Electronic Calculations (eCalc's)

Over on linkedIn group:  MathCad & Excel Structural Design files share Group there is discussion about MathCAD versus MS Excel.. Some of the MathCAD users seem to have similar knowledge of spreadsheets as the advertisers who work for MathCAD: namely zip.

Advertising they push the idea that in a spreadsheet a formula always look like:

=D2*D3^2/8

rather than

=w*L^2/8

Ok! may prefer it to look something like the following, but for such a simple expression I'm not going to pay the price of MathCAD for such limited additional benefit.

But consider something more complex say a formula from the cold formed steel structures code AS4600, say equation EQ3.3.3.2(5). So first off, the code standardises the equation and provides a reference number, therefore if want to see the equation using infix notation and text book style presentation, persons who are interested can go look at the code. Secondly we are here on the Internet, so how can we present the equation, in that text book like mathematical notation like manner. One option is MathML, though according to the following article it is not supported by all browsers.

Google subtracts MathML from Chrome, and anger multiplies

Therefore first presentation of the formula is by image, the formula above was copy pasted using the MS Equation object in MS Excel and copy pasted into MS paint. So the more complex formula presented as an image is:

This image was obtained from a screen capture from Amaya, where I created the equation using its equation editor for MathML as can be seen in the following:

Given the hassle of trying to write the equation using the equation editor and get each variable into the right box, I rapidly concluded that if MathCAD is anything like the equation editor in Amaya or MS equation editor then I will stick with the single line Excel or vba like notation:

foz = ((G * J) / (A* r_01 ^ 2)) * (1 + ((PI() ^ 2 * E * Iw) / (G * J * Lez ^2)))

Any case if MathML works in your browser (firefox, Amaya should work) then the following should render similar to the image. If MathML not working then ignore, the equation is not valid.

$f_{\mathrm{oz}}=\frac{G.J}{A.r_{01}^2}\left(1\; +\frac{{\pi }^{2}E.I_w}{{G.J.L}_{\mathrm{ez}}^2}\right)$

So that is the equation, and arguably the vertical style presentation is more readable than the single line presentation. When reading text books and industry manuals its definitely more familiar and recognisable, that doesn't however make it any easier to understand.

One thing I skipped in my little exercise was to actually time each application used. Any case since I am more familiar with Excel than any of the other applications, timing wouldn't really be fair any case. So the following are use of SMath Studio , its similar to MathCAD, I also used FreeMat which is similar to MatLAB , there are other free software variations of MatLAB: these are Octave and SciLAB. here is an interesting article on MatLAB alternatives.

Speed wise in terms of getting the number I wanted, the order was MS Excel, FreeMat, then SMath.

Here is the screen capture of the MS Excel version, which does highlight one problem, the equation presented above the result doesn't relate to the actual formula in the cell. That is largely because I copied the formula from my vba code to set up the calculation, not from the result cell.

Correcting for this here is a better view, this time with the cursor highlighting the edit bar for the calculated work cell, so that the calculated formula can be seen.

Has can be seen the formula is not based on cell references but on named ranges. Though it could be presented in the following manner. This time the formula is based on cell references, but an equation object has been presented above the result. However it is only an image of the formula, and if change the formula then it doesn't update. A better option would be too make use of XLC.

The following is the screen capture for SMath Studio, unlike Amaya, the equation editor worked by my typing in the single line format of the expression. As you type it reformats the expression. Copying and pasting an expression from elsewhere doesn't seem to work well, and editing had some problems. All up it gave me the wrong answer in the first instance. At first I thought there may have been conflict between Young's modulus E and the base of natural logarithms e, so I replace E with Y. But that didn't help, I also made some of the variables single character (I couldn't figure out how to subscript variable names, I could only find options for matrices, and I figure that would produce a different type of variable), and I removed underscores from the names. None of it helped. So I deleted the expression and started again from scratch, and that worked. I then started editing and changing the variable names back. So first lesson, just because the expression is nicely formatted, doesn't mean that it bears any relationship to the behind the scenes translation into a calculated result. The answer I got in the first instance was 9 point something: way out.

Since I got an error, I decided to track down what part of the calculation was potentially at fault. So I extended the Excel spreadsheet and the SMath calculation page, rewriting the expression as Eq1+(1+Eq2) as follows:

Excel Calculation

SMath Calculation

From this it appears that second part of the expression was not a valid calculation in my first attempt to type the equation in. It seems very sensitive to the operators used: just because it displays a given symbol doesn't mean that is the symbol to be typed in. For example need to type * for multiply not a simple dot (.) . The few times I have tried SMath it hasn't given me much confidence in the validity of the calculated result.

Whether we use Excel or MathCAD or even Mathematica , it important to understand that a great deal of translation is going on behind the scenes. To start with, we may prefer the use of infix notation for our mathematical expressions, computers on the other hand prefer post fix (or reverse Polish notation RPN), so arithmetic expressions we type in have to be parsed and translated into a stack of numbers and operators to be fed to the computers arithmetic processing unit. How the expression is displayed and how it is represented behind the scenes for calculation purposes are two different things. Similarly how the expression is input can be yet another data structure and processing task.

Our regulations require independent technical check. I would contend that independence is not achieved if designer and certifier are using the same software. Therefore both should not use MathCAD. Further just because have expression displayed, and results supposedly calculated from that expression doesn't mean that it is all valid. I contend that an independent technical check does not take place if the certifier has the designers calculations, and is reading those calculations. The certifier should have a blank sheet, for a valid independent check.

So having identified that one part of my expression was messed up by attempting to copy/paste equation segments. I decided to have a try with FreeMAT to see how cumbersome that would be, the following is the screen capture from.

Calculation in FreeMat

Other than it took a months of Sunday's for FreeMat to open in the first instance. It otherwise did the calculations quickly and with minimum problem. The only issue is that straight out, its a poor presentation.

But from another viewpoint, I spend as much time in the vba editor behind Excel as I do using Excel worksheets. So the FreeMat type environment does have its attraction, the issue for me is I have a lot of vba code, and translating to yet another application language is wasteful of my time.

There are also a multitude of simpler computer based calculators with tracing of the calculations, or specialised versions of interpreted Basic programming environment for engineering. For example to name a few:

1. ATCalc, version 4.0 (This now taken over from my use of Calc98 using RPN via my computer keyboard or the mouse was/is a pain: it also doesn't trace (But it does have a lot of other useful features). Just typing expression into ATCalc is easier and get a trace if have results panel switched on.)
2. Microsoft Mathematics
3. GraphCalc
4. Math Mechanixs

Some of these are more educational than anything else, but some times the only requirement is to get some insight into a formula by either tabulating input/output or otherwise graphing various parameters, and the educational tools are probably good enough for that purpose.

ATCalc with results panel switched off, simple and unobtrusive

Now personally I believe cell formula are error prone and cumbersome, especially given that the codes of practice have a multitude of conditional tests to carry out to select the most appropriate formula. Additionally compared to a database management system (DBMS) like MS Access, using MS Excel is actually inefficient for generating tables. It is better to define the formula once and apply to consistently to every record rather than replicate the formula. An alternative to this is to use something like VCmaster which has a better approach for tables and avoids the bloat of using a DBMS for simple tables.

