Software to Automate Engineering for Structural Products

Given an idealistic view that engineering takes place at the frontiers of science and technology, no engineering should be required for structural products as they are largely established technologies. However there may be need to make prototypes and physically test them to get some empirical data, and calibrate mathematical models. Additionally there are issues of interpretation of codes of practice, regulations and the need for judgement due to limitations of the available information which is published in the literature, and the need to compromise based on risk, expense and practicality of conducting further research.

Structural Design can be broken down into the following tasks:

1) Specify Dimension & Geometry
2) Determine Applied Design-Actions
3) Calculate Design-Action-Effects
4) Assess/Design Suitable Member
5) Assess/Design Suitable Connections
6) Assess/Design Suitable Footings

The Structural Model
A basic structural model needs to define, the dimension & geometry, the applied design-actions, and the initial trial of materials and structural sections. A structural product like a balustrade is relatively simple, a shed has a few variations of structural shape but a lot more components to design than a balustrade, whilst canopies (carports, awnings and verandahs) have a complex variety of shapes but fewer different component types than sheds.

Using computers, the shape of structural products with simple shapes can be auto-generated from a few defining parameters. For example a gable roof shed can be defined from the following few parameters:

1) Over All Width (out side face of girts)
2) Height to Eaves
3) Roof Pitch (Alpha)
4) Bay Spacing or Frame Spacing
5) Number of Bays

Whilst such information is adequate for creating simple stick diagrams, it is not adequate for determining the centre line geometry of the primary frame. To calculate such dimensions the following additional information is required:

1) Girt Size
2) Purlin Size
3) Rafter Size
4) Column Size

Which poses us with a chicken and egg problem, as the purpose of the structural design exercise is to determine the size of these structural components, but we need to know the size in the first place. It therefore becomes a trial and error exercise, we start with a section size, do the structural calculations, check if its suitable, if not suitable, then change the section and repeat the process. With a fixed over all width, then larger structural sections for the column and girt, will result in smaller centre line span of building. Whilst smaller structural sections result in larger span of the building and thus typically higher stresses. It is therefore preferable to start with smaller sections and step upwards rather than start with large sections and step down towards the most suitable section.

A monoslope roof can be generated from a gable roof by adjusting one of the eaves intersection nodes so that it is higher than the ridge node. Whilst a canopy with similar roof shape to a shed can be generated by removing the wall cladding support structure. For structures with more complicated plan and elevation shapes, it is preferable to manually build the structural model from common component types (columns,rafters,struts), or auto-generate common shapes and then manually merge them together to put appropriate structure at the interface. This is where software which supports a parametric constraint based building information model (BIM), with nested assemblies, becomes useful. With such software can define an outline sketch of the dimension and geometry, and then relate by constraint points and formulae. Whilst such software enables flexibility in dimension and geometry, it does not generally permit change of shape or structural form. For example a triangle cannot be changed into a rectangle, and a rectangular floor plan cannot be changed into L-Shaped, T-Shaped or U-Shaped floor plans, nor can the more complex shapes be reduced to the simpler by setting dimensions to zero. Similarly as the span of a building gets too great for the use of a single structural section, the simple rafter cannot be replaced by a triangulated roof truss. Such alternatives thus need to be offered as starting templates or auto-generate wizards. Templates would be complete models which allow editing of the dimension and geometry. Auto-generate wizards would take a few parameters and then automatically create the model.

However, whilst dimension and geometry can be calculated from a few independent parameters, and primarily comprises of calculating node points for insert of component parts, or otherwise end points for the lines of stick diagrams, the assignment of loads is far more complicated. Not all parts of a building are structural components, and not all structural components have directly applied loads. Further more whilst structural design would primarily be concerned with following the load path from the roof to the ground,repetitively going from action, to reaction, to action on support, the process is complicated by codes of practice which modify loads on certain component parts. For example purlins are designed for locally magnified wind loads, these magnified loads are not required to be used to design the rafter and primary support frame. Therefore need to consider different loading scenarios, each having multiple load cases.

When auto-generating the structural form, its is relatively simple for the author of the program code to decide that a member is of a given type and needs to have certain loads applied. Whilst such hardcoding may be simple and convenient, it is not very flexible or adaptable, further more it is not suitable for automatically applying loads to an existing dimensional model. To cater for such situation the structural elements need to be identified, and various structural properties/characteristics assigned to the element so that the appropriate loads can be assigned to the element. That also poses two alternative situations. If structural properties are not added to a building element when inserted in the building model, then such properties have to be manually added to each element later, so that loads can be auto-generated. Alternatively rather than add properties to the building elements so that loads can be automatically applied, the loads can be manually added instead. The former is probably preferable with the ability to manually edit the loads as need be to clean them up.

