MODELLING FOR ENERGY MANAGEMENT A SHIPYARD S PERSPECTIVE EDWARD SCIBERRAS & ERIK-JAN BOONEN

MODELLING FOR ENERGY MANAGEMENT A SHIPYARD S PERSPECTIVE EDWARD SCIBERRAS & ERIK-JAN BOONEN

HISTORY 1927 DAMEN IS ESTABLISHED BY BROTHERS JAN & RIEN 1969 K. DAMEN TAKES OVER & INTRODUCES STANDARDISATION 1978 DAMEN STARTS BUILDING OUTSIDE OF THE NETHERLANDS 1939 YARD NUMBER 100 IS DELIVERED 1975 EXPANSION DAMEN SHIPYARDS GORINCHEM 2

HISTORY 1988 DAMEN ACQUIRES DE GROOT NIJKERK 2000 NAVAL SCHELDE JOINS DAMEN GROUP 2016 DAMEN REACHES 9,000 EMPLOYEES WORLDWIDE 1992 AMELS HOLLAND IS ACQUIRED 2006 DEVELOPMENT OF AXE BOW 3

INTRO ASD TUG 2810 HYBRID Hybrid Tug 2x 2,000 kw propulsion 2x 230 kw shaft generators Batteries 2x 118 kwh ASD TUG 2810 HYBRID 4

INTRO SeaExplorer Diesel-electric propulsion Main DC distribution Dynamic Positioning Variable speed generators Batteries Optional SEAXPLORER 5

INTRO DRFE8117 Double ended ferry 2x 825 kw propulsion drives 2x 1,080 kw Diesel-generators Batteries 2x 400 kwh Future full battery powered Main DC microgrid PUBLIC TRANSPORT 7

INTRO CONTENTS Introduction to Damen Modelling at Damen Modelling Efforts System Design Tertiary Control Hardware In the Loop Testing Modelling path Outlook Conclusions

INTRO WHAT WE MODEL Traditional shipbuilding modelling Hydromechanics - CFD Structures FEA Sound & Vibration SEA Significant effort and capability Machinery modelling is the concern here

INTRO SYSTEM DEVELOPMENT PROCESS Requirements analysis System testing High-level design Integration testing Detailed design Unit testing Implementation

INTRO DEVELOPMENT PROCESS NOW Requirements analysis System Topology High-level design System design Acceptance trials Integration testing System testing Detailed design FAT Unit testing Component Calculations Implementation

INTRO THE ROLE OF MODELLING Requirements analysis System Topology High-level design System design Acceptance trials Simulation testing Integration testing System testing Detailed design FAT Unit testing Component Calculations Implementation

INTRO MACHINERY MODELLING Component Calculations THD estimations Shafting calculations Power-speed calculations Design for specification point Static design point kw Design for service condition Design with operational profile in mind Design by simulation Energy storage kwh

INTRO CONTENTS Introduction to Damen Modelling at Damen Modelling Efforts System Design Tertiary Control Hardware In the Loop Testing Modelling path Outlook Conclusions

SLD SYSTEM LEVEL DESIGN Design for an objective! Design by Simulation for a given objective Capability to consider complete system Operating profile Control Topology Sizing of energy storage is prime motivator User constraints Hull information Machinery configuration Objective function is system model Sizing Components Need to consider Energy in time 17

SLD SYSTEM LEVEL DESIGN ARCHITECTURES

Hybrid Fuel cell TOPOLOGY SELECTION Work done in JOULES project Large effort on modelling Evaluation of different concepts for Battery ferry Hybrid Tug Wind-assisted Cargo vessel 19

SLD BATTERY FERRY Goal: Balance between battery size, lifetime & weight Objective function: Total cost = cost operation + cost components Variables Gear ratio Motor size Battery size Charging rate Wessel van Zon Multivariable optimization methodology for an optimized design of electric powertrains of vessels (2016)

SLD NESTED APPROACH Particle Swarm Optimisation as global algorithm Within each evaluation, optimal tertiary control is evaluated Energy management concerned with charging rates DP algorithm But very computationally expensive!

SLD MODELLING Feedforward simulation (Matlab) Static models Linear approximations and LUTs Scalable Battery ageing model based on DoD cycling Weight increase effect on propulsion Simulation Inputs: Shaft speed Shaft torque Battery SoC Rule based energy management

Switch Board [400V, 60Hz] SLD HYBRID TUG Goal: Investigate feasibility of global design tool ~ = = ~ Shor e Objective function: EIM 230kWe DE 1600kW Total cost = cost operation + cost components DE 690kW G 650kWe ~ = H BAT 240 kwh Variables EIM 230kWe DE 1600kW Electric machine size Battery size ~ = = ~ Torque sharing EM/DE Electric power split DG/Batt Mark Naaborg - System-level Design optimization of a hybrid tug (2017)

SLD SIMULTANEOUS APPROACH Manual selection of DE and Genset (Exhaustive search) Convex optimization as global algorithm Integer optimisation of DE/Genset on/off

SLD MODELLING Feedforward simulation (Matlab) Static models Convexified component models Fitting of quadratic functions to manufacturer or measurement data Scalable component models Simulation inputs Shaft speed Shaft torque Battery SoC Energy management parameters are part of optimisation e.g. for Diesel engine

SLD SYSTEM LEVEL DESIGN - SALIENT POINTS Quantification of optimality of results Hard to validate meaning of global optimality in a practical case Complex to build Quite a bit of adjustment of data needed to make it robust Alternate architecture is a lot more tangible Rule-based control easier to fit Attraction of separate optimisation algorithm and system model Energy management is an integral part of system design Chicken and egg situation!

