Engineers and Modelers

Today I had an interesting conversation with Shyamal Roy, Founder of GEOMATE, a company that develops GrafiCalc. GrafiCalc is basically a 2D graphical calculator meant to be used by engineers to help them come up with an optimal design before they hand it over to modelers to model in CAD. Shyamal has an interesting view on Digital Prototyping, which is a bit different from that of Autodesk. Well, not exactly different, but his idea of Digital Prototyping takes root a couple of steps before that of Autodesk does.

Autodesk’s idea of Digital Prototyping basically revolves around giving an Inventor user the ability to test his digital prototype for failure in Inventor itself before he creates a physical prototype and subjects it to actual real world testing. The idea is to minimize the number of physical prototypes, thereby reducing cost and saving time, both of which save money.

According to Shyamal, “post-modeling analysis shows how the product will fail whereas pre-modeling engineering helps to make sure the design will work“. To understand this you need to know what GrafiCalc does. Suppose I want to design a mechanism that is subjected to certain known loads and which must yield certain movement without failing. I can simply draw a 2D drawing of the mechanism in GrafiCalc, specify the loads, constraints, etc. and it will do the math to give me the dimensions of the parts in my mechanism (see demo). On the other hand Inventor needs you to model the mechanism completely and then test it for failure. There is a difference here. Inventor does not modify the mechanism for you to make it work. It simply tells you whether your model will work or not.

However, Inventor and other solid modeling systems that offer post modeling analysis can offer far more analysis features as compared to GraphiCalc, quite simple because GraphiCalc is basically a 2D system, whereas Inventor resides in the real 3D world. Which brings me to the real topic of this post. Shyamal is of the opinion that less than 5 percent of CAD users in the USA have a degree in Engineering. According to him, “Here you can get a 2 year course on how to fly a 3D CAD system and get a job as a designer without having to know anything about engineering.

Which begs the question. Is Autodesk barking up the wrong tree? For Digital Prototyping in Inventor to work, you need to be an Engineer, not just a Modeler (or CAD Operator), who knows how to come up with geometry by either cooking up a feature tree or directly moving faces around in a solid model. I wonder how much analysis you can do without knowing the basics of analysis, which is what engineering is all about. Personally, I find the figure of 5 percent rather shocking. On the other hand, Shyamal thinks that the number is actually quite generous. Thats why he is of the opinion that his 2D solution works in 90% of cases because engineers think and plan their designs in 2D and hand them over the modelers to model them in 3D. In his view the Digital Prototyping needs to start at the Engineer’s level, not at the Modeler’s level.

Recently SpaceClaim and ANSYS announced an integration that would allow “simulation driven product development”. After the announcement I had a conversation with Mike Payne, Founder and Chairman of the Board of Directors of SpaceClaim. Mike seems to share Shyamal’s view on Engineers and Modelers. This is what he told me:

“This coupling of SpaceClaim and ANSYS allows the person who owns the design (a work in process) to hand it to the analyst who can iterate the starting design in his world until he is satisfied. He can then go back and tell the design department what he needs to be done to their idea of the model. Remember that it is in the analysis area that a lot of the real engineering is done, and not by CAD operators. When a bridge sways, or a building collapses it is not the fault of the CAD operator, but rather the fault of the engineer who did not do his sums correctly. We have all been brain-washed into believing that today’s solid modeling systems are all for engineers. They are not. Engineers do not sit in front of CAD systems all day long, and have extensive training in how to use the CAD system. They deal with requirements, materials, heat transfer, disassembly for maintenance, wear, manufacturing tolerances, etc. etc. Put simply, today the aircraft companies have a pretty good idea of how a plane will perform before it makes its first flight. That is made possible by analysis, and other engineering tools, and not by CAD or PDM or PLM.”

I guess that explains why SpaceClaim has two versions of their software: SpaceClaim Style for Modelers and SpaceClaim Engineer for Engineers. So as I see it, Autodesk is helping modelers wear an engineer’s hat, whereas SpaceClaim is helping engineers wear a modeler’s hat. And Shyamal is looking that them both, shaking his head and wondering, “Engineers need to first use software that gives them a good first design instead of wasting the modeler’s time in modeling and remodeling“. Pretty interesting, I must say.

But whichever way I look at it, I still cannot come to terms with Shyamal’s 5 percent figure. Is the situation that bad in the US?

  • Anonymous

    I think graphicalc is interesting but it would be nicer if it could be used on top of a cad kernel like opencascade like ledas did with their constraint solver.

