Steel fabrication thoughts and insights
Brian Williams has spent his adult life working in steel fabrication and technology. Starting as a grinder/wire brush operator in college he worked as a fitter, CNC programmer, production control manager, and VP of Operations for a 4-shop industrial plate and structural fabricator. At FabTrol he has delivered training and implementation services to some of FabTrol's largest customers and designed software as FabTrol's Product Manager. Today, he continues to work in product management and serves as the General Manager of FabTrol Systems.
FabTrol uses an “Agile” development process. This doesn’t mean we have no process—in fact it requires knowing exactly where we are on each of the tasks at hand, delivering quality completed work more often and knowing what’s next if we finish this task.
Process agility is:
- Accepting that change is inevitable and driving rather than resisting it
- Using a process that divides tasks into smaller phases of work to deliver some progress sooner and allow for easy re-prioritization
- Making a habit of incremental changes to the process itself, so those can be delivered sooner, results measured and new adjustments made
- Embracing the discovery that you've taken a small step in the wrong direction, as positive.
Many people and processes assume that the more prep, review and depth of work achieved, prior to delivery--the lower the risk. Perhaps the prep and review helps, but making too much progress prior to delivering something often buries problems and assumptions deep under further progress. Mistakes become more and more difficult--expensive--to unravel and are often so ingrained by the time of discovery that entire projects and products are abandoned or the problem is institutionalized and the cost of dealing with it accommodated for months and years to come.
Process agility does not mean that you can’t have a six-month or two-year plan for your fabrication shop. It does mean that no one can work on something for months before you acknowledge the impact of unforeseen events or find out whether it will take another six months. The point is to deliver many small changes sooner rather than delaying change until it is so large and scary that people over-complicate and resist it.
In previous posts I've talked about integrated, instantly analyzed data that connects feedback from things, people and processes throughout your business. Process agility is what makes it all worthwhile.
Without instant analysis of integrated data you pretty much have to follow the common course of tossing work over the wall, into the shop as soon as the documents are approved, and materials are inbound to match the cutting plan. Rather than driving and managing your delivery and production schedules with well-timed releases of work, all work is released ASAP and paperwork and meetings convey prioritization to the shop, often after the shop has made its own decision about what to do next.
With accurate real-time feedback about the current shop load, anyone can open the left-to-do form and click on the bottleneck in their shop (yours will be different from the next guys) and see the key indicator of how effectively the shop is loaded right now. This means work can be released precisely when you know it is needed--one week, one day or one hour before it will impact your bottleneck. This preserves scheduling flexibility, increases efficiency as everyone collaborates around the same priorities and gives late-breaking revisions their best chance of being known before work is started.
Document Example: "On Demand"
Shops receive approved documents ASAP to "make sure no one in the shop is waiting on drawings (work)". The downside is needing to revise the documents one or more times before anyone looks at them and the possibility that this process will fail and the old revision be left in production. Enabling the shop to print the drawings they need (right now or perhaps for the day at hand) or better, fetch them to an iPad, removes the entire process of maintaining the right revision on the shop floor. If you retrieve the drawing as late as possible and the software is tasked with maintaining the latest revision every time there is an update from the model, you increase the likelihood that workers have the right revision while eliminating the process of destroying or removing the old copy and replacing it. Plus you achieve the original goal of making sure they always have work--a "Drawing Release" or "Work Order" document that lists drawings or assemblies that they are authorized to use/work on is all that is required. No more waiting for the drawing office to print or copy drawings, the shop can pull the drawings to a printer or their iPad at will.
When is On Demand better?
This applies to nc data, cut lists, anything you have a formal process for "revising and replacing". Delivering the lengths to the saw when it's time to cut the bar means the parts will be cut according to the latest, greatest information. Plus it takes less work to fetch the cut lengths as bars are loaded on the saw than it does to sift through a stack of cut lists and manage revisions to a partially cut page.
Drive your shop
When you release work at precisely the moment that will drive production efficiently--but no sooner--and continuously deliver process improvements, one small step at a time, you don't just embrace change, you drive it. Process agility means change is your tool, not something you resist or only react to.
Many shops gather and crunch data once per week. Meetings are held, courses are corrected and everyone guesses how adrift they are until the next weekly update and new course correction. Others go through this every day—obviously they see great value in the integrated data, regularly updated.
If the process was automated they could retrieve it on demand at any time, with less effort. Formally reviewing it every day would be easy and it would be just as easy to get up to date information, on one critical project, a few hours later.
Without automation, what does the summary report mean? When the shop reports that production batch 12 is 80% fit, does this mean that 80% of the assemblies are fit or that 80% of the fit-up hours are expended? Since it’s necessary to track man-hours for payroll, it’s often the latter and no one really knows if batch 12 is actually complete and ahead on the hours or the progress estimate is nonsense and batch 12 is barely underway.