Custom Software or Off-the-Shelf Software For Structural Products
Manufacturers of structural products, and their distributors have a tendency to want design software which operates at the point-of-sale (PoS). Part of the reason for wanting such software is to allow flexibility with respect to potential customers needs and rapid pricing of the proposed structure. However design at the PoS by sales people who are neither architects, engineers or others with suitable knowledge in such fields, is cumbersome and problematic process. Actually point of sale, may be optimistic, it is more of a sales enquiry, and needs to generate a price in less than a one hour consultation. That consultation may take place face-to-face or over the phone. The more complicated the floor plan, the less likely it is practical for the consultation to take place and be completed over the phone. The greater the complexity, the greater the need for the buyer to have architectural plans and elevations prior to contacting the manufacturer.

So first some background information. Consider in the past timber suppliers produced material take-off’s to the Australian timber framing code (AS1684) so that they could determine the price of the timber supplied. New players entered the market, and these players didn’t provide free material take-off’s, they used those produced by the more established players and then undercut the prices. The timber suppliers ceased to provide free material take-offs, people are now directed to timber estimators. However small builders of timber carports and verandahs, still practice such things, making use of sketches and material take-off’s produced by other builders giving quotes. If the proposed building work is more complicated than a simple rectangle, then the owner really needs to get drawings produced by an independent drafter or designer.

Historically, architects and engineers became independent consultants, as the builders and other makers did not necessarily have adequate skill to build something which was suitable for purpose unless the design was supplied by others more capable in the area of design. That is whilst they had high capability to build, they had inadequate knowledge of what was required to make the building suitable for purpose. These consultants however emerged with focus on defining the end-product, and with little concern for the practicality and safety of fabrication and construction. So today there exists a conflict between designers and makers: things which are designed which cannot be built, and things built which are not fit for purpose.

To replace these designers and makers by sales people who have even less knowledge, does not bring about any improvement in the situation. A manufactured structural product (MSP) does however allow more effort to be put into the relationship between the product form and its manufacturing and construction processes: so that components and assemblies can be made faster and more strictly conforming to specification. The product can be continuously improved and made more suitable for purpose in the long term, compared with one-off construction where the owner gets all the defects inherent in the design and construction process. These defects have typically been filtered out off the processes for the manufactured product.

So the manufacturer of a structural product is supposedly providing a superior product as a consequence of continuous improvement of both the end-product specification and the production processes. It is the proactive design of the product, and its repetitive production which allows effort to be put into improvement. Compared to one-off custom design, where all the effort goes into just bringing the idea into existence. Contrast this with a typical small building that could take 2 weeks for architectural, engineering, and workshop details, against the smallest industrial widget that can take over a year to get from drawing board to first production. One of the major differences is the engineering effort put into the production process of the industrial widget, whilst the small building is built on the basis of, figure it out as you go.

It is also important to note that design and engineering effort is not proportional to the size of the structure but to the complexity. That is structures having the same structural form require the same engineering effort no matter the size. Thus a small garden shed requires the same engineering effort as a large industrial shed, if they have the same structural form. Consulting architects and engineers typically try to get fees based on percentage of the capital value of the building: consequently large buildings can generate high fees exceeding actual value of engineering effort, whilst small buildings barely cover the cost of living for the time spent on the engineering effort.

It is therefore important than small structures and also complex structures are engineered once, and made many times so as to get the greatest value from the engineering effort.

A fixed form product only needs engineering once, but a variable form product may impose additional engineering requirement if the permitted variations are not constrained to avoid the need for engineering. At the point of sale or at sale enquiry can consider three situations with regards to manufactured structural products:

1) Fixed form Structural Product
2) Variable form Structural Product (constrained to avoid additional engineering)
3) Variable form Structural Product (extra engineering required)

A fixed form structural product is the easiest to handle during a sales enquiry and during a sale, as the price is predetermined and typically non-negotiable or if negotiable only with in pre-determined constraints. Both customers and sales people just need price lists and appropriate product information. The suppliers mostly just need materials requirements planning (MRP), manufacturing resource planning (MRP II), enterprise resource planning (ERP) or some other form of management information system (MIS) or computer aided production management (CAPM) software. The type of software would depend on whether just a retailer or wholesaler, or a manufacturer making to order or making to stock. All design and engineering is complete, and can be done by a backroom design office, or designs can be bought in and produced under license.

If have a variable form structural product constrained to avoid additional engineering, once again, all the design and engineering can be completed by a backroom design office, or designs can be bought in under license. At sales need product configurator type software, to allow the optional features to be selected and priced and added to the over all price of the proposed building. All other software is similar to that required for the fixed form structural product.

