tained that takes into account the state of the system and, based on the latest information at each time step, suggests how the operation of the assets should be optimally adjusted while keeping the numerous constraints fulfilled. Furthermore, by setting appropriate constraints in the MLD model, the ex tent lo which plan changes are al\owed can be specified. This is a significant improvement over the current approaches, because the (closed-\oop) plan is reactive and is thus a va\uable help in facing changes. A practical example The new methodology is illustrated in2 with a production pro cess !hat requires both electrical and thermal energy. Electricity can either be generation on site by a combined-cyc\e power planı, or can be purchased from the power grid. Steam is produced only when the power generation planı is running. The production consists of five steps: primary process, intermediate storage, secondary process, blending, and final stor age. The primary process requires e\ectricity and steam. This process, which can be run on two different units (Process A, Units 1 and 2), creates a first intermediate product. This is accumulated in intermediate storage. A second type of pro cess, which does not need steam, is run in paralle\ (Process B). Both intermediate products are mixed in a blending stage. The end product is stored and finally delivered according to demand. Several constraints rule the operation of the produc tion units. Starting up an idle unit incurs extra cost. When a unit is operating, the production rate cannot drop be\ow a given minimum bound. lf it is shut down, it must remain idle for a given minimum period. The intermediate and the final (limited) storage volumes must be managed in such a way that given mixing proportions as wel\ as orders are respected. The power generation planı must also fulfill several operational constraints. Startup times, minimum up-and down times, \ower and upper bounds on generation, as well as the time-dependent price of e\ectricity ali contribute to the com plexity of the optimization problem. in !his problem the decision variables, as shown in 2 are: 1.The amount of e\ectricity to be generated by the loca\ power planı; 2.The amount of steam to be produced; 3.The amount of electricity to be bought from or sold to the power grid; 4.and 5) The operation parameters of both primary process Units; 6.The operation of the secondary process. Delivery of the final product is dictated by the demand (orders). By optimizing the overall objective function, the besi com- =� Hoat/Powor plant ı .a Sleam Pr��ı :� A 1 • li Procoss A Urıit2 . l!A ProceH B Unl t1 lntormod1aıosıoragc Costs Blondıng .Ro,onoo,■ Fınal producı Objectı vo sıorago function Logond' • �a����ı�� ARTICLE / MAKALE The performance of a production system is influenced not only by its controlled (hence: certain) variables, but also by unmeasurable (hence: uncertain) perturbations such as changes i n machine condition, or i nput product quality. bined operation of the assets (3 to 5) is obtained. Ali plots correspond to a receding horizon of one week. in 3, the upper plot represents the given price of grid electricity. The second plot shows the power planı generation derived by the optimization procedure. The exchange of e\ectricity can be seen on the lower plot-negative values mean that e\ectricity is sold to the grid. it is interesting to note !hat the power plant is shut down only during prolongated periods of \ow price. Converse\y, maximum power is generated during high price peri ods and the excess is sold to maximize revenues. 4 displays the planned operation of the process units. Be cause it has lower production costs, Unit 1 is run more frequently !han Unit 2. Note: Because both units require steam, !hey are active only when the local power planı is in operation (ie, when thermal energy is available). 5 shows how the inventory \eve\ of the end product varies over lime (middle plot). We can see !hat the optimization exploits the period of \ow energy price to increase the storage. The \ast plot represents the demand for the end product. Hedging against uncertainty On which issue of practical relevance shou\d research pursue its efforts? One of the major effects of market liberalization is that industry is increasingly facing uncertainty. it affects vari ous aspects of the supply chain: supp\iers, commodity prices, quality, demand, financial market, and other parameters. This is leading to a shift from deterministic planning towards opo 20 40 60 80 100 120 140 160 ı 101YP11 1 1 pttr�r ı ı 111 111 t ı ı O 20 40 60 80 100 120 140 160 f :;! bf ;.:f '"; F'"f '.'irf 1 :;r {>:) ,f' I � O 20 40 60 80 100 120 140 160 Time (h) .A. ENERJİ & KOJENERASYON DÜNYASI -EKIM 2005 "AB'ye Giriş Sürecinde Türkiye'de Kojenerasyon-Yeni Gelişmeler" 'V' ========----_:::'..:'..=��'.:'.:'.:'.�::'.'.:'.:'.�����5 -9 -
RkJQdWJsaXNoZXIy MTcyMTY=