I don't use cell formula I use vba functions and central library TechLIB, as mentioned in earlier posts. Using such the spreadsheet would look something like:

Yes I know, no formula present and cannot check. But I reiterate, reading it, is not a valid way of checking the calculation. Need to check that the calculation is relevant to the technology being assessed and that the arithmetic results are correct. The certifier or reviewer should redo the calculations from the input parameters. The designer has to carry out multiple iterations to find a potential solution, the certifier only needs to check that the proposed solution actually is a valid solution: one iteration. Software allows rapid checking, no need to start writing calculations out with pencil and paper and start bashing numbers through a calculator. We need to build solid foundations, reliable heritage to build up on for the calculation process.

Now in using a function library, can use the following method. First insert function, by pressing fx on the edit bar, to get the following dialogue box.

Choose user defined functions and select the function. Which if they are you own functions are easier to find, than if someone else's. If I knew how to add the help to the functions, I'd might consider including it. Any case from here can choose the function, and get the following assistance.

Press the right vertical scroll bar to see additional parameters.

From here can then go and select the appropriate cells, and get to see the values selected as well as the final result of the function, before accepting and committing it to the worksheet cell.

Now as far as I know at the end of the day when want to get a real job done, then mathematical notation is thrown in the bin, and the constructs of high level programming languages like Fortran, Pascal, Basic are adopted. The difference comes down to the available function libraries available to the application, and what want to do with the resultant application.

For me, the important criteria is being able to write a stand alone application which can be handed over to persons who are not engineers. Further, the custom application shouldn't have a massive runtime application to work: Java and .net have runtime application's but these are more readily available than say the runtime for MatLAB.

Currently I have the hassle that I thought I could make use of my Excel/vba, by adopting vb.net as stand alone programming environment, but too many differences between the languages. So become an issue a of whether to go back to Delphi, or move over to Java or C#. I don't like the idea of C++, pointers in Turbo C always gave me a problem and the syntax for record type data structures seemed untidy and cumbersome.

Put simply I disagree with MathCAD's white paper that I read some years back, and the benefits of using MathCAD. The benefits have very little to do with the text book type presentation of mathematical expressions and such presentation is not all that clear in any case. Personally for practical purposes, I prefer integrals presented as summations, and further I prefer summations presented as "for loops" rather than using a sigma symbol. And that's another issue. It does not help very much to use the lower case sigma symbol for stress and standard deviation. Now some who write programs would tend to create a variable called sigma, rather than name it stress. Similarly they would call it sigma rather than standard deviation, and additionally call the summation function sigma. Thus got a real mess when statistically assessing the results of experiments involving mechanical stress: as everything is called sigma.

Dreaming up new variable name, or creating arrays and matrices is largely avoided in MS Excel and other spreadsheets, as working in an array of cells in the first place. So simply point to and use a cell value on an as needs basis. MS Excel fits in better with what the individual was going to do with their calculator, especially their RPN calculator, as the cells represent the stack or registers of such calculator, but have the advantage of labelling each register. The calculations can be saved as a template, and can be modified on an as needs basis to be made more and more presentable. However even with Excel, still need to understand that their are different presentation requirements for different purposes and different audiences. Those who push MathCAD only seem to recognise one audience and one purpose for calculations.

And that narrow view represents a major obstacle in moving forward with calculations, as we should be enabling and empowering others to get the calculations done, with out being concerned about the details of how got from input to desired result. Most people just need results with which to make decisions.

This IStructE forum thread for example includes a lot of annoying people. It does not take 4 years to learn how to calculate the base moment to a simple cantilever post: its high school physics stuff. Further more now that we have limit states design and use member resistances rather than check stresses, once the design moment is calculated, can often look up member resistance in manufacturers tables. My question would be why the drafter asked the question if they already had calculations from an engineer? It would seem either his employer never receives the printed calculations: or maybe for some reason they are not easily read. Chances are they just get a certificate. My view is that they should have someone on staff to do the calculations, and it doesn't require an engineer to do the calculations: Australia's 2 year qualified engineering associates are more than capable. But then again engineering associates are not drafters, nor are they WFEO technicians.

Any case the attitude shown was largely obstructive not helpful. But it does highlight need to consider audience and purpose of calculations. In-house the calculations need to present the theory, and detail, as informative training materials for each new generation of designers taken on by the business. But for the regulatory system such repetition of calculations represents a waste of paper or hard disk space if printed to pdf, and a waste of  time to read and to check. If have appropriate software, then both designer and certifier can get their answers and reach their respective decisions quickly.

---

Checking using  firefox, the MathML in the blog post displays correctly, though possibly too small.

Software for Barrier Design Calculations

A simple program for calculating the base reactions for a post forming part of a barrier designed to be compliant with Australian loading code AS1170. It only calculates reactions based on imposed barrier loads to AS1170.1, it doesn't consider wind loads to AS1170.2. The wind loading won't be critical if the wind pressure is less than the imposed barrier pressures.

It can be downloaded here as a zip file.
To install just unzip into any suitable folder, it doesn't touch the windows registry, and can run from a memory stick. The main program file to run is dsgnBarrierPost.exe.

Once the base reaction for the post is known, then a post with equal or greater resistance needs to be selected. Unfortunately the methods of connecting posts to support structure typically have less resistance than the posts, and thus become the critical factor for selecting suitable post spacing and post heights. Additionally the support structure to which the barrier/balustrade is to be connected may be unsuitable, note the barrier is to be loaded along its entire length or otherwise pattern loaded to check the suitability of the support structure.

For example the loads on multiple posts attached to a fascia beam are likely to tear the flimsy cleats and the fascia beam from their support posts. Also if attached to a fascia beam then the barrier post bending moments are transformed into torsional moments in the fascia beam. Torsion is a shear stress issue, not a normal stress issue. Rough rule of thumb is that shear strength is 50% of the tensile yield strength. Therefore generally want to avoid torsion in a structure, and to do that requires careful attention to the perimeter framing of a balcony/decking floor.

The software is made available to assist designers to check the suitability of support structure, by providing the magnitude of the actions the support structure needs to resist. An equivalent horizontal action can be obtained by dividing the base moment by the post height.

Suppliers of barrier/balustrade systems, should preferably have technical specifications stating the resistance of their posts, rails, and connection systems.

The software doesn't produce fancy reports or save data, its a simple design tool. A sample of more detailed calculations can be found here.

Barrier/Balustrade design is also not all that clear cut. AS1170.1 specifies barrier loads to be applied to the top edge of the barrier: this does not make sense for barriers of any height or any structural form. Further commentary on balustrade/barrier design can be found here:

1. Balustrades-pt1
2. Balustrades-pt2
3. Balustrades-pt3
4. Barriers
5. Barrier Design Calculations (Previous mention of the application)

EXPERIMENTAL
The software is primarily experimental. It is a simple program with few features, I can dream up a multitude of additional features for the program. The big question: is it worth developing these features, and should such features all be embodied in the one application?

The display window/form for the application is intentionally small: roughly the size of a smart phone. How features are added for use on a smart phone is entirely different than would be required for a large display monitor. Also the program is written in vb.net, so it would likely need translating into java to port it over to an android compatible smart phone.

