The following study paper was written by Richard Taylor about the effects of flywheel imbalance on a Lambretta engine and is published on Lambretta-images.com with his consent.
For many years I’ve been merrily bolting on Indian flywheels without a care. Flywheel balance is something I never bothered to consider in all honesty as I’m sure many have done and will probably continue to do so, but I won’t be one of them. I’ll be dynamically balancing all flywheels going forward.
Please read through this information and perhaps replicate some of the bench tests. In addition I would recommend you find a local company who can perform dynamic balancing.
Some people may tell you that flywheel balance only affects certain RPM’s, this is NOT true. If the flywheel is out of balance then it is out of balance throughout the entire rev range. Importantly the loads get exponentially worse with higher revving engines.
What is dynamic balancing
Static balance is achieved by balancing the object between centres and detecting which edge drops under gravity. Dynamic balancing spins the object at a known RPM. This has the advantage of increasing the measured force and defining angular position by means of a rotary encoder. It’s much the same as wheel balancing but it is typically done vertically (to negate gravity) and much faster, typically 800-1000 rpm.
What is the balancing unit of measure
The typical unit of measure is g/mm or Kg/mm (conversion between the two is obviously 1000). In layman’s terms if a flywheel has an imbalance of 200g/mm then it would require 200grams of material to be removed from the object 1mm off the rotational axis, at the point of imbalance. The weight of material required to be removed gets less the further away from the rotational axis. For example 200g/mm would require the following weights removed at different radii.
200 grams @ 1mm
20 grams @ 10mm
2 grams @100mm
How to calculate out of balance load
Out of balance g/mm or Kg/mm can be converted into a load. You can look this up on the internet and create a graph on excel. The load is not a linear relationship to RPM, this is very significant. Below is a graph that shows the load imbalance for a number of flywheels imbalance values.
As an example you can see that the flywheel with an imbalance of 0.4Kg/mm (or 400g/mm) creates a load of 19Kg @ 6500 rpm. It is also important to understand that this is not a constant load; it oscillates at the same frequency as engine RPM (108Hz in this case) because the crankshaft is horizontal and therefore the load is influenced by gravity at the lowest part of the rotation. In this example the crank experiences a peak load with 19Kg, 108 times a second.
From the graph you can clearly appreciate that high imbalance and high RPM greatly increase the load. The load imparted by the flywheel also deflects the crankshaft which has further consequences that are significantly influenced by flywheel weight.
Crank deflection load
This was a real surprise, just how little load is required to deflect a Lambretta crank shaft. I duplicated this test in 4 cranks and they were all the same. MEC, GT, Indian and PM. Please try it yourself.
I made two tests, one on the crank jig with a fixed load hung off the taper end and clocked with the load applied and unloaded (see pics below).
There is a case that the bearings provide support to crank so a deflection test was performed with an Indian crank, with new bearings, in a casing (the deflection was reduced but still reasonably linear).
The load and deflection were measure on the threaded end of the taper but the readings were proportionally reduced by 25% to reflect the load and deflection occurring 20mm further inboard than the measurement point. This makes the loads and deflections representative and not exaggerated by leverage. You can see this by comparing the displacement of 0.01mm @ 1.8Kg in the graph below with the same load showing a deflection in the picture above of 0.025mm (The graph is using corrected values).
The key thing is that more load = more deflection. More deflection = more imbalance load (especially if you have a heavy flywheel).
Effects of heavy flywheel
If a flywheel has a very low imbalance measurement then the effect of its weight is greatly reduced (see the Imbalance load graph example 0.01Kg/mm). For example, during this work I measured an original Ducatti 3.2Kg electronic flywheel. Its imbalance was 9g/mm, staggeringly low. Clearly Innocenti were precisely balancing flywheels over half a century ago.
Should crankshaft deflection occur, an additional imbalance is created and this increases the overall imbalance load. Below is graph to calculate additional imbalance if a flywheel is running off centre (either through deflection or bearing runout). For example If a 3Kg flywheel is running off centre by 0.025mm (1 thou “) the additional imbalance is 0.08Kg/mm (or 80g/mm).
You can now appreciate the relationship between Load imbalance, crankshaft deflection and flywheel weight.
Bearing run out
Worn or second hand bearing have a big influence of crankshaft load. Even if the flywheel is perfectly balanced the use of a worn out bearing immediately adds runout which can be considered as flywheel offset. The geometric relationship between drive bearing, mag bearing and the CoG of the flywheel is approximately 1:1. Example, a 0.02mm drive bearing runout would add 0.02mm offset to the flywheel. If both the mag and drive bearings had 0.02mm run out the two values are added to give an overall flywheel run out of 0.04mm. We could then calculate the additional load and deflection of the crankshaft through the RPM range plus the influence of bearing runout.
