Ayrton: Today is all about CAD, Computer Aided Design, which is critical to mechanical engineering and product development. I'm here again with Daniel Berti, our Senior Mechanical Engineer, to look at the key CAD software and processes we use at Element.
SOLIDWORKS was the first tool that we purchased in 2010 when I opened the business up. It’s probably the most mainstream, widely adopted, and the easiest choice of CAD software for different companies to exchange files and things like that. It's our primary CAD software that we use here at Element.
A lot of the things we'll talk about today will probably be in SOLIDWORKS, but the principles will apply to CAD in general, computer aided design.
What makes for really good CAD?
Daniel: I would say that having a logical structure and a well-organised feature tree that makes sense is a big one. For us, having a model that can be picked up even by multiple people is important. There's nothing worse than having something that one person has done that's really messy. They're away, and someone has to jump on that model and fix the mess.
Every model is different. There are a few ways of going about it, and a few different types of modelling as well. You’ve got to gauge what the customer actually wants with that CAD model.
Ayrton: Yeah, it can be done in a number of different ways. In our experience, doing CAD using the tools in an unstructured way can lead to fast results for small models. But once you start working on a bigger model or have multiple people working on the same model at the same time, you've got to have some sort of consistent way of doing things.
One of the biggest things that we've worked on or developed over the years is just ways in which we continue to do CAD. Seeing other people's models as well, some companies we work with do great CAD, very similar to the way we do it, which aligns with how the software works best. Then there are other ways we've seen that are pretty bad.
Daniel: A well-organised feature tree is a big one. Just being logical in how you lay out the CAD model. Your sketches define or form the baseline for your model. A big one that we push these days is making sure sketches are named clearly so it's obvious what features are being driven from that sketch.
In this case here, we have the model of this tool mounted to a vehicle we've done for a client. I have the driving part open, which shows how we've laid out the layout sketches or the driving sketches that form the baseline of the model. You can see here on the feature tree, everything's laid out quite nicely. This sketch here is the chassis tray reference, a side view of it. The winch, the different toolboxes laid out on top, it's basically easy to follow.
Ayrton: One thing that a lot of people don't understand about CAD, especially 3D CAD, is that a lot of the time we're working in 2D. We're laying things out in 2D with different planes and such. In programs like Blender, you're kind of moulding clay. But in CAD, you're structuring a model in a way that's manufacturable, which means you need reference planes and so on. Parametric 3D modelling is key to doing really good CAD in our experience. Having sketches where one sketch references the features of another is how we've found SOLIDWORKS works really well.
If you have a feature referencing another feature, that edge or vertex or feature changes. We've found over the years that it can basically crash the model and cause a lot of issues. So sketches built on other sketches, put into a driving part, is how we've been successful in doing our CAD. That's what we've seen from other clients and customers when they're doing good CAD as well.
Daniel: For the most part, this is the method we use for top-down modelling, where everything is driven from your driving part. All your critical features and dimensions are driven off the sketches. This makes it really easy to change dimensions later on without breaking the model.
Ayrton: That one feature you change in CAD or in the sketch, anything that's driven off that will update accordingly. This is particularly important for us because if we're doing iterative analysis and simulation, which is basically how we built the business, we start with an initial design. That design is the best we can think of, whether on paper or in 3D. Then we run it through a simulation package, find the hotspots, and have to change some of the geometry. Moving a part in a sketch and having everything else update is magical. It allows us to turn around the iteration time. But if we change something and the whole feature tree crashes, we can't turn around that FEA.
Daniel: Even the thicknesses of plates and stuff, if you define it in the sketches and it's all driven off the sketch rather than putting your dimension in the Boss Extrude feature, it makes it really easy to tweak.
Ayrton: You can, of course, drive off the thickness of the Boss Extrude feature, but it doesn't always work. Good CAD is a compromise, I guess. This might seem like a lot of upfront work, but it's horses for courses. Doing this amount of work for a complex model that needs to be changed and improved over time makes sense for this model.
Daniel: There are other general things too, like minimising your feature count where possible. Back when computers weren't as powerful, you didn't want to use 300 features to create something simple. You want to group things smartly. A good example is Magna, remember the old tyre models we used to do? The sidewall text was modelled with thin lines.
Ayrton: It looked great for the renders but was horrific for the feature tree.
Daniel: There are different ways you can speed up the model. Making sure everything's well-defined, context, references, and so on, makes your model run better as it gets bigger.
