Either system can be good or bad in terms of durability depending on how well engineered it is.

As a generalisation, PSA design their belt drives very well using common sense detailing such as placing the water pump on the slack side of the belt and ensuring that the number of belt teeth is not wholly divisible by the number of teeth on any significant wheel. Some other manufacturers (VAG for instance) are not so clever.

Much the same can be said of chains, except that over the years chain drives have become much lighter (duplex chains are not now common) and engines generally last longer.

It is this last observation which tips the balance in favour of a good belt drive. When an engine has run to a high mileage and is worth very little, the labour of replacing a chain is disproportionate and the engine will be left to destruction. A considerately designed belt drive system will often be fitted with a new belt.

Belts are lighter and quieter than chains and don't stretch. Chain stretch can be accomodated by a tensioner but the result is a timing shift. An advantage of a chain over a belt is that it reduces the overall length of an engine for a given drive torque capacity - it's narrower and internal shaft oil seals are eliminated. This can be crucial in some installations where a large engine is "shoe horned" into a small chassis.

Vehicle builders know this and take advantage of it - rendering the access in situ to these chains almost impossible as they are not then classed as a maintenance item. These vehicles will be effectively be written off by a worn chain.

The death rattle of a worn chain prior to failure can indeed be a point in its favour - but a surprising number of people will ignore it. A good belt drive will run to at least twice the maker's stipulated change mileage - so the owner usually has himself to blame if failure occurs.

Unfortunately, there are some bad belt installations with poor quality idlers and inappropriate geometry which will often fail prematurely. The belts themselves are durable.

659.

Collos, you seem very certain of yourself.

There's probably not much point in debating things further, but, just to highlight one of the points you make:

"Chains usually give a warning before they break yes about a microsececond."

If they stretch hugely, as you asserted in a previous post, I would expect there to be some very obvious noise long before a chain actually failed.

Edited by FocalPoint on 10/06/2011 at 16:57