Mathcad vs. Excel Prizefight

[DanMarotta]

If we put Mathcad and Excel in a boxing ring, toe-to-toe, who would come out victorious? What can you do in one, and not in the other? What can you do better in one, and not the other? Why do you like one over the other?

Photo credit: Generationbass.com

Ready? Let's get it on!! [ding ding]

[RichardJ]

The maths behind most engineering problems is quite simple. We don’t spend much time solving integrals or finding solutions to second order differential equations.

If you think that then you must have a very narrow view of "engineering". Engineering includes everything from designing a simple spring to designing a jet aircraft (or even a space shuttle). Or an optical system (I do some such engineering). Or an electronics system (the thread I pointed you to above with the large vector of points is an electronic engineering problem). Or chemical engineering.

Most engineering calculations require us to follow standard design methodologies which are published as codes of practise and national standards.

That is true only in certain fields of engineering, such as civil and structural engineering. There are no such codes (in general) in mechanical engineering, electronic engineering, optical engineering, etc etc.

Like I said before, Mathcad is used for a very wide variety of science and engineering problems. If you are lucky enough to need nothing more than a few simple formulas that are given to you in codes of practice and national standards then perhaps the calculation capabilities of Excel can indeed be considered adequate. For the rest of us, they are not.

[JohnBDoyle]

Richard I am a mechanical engineer and I use published codes of practice, national standards and text books all the time. It’s an incredibly rare event that I need to solve an integral or second order partial differential equation. Solutions are for the taking in reference material and it is pretty straight forward encoding these into spreadsheets. I know little about the work of electrical engineers or optical ones but I dare say they will be using their own reference material and simulation software not deriving formulae from first principles.

I once gave a talk to a university Physics department and I observed that Physicists are looking for fundamental truths and the big ideas whereas engineers are just looking to make something and we’ll take any shortcuts we can to get there. I have a very well-thumbed copy of Roark’s solutions for Stress and Strain for example I never have to derive all the complex formula myself.

[RichardJ]

Richard I am a mechanical engineer and I use published codes of practice, national standards and text books all the time. It’s an incredibly rare event that I need to solve an integral or second order partial differential equation. Solutions are for the taking in reference material and it is pretty straight forward encoding these into spreadsheets.

Mechanical engineering could involve the design of absolutely anything mechanical. Sometimes you can get what you need out of tables of reference material, but sometimes that is not possible. I find it astonishing that you can get through a career in any field of engineering without running into an equation that can only be defined in terms of an integral (that has no analytical solution). What about centers of gravity? Or moments of inertia? The same could be said for differential equations. What about a basic damped oscillator? Take a bucket of water and drill a hole in the bottom. The flow rate is given by a differential equation. I guess if you can solve all your problems by just pulling a few formulas out of tables, and never need an integral, then you don't deal with very complex problems.

I once gave a talk to a university Physics department and I observed that Physicists are looking for fundamental truths

That's a very stereotypical, limited, view of what physicists do. Everything I do involves solving real, practical problems.

[JohnBDoyle]

Richard, the problem you refer to is Plate Buckling using Rayleigh Ritz method – First of all I would generally turn to a large deflection non-linear analysis using finite element software because your analytical solution will only work for simple geometries. Secondly I would benchmark this against reference books many of which are already encode in Excel. The very last thing I would do start doing the maths from first principles.

I'm not paid to reinvent the wheel.

Centers of gravity and moments of inertia… it’s all done already - and even complex shapes readily solvable using a basic CAD package.

Flow rate from a bucket of water…its all done already.

Basic damped oscillation …it’s all done already.

http://youtu.be/RNoUdmbAYVo

There is little point telling me what Excel can’t do because I’ll keep on showing you it can. I deal with complex engineering issues. I do I just go about solving them in a different way to you. I am happy for you to go your way with Mathcad, I’ll go mine with Excel. The knub of my argument is that an Excel solution can be used more widely than a Mathcad solution.

[RichardJ]

Centers of gravity and moments of inertia… it’s all done already

No its not.

and even complex shapes readily solvable using a basic CAD package.

Exactly. Because it's not "all done already", what you are saying is that when it's not done you will have to use some different software, such as a CAD package, to do it, because doing it in Excel would be too painful to comtemplate. Then how do you get the number back to Excel? Copy and Paste? And then what if you chnage the design?