Ayrton: Good CAD is a lot of trial and error, figuring out what the program can handle.
What about bad CAD?
Daniel: We've seen some shockers. A recent one was a client that came to us with a set of drawings that a client asked us to quote for, a custom spearfishing gun. They had actually made a few batches overseas. They called up and complained that what they received back between the batches had varied massively and there was just no consistency to it. I asked him to send through the manufacture package and I went through it. And it was pretty clear from the outset why that was the case.
The drawings were missing critical features such as the surface finish, the chamfers weren't spec'd properly, dimensions were missing. That was just a case of someone doing the work that were architects who didn’t know what was important to include.
Ayrton: That's back to a DFM piece, right? Designing for the manufacturing processes, understanding what the manufacturing process might be, and what the people on the shop floor will be using to make that. And then specifying in the drawings what that person needs to know.
Daniel: Which is why having an engineer do that helps so much, because we know what is needed. We have review processes, checklists, with everything like surface finish, dimensions, tolerances, etc.
Ayrton: We've come into jobs where we're extra hands to do a certain part of the job. And I remember one which was a very similar project to this one. The model was done in a way we call ‘smash CAD’ to start with. You're just fleshing out the physical features. And the model gets bigger and bigger and bigger and gets slower and slower and slower. So for us to have any useful impact to be able to change designs or create drawings, the model was so slow that we had to actually go back and redraft it or re-CAD it, using a top down modelling technique.
Daniel: It was basically going through the design, analysis, optimization process, and the geometry. And every time we would try and change something, the geometry would break, essentially.
Ayrton: It's horses for courses. You've got to look at the scope of a job and figure out, is this something that the customer can get some quick wins with a single design process with no iteration, or are we going to a full production drawing version?
Daniel: Are they even building the model? Are they using that for manufacture, or is it a marketing thing, or just for analysis? Which happens in many instances where we're just, we're using a model that's given to us. The manufacture drawing's not even done on our side.
Ayrton: There's CAD design, and then there's CAD drawings. As engineers these days, we seem to do the whole lot - coming up with hand sketches, doing the calculations, and putting it into 3D CAD, often as you're doing the engineering.
And then a lot of the time our engineering team will actually output the drawings. Some businesses will have specific people for this, but we find that the engineers who are considering the design for manufacture processes involved will probably be the best ones to output the drawings. In SOLIDWORKS we've got templates that make it easy to put drawings out. We have a specific style that we like to follow which is easy for someone to read on the shop floor, using isometric views, zoomed in views, etc. But the person who's designed it is usually the best to tell that story in drawings.
How does CAD fit into our product development process?
Daniel: When developing any product, we always advocate starting with sketches. CAD is fantastic-it's the first step to creating a 3D model. From there, we move on to manufacturing drawings. CAD also enables us to create marketing content, like renders, and conduct analysis.
For manufacturing, CAD allows us to output DXFs for laser cutting, engraving, and machining, as well as step files for other processes. Without CAD, the old-school method was to draft everything on paper, which was much more laborious. Often, a drafter would complete it on paper before transferring it to CAD, but it all begins with sketches.
Ayrton: We're strong advocates for doing as much as possible with pen and paper or on the whiteboard. However, that only gets you so far. Even after measuring and fleshing things out, moving to 3D is the next essential step. 3D modelling helps identify clashes and fitting issues early on. After that, analysis is crucial to understanding stresses and structural integrity. From there, iterative changes are necessary. CAD is integral to our entire product development process, but we make sure not to jump into it too soon.
Daniel: SOLIDWORKS is our primary CAD package for design and manufacturing preparation. However, we also use other CAD tools for tasks like FEA preparation, which can significantly speed up the process. For example, when performing FEA, we now quickly transfer models from SOLIDWORKS to SpaceClaim, which allows us to simplify and modify features more efficiently.
Ayrton: Before that, we would stay in SOLIDWORKS, creating simplified variants of models. But SOLIDWORKS wasn't as effective for simplification as newer tools are now.
Daniel: Another key tool is SOLIDWORKS PDM, which we've used for the past few years. It enables us to collaborate on projects as a team—something that wouldn't really be possible in SOLIDWORKS without PDM. It also standardises the review and approval process within the software package.
Ayrton: Our team has grown a lot over the past few years. Before that, we didn't need PDM because only one or two people worked on a project at most. Back then, sharing files on a network drive was fine. But now, PDM is vital for a larger team. Of course, it's not for everyone, if you don't need it, there's no point investing time and effort into it. But if it integrates smoothly into your workflow, it offers many advantages.
