The present research has been carried out to provide a failure prediction tool for a Chilean concentrated juice company. The thermal energy required in the company’s processes is supplied through a liquefied petroleum gas (LPG) boiler, whose feed water is preheated by a parabolic trough solar collector. According to monitored data and computer simulations carried out during 2017, it was possible to identify a low performance of the solar field. For the low performance of the solar field, it has been possible to identify the following faults: inaccuracies of the solar tracking system, low cleaning frequency of the solar field, low effectiveness of the heat exchanger between the solar field and the processes feed water circuit, among others. A condition-based maintenance tool was developed to detect failures in the solar field using machine learning techniques. The tool uses a set of four machine learning models to detect and identify the existence and source of faults such as soiling factors in the solar field, problems with the solar tracking system, problems with pumps and faults in the heat exchanger. For faults greater than or equal to 20%, the tool can identify the source of the fault 80% of the time if it comes from the solar field or heat exchanger, however, if the fault comes from one of the pumps the performance is lower. The tool generates false positives 21% of the time when it is used the model to detect faults at a global level of the solar thermal plant. This tool could be used to optimally manage the solar plant and maximize the cost savings.

1.
División de Prospectiva y Política Energética
. (
2015
).
Reporte anual del balance nacional de energía 2014.
Ministerio de Energía.
2.
REN21.
2018
. Renewables 2018 Global Status Report (
Paris
:
REN21 Secretariat
). ISBN 978-3-9818911-3-3.
3.
Escobar
 et al (
2015
).
Estimating the potential for solar energy utilization in Chile by satellite-derived data and ground station measurements
.
Solar Energy
,
121
,
139
151
. 2016.
4.
Strategic Research Priorities for Solar Thermal Technology
. (
2012
).
Strategic Research Priorities for Solar Thermal Technology
.
Bélgica
:
European Solar Thermal Technology Panel (ESTTP
).
5.
Dirk
Pietruschka
. (
2015
).
Report for self-detection on monitoring procedure
.
Germany
:
Solar Heating & Cooling: Task
48
.
6.
Timma
L
,
Blumberga
D.
An Algorithm for the Selection of Structure for Artificial Networks
.
Case Study: Solar Thermal Energy Systems. Energy Procedia
2015
;
72
:
135
41
.
7.
S.A.
Klein
. et al,
2010
, TRNSYS 17: A Transient System Simulation Program,
Solar Energy Laboratory
,
University of Wisconsin
,
Madison, USA
, http://sel.me.wisc.edu/trnsys.
8.
Boukelia
TE
,
Arslan
O
,
Mecibah
MS
.
Potential assessment of a parabolic trough solar thermal power plant considering hourly analysis: ANN-based approach
.
Renew Energy
2017
;
105
:
324
33
.
9.
Bugedo
,
N.
 Soiling effect on reflective mirrors of concentrating solar plants under Chilean climatic conditions.
Pontificia Universidad Catolica de Chile.
2016
. https://repositorio.uc.cl/handle/11534/16878
This content is only available via PDF.