Designing and Achieving Geothermal Power Plant Performance with Confidence
Guofu Chen, TAS Energy Inc. 6110 Cullen Blvd, Houston, TX 77021
The design and the actual performance of a geothermal air-cooled power plant utilizing a supercritical refrigerant of R134a as the working fluid are discussed. A supercritical Organic Rankine Cycle (“ORC”) in many cases outperforms a sub-critical cycle, from the net kilowatt (kW) generated point of view. Additionally, the plant configuration is simpler to design and easier to operate. An additional advantage of using non-flammable working fluid in the cycle, such as R134a, eliminates the risk of fires. During the design stage, a preliminary process flow diagram is established based on the standard process engineering practices in HYSYS, a simulation software from ASPENTECH. Based on the preliminary process requirement, the components of the cycle, including the shell and tube heat exchanger(s), expansion turbine, aircooled condenser, and working fluid feed pump are sized and selected. A true simulation model is built to analyze the off design performance of a “virtual plant”. Given the geothermal heat source information and the ambient conditions, the power output is maximized and committed to the customer (Model 1 with geometries). After the plant is successfully commissioned, by measuring the flow rate, temperature and pressure, a plant reality model is built to reflect the actual plant operating conditions (Model 2 without geometries). Normally the process conditions of Model 2 are different from Model 1. To validate Model 1, developed in the design stage, the process conditions of Model 2 are extracted and input into Model 1, thus Model 3 with actual process conditions and actual geometries is established. Model 2 is the reality, while Model 3 is used to predict the reality with actual process conditions and actual equipment selection. By comparing these two models, Model 3 accurately predicts the gross power and net power generated at various operating conditions.
The Challenge
When it comes to the bottom line of a geothermal project, a high confidence calculation of the plants performance over the projected economic life of the project with varying plant conditions is a fundamental input in the project proforma and a key driver on whether the project moves forward or not. There are two (2) important inputs to the performance calculation: the first one is the output kilowatt (kW) at various conditions, and the second one is the capital cost of the plant equipment to achieve that performance. In most cases, manufacturers are able to get accurate plant capital costs at the plant design point. However, the challenge is to accurately predict the performance output during the annual ambient temperature conditions and varying resource input conditions. These “off design” assumptions can be too aggressive on the output which results in the guaranteed performance at other conditions not being met, or too conservative so that the projected performance doesn’t support further development. TAS Energy has delivered several geothermal ORC plants based on their refrigerant based cycle designs, and after successfully passing the performance tests of all the commissioned plants provided, a consistent approach to the design and modeling has been established to deliver the performance with confidence.
The Solution
For illustrative purposes, an example of a project which utilized a supercritical R134a Organic Rankine Cycle (ORC) system was selected to describe the modeling and performance prediction process. In comparison with a subcritical pentane ORC system, there are three (3) principal benefits: 1) a supercritical cycle produces more net output than a subcritical system for many geothermal resource conditions; 2) R134a is non-flammable so it improves the operational risk of the project; and 3) the supercritical cycle has only one (1) conceptual vaporizer, while subcritical has potentially as many as four (4), thus the supercritical cycle is easier to design and simpler to operate.
After the conceptual budgetary phase of the project development cycle was complete and resource conditions were validated, a firm proposal inquiry was requested. At this point a thermal calculation (Model 1) was created to provide a general idea on how many kilowatts could be generated, based on sound and proven engineering practices and assumptions, such as the heat exchanger pinches, turbo expander isentropic efficiencies, and other commercial component performance expectations. Process engineers utilized the theoretical thermal performance data of Model 1 to size the shell and tube heat exchangers, the air cooled condenser array, the turbo expanders, and working fluid pumps. In an effort to leverage the best practices of past efforts and the economies of repeatability, if the equipment duty was similar to comparable equipment utilized on prior projects, then an economic evaluation of the component from the past project was performed balancing the performance versus cost benefit. After all of the equipment selections were finalized, Model 1 integrated all geometries from the actual selections and then off design performance runs were commenced to predict the performance at various conditions, including the net output of the plant at various ambient dry bulb temperatures. The end result was performance correction curves that became the basis for the purchase contract of the plant.
After completion of the commissioning and startup of the plant, TAS Energy analyzed the performance output at various conditions to verify how accurate our predictions were. Based on the measured resource flow rate, operating cycle pressures and temperatures, a thermal model (Model 2) in HYSYS was established. The goal of Model 2 was to get the actual process conditions at the various component operating boundaries, since measurement devices were not installed at every boundary. In addition, Model 2 was able to identify areas of inconsistent data based on the analysis of data measured and calculation.
The third modeling exercise was to combine the real geometries in Model 1 and the actual process conditions identified through plant measurement and used in Model 2 to create an operationalized Model 3. Additional minor refinements of Model 3 were required to match Model 2 exactly, due to some factors such as fouling and other operational factors. Once Model 3 was calibrated to match Model 2, it could then be used to accurately predict the performance at other conditions.
The actual operation data was extracted from the plant and plotted in the following chart, all blue diamonds are actual operation points. The dark red straight line is the predictions calculated from Model 3. On average, the prediction matches the actual plant operation well. However, we do see a wide range variance of output at the same dry bulb temperature. The thermal inertia of the large heat exchanger surfaces is believed to be the principal reason for the variation. Further analysis of the dynamic response to this thermal inertia is being investigated.
The Conclusion
By using the “3 Model” approach mentioned above, TAS Energy is able to confidently deliver a plant that met its design performance guarantees but just as importantly accurately predict the off design performance. Model 1 is created during the proposal phase to predict the off design performance and provide the process data to accurately provide a capital cost for the project economics. Model 2 is established after the plant is commissioned to reflect the actual plant operation. Finally Model 3 combines the real geometries and the actual plant process conditions to predict the plant performance at other conditions. Through this illustrative project, the “3 Model” approach proves to be the solution to deliver geothermal power plant performance with confidence.
Note: This paper was awarded "Paper of the Year" 2013 in Renewable Energy World Conference in Orlando. Click the link for more information http://www.renewableenergyworld-events.com/2013-award-winners.html.