Daniel: Another benefit is how CAD ties into the electronics side of our work. It allows us to collaborate seamlessly with the electronics team.
Ayrton: Yes, we have SOLIDWORKS Electrical, but we also use...
Daniel: Altium CoDesigner?
Ayrton: Exactly. Altium CoDesigner with SOLIDWORKS has become incredibly powerful. I've used it since it first launched, and initially, it worked well. But then something went wrong, and we ended up importing step files of electronics again, which was cumbersome. However, over the past six months, something changed, and now it just works. You can import an Altium file into SOLIDWORKS CoDesigner, and it automatically generates a SOLIDWORKS 3D model of what you've designed. You can then insert that 3D model into your mechanical design, check for clashes, and make adjustments in SOLIDWORKS that synchronise back to Altium. It's been a game-changer for our electronics workflow.
For more electrical-focused tasks, like building control boxes or routing cables, SOLIDWORKS Electrical is excellent. We now have a licence for SOLIDWORKS Electrical, which allows us to quickly assemble all the components for a control box. Whether it's integrated into a vehicle solution or another system, we can efficiently route cables, calculate cable lengths, and generate a bill of materials. It's a fantastic tool.
What software do we use at Element for creating CAD models and drawings, and why?
Daniel: We use SOLIDWORKS, which is probably the most widely adopted software globally. There are other products out there, but for us, it was a natural choice. When I started, everyone was already learning SOLIDWORKS at university. It was the standard. So, when we bring on graduates or new hires, they are typically trained in SOLIDWORKS. Any mechanical engineer in our region knows how to use it.
For us, it's crucial that the CAD package is easy to use, has a clean interface, and offers comprehensive features. It must handle everything from simple tasks to complex surfacing efficiently and neatly.
Ayrton: Cost-effectiveness and time efficiency are critical too.
Daniel: That's why we're always on the lookout for alternatives. SOLIDWORKS isn't the cheapest option.
Ayrton: But it gets the job done and has all the features we need. Many in our industry, including international clients and suppliers, also use it. When we need to send drawing packages, sharing SOLIDWORKS files makes collaboration seamless. They can view all the features and even make adjustments if necessary.
We've also explored cloud-based solutions like Onshape over the past few years. Onshape was developed by the original creators of SOLIDWORKS after they sold it to Dassault Systèmes. It has a great workflow, but adopting it would require a major shift in our business processes. Given our team size, that's a significant undertaking. We've used it for smaller projects, but it lacks some of SOLIDWORKS' features. For delivering the best value in time and cost to our customers, SOLIDWORKS remains our preferred tool.
Daniel: Tried and tested.
Ayrton: SOLIDWORKS has improved in some areas, but overall, it has been pretty stagnant.
Daniel: For about a decade!
Ayrton: Our first license here was in 2010. By 2014, workflows and capabilities had improved significantly. I remember working on complex tyre models with intricate tread patterns and large welded steel structures for truck trays, often involving hundreds of parts. Back in 2010, we had parametric, dimension-driven models where changing a few sketches could update the entire drawing package. It was impressive for the hardware we had back then. But today, it feels like the software hasn't progressed much, and now we need supercomputer laptops to handle the same tasks.
Daniel: They've essentially forced users to upgrade hardware every two to three years to maintain performance, which is the main gripe I have. It's frustrating. You open a model today that you opened ten years ago on much slower hardware, and it performs the same or even worse. It makes you wonder, why?
Ayrton: It would be great if SOLIDWORKS would tell us. I'm sure many in the SOLIDWORKS community are asking the same question. But at the end of the day, if you keep your hardware updated, performance has improved. The graphics are fantastic, and there are many features included in the licensing—though most go unused. I wish they focused more on improving the core parametric performance, bringing back the efficiency we had in 2014. That would be amazing! Still, SOLIDWORKS gets the job done, and we love to hate it.
Daniel: It works on our computers. We have decent systems, so...
Ayrton: Well, we have to. There's no choice.
Daniel: Just need to budget in the upgrade next year!
What would ideal CAD software look like?
Daniel: I think it would have a simple, intuitive interface that's easy to use. Robust 3D modelling capabilities are essential, handling everything from simple to complex shapes. It should include all the features we need for our work. For us, that means efficient tools for sheet metal, weldments, and structural members.