A variable form structural product allowing custom features which imposes additional engineering is the most complex product to configure at the point of sale or during a sales enquiry. Roughly the sales enquiry needs to be less than 1 hour duration, preferably around 5 minutes. The whole purpose of going direct to the manufacturers is to avoid the delays caused by consulting architects and engineers brought in at the start of the project. The manufacturer has supposed to have brought the architects, industrial product designers and engineers in, long before the start of the project. The manufacturer is meant to provide for rapid design and fabrication, followed by rapid transportation and construction, so that the building can be operational in the shortest possible time.

So when the custom product design process involves more than a simple product configurator to select component parts assembled in predefined assemblies, then no longer simply supplying a physical object , also supplying extensive design service. Which then raises the issue should design be independent of the manufacturer? The sales enquiry requires far more design and engineering input, and then have the risk that such effort being used by other suppliers not providing the design service and undercutting prices. Sales people have to either sell the design and fabrication package before any design takes place, or direct people to a list of independent designers. For the full benefit however, independent designers need to work closely with the manufacturer: otherwise more or less back to the design and tender process. The process is slightly better than design and tender as rather than requesting suppliers to tender, a buyer can simply shop around to find a suitable supplier. However, such design will be generic and exclude proprietary specific features, requiring customisation to the chosen manufacturers specific system. But the manufacturer will still need some system to get an accurate cost of the proposed building, each manufacturer starting from scratch building a computer model to get a cost would be inefficient. Therefore it would be preferable that the design model is available in a common data format preferably using xml file format, and industry data formats for the structure of the data (STEP,CIMSteel,IFC).

Software Tools Paralleled with Factory Automation Hardware
CNC Flexible machining centres are highly versatile machines, but it would not be appropriate to replace all traditional machines tools with such machinery. The flexibility of the machining centres is also a problem: it creates too much complexity in setting up the machine and too many settings which can vibrate loose and need resetting. As a consequence, one common suggestion is that such machine tools are not suitable for batch sizes greater than about 10,000 units,though frequency of the batches may also need taking into consideration. For larger batch sizes custom purpose made machines are typically more economical. For batch sizes less than 1000 units, traditional machine tools or manual methods may be more appropriate.

So for example if want to make bolts at the rate of several hundred thousand units per year, then a purpose made bolt threading machine would be the better option. Whether such machine uses computer numerical control (CNC) or is a simple mechanical system depends on the over all nature of the process. A simple mechanical system would require large stores of taps and dies, or cams if cam operated machine. A software library of thread forms would enable a CNC machine tool to cut a large variety of threads using a single point cutting tool. Whilst a die with the specific thread form required could be expensive to replace once worn.

As technology evolves it becomes available in a variety of forms, an evolutionary heritage can be traced. On the one hand general purpose components are specifically adapted to the technology, for example electric motors are integrated into the cases of the specific technology and the electric motor ceases to have its own case. Parts become integrated and blended together to create better aesthetics and safer machines. On the other hand the machine can be broken down into subsystems and each subsystem is made as an independent component, these components can then be assembled into more complex machinery. For example whilst could build a machine using an off-the-shelf electric motor, it is better to use and off-the-shelf drill spindle, such spindle can have a built-in motor or be designed to be connected to specialised motors. So for example could have a single drill spindle with own drive, or could have a bank of 6 drill spindles driven by a single drive. Using such heritage, what would once have been a complex task and produced ugly cumbersome looking machines, now becomes a simple task producing more elegant looking machines. Such heritage is the consequence of proactive design and engineering of products.

Now software tools are similar to machine tools. Software like Microsoft Office provides tools similar to the traditional machine tools, such as drill presses, lathes, shapers and planers, and mills. Each such tool performs a specific task well, but little use for anything else. The high end ERP systems and engineering design software systems are similar to flexible machining centres, such are not necessarily suited to the needs of a specific business. Now software developers, computer scientists and the likes study and program systems more directly connected to Computer Aided Production Management (CAPM) than they study systems relating to engineering product design. Consequently plug and play type software for planning and management type functions is more readily available than for design, science and engineering functions. Scientists, Designers and Engineers largely get an education using software tools not building them.

Consequently it is a lot easier to build custom planning and management systems than it is to build custom design and engineering systems. Software distributed under the GPL does provide some building blocks for science and engineering, but it is not exactly plug and play, nor well documented, and even if it was reasonably documented it would take a significant period of time to become familiar with. People studying for Masters and Doctorates are more likely to build custom tools from such, not practising designers.

So developing custom software for the planning, design and management functions of a manufacturer of structural products is not a simple exercise, most especially if the product has a variable form imposing need for additional engineering.

