Investment and operational optimisation of an energy recovery engineering plant
Burnable off-gases generated from operational processes in engineering plants are regularly utilised as energy sources. A common practice is to use this for steam production in boiler houses, where excess steam is allocated to power generation turbines. Fluctuations in off-gas productions may, however, result in turbines shutting down, due to insuficient steam. Some investment models exist, which are typically based on cost minimisations or for the purpose of meeting energy demands. These models do not, however, take into account plant-specific steam flow patterns and typically use average-based pro les for decision making. The operational control of turbines are typically performed by means of a fixed-sequence philosophy. Turbines are loaded in a predetermined order until a fixed set point. This operational philosophy incurs signi cant power generation losses from turbine shut-downs, as a result of the inability to distribute steam dynamically. Such an operating philosophy is easy to implement and, therefore, commonly used in industry. Another philosophy is that of dynamic control where steam distribution between the turbines are computed at each time period by means of an in-time operating control algorithm. In this thesis, a number of novel model formulations are proposed, which addresses optimal power generation and turbine investments under a fixed-sequence philosophy, as well as dynamic control. Seven conceptual formulations are applied to demonstrate basic power generation optimisation. These conceptual formulations are to incorporate either a fixed-sequence philosophy, dynamic control or both. The main contributions of this study entail a further seven formulations where six are optimisation models. Each of these six formulations utilises plant signature steam flow profiles in order to determine, either optimal power generation, or optimal turbine investments in terms of the net present value (NPV), or both. All investment formulations include turbine shut-downs in the decision making process by penalising the NPV with each occurrence. The first three of these model formulations are for turbines operating under a fixed-sequence philosophy, two for optimal power generation and one for optimal turbine investments. For optimal power generation the formulations determine in which fixed order the turbines should operate. Optimal turbine investments, as determined by the third formulation, is the combination that yields the highest NPV. For this combination the optimal fixed turbine order and power generation are determined. Optimal investment results indicate that when future trip costs are taken into account as an NPV penalisation, a single turbine should rather be procured. The final three of these proposed formulations are for turbines operating under dynamic control. The first formulation optimises power generation between any number of turbines. The second formulation optimises turbine investments in terms of NPV. Comparing optimal power generation results, increases between 3.5% and 13.2% are observed for turbines under dynamic control compared to a fixed-sequence philosophy. Optimal NPV's under dynamic control are between 7.3% and 19.3% higher than those of a fixed-sequence philosophy. All optimal outcomes yield that two turbines should be procured. A formulation is proposed that optimises turbine investments, which incorporates the procurement of a supplementary energy resource to assist during low o -gas, and therefore, low steam flow periods. Such a resource is typically very expensive and does not make sense to procure under normal operating conditions. However, in a uctuating steam flow environment it proves to increase the NPV, while safeguarding turbines from shut-down occurrences. Depending on the procurement and projected shut-down costs, results indicate that an investment into a supplementary resource under optimal investments can yield an NPV improvement 10.6% to 118.0% versus a fixed-sequence philosophy, and 3.0% to 82.7% compared to dynamic control. Results further indicate that involuntary turbine shut-downs, owing to low steam flow periods, are reduced up to a 100%.