Skip to main content

How to identify best approach to solve the problem

Answered

Comments

3 comments

  • Ronald van der Velden
    Gurobi Staff Gurobi Staff

    Dear Sudheer,

    Should this problem be approached as a vehicle routing problem (VRP) where the goal is to optimize the routes and truck assignments for deliveries, or should it be modeled as a logistics network design problem

    The answer should usually come from the problem statement. I would first try to figure out which decisions you (or your team) need from your model and what objectives matter. For example: do you only need to know which customer is served from which location (assignment), or do you need to know specific days as well (planning), or does the actual route during each day also matter (scheduling/sequencing/routing)? Only when you have those answers, you're ready to start modeling.

    Additionally, would it be more efficient to solve this problem in two stages: first by assigning customers to DCs, and then by optimizing the delivery routes and truck assignments?

    This is usually a trade-off: if you decompose the problem like you describe, then the second step becomes easier in the sense that you can solve it independently for each DC. At the same time, the question is if you can take the right decision in the first step: what does a "good" assignment look like?

    Kind regards,
    Ronald

     

    0
  • Sudheer Reddy
    Conversationalist
    Gurobi-versary
    Investigator

    Below is the problem statement I am looking to solve

    1. Customer Assignment: Determine which customers are served from which locations.
    2. Delivery Days: Schedule specific delivery days for each customer.
    3. Transportation Mode: Decide between FTL and LTL based on shipment volume and costs.
    4. Cost Segments for LTL: Different costs for different volume segments (< 5m³, 5-15m³, >15-20m³).
    5. Truck Costs for FTL: Which trucks to use on specific day different truck sizes (5m, 8m, 9m) which has different costs
    6. Milk Runs: Incorporate extra costs of milk runs into planning and decide the most cost-effective delivery method and schedule.
      Above are few I am trying to answer from optimization model
      Routing is not priority if we are able to find Milk runs and correct LTL and correct FTL from optimization model 
      @Ronald van der Velden  
    0
  • Ronald van der Velden
    Gurobi Staff Gurobi Staff

    Hi Sudheer,

    Here's a few more thoughts. Unfortunately we cannot go much further with modeling assistance, but other users of this community forum might be happy to jump in.

    • If the number of loads per truck in an LTL scenario is very limited (say 2 or 3), you could perhaps generate all combinations per truck upfront and solve the problem as a single MIP.
    • You can reduce the number of variables further by only combining loads for customers that are located close to each other. Similarly, only consider combinations of DC/RDC and customer(s) that are close enough to each other.
    • You could also ignore routing completely and generate an assignment problem, e.g. what volume does truck X bring to customer Y on day T.

    Good luck!

    Kind regards,
    Ronald

    PS. If you're a commercial Gurobi user already, don't hesitate to open a support ticket for direct access to our Experts. Still there will be a limit to how much modeling support we can provide - ultimately you will need to pick an approach and try that out in practice to see what works and what doesn't.

     

    0

Please sign in to leave a comment.