Integrating Cost Analysis into Simulation-Driven Design
Simulation software has been helping engineering analysts identify potential performance problems in product designs for many years. Recent technology advancements in this field have concentrated on simplifying interaction of the end user with the software so that a wider range of users can benefit from the outputs the product can deliver. Specifically, there has been a significant push to enable the movement of performance analysis of product designs into the earliest stages of product design.
What do Simulation Toolsets Provide Engineers?
Simulation toolsets have provided engineers with a powerful ability to understand a product’s performance earlier in the development cycle than ever before. Furthermore, software vendors and analysts have been collaborating on how to accomplish product design optimization more efficiently by combining different types of simulation software; for example, stress and thermal analysis, and enabling the setup and execution of multiple simulation programs simultaneously. Often referred to as multi-variant analysis or design of experiments, this new approach provides significant potential benefits to companies that are able to incorporate the strategy into their product development process.
A more recent introduction into this type of computer design of experiments has been to include product cost analysis as one of the vectors for optimization.
Manufacturing cost is the most critical non-performance constraint on a product’s design.
Imagine being able to simultaneously calculate the perfect intersection of a material thickness that maximizes the strength of a product with the cost to purchase the raw materials for varying thicknesses of rolled sheet stock? Imagine if you could also just “set and forget” this type of analysis, where it runs in batch mode, plotting all of the possible variants as the software seeks to discover that perfect balance of strength and cost while you are out having a cup of coffee? Benefits of this type of approach could include:
- Analyze many more design alternatives
- Faster time to market
- Improve product value
- Lower product costs
- Less late stage churn
The Benefits of Integrating Cost Analysis into Simulation-Driven Design
The following presentation, covering all of the intriguing possibilities described above, was delivered by Amanda Bligh, Solution Architect & Consulting Engineer at aPriori Technologies, in October 2016. The total presentation is less than 20″.
A Product Cost Estimation Toolkit that Reflects the Cost of Design Choices
The right product cost estimation software should provide comprehensive analysis that reflects the full breadth and interactivity of manufacturing and design choices and their cost implications.
See a Demonstration of aPriori
Design to CostWatch Now
Complete Transcript of the Presentation
“I’m going to start this presentation by making two uncontroversial statements. The first is that simulation toolsets have provided engineers with a powerful ability to understand a product’s performance earlier in the development cycle than ever before. This is something that we see with our customers as we talk with them about toolsets they use during their early development process. Next is that manufacturing cost is the most critical non-performance constraint on a product’s design. This is something you’ve probably heard over and over again, and something we hear as well as we talk to our customers and prospects regarding cost in product design. I want you to hold onto these two ideas as we start going through this presentation. What we’ll be covering is how we can bring cost into early design optimization and early simulation.
My name is Amanda Bligh. I am a solution architect and consulting engineer with aPriori. What we’ll be going through today is a short review about simulation and optimization and how those are used in the early development process. Next we’ll go into aPriori and 3D costing and how those can be used in order to better understand the cost of your designs. We’ll talk about how 3D costing and simulation and optimization can work together to provide you with the best design early in your development process. Then we’ll talk about how these toolsets can fit into your design process itself.
The general trend in optimization is to move simulation and optimization efforts earlier in the design cycle. We want to do this when our design space is more open and we have more opportunity to find, understand, and fix issues that we might find with the performance. One of the goals though as you’re doing this design and simulation is to do this in concert, not one right after the other. As we have Jeff Waters, CAE process expert, “Design-driven simulation is backwards.” You want to do both of them at the same time and use the results from both the design and your performance evaluation in CAE to help to drive to a better product.
Many teams have started to do this type of investigation with PIDO toolsets. These are Process Integration and Design Optimization (PIDO) toolsets. These might be toolsets such as Isight or Optimus or Phoenix that allow you to take and integrate various CAD and CAE tools together in order to take and figure out the best results. We have an example here of what a flow might look like. In this case we have our CAD software and some simulation software that we’re using. Then, our PIDO toolset is controlling a number of other steps. We start with our first parameters that updates our CAD file so it actually updates the geometry of the CAD. This information gets sent to an FEA toolset, for instance, or another CAE tool, where the mesh might be updated, a simulation might be run or would be run, and then the results would be output to our PIDO toolset. Depending on the number of runs that we’re doing within our design of experiments or our optimization, we’ll move on to either define our next parameters and update the CAD again, or we’ll go and take all of our results and create a graphical output.