Any case the real experimental nature of the application is getting away from all the calculation printouts and all the certification nonsense. The exercise is to develop a box of tools and features which aid the design process, and otherwise aid independent and automated testing of the calculation process. Printing out a calculation report would be a useful aid for checking the calculations: however printing it out and distributing such report shouldn't become a ritual. If I don't provide such report then it cannot become a ritual, and therefore alternative means of validating the calculations buried in the application need to be developed.

DISCLAIMER :
Users of the software must accept this disclaimer of warranty :
The software is supplied as is. The author disclaims all warranties, expressed or implied, including without limitation, the warranties of merchantability and of fitness for any purpose. The author assumes no liability for damages, direct or consequential, which may result from the use of the software.

---

Revisions:
[28/11/2013] Original

[23/04/2016] Changed download links to MiScion Pty Ltd Web Store

On Automating Technical Specification and Assessment

On Engineering: Engineering is not Required in the Building Industry.
As indicated elsewhere I try to avoid the use of the terms engineer and engineering: as I consider engineering to take place at the frontiers of science and technology, and such is where I consider engineering to have been through out history. The only frontier that the building industry encounters is the wilderness where it constructs buildings. Prior to the imposition of any regulations the building industry would just get on with building stuff. Now with regulations in place they are being told to go get engineering, go seek an engineer, get an engineers report, get an engineers certificate. This is not a recent requirement however, modern type building codes have been around since the mid 1800's.

However early building codes were based on prescriptive requirements derived from past successes. If something worked in the past then expect it to work in the future, as long as have knowledge of the conditions under which it worked. If the scientific method and/or scientific knowledge was applied, then it was behind the scenes conducting tests and producing the prescriptive solutions placed in the codes. Fundamental scientific principles were not applied on each and every project: it is not sensible to do so.

However it is important to understand the limitations of the prescriptive solutions and the conditions under which they are valid: this is what I consider to be technical science, and decisions made on the basis of such science to be technical design. Using the timber framing code AS1684 to specify the structural requirements of a building, and assess valid specification of a building is technical design. Using AS1170 loading code and AS1720 timber structures code to specify structural requirements and assess suitability of a structure is also technical design: it is not engineering. Neither is using AS4100 or AS4600 to design steel structures to be considered engineering.

The activity is technical design of established technology through the use of technical science carried out by technical officers. The important distinction between technical design and engineering design, is that there is little uncertainty and little risk associated with technical design.

No one expects the lintel supporting the roof above a window to sag and crack the window. No one expects a crane to collapse whilst lifting its load. These are established technologies and the public has certain expectations of performance.

Engineering is not concerned with established technology nor for that matter with established scientific knowledge. If a project really involves engineering then there are no national standards, no industry manuals, no text books, and no academic programmes to attend where people can be taught: there is no established body of knowledge: not for the technology and not for the science. The project is going to involve a considerable amount of fundamental research. Simply testing prototypes and saying: yeah! it works, is not engineering. The prototypes tested for engineering bare little relationship to the final objective: the purpose of the tests is to derive fundamental relationships which can be extrapolated to larger much riskier objectives. When the final technology is built, it is still a real world experiment, and likely to fail or otherwise turn out defective. Because even with all the testing there is still the risk of some new unexpected phenomena becoming apparent when the prototypes are scaled to required size (either scaled up or scaled down).

For example cannot just decide we are going to place a human habit beyond Pluto, or for that matter on the bottom of the ocean floor. Whilst it is feasible to come up with an intial proposal based on the established technologies, such proposals are likely to be impractical. Having decided such is impractical we will then encounter a series of problems which are beyond our established technologies: but otherwise potentially within the scope of our established scientific theory. However to apply that scientific theory, experiments will need to be carried out to collect fundamental scientific data. Collecting the scientific data however may require new technologies. This interaction between science and technology, to push forward to reach a specific objective is engineering. Engineering originates, it establishes technologies, it does not concern itself with the routine implementation of the established. Fundamental to establishing a technology is developing the required technical science for the routine application and adaptation of the technology. This routine application and adaptation is the work of technicians who employ established tools and techniques. Here technician is not taken as some inferior subordinate to an engineer: an engineer has to be a highly competent technician otherwise they are just dreamers. The ingenuity comes from transforming imagination into reality.

Structural Design
The importance of this distinction is that there is little to no engineering associated with the design of bridges, buildings and most other structures. The technical science of structural design is well established. Therefore when someone is seeking development approval for a building they do not expect to encounter any difficulties or delays in obtaining such approval. Further more if they go to a supplier of the specific structural product they seek, rather than to an architect, then they expect the approval process to be even smoother, not more hassle. Unfortunately going to a supplier of structural products is likely to result in more hassle. This is because the suppliers of the structural products are largely ignoring the established technical science.

The suppliers are thus sent off by regulators to go get engineering. Since need to get engineering from an engineer the expectation is that the alternative is to employ an engineer on staff. Employing an engineer on staff is seen as too expensive, or just not compatible with the organisation. Since the so called engineers primary role in the regulatory process is merely supplying the numbers, and the engineers are increasingly using software to produce the numbers, it is to be expected that the persons paying for these numbers should seek software which they can operate themselves. They thus consider getting software to automate the engineering. Doing the numbers is not engineering. Engineering cannot be automated, it cannot be carried out by computer.

Automating doing the numbers, is automating the calculations, and otherwise automating the technical aspects of the structural design. Other aspects of structural design are creative and cannot be automated. Also the output of any creative contribution is also likely beyond the established calculation procedures and thus demand fabrication and testing of prototypes to validate and calibrate new calculation procedures.

Now the input parameters for one structure can be the same as those for another structure. For example the basic dimensional parameters of a shed are the height and span, whilst the basic parameters for a guard rail system are the height and span. What happens between the input and the output of suitable structural sections however is different. Comparing differences between a shed and a guard rail is probably obvious. Where it is not so obvious is comparing one structural connection with another which have significantly different behaviour but similar input parameters. Such may give the user of such software the impression that they can dream up new connections for, say a shed, and still use the same software to assess the suitability of the building. Such judgment would be hazardous: and such new connection is highly likely to require testing.

The technical science for structural design is well established, however the use of the technical science to develop mathematical models is not always practical. Either the mathematical model takes too long to develop and then test and validate. Or the mathematical model once developed takes far too long to calculate, even with the fastest computers. Either way a prototype is typically required to be fabricated and tested. Mathematical modeling does not remove the need for physical testing it merely reduces the number of tests and increases the likelyhood that the protyotype will pass the tests. A qualitative understanding of the technical science also assists in being prepared to deal with any hazards which may arise during testing.

So with respect to suppliers of structural products, there is no point-of-sale engineering, there is no point-of-sale design. No design and no engineering can occur at the point-of-sale via the use of computer software by salespeople having a discussion with the customer.

Point-of-Sale
All that can occur at the point-of-sale (PoS) is capturing the parametric variation of highly established technologies and adapting those technologies in an extremely limited way to better meet the needs of the customer. It is not design and it is not engineering. Design is an unrestricted creative activity, engineering operates at the frontiers of science and technology. Technical design places constraints on the freedom of creativity.

