World Records

Optimization is very complex and requires advanced algorithms. To ensure the quality of the optimization software it has been developed international, standardized test cases. In 2011 we analyzed 400 such test cases *. We beat the top score of 189 cases and 111 cases had a touch / near tangency. We had 100 cases just behind the best result ever achieved.

 This means that we have outperformed the overall environment of researchers and competitors worldwide. This is unique for a commercial product, and we are very proud of these results. Testing verifies that we are in the lead of optimization quality.


  We have no competitors who achieve these results.
  We like to run comparison with any competitors.

More detailed information on the results can be provided on request. Contact us

 


Why is this important?

Because different optimization products gives results of varying quality, and SPIDER performs well. Documented.


Let us consider a simple example to illustrate the complexity of optimization. With only one tour and one set of 3 customers to be visited, there are six possible routes to visit the customers. With 4 clients, there are 24 possible routes. With five customers, there 120 possible routes. Already by 10 customers there are well over three million possible routes.

With 20 customers, there are over 2 billion of billion different possible routes!
 

Number of customers in one tour

Number of possible routes.

3

6

4

24

5

120

6

720

10

3.628.800

20

2.432.902.008.176.640.000

50

304.140.932.017.133.780.436.126.081.660.650.000.000.000.000.000.000

 

This shows that manual route planning is difficult, but it also shows that automatic optimization is challenging for large cases with many factors and requirements that needs to be taken into account. SPIDER makes it particularly well in the standardized test cases implying increased control and efficiency for our customers.
 

What are the practical implications? 

It is difficult to give any definitive answer. However, for simple cases 5% better optimization means that you can save two cars rather than one vehicle in a case with 20 cars. I.e. that optimization reduce the number of trucks from 20 to 18, instead of 20 to 19.

In cases that are more complex the difference between the best and medium optimization systems will be even larger. (See below.)

Complex cases

Very good basic optimization ensures SPIDER-users efficient optimization for complex cases. Examples include:
• Different types of cars
• Linked multiple trips
• Pause Handling
• Frequency Orders

In complex cases, the difference between two optimization products will be significantly more than 5%. For complex cases, it is not unreasonable to find differences between providers of 10%, in some cases even more. It provides substantial differences in saved drive time and number of vehicles utilized.

Background

Bjørn Sigurd B. Johansen is responsible for the development of algorithms and optimization in SPIDER Solutions. He has worked as a researcher at SINTEF Applied Mathematics, which is among the leading research groups in the world in this type of optimization. Johansen has in 20 years worked in both academic and commercial companies to create the best possible optimization. Optimization which takes into account all important factors in "real-world problems."  

 

* The used test cases are well-known and well-established test cases from Gehring and Homberger. The cases includes:

• Both tight and more flexible time windows on trips and orders
• Varying capacity of vehicle and quantities on orders
• Orders that are randomly dispersed, and orders which is more grouped in industrial areas
• Varying number of trips per case (from 4 to 100) and varying number of order pr. case (from 200 to 1000)

More info on: http://www.sintef.no/Projectweb/TOP/VRPTW/Homberger-benchmark/