Compound effects on balance, deflection and bearing run out
Below is a graph where the initial flywheel imbalance of 0.048 g/mm is plotted against RPM (red). The green plot is the initial flywheel load plus the additional load created from the crankshaft deflection.
The purple line is the additional bearing runout load based on 0.05mm runout of a 2.67Kg flywheel. What started out as a comparatively low 4Kg load at 6,500 rpm ends up at 12Kg. The worn bearing is tripling the crank load!
The blue line at the bottom it the original Ducatti electronic flywheel with 9g/mm imbalance. If I run the additional curves for crank load deflection they virtually superimpose each other because the values are so low.
The typical Indian mid weight flywheel is between 0.248 – 0.277Kg/mm imbalance. Between 7 and 8k rpm they have between 20-30Kg load before crank deflection is added. It is also worth noting that the balance holes drilled in the Indian flywheels bore no resemblance to where the imbalance was.
Collected data and notes
Andrews Precision advised that the flywheel weight should be 25% of the crank weight or less. In the majority of cases the Lambretta flywheels are considerably heavier than the entire crank and it begs the question how this could have come about (after all Innocenti were no fools).
There are examples of Lambretta flywheels being much lighter, for example LD’s. LD’s also have an advancing cam which allows the ignition to retard for starting and tick-over. It is possible that with the introduction of the Li engine came with an aggressive cost down exercise where the engine design was optimised for fast production and low cost. Perhaps in this process the cams were ditched because of cost and replaced with a fixed ignition which relied on an increased inertial mass for tick-over. Innocenti may have mitigated reliability concerns by precisely balancing the flywheels. I’m not aware of any documentation to confirm this; it’s a hunch.
Collated data from testing to date.
Importantly, all of these flywheels can be re-balanced to less than 10g/mm. This will resolve reliability issues such as crank twisting, splaying and spurious ignition timing problems such as miss-fires at high revs, unexpected contact between stator/pickup and magnets, curious timing issues.
During this process I’ve revisited several other crank and flywheel combinations with regard to the ratio between flywheel and crank weight. CR250/YZ250 flywheels are approximately 18-20% of crankshaft weight. RS125 flywheels are approximately 25% of the crankshaft weight and many of the BSSO racers are using much lighter flywheels and or internal rotors. One particular champion has always been lightening dynamically balancing his flywheel. The general guidance from alignment with other manufacturer’s product is that performance engines must use lighter, balanced flywheels in order to increase reliability and prevent crankshaft damage.
It’s been recognised for many years that the quality of Indian parts has got exponentially worse. Indian flywheels seem to have escaped scrutiny until now, but I’d put this down to the general practice of assumption and the ability/process for checking not being widely known or practiced.
It is also worth mentioning that the calculation for imbalance and deflection would end up as a circular argument, forever one process increasing the other to the point of failure. Clearly this does not happen which is why I only applied one level of crank deflection as a best case. The reasoning behind using one level of crankshaft deflection is that there must be some level of specific frequency in the crank web and pin assembly which prevent the effects from being immediate. For example, if specific frequency is achieved there may be some rapid damage which is not evident from reving the engine through. To try and investigate and substantiate a piece of work like that would be hugely expensive and ultimately require a lot of high speed data acquisition and measurement from many running engines of exactly the same build, something that just does not exist in our market.
I used the company Andrews Precision Balancing in Bromsgrove. Suitable testing fixtures for GP tapers had to be made for their equipment (relatively low cost fixtures £100 or so). They charge £30 set up and £25 to balance. Money well spent IMO but please look for a local supplier so perhaps you can form a relationship with them so you can develop and confirm your own opinions.
The learnings I’ve taken out of these tests are quite simple.
1. Personally, I’ll be dynamically balancing every flywheel, regardless of make. The cost is quite small, typically a £30 set up and £25 to balance an individual flywheel.
2. I’ll always be using the lightest flywheel available because the comparative loads are much smaller should any imbalance occur
3. Tick-over with a light flywheel is achieved with an ignition that retards at start-up and tickover. My preference has always been RS125 conversions for the last 8 years or so but there are others, Kheper, LTH over-rev and the soon to arrive SIP.
4. Ignitions timing boxes that only retard are not good as they require inertia mass to throw the crank over TDC with increased advance at low RPM.
5. Even modern production flywheels had quite large imbalance (50 – 140 g/mm). IMO everyone is worth checking, the presence of balancing holes means nothing especially on Indian products.
6. Indian flywheels, especially those purchased on Ebay or other various routes from India have been massively out of balance 240 – 690 g/mm. The tapers have also been distorted by welding to the boss. It is essential they are balanced.
7. These findings are not unique to GT cranks; they apply to all 40.5mm wide cranks. I tested GT, PM, Indian and MEC, the effects are the same.