If you are using past inventory usage data on a sample of six items to calculate safety stock levels, then I would suggest a couple of possible issues you will need to consider:

• Railway inventory will cover a diverse range of stock items from track, sleepers (which may be wooden or concrete), electrical signalling, electronic components, rolling stock wheels, axles, brake pads etc etc etc. Each will require replenishment on very different cycles due to varying wear and tear. So a sample of six items will not be enough.
• Rather than trying to use mathematical estimation based on limited historic data, I would suggest a qualitative approach based on the knowledge of the various engineers in the railway on the expected useful lives for each major component of inventory.
• All of the above will be affected by timetable changes which will result in differences to the future rate of wear & tear. So that will also need to be taken into account when calculating safety stock levels.

Thank you, Christopher C Kelly for your insightfull response.

i forgot to tell the audience that im only focusing on the Rolling Stock parts.

beforehand i have ever consider using qualitative approach to categorize part into three classes (Cirtical, Needed, and Others) by taking part : Exchange rate, Complexity of the Exchange process, Cost, Lead time, Storage spec condition (as its factors). im hesistant because i thought it will took very long time since they (Rolling stock item) are a lots . still, thank you now im considering the expected useful lives in my project.

i agree with you that i have to only consider the major components. therefore im using ABC method in order to determine the most valuable component in the inventory . am i rigth at this point?

thankyou, fortunately me and my intern team have ever developed excell (xls) model for future rate of wear & tear of Railway part. now im considering using it again.

lastly, do you have any perspective how to include probabilistic (in our cases the uncertain wear and tear or the lead time replenishment) aspect into my research besides using heuristic apporach that im not very fammiliar with? :)

Thanks.

A couple of thoughts there Muhammad Irfan Kemal

https://www.railjournal.com/regions/australia-nz/excessive-wheel-wear-cripples-v-line-passenger-services/

Article Prioritizing risks for the future using Monte Carlo simulati...

By the way if you find the article useful please give it a 'recommendation' to improve its circulation value here on Researchgate.

All the best with it.

1. determine the rule of thumb for categorizing inventory model/system

Categorizing as continuous or periodic is up the reviewing system in the company. If they can monitor the stock level continuously, then it should be continous review and you will have a demand rate. But, if they review the stock only at certain times and have a discrete demand, then it is periodic. For the continuous review formulation, this is easy to apply some analytic application. Taking the derivative of the objective function will tell you so much. And if the cost is convex, you can get the optimal by making the derivate zero. Periodic review on the other hand, might be a little bit difficult. You may need to use dynamic programming (backward induction).

2. determine the mathematical ROP function

This is up to whether you have a continuous or periodic system. You need to check whether demand is stationary (doesnt change by time) or not. The ROP is the stock that satisfy the lead time demand to minimize the expected cost during lead time.

3. any idea what heursitic for in this context?

In case if you cannot take the derivative of your cost function, you may use enumeration methods (exhaustive search) since there are not too many variables for the continuous review inventory problem.

You should check the followings,

1- Check the demand structure? Does it fit to any distribution? Is it stationary or changing by time?

2- Then you need to decide on continuous review or periodic?

3- Draw the inventory-time plot for continuous review and try to write the objective function

If you go with periodic, start looking at backward induction algorithm.

4- After you write the cost formulation, then you can choose the appropriate mathematical method to solve the problem. You may not need complex mathematical methods or heuristics.

You can check the following article for exhaustive search algorithm,

Article Joint inventory and distribution strategy for online sales w...

Best of luck