Latest Contributions by HeatherMoe

‎02-26-2020

Hi Daniel, A few questions about the set up of your problem. Are you using order pairs? The excess_transit_factor being set to High will only be relevant if using order pairs.If you have order pairs it might be passing a location because it has not picked up the first in the pair. By saying equal time windows do you mean that all orders have exactly the same time windows (such as 8:00 am to 5:00 pm)? If this is the case you should be able to remove the time windows from the orders and put the constraints into the Route using the earliest and latest start time parameters to say the beginning of the time window and the max total time to indicate the ending of the time window. I suggest this because taking time windows out of the orders (even with them all the same) indicates to our solver that it can go through the algorithm steps for a non time window case which is an easier problem to solve. This might improve the results some. If instead you mean that the time windows are of equal duration such as 2 hour time windows then maybe the time window hasn't been reached and so it is best to pass by the order and get it later instead of waiting. Thanks, Heather Moe

‎02-26-2020

Hello, Order pairs is how we have mostly seen school bus routing accomplished in the past. I would have expected with the order pairs that all that are on the bus at the time it reaches the location of drop-off that everything would be "delivered" or bus emptied. A few thoughts on this. Do you have an order pair set up for each student, the pickup location at home (or near home) and the drop-off location at the school? This will have several locations overlapping at the school. Do you have any time windows on the second order in the order pair? If an order has a time window after when the bus arrives there the first time it would stay on the vehicle. Thanks, Heather Moe

‎02-26-2020

Hello, Thank you very much for your interest. We are planning a demo on school bus routing and will work to include as many of your topic points as we can. Thanks, Heather

‎01-23-2020

Hi Ihab, The CostPerUnitDistance and CostPerUnitTime values are used by the VRP solver during its solving both with multiple routes or just a single route in the problem. For the single route case it will be used in the sequencing of orders and for the multi route case it is used in both deciding which route the order is assigned to as well as the sequencing of the routes. Yes increasing the CostPerUnitDistance above the CostPerUnitTime will influence the solver towards options with lower distances if possible considering the other constraints in the problem such as time windows and maximum route constraints. Hope this helps, Heather

‎11-19-2019

Hello, The only other route limiting parameter that Plan Routes has is the total time for the route. You might be able to limit things using that as an approximation but that would probably take a lot of trial and error to get a distance constrained problem. The better option is to switch to using the full Vehicle Routing Problem solver. This allows the routes to be constrained by distance directly. You can still access this through your ArcGIS Online account but instead of using the Plan Routes tool, you make a REST Request. This tutorial should help you get started with it. Route multiple vehicles | ArcGIS for Developers

‎08-23-2019

Hi Rong, For on demand type work flows, I would recommend getting an initial solve of the problem from what you know at the start of the time frame. Then as new orders come in (depending on the frequency of this you might wait until a set number have come in or a set interval of time has passed) you resolve the problem. The depot locations can be changed to the vehicles current location, the orders already completed can be removed from the problem, and if orders need to stay on a specific route because say the product for it is already on the vehicle then you can change the assignment rule for the order to preserve route. The next solve will then try to optimize with the new state of the problem. Hope this helps, Heather

‎08-05-2019

There are several reasons a solve could be infeasible. Could you give some details about what you are trying to model and the parameters you have set?

‎07-30-2019

If running as a VRP layer in ArcMap the Orders attribute table will tell you the reason for not routing the orders. Look for the field "ViolatedConstraints." If using the VRP GP tool or service look at the Unassigned Stops table for that same information. It will be written as an index but those can be referenced here. If I were to make a blind guess on what is limiting these orders from being routed it would be the MaxOrderCount field on the Routes. This is by default set to 30 as it is a required field for the solver.

‎07-30-2019

It looks like the problem you are solving has a pretty high density of orders (multiple on a street edge and most edges). The VRP solver has been designed for a more spread out use case so for helping this problem solve better the recommendation is to consolidate the orders on a street edge to a single order to help spread out the problem. The curb approach and global u-turn policy can then be used to have the routes travel that whole edge without turning around. You can find more information on this here from the material on a high density routing presentation. Of note is this is presented on 10.7.1 where several improvements have been made to the solver to help with the solution and more are planned for coming releases. The concepts should help with 10.4.1, still.

