SE MS Thesis Presentation of Aykut Turkoglu

Analysis of Parabolic Trough Collector Cleaning System Under Adaptive Scheduling Policy

The purpose of this study is to investigate the effects of stochastic dust accumulations and rain events on the cleaning schedule of the parabolic trough collectors that are used to generate power at concentrated solar power (CSP) plants. The level of cleanliness is proportional to the power produced, and thus it affects the economic pay off at CSP plants. Current practice to address this dust problem, termed as conventional cleaning, is to follow a periodic cleaning schedule that entails a fixed setup cost for each cleaning event. The frequency ofcleaning under such conventional (periodic schedule) policy is selected based upon a tradeoff between the setup cost and the payoff from improving the cleanliness factor. The conventional practice is to have a constant andperiodic cleaning schedule over an entire season (e.g. either severe or mild combination of the dust and rain overa 180-day cleaning season, with either 8 or 4 cycles scheduled for the severe and mild seasons respectively).

This thesis draws upon evidence from recent literature to show that presence of random rain events improves thecleanliness of parabolic troughs in CSP plants. Upon analyzing such evidence, this study models rain event as a compound poison process that replenishes the level of cleanliness. In this scenario, it is possible to establish an adaptive threshold policy for scheduling plant cleaning that analogous to the formulation of a (s,S) inventorymanagement policy, subject to random replenishment of inventory. The study offer a review of related literature toestablish that such formulations are not amenable to a close form solution.

The second half of the thesis describes a numerical study that has been conducted using Arena Simulation packagefor characterizing the adaptive cleaning policy. The parameter of interest for assessing system performance is theaverage payoff over the average cost of cleaning for a 180-day cleaning season. Numerical study shows that adaptive cleaning policy outperforms the conventional (periodic) cleaning policy under reasonable assumptions for dust and rain event distributions. As an extension, the simulation study also examines the use of alternativecleaning system, known as electrodynamic screening (EDS), for different rain scenarios that may be used in conjunction with either conventional or adaptive cleaning policies to improve the overall system performance.

COMMITTEE: Advisor: Erol Pekoz, SE/Questrom; Nitin Joglekar, Questrom; Malay Mazumder, MSE/ECE

When 11:00 am to 1:00 pm on Tuesday, April 18, 2017
Location 8 Saint Mary's Street, Rm 339