So what is VRP, exactly? It stands for the vehicle routing problem, and it’s the three-letter answer to a question every delivery operation asks itself before sunrise: which vehicle should visit which stops, and in what order? I’ve spent a good chunk of my career as an operations research analyst building routing models for real fleets, and I still think VRP is the most practical piece of math most people have never heard of. If a truck brought you anything this week, a VRP solver (or a dispatcher sweating over a map) decided how it got there.
The vehicle routing problem, in plain English
You have a depot, a fleet of vehicles, and a list of stops that need visiting. Each vehicle can only carry so much, drivers can only work so long, and some customers will only take a delivery between, say, 8 and 11 in the morning. The vehicle routing problem asks: what set of routes serves every stop at the lowest total cost? That’s it. The definition fits in a sentence; the solving is where careers go to hide.
VRP is the bigger, messier sibling of the traveling salesman problem. The TSP asks for the shortest tour for one traveler; VRP splits the work across a whole fleet while juggling capacities, time windows, and driver limits. The formal version goes back to George Dantzig and John Ramser, who wrote it up in 1959 as the “truck dispatching problem” for gasoline deliveries. The math has evolved a lot since; the trucks, less so.
Why it’s so hard (the honest math)
The difficulty is pure combinatorics. A single route with 10 stops can be ordered 10! ways — that’s 3,628,800 possibilities for one vehicle. Add a second vehicle and you’re not doubling the options, you’re also choosing which stops belong to which truck before ordering them. By the time you’re at a realistic size — my free route optimizer ships with a 113-stop, 10-vehicle sample dataset — the number of possible plans exceeds the number of atoms in the observable universe by a margin that stops being funny.
Computer scientists classify VRP as NP-hard, which in practice means nobody has a method that’s guaranteed to find the single best answer quickly for large instances. (This is also why Google Maps politely caps you at 10 stops — it’s not being lazy, it’s avoiding a fight it can’t win.) The industry answer is to stop demanding perfection and get very good at finding excellent answers fast.
The flavors you’ll actually run into
Textbooks list dozens of variants, but four cover most of real life. CVRP adds vehicle capacities — the truck fills up. VRP-TW adds time windows — the loading dock closes at noon. Pickup and delivery means goods move between locations, not just outward from a depot (I wrote up how that changes the model in my post on pickup and delivery route optimization). And multi-depot lets routes start from more than one warehouse. Real operations are usually a stew of several at once, which is why off-the-shelf spreadsheet tricks run out of road quickly.
How it actually gets solved
For small instances you can write VRP as a mixed-integer program and let an exact solver prove optimality. Past a few dozen stops, exact methods take longer than the workday you’re trying to plan, so practical solvers switch to metaheuristics: construct a decent solution, then improve it relentlessly with local search moves — swap two stops, reverse a segment, shift a stop to another route — while a strategy layer keeps the search from getting stuck. My own tool uses Google OR-Tools with guided local search, a combination I settled on after years of building these models in other stacks (I’ve explained why I use OR-Tools instead of the Excel Solver at length).
One thing the textbooks undersell: the distance matrix matters as much as the algorithm. Straight-line distances produce routes that look great on paper and terrible on roads. A serious solver feeds on real drive times — mine pulls them from OSRM, the open-source routing engine — because a river between two points that are 500 meters apart will ruin an optimal-looking plan (ask me how I know). If you want to go deeper on the mechanics, Google’s OR-Tools VRP documentation is the best free reference I know.
I’ll admit the field has humbled me more than once. The biggest routing model I ever worked on was part of a $2M logistics project that evaporated when a pandemic hit — years of careful modeling, gone with the client’s budget. The code survived, though, and pieces of it live on in the free tool below. Expensive lesson, cheap consolation.
Quick answers to the questions people actually type
What does VRP stand for?
Vehicle routing problem. In logistics software marketing you’ll also see “route optimization,” which is the same idea wearing a suit: VRP is the mathematical problem, route optimization is the product category built on solving it.
What’s the difference between VRP and TSP?
The traveling salesman problem routes one vehicle through all stops and back; that’s a pure ordering puzzle. VRP routes a fleet, so the solver must first decide which stops go on which vehicle and then order each route — all while respecting capacity, time, and driver-hour limits. Every TSP is a VRP with one uncapacitated vehicle, which is why VRP is at least as hard.
What software solves VRP?
At the heavy end: commercial platforms and solver libraries that companies build into their dispatch systems. In the middle: open-source libraries like Google OR-Tools, which is what powers my tool. And for a spreadsheet-sized operation, a free web tool is honestly all you need — you don’t have to buy a routing platform to stop hand-drawing routes (most small fleets I’ve talked to are still doing exactly that).
Try VRP on your own stops
Definitions only get you so far — the fastest way to understand the vehicle routing problem is to feed one to a solver and watch what it does. My routing optimizer is free, runs a real VRP-TW model with capacities and time windows, and takes a plain Excel workbook or Google Sheet as input. Download the sample, swap in your own stops, and see how many driving minutes the math finds that a map and a marker never would.
Things that I use, like, and am affiliated with:
Mint Mobile offers great cell phone service for $15 flat, get $15 off using the link. Get discounted phones with service activation and no contract.
I never spend money before I check Mr Rebates or Rakuten to get cashbacks, rebates, discounts, coupons or cheaper gift cards.