Let’s take a look at an example use case of this. We start with a suspension arm. What we want to do is we want to understand what the best thickness is based on the loads that we have on this particular part. The first thing we need to do is we need to set up our starting point in our FEA toolset. We load our part with our particular loads that we might have, make sure that everything runs correctly, and then we’re going to take and make this an automatic execution. We use a PIDO toolset. In this case we have Isight. We have our SolidWorks file updated and simulation updated. Then we have Abaqus next to it, which will run our FEA. Now we set up a screening design of experiments with a number of standard gauge thicknesses to adjust the CAD file. We run these results, and what we see is the intuitive result that, as our part thickness increases, our max stress decreases. We can overlay our yield criteria as we see in this green box, and we can see that we have two points that satisfy our yield criteria, our two thickest materials.
If we’re an analyst or an engineer and we’re looking at this, we’d naturally select the thinnest material that satisfies our yield condition. From there we would move forward with other portions of our design or other parts of our project and settling on that thin material. The reality is that our analysis is being completed inside a larger system. We have a purchasing team and buyers working with our supply base to get the best price for all of our parts and also negotiating for raw material costs. That’s going to be based off of our current usage, so we may be able to get some particular deals on a selection of thicknesses based off of our current volumes on our current parts. Based on this particular redesign we might be doing, we may be able to leverage these particular thicknesses as an opportunity. We see a savings for our thicknesses of high volume, but for all other thicknesses we’re at a dollar per pound.
The question becomes, how do we get the cost information into the hands of the engineer or analyst so that we can avoid early decisions that might drive downstream costs? As a way of starting to discover how we can tie these two ideas together, let’s take a look at 3D costing in aPriori. Many of you have probably seen a version of this slide before. Most lifecycle costs are determined by decisions that happen early during the conception phase, and we can see that here with the red line. The reality is that the cost reduction opportunities happen also during that timeframe. As we get further into our design cycle, we’ve made decisions that cannot or are difficult to be reversed or turned around. We may have gotten to the point where we’re cutting steel on our molds or that we can’t hit schedule and cost because of early lead times for different portions of our processes or our products. As a result, we have a short innovation window early in conceptual design where we both have the opportunity to reduce cost and the opportunity to impact it.
Though this is really well understood, internal systems and processes are often not optimized for cost management challenges. Different groups have different cost objectives. There’s many tools amongst all of these different groups. These systems are not shared with the different groups. As an example, you may have engineering, where they have cost information that comes directly from quotes or talking to suppliers, and they can only get feedback at limited points in time. It may take quite a while for them to get feedback and they miss that innovation window.
What we’ve done with aPriori is we have unique value, and that is to drive cost assessment from the CAD file. Many of you may be familiar with this type of idea from a master model, where the 3D model is where all of the key information about the part and the product is stored. We’re using that same type of idea with aPriori. The first thing that we do in order to understand the cost of a product is to extract the geometry based on the inputs that we have provided. We have the design geometry itself, the material type, and the production volume as those key inputs.
The next step is that, based on the details of the CAD model, we automatically evaluate different ways the part could be manufactured. This includes mechanistic manufacturing models, machine and feasibility rules, and looking at key constraints in order to find a feasible routing. Based off of the selection of the feasible routings, we automatically calculate the cost based off of labor rates and overhead rates for various regions as well as material rates for the different material types that are available. We had data from more than 60 different global geographies in order to support these calculations.
One of the key things is that, since our cost is based on geometry, just like an FEA tool, as the geometry is refined, so is the cost. As we tighten those decisions around the design and our space starts contracting, we’re gathering information that also allows the refinement of the cost. Additionally, we’re providing a common and consistent framework for cost across the product lifecycle. Engineers can take and have early visibility to cost as they do the early estimates, and cost and manufacturing engineers can provide detailed inputs on the realities as the cost gets further refined.
We support this in a couple of different ways. The first is through the GUI, where we can provide summarized and detailed cost and manufacturing information. We can see cycle times and cost breakdowns for every step within the process. Additionally, we have a number of ways to automate this process. The first is by allowing an automated way to extract geometry. Next, we can alter a number of those inputs, again in an automated fashion. Finally, we can outdate output data on individuals or groups of parts.
Many of you may be putting two and two together on this idea of 3D automated costing and a simulation workflow. Keep that in mind, but before I do that I want to highlight that aPriori does not only handle automated manufacturing costing. We are a complete product cost management solution with a number of key capabilities that include flexible methods of costing based on user types, the ability to generate cost roll-ups, the ability to understand detailed manufacturing routings, as well as being able to view and report data at all different levels. One of the key things is that you can do this across multiple sites in order to support a global roll-out.