SOLIDWORKS works well for this—it's super simple to use. Lately, we've relied more on 3D sketches to simplify designs. The model I showed earlier was mostly done with 2D sketches, but now we'd probably use a combination of 2D and 3D sketches for better efficiency. Access to comprehensive design libraries with standard fasteners is also crucial.
Ayrton: Ideally, it would connect directly to manufacturers or suppliers. While this is possible now, it requires different plugins, and each works differently. A standardised way to import parts would be a game-changer.
Daniel: One great tool we use is the McMaster-Carr add-in. It's a tab in SOLIDWORKS where you can search by part number or browse their catalogue, download a part, and drop it straight into your assembly.
Ayrton: I saw a meme the other day saying McMaster-Carr holds up the whole of the entire US innovation ecosystem. It's helping us in Australia too. We buy from them or integrate their parts into our models because it's so easy to find.
Daniel: And it is so well laid out.
Ayrton: They have everything there, which is great.
A multi-user ecosystem where multiple people can work on a project simultaneously is also important.
Daniel: Simulation and understanding motion kinematics are vital for us. SOLIDWORKS has features that let you analyse forces across motion ranges and optimise designs accordingly.
Ayrton: That has improved significantly in recent years.
Daniel: Setting up mates and running motion studies is now quite straightforward.
Visualisation and rendering are also critical. SOLIDWORKS Visualize is now integrated, making it easier to produce high-quality renders for clients throughout product development.
Ayrton: Some clients respond better to photorealistic renders than CAD models.
Onshape has cool features like Git-style revision control where you can see real-time collaboration. It’s cloud-based, though. I'm still not sold on cloud systems—I'm old school.
Daniel: We benchmarked it, and it performed better than I expected. Their cloud service, likely through Amazon, was surprisingly good.
Ayrton: Was it comparable to a desktop?
Daniel: For most models, yes. The consistency of access anywhere was a plus, especially for me as a laptop user. My graphics card is decent, but cloud performance might even be better. However, it didn’t perform as well on complex models like Magna's tyre designs.
Ayrton: Full surfacing geometries and large multi-patterns are tricky.
Daniel: It auto-scales model quality based on complexity, but we couldn't get it to look as good as SOLIDWORKS.
Ayrton: That was a while ago. Maybe we should test it again.
Daniel: Let's request another trial or even get a licence to explore it further!
Ayrton: Many CAD providers are pushing for cloud adoption. I’m not convinced yet. We've built a strong infrastructure over the years and we've got perpetual licences. We'll stick with that until we're forced to change. Cloud solutions would need to offer a clear workflow improvement for us to switch. So far, I haven't seen that. Marketing teams pitch a lot of hype, but we often know more about the product than they do. We're in it every day, so we understand its real capabilities. If a solution lowers costs and speeds up delivery for customers, we'll adopt it. But I haven't seen that yet.
Daniel: I agree. We're always thinking about improvements. In the mechanical team, we're keen to explore new tools that outperform our current methods.
Ayrton: It's a bit of a let down, right? Like with Altium, we adopted it early, over ten years ago. It has steadily improved every year with useful features. SOLIDWORKS feels stagnant. When I return to it for conceptual design, it feels the same as it did in 2014. That's frustrating. We're advocates for it, but it’s a love-hate relationship. Why pay for upgrades if the 2014 version still works on a 2014 PC?
Daniel: Why can't we just stick with that?
Ayrton: Sure, the 2014 PC will eventually fail, but still.
We have a love-hate relationship with the vendor. Sales reps often oversell SOLIDWORKS features, and we've seen it for years. They should be honest about what the software can actually do. We work with it every day and know its strengths and weaknesses. If you want to sell us on improvements, show us real, impactful upgrades—not features we don't need.
Daniel: I remember a SOLIDWORKS event a few years ago where a rep told me their simulation tool was better than ANSYS.
Ayrton: They've been saying that for years. Back in 2010, they claimed Abacus would integrate into SOLIDWORKS, but it never happened. The industry doesn't need that kind of misinformation. We need clarity so we can improve outcomes for customers and drive innovation.
SOLIDWORKS user groups used to help with that, but they've faded, especially in Perth. Resellers turned them into sales pitches. I'd love to revive those groups. There are great user groups elsewhere, especially on the East Coast, where users support each other. We need that, sharing tips and overcoming software limitations. Just like our mobile app developer meetups. In engineering, we shouldn't compete over software knowledge. Collaboration grows the innovation ecosystem.
