Evaluation of the ICEngineworks Modular Header Modeling System

           "Lego's for boys with toys...."

Author:  W. Kurt Dobson, CEO - Spectrum5 Racing

Spectrum5 has designed and fabricated hundreds of headers for Formula 1, NASCAR, Busch, offshore racing, aeronautical and custom applications. All prior design work was accomplished with our own in-house automatic tubing design software.  This powerful software provides optimal designs, but requires the customer to already have full 3d CAD models of the engine and chassis or have the deep-pockets required to develop them.    

The header design business hasn't seen a major design innovation since Spectrum5 introduced it's automated tubing software design technology to the industry -- so when I  first saw the ICEngineworks modular design system I got pretty excited.  To me, this seemed like an excellent design solution for customers who do not have 3d CAD models to work with or the time and financial resources to develop them.  This group represents the largest segment of the design market who typically have the project vehicle in their shop and need to design and fabricate using only the vehicle itself.

We decided to evaluate the ICE system, by attempting to re-design an existing 4-cylinder Honda drag racing header system we originally developed a few years ago.  This particular design was chosen partly because at the time of writing this article, the only available tubing OD from ICE is 1.625 inches.  We look forward to future design examples as new kits arrive on the market from ICE supporting larger tubing OD's for our other customers. 

The ICE Kit: 

When the package arrived from UPS and was opened, I was really impressed.  Everything about the kit is professional and all the components are top quality.

Each kit comes in a durable foam padded travel case which helps keep everything organized in the shop.  This particular kit is the " 4-cylinder Transverse 1.625 OD Kit".  Inside the box are 4 zip lock baggies (one for each primary tube) with an ample assortment of the available bend radii (2, 3 and 4 inch CLR for this kit), straight sections, a cylindrical box which contains 4 fasteners (picture below) to attach the ICE primary tubes to a prototype exhaust flange, as well as an excellent users manual with everything you need to know to get started (which I admit I didn't even read as the ICE system is so simple).

Fit, finish and fitment of all the parts is excellent.  The modules are made of a very durable plastic and snap tightly providing solid mechanical integrity.    Witness marks every 30 degree's around the circumference of each part, makes measuring actual rotations of a design very straightforward.  

Evaluation:

We did our evaluation purposefully backwards, because in the process of prototyping a design, we wanted to compare the results obtained with the ICE blocks with a highly optimized computer solution.  We started with an existing Honda VTEC drag race Turbo design that had been computer generated using the Spectrum5 header design software.  This particular design was optimized for flow, primaries within 1 inch of equal length, and tight packaging constraints.  The design used 2, 3 and 4 inch CLR tubing radii which is identical to what is provided in the ICE kit.

We were interested in how fast a prototype could be put together using the ICE modules and whether or not having the exact arc angles specified in our design would still work.   The Spectrum5 automatic header design software, although written in MatLAB (a widely used scientific programming language)  produces 3d Solidworks models --  a popular 3d CAD software package.  The two top pictures show an earlier prototype of the design.  Note that there is only one weld on each of the primary tubes -- easy for the computer software to compute, difficult to do achieve manually.  

The two pictures below show a slightly improved equal length CAD design.  The picture on the right shows a fabricated header on the particular drag race car (a 9 second Honda civic race car).  On this slightly modified design, there is one weld required on the two inner primaries, and two welds required on the outer primary tubes.  We only had photo's of the car for the earlier design, which is almost identical to the newer design we prototyped.

The photos above are of the primary tubes in the Solidworks design environment and show a dummy tube where the merge collector would normally start.  A merge collector with an integrated turbo wastegate flange was also designed first in Solidworks, and then later fabricated.  Photo below is of the CAD design.

ICE Prototype:

Next, using the CAD design I put together a prototype using the ICE modular blocks.  Since we had a computer printout with the exact parameters for each primary, we had a good starting point.  One obvious problem was that while the computer designed solution specified exact dimensions for straights and bend arc angles, only discrete lengths of straights and available bends can be achieved with the ICE modules which are always 1.000 inches in length.  For example, the outer tubes of this design (starting from the merge collector) start with a 1.625 inch straight piece, followed by a 2 inch CLR 180 degree bend, followed by a 2 inch CLR 128.478 degree bend, followed by a 4 inch CLR 36.891 degree bend, followed by a 0.500 inch straight section going into the exhaust flange.