With all this in mind, let’s take a look back at our example. First, we’ll take and we’ll add cost as an output variable in our PIDO toolset. Here we see our original flow for the first part of our example. Now let’s add in aPriori. Now when we update our CAD file, we not only feed it to our FEA or other simulation toolset. We also feed it into aPriori, where the geometry is loaded and the overrides are completed and the results is outputted. Now when we’re recording results we’re not only doing this in the area of stress but we’re also doing this within the area of cost.
Let’s go back to our suspension arm example. As a quick reminder, we have a deal with our suppliers on our most common material thicknesses and a premium for all other thicknesses. This information we can use in order to set up aPriori. What we’ll do is we’ll start coordinating aPriori with the negotiated costs in our ERP or other material system. Here we have our negotiated costs for our specific gauges included, and then our generic costs for all others. We can load this data into aPriori on a regular basis so that we make sure that we have the most up-to-date inputs for these early-stage evaluations. Next thing we do is that we add the aPriori steps into the PIDO workflow. Here again we’re doing this within Isight. We can notice that we’re feeding our aPriori steps off of the same CAD file updates.
Now we run this simulation with our input gauges. We’re outputting stress, as we see here, but also cost. We can see again that, as our max stress decreases with thickness, so as we increase our thickness, we get a lower stress. Then also we see that our cost generally increases with thickness. We can see that we have a few points that fall a bit below the expected response service. Let’s take a closer look at these in a two-dimensional view. We can see we have a general trend. As our part thickness increases, so does our piece cost, but again we can see these two points a little bit more clearly that fall outside of the expected response line.
If we look at the selection of thicknesses that satisfy our yield condition here on the right, we can see the part with the thinnest material, the one that we would expect to be the best choice, comes in at $2.39. If we actually go to the thicket material, we see that that comes in at about 30 cents less. We can see by adding this bit of cost information based off of the realities of the purchase price of the material we can start to see that there’s some potential cost savings here. If we’re talking 500,000 parts per year, we’re seeing a $150,000 cost avoidance on this single part.
Now let’s go back to that earlier question: How do we get the cost information into the hands of the engineer or analyst to avoid early decisions that drive down stream costs? Now we can do that with aPriori. In this case we looked at a single design engineering example, but there are many others. We have, for example, the ability to see what design gives the best material usage, either on a sheet or on a coil. We can see which geometry is the most cost-effective by running a selection of different geometries very quickly. Here we looked only at thickness, but perhaps you might look at hole placement or bend placement or material removal areas. Next we can understand which part features drive part and tooling costs by running many different geometries through a automated costing sequence like this. Then, finally, we can understand what the cost impact of tolerances might be.
With cost engineering we can do some basic sensitivity analysis to understand how your cost changes in labor, materials, and overhead based on those impacts. We can understand which manufacturing processes drive the best results. We can look at more detailed sensitivity analysis around electricity and other overhead inputs and how those can change impact cost. Finally, with a buyer or a purchasing team we can understand what the best batch size is for a part. We can understand what the impacts of regional sourcing and manufacturing and cost are by running a large number of these different scenarios very quickly. Then, finally, we could understand the impacts of regional sourcing on tooling costs.
Let’s take a look at how these types of ideas can be used within a development process. We have a couple of quick case studies. The first is of Krauss-Maffei-Wegmann, where they took and they ran aPriori in parallel, so not within the context of a PIDO toolset, but they ran them in parallel, and were able to use insights from both of them to understand which part satisfied not only the cost criterias but their performance criterias. They were able to get to the best design much, much faster, in weeks instead of the months that it would normally take to do physical testing as well as working with the supplier to understand the costs. As a result, they were able to see a cost savings of about $415 per part. This ended up being a weight savings of 29 kilograms, and so a five-year savings on this part would be just under half a million euros.
Another example is a Fortune 25 manufacturer that started looking at integrating design and costing together. This was a proof of concept to start to understand very similar type of analysis as we looked at today, where they took and looked at a computer design of experiments using CAD, FEA, and aPriori to really understand those cost performance trade-offs and understand where the most optimal design might be. At the end of the proof of concept they saw that there could be a 30x increase in the number of different options that would be investigated for their designs, with a expected 15% to 25% reduction in cycle time.
Coming back to the same idea of driving changes earlier in the design, from what I’ve shown today we can see how automated costing tools in connection with simulation can really help in this early design innovation window to lower costs, improve time to market, improve product value, lower product churn, and really just overall improve productivity and collaboration throughout the development process.
I want to thank you very much for your time today. If you have any questions, feel free to contact myself, Amanda Bligh (email@example.com), or our VP of strategic marketing and product management, Julie Driscoll (firstname.lastname@example.org). Thank you.”