If the structural product is fixed form, then a general purpose tool such as a spreadsheet or general accounting software would suffice. If the product has variable form but no need for additional engineering, then general tools such as a spreadsheet or data base management system (DBMS) would be suitable for creating a product configurator. Such software could be custom designed to suit the needs of the individual business, the cost being heavily dependent on which tools are chosen as the foundation. For example a lot of software used by the carport and verandah industry are built around MS Excel (spreadsheet), when they would be better being built around MS Access (DBMS).

If the product has variable form imposing additional engineering then have some major stumbling blocks and major design effort required to produce custom software. Most of the engineering heritage for the structural products would have been carried out using off-the-shelf software for structural analysis, with member and connection design carried out using spreadsheets (QPro, MS xcel) or other electronic calculation books (MathCAD,MatLAB). These electronic workbooks are cumbersome to integrate with other parts of a manufacturing production system, and are not suitable for use by persons who have no engineering background. From another perspective however it is cumbersome to format calculation reports in high level programming languages.

The introduction of building information model (BIM) enabled software also adds another perspective to the over all picture. Such software aims to connect as much information as possible to 3D graphical entities, and have as many tasks as possible driven by such centralised data. So the dimensional and geometric model needs to be connected to the structural model for structural analysis, and also the workshop detailing. The structural design software has to be capable of more than structural analysis of a model in single form: that is components of the structure which do not fall into the stick framework concept, still need to be designed from a central model. So for example want to be able to design connections and footings from such model. Such facility is typically not available in structural analysis software and hence spreadsheets or pencil and paper are used to do such additional calculations.

So whilst buildings are an established technology and design procedures fairly well determined, such design procedures are not well documented, nor has much effort gone into full automation. Where automation is available it is focused on presentation of calculations carried out, it is not focused on automated design and production, and has tended to be material specific. For example whilst those using structural analysis software for hotrolled steel have had automated design for steel members and connections, for some twenty years, such facility has not been available for cold-formed steel, aluminium, stainless steel or glass. This is primarily because the largest consulting structural engineers make most use of steel reinforced concrete, hotrolled steel, timber and masonry. Other materials tend to be specific to manufacturers of specific structural products, and they typically are small business with out RD&D departments and rely on small consulting practices, or otherwise product testing. So whilst the demand for the structural products may be high, the demand for application of technology for the design process is low: and the developers of the high end software packages are slow to develop such design tools. More over the developers of such tools have poor support for Australian codes of practice irrespective of the materials, and Australian software developers haven’t exactly developed their software in the past 20 years.

So we have the problem that there is advanced software out there, but chances are it is not suited to Australian conditions, even if it is suited to Australian conditions it isn’t suited to the needs, wants and whims of manufacturers of structural products.

Another problem is that the manufacturers do not put any where near enough work through their external consultants for it to be worthwhile for such engineering consultants to purchase BIM enabled software. But a manufacturer potentially does have enough work to justify setting up an in-house design and estimating office using BIM enabled software.

If a manufacturer does not have an in-house design office, and wishes to by pass the need for such by the use of custom developed, highly automated design software, then the first apparent problem is that the manufacturer is an uninformed buyer of a complex and expensive system: the costs and timing of which can get completely out off control. Such software is vastly more complex than the structural products being sold, which the manufacturer is also uninformed about. The manufacturers and their distributors need a lot more than simply automated engineering design software, and fancy 3D graphics to help to sell their structural products: they themselves need to be far better informed. A salesperson that was selling plastic toys last week cannot simply jump in and start selling a product dependent on engineering design next week. Significant training is required as they are meant to be the informed guide to the buyer of the structural product.

The development of such software also has to take several stakeholders or affected parties into consideration, and these include:

1) The end-user
2) Sales People
3) Buyers
4) Developers (Structural Design)
5) Developers (Production Management)
6) Certifiers (Structural for the Software)
7) Certifiers (Structural for project put through the software)

…o0o…

{Run out off steam, so that’s it for the moment. Future articles will cover the stakeholders and more specific details about the problems of automating the engineering design process, in particular issues regarding the limitations of the wind loading code and the judgements which are made. In reality whilst numbers may be pulled out off the wind loading code to determine wind pressures, no building is actually compliant with the wind loading code, as most buildings have shapes more complex than allowed for by the wind loading code and are therefore beyond the scope of the code, but it would currently be too expensive to do wind tunnel testing on every building . Due to such judgements being made, little additional wind tunnel testing is done to check the complex fluid dynamics associated with these complex shaped roof and wall surfaces.}


Notes:

  1. Reading Time Approximately 25 minutes.

 


Revisions:

[19/08/15] Original: unstructured associative free writing.

[20/08/15] Minor changes to spelling, inserting missing words, fixing up sentences, and formatting.