How can we generate CAD models quickly?
Daniel: You need to understand what the CAD model is being used for. If it’s a complex project going through multiple design iterations or leading to a manufacturing package, the approach is very different compared to creating a quick model for visualisation. Over the years, we’ve developed a collection of template models. For example, we have a client we produce tyre renders for-sometimes these models need precise, millimetre-level accuracy for production, while other times they’re just for visualisation.
Our approach changes depending on the need. For projects like transport frames for mining equipment, we rely on template models with features like fork tines, tie-down points, and lifting points. This allows us to adapt existing models quickly, sometimes even with partially completed drawings, saving time and reducing costs for the client. Smash CAD, as we mentioned before. If the model won’t be revisited, we can create it quickly without worrying about long-term usability.
Ayrton: That’s mainly when we’re translating sketches from paper or a whiteboard into 3D to understand the geometry. If it’s not going to be manufactured and it’s a one-off, that’s fine. But if revisions are needed, it’s best to switch to proper top-down modelling,either simplified or detailed, depending on what’s required.
Daniel: Sometimes we start with a quick model for early analysis and then move it into SpaceClaim for FEA preparation. SpaceClaim has great optimisation tools for quick parametric analysis. Once we optimise and finalise the design, we rebuild it properly in SOLIDWORKS, especially if it’s going to be used for manufacturing and might evolve into later versions. That’s what we did with this model.
Ayrton: This is a production-grade model, and you can see the feature tree is a bit messy. It’s acceptable as long as it delivers what the client needs.
Daniel: In this feature tree, everything is named according to our part numbering scheme, which we apply across all jobs unless the client provides their own. We also follow good modelling practices, using assemblies and subassemblies to separate components that will be manufactured independently.
Ayrton: The driving part contains sketches that define the entire model's geometry. From there, we build parts within the assembly or subassemblies. In multi-user environments, segmenting product areas into subassemblies is essential. The driving sketches outline these sections, and subassemblies handle specific details.
Daniel: In this case, we laid out all the toolboxes and components on the truck. We had to model the truck accurately, which involved on-site measurements. Some of our team were physically present to build the model alongside the assembly.
Ayrton: We often use tape measures, levels, and string lines for quick measurements. Laser scanning has its place, but manual measurements can be faster and more practical for some tasks. Scanning and processing point clouds can be more work than it’s worth unless it benefits the end user. Our teams take measuring kits—pen, paper, and tools—to site and build models directly on laptops as they measure.
Daniel: Once the truck and tray are modelled, we work on the layout. The client shares their preliminary designs, and we collaborate to position components. Early layouts use simple block sketches, like rectangles, to represent items like toolboxes.
Ayrton: Any interfaces between subassemblies and the main assembly—like bolt holes—are detailed later.
Daniel: Exactly. After finalising positions, we create the detailed subassemblies in the correct locations.
Ayrton: And we detail those directly within the subassembly.
Daniel: Right. This method ensures that changes in one area don't impact the entire model.
Ayrton: It also enables multiple team members to work on different subassemblies simultaneously without causing conflicts.
Daniel: That’s exactly how we manage complex projects efficiently. This process works very well with PDM.
Boss Bros Water Tanks
Ayrton: This is one of the projects where we used parametric modelling.
Daniel: It was fully parametric with equations and everything. It was very intelligently built.
Ayrton: A cylindrical water tank made from light sheet metal with a standardised bolt pattern on either end. This customer wanted different-sized models, so we created a parametric model. This was from around 2010 or 2012 when you first started, Dan. It took some time, but we developed a top-down model defining the overall geometry.
You can see different diameters and parameters that are all driven. Parameters like panel length, panel width, number of panels, and panel height allow us to generate various models. By using standard panel sizes with consistent bolt patterns, we could create numerous size variations. I’ve done this before for other customers needing production versions or design variants, like DT HiLoad Australia and Boss Bros, which later became a different company. Instead of redrawing every version, we’d adjust the parameters, and the model would update. Ideally, we could produce drawings directly, though some cleanup was often needed due to occasional model breaks.
Daniel: But it was relatively fast. I remember resizing a model could take just one or two hours. Adjustments like increasing diameter or height were simple. I think there were instances where the ladder might not have even had a ladder. It was almost a template model, we invested extra time upfront to make it easier to configure multiple tank sizes later.