After having varied the parameters of a generic technology, the task is assessing the suitability of the technical specification for a specific purpose. Or is the task determining a technical specification which is suitable in the first place?

The question may be: I want to make this shed, what size c-section is suitable for the frame? The answer to that question, requires guessing some size of c-section, and then seeing if it is or is not suitable for purpose. If not suitable for purpose we then try another c-section and check its suitability. So behind the scenes the exercise is assessing the suitability of a proposed technical specification.

Some structural forms are known as determinate and others are known as indeterminate. To simplify the meaning, rather than use the strictly correct meaning, those structures for which we can directly calculate the required size of structural section are considered to be determinate. Those structures which we have to determine suitable structural sections for by trail and error are considered indeterminate. Indeterminate structures require knowledge of the materials and structural sections to determine the forces in the structure, whilst no knowledge of the materials and sections is required to determine internal forces for determinate structures. {For some simple strictly determinate structures, the materials properties have cancelled out to leave a simple expression which can be used to directly size the section. (eg. fixed ended beam). On the other hand the materials codes are so convoluted that it is typically not possible to calculate required section properties directly, and all structures need suitable sections selected by iterative trial and error.}

Commercially Available Structural Analysis Software
Most structural analysis software available in the market place is setup for indeterminate structures, so even if wish to analyse a determinate structure it is still necessary to specify structural sections and material properties before any analysis can occur to determine internal forces for the structure. {Actually I'm not aware of any readily available software purely for determinate structures.}

Further commercially available structural analysis software typically has menu options for checking or designing members. Once again, I contend the design option doesn't design anything. The design option merely iterates through a list of available sections and determines which are and are not suitable for purpose. The software may stop when its found the first suitable section, or it may determine all the sections suitable from the available sections. The design option is therefore just a variation of the check member option.

Why have the check member and design member options? The basic answer is because its generally a bad idea to have the computer select the structural section. If the computer selects the section, then it will either be unsafe, uneconomical or impractical. Letting the computer select the section is just a starting point and reduces the number of manual iterations.

How can it be unsafe? It can be unsafe because the materials codes are incomplete. So for example if assessing a steel structure to AS4100, a structural section may be considered acceptable, even though it is subject to a high level of torsion. This is because there is no direct torsion check in AS4100, but the torsional resistance still need to be checked. The traditional task of the designer is to qualitatively configure the structure so that members are not in torsion: and therefore torsion is not an issue for calculation. Assuming a structural designer has followed tradition and removed torsion, a code check by computer to AS4100 may still accept a defective structure. This is because the 3D stick model of the structure does not represent the real world structure.

Roof purlins for example in the real world typically sit above the rafters, but in the 3D stick model the purlins are in the same plane as the rafters. Further the purlins are rigid and continuous over the rafters, not rigidly connected into the sides of the rafters. The default behaviour of the software is typically to create rigid moment resisting connections where ever two members intersect. This does not  necessarily reflect the real world model where the real connections may be pinned. Assuming members are rigidily connected, that is cannot rotate about a pin, typically reduces section sizes, but can otherwise increase stresses in the support structure. The result is that sections are not checked for proper stresses. The other issue is that there are eccentricities and localised affects not considered by the software, and need to be dealt with manually, or by other means.

When it comes to member checks, the software is also likely to assume full restraint of a section where there is no such restraint, and therefore the structural section selected by the design option is likely undersized.

But this leads to the more common problem of uneconomical and impractical. The software will typically find the minimum size structural section suitable for each member in a 3D model. Since the applied forces are not uniform across the whole structure the resultant internal forces and stresses are also not uniform. The result of allowing the software to select member sizes is common members all having different sizes. For example rafters varying in size depending on their location in the roof. This is not typically practical. Further if lateral torsional restraints are present in the real structure, but not known to the computer software then, the software will over size the steel sections. These restraints typically have to be manually specified before executing the design options of the software.

Thus from the perspective of using the available structural analysis software, automating the technical specification and technical assessment of structures doesn't seem all that practical, as it is a highly interactive and iterative process requiring learned judgment. Learning, which a salesperson and customer typically do not have.

But what exactly is it they want to automate?
The difficulty is caused by the suppliers having no proper technical specification of their structural product in the first place. Sure they have some drawings of a specific instance of the generic product, they may also have standard calculations for that instance. Some even have spreadsheets with specific instances of the generic product buried and hidden inside the software. None of this however represents a proper and valid technical specification of the generic technology they wish to adapt to the needs of the customer. The suppliers most certainly have limited fabrication capabilities, and they also have restrictions on what they are willing to do. Yet the salespeople typically say yes to just about anything the customer requests, on an assumption that such proposal is feasible.

The first and foremost problem doesn't have anything to do with calculations. The first and foremost problem is accepting a project for which the suppliers technology is ill suited. Accepting a project which imposes demands beyond the capabilities of the supplier and their supply chain.

If going to supply anything the customer wishes then need technical officers on staff to design a suitable structure, and such design is not going to take place at the point of sale. People do not go to suppliers of structural products so that they can wait for design to occur, they go to such suppliers so that they can effectively buy off-the-shelf. However, they also want to change things, things which are outside the scope of the existing technical specifications, things which would make all the existing documentation invalid. When the existing documentation becomes invalid, then it is no longer possible to communicate intention to all the people in the supply chain. Further more if the documentation describing the proposal is invalid then it is no longer possible to assess the suitability of the structure.

For example say we have a standard design for a 3m high by 7.2m span portal frame shed fabricated from cold-formed c-section. This standard design will typically be accepted by city council for any design which fits with in the dimensional envelope. So if a customer wants 2.4m high by 6m span, then the standard design will be found acceptable. But none of the workshop drawings would be valid. Then again who as workshop drawings for a cold-formed steel shed: just need some hole punching sheets. In which case the punching sheet for the columns and rafters would be incorrect and new sheets would need writing up. Further all sheets would need revising to match the quantities needed for the project.

Under such circumstances no technical assessment is really required, all that needs to be automated is the revision of quantities in the bill of materials, and the dimensions of various members along with generation of hole punching sheets for the roll-former. We have a structurally valid design solution in the first place, and we haven't made it invalid.

Now consider the situation where the customer wants to rip out the end wall mullions and install a large door. The existing structural specification is no longer valid. An alternative structure is now required to support the end wall cladding along with the door.

Another situation is removal of portal frame columns along the side of the building. Removing such coulmns cripples the portal frames and the frame check for the columns and rafters is no longer valid. It is not just a simple matter of providing carry-beams to support the rafters over the opening. The primary structural form the rigid moment resisting frame, comprising two rafters and two columns, has been crippled. This structure resists the wind attempting to push the building over, and can no longer do so. The load pushing against the door and the building now has to be transfered to other complete portal frames, these frames thus now carry more load. Take out too many columns on either side of the building, and just leave rafters, then the whole structural form of the building may have to be changed to stop it from being pushed over. Such changes do not represent simple parametric changes to a generic technology, rather they represent a change of the generic technology itself.