Flow rate from a bucket of water…its all done already.

Basic damped oscillation …it’s all done already.

I thought it was obvious that my point was not whether a couple of trivial problems have already been done, but rather that DEs (and equations based on integrals and derivatives) are extremely common. And not every such problem "has been done".

There is little point telling me what Excel can’t do because I’ll keep on showing you it can.

It's not only a question of whether or not it could be done, but how painful it would be, and/or how much of a hack job it would be. Excel can't handle equations written in terms of integrals or derivatives (which doesn't mean there is no workaround, but a workaround means it couldn't be done the right way), has no DE solvers, no user defined functions (unless you program them in VBA), etc. It can't even do something as basic as polynomial regression (except by using the solver, which s a lousy way of doing it), it has no spline interpolation functions, no Chebyshev polynomials, etc. It doesn't handle units, can't do symbolic math, can't handle multidimensional arrays (except, once again, by resorting to VBA), can't handle large data sets. Computationally, it is light years away from the capabilities of Mathcad.

The knub of my argument is that an Excel solution can be used more widely than a Mathcad solution.

That does not make it a better tool.

[wayne]

John,

Thanks for your participation, to continue:

Traditionally is true, design engineers typically apply set formulas and go to a FEM package for anything complicated. Mater of fact I am sure that only a small percentage of structural engineers that have been out of school for some time have more than vague memory of integrals or matrix math.

But the reason is that hisorically, the tools to use thoses skills have not been available, "use it or loose it". Even though Matcad has been around for awhile, I hvae only seen it used by Enginerings in my profession for short time.

It does not have to be this way though. For example, for wind loads, Engineers normally calculate the wind at intervals over the height of a building and apply as a step function, just as they did by hand, even before Excel.

I haven't done it that way for a long time, I set up a function (for example, use an interpolation function) and then just use numerical integration from floor below to floor above, A can easily include variable building widths (like for a sign) and calculate horizontal eccentricities, total shear and overturning moment at any point, all as functions of height (the width is also a function of height). I can easily plot to check. Also, I can easily use approximation calculations on the right hand page for checking.

I numerically integrate all the time, I define functions (programs) in Mathcad that are easy to follow, and do integration and matrix math inside these functions. These are normally in a file that I reference into the job file, then I can use any of them at will.

For beam design, I have functions available for applied moments, uniform loads, point loads, partial length trapezoidal loads, in any combination or number, The functions calculate deflection using virtual work and numerical integration based on the moment function. I then have templates for different problem types which use these functions, that I insert into the calculations as needed. If I have a wood beam that I need to change to steel, I just delete the wood section and insert the concrete section, or keep both options, the analysis part stays as is. If I need to add a beam, I just insert insert into the calculations where I want, I only need to assign a beam number high enough that the following designs are not effected by a shift in beam numbers. (Reaction at end 1 of beam bm=2 that is applied as a load on beam bm=6, simply applied as a point load R1.1 on beam bm=6) Other than that I don't have to worry about screwing up cell references.

I know that Excel can deal with inserting and moving things around, but if it can be screwed up, I am the one to do it.

The XLC is great for Excel. I have seen this before, I don't know if it was this one or if there are others. But If I had no investment in either Excel with XLC or Mathcad, and I was looking at each for the first time, with the purpose of performing engineering designs (at least in Structural Engineering), I can't imagine why I would choose Excel over Mathcad.

If I were in research or developing large applications, I might be looking at Matlab, but not Excel, even with VB.

I do agree that XLC is of great value to users of Excel, but it does not make Excel excell at what Mathcad does.

[JohnBDoyle]

Wayne, I think our view of what maths for engineers involves is pretty much in agreement. The choice of mathematical tool to use, Mathcad or Excel, I believe is fairly arbitrary. For every neat trick you can do with Mathcad I believe I’ll have an equivalent in Excel. Or maybe more to the point is that we become proficient using the tools we are given. To be fair I would have to say Excel with the XLC add-in matches Mathcad – I do think if you are an engineer using Excel then the XLC add-in is essential.