Without automation, information isn’t available to check when you are first aware of a potential problem. If the purchasing agent is over budget and keeps hearing that “Change Orders will cover it”—can they look and find out or must they remember to check the report at the end of the month? What about a project manager who just won the production meeting battle and got his critical production batch moved to the top of the list? When he comes in tomorrow, how much effort will it take to learn if the shop is acting on the new plan? The load just left, when and how confidently will the project manager know whether that critical assembly was on it?
Having the data together in one place is fundamental—but lists have to be crunched into summaries and the meaningful information presented in the right context. This can be available for the asking, all of the time.
One you commit to connecting feedback from machines, people and processes like change order submittals, shipping and purchasing, together in one place, software like FabTrol Pro can mine that data for instant, up-to-date, analysis. Hundreds of reports, included in the software or tweaked by you to address your unique concerns, are possible. FabTrol Pro also saves your filters or "queries". You can design a report, for instance, that lists assemblies completed through welding today and FabTrol can re-use your "0 to -1 days" filter again and again.
Some shops drive production based on material cutting progress—because that is something that they track well. Others are fundamentally drawing approval based. Drawings go to purchasing ASAP and are then thrown over the wall into the shop as soon drawings reach a certain status (like "approved") and materials are in-bound. Still others monitor shipping closely and drive production based on that.
Most shops review a spreadsheet, every week, in a production or planning meeting. The spreadsheet communicates progress on erection sequences or production batches (or "drawing releases"). A percentage of progress is represented for things like cutting and painting. Human judgement is used to correlate whatever data does exist for those things into a percentage of completion. Sometimes this is based on labor expended vs. labor in the budget but just as often it is the result of a seasoned weather eye summing up what is visible on the shop floor.
Previous posts covered how inexpensive it is to collect data from machines, every minute of every day and how easy it can be to sell people on processes and tools that shrink their workload while collecting more data. But what's the point?
- Push-button access to a one page, real-time overview of your project covering: RFIs, labor hours, change orders, shipping status, materials on hand, purchasing and more.
- Instant access to real-time production progress summarized by erection sequence or production batch. Questions about the details? Need to know which specific assembly is keeping that Batch marked "in process"? Just a click away.
- Constant, effortless dashboard feedback about which jobs are shipping and how the work that you need for your shop is flowing through engineering and approvals.
- Confident material multing/nesting/allocation knowing that material-on-hand data is accurate, right now.
- Click on a part or assembly in the BOM and find out its production status.
- Click on a task or work area on the Production left-to-do form and know how much work is currently available to process through the saw or fit-up bay.
- A list of assemblies welded, today, so far.
And it's all available from one source, full of insightful reports that can easily be customized to deliver even proprietary information that you know is the key to your particular business and customers.
Data is valuable, in and of itself. But it's also overwhelming in this new world where you can collect so much, so fast. This can discourage steel fabricators from accumulating the information. If you can't get on demand, automated analysis and easily add more reports to answer different questions, then there's little point in collecting more data than you can easily review and understand in a few minutes. Instant analysis is possible though. For more, read the post coming next week: Capturing insights by connecting things and people (5 of 7): Automated analysis.
In the last post about connected things, I hope it was apparent that machines can quietly increase visibility into your part production status and performance with no impact to their productivity. If this data is accessible in conjunction with what you know about change orders, shipping, labor hours and other factors from your business, the advantageous cost (virtually none on a day to day basis) to benefit (important facts at your fingertips in the context of other important facts) ratio is obvious.
People are different. They get tired, experience stress and forget things. Often, and certainly in my case, in that order. Add something else for the machine software to collect and communicate and you won't stress it or impact it's productivity. For a person it can be the straw that breaks the camel's back. People know this and will resist new tasks, especially if they can't see the benefit to them and their work.
There is a magic bullet
I recently read an interesting blog that claimed there was a magic bullet for learning multiple languages. Nonsense, right? Actually you probably already know that they are right. Take any 2 year old and have them spend much of their time with someone who only speaks an alternate language. No special curriculum or teaching method is required. The child learns both languages because that's the period of their life when they absorb these things as part of all of the necessary daily interaction.
Capturing data about people and processes can also be a side-effect, something people can do without even thinking about it. This is easier to set-up than a machine and there is almost always a direct effort-saving benefit to the person doing the work.
It requires a different way of thinking though:
Stop using software after the work is done, to record and communicate your work. Instead, insist that the software is used to do the work and expect logging and data capture to be automatic.
Already doing this? Let's find out:
- Does someone in your office find and gather the latest revisions of fabrication drawings into a single file folder from time to time?
