I work in a refinery, but cannot think of any services that are as severe as you describe.  I am picturing a sample loop taking a slip stream of some product from the process to be analyzed.  We commonly do this with finished products.  But, finished products tend to be cool.  For hot sampling we have a few options.  Our preferred coolant would be cooling water.  But your temperatures are much too hot for that.  Our second choice would be air cooling using a finned tube exchanger with a cooling fan.  But the temperature drop you require would require a very large air cooler.  For some very hot products, we sample in batches and use steam to cool the product initially.  This is done on asphalt in some cases.  A fixed volume of the product is allowed to fill a small pot that is surrounded by steam coils.  An operator monitors the temperature of the product until it drops to the desired sampling temperature then they open another valve and pass this cooled sample on to the sample station.  If your process requires a continuous stream rather than a batch process, you may need to combine systems.  A steam cooler could drop the temperature initially followed by an air or water cooler to get the final temperature required.  I assume any refinery would have a source of low pressure steam and cooling water.  If you could use boiler feed water as the coolant, you might be able to do this in a single cooler.  

This stream is so small that, at least IMHO, you don't need to spend much time looking for other streams for heat integration. That means a closed cooling system. If you find an appropriate coolant for these temperatures, you'd have to cool the coolant afterwards. I'd tend to think no matter what you'd end up with an air cooler somewhere. Why not limit maintenance and operator attention by being lazy and just cool your 20 litres/minute with a couple of finned tubes, if possible integrated in a larger finfan air cooler bay that belongs to the same unit?

(I am assuming that the 100 deg is just the maximum for the analyser and that 80 or 60 would be fine. If not, you'd need a dedicated fan and temp control)

Thanks guys,
helpfull comments and though I'm sure the problem will return in the future, in this case the end user has now decided to install after his main process cooler...  thankfully, I now have an inlet temp of below 100degC and am back to using cooling water.

JMW
www.ViscoAnalyser.com