DIY Flowbench project

[Roman]

For ages I've wanted to dip a toe into the world of porting heads, which of course means you need some way to measure results. 
So a flowbench is needed.
There are lots of plans online, however most of them use water filled tubes and other completely valid forms of measurement which arent any good for drawing graphs on the internet with.

I've now got an okay handle on programming arduinos and touchscreens and understand sensors well enough that I'm gonna try put something together that's fairly well automated. 
It will probably start out simple and then I will get bored and not finish it I will add more bits to it

So the first part of the problem.

Valve actuation  
The results of a flowbench usually show a graph that demonstrates airflow from the valves being mostly closed, through to right open. 
Something like this:



So you can see that this particular test opens the valves and tests in 0.05" increments. 
I thought it would be nice to be able to both measure how far the valve has gone down, and also have a mechanism for pushing the valve down, automatically so it can just take measurements at set intervals.
So this way there's good repeatability without human error but also it helps automate the testing to be less laborious. 

Even with soft-ish valve springs fitted it's still a reasonably strong force to push the valves down. Speed isnt important but high torque and being able to hold in place is. 
So a stepdown geared worm drive motor should be good for this. Not a stepper motor, just DC but doesnt matter as the worm drive gearing holds everything in place (hopefully) 



For measuring how far the valves are depressed, I found out that some dial indicators have a data port on them, so they can spit out measurements over serial. sweet.

 



In order to get the USB connection into an arduino, need a USB to serial converter like so.  



So the idea is that the dial gauge will sit on one of the valve buckets, then the stepper motor will be a fork shape that can push down on both of the buckets separately. 
So you can find "zero" position by reversing the motor until the value on the dial indicator stops changing. Then you know you're at home position.

So I'm going to try make an assembly that holds all of the above that can easily be moved from cyl to cyl. So it will likely attach to something that can bolt in place of the regular cams, or use the cam cap bolts.

It's important to note at this point that I've ordered all of the parts, but not actually done any of the work apart from percolate ideas in my head for a long time. 

[Roman]

Airflow 
The next issue is being able to pull enough air through a system in order to create a big enough pressure drop that it actually simulates the properties of a port/valve. 
What I mean by this is that if you are at max valve lift but you are only flowing a tiny trickle of air through, it's not going to illustrate differences in flow because nothing's a bottleneck. 

Generally flow benches that use manometers (water tubes) for measuring inches of water as a pressure drop, run at 28 inches of water which is something like 7kpa below atmospheric pressure.
Any less than this, and it's not really going to be illustrative of the shape of a port etc. 
I think basically it's a case of higher vac the better, but it just becomes impractical. So 28" has become the happy medium that's adopted as industry standard.
Another reason for standardizing the vacuum level is that it makes the maths much easier to work out CFM from manual calculations.
As flow benches have existed long before computers have.

So the less restrictive your head is, the more air you need to flow in order to make 7kpa of pressure drop.
What I'm wanting to test for starters is a garbage spec 1NZFE head so I'm hoping I might get away with a single motor setup. 
However some use multiple vacuum motors in order to get enough flow.
In this case I've got a fairly decent sized motor in a commercial vacuum cleaner that I'm hoping will do the trick.

Since its not 1947 and I am not planning on doing any manual calculations for airflow.
I have no intentions of sticking to a fixed pressure drop for sake of easy maths.

I am thinking that if the valves are only partially open, and this creates a higher pressure drop than when they are fully open, this is fine!
It will be a closer approximation to real life situation inside the engine at low valve lift.
I will still have an accurate mass airflow number, and the higher vac level should show any changes with better "resolution". 
My thoughts on this may change once I get it working though.

The sensors I am going to use for airflow are...

-Normal Toyota 1 bar map sensor 
-Diesel particulate filter sensor (which is a cheap differential pressure sensor) 
-Toyota MAF sensor (For mass airflow results, and temp sensor) 

I'm thinking that the MAF sensor will do most of the work, but I'm not betting the house on it. 
Since a normal map sensor reads 0-5v from say 0kpa to 150kpa, it doesnt have a huge resolution in the relevant range if we're only going to something like 7kpa below atmospheric pressure. 
So a differential pressure sensor can be useful for measuring vac chamber vs atmospheric with good resolution, as they are 0-5v output but over something like +- 10kpa range.
However before you turn on the vac source, the regular MAP sensor can also earn its keep as a barometric pressure sensor. 