  • Anonymous

    I think graphicalc is interesting but it would be nicer if it could be used on top of a cad kernel like opencascade like ledas did with their constraint solver.

  • I agree completely. I also keep telling Shyamal that GraphiCalc working side a CAD system as a plug-in would be fantastic as well. But like he says, CAD users are not his target audience, since only 5% of them will understand what his software is doing.

  • I agree completely. I also keep telling Shyamal that GraphiCalc working side a CAD system as a plug-in would be fantastic as well. But like he says, CAD users are not his target audience, since only 5% of them will understand what his software is doing.

  • There are technical and workflow reasons why we have not integerated inside a CAD system.

    Technical reasons:

    Most CAD system constraint managers were developed to capture “form” whereas digital prototyping for predictive-engineering requires to capture the “functional” intents of the design where the geometry and mathematical aspects of the challenge are inseparable and have to be contained in the same intent model. We had to develop from the ground-up and refine the constraint manager in Graficalc over several years to do function modeling that requires resolution of circularities in constraints, resolution to original intent when plural solutions are possible, and ultra-fast response time. Simply put, the engines inside most CAD applications are unsuitable for predictive-engineering and we are not resourced to fit square peg in round hole.

    Workflow reasons

    Desicions made during the per-modeling stages have the most impact on the final cost and performance of any product. Once modeling has started the user has already committed to a design path. If errors are detected by post-modeling analysis, the cost to come back and unravel the design is substantially higher.

    GrafiCalc is positioned to assist the users in pre-modeling stages of design when they have maximum questions and minimum information while conflicts are the easiest to detect and the least expensive to fix. With GrafiCalc our users have been able to considerably shorten the expensive “build-test-fix” product development cycle.

    Optimal solution:

    I beleive a CAD system should offer the user a choice of sketchers – one for form modeling like they have now for creating the solid model and an alternate sketcher for predictive-engineering, like GrafiCalc to make sure the design will perform as expected the very first time.

    Hello CAD companies – talk to us!

  • There are technical and workflow reasons why we have not integerated inside a CAD system.

    Technical reasons:

    Most CAD system constraint managers were developed to capture “form” whereas digital prototyping for predictive-engineering requires to capture the “functional” intents of the design where the geometry and mathematical aspects of the challenge are inseparable and have to be contained in the same intent model. We had to develop from the ground-up and refine the constraint manager in Graficalc over several years to do function modeling that requires resolution of circularities in constraints, resolution to original intent when plural solutions are possible, and ultra-fast response time. Simply put, the engines inside most CAD applications are unsuitable for predictive-engineering and we are not resourced to fit square peg in round hole.

    Workflow reasons

    Desicions made during the per-modeling stages have the most impact on the final cost and performance of any product. Once modeling has started the user has already committed to a design path. If errors are detected by post-modeling analysis, the cost to come back and unravel the design is substantially higher.

    GrafiCalc is positioned to assist the users in pre-modeling stages of design when they have maximum questions and minimum information while conflicts are the easiest to detect and the least expensive to fix. With GrafiCalc our users have been able to considerably shorten the expensive “build-test-fix” product development cycle.

    Optimal solution:

    I beleive a CAD system should offer the user a choice of sketchers – one for form modeling like they have now for creating the solid model and an alternate sketcher for predictive-engineering, like GrafiCalc to make sure the design will perform as expected the very first time.

    Hello CAD companies – talk to us!

  • “I guess that explains why SpaceClaim has two versions of their software: SpaceClaim Style for Modelers and SpaceClaim Engineer for Engineers.”

    Deelip, that is one gratuitous comment if I ever saw one.

    SpaceClaim markets the Style version to industrial designers, not “modelers”.

    The differences between Style and Engineering are that the latter offers optional import and export modules, and sheetmetal features. At least according to the comparative chart on Novedge:

    http://www.novedge.com/products/2933

    I am not an engineer. But I read job listings, and here in Eastern Canada, there seem to be plenty of “lowly” modeling jobs that require an engineering, or industrial design degree…

  • Norm C.

    “I guess that explains why SpaceClaim has two versions of their software: SpaceClaim Style for Modelers and SpaceClaim Engineer for Engineers.”

    Deelip, that is one gratuitous comment if I ever saw one.

    SpaceClaim markets the Style version to industrial designers, not “modelers”.