Product Structure Tree and Variations
With respect to the bill of material (BoM) or the product structure tree, the difference is that between merely changing counts of quantities versus pruning and grafting branches from the tree. If merely change a part count then it seldom generates a need for further technical assessment, but if prune a branch or graft in a new branch than additional technical assessment is required. Grafting and pruning branches changes the generic form of the product: it changes the product-structure, and if change the structure then require further structural assessment.{NB: Two different but related concepts of structure}

Variation of a product can be brought about by:

1. Changing part count
2. Changing parts
3. Changing part dimensions
4. Removing Subsystems (Pruning branches)
5. Adding Subsystems (Grafting Branches)

If a part is a subsystem, then setting its part count to zero will prune that branch from the product-tree, if change from zero, then will graft that subsystem into the product-tree. Depending on the product any of these changes could impose a need to reassess the suitability of the product for a given purpose.

The shed industry typically considers a shed comprises of columns and rafters: it doesn't. The typical cold-formed steel shed comprises of portal frames, rigid plane frames, or moment frames. These frames are the primary structural element. These frames comprise of columns and rafters. However the frames cannot have parts pruned from them and still be expected to function as required. Nor can these assemblies have parts grafted into them and still be considered suitable.

A shed can be made longer by increasing the part count of the frames. But it cannot be made longer by increasing the spacing of a given frame. A given frame has a specified maximum spacing.

I contend that in the first instance the industries have no need of structural analysis software, and that so called engineering at the point-of-sale is just a sales gimmick: it has no real value. Why have a computer spend an hour assessing a series of similar frames, when already no the answer. An engineer wouldn't assess every frame in the structure, they would only assess the critical plane frame. They would only assess a 3D model if the structure was behaving in three dimensions, and then they would only build a model large enough to assess that portion of the whole building. Structural design is based on simplified models, not an attempt at creating a complete virtual reality. Further after an hour at the point-of-sale, why have the computer come back and say not possible, when its a simple exercise for the engineer to make it possible.

The larger the building and the more complex the structure the longer it will take the computer to complete the analysis and select suitable parts. An experienced engineer or technical officer can reduce the time it takes the to find a solution, because they already know the solution, they just have to get the computer to check, rather than have the computer iterate through all available sections and find a suitable section. The computer can complete 3D analysis fairly quickly for most common structures, it is the selecting of suitable section sizes and specifying and connections which is a time consuming exercise for both computers and humans. Humans can however fast track, and potentially get there faster than a computer. The problem is going to a consulting engineer is going to place the project in a long waiting line: and cause delays. So that the 1 hours it may take a computer to find an answer is better than the 4 week wait to get answer from a consultant.

The issue here is the calculation versus database approach to the specification. Typically the database approach is frowned at, and preference is given to the calculation approach. However humans can fast track to a solution because of their memories: a database of past projects and known solutions.

Also the calculation approach is actually based on a hard-coded structural model, which generates a bill of materials. The bill of materials doesn't drive the system, and therefore changes to the bill of materials results in an appropriate structural analysis and assessment of the structure. The structural model is locked, therefore the bill of materials is locked. No parts can be added, and no parts can be subtracted. The software typically used is simplistic and built around MS Excel spreadsheets: and typically is an extremely inefficient use of computing resources. {This is because data and calculations are combined in the one file. Add the need for MS Excel, and the systems are using a lot of hard disk space and RAM for very little capability.}

Few of the suppliers of structural building products are using software built around a building information model (BIM): for the most part they don't need such. The main task they need to automate is the material take-off and associated costing. Except there isn't anything to take the materials off. There are no workshop drawings. At most, likely to be a simple sketch on an order form. The order forms are mostly descriptive of requirements.

The issue is that its not quite simple retail. If people just turned up placed an order, and were given a package with fixed content then there would be little problem. The problem is that customers can change the content of the package. That in itself wouldn't be too much of a problem if it was a simple part count of independent parts. Just fill the shopping cart with items, and determine total cost.

The problem is want to change the parts. Consider similar situation outside the building industry. The customer being able to change a car from 4 wheels to 3 wheels. Consider a customer wanting 24 hours of recording on video tape, but without increasing the space occupied by the typical 2 hour recorder. Such customer requirements at not things which can be designed at point of sale by a salesperson. Such desires may require years of research design and development. There are clear restrictions on what can be supplied, and the variations permitted.

The building industry doesn't seem to understand the restrictions on variations to the technology, nor is it particularly capable of meeting the demand for essential buildings: global shortages of housing, schools and hospitals to name a few. Architects oppose industrial building systems, yet we end up having to use such systems to meet demands, and yet the industrial systems are not as efficient at supply as they could be. Unwarranted variety needs to be constrained.

There is a high incidence of owner-builders in Australia, and otherwise custom designed houses, yet there isn't that much difference between the houses. Yet the differences could significantly increase the cost of supply without increasing the benefit obtained. It is false economy to save on materials. Often the customer is wanting changes to reduce the cost of materials, but increasing the cost of labour significantly more. Standard designs have a design heritage, a foundation to build upon. Move away from the standard and introduce a lot of additional work which wasn't otherwise needed. A lot of the work only needs doing once: like design and documentation. Standardising parts also means automation is possible for the fabrication and production, this in turn reduces production time and achieves higher levels of consistency across repetitive parts. There is typically greater variation in repetitive parts made by humans than those made by machine.

If go to a supplier of structural products, then expecting faster supply times, higher quality product and lower costs, compared to custom designed and fabricated. These expectations are not always met: and that is largely due to unnecessary variations or customisations to the standard product.

House builders with large markets, supplying relative standard designs have tended to opt for the use of BIM based software. They can afford the software and have supply chains which can benefit from its use. Other suppliers of building and/or structural products would not get any benefit from the software and its too expensive to go experimenting with.

Most suppliers just need parametric based bill of materials (BoM) software, no 3D graphical picture is required. They just need to modify part counts and dimensions subject to certain constraints. The BoM needs validity checks on each data field relative to the specific item. Software like MS Access and MS Excel can have validity checks on data fields, such as whether a date in input, or a telephone number is formatted correctly. A BoM however requires more than this, each data record needs unique validity checks on each of its fields or at least on the quantity field.

For example with respect to the previously described standard shed, the column cannot be greater than 3m, but can be less. Similarly the structural section used can be increased in size to a higher structural resistance, but it cannot be decreased to a lower structural rsistance. Once we know the required structural resistance for the shed in the form of specified structural section: then it is not necessary to re-analyse the structure, an alternative section of equal or greater resistance can be substituted. But such substitution is likely to lead to a need for new work shop details.

The perceived need to carry out structural calculations is because of wind loading and earthquake loading and other loads which change with the environment. However these design loads are not really site specific, they really belong to a class of loads. Wind loads to AS1170 for example are regional, sure the regional wind speeds can be reduced to site specific wind speeds: but is there any real economic benefit in doing that for repetitively manufactured products? More over just because the wind load does change for the site, it does not mean that it is necessary to redo the structural analysis and recheck all the members and connections.

