I suggest a 2 stage approach:
1) I would start by reviewing all your applications using N2. As suggested above, there should be no need to use N2 on diaphragm pumps, and this is likely to be the biggest cost saving. How many valves etc. do you have operating on N2? This can be used, because it is dry, and usage is generally small, but dry air is most commonly used here, and is more economic for larger sites. Look at your purges as well. Are you being over conservative with the amount of purging done? What about overpressure using N2 for vessel transfer - can this more cheaply and effectively be done with pumps, gravity etc.? Get a list quantifying all N2 usages, identify alternatives, and determine the cost implication of the alternatives against the cost savings, this will show you where to concentrate your efforts.
2) Do you always use liquid Nitrogen in addition to your membrane system, or is it only used in emergencies, membrane shutdown and uncommon very high usages? The costs of liquid N2 are likely to be much higher than the N2 generated on site. If you are constantly using liquid N2, then this will add significantly to your costs. In this case, and after carrying out the above exercise, you might want to look at increasing the capacity of your membrane system, either in terms of peak capacity or storage capability. If this is run under contract (and it sounds as if it is), increasing capacity may be a way in to re-negotiate your contract.
Finally, there is a general perception that N2 is totally safe. It is not. Although it has many usages and advantages in the process industries, it is an asphyxiant, and there are many cases of injury or death from Nitrogen asphyxiation. Usage of N2 should therefore be carefully considered. Use in diaphragm pumps etc. is not necessarily therefore even the safest option.