Also, I think its incredibly important to have as fewer variables as possible away from your actual motor as it sits in the car. 
One of the "gotchas" of porting a bare head, is that your car never drives with a bare head. So it's kinda pointless knowing how a bare head flows.
The sharp corners at the entrance of the port can bias the flow away from areas which are more important in real life when the flow into it has been straightened by an intake runner. 
So what is common is to build up a bellmouth entrance onto the head using putty or something, like this



Which I think is a bit silly, because with a normal manifold on you're likely gettting a much straighter shot of air into the port than an aggressive turn in, right at the head. 
You could be introducing weird swirling or turbulence or whatever else that isnt accurate to normal operation.
Also, the repeatability of making a shitty bellmouth entry like this is debatable. 
You might see changes simply from how much you've smooshed it in place, or how tall it is each time. 
So I'm planning to run it with a complete intake manifold on, as it would run in the car.
Because ultimately that's what is going to be correct. 
If your test shows that the head flows way better than your manifold does. Then there's hardly any point being a fuss pot over the state of the head.

[nzstato]

This thread interests me alot...

Be wary that some vac cleaners end up 'pulsing' a bit and may need to be accounted for in your measurements.

 

[Roman]

Good point. 
This vac has quite a large internal chamber and a big grunty motor so I'm hoping it'll be fairly steady in that regard. 
Hopefully I will be able to pick that up with sensors if it's happening. 

[ajg193]

Why not use servomotors for the valve actuation? You probably don't even need a valve spring. Just 3d print a jig that can move from valve to valve and clamps on where the collet groove is. Just need to calibrate once with the dial gauge.

Could also use the map value to control the vacuum with pwm. The vacuum motor is probably a universal motor and would run fine on DC.

[Stroker]

I'm keen to see how this turns out. I'd love to make my flowbench less manually dependent.

I built my flow bench a couple of years ago. I based mine on this design.  http://dtec.net.au/Flowbench Design Guide.htm

But used a couple of cheap Ebay  digital manometers for measuring pressures. I downloaded a spreadsheet that had all calculations done for head flowing. I just had to enter the pressures in manually and I could print a graph from that. Very low tech.

I use 4 vacuum motors in mine which is just enough to flow my Ford 2.8 heads at 28" so I suspect you'll need a lot more than a single vacuum motor for your 1NZFE head.

My flowbench is ugly but works and is repeatable. Great tool and fun to build and learn from.

8 hours ago, Roman said:

If your test shows that the head flows way better than your manifold does. Then there's hardly any point being a fuss pot over the state of the head.

I've found that even with a low flowing inlet runner, improving the head flow usually gives an increase of flow at the valve with the manifold attached. It's just a smaller increase than the bare head. You will always get a decrease in flow to the valve with a inlet manifold attached, no matter how well the manifold flows.

[Muncie]

Could you use a centrifugal blower or even a roots supercharger and spin it with a briggs or something? Even a small one should move plenty of air for a single cylinder.

[fletch]

Possibly could sponsor a differential pressure transmitter. 

4-20mA. I think I have a 10kpa one that is surplus. It has damaged thread on the cable entry so its certification is not valid.

[flyingbrick]

15 hours ago, Muncie said:

Could you use a centrifugal blower or even a roots supercharger and spin it with a briggs or something? Even a small one should move plenty of air for a single cylinder.

@Roman if you scope creep to include a roots blower driven by the old eco engine- I'll buy you dinner.

[Roman]

Haha, that would be a laugh. 

Nah ive actually got 4-5 1500w vacuum cleaners here so theres room to upgrade if needed.

Thanks for your feedback @Stroker thats all very good to know. its handy to know right from the start that I will need multiple motors.

Ive been having some thoughts about just getting my ECU to log the data, and use megalog viewer to interpret and compare results. 

As I can send all of the flowbench data from the arduino to ecu over canbus. I just have to park the car near it and plug a cable in haha.

 

So the arduino would still control valve distance etc but just saves reinventing the wheel a bit with logging etc.

It also means I will have before/after engine logs in a compatible format with the bench data. So I can directly see how its changed things in real life (if at all)

[anglia4]

50 minutes ago, Roman said:

So the arduino would still control valve distance etc but just saves reinventing the wheel a bit with logging etc.

I was only here to pinch your methods for data-logging with the Arduino 

[Roman]

20 hours ago, anglia4 said:

I was only here to pinch your methods for data-logging with the Arduino 

I've written some stuff before, for logging to my digidash. 

But it depends on what board you're using. 
If using Teensy 3 or 4 then its easy as you've got onboard eeprom or flash memory. 
But if using an arduino mega or something, need to write to SD. Which isnt too hard there's code for it. 

It becomes harder when you want to transfer the data to anything else though.
 

[Roman]

On 21/06/2021 at 15:18, ajg193 said:

Why not use servomotors for the valve actuation? You probably don't even need a valve spring. Just 3d print a jig that can move from valve to valve and clamps on where the collet groove is. Just need to calibrate once with the dial gauge.

Could also use the map value to control the vacuum with pwm. The vacuum motor is probably a universal motor and would run fine on DC.

I've been thinking about this, and yeah servos could be a really good option. 