    The differences between Style and Engineering are that the latter offers optional import and export modules, and sheetmetal features. At least according to the comparative chart on Novedge:

    http://www.novedge.com/products/2933

    I am not an engineer. But I read job listings, and here in Eastern Canada, there seem to be plenty of “lowly” modeling jobs that require an engineering, or industrial design degree…

  • BTW, to a limited extent, Solid Edge has included since V20 some functions similar to GraphiCalc in the “Goal Seek” tool. There was a demo that showed it could be used in an assembly context to determine the length of a drive belt.

  • Norm C.

    BTW, to a limited extent, Solid Edge has included since V20 some functions similar to GraphiCalc in the “Goal Seek” tool. There was a demo that showed it could be used in an assembly context to determine the length of a drive belt.

  • Kevin Quigley

    I’ve followed Graphicalc for many many years and for many many years it has remained unchanged, difficult to demo (as in there isn’t one), and (in my opinion) overpriced (1395 Euros). I read the comments about the sketcher but I’m not convinced as I’ve heard the same arguments from other CAD companies who created their own sketcher for the self same reasons mentioned here…and most now use the D-Cubed one (Think3, VX etc).

    Where I do agree is in using 2D as a precursor to 3D modelling, and goal seek in 2D…but the problem is that this only works for some design scenarios. These days, with more and more complex shape design, weight optimisation and other factors a tool like this will not always help (and yes I am a qualified mechanical engineer).

    But I agree that if this was available as an add on built into my 3D CAD system (from where I could directly continue to build the model from whatever 2D geometry we ended up with), and it was a more reasonable price, I would buy it.

    The problem for Graphicalc though is that you can actually do a lot of what you can do in Graphicalc in the mainstream CAD systems – it is just that most users do not know it is there! For example, VX has had advanced 3D goal seek options for years, if you know what you are doing you can drive the D-Cubed sketcher with equations in pretty much every CAD system, then you get into the more recent releases of FEA systems where goal seek is very much a part of the picture – for full on 3D content.

  • Kevin Quigley

    I’ve followed Graphicalc for many many years and for many many years it has remained unchanged, difficult to demo (as in there isn’t one), and (in my opinion) overpriced (1395 Euros). I read the comments about the sketcher but I’m not convinced as I’ve heard the same arguments from other CAD companies who created their own sketcher for the self same reasons mentioned here…and most now use the D-Cubed one (Think3, VX etc).

    Where I do agree is in using 2D as a precursor to 3D modelling, and goal seek in 2D…but the problem is that this only works for some design scenarios. These days, with more and more complex shape design, weight optimisation and other factors a tool like this will not always help (and yes I am a qualified mechanical engineer).

    But I agree that if this was available as an add on built into my 3D CAD system (from where I could directly continue to build the model from whatever 2D geometry we ended up with), and it was a more reasonable price, I would buy it.

    The problem for Graphicalc though is that you can actually do a lot of what you can do in Graphicalc in the mainstream CAD systems – it is just that most users do not know it is there! For example, VX has had advanced 3D goal seek options for years, if you know what you are doing you can drive the D-Cubed sketcher with equations in pretty much every CAD system, then you get into the more recent releases of FEA systems where goal seek is very much a part of the picture – for full on 3D content.

  • Kevin, the price of GrafiCalc in Europe is $1395 not 1395 Euros. The price in other markets is $995

    There have been thirty (30) significant capability enhancements made to GrafiCalc software in the past one year – many were based on customer inputs.

    In mechanical design geometry and mathematical issues are inseparable. D-cubed sketcher does not hold ANY mathematical attributes that represent the functional intents of the geometry.

  • Kevin, the price of GrafiCalc in Europe is $1395 not 1395 Euros. The price in other markets is $995

    There have been thirty (30) significant capability enhancements made to GrafiCalc software in the past one year – many were based on customer inputs.

    In mechanical design geometry and mathematical issues are inseparable. D-cubed sketcher does not hold ANY mathematical attributes that represent the functional intents of the geometry.

  • Kesava prasad

    I have tried out Space Claim and feel that it is truly an innovation. But still need some validation of the capability to handle large models/assemblies.

    I really felt the need for a fast modeling tool when I recently had to offer a quick mock up of a fairly complex machine. It took me approximately 25 hrs of work in Solidworks. But I think Spaceclaim or other ‘Direct modeling’ softwares would have made my life easier.

  • Kesava prasad

    Hi Dileep,

    Could you give me an idea of openCAD? Is it a platform SYCODE has developed?

  • Kesava prasad

    Hi Dileep,

    Could you give me an idea of openCAD? Is it a platform SYCODE has developed?

  • OpenCAD is an application built over the DWGdirect platform from the ODA. For more details and to take this offline, click the “Contact Deelip” link towards top right and send me an email.

