THERMOCHEMICAL HEAT STORAGE FOR CONCENTRATED SOLAR POWER

Abstract

Thermal energy storage (TES) is an integral part of a concentrated solar power (CSP) system. It enables plant operators to generate electricity beyond on sun hours and supply power to the grid to meet peak demand. Current CSP sensible heat storage systems employ molten salts as both the heat transfer fluid and the heat storage media. These systems have an upper operating temperature limit of around 400 C. Future TES systems are expected to operate at temperatures between 600 C to 1000 C for higher thermal efficiencies which should result in lower electricity cost. To meet future operating temperature and electricity cost requirements, a TES concept utilizing thermochemical cycles (TCs) based on multivalent solid oxides was proposed. The system employs a pair of reduction and oxidation (REDOX) reactions to store and release heat. In the storage step, hot air from the solar receiver is used to reduce the oxidation state of an oxide cation, e.g. Fe3+ to Fe2+. Heat energy is thus stored as chemical bonds and the oxide is charged. To discharge the stored energy, the reduced oxide is re-oxidized in air and heat is released. Air is used as both the heat transfer fluid and reactant and no storage of fluid is needed. This project investigated the engineering and economic feasibility of this proposed TES concept. The DOE storage cost and LCOE targets are $15/kWh and $0.09/kWh respectively. Sixteen pure oxide cycles were identified through thermodynamic calculations and literature information. Data showed the kinetics of re-oxidation of the various oxides to be a key barrier to implementing the proposed concept. A down selection was carried out based on operating temperature, materials costs and preliminary laboratory measurements. Cobalt oxide, manganese oxide and barium oxide were selected for developmental studies to improve their REDOX reaction kinetics. A novel approach utilizing mixed oxides to improve the REDOX kinetics of the selected oxides was proposed. It partially replaces some of the primary oxide cations with selected secondary cations. This causes a lattice charge imbalance and increases the anion vacancy density. Such vacancies enhance the ionic mass transport and lead to faster re-oxidation. Reoxidation fractions of Mn3O4 to Mn2O3 and CoO to Co3O4 were improved by up to 16 fold through the addition of a secondary oxide. However, no improvement was obtained in barium based mixed oxides. In addition to enhancing the short term re-oxidation kinetics, it was found that the use of mixed oxides also help to stabilize or even improve the TES properties after long term thermal cycling. Part of this improvement could be attributed to a reduced grain size in the mixed oxides. Based on the measurement results, manganese-iron, cobalt-aluminum and cobalt iron mixed oxides have been proposed for future engineering scale demonstration. Using the cobalt and manganese mixed oxides, we were able to demonstrate charge and discharge of the TES media in both a bench top fixed bed and a rotary kiln-moving bed reactor. Operations of the fixed bed configuration are straight forward but require a large mass flow rate and higher fluid temperature for charging. The rotary kiln makes direct solar irradiation possible and provides significantly better heat transfer, but designs to transport the TES oxide in and out of the reactor will need to be defined. The final reactor and system design will have to be based on the economics of the CSP plant. A materials compatibility study was also conducted and it identified Inconel 625 as a suitable high temperature engineering material to construct a reactor holding either cobalt or manganese mixed oxides. To assess the economics of such a CSP plant, a packed bed reactor model was established as a baseline. Measured cobalt-aluminum oxide reaction kinetics were applied to the model and the influences of bed properties and process parameters on the overall system design were investigated. The optimal TES system design was found to be a network of eight fixed bed reactors at 18.75 MWth each with charge and discharge temperatures between 1200 C and 600 C, which provides a constant output temperature of 900 C. The charge and discharge time are 8 hours each respectively. This design was integrated into a process flowsheet of a CSP plant and the system's economics were determined using AspenPlus and NREL's Solar Advisory Model. Storage cost is very sensitive to materials cost and was calculated to be based around $40/kWh for cobalt based mixed oxide. It can potentially decrease to $10/kWh based on reduced materials cost on a bulk scale. The corresponding calculated LCOE was between $0.22 and 0.30/kW-h. The high LCOE is a result of the high charging temperature required in this first design and the cost of cobalt oxide. It is expected that a moving bed reactor using manganese oxide will significantly improve the economics of the proposed concept.