Sign In

Feel the force to design world-beating products

Imag­ine a 3D com­put­er-aid­ed design (CAD) on screen, cov­ered in a fine mesh of small tri­an­gles. “FEA [finite ele­ment analy­sis] nor­mal­ly reveals all the safe areas of your design as blue and green, and the ‘at risk’ areas, where fatigue could cause fail­ure, in yel­low and red,” says Steve Cory, com­put­er-aid­ed engi­neer­ing man­ag­er at Thatch­am-based Xtrac.

The £40-mil­lion engi­neer­ing busi­ness makes fail-safe parts, such as gear­box­es, for top-end motor­sport appli­ca­tions, includ­ing For­mu­la 1, mak­ing exten­sive use of FEA as a fun­da­men­tal tool.

Typ­i­cal­ly, a design engi­neer builds a com­po­nent in CAD, tests it in sim­u­la­tion soft­ware, which is often, but not always, with­in the CAD suite, and imports it into a prod­uct life­cy­cle man­age­ment (PLM) pack­age when that part needs to be man­aged across users as data with­in a larg­er assem­bly across mul­ti­ple users.

Sim­u­la­tion soft­ware func­tion­al­i­ty has improved great­ly in recent years, dri­ven by intense ven­dor com­pe­ti­tion and the demands of very com­put­er-lit­er­ate engi­neers. “Much sim­u­la­tion can be per­formed with a cur­so­ry under­stand­ing of physics; you don’t need a mechan­i­cal engi­neer­ing degree,” says Mike McDer­mid, senior design engi­neer at Sheffield firm Per­for­mance Engi­neered Solu­tions (PES), which is used to design­ing any­thing from elec­tric motor­bikes to sports equip­ment.

Using sim­u­lat­ed com­put­er-aid­ed design with­in a PLM sys­tem, in a fast-mov­ing and flex­i­ble engi­neer­ing func­tion, is fun­da­men­tal­ly impor­tant

This year Siemens PLM Soft­ware launched its CAD plat­form, NX 8. A favourite with the motor­sport indus­try – Williams and Red Bull both use NX with Team­Cen­ter, its PLM suite – NX 8 boasts 200 new capa­bil­i­ties over its pre­vi­ous ver­sion of NX, includ­ing improve­ments in the speed of geom­e­try and analy­sis mod­el­ling, and topol­o­gy opti­mi­sa­tion and mul­ti-physics analy­ses.

What does this mean for the design­er and man­u­fac­tur­er? “The biggest improve­ments in sim­u­la­tion are in their usabil­i­ty and analy­sis pro­cess­ing time,” says Dan Fleet­croft, engi­neer­ing direc­tor at PES. Core func­tions have remained fair­ly con­stant, but new mul­ti-physics capa­bil­i­ty, such as ther­mo-mechan­i­cal solu­tions and non-lin­ear analy­sis, is a step-change, he says.

In the real world, very few prod­ucts are sub­ject­ed to a sin­gu­lar phys­i­cal force. “By sim­u­lat­ing the inter­ac­tion of these dif­fer­ent forces we achieve greater fideli­ty between our analy­sis and real­i­ty, ulti­mate­ly giv­ing us greater con­fi­dence in the results,” says Mr Fleet­croft. “Mul­ti-physics soft­ware can com­bine a range of mechan­i­cal, flu­idic [aero­dy­nam­ic-hydro­dy­nam­ic], ther­mal and elec­tro­mag­net­ic forces that may influ­ence prod­uct per­for­mance.”

One improve­ment in FEA is linked to the pro­cess­ing pow­er of the hard­ware. Soft­ware devel­op­ers have enabled these mem­o­ry-hun­gry pro­grams to be more effi­cient and more “hard­ware-aware”, says Mr Cory.

“Our work­sta­tions use graph­ics cards and the recent ver­sion of Ansys [sim­u­la­tion tool] can recog­nise GPUs – graph­i­cal pro­cess­ing units – that live on the graph­ics card,” he says. “Because the card is con­cerned sole­ly with pre­sent­ing the best pos­si­ble image on screen, it has to do a lot of num­ber crunch­ing to keep up. Par­ti­tion­ing the pro­cess­ing between core CPU [cen­tral pro­cess­ing unit] and card real­ly speeds up the sim­u­la­tion.”

Sim­u­la­tion can­not be per­formed with­in PLM. But using sim­u­lat­ed CAD with­in a PLM sys­tem, in a fast-mov­ing and flex­i­ble engi­neer­ing func­tion such as Xtrac’s, is fun­da­men­tal­ly impor­tant, says Mr Cory.

“It pro­vides a sta­ble plat­form for the ‘mas­ter mod­el’ con­cept; that is a sol­id mod­el of a com­po­nent or assem­bly that dri­ves all oth­er data – draw­ings, CAE [com­put­er-aid­ed engi­neer­ing] struc­tur­al analy­sis and CNC [com­put­er numer­i­cal con­trol] man­u­fac­tur­ing pro­grams,” he says. “Re-use of exist­ing data is sim­pli­fied, stream­lin­ing the devel­op­ment process, and dri­ving the rapid adop­tion of best-prac­tice sol­id mod­el­ling tech­niques and com­pa­ny stan­dards.”