BALANCED MANUFACTURING Modeling, Simulation and Optimization of the Operation of Electric Chillers Benjamin Mörzinger August 15 th, 2016
Motivation Source: science.sciencemag.org 2
Architecture of BaMa Tool Chain 3
Cube Methodology Air compressor Mass balance Energy balance Time balance Cost balance Information- & resource flows Cube Air compressor Production machine Production machine Cube Production machine Cube Production machine Production process Waste disposal Cube Waste disposal 4
Cube Categories 5
Cubes Production plan Operation mode Control signal etc. Material flow Work piece, goods, etc. Footprint (Time, Costs, CO 2 ) Supplied energy (electrical, thermal, etc.) CO 2 conversion factors Control action Energy flow Measured data Model parameters: Dimensions Capacity Efficiency etc. Energy flow Energy demand Operational state etc. Material flow Work piece, goods, etc. Updated footprint Waste energy Recovered energy, Produced energy CO 2 conversion factors Advantages of Cubes Decomposition approach helps tackle system complexity Modularization offers flexibility and reusability Boundary specification enables encapsulation and coupling of different modeling paradigms Generic description enables applicability to different industry sectors Congruence of real and virtual cubes facilitates data comparability Ontology facilitates communication Challenges of Cubes Uniting discrete-event and continuous simulation (DEV&DESS formalism) Choice of optimization algorithm (no derivative of objective function available) 6
Overview Use Cases Facility Management 36 MW Cold 10 MW Heat Production Hall 1800m² Clean Room 2 MW el. Power 3 MW Cold 2 MW Heat Cooling Tower Machine Production Hall Machine Machine Chiller Air Conditioning 7
Model Approach PP iiii : EnergyPort PP oooooooo TT CCCC TT KKKK Parameter: Chiller Specified Capacity(CCCCCC 0 ) Specified EER (ηη 0 ) Part-Load-Coefficients: aa = (aa 0, aa 1, aa 2 ) EER-Coefficients: bb = (bb 0, bb 1, bb 2, bb 3, bb 4, bb 5 ) Capacity-Coefficients: cc = (cc 0, cc 1, cc 2, cc 3, cc 4, cc 5 ) PP oooooo : EnergyPort QQ AAAA : EnergyPort PP iiiiii QQ WW : EnergyPort QQ AAAA : EnergyPort QQ HH : EnergyPort QQ KK : EnergyPort Parameter: Thermal Zone Number of Waste heat-inputs (NN) Volume (VV) Air Density (ρρ LL ) Specific heat (cc pppp ) Infiltration air change rate (ii) Ventilation air change rate (vv) Lower set temperature (TT ssssssss ) Upper set temperature (TT ssssssuu ) State Variables: EIN : EntityPort Pel : EnergyPort Qw : EnergyPort Qk : EnergyPort Parameter: Production Unit Capacity (NN) Production plan (Pplan) Power levels el. (PP EEEEEEEE ) Power levels heat ( QQ HHHHHHHH ) Power levels cold ( QQ CCCCCCCC ) Holding time (tt BB ) Inside temperature (TT) Waste heat recovery rate (ηη) Waste rate (αα) Area weight (da) Production weight (dp) Value weight (dv) State Variables : Inside Temperature(TT) EOUT : EntityPort EA : EntityPort QAW : EnergyPort Qrec : EnergyPort TT oooooo QQ HHHH QQ KKKK VV BB State Variables : Operational state (p) List of entities (ent) ID-Counter (idcount) PelB QwB QkB EINcom 8
Detail: Chiller PP iiii : EnergyPort PP oooooooo TT CCCC TT KKKK Parameter: Chiller Specified Capacity(CCCCCC 0 ) Specified EIR (ηη 0 ) Part-Load-Coefficients: aa = (aa 0, aa 1, aa 2 ) EIR-Coefficients: bb = (bb 0, bb 1, bb 2, bb 3, bb 4, bb 5 ) Capacity-Coefficients: cc = (cc 0, cc 1, cc 2, cc 3, cc 4, cc 5 ) State Variables: PP oooooo : EnergyPort QQ AAAA : EnergyPort PP iiiiii CCCCCC = CCCCCC 0 cc 0 + cc 1 TT KKKK + cc 2 TT 2 KKKK + cc 3 TT CCCC + cc 4 TT 2 CCCC + cc 5 TT KKKK TT CCCC ηη = ηη 0 bb 0 + bb 1 TT KKKK + bb 2 TT 2 KKKK + bb 3 TT CCCC + bb 4 TT 2 CCCC + bb 5 TT KKKK TT CCCC (aa 0 + aa 1 PPPPPP + aa 2 PPPPPP 2 ) PPPPPP = PP oooooo CCCCCC PP oooooo = min(pp iiii ηