Making Turbo Manifolds, Part 2

Cost

The total cost of making the manifold was about AUD$500. This can be broken down to:

Steel plate (0.5 metres x 12mm x 100mm) - $16

Buttweld fittings (5 x 45-degree, 4 x 90 degree) - $72

Welding and cutting (6 hours at AUD$50 per hour) – $300

Sandblasting - $20

Facing of both mounting plates - $90

The Prius Manifold Design

This manifold was made for an NHW10 hybrid Prius. The Prius uses a 1NZ-FXE 1.5-litre, 4-cylinder engine developing 43kW and 102Nm, both at 4000 rpm.

The valve timing is arranged to provide an Atkinson cycle. In this design, the closing of the intake valve is delayed such that the rising piston pushes some of the intake air back into the manifold, reducing at lower rpm the amount that is inhaled. This results in an actual compression ratio much lower than is indicated by its 13.5:1 geometric compression ratio. However, the expansion ratio that occurs as the fuel/air mix is burned remains as high as the compression ratio indicates, so improving efficiency.

Intake valve opening is variable from 10 degrees Before Top Dead Centre to 30 degrees After Top Dead Centre with a duration of 270 degrees, while the exhaust valve timing is fixed with opening occurring at 32 degrees Before Bottom Dead Centre and closing 2 degrees After Top Dead Centre, giving an exhaust duration of 214 degrees. The overlap is therefore variable from -28 degrees (ie 28 degrees of underlap!) to 12 degrees. This exhaust duration is much shorter than found in most engines and the inlet duration much longer. The amount of overlap is considerably less than found even in engines with variable valve timing. In addition to varying idle smoothness and power, the variable intake valve timing is used to provide at times internal Exhaust Gas Recirculation (EGR).

The standard intake manifold is not a tuned-length design, with Toyota stating “it is not necessary to improve intake air efficiency through inertial intake due to the adoption of the Atkinson cycle”. The description also states the intake pipes “have been integrated midstream to achieve large scale weight reduction”. As this diagram shows, the intake manifold uses a 1>2>4 design with a plenum chamber volume fairly typical of four cylinder engines.

The standard exhaust manifold is a cast iron design that uses short runners grouped 4>2>1 and primary runners about 2¼ inches long. These flow into a large volume before exiting via a 1.5 inch outlet.

The lack of traditional tuned-length intake and exhaust manifolds suggests that this type of tuning is not particularly important in this engine design. (Perhaps it’s also a case that adequate resonant tuning cannot be achieved in the space when peak rpm is so low.) The non-traditional grouping of the exhaust outlets (cylinders 1 and 2, and cylinders 3 and 4 paired, rather than cylinders 1 and 4, and 2 and 3) is also significant.

The use of internal EGR and the Atkinson cycle, and the ability to vary valve timing to provide negative overlap, makes traditional design approaches to turbo exhaust manifolds rather problematic. We thought it possible that considerably altering the runner length or layout could upset aspects of the engine’s operating system.

In the absence of any other guide, when building the turbo exhaust manifold we chose to follow the runner layout of the standard manifold.

Turbo Manifold Design

Start talking about turbo exhaust manifold design and people have all sorts of theories. Most say that equal-length, long runners should be used – irrespective of the length of runner that then results. But others say runners should be grouped on the basis of firing order. Sounds easy – until you ask some questions. Like, grouped exactly how on the basis of firing order? Or, how important is it that the runners are of equal length? For example, is it more important that runner length be equal – or the runners are organised to provide the best flow? After all, the longer the runner, inside a typical engine bay the more bends it’s likely to have in it and the greater resistance it will pose to flow.

Let’s take a look at what the experts actually have to say.

The original bible of turbocharging is Turbochargers, by Hugh MacInnes (published by HP Books). Despite being first published in 1978 – and so containing almost nothing that relates to EFI engines – the core content of the book has stood up surprisingly well in the years since. MacInnes suggests that turbo exhaust manifolds should use small diameter runners with about the same internal area as the ports and that in turbo engines, the use of “smooth flowing exhaust headers with beautiful swerving bends.... is more aesthetic than power-increasing”. Except for V8 engines, he makes no comments at all about grouping the flow from cylinders in any particular manner.

Another old book is Turbocharging and Supercharging, by Alan Allard (first published by Patrick Stephens in 1982). Allard says: “The main criteria when designing and fabricating an exhaust manifold are: firstly, to build in sufficient strength to take the weight of the turbocharger system and to remain rigid without distortion or fracture even when working up to 1000 degrees C; and secondly, to have sufficient wall thickness (3.0mm minimum is recommended) to withstand the corrosion effects of running up to high temperature over a long period.”

Allard suggests the use of a log-type manifold pipe of not more than 2.5 times, and not less that double, the area of one exhaust port. The log is joined to the individual exhaust ports with stubs with the same inside diameter as the exhaust ports, each as short as possible and of equal length. The stubs can enter the log at right-angles or be angled towards the turbo.

However, while not mentioned in the text, a diagram shows a 1-3-4-2 firing order four cylinder engine using a manifold where cylinders 1 and 2, and 3 and 4, are paired and fed to a split-pulse turbine. In addition, again while it is not discussed in the text, many turbo racing engines are shown where equal-length long runners join at a common collector just prior to the turbo.

Automotive Supercharging and Turbocharging Systems was first published in 1992 by Motorbooks International. The author is John Humphries. Of my references, this book provides the most detailed treatment of turbo exhaust manifolds. However, rather than making things clearer, if anything it further muddies the waters! The book suggests that there are two fundamentally different approaches to turbo exhaust manifold design.

The first is to use a manifold with sufficiently large internal volume that the exhaust output pulses of each cylinder are damped and a more or less constant pressure is available to the turbine. The internal volume of the manifold sufficient to obtain this pulse dampening can be 1.4 – 6 times the swept volume of the engine. That rules out pretty well all long runner exhaust manifolds, although a log-type one of the sort suggest by Allard may fit into the bottom end of this scale, and the current fashion in the US for mounting the turbo at the back of the car (in a car with a front engine!) would also conform to this approach.

The second approach is a pulse system, where the exhaust pulses provide additional short-term energy to the turbine. In a pulse-type manifold, Humphries suggests that the pipe runners should have a “cross-sectional area....not significantly greater than the geometric valve area at full lift [and] these connections should be kept short and free of sharp bends”.

He says the reflection of pulses within the system will be determined by pipe length, exhaust temperature and the status (ie open, closed or partially open) of the exhaust valves. In addition, at pipe junctions the exhaust pulses will split, with smaller magnitude exhaust pulses travelling down each pipe. “The overall pressure wave system that occurs in such a manifold will be very complex, with pulses propagating from each cylinder, pulse division at each junction, total or partial reflection at an exhaust valve...and reflection from the turbine.”

In order to take advantage of this pulse flow, “narrow pipes from several cylinders can be connected through a single branched manifold to one turbine....a four stroke engine which can have its cylinders grouped into threes is particularly attractive.” This is because “the opening periods of the exhaust valves follow successively every 240 degrees with very little overlap between them.... thus a sequence of pressure pulses arrives at the turbine...”

Humphries suggests that the use of twin turbos on a six cylinder engine allows for efficient pulse operation, and where cylinder multiples are not in threes, a single turbo entry can be linked to multiple cylinders through “pulse converters”. Pulse converters are suitably shaped junctions which prevent reverse pulse flow. Humphries shows a four cylinder exhaust manifold with cylinders 1 and 2, and cylinders 3 and 4, paired and then coming together through a pulse converter junction.

One of the more recent books on turbocharging is Corky Bell’s Maximum Boost (published by Robert Bentley, 1997). Bell suggest that it is important the manifold retains heat, prevents reverse flow (eg by the use of so-called reversion cones in the first section of each runner), and is designed to minimise thermal loads on each section of the manifold. The latter can be achieved by the use of runners from each cylinder travelling separately to the turbo inlet – that way, each runner is subjected only to the heating loads of one cylinder. It is implied but not stated that controlling these heating loads is more important than flowing the individual pulses in a sequence to the turbine – in the diagram the pipes are of unequal lengths.

Bell also says that the experience with turbo F1 cars suggests that “the best manifolding is multiple-tube, individual runner style”. As with the other authors, he recommends the use of relatively small diameter runners with large wall thicknesses. With regard to pulse tuning, he says “a design that allows exhaust gas pulses to arrive at the turbine at regularly spaced intervals is ideal but difficult to achieve”.

So what does one make of all of that?

Firstly, it’s clear that these authors agree that the use of heavy wall tube (“steam pipe”) bends are preferable to thin gauge materials. Secondly, the individual cylinder runners should be sized smaller rather than larger, being near to port size. It also appears that if it is possible within the confines of the engine bay, equal-length runners that join at the turbo are to be recommended. In six cylinder engines, the grouping of two pairs of three cylinders to feed either two turbos or a single split-pulse turbine housing is to be favoured.

However, unequal length runners are extremely widely used (few if any factory turbo cars have equal length runners in their cast manifolds) and some aftermarket tubular manifolds use branchings of unequal length runners. (Most of the latter are dubbed ‘pulse converter’ manifolds but whether the internal junctions conform to pulse converter geometries is not known.) Not one of the best known references is particularly critical of exhaust manifold designs which on a naturally aspirated engine would be seen as fatally flawed.