‎04-04-2019

Hi João, There are two Network Analyst patches for 10.7. I believe since you are using both VRP and StreetMap Premium you will need both of them. Here is the link to the patches. Please install these and see if it resolves your issue. If not could you give a little more detail of what your VRP problem set up looks like so I can try to troubleshoot with you. Thanks, Heather

‎03-12-2019

Hi Rafael, Thanks for sharing your use case. Looks interesting. To accomplish the Courier/Rideshare model you would use order pairs. Here is our basic VRP documentation. If you search for "Order Pairs class" you will move to the part of the doc that is about the Order Pairs to get some more information on them. It is pretty far down the page. There is also a tutorial for setting up a problem with Order Pairs. This will walk you through the steps and provide a bit more discussion on different settings you can adjust. Exercise 8: Finding best routes to service paired orders. Hope this helps and reach out if you have some more questions. Heather

‎03-11-2019

Hello Denise, Although you point out VRP isn't needed for identifying which inspector is go to each facility, it has some added features that I think would be helpful for your problem. You can run the VRP problem with just a single route and then it is essentially just optimizing the single route's sequence like the route solver. By doing this, within the VRP solver you can specify on the Routes feature class that the route is only to visit x number of facilities by using the MaxOrderCount field. You could also specify that the MaxTotalTime is 8 hours and restrict the total length of the workday. Another option that you might consider if you have 100 facilities that a single inspector is to visit for the month. You could run this whole problem as a single VRP model and better optimize the distribution of facilities being visited throughout the entire month. To do this simply put a route in for each day the inspector is going to work. If some of the facilities have specific day of the week type constraints on top of the time window they are available this can be modeled with the specialties. If this is something you are interested in you could watch the second demo presented in this video on optimizing routes for two drivers for a week within one problem. This demo starts at 13:25 on the video. Hope this helps, Heather Moe

‎03-04-2019

After the routes have been created the solver has a post processing routine that attempts to move them to cheaper routes where possible. However, there is a known issue for this post processing routine where it does not happen if there are breaks on the routes. For now the work around option is to remove the breaks if the routes can be adjusted time wise because of not having time windows, reduce the max total time by the time of the break, and then add in a fixed cost that shows the difference in capacity making the smaller capacity trucks cheaper. The solver is looking to make routes cheaper so there needs to be a cost function way of knowing that smaller capacity vehicles are preferred so a higher capacity percentage is reached. Alternatively, since you know that the routes have the same characteristics other then the capacity that you can let the solver come up with the optimized assignment of orders to a route and sequencing and then adjust which vehicle is actually assigned to it afterwards. Hope this helps

‎03-04-2019

Hi Deivid, For the different routes that you have in the problem are there other constraints that are different between them? Starting/Ending Depot, max total time/travel time/distance, max order count, etc? Also do you have breaks set up for the routes?

‎03-04-2019

Hi Deivid, For the different routes that you have in the problem are there other constraints that are different between them? Starting/Ending Depot, max total time/travel time/distance, max order count, etc? Also do you have breaks set up for the routes?

Online Status	Offline
Date Last Visited	a week ago

My Ideas

Latest Contributions by HeatherMoe

Re: Inconsistency sequence of order visit.VRP

Re: VRP: Order Pairs to One Location - Vehicle Not Emptied on Visit

Re: Topic Suggestions for Fleet Routing with the VRP Solver

Re: VRP Costs for a single route problem

Re: Route planning with distance constraint in ArcGIS Online

Re: Topic Suggestions for Fleet Routing with the VRP Solver

Re: How to force the VRP solver to assign all orders

Re: VRP extension does not allocate all orders

Re: VRP returns to the same road

Re: VRP error

Re: Topic Suggestions for Fleet Routing with the VRP Solver

Re: Routing for an 8-hour workday

Re: How can occupation of route capacity be improved?

Re: What VRP problem types are you modeling?

Re: How can occupation of route capacity be improved?

Re: Setting order for each route in vehicle routin...

Re: Set VRP cluster using overrides parameter

Re: CumulDistance "null" although Distance Attribu...

Re: VRP Solver - 030092: failed due to invalid inp...

Re: Someone expert for network analysts!