  • OpenCAD is an application built over the DWGdirect platform from the ODA. For more details and to take this offline, click the “Contact Deelip” link towards top right and send me an email.

  • Kevin Quigley

    Apologies on the pricing. Then again maybe not. Why do we pay significantly more in Europe?

    I have looked into the app before but I have never been able to actually try it out in my own time. Why is there not a demo version available, time limited or save limited? In this day and age very few are going to shell out £1000 for soemthing that might help them.

    From what I can gather, speaking to a colleague who actually does have it, the only way to get the data into a CAD system is via dwg or dxf?

    The benefit of a CAD integrated sketcher is that we can link the model to the analysis, so that if our parameters change we can go back, edit the profiles that drive the model, and update the analysis and models. Going back to another app and re-exporting a dxf file then have to mess around trying to relink profile to model is not a solution that makes life easy.

    If the app was a plug in, we could sketch using the Graphicalc sketcher, then reference that sketch to the CAD system (D-Cubed) sketcher.

  • Kevin Quigley

    Apologies on the pricing. Then again maybe not. Why do we pay significantly more in Europe?

    I have looked into the app before but I have never been able to actually try it out in my own time. Why is there not a demo version available, time limited or save limited? In this day and age very few are going to shell out £1000 for soemthing that might help them.

    From what I can gather, speaking to a colleague who actually does have it, the only way to get the data into a CAD system is via dwg or dxf?

    The benefit of a CAD integrated sketcher is that we can link the model to the analysis, so that if our parameters change we can go back, edit the profiles that drive the model, and update the analysis and models. Going back to another app and re-exporting a dxf file then have to mess around trying to relink profile to model is not a solution that makes life easy.

    If the app was a plug in, we could sketch using the Graphicalc sketcher, then reference that sketch to the CAD system (D-Cubed) sketcher.

  • In response to Shyamal Roy’s comment that, “D-cubed sketcher does not hold ANY mathematical attributes that represent the functional intents of the geometry.”

    No, not directly. However, as Kevin alludes to above, the D-Cubed constraint solvers (2D DCM, 3D DCM) both have the facility to solve equations. These equations can be quite general, and in particular allow the position, orientation or size of geometries to be variables.

    Equations, which could be defined by the end-user or by application code, will be solved by the DCM at the same time as the rest of the model. Geometry will be changed so that it satisfies the dimensions, constraints and any equations.

    So, for example, an application could create an “area constraint” by defining an area equation that included the geometries of the profile as variables. This level of flexibility allows complex engineering relationships to be solved simultaneously with the rich set of geometric constraints directly supported by the solvers.

  • In response to Shyamal Roy’s comment that, “D-cubed sketcher does not hold ANY mathematical attributes that represent the functional intents of the geometry.”

    No, not directly. However, as Kevin alludes to above, the D-Cubed constraint solvers (2D DCM, 3D DCM) both have the facility to solve equations. These equations can be quite general, and in particular allow the position, orientation or size of geometries to be variables.

    Equations, which could be defined by the end-user or by application code, will be solved by the DCM at the same time as the rest of the model. Geometry will be changed so that it satisfies the dimensions, constraints and any equations.

    So, for example, an application could create an “area constraint” by defining an area equation that included the geometries of the profile as variables. This level of flexibility allows complex engineering relationships to be solved simultaneously with the rich set of geometric constraints directly supported by the solvers.

  • Ralph Bennett

    As a consulting engineer I find Graficalc an invaluable contributor to my practice. I provide solid modeling and FEA services to clients using Autodesk Inventor. My primary focus is machine design and product development.

    I just finished the development of a mechanical/hydraulic system. This system consists of many complex parts such as pulleys, cables, cylinders etc. The parts move in both linear and non-linear paths. My standard procedure is to model the movements in Excel and then migrate that information into Inventor, in this case, for animation. I find it useful to model part movements in Excel which allows me to calculate forces, velocities etc. Because of that, I found it quicker to model all the moving parts, linear as well as non-linear. An additional limitation with Inventor for this project was that it does not model actions using cables.

    The project, however, required math beyond the scope of Excel. Several geometric dimensional requirements could only be derived by setting up and solving multiple simultaneous trigonometric equations. I could have used Inventor for some of the calculations but the parts that moved in a non-linear motion were beyond the scope of Inventor’s animation capabilities.