Obviously using the modular blocks, one is constrained to use combinations of the angles available in the kit.  The arc angle per block in this particular kit works out (by CLR) to:

Clr = 2.000 inch Degrees = 28.648

Clr = 3.000 inch Degrees = 19.099

Clr = 4.000 inch Degrees = 14.324

So, while we'd like a 180 degree bend, the best we'll get using 2.000 CLR ICE blocks is 6 pieces = 171.888 degrees.  Similar compromises in exact angles for primary segments using 3.000 and 4.000 CLR blocks were obtained.   It took about 10 minutes to snap all the correct pieces together to make up the 4 primary tubes,  10 minutes of which were spent trying to scotch tape the four primaries together at the merge point since I also didn't have a merge collector handy.  

When all this was put together, the design looks visually very much like the computer design.  The two photo's below show the actual ICE modular block prototype design from two different angles alongside a plot of the computer design.

Fortunately, we did have a CNC'd production assembly jig available for our Honda design which provide round 1.625 inch ports with the proper spacing and a 10 degree downward angle at the exhaust plane.   Using this, we inserted the modular design into the jig, anticipating a less than perfect fit with the slightly different arc angles.

With just a few twists of the bend rotations (like playing Rubik's cube), it took only a few minutes to coax the design to fit properly into the jig and achieve reasonably good perpendicularity at the exhaust ports (not perfect, but again just a few minutes of tweaking and without the exact arc angles for every bend in the design).

Wow Factor:

Consider what's happened to this point:   30 minutes elapsed time since we first opened the ICE Modular Kit box and we have prototyped an existing design with modified arc angles and made it fit into a production jig.  This is very promising.  

In contrast, the first set of headers I built I spent 3 weeks cutting metal, bleeding and swearing trying to get all the pieces to fit and I was never satisfied with the results and ruined about $1200 worth of stainless steel material in the process.  Waste material savings alone on a header project will pay for the ICE kit, and you'll save the three weeks of design time and never get your hands dirty.

ICE claims an average guy can prototype a header in 3-4 hours from scratch.  I tend to believe this based on my initial experience with the kit.  

Technical Caveats:

Available ICE modules are always 1.000 inch measured through the centerline.  The folks at ICE designed the blocks this way to make determining the primary length as easy as counting the blocks.  There are a few trade-off's with this approach you as a designer need to be aware of:  

• Available bend arc angles are also fixed in discrete angles.  Intuition tells me this is probably good enough in most cases as there are enough degree's of freedom in most solutions by rotating the components to achieve a workable geometry.  We will prove this point later in this article.

• Straight pieces are also fixed to 1.000 inches.  This is not a problem at the merge collector because if you need an odd length, you can simply cut a straight piece to the required length.  If you need an odd straight piece going into the merge collector, this can be a problem because cutting off a piece to length removes the ability to snap the parts together.  One possible solution is to simply glue two pieces together if needed. 

The overall technical issue that I wanted to answer with respect to the ICE modular system is whether having only these discrete lengths imply any compromise relative to a computer solution where exact arc angles and lengths can be used.  To answer this question, we note above that our ICE prototype is not quite the same as our computer generated design.  Rather than spending 3-4 hours twisting and turning rotations on the model (and me being both mathematically inclined as well as lazy), I resorted to the computer software to provide the answer to this question.

What we did is to take the original Honda VTEC computer design, and fix the lengths of the straights to discrete 1.000 inch sections, and also to fix the arc angles to discrete values obtainable with the modules.  We can then re-run the computer design software, leaving only the rotations between the primary pipe segments and the lengths of the straight sections as degree's of freedom in the solution and look at the results.  