If there is no smaller structural section available then the building gets made stronger than it otherwise needs to be. How strong it needs to be is a matter of personal opinion in any case: sure it may be in the code: but the code is someone's opinion. So its actually silly to consider any structure is stronger than it needs to be. The structures are simply strong enough to be acceptable to someone somewhere. As for not being strong enough: once again that is a matter of opinion. Clearly if the structure collapses under its own self-weight then its not strong enough. But how much stronger than standing up under its own self-weight does it need to be? Ok! So don't want it to blow down in the wind: but what wind? People are always saying South Australia doesn't have cyclones, and basically consider the wind loads are too high. But what lower wind load is acceptable to the community? More to the point what level of wind loading is acceptable to the insurance companies: they seem to think the loads should be increased.

The design loads are not so much a response to the natural environment but more a reflection of human desire, and are otherwise misleading. No structure is hurricane or earthquake resistant, there is always some magnitude of hurricane or earthquake which will exceed the design loads and collapse the structure. So making the buildings stronger after a disaster is not all that sensible. I contend the primary defence of the nation is not the resistance of its buildings but its ability to rebuild in the minimum of time. Buildings which behave like mechanisms and collapse to a more robust form during an event, and then be erected again after the event: seems far better use of materials, than trying to resist massive loads and ultimately not being string enough in anycase. It is nonsense to be declaring that buildings are unsafe if they are not built to the codes: the buildings are unsafe when built to the codes. The issue is that if not built to the code then it is unacceptable and is therefore illegal construction. Being illegal doesn't make it unsafe, it just makes it illegal: not compliant with the letter of the law. Not compliant with someones opinion. It is emotional manipulation to declare one brand of sheds is safe and another brand is unsafe. In the real world, they are all unsafe, no buts doubts or maybe's about it they are all unsafe, and they cannot be made safe. The risk of collapse will always be there, and the occupants will not have any awareness of when the environmental loading is likely to exceed the resistance of the building. Would you like to operate a submarine without a depth gauge? Because that is effectively how we operate our buildings.

So we could consider getting rid of the regulatory framework and the standardised minimum performance criteria. For having such doesn't serve the community well. The building industry seems to be the only industry in which suppliers take pride in complying with the code. Like finding someone who complies is hard work. Merely complying however equates to low quality rubbish. Most other industries based on engineered technology take any standards as the benchmark to be exceeded without significantly increasing cost to the buyer. If only meet the regulations then a low quality supplier.

The point is: that just because there is a smaller structural section available which can be used to meet the code requirements it doesn't make it sensible to select and build the structure from such smaller section. Further more just because used a larger structural section that exceeds the requirements of the regulations it doesn't make the structure stronger then needed. Nobody knows or can know how strong the structure needs to be.

It is nonsense that a supplier cannot compete because their structural sections are over sized and the competitions are considered under sized. The supplier cannot compete because they do not know how to sell value, and because their sales people have zero understanding of the product they are selling. The consequence is that the only thing they can do to make a sale is drop the price. Since they only really cost the materials, not the labour and overheads, it is apparent the only way they can drop the price is to drop the cost of materials by using less steel. In short sales are based on the principle of my rubbish costs less than their rubbish. Not really sensible.

So most shed suppliers start with a single standard design which covers a large portion of their initial market, then as people ask for larger buildings the supplier gets standard designs for larger sheds. Or they get custom engineering for what ever the customer wants. The problem is they seek the custom engineering at the wrong time. Further if they really need to be providing custom engineering then they really need engineers on staff to meet the time constraints, rather than rely on consultants. However, need to differentiate between designing a standard product and custom design to the project.

There are about 18 standard size c-sections, for the typical gable frame shed these all form about 16 envelopes, that is 16 standard designs for a given wind loading. There are only 16 envelopes because 2 of the sections have near equal capacity to other sections: so its a question of preference. Now 18 standard designs is not a large database, especially if it is a parametrically driven database. The variety decreases even further if consider that there is little difference between the dimensional envelope achieved by one c-section and another. So can toss a lot of the variety out, thus reducing strip widths and strip thicknesses required to produce the c-sections. Thus reducing the variety of c-sections, which reduces material inventory costs if rollform own c-sections. Those who order c-sections probably don't care: but they would if the roll-formers adjusted prices to match demand rather than steel content. That is the roll-former offers 18 sizes, but prices such that, bulk of supply is only for about 6 of the sizes.

A shed fabricator for example could rollform C10024 and supply sheds across most of the market. The sales people right now are saying nobody rollforms such section and its too small. Yes, its too small to se as a rigid moment frame. So is a C7510, but such section is fine to use as the studs in a house. Houses are made from cold-formed steel but they are not moment frames. The structures are pinned and braced, and otherwise triangulated. Roof trusses are triangulated, and the small C7510 can span large distances when fabricated into a truss. Similarly the C10024 can be fabricated into deep parallel chord trusses which can be used to span distances greater than achieved using a C35030 as a moment frame. Structural form is important.

One supplier was always phoning up and saying why can't we span such and such a distance? We'd get an address and drive by to find a structure which wasn't the typical moment frame the supplier produced. Our response was thats not the kind of structure you normally fabricate. We can design a structure using the c-sections, if we know what you are willing to fabricate.

Other structural forms have far more labour content than the moment frames, and so take longer to fabricate and tend to cost more. However bolted cold-formed c-sections requires less skill than welded hotrolled sections. So a cold-formed structure has the potential to cost less than any hotrolled frame.

So we get back to the first and foremost problem. Are these businesses suppliers of structural products or general steel fabricators? Since most don't have workshops they are not fabricators. They also don't have design offices and they are not really project managers either. They just have sales offices, but they are not sales agents for some big supplier either: and they don't wish to be. They are independent agents, sometimes fabricators, but mostly sales, builders or steel erectors. They are a kind of organiser/broker. Certainly not coordinators, but possibly facilitator's, arrangers, overseer, directors, handlers. Basically they are the people with the right contacts to get the job done eventually. The desire and task is to make that process smoother for everyone.

But automating the detailed structural calculations is not really part of the exercise, and serves no real benefit. Sure if such software is available, affordable and operating such software produces quick answers at the point-of-sale then that is one way to go.

However, if software is not readily available off the shelf, and it needs to be developed. Then the first thing to note is that in the first instance it is not going to be quick, and it is likely to have a cumbersome if not apalling user interface. Further more it is to be noted that engineers now largely use computer software themselves, and they wouldn't typically consider writing such software themselves.

It is one thing for an engineer or technical officer to consider using a spreadsheet to write a few calculations, but they wouldn't consider writing their own version of say AutoCAD nor their own version of say MicroStran. Can therefore consider that the typical consulting engineer producing calculations and drawings for suppliers of structural products does not have the resources to write software. They may write their own software so that do not need the likes of say LIMSTEEL. They may automate the inputs to AutoCAD and MicroStran.

To consider that the average civil engineer designing structural products can produce adequate software, especially to automate the design of an entire building is just plain ludicrous. They may be able to produce tools to assist themselves, but not produce anything anywhere near robust enough to be placed in the hands of unqualified sales people.

The need for structural calculations at the point-of-sale is perceived as necessary largely because no one has sat down and properly specified the structural product being offered. Clearly the sales people are not architects or engineers, and are thus not capable of providing the customer with any building they desire. There is significant constraint on what they are able to supply. There is also significant limitation on what they are able to do for the customer.