So in terms of functionality they are equally matched then the question becomes how useful is a Mathcad document compared to an Excel document? Excel has an advantage purely in terms of the number of people who can use it. I choose it because more people can use it. So if a template calculation is written in Excel I can issue this as a procedure for all my colleagues to use and I am reassured that we are literally all working from the same sheet. I don’t think I could manage this with Mathcad particularly when I have to send calculations outside the company to partners or clients. Here is a slide from one of my Powerpoint presentations it pretty makes my case for Excel. The prize-fight is not won on functionality alone, to my mind, it comes down to how many people can access the work.

[RichardJ]

For every neat trick you can do with Mathcad I believe I’ll have an equivalent in Excel.

We have established that is quite clearly not the case.

Do this in Excel:

http://communities.ptc.com/message/165160#165160

[JohnBDoyle]

The problem you refer to is Plate Buckling using Rayleigh Ritz method – First of all I would generally turn to a large deflection non-linear analysis using finite element software because your analytical solution will only work for simple geometries. Secondly I would benchmark this against reference books many of which are already encode in Excel. The very last thing I would do start doing the maths from first principles.

[Fred_Kohlhepp]

The very last thing I would do start doing the maths from first principles.

I believe that is exactly Richard's point; in Mathcad you CAN start from first principles. Using EXCEL that is very difficult and complex.

[JohnBDoyle]

It is not difficult or complex using Excel. I have inserted many links where Excel solves engineering problems which I hoped would serve to demonstrate this. Anything problem Mathcad can help you solve Excel can solve too. It’s a choice. I choose Excel so more people can reuse the calculation.

I know that I have walked into the lion’s den here but I think I have made my argument and for fear of bludgeoning my head further. I think it is about time I bowed out gracefully. Thank you all for your hospitality; it has been an interesting discussion. I am just glad we live in a time when we can all at least celebrate losing the slide rules!

[RichardJ]

Anything problem Mathcad can help you solve Excel can solve too.

Sorry, but really, that is not the case. Out of curiosity, have you used Mathcad at all?

Thank you all for your hospitality; it has been an interesting discussion.

Yes, I agree

I am just glad we live in a time when we can all at least celebrate losing the slide rules!

I still have one. Just in case Mathcad can't solve something for me

[wayne]

John,

If you leave, Richard and Fred might turn their attention to me, please come back.

[wayne]

HI John

I appreciate your patience. I know that Excel is ubiquitous, but things are changing. I see more and more use of Mathad in Universities. As this continues, you will see Excel used less and less. You are correct in that professionals tend to stay with what they know and are invested in, but this investment is slowly changing.

I have attached three worksheets that I did for some course work some time ago. Could you get the same value in Excel with the limited time? I can look at them now and clearly follow what I did (I am showing these as examples of use, if there is a mistake, please remember that ignorance is bliss). The one for Seismic Isolation could easily be adapted for design, the others are pretty much educational.

I would bet that most structural engineering firms have Mathcad; though not used effectively in most firms now, that will change with the younger Engineers. I am equally sure they don't have enough licenses, yet)

With regard to tables, thoses are disappearing also, How many younger Engineers have a copy of CRC Standared Mataematical Tables? (uh...I don't have one).

Wind loads, for example, have gotten very complicated with tables and figures, but I have them all as functions, I don't use the figures and tables any more (any more than once).

Wayne

[JohnBDoyle]

I use whatever I have to hand to solve problems CAD, FE and Excel (yay!) – they serve me well and I have yet to find an engineering problem I can’t lick using them.

I have used Mathcad a long time ago – I hated it (no surprise).

I would like to keep refuting your arguments but you won’t find them satisfactory and I am still trying to bow out gracefully. In fact I feel like I am being chased out of town like some kind of heretic. I can feel the heat of the burning stake all around me but I am screaming “I STILL LOVE EXCEL!”

Goodbye and thank you for at least allowing me to voice my opinion.

[RichardJ]

I have used Mathcad a long time ago – I hated it (no surprise).

Out of curiosity, why? I can see that if you had a large amount of legacy work in Excel, or needed to transfer the worksheet to someone that had only Excel, not Mathcad, you would have good reason not use Mathcad. But I don't see what there is to hate about it.