Plugging these slightly different values into the computer header design software, the solution prior to optimizing the design looks like the picture below, with the primary tubes misaligned in both elevation and exact location and perpendicularity at the exhaust ports.  Again, this occurs because exact arc angles cannot be achieved with fixed ICE modules.

The header design software was then used to optimize this design and converge on a solution.  Two plots below show the results which achieve a perfect result with exact location and perpendicularity at the exhaust flanges.  The straight sections coming out of the merge collector ended up being 1.625 inches for the outer tubes (thus these would need to be trimmed in a real prototype).  The straight sections at the beginning and end of the inner tubes ended up being exactly 2.000 inches, so no trimming would be needed at the exhaust flanges.  First two pictures are the MatLAB renderings, followed by the two blue renderings after exporting the design into Solidworks.

Lengths of the primary tubes ended up being 15.625 inches for the outside tubes (due to trimmed straight section at merge collector), and 16.000 inches for the inside tubes.  This is better than most hand built headers ever achieve and one of the advantages of the ICE system.    

The resulting solution has several more sections per primary tube.  Looking at the photo above left, you'll notice all 4 primary tubes are made up of 6 separate segments.  Considering the actual metal part will use mandrel bends with straight sections combined with the first bend, this implies that the two outer tubes will require 5 welds, while the inner tubes will require 4 welds.  If you're willing to cheat a bit on the perpendicularity achieved at the exhaust ports, you can probably reduce this by 1 weld per tube in many cases.  What's important here is that these results are as good or better than most hand designs we see created by experts and this entire process can be done with the ICE system for a design like this in 2-3 hours without any computers or software.

In essence, what happens in an ICE design is that additional segments rotated slightly will sometimes be required to achieve a particular angle.  This compromise occurs only because ICE bend modules are fixed angles.  Using our computer software to optimize an ICE design proves that a user could accomplish the same thing by adjusting the rotation between primary segments until successful results are obtained.  The software just makes this process automatic and very fast.  To illustrate this point, shown below are seven 2.000 CLR bends snapped together (shown in a computer model), with one end sitting on a flat surface.  You'll notice 6 segments don't quite equal 180 degree's, while 7 segments are greater than 180 degree's and pass through the surface.   Can we make these 7 pieces into a 180 degree bend?  Sure, just rotate a few of the segments as shown in the photo on the right.   It doesn't matter which segments are rotated.   As an example, it turns out that a 180 degree bend for the 2.000 inch CLR ICE components can be achieved by rotating one section about 50 degree's counter-clockwise and the next section about 50 degree's clockwise.  This implies a rather odd looking 180 degree overall bend, but also 3 extra welds due to the rotation.  

Another consequence to adding these rotation is that the overall width of the 180 degree arc is now closer to 5 inches in center to center spacing rather than 4.000 inches (twice the CLR of the bend radius).  This is an unavoidable mathematical consequence of these additional rotations.  The photo on the right illustrates a computer model of this effect showing the witness lines on the ICE modules.

Given this characteristic of the ICE system, a good strategy is to attempt to get the required angles by rotating first only at the intersections of tube segments that you know will already be a weld seam in the final metal product.  If this fails, then you can rotate one or two other sections somewhere in the middle of the solution and achieve the results you want albeit with an extra cut and weld for each.

Now, back to the ICE prototype -- while the original fully optimized computer design used only 1 weld for the two inner tubes and 2 welds for the two outer tubes, this is usually impossible to do by hand even in metal where exact angles can be cut.  We know the best NASCAR Master header builders who've been designing headers for 20-30 years and their solutions always use an extra few welds per primary tube compared to a computer solution.  You can achieve the same results as the Masters with the ICE system without investing the 20 years.

Can the computer software generate a better solution?  Almost always, but unless you are going to build a production run, you probably don't care about a few extra cuts and welds.  If you are going into production with your design, then all it takes is sending your ICE design to Spectrum5 and we'll optimize it for you using our software and you'll have an exact solution that achieves the lowest manufacturing cost.  This can also include optimizing the design for production on a CNC bending machine.    