Ayrton: It’s an investment piece. The first model might take 10 hours. The second could take 8 to 10 hours. But creating a fully parametric model might require 40 hours of work. You make it once, but when adjusting sizes, things may break. You return to the model, refine it, and repeat this for several variants. Once stable, each change might only take an hour or two, depending on the model's size and complexity.
For customers like Boss Bros and DT HiLoad Australia, there were countless size variations. DT HiLoad would specify truck trays with specific dimensions to carry loads from massive diggers or specific ore quantities to maximise payloads. Even 50mm changes in width or height could differ between mine sites. Initially, it could take months to build and refine the model. After that, it transitioned to generating production drawings while improving the parametric model. The DT HiLoad team continued refining the model over the years. These models can become cumbersome if not regularly maintained, especially with changing SOLIDWORKS versions.
Daniel: When creating these models, it’s critical to understand how they’ll be used. Even with dimension-driven parametric models, you need to anticipate changes and design efficiently. That means capturing those details in your sketches and layout drawings.
Ayrton: This model is a solid example of a productionisation system. You can build this in SOLIDWORKS or other parametric CAD software. But sometimes, you only need to flesh out ideas in 3D when it can't be done on paper. That’s where quick 'Smash CAD' models come in.
Magna Tyres
Daniel: Magna models are some of the heaviest we handle graphically. There are lots of small, intricate features. Our main goal has been to create high-quality renders for marketing, while still accurately reflecting how the tyres would be manufactured. Knowing we’d produce these models regularly, it was essential to establish solid baseline sketches for the tyre's base geometry, which we could then build the tread onto.
Ayrton: It's interesting to see some of the sketches I did during my thesis still being used.
Daniel: They’re still there! At least in this model.
Ayrton: Magna Tyres was one of our first clients back in 2010. Part of why I started the business was because I got busy with DT HiLoad, and Magna approached me due to my previous tyre engineering work. My thesis focused on tyre design, specifically how heat builds up in the tyre due to different compounds and how the casing affects deformation. The tread pattern plays a role in dissipating heat through air convection.
This is a radial tyre, with steel casing wires wrapping around the bead. We modelled the rim, flange, bead, casing wires, radial plies, and the different rubber compounds. Then we added the tread pattern, which provides traction and dissipates heat. This model could be used to produce tyre moulds, but our main use was for tyre renders. Over the years, we’ve also done simulations, certifications, and iterative improvements on casings and tread patterns.
Daniel: Here, you can see our layout sketches on the front plane, forming the tyre's base.
Ayrton: This part represents the casing’s rubber compound but not the tread. Next, we start adding more features to the model.
Daniel: We extruded the main lug pattern and patterned it around the tyre. Then we cut sections away to form the outer tread profile.
Ayrton: That’s one method. There are others.
Daniel: I think our current method is different. This is the older approach.
Ayrton: It’s a simplified process that works for basic designs. For chunky lug patterns, this method isn’t ideal because the extrude wraps around the radial profile, but most lugs should radiate outward.
Daniel: Now we use the wrap feature with surfaces, which is more effective.
Ayrton: Again, horses for courses. Different methods suit different needs. This approach was perfect for generating visually appealing renders quickly.
Daniel: It’s also easy to follow. The simpler method makes sense, though the newer one uses more complex features. You can see extra detail here with the cuts and small ridges, though fillets haven't been added yet.
Ayrton: We're gradually cutting away sections and building up the model step by step.
Daniel: Next, we add fillets.
Ayrton: Fillets can be very resource-heavy.
Daniel: Exactly. Our current process is different. Now, we focus on completing one section before applying the pattern, whereas before we applied features across the entire model.
Ayrton: We’re adding detailed features to make the model look more realistic. It’s common to have many sketches visible, which might look messy. But when you're designing, those sketches guide features and don't get in the way. For others, it can be overwhelming. Structuring the model so sketches can be hidden or revealed allows others to explore and understand how the model was built. That’s crucial in CAD modelling.
Daniel: This is a good example of a model that could benefit from better feature naming and grouping. Early on, it was just a few of us working on these, mostly one at a time. You started, then Ben and I joined in, and we quickly produced about 50 models over a few months. At that pace, we didn’t think much about how others might use the models.
Ayrton: Well, you were giving the customer the best bang for buck, right? Naming every feature takes time, and if it’s not necessary, it’s often fine to leave it. But if you’re in the flow, it’s easy to name features as you go.
Daniel: Now we have a document outlining each step, what to watch for, and how to prepare models for rendering. This covers everything from dimensions to sketches and even sidewall text, which now has three variations. In this model, the text is a simple white extrude, but there are versions where it’s striped or matches the rubber colour.
