of what you suggested, then perhaps it would be worth improving our system based on what's necessary for that particular part?

 

[ncwalker]

It's dies, not dyes. ;-)

It sounds like you are putting marks on the parts. So I am now a little unclear what you mean by traceability problems.

In the one instance where you make the same part for two customers and one has more requirements than the other, it would be wise to conform the entire cell to meet the more stringent requirements always. Trouble starts when you run a part one way for one guy and a different way for another.

In most cases, the guy with the more stringent requirement has been burned by a casting problem and added this requirement. What that means is, his "extra" requirements are probably a good idea. I have admittedly seen some of these extra requirements get out of hand, but without specifics, I cannot judge either way.

But if you have all the markings, like you say, and it is not helping with traceability, then you may need to look at your procedures.

I'm pretty sure I know where you work from what you have said so far. (The casting industry is very small). And you have a bit of a battle ahead of you. First, cast in date codes should be a must with traceability back to the shift. Anything less is not suitable for automotive work. It's mainly because everyone is JIT and when you DO have a spill, you need to be able to quickly and accurately throw a rope around suspect parts. Otherwise you will find yourself constantly expediting and running on Sundays to replace good parts that have been contained because your traceability lacks appropriate resolution. Even if your customer doesn't require it, you should do it for your own protection. AND you should do it in a consistent way. That makes it easier on your shop floor. All parts done the same.

Now, if you are doing some root cause work and your traceability doesn't help, then your process engineer needs to up his game. We've had excellent process monitoring available on the casting machines since the 80s. And especially after the 2008 automotive downturn, anyone worth their salt that survived it has it retrofitted on every machine. So if you can't track back a flaw to some investigation of process parameters, your process engineer needs training. You should be able to use the cast in date code and go back to what the process was doing via the process monitor and to what the die was doing with maintenance and repair records. You can't just run the job. You have to run the job and make note of what it was doing. Example: I diagnosed a porosity problem with on of my suppliers and they were able to tell me what their cooling towers were doing when the spike in defects occurred.

On the machining side, I'd be very interested in your procedures in how you dial in your fixtures. Bad shops dial in the fixtures until the part is in spec. Good shops CENTER them so they are all running the same. If you can measure a part and from the measurements identify which fixture it came from, you aren't setting the job up right. The only way you should be able to tell the difference between parts is the traceability markings. I find this to be rampant in the industry (machining castings) and that trail usually leads back to your CMM guy not understanding how to query 3 dimensional cast datums properly to properly feed accurate offsets to your setup guys. If your setup guys play this ongoing game of machine a part, check, make and adjustment, repeat and it takes more than 1 hr of time, you're doing it wrong. It should be machine a part, measure a part, based on this measurement put in accurate offsets, next part is good. The training the CMM guys get in this from the CMM manufacturers is laughable. You go to Brown and Sharpe school and learn how to set up datums on a perfect, simple part. That isn't the castings world. In that world, nothing is parallel. Nothing is square. And there aren't any true planes to query as a level except in the most simple of parts.

But at the end of the day, if your marking scheme for both cast and machining isn't consistent and tied into your process records, your going to have a lot of 8Ds that say "operator error." Not a good situation.

 

[AU_Fan]

Definitely a lot to think about. We have some set ups that are more difficult than the others, but for the most part we have them down to a science. It seems that most of our difficulty in machining comes from running parts on machines we don't normally use for that particular part. So when we upload the latest saved information, even though the machines are identical, they still run into issues. Between common settings, sub programs, and casting difficulties it can set us back.

The company is making great strides in the right direction, but the mess they bought is still relevant. I do have a better understanding of what to look for, and tonight I will look through our procedures set forth by our Process Engineer. I will follow the entire process along the line (which they seem to do frequently), anf record what I find. I'll then mull over what I believe could use improvement, whether is be more structure or enforcing our current procedures. Either way, there's a big hiccup somewhere along the way and I'm losing sleep over it.

 

[ncwalker]

No worries. I can be of more help with more specific information. PM me if you want.