Only thought about not running a valve spring is that at low lift there can be a fairly grunty pressure differential on either side of the valve, trying to pull them open. So having a valve spring seems useful to make sure there arent tiny oscillations or whatever. 

[Roman]

This has been slow going, waiting for parts to turn up / other shit going on. 
But I managed to get my MAF in a pipe hooked up to my arduino nano.
So I could quantify how good/shit each of my vac cleaners are as a vac source.

I've already got a MAF scaling for this pipe size, from my engine tune. So I can just refer to this for voltage to airflow conversion:




So with my big industrial shop vac, it reached 2.68 volts on the MAF. 
so this is about 40 grams/sec of airflow. 
Which is about 60hp worth of air. 

I tested the other vacs that I've got, and they were worse than this.
Using the best two running at once, I could get the MAF up to 3.0 volts. So around 70 grams per sec / 84 horsepower worth of air. 

So is that enough to simulate a single runner? 
84hp x 4 runners = 336hp 

Which sounds good but unless you have a 720 deg intake cam, this is not representing anything close to the peak or average airspeed at all.
268 deg cams out of a 720 deg stroke = 37% of the time it's breathing in air. And then much of this time is spent at less than full lift obviously.

So to truly simulate the airfow through a single port on a 4 cyl 150hp motor I'm thinking I'll need at least 100-120 grams per second capability.
So I will look at getting some vac motors all of the same and wiring them up. Maybe 4 to start with. It's a better situation if all of the motors are identical rather than mix and match. 
But I can rule out the idea of running just a single vac to get meaningful results anyway, even if its a grunty one.

On my previous search I only found more expensive ones, but 2kw vac motors are $50ish each delivered from aliexpress so might start wth 4 of these, and hope that it's overkill! 
https://www.aliexpress.com/item/32966020456.html?spm=a2g0o.productlist.0.0.7e675d6fuVKTf7&algo_pvid=8dca2d00-4acb-4a6d-b88a-a9401a618dcd&algo_exp_id=8dca2d00-4acb-4a6d-b88a-a9401a618dcd-3

 

[h4nd]

You have them in series or parallel? (i.e. are you running up against a pressure differential limit on the vacs?)

[Roman]

So I've just got my intake pipe with MAF in it, attached to throttle body, attached to intake manifold. 
Then everything on the manifold blocked off. 
Then printed an adapter so I can shove a vac hose into an intake runner.
To test two, or more, I unblocked off the other runners and jammed more vacs in. 
But also tested them one by one in a single port. 
The worst had about half the flow capability of the industrial one. 
The other two somehwere in between. 
 
 @h4nd do you have any good thoughts about wiring up 4x 240v motors? 

I've been looking at some 240v speed controllers, one I saw had 4000w capability. 
So could potentially run 2 motors full blast (or just on/off) then vary the other two at the same time to reach a goal flow amount
240v wiring makes me nervous!

[NickJ]

rather than airflow, what static vacuum can you achieve with those fans?

I get the feeling you may struggle to generate the vaccuum and subsequent flow required without serious spending 

[h4nd]

3 hours ago, Roman said:

wiring up 4x 240v motors? 

 

3 hours ago, Roman said:

240v speed controllers

Light dimmer style? Probably OK, best ask a tame sparky. @UTERUS etc?

Chuck'em in parallel, use sensible fuses, make sure chassis are securely grounded. Insulate them for another layer of safety.

4kW is getting pretty heavy, you're gonna maybe melt your plugs / sockets / wires in the wall / set the suburb on fire / raise the CO2. Maybe get a (safe) current meter so you know how reckless you're being?

I think @NickJ comment may be worth considering. Pneumatic engineer, I ain't

[Stroker]

Just for reference, my vacuum unit uses 4 x 1200w vacuum motors that I have in parallel, wired in 2 x pairs. That way I only run the single pair when testing low valve lifts. With 4 going it's just enough on a hot day to achieve 28in vacuum on my 2.8 v6 heads at peak lift. Something else to be aware of is heat, the vacuum motors usually require airflow over them for cooling. So in vacuum they tend to get hot quite quickly. That's one reason I use a air bypass setup to adjust my test pressure, that way it increases the airflow for the vac motors.

Also my flowbench flows a max of 247cfm@10in.

[flyingbrick]

20 hours ago, h4nd said:

 

Light dimmer style? Probably OK, best ask a tame sparky. @UTERUS etc?

Chuck'em in parallel, use sensible fuses, make sure chassis are securely grounded. Insulate them for another layer of safety.

4kW is getting pretty heavy, you're gonna maybe melt your plugs / sockets / wires in the wall / set the suburb on fire / raise the CO2. Maybe get a (safe) current meter so you know how reckless you're being?

I think @NickJ comment may be worth considering. Pneumatic engineer, I ain't

I'm running 4000w of heater off a multi plug in our garage- doesn't even get warm  

Edit. The heaters get warm... the plug/socket does not.