For example attaching a verandah to a house requires significant strengthening of the house, and attachment of a balustrade may also require significant strengthening of the support structure. Checking and strengthening the existing structures is not the business of these specialist suppliers. The existing structure needs to be adequate in the first place. Potentially strengthening the existing structure is a role to be carried out by another small business.

Further should the supplier have and operate the software, or should the customer operate the software?

In terms of balustrades for example as a buyer or designer I would really want the supplier, I wouldn't want to keep phoning the supplier up to find out what I need to do for my structure to be compatible with the balustrade installation. The architect and engineer designing a building need to know what is required to provide a structures compatible with a balustrade installation. Not wasting time between a supplier and a suppliers external consultant, nor wasting time with the supplier pushing numbers through software. Just need to know what has to be done at the start of the project. The problem is the architects ignore the issue until the last minute and expect the balustrade suppliers have a solution. The architects then end up with a structure that is not compatible with their desires for a balustrade. Its the architects fault because they thought they could dismiss a junk of structural design from the engineer to a salesperson. Whilst I contend doesn't require an engineer, and the task is not engineering, it does require someone with appropriate technical qualifications.

Similarly attaching a verandah, extending a shed, all have similar problems, or dealing with the existing structure: a structure which may not comply with current codes, and therefore is not adequate in its own right to start with, before even consider applying additional loads.

So the issue is what are these businesses really supplying? What are they really in the business of providing?

Because if throw computers, software and the Internet into the mix, with the real needs of the customers. Then have an entirely different business model, and these suppliers are gone. So their investment in half-baked software is not in their best interests. Buying readily available tools which assist at low price is one thing, investing in development of a large system is another.


{I got interrupted and lost my track way back there somewhere. But identified some smaller chunks for more detailed assessment.}

On Calculations and Software Part 1

None of the spreadsheets I have uploaded to ExcelCalcs make use of XLC . I like XLC, but after some 20 years of creating spreadsheets without the benefit of XLC, I am not about to start a massive exercise of converting my spreadsheets. More over I have a developed a different approach to producing calculations. When I started I did want to present calculations in similar manner to how I wrote out calculations with pencil and paper, and so every now and again I would take a look at MathCAD, and then change my mind about the whole thing. I also read a white paper by MathCAD which identified a problem of design solutions being scattered between handwritten calculations, Fortran source code, spreadsheets and the likes, and how this was not otherwise readable by all.

However for the authors of MathCAD, calculations are an end in themselves, not a means to an end. As I keep pointing out, real engineering takes place at the frontiers of science and technology. Calculations as an end in themselves are important for real engineering, because such calculations document the design-solution, how to get from concept to reality. There are no national standards, no industry handbooks, no established technical science. The engineers calculations will define the technical science and the foundations for writing future standards and industry handbooks. But this is not the environment in which the vast majority of modern professional engineers operate.

Most so called engineering, is not what I would describe as engineering, it is technical design based on firmly established technical science, making proper use of national standards and industry handbooks, for the purpose of applying and adapting established technologies. The community will not tolerate failures or defects in such established technologies. It is therefore important to get the calculations correct, but it is also important to get the calculations done quickly and efficiently.

The regulatory system typically requires documentation to be issued in at least triplicate. The more pages of calculations the more paper to print off and the more documents to store and otherwise manage, and mostly for no real benefit to anyone. When I look at the MathCAD templates, of TEDDS, Master Series PowerPad, EnerCalc to name a few, all I see is software that produces pretty calculations and potentially a large pile of scrap paper. Sure we can now print the calculations to pdf file and therefore save the paper and the forests, but the pdf files consume hard disk space, and someone has to read through and check the content. The question is why, does all that documentation need producing and checking?

Here in South Australia we now have a ministers specification for structural software used by persons who are not engineers. It is largely a result of failures of nail plated roof trusses. These trusses are typically "designed" by timber estimators working for truss fabricators using proprietary software. The South Australian regulations require independent technical checks for development approval and obtaining building permits. The software can design a truss in a few minutes, the reports traditionally produced by the software were relatively deficient. No certifying engineer had similar tools, therefore they could not independently check and relied on information in the reports to make some checks.

The first problem is that analysing a truss is time consuming, the second problem is inadequate information is provided with respect to connection details. In particular whilst the software may have sized the nail plates required at the connections there was little evidence, that any checks had been made that such nail plate would fit, and that there was adequate timber to fasten into. For shallow trusses this can be a problem, and really need to draw the connections out. The old design manuals had templates which could be over laid on drawings to check for fit.

There is an increasing number of structural products in the market, and suppliers wanting software to do the engineering, and therefore some controls have been put in place for this software.

However, my view concerns the certifiers. Nail plated roof trusses have been around now for more than 30 years, and the certifiers still haven't got rapid design and/or assessment tools. I contend that in this age of personal computers and electronic calculation pads, it is nonsense that only the big truss companies can afford to develop rapid design software. I also contend that it is unacceptable that there often no alternative software for proper independent checks. For example LIMSTEEL basically holds a monopoly for hotrolled steel design. Not only is it integrated into MicroStran and SpaceGass the two main frame analysis software applications in Australia, but in modified format it is also the basis of ASI design capacity tables (DCT). Whilst this produces consistency, its lacks independent assessment, no alternative checks and balances, and that is not a good thing.

In the past our office here, has carried out independent checks. Initially I looked through hand written calculations submitted and tried to follow them. Often difficult as sometimes the calculations are just numeric expressions, with no algebra and no descriptions as to what is going on. One set of calculations I checked kept moving from limit state to permissible stress using a factor of 1.5. The calculations also involved spans of 1.5m and load widths of 1.5m . With no explicit description of which each calculation was for, they were difficult to read.

Other calculations, I had different views on application of local wind pressure factors. At the end of the day however the design engineers calculations are irrelevant to the certifiers independent assessment. Why would a certifier waste 5 hours trying to decipher the design engineers calculations when they can use their own software to check the specifications and drawings in 5 minutes?

Which reversed the situation for me. Normally my view was, if I can complete the calculations in 5 minutes then I expect the certifier to be able to check in significantly less time. As designer I have to go through various alternative solutions, or seek a solution through many iterative trials and errors. The certifier only needs to check the validity of the final solution: and either accept or reject.

If I can produce software then the certifier can equally well produce similar software, and has far greater justification for doing so: they have more pressure to check in short time frames, and also greater recurrence of common design solutions.

If a supplier of structural products phones a consulting engineer up to get an estimate of member sizes, so that they can quote a realistic price for supply: that answer needs obtaining quickly. Preferably in 24 hours if not 1 hour. A week is likely the upper limit and not typically acceptable.

The calculations, to AS4600,  for checking a segment of a cold-formed steel section are fairly extensive. One engineer who produced calculations by hand and moved over to using MathCAD increased the thickness of his reports by about five times: it served no real purpose and was potentially less readable. If produce calculations by hand then there is greater tendency to be more efficient and reduce repetition. If use a computer then there is a greater tendency to opt for flexibility. The problem with flexibility is it increases the potential for error on each project. Some times it is just far more efficient and reliable to have a custom tool made for the job, rather than have multi-purpose tools. When it comes to computer software it is often possible to customise multi-purpose tools for a specific purpose, but that also tends to be an expensive approach.

