Eugeniusz Toczyłowski

A new pricing scheme for a multi-period pool-based electricity auction

A new pricing scheme is proposed for determining the social welfare distribution in a centralized pool-based auction in the context of solving the unit commitment problems under competition. A significant contribution of this paper over previous publications on this subject is the inclusion of the price-responsive demand side for the multi-period auctions with dynamic commitment characteristics. The model allows every thermal unit and every consumer to obtain individual maximum profits, and at the same time it gives the market coordinator an adequate tool for solving the ensuing technologically constrained unit commitment problem with fair market clearing. The pricing model is in the form of a mixed linear programming model that minimizes the sum of the compensation costs. The accompanying case study illustrates the approach proposed.

Electronic Trading on Electricity Markets within a Multi-agent Framework

Specific features of trade on electrical energy markets make the automation of certain negotiation and contracting processes attractive and desirable. Multi-agent systems are a natural solution for automating processes in a distributed environment, where individual market entities posses their own objectives. In the paper we consider a concept of a multi-agent framework for an electronic market trade. Since both energy markets and trade mechanisms are complex and permanently evolving, we require the highest expressiveness from this concept. Development of such a system involves solving several design issues related to embedding the agents in the environment, agent communications schemas, language, and offers modeling, especially in the case of the distributed, multilateral trade. We present a solution for the most important design issues. Our concept, including new information technologies supplementing multi-agent systems and Web services, are capable to support enterprises in the decision processes, to facilitate the buy/sell offers preparation, to select parties and enter into business relations with entities, and to support contract negotiations.

Profit-based FMS dynamic part type selection over time for mid-term production planning

The dynamic part type selection problem for mid-term scheduling to maximize profit over time in a general flexible manufacturing system is considered. Some mathematical programming models are developed to address this problem and their method of solution is based on a column generation technique. In the solutions to these models, the production plan is represented as a sequence of steady-state, periodical, cyclic schedules. Each cyclic schedule allows a subset of parts of different types to be released periodically in appropriate production ratios. Parts can be produced according to these ratios and cycles until the production requirements for some part type are completed. Then new part types can enter production and new ratios and cycles can easily be found. A two-level procedure is developed. At the upper level, a large-scale linear programming Master Problem is solved, the columns of which are generated by solving a nonlinear, mixed-integer, profit-based part type selection subproblem that selects part types for simultaneous production over a period. Numerical experiments demonstrate that this model is computationally tractable to solve problems of practical size and over several periods of time.

On Multiple Criteria Decision Support for Suppliers on the Competitive Electric Power Market

As an active participant of a competitive energy market, the generator (the energy supplier) challenges new management decisions being exposed to the financial risk environment. There is a strong need for the decision support models and tools for energy market participants. This paper shows that the stochastic short-term planning model can be effectively used as a key analytical tool within the decision support process for relatively small energy suppliers (price-takers). A self-scheduling method for the thermal units on the energy market is addressed. A schedule acquired for given preferences can be used as a desired pattern for bidding process. The uncertainty of the market prices is modeled by a set of possible scenarios with assigned probabilities. Several risk criteria are introduced leading to a multiple criteria optimization problem. The risk criteria are well appealing and easily computable (by means of linear programming) but they meet the formal risk aversion standards. The aspiration/reservation based interactive analysis applied to the multiple criteria problem allows us to find an efficient solution (generation scheme) well adjusted to the generator preferences (risk attitude).

Detailed scheduling of batch production in a cell with parallel facilities and common renewable resources

Abstract The problem of scheduling production in a cell consisting of several parallel facilities, which arises in many process industries, is addressed. The cell produces a variety of items in required quantities. The production of an item requires an operation, which may be performed on one of a subset of the parallel facilities, and it consumes some base products at given rates. There are constrained resources and structural constraints. Resetting a facility to change from one operation to another may necessitate time-consuming changeover. The problem is solved first by assuming that changeover times are negligible. This leads to forming complete schedules by combining in sequence a number of elementary production schedules (EPS), such that an objective function related to production cost is minimized. During each EPS, and therefore throughout the schedule, all constraints are respected. Subsequently, the EPSs are sequenced to minimize the number of changeovers among operations. Finally, all operations, including changeovers, are scheduled such that the makespan is minimized, while preserving precedence relationships among the operations such that the feasibility of the overall schedule is maintained. The paper presents the overall approach, the models constructed for each sub-problem and the algorithms developed to solve these models. Illustrative examples and results of computational testing are given.

A flexible auction model for virtual private networks

We consider the resource allocation problem related to Virtual Private Networks (VPNs). VPN provides connections between geographically dispersed endpoints over a shared network. To realize VPN service, sufficient amount of bandwidth of network resources must be reserved for any traffic demand specified by a customer. We assume that there are many customers that want to purchase VPN services, and many network providers that offer their network resources for sale. We present a multicommodity auction model that enables the efficient management of the network resources in the market environment. On the other hand it is very convenient for the customers as it allows them to specify the bandwidth requirements concerning VPN in a very flexible way, including pipe and hose VPN representations. The proposed model has also many other valuable properties, such as individual rationality and budget balance. The auction model has a form of LP for which the computational efficiency can be improved by applying the column generation technique.

M3 Applications on the Electricity Markets

The electrical energy markets illustrate very well complexity of the infrastructure markets. Achieving the market balance, while maintaining its high efficiency, is a challenging task which requires both a rich commodities structure and a complex structure of the market processes. Market balancing is being obtained by multi-step processes of the multi-commodity trade. In this chapter we show that the fulfillment of various specific electricity market requirements is possible by applying model M3 . We focus on the balancing segment of the market and present relations and mappings between various concepts of a real balancing market and notions of M3 . We discuss some potential ways of market development to demonstrate how M3 notions facilitate modeling of the new solutions.

M3 – Motivations and Formal Model

In the chapter, first we discuss needs and motivations for market-based mechanism for complex, distributed systems, especially focusing on infrastructures sectors. A great number of issues related to market mechanisms developing is modeled by decisions models. There are many completely different attempts to organizing the mechanisms of finding market equilibrium. These attempts are hard to reliably compere on the common test cases due to huge effort needed for numerical data adaptation to each approach. We give strong foundations for Multicommodity Market Model (M3), which describes the data on the input and output sides of the market clearing process. Its main advantage is very high expressiveness. M3 covers most known trade mechanisms, including different auction types, multi-commodity mechanisms, bilateral contracts and others. In the chapter we present architecture of M3 , including formal model.

Multicommodity exchange model for trading bandwidth in undirected networks

The multicommodity exchange for trading bandwidth commodities in undirected networks, in the form of the BUCBT bandwidth exchange model, is presented. The BUCBT model allows for the most efficient allocations of undirected resources. An associated allocation model is formulated as a multicommodity network flow problem with directed flows. The equivalence of the associated model to the undirected bandwidth problem is shown.

Preemptive Jobs Scheduling on Parallel Machines with Setup Times and Renewable Resources

A two-stage algorithm for scheduling preemptive jobs on parallel machines with minimum makespan criterion and requirements for limited renewable resources and existence of sequence dependent setup times is investigated. In the first stage of the algorithm a set of best elementary feasible plans is obtained through column generation. For the second stage we compare genetic algorithms for sequencing elementary plans, where various approximate criterions for calculation minimum makespan are used.