- Is there a regular process of pulling change order summaries together into a document that accounting or management can consume?
- Does someone update a spreadsheet with percentage of completion information (data? or guesses?) before every production meeting?
- Do you find out that sometimes fit-up is reported as 80% complete because 80% of the labor budget is spent and in fact you do not really know where your are on fit-up of that critical batch or sequence?
- When you send a transmittal, is the status impact on each drawing tracked? If so, does someone go on to note the change in status for each of the drawings on that transmittal?
- When a machine operator processes a part or cuts raw material, do they search on the network (or a thumb drive) to find the CNC files to match the paper cut list or production list they are looking at?
These are all signs that you use software to record and communicate your data--but not to do the work itself.
You can eliminate much of this effort and gather more accurate and up-to-date data as a side effect.
If software was doing the work:
- The latest drawing revision would be available to the iPad in your shop without relying on someone to organize the latest revisions into file folders.
- Budget impact of change order requests would automatically be apparent as they are submitted and approved or denied.
- Percentage of completion could be fetched on demand and would be based on actual work complete vs work scheduled on a batch.
- Sending a transmittal would automatically update the status of drawings on that transmittal and add a record of that status change to the drawing log.
- Files to process the next beam into parts would automatically arrive at the machine together with the relevant line on the cutting plan or production plan.
Thinking this way, it's easier to connect feedback from people than to connect feedback from machines. Why? The processes and tools that collect this data make it easier to do their daily work. They see the direct positive impact to the speed and quality of their own effort, and don't have to remember to pull together summaries or log what they did. Then the connected data starts flowing and delivering the valuable insights you need to drive your business.
We've been assuming that the data from these connected things, people and processes isn't isolated by vendor, brand or department. For more about that, read the next post, Capturing insights by connecting things and people (4of 7): Integrated data from the things, processes and people.
Connected things are everywhere but steel fabrication shops are full of disconnected things, even when those things are driven by a computer.
Our photos are automatically uploaded to cloud storage and our calendar sends text messages to our phone. We can see graphs online representing how many steps we have climbed today. Cameras capture our speed and license plate number and then computers automatically send us traffic citations.
In fabrication shops, though, we see people programming by hand, right at the CNC machine controller, while the attached, tireless machine waits for instructions. Files are delivered to the machine on a thumb drive, much like we used to receive faxes in a physical in-box on our desk.
Other steel fabrication shops have machines that know the next dozen beams lined up to go through the saw. Sometimes the machine, not the operator, guides cut beams on to the drill line and others past it, straight to fabrication. As the beams are processed, the machine removes them from the queue and knows not to process that beam again. A network based server is updated as more beams are loaded and lined up for processing. When it's all working, the operator looks bored, may not even be near the machine or is doing work to get ready for a part or some raw material that will be processed hours from now. Progress, and the time it took to do it, is logged.
This is pretty sophisticated for a job shop but the data collected from the CNC machine work cell, while valuable, is often disconnected from the rest of the business.
The CNC machine server knows that a part is complete but can you view that together with the rest of the assembly or project story? Are the other parts, not processed by a machine or at least this brand of machine, ready for fit-up? Are the latest revision of relevant documents available to the shop? Anything on hold? How long did it take to complete that complicated column, including the processing time the machine captured, the fit-up, painting and welding?
It's possible to capture all of this from machines and bring it together with other data about a project, assembly, or overall shipping progress across all projects. It's almost effortless after you set it up.
And there's the rub, investing time in the original setup. If you had this data for the next three months, with dashboards and reports making it unnecessary to get out of your chair, pick up a phone or send an email to find out everything that happened to 100B1 (onsite and too short) or whether loads for any of the projects you monitor shipped this week, you would never go back. Once the system is working, it's better than free. It takes minimal effort to use everyday and repeatedly helps you avoid mistakes, save money and save time.
What about delivering data to the machine? Is the "thing" connected to the rest of your material and production plan? Can it receive cut lists and production batch plans electronically? How does the operator find this data and how much effort is expended making sure that the right CNC file is used? Is this in sync with how documents and data are delivered to other processes and machines? With the proper setup, much of the effort to deliver this data to the machine is eliminated. This means better quality and better coordination between all of the people, departments, fabrication bays and machines working together to deliver multiple projects to different job sites.
FabTrol's current software implementation staff has used FabTrol software to deliver this kind of value to hundreds of steel fabrication companies world wide. Once it is working, it does need revisiting every year or so, but day to day data capture from machines just happens quietly, with no extra human effort, as a side effect of doing the work you have to do anyway. The software literally does this data capture work for you and pulls this feedback together with data from processes and people throughout your company.