Conclusions:

The ICE Modular Header Design technology is the biggest breakthrough in header design in years.  It provides the best, quickest, cleanest and most cost-effective way to rapidly prototype a header design manually.  More importantly the ICE solution addresses the largest market segment in the performance industry requiring in-shop design where the vehicle is available, but not expensive 3d CAD models and software tools.

Results achieved with the ICE system will usually contain a few more cuts and welds than a computer solution, but this is as good or better than any other known hand design method and vastly quicker.  We love it!

We think the ICE system is a huge win for the vast majority of header designs done in the shop.  There are also some great synergies created in some cases using the ICE system in conjunction with Spectrum5's header design software.  Spectrum5 capabilities include:

• Professional Fabrication of your ICE prototype:   Let Spectrum5's fabrication experts build your ICE design.  It takes special TIG welding equipment and many years of practice to properly weld, especially using stainless.  If you can't weld stainless and get shiny gold welds, you're overheating the material and it will crack.  Just send your ICE prototype to Spectrum5 and let the professionals do this for you.  This hasn't been possible until the introduction of the ICE prototyping system.
• Optimization of an ICE prototype:  For the best quality prototype or a production design with minimum cuts and welds, and/or CNC tube bending.
• Conversion of ICE design to other OD, or other CLR's:  At the time of this article, not all OD's or CLR's were available from ICE.  Prototype now in 1.625 modules, and Spectrum5 can automatically create a design in any other OD and still meet packaging constraints.
• Conversion of ICE Model designs to 3d CAD models:   Some customers may want to have a 3d solid model of their product for documentation, marketing, customer support or other reasons.  Spectrum5 can provide this.
• ICE Design Jump-Start Kits:  Spectrum5 has hundreds of existing designs which can be used as a jump-start for your product.  These will be made available online allowing one to quickly prototype a NASCAR header, many popular hot-rod headers, dyno headers, etc.
• ICE Prototyping Flange Kits:  Spectrum5 offers aluminum prototyping flanges that mate exactly to the ICE modular blocks.  Production parts are available in mild and stainless steel.  Many exhaust flanges are not round in shape.  Spectrum5 offers special flanges designed especially to support the ICE modular prototyping system.  These flanges are CNC machined in aluminum, and have a built-in counter-bored fastening system with an M6 threaded boss to secure the first ICE module.  After your prototype is finished and you're ready to fabricate in mild or stainless steel, Spectrum5 offers a full line of identical production flanges.  The photo below shows both an ICE prototype flange as well as the production flange as an example for the GM small block engine with AirFlow Research cylinder heads (AFR-220).

Here at Spectrum5, we plan to use the ICE system on any design that isn't CAD based.  It's simply great stuff from the great folks at ICE with a completely innovative approach to manual header design that will save us time and money in our operations.  In many of our ICE designs, we'll still use our software to optimize to a minimum number of cuts and welds.   We're proud to be a distributor for these fine products.

Vendor Information:

For more information on ICE modules and components, see the ICEngineworks web page at:   www.icengineworks.com 

For more information on Spectrum5 Automated Tubing Design Software, see:    www.spectrum5.com  

For more information on Spectrum5 Racing products, ICE prototyping flanges, fabrication services, please see:  www.spectrum5racing.com 

Author's Bio:

W. Kurt Dobson is CEO of Spectrum5 and Spectrum5 Racing (www.spectrum5.com  & www.spectrum5racing.com), developers of automated tubing design software and fabricator of racing headers and related components, respectively.   Mr. Dobson has served as Adjunct Prof. of Electrical Engineering at the University of Utah, was IEEE 'Engineer of the Year' 1997/1998 Utah, and has authored 42 U.S. Patents (12 granted) primarily in the mathematics of and signal processing for wireless communications.  Mr. Dobson is a road racing enthusiast and is a licensed SCCA and NASA competitor and road race instructor and has raced professionally in Sports 2000, CSR, GT2 and GT1 racing venues.   Photo below shows author with his youngest son, and two race cars, a 1997 Porsche twin-turbo (GT2 Class), and mid-engine Ultima GTR with 1300bhp twin-turbo power (Unlimited Class).