1 NOWITECH models for O&M and fleet optimization Thomas Welte SINTEF Energy Research Reliability Data Workshop, Berlin 2015-09-23
www.sintef.no 2
SINTEF Energy Research
NOWITECH Norwegian Research Centre for Offshore Wind Technology Research partners: SINTEF ER (host) IFE NTNU MARINTEK SINTEF ICT SINTEF Material & Chemistry Associated research partners: DTU Wind Energy Michigan Tech Uni. MIT NREL Fraunhofer IWES Uni. Strathclyde TU Delft Nanyang TU Industry partners: CD-adapco DNV GL DONG Energy Fedem Technology Fugro OCEANOR Kongsberg Maritime Norsk Automatisering Rolls Royce SmartMotor Statkraft Statnett Statoil Associated industry partners: Devold AMT AS Energy Norway Enova Innovation Norway NCEI NORWEA NVE Wind Cluster Norway O&M Grid Wind turbine Sub - structure 2009-2017 budget: NOK 320 millions (~M 40) cofunded by the Research Council of Norway and NOWITECH partners
5 Outline NOWITECH (strategic) models for O&M and fleet optimization Typical decision problems Examples and case studies Data requirements
NOWITECH (strategic) models for O&M and fleet optimization Simulation model: NOWIcob Developed by SINTEF Energy Research Optimization model: Vessel fleet optimization model Developed by MARINTEK
NOWITECH maintenance logistic models - Team and cooperation NOWITECH NTNU IØT (Post doctor, master students) MARINTEK 1. Vessel fleet optimization model 3. Routing and scheduling SINTEF ER 2. NOWIcob - O&M simulation model NTNU Marin 3. Routing and scheduling LEANWIND EU funded project (ongoing) FAROFF NFR funded (ended)
Simulation and/or optimization? Simulation: evaluate a given possible solution Optimization: searches for the optimal solution among a large set of different possible solutions Solution space Possible solution
Simulation and/or optimization? Simulation: evaluate a given possible solution Optimization: searches for the optimal solution among a large set of different possible solutions Possible solution Considered solution(s) (suboptimal) Optimal solution
Simulation and/or optimization? Simulation: evaluate a given possible solution Optimization: searches for the optimal solution among a large set of different possible solutions Possible solution Considered solution(s) (suboptimal) Optimal solution
NOWIcob: O&M simulation model Model developer: SINTEF Energy Research
O&M vessel fleet optimization model? Model developer: MARINTEK
Transport logistics for offshore wind farms Characteristics of offshore wind farm: Large number of turbines Large distances Marine operations Turbine accessibility Weather restrictions Infrastructure Ports, offshore platforms, mother vessels Means of transport Vessels, helicopters Maintenance personnel GL Garrad Hassan (2013). A Guide to UK Offshore Wind Operations and Maintenance. Scottish Enterprise and The Crown Estate.
Typical applications of the models Decision support for choosing vessel fleet Estimating O&M costs Calculating wind farm availability Analysing O&M and logistics strategies Cost-benefit analysis of O&M innovations Studying sensitivities and drivers of offshore wind O&M Evaluating the effect of uncertainties / stochastic elements
Application to a reference case Test case with 80 turbines, 50 km from shore Conventional CTVs or surface effect ships (SESs) Failure rates based on Dinwoodie et al. (2015) Weather data from the FINO1 metocean platform El. price: 90 GBP/MWh Dinwoodie, I., Endrerud, O.-E., Hofmann, M., Martin, R., Sperstad, I.B., "Reference Cases for Verification of Operations and Maintenance Simulation Models for Offshore Wind Farms.",Wind Engineering, 39 (1-14), 2015.
Result from vessel fleet optimization model 2 Surface effect ships Availability - energy based 95 % O&M cost (absolute value) 9 696 380 Loss of production (absolute value ) 5 400 635 Vessel cost 3 650 000 Spare part cost 645 745
Result from NOWIcob simulation model 17 1 surface effect ship + 1 conventional CTV 160,00% 140,00% Personnel cost Vessel cost Spare part cost Lost income due to downtime 120,00% 100,00% 80,00% 60,00% 40,00% 20,00% 0,00% 2 CTV 3 CTV 2 CTV + 1 SES 1 CTV + 1 SES 2 SES 1 CTV + 2 SES 3 SES 17
Using simulation and optimization model 18 Inputs: Optimization model Simulation model Outputs: Added value 18
Advantages/disadvantages and use of the models Technology for a better society 19
Optimization model: Data/structure 1/2 Vessel type Max Hs and wind speed Sailing speed No. of technicians Docking and transfer time Fuel consumption rates Base Distance from wind farm Maintenance activities Type Time to execute Personnel requirements Vessel requirements Weather input One historical year of weather data Compatibilities Vessel/base Vessel/ maintenance activity Possibilities Weather, time and personnel constraints Maintenance patterns Describes which maintenance activities a given vessel type, based at a given base can execute during a work shift during a given time period Technology for a better society 20
Optimization model: Data/structure 2/2 Maintenance patterns Cost of executing pattern Vessel type Time charter costs Availability Base Max personnel Base incompatibility Maintenance activities How many to execute When they can be executed Penalty/downtime cost Optimization program Selects the overall cost-effective combination of vessels, bases and maintenance patterns Output Vessels and corresponding bases Which maintenance patterns that are executed when Technology for a better society 21
22 Data requirements Most important data requirements optimization model Weather data Currently use significant wave height and wind speed (other weather factors may be used) Vessel data Transfer speed Weather requirements ("go/no-go input" with respect to accessibility for individual vessel types) Time charter cost Failures data and maintenance activities data Failure rates Number of, time for, and necessary resources for both corrective and preventive activities
23 Reliability of data Important to analyze potential effects of changes in the input data Can experience that changes in input data result in a different optimal fleet: - Varying time charter prices for vessel types - Varying weather input - Varying number of corrective maintenance activities A stochastic version of the model implicitly evaluates some of these effects But still relies on scenarios being generated from correct input data Constraints Objective function Best solution
24 Reliability of data Important to analyze potential effects of changes in the input data Can experience that changes in input data result in a different optimal fleet: - Varying time charter prices for vessel types - Varying weather input - Varying number of corrective maintenance activities A stochastic version of the model implicitly evaluates some of these effects But still relies on scenarios being generated from correct input data Constraints Objective function Best solution
25 Acknowledgements MARINTEK Lars Magne Nonås Elin Espeland Halvorsen-Weare Magnus Stålhane SINTEF Energy Research Iver Bakken Sperstad Magne Kolstad Matthias Hofmann In NOWITECH and other projects (Faroff, LEANWIND)
26 Trondheim, 20-22 January 2016 Deadline for abstract submission: 15 October 2015 Call for papers and further information: http://www.sintef.no/projectweb/deepwind_2016/