INTRO CONTENTS Introduction to Damen Modelling at Damen Modelling Efforts System Design Tertiary Control Hardware In the Loop Testing Modelling path Outlook Conclusions

TERTIARY CONTROL TERTIARY CONTROL OPTIMISATION Complex systems controlled by deterministic rule based controllers Safe Well understood Transparent Convenient Different modes operator selectable Predefined rules for different operational conditions Energy management needs to be taken out of the operator s hands

Switch Board [400V, 60Hz] TERTIARY CONTROL HYBRID TUG Objective function Fuel consumption Variables Torque sharing EM/DE EIM 230kWe ~ = = DE 1600kW ~ Shor e ~ = BAT 240 kwh Electric power split DG/Batt DE 690kW G 650kWe H Implementation as realtime controller EIM 230kWe DE 1600kW ~ = = ~ Militiadis Kalikatzarakis Tertiary Control of hybrid Tugboats (2017)

TERTIARY CONTROL ECMS APPROACH Equivalence factor assigns equivalent fuel consumption value to energy stored in battery This influences decision on whether to charge or discharge battery. Part of the propulsion part model approach Speed setpoint Power Output SOC Propulsive Load Statistics Equivalence Factor Optimizer (Branch & Bound +Lagrange Multipliers) Propulsion Plant Representation Induction Machine Diesel Engine Diesel Generator Battery System

TERTIARY CONTROL MODELLING Detailed models used to derive 2 nd order polynomial fits Static models to calculate fuel consumption/losses Fast simulation required for realtime application. ሶ m f,me = b ME (t) a 0 + a 1 n ME (t) + a 2 P ME (t) + a 3 n 2 ME (t) + a 4 P 2 ME (t) + a 5 n ME t P ME t

TERTIARY CONTROL SALIENT POINTS ECMS looks like very promising for realtime application ECMS did not prove more efficient than RB controller BUT utilises battery capacity better Tuning and sensitivity of controller still question marks Variability of operating profile gives large spread in results Possibility of using offline procedure to tune ruleset Improves performance Maintains familiarity

INTRO CONTENTS Introduction to Damen Modelling at Damen Modelling Efforts System Design Tertiary Control Hardware In the Loop Testing Modelling path Outlook Conclusions

HIL HARDWARE IN THE LOOP Physical part of a machine or system is replaced by a simulation.

Thruster control Thruster control PCS PCS BMS AVR + Governor AVR + Governor BMS HIL APPLICATION EXAMPLE G G Power Management System M AC grid M HIL machinery model Ship model Control system hardware Inputs

HIL simulation HIL CONTROL SYSTEM OVERVIEW Operator Hardware PCS PMS Thruster control Propulsion drive DC/DC converter BMS AVR Engine control Software Thruster + Motor Batteries Generator Ship model Virtual ship

Thruster control PCS AC grid BMS AVR + Governor HIL IMPLEMENTATION User input Power Management System Machinery simulation Ship model Modelling complexity dependent on application G Level of detail determines what we want to test M Goals Testing of integration of control systems User input Control System under test Machinery simulation Ship model Testing of correct functionality of energy management

INTRO CONTENTS Introduction to Damen Modelling at Damen Modelling Efforts System Design Tertiary Control Hardware In the Loop Testing Modelling path Outlook Conclusions

CONCLUSIONS MODELLING REQUIREMENTS Requirements analysis System topology Increased model complexity High-level design System design Detailed design Component Calculations Increased model scope Implementation HIL testing Scope and complexity depending on what to test

INTRO THE ROLE OF MODELLING & OPTIMISATION Component LUTs 1 st order approximations? Heuristic optimisation tools PSO? GA? Simplified models 2 nd order approximations? Numerical methods? Requirements analysis System Topology Detailed models High-level design System design Detailed design Component Calculations Hardware In the Loop testing Implementation

THE OUTLOOK FOR ENERGY MANAGEMENT Energy storage introduces the time factor Design for service rather than specification (only) Needs to be considered at design stage to fully exploit benefits Hand in hand with plant design Rule-based energy management is the norm Modelling helps to tune rules 45

Testing by simulation CONCLUSIONS GENERAL OUTLOOK Global system level design Optimal energy control Ad hoc modelling Established as mainstream tool in the industry Now Ongoing projects Longer term research effort

CONCLUSIONS CONCLUSIONS Design by Simulation requires multidimensional modelling effort Holistic energy flows! Time domain simulations are now being used for system development System design System operation Hardware In the Loop testing of complex systems required for derisking Rule-based energy management tuned by modelling ECMS Machinery systems simulation scope will keep on increasing in time

48 Thanks for your attention!