    To simulate the system, I develop 820 incremental driving force movements, moving 0.0811267” per increment. I then programmed Graficalc to match this and was able to record and copy precise geometric data directly into Excel. I did some additional calculations in Excel then copied and pasted the final data into Inventor’s animation timeline. Because of the non-linear movement of some of the parts, I had to copy/paste data those parts for each of the 820 increments in Inventor’s timeline. All total, I was able to load 17 columns of data from Graficalc into Excel then into Inventor, or just under 14,000 incremental positions of the many parts. It was a great time saver.

    I think anyone involved in tracking complex machine movement or forces or a myriad of other engineering tasks would find a great advantage to employ the power of Graficalc. I might add that this project alone paid for the initial cost of Graficalc and the many upgrades I’ve paid for over the years.

  • As a consulting engineer I find Graficalc an invaluable contributor to my practice. I provide solid modeling and FEA services to clients using Autodesk Inventor. My primary focus is machine design and product development.

    I just finished the development of a mechanical/hydraulic system. This system consists of many complex parts such as pulleys, cables, cylinders etc. The parts move in both linear and non-linear paths. My standard procedure is to model the movements in Excel and then migrate that information into Inventor, in this case, for animation. I find it useful to model part movements in Excel which allows me to calculate forces, velocities etc. Because of that, I found it quicker to model all the moving parts, linear as well as non-linear. An additional limitation with Inventor for this project was that it does not model actions using cables.

    The project, however, required math beyond the scope of Excel. Several geometric dimensional requirements could only be derived by setting up and solving multiple simultaneous trigonometric equations. I could have used Inventor for some of the calculations but the parts that moved in a non-linear motion were beyond the scope of Inventor’s animation capabilities.

    To simulate the system, I develop 820 incremental driving force movements, moving 0.0811267” per increment. I then programmed Graficalc to match this and was able to record and copy precise geometric data directly into Excel. I did some additional calculations in Excel then copied and pasted the final data into Inventor’s animation timeline. Because of the non-linear movement of some of the parts, I had to copy/paste data those parts for each of the 820 increments in Inventor’s timeline. All total, I was able to load 17 columns of data from Graficalc into Excel then into Inventor, or just under 14,000 incremental positions of the many parts. It was a great time saver.

    I think anyone involved in tracking complex machine movement or forces or a myriad of other engineering tasks would find a great advantage to employ the power of Graficalc. I might add that this project alone paid for the initial cost of Graficalc and the many upgrades I’ve paid for over the years.

  • Mike Poccia

    In my experience, design engineers use their PCs for much more than running CAD, and, indeed, spend more of their time in the Windows Desktop environment than in their CAD environment. For this reason, I believe it is considerably more efficient for GrafiCalc to run stand-alone, outside of the CAD environment. I don’t run my word processor, spreadsheet, MathCad, Working Model, email client or most specialized analysis software (with the exception of some FEA, CFD, and tolerance analysis software – all of which like complete or nearly complete geometry as input) inside CAD, nor would I want to. For one whose PC is not continuously running the CAD environment, how much of a pain would it be to have to run and enter that environment in order to to grab what would otherwise have been a quick answer from GrafiCalc? The extra convenience of not having to leave the CAD environment (along with the necessity to use dxf import/export) on the few occasions when I might want a quick answer while I happen to be in the CAD environment, is greatly outweighed by the inconvenience of having to enter the CAD environment in order to get that answer.

  • Mike Poccia

    In my experience, design engineers use their PCs for much more than running CAD, and, indeed, spend more of their time in the Windows Desktop environment than in their CAD environment. For this reason, I believe it is considerably more efficient for GrafiCalc to run stand-alone, outside of the CAD environment. I don’t run my word processor, spreadsheet, MathCad, Working Model, email client or most specialized analysis software (with the exception of some FEA, CFD, and tolerance analysis software – all of which like complete or nearly complete geometry as input) inside CAD, nor would I want to. For one whose PC is not continuously running the CAD environment, how much of a pain would it be to have to run and enter that environment in order to to grab what would otherwise have been a quick answer from GrafiCalc? The extra convenience of not having to leave the CAD environment (along with the necessity to use dxf import/export) on the few occasions when I might want a quick answer while I happen to be in the CAD environment, is greatly outweighed by the inconvenience of having to enter the CAD environment in order to get that answer.

  • Kesava prasad

    I have tried out Space Claim and feel that it is truly an innovation. But still need some validation of the capability to handle large models/assemblies.

    I really felt the need for a fast modeling tool when I recently had to offer a quick mock up of a fairly complex machine. It took me approximately 25 hrs of work in Solidworks. But I think Spaceclaim or other 'Direct modeling' softwares would have made my life easier.