Ayrton: The logo was a real challenge back then. We had to import Adobe Illustrator files into SOLIDWORKS and create features from them. If you zoom in, you’ll see the fine detail. It doesn’t need to be that detailed, but it works and looks great in renders.
Daniel: To keep models running smoothly, the logo is a separate part or subassembly that we add later. Splitting the model into modules or subassemblies makes it much easier to manage. Trying to do everything in one part or directly in the assembly would be a nightmare, at least for my computer.
Ayrton: For most computers, really!
How did scalable models help us generate fast production drawings for DT HiLoad?
Ayrton: We helped DT HiLoad design and optimise massive dump truck trays for mining operations. The model we have here is from around 2012 or 2014. This is an underground truck tray we developed, and it was very successful. Ben (our first employee), Frenchie, and I went out with measuring tape and hand tools to measure an underground truck, and built this model. We simulated it using Abaqus and iteratively improved it by eliminating stress hotspots.
This tray served thousands of hours in the field, with only one issue at the back where the angle of repose caused hang-ups, which we fixed with an additional plate. This model was initially developed for FEA but evolved into a full production model. Here, we're looking at one stage in the production process, showing how different parts of the structure are installed in phases.
You can see lugs on the red piece. These lugs were likely used to hold or assemble that plate during manufacturing. We model each production stage in CAD and adapt it for the next configuration, which leads to the next set of production drawings. This looks like the complete production model. Our naming conventions were different back then, and we’ve just opened this from an old server. This top-level assembly includes routed lines for installation and guides how we created production-grade drawings for each stage of manufacturing by nesting new parts onto previous assemblies. Here, we’re adding a floor subassembly, two wheel arch assemblies, two sidewall assemblies, a front wall assembly, and individual parts. For the floor assembly, we would produce an assembly drawing to bring all these parts together.
Daniel: When building a model, you must understand how it will be assembled.
Ayrton: In this model, we’re adding a rolled floor section as a subassembly, several rolled parts welded together, along with multiple beam and crossbeam assemblies. Each subassembly had its own drawing package.
Another crucial aspect is jigs and fixtures. Many people overlook the need for fixtures to hold components in place during welding. Fixtures ensure stability and allow workers to safely assemble parts using cranes and lifting tools. We also created jig and fixture models that were part of the drawing package for the customer.
Daniel: We collaborated closely with DT HiLoad throughout the process, especially with the shop floor team, to ensure everything could be built efficiently.
Ayrton: Yes, after completing the FEA and the design, we worked directly with the workshop manager. We reviewed each subassembly and discussed how fixtures should be made, sometimes starting with hand sketches or whiteboard drawings. Engaging with the fabricators is often overlooked but essential. You must value their skills and the time they invest. Assuming things leads to poor outcomes and strained relationships. When shop floor teams are involved, they contribute their experience and ideas, which improves the project. Their input allows us to engineer the best solution for time, cost, and weight. It becomes a collaborative effort. Many young engineers don’t appreciate how much knowledge fabricators, machinists, and suppliers offer. Building strong, collaborative relationships is key.
Daniel: Another important point is cost-effectiveness. Producing manufacturing packages is time-consuming—drawing assemblies, parts, and outputting DXFs and STEP files. If changes are needed after submitting these, it can be a lot of extra work. That’s why collaboration is so important. Everyone agrees on assemblies and designs before we finalise the manufacturing package. Most of our clients prefer this integrated approach because it ensures they get exactly what they need.
Ayrton: Exactly. We need to respect their expertise. We’re not experts in their field. Often in FEA, we’ll report a hotspot in a model, and the client will confirm they’ve seen cracks in that exact spot during field tests. That feedback helps us understand the real-world factors behind it. Collaboration is essential for cost-effective and timely engineering. Unfortunately, I’ve seen engineers who think they know more than shop floor teams or clients—and that mindset doesn’t work. We’re always learning on the job too. This model uses some of our older Element workflows.
Daniel: But it still follows logical, organised structures. Everything is clearly named, making it easy for anyone to understand. If it wasn’t, navigating the model would be chaotic.
Ayrton: Nowadays, we organise everything into folders for better structure. These chassis sketches are based on the on-site measurements we took, defining part placements. We used simple tools like tape measures and rulers, creating layout sketches to drive the product design. This particular model replicates an OEM dump box. The different profiles and references are much clearer now compared to how we used to work. Showing all of the different references wasn't a thing back in my day.
