Los Alamos High School (64)/Interim Report

Problem Definition
Online retail stores such as Amazon use massive warehouses to move, sort, and store goods until they are ordered by a customer, and shipped by companies such as FedEx. However, due to the incredible complexity of these warehouses, even the simplest things, such as a truck being late, can have major impacts on the movement of packages in these facilities. Since the impacts can be so varied and seemingly random, a computer program that could show the results of every possible disruption that could happen in a set facility would have great commercial value.

Problem Solution
We will create a computer program written in Java, using an IDE known as Eclipse, that will visually show the boxes moving along a virtual representation of a real life warehouse. The warehouse layout and specific data related to precisely how the boxes move through the facility, (travel times, employee time spent on each box, etc.), will be provided to us by a Los Alamos National Lab contact. The program will first create a model of how the warehouse operates in the status quo. The results of this run will be compared to real life data to see if we adequately modeled the warehouse. The program will then keep rerunning itself, each time with a different error until every possible error combination has been simulated. It will record data during each rerun and we will use that data to see how the warehouse was affected.

Progress to Date
Thus far a demo engine which may be used as a starting ground for the simulation has been created. Our hope is to finish most if not all of the coding over later December into Janauary, and have the program finished at latest by February 5th. We have completed all of our research related to what business as usual is like in these warehouses, and have even located a few online retail stores centered in New Mexico. Our contact at LANL spent a lot of time working on a project that dealt heavily with warehouses of this kind and has agreed to help us find data on warehouses in New Mexico.

Expected Results
Our program should be able to perfectly replicate a real life facility that is preferably in New Mexico. The program then will then rerun itself changing the parameters of the baseline facility each time. For example, the program will make a truck carrying goods that need to be stored come in an hour late. The magnitude of the change in efficiency that occurs during each run will help us figure out what would effect our simulated warehouse the most.

Interim Comments
My name is Jeff Grantham and I am a graduate student in the Computer Science Department at New Mexico Tech and I'll be providing some feedback on your interim report.

It looks like you have you have a good handle on the model you will use and how to compute it. It would be useful to have the details of what errors you will be simulating as no simulation will cover all possible errors (for example, it likely isn't useful to simulate an asteroid randomly hitting the warehouse) and it will also allow readers to better understand the extent of the project. In this same vein I'd caution you against using the word "perfectly" as you did in "Expect Results" section as simulations are rarely, if ever, perfect and its good keep that in mind so that you can be aware of the exact capabilities and strengths (as well as weaknesses) of a simulation. Additionally it looks like from your references that you are looking to simulate some rather large scale warehouses and it would be useful to define the scale warehouse you are looking to model as different scale warehouses are likely to operate very differently and errors will likely have different effects on them.

You have made some solid progress and hopefully have made some more good progress since you wrote this as the time table you have described seems like it might be a bit tight. It does look like, with your mentor and references, that you have all the material you need to get the project done and have definitely done a good job there.

Finally, I'm supposed to remind you that you will have a face to face evaluation in February.

Good luck and I look forward to seeing your completed project.

Hi Team 64, my name is Christopher Morrison. I am a challenge alumni from the years between 2003-2007. I am currently an senior at Embry Riddle Aeronautical University in Prescott, AZ studying aerospace engineering. I've just been accepted into a PhD program where I will either be involved in nuclear powered rockets or light propulsion (shining lasers at shiny discs to propel them into space). The Supercomputing Challenge really helped me and I know it will help you too. Knowing how to program and problem solve is a great trait that has helped me immensely.

I am reading your project and I believe that understand the gist of what you are attempting to model. You are attempting to model a single very large warehouse and determine the effect that an "error" would have on the operation of the ware house.

I am not sure about the rules behind the warehouse simulation. You mention that the information about warehouses will be provided by someone else? What have you learned?

"The warehouse layout and specific data related to precisely how the boxes move through the facility, (travel times, employee time spent on each box, etc.), will be provided to us by a Los Alamos National Lab contact."

You mentioned that you have been working on the model and plan to get to finish the model by the 5th of February? What are the rules behind your model? You mentioned that you will be modelling the boxes individually as they travel between areas. So, I'm guessing you are using agent based modelling? What are your metrics (the numbers you are keeping track of)? You mention efficiency? What is efficiency? I think you have a great topic, and I want to know how you are tackling the problem.

My advice is to write down the rules for your program very clearly. Make a flow chart and since you are using java make a javadoc for your program ([][] ). When you have your project evaluations in the next few weeks you can really shine if you can easily explain to the judges how your program works and you will be able to make your results more authoritative.

It's awesome you have a mentor/contact at the Labs. Rely on your contact for the information, and try to model the factory off of his description. However, don't get sucked into making your model "perfect." There is no such thing as a perfect model for a problem as complex as this.

I'm guessing you have made progress since your interim report, but I thought I would throw in a few of my brainstorming ides.

I'm guessing packages arrive on a truck, are sorted and stored for some short period of time and then sent away on a truck.

I could see you putting together an agent based model where you have a set number of vehicle loading areas and each loading area can have so many trucks. Each truck has a certain number of packages that it delivers and also can carry a certain number of packages to a location. Trucks arrive at certain times.

Employees process the packages and can only move them at a certain speed.Perhaps you could process errors as employees and trucks coming late, packages being incorrectly sorted, and perhaps packages getting dropped and broken. Your main metric for the package delivery system is simply the number of delivered packages.

Again you have a great topic and I think your project has a good chance to do really well.

-Chris