An agent-based model is used to simulate the passenger car fleet of a country rich with domestic renewable energy. With data on peoples sensitivity for car features and a set of cars the results are given in percentage of electric vehicles (EVs) compared to conventional vehicles for a given number of years. Incentives are then added to increase the shares of EVs in order for less dependence on oil and more environmental friendly transportation. Different policy decision of car purchase price and fuel price are made and the model is run again and compared to previous results. In this study the Icelandic car fleet is simulated and both Icelandic and Danish data are used.

Iceland is a country rich in both geothermal and hydroelectric resources but remains dependent upon foreign gasoline to supply its car fleet. Electric Vehicles (EVs) have been hailed as a cost-efficient solution for Icelanders due to relatively low domestic energy prices, but since EVs are generally more expensive to purchase than traditional cars the average consumer must drive for years before receiving the benefits. The relatively higher list prices (and the fact that all Icelandic cars are imported) indicate that EV implementation today would contribute to a national trade balance deficit. The price difference for EVs would cause government tax income to increase somewhat over a period of an EVs suggested lifetime. Required infrastructure to accommodate the increased energy demands following EV implementation would be minimal.

The study looks at scheduling jobs for a pharmaceutical company inspired into prepared plans. Comparison is made between three methods to assign the jobs to a prepared plan, optimization, greedy algorithm and the current manual assignment in use today at the pharmaceutical company. The goal is to see if the process can be automated. Both optimization and greedy algorithm can be used to solve this problem automatically.

Merchandise promotions are used to introduce new products or boost sales of existing products. Successful promotions can multiply the expected sale of products. They are therefore of great importance to account for the planned promotions when calculating inventory levels in distribution centers or stores at promotion time. The performance of promotions depends on several stochastic factors that can be difficult to predict accurately. It is however possible to analyze past promotions and use the results in planning future promotions. This project develops intelligent methods that can analyze and track promotion performance both during and after promotions to learn the overall affect and utilize for future planning. The methods will form the base for an intelligent expert system that can automatically predict the effect of various types of standard promotions and ensure adequate inventory levels without ending up with excess stocks once the promotion is over.

The objective is to develop intelligent and effective processes for demand planning that helps executives and product managers alike to continually achieve focus, alignment and synchronization among all the functions of their companies. The sales of products or services can be hard to predict over longer time horizons and there are various factors that interplay and make it a challenging task to create reliable demand plans. There are many players on all management levels that need to participate in the planning process and analyze chronological, geographical and product dimensions. The objective is also to define the function of the processes needed to make accurate forecasts and who is responsible for each process.

This project also develop sales forecasting audits to help companies understand the status of its sales forecasting processes and identify ways to improve those processes. This methodology revolves around three district phases: the „as-is“ phase, in which the audit team seeks to understand fully company´s current forecasting process, the „should-be“ phase, in which the audit team presents a vision of what world-class forecasting should look like at the audited company, and the „way-forward“ phase, in which the audit team presents a vision of roadmap of how the company can change its forecasting processes to achieve world-class level.

Today’s modern warehouses are data rich environments with warehouse management systems that log employs every tasks within the warehouse and keep information on orders from customers. The warehouses have to be efficient and responsive in order to minimize costs and fulfil customers expectations. This study analyses the option of using data from warehouse systems to model employees performance and predict how long it will take for them to complete picking the orders. The prediction is used as an input for an optimization model that minimizes orders delays by allocating task to employs in the optimal manner.

In recent years hardware manufacturers have shifted their focus away from CPU speed to putting more than one processing units on the CPU and thus effectively put the pressure of ever increasing speed onto software developers. For many years the choices to run a Meta-heuristic algorithm have been on a single CPU computer, on a computer with many CPUs (a SMP computer) or on a Cluster. Clusters and SMPs are expensive and in many cases are limited by communication bottlenecks, such as Ethernet, if the problems are not fragmented correctly.

The main goal is to find the benefit of using multi core programming vs. not using it for three Meta-heuristic algorithms which are tested and compared both in a single core implementation and in a multi core implementation. The three algorithms tested are simulated annealing, tabu search and ant colony. Each is tested on a different problem, the Ackley function, the traveling salesman problem and vehicle routing problem respectably. In all cases the amount of coding to be done to implement parallelism was negligible but the performance gain was substantial.

Every company that has employees working on irregular schedules must deal with the difficult and time consuming problem of creating feasible schedules for the employees. We are working on an algorithm that takes a partial schedule created by requests from employees and creates feasible schedule where most of the employee's requests are unchanged, while still making sure that rules and regulations are not violated. The algorithm is based on independent modules, which can be executed in any order, and each module tries to emulate some action taken by a staff manager.

Our goal is to create a transparent and fair system that creates feasible schedules of high quality, but also a system where the employees can get an explanation and justification for every change that the algorithm makes to the employee requests.