Daniel: It’s definitely useful for troubleshooting.
Ayrton: I feel like an old fart saying that!
KLS & Epiroc OreSight
Ayrton: We worked alongside KLS's team to model and engineer the Epiroc OreSight downhole logging platform to gather accurate, high-resolution data from drill holes for the mining industry. How did we approach that project?
Daniel: The project began through a contact who needed help with analysis work for a client. We assisted with the analysis and refined the design of some sub-critical components. As we got more involved, we learned the project was focused on developing a downhole assay tool, a logging tool for drill holes at mine sites. Essentially, it’s a large sensor sent down the hole to collect data on material compositions. They aimed to create an autonomous version of this tool.
Previously, this type of data collection involved coring, which was more expensive and time-consuming. In the early concept phase, the idea was to mount the tool on a roadworthy truck, take it to different sites, and run it through various drill holes. We initially supported their draftsman with design work. However, their time-to-market goal was tight, around six to nine months, so we proposed helping them accelerate the process.
Ayrton: But we found that the SOLIDWORKS model was pretty ****.
Daniel: Exactly. So at first, we worked alongside them, but as they were building it, we realised the way their draftsman was modelling things was not ideal. It was kind of Smash CAD. It wasn’t built for on-the-go changes or for planning a version two later on, which I think was always part of their plan.
Ayrton: This happens when you're creating something conceptually from hand sketches or ideas in someone’s head. Smash CAD helps to get a quick prototype, but when moving to production or building several prototypes, you often need to revisit the model.
Daniel: Their goal was to have someone help get the prototype working and provide the manufacturing packages, but also streamline the process for refining designs in future versions—V2s and V3s—to cut costs later down the track.
Ayrton: We had several of our guys stationed in their workshop for six months.
Daniel: Six to twelve months, I think it was. Pretty much full-time.
Ayrton: Yeah, our guys were on the ground with the boilermakers, fitters, and other trades. They got direct feedback, made changes on the fly, and came up with new solutions. New parts would arrive the next day, and they'd work on them together.
Daniel: That was definitely a case where we spent more time upfront, knowing their plans, to create a solid baseline driving part with well-laid-out sketches that could be changed and refined later. We built it in a much more effective way.
Ayrton: Scaling.
Daniel: Yeah, scaling the whole thing.
Ayrton: That’s leveraging our experience in getting products to market. Smash CAD helped get the initial concept up. But moving from that to a manufacturable pre-production model requires going back and doing more foundational work. To scale and create different variants or refine the model with FEA, it’s a different CAD modelling process.
Daniel: We started with layout sketches, agreeing on the sizes for different toolboxes and components. Everything was probably oversized to allow extra room for the first version. To speed things up, we focused on one subsystem at a time, getting it finished and approved so they could start building while we refined the next part. The high-level layout was agreed upon, ensuring everything would fit on the truck tray.
Ayrton: This client didn’t have in-house engineers. We came in as engineering consultants and designers, leveraging our experience with everything from tiny injection-moulded tyre sensors to large metal structures. We were their engineering team, integrated into their workflow to get them to market fast. This approach also benefited them because they didn’t need to hire or fire anyone—they brought us in for six to twelve months, and we scaled down when the work was done.
With projects like this, we understand the timeframes and costs. Bringing in someone inexperienced can drag out the timeline. Time is often the most important factor. Training someone is fine for long-term projects, but for fixed deadlines and costs, our experience makes the difference. This was a great project for us, and we’re still doing their engineering.
Daniel: Continuous improvement was a huge focus for them. They developed two or three other variants after the initial design. Knowing their fast timeline, we helped them focus on critical components that could have caused trial failures. Some less critical parts weren’t analysed upfront and failed sooner, but those were addressed later. If we had analysed everything from the start, the project would’ve stalled.
Ayrton: They might have missed their opportunity. Maybe the structure would’ve been perfect, but they wouldn’t have tested the tool. Engineering needs to focus on the end-user experience and business goals. Sometimes shortcuts are necessary—as long as they’re documented and revisited later. Our goal is to innovate quickly and cost-effectively. To support local industry, we need clear communication and collaboration.
Daniel: Exactly.
Ayrton: That’s all for today. Thanks for listening. Don’t forget to like, subscribe, and leave a review. Hot takes are always welcome. Until next time—make it good, make it fast. See you soon!