[JohnBDoyle]

Richard, maybe hate is too strong a work I found it cumbersome. I am old enough to come from a time when we used to write calculation sheets. When I made pen and paper calculations I could use a variable name many times (say, y was the distance between centroid and neutral fibre and its repeated use in a set of calculations was not ambiguous because it was used in context). When I calculated a bending stress I write simply M*y/I and it would be obvious to the reader what I was doing. At the time I used Mathcad (which admittedly is about 20 years ago) and I had a number of bending stresses to calculate I would have to define y1, y2, y3… etc. and I would also need to separately define M1, M2, M3… etc. and I1, I2 and I3. As a consequence my formulas became tagged with infuriating suffixes and what I used to fit on a calculation sheet seemed to run off into the margins. I did not like the way you had to loop through and index matrices – I might as well have been writing in Fortran. I never liked the graphics much either. One time I had a tedious job of scrutinising supplier’s calculations and occasionally they used Mathcad. These submissions were extremely voluminous, unwieldy and characterless. At the time I had no vested interest in Excel so had I liked Mathcad I could have continued using it but I stayed away (and I have probably missed out on 20 years of product development too).

I like the way I can throw down a few numbers in Excel and click around a bit to build some formulas up. It’s a bit like the ‘back of a fag packet’ calculations older engineers liked to kick around. Often I would be sent Excel calculations and I always thought it let engineers down by not being able to present equations easily (hence XLC). XLC gives me the flexibility of being able to reuse variable names like my old calculation sheets did. Now I use XLC to formalise the ‘back of the fag packet’ Excel sheets into text book like calculations. It also works as a great check on the excel formula because if I didn’t get the equation I was copying from the book – I knew I had goofed up somewhere.

The best thing about running the www.ExcelCalcs.com site are the contributions from members. I learn a lot from these people both in terms of theory but also in the ingenious ways in which they use Excel. We run a contribute or pay system on the site so contributors get it all for free but most people choose to pay.

Anyway I am supposed to be bowing out. Perhaps you and Fred could vent your spleen on Wayne Reid he seems to have stepped off the ‘Mathcad Rules’ line a couple of times. Take it easy on him guys. Perhaps I’ll venture back if anyone else starts picking on Bill’s big green monster…. Next time I’ll bring some friends!

Take care

John

[RichardJ]

I'm not sure why you came to the conclusion that you can't reuse variable names in Mathcad. I have been using it since version 3.1, and that has always been possible.

Your description of throwing down a few numbers and using Excel like the back of a fag packet is interesting, because it's exactly what a lot of Mathcad users like about Mathcad. It's a whiteboard, just like a piece of paper, and I can drop numbers, functions, graphs, etc on it, and they are just as readable as if I wrote them on a piece of paper (or the back of a fag packet). The difference to the paper (or fag packet) of course, is that they actually work.

[PhilipOakley]

I think that Hammers are better at inserting screws than Screwdrivers are at inserting nails.

However the whole thesis of the thread asks the wrong question: It presupposes a fixed arena with only two competitors, rather than the reality of an open market with many tool choices [Just look how big HomeDepot / B&Q warehouses are!]

There are many competitors who are doing 'roaring' trade, whether that is C, C++ or Matlab, or the Mathematica, Haskell or F# arena; each picking up folk who find their needs and the tool paradigm are broadly aligned (and the tool is available to them 😉

So the question should be why the fight?

Excel can do loads of stuff rather effectively, such as vast tables, but it falls down when it steps outside its conceptual boundaries - lots of folk are schooled on using tabular methods, it works for them.

Mathcad likewise can do loads of stuff rather effectively. It's schooling is from engineering maths, and it shows, and it's very very good at it. You get real equations. You get integrated units and dimensions checking. You get proper vectors and matrices.

Neither do 3d matrices though, but C, C++, Matlab, Mathematica, Haskell and F# do, and we live in a 3d world, so clearly [ 😉 ] we should neither use Excel nor Mathcad. Shouldn't we [rhet].

The reality is that it is a Scissors, Paper, Stone argument. The Excel extensions are useful for some but in no way meet the needs of real Mathcad users. Mathcad is still unique in the high ground it holds, but does need to tend to its hinterland if it isn't to be undercut, when it could be expanding and cooperating with its key neighbours.