From a mechanical view point, it wouldn't be sensible to customise a CNC flexible machining centre to make bolts all day every day. It is far more efficient and less expensive to design and build a specific machine for making bolts: such machine would also require resetting less often.

Similarly with software, custom design tools are more efficient than multi-purpose design tools. Microstran and Spacegass have hardly changed since they were ported from MS DOS to MS Windows. There are many suppliers of structural product who want custom frame analysis software, but go to small consultants who don't have any such software to use as a foundation for building custom software products. The software then becomes time consuming to produce and also expensive.

The other problem is the software becomes proprietary and the consultants also get their hands tied and prevented from supplying similar to other suppliers of structural products. Due to the problems with the nail plated roof trusses I don't believe such mode of operation is acceptable. A variety of software tools need to be available to manufacturers, builders, designers and certifiers. Further no certification should be based on using the same software as the designers used. It is not all that sensible to jump on the upgrade band wagon: the upgrade may introduce errors which were not previously there.

So calculations have different audiences and different purposes and appropriate tools need to be adopted for each combination of audience and purpose. Mathematical type face or text book presentation of calculations has one purpose: but it is not necessarily more readable. Personally I find programming source code more readable than some of the symbolic mathematical notations. Some times concise and succinct can become obscure and meaningless. Why use the Greek symbol sigma for stress and standard deviation: especially if conducting statistical calculations on stress? However in programming need to balance between clearly identifying meaning of a variable versus unreadable over long mathematical expressions.

However the two main problems with programming, concern input and output. The calculations transforming the  input into the output is usually relatively easy. Spreadsheet software has already solved the problem of data storage, file format, and editor to get data into the file, and the means of presenting results and sending to a printer. All the user has to do is create their own report in their own style.

One style would be to move from the top of the screen to the bottom, showing algebraic expressions, numbers substituted into those expressions, and the final calculated result: the use of XLC would help with such presentation.

The use of a spreadsheet however combines transforming calculations and vba source code with the data, and is not very efficient use of computing resources. It is easy for just about anyone to set up, but otherwise extremely wasteful.

Calculating sectional moment capacity (phi.Ms) to AS4600 involves a large number of intermediate calculations, and calculating member moment capacity (phi.Mb) involves even more. These calculations need repeating several times for different segments and load cases of a single member, and again for each member in a project. My original AS1538 (permissible stress version of AS4600) required a single printed page for each segment. If all the algebraic expressions had been included it would probably stretch to 5 pages, as it was I thought one page was too long. Noting that I also had to photocopy each page and bind into a report. so being able to check multiple segments, multiple load cases and multiple members on a single page: was something of an important objective.

There is also an issue of how spreadsheets are used. My original spreadsheets were centralised. Use the spreadsheet on a project and print the results out. If results were saved, they were over written by the next project. There was no copy of the spreadsheet for the project. Also had different spreadsheets for different codes and each of these linked into larger spreadsheet for specific structural forms. So if changed the modules, the results from the larger module also changed, and that also meant low potential of repeating the results obtained on previous projects. But hard disk space was in low supply, so it kept disk space consumption low.

The linking of the spreadsheets is also another issue. Initially just use different spreadsheets for different dependent purposes and manually type the results from one into another. Whilst that provides flexibility with many small tools adaptable to a variety of structural projects, its not very efficient and can result in transcription and round off errors. It is far more productive to have a spreadsheet which does all the structural calculations for a given structural form or whole structural project. But that can cause revision problems for common pages of calculations across each project type: any error has to be fixed in multiple locations rather than one central location.

Another problem with using spreadsheets is the use of common data such as material properties and section properties. A simple structure like a cold-formed steel shed, involves several structural members: columns, rafters, mullions, girts, purlins, struts, bracing. Bringing all the section property data into a spreadsheet can be cumbersome, as is referencing the correct collection of such data. Using a database is more efficient than using a spreadsheet, and MS Excel permits access to MS Access data using vba and DAO. Using DAO the spreadsheet only needs to know the section name such as C300-30, with that vba can get all the data and calculate phi.Ms and return the value directly to the worksheet: there is no need to show all the intermediate calculations.

Whilst it is important to be able to see all the intermediate calculations if something goes wrong, it is not necessary to display all such calculations for each and every project. It is cumbersome scrolling back and forth through long tracks of calculations, to change input parameters and see final results. Whilst it is possible to split screens, freeze panes or open new windows and tile: it is still using more screen space than really necessary to get the desired results.

Compared to many industries where the technology would not be possible without engineering design, the building industry has had engineering imposed upon it. Buildings were constructed for 100's if not 1000's of years with out the use of engineering calculations. The regulatory imposition causes delay in supply and often achieves no real  benefit.

Sometimes the more detail submitted to regulator the more rigorously they check the design, and the more questions they ask. Send less in they make fewer checks and ask fewer questions: too little and they ask lots of questions. Other occasion's however get the impression that the more extensive the documentation, the less likely any checks occur. The regulator just looks at the pile of paper says, they must have done a thorough check: and immediately approve. It is thus difficult to assess whether should provide more or less detail in the calculations.

Therefore need a certain amount of flexibility in automation of calculations with ability to create multiple presentations of the same calculation process. Spreadsheets with application programming language such as MS Excel and Openoffice/Libre Office provide such flexibility.

If I write visual basic for applications (vba) user defined functions (udf), I can easily develop MS Excel workbooks which use the worksheets for collecting, storing and presenting information, or write an application which just uses the workbook for storing data, and using dialogue boxes for collecting and displaying information. I can also move away from MS Excel altogether and write stand alone programs.

Further more I can build on the vba code and further hide information. For example I want to know what phi.Ms to make a decision, most other people don't care, they just want to know if a given structural section is suitable for their needs.

So engineers preferred presentation of calculations largely stems from a tradition of working calculations out with pencil and paper. If using a computer to carry out the calculations, then such presentation is no longer overly relevant to the majority of people who need to know the results and conclusions reached from such calculations.

To enable and empower people to get on with making and applying established technologies, we need to rethink the calculation and assessment process, and provide suitable automation tools. I consider that vba code has far greater flexibility in developing alternative design tools, than in cell worksheet calculations.

Compare the following examples of AS4600 calculations for phi.Ms against those in the the previously mentioned calculator application. The calculator has several screens to get to the calculation of phi.Ms, where as in all the following spreadsheet images, the intermediate calculations are hidden, whilst sectional property data is also hidden in an MS Access table. All of the worksheets make use of the same vba function to calculate phi.Ms.

Once schTechLIB is installed and available, making such alternative workbooks is a lot easier than replicating all the intermediate calculations and then meshing with all the additional calculations required for the design of a structural member.

Text book presentation of calculations is the way to understand calculations but it shouldn't dictate presentation for all purposes. Calculations are a means to an end, and a means of reaching decisions, and we shouldn't forget this.

Simple Calculator to get phi.Ms

Simple calculator to get phi.Ms and phi.Mb for each available c-section

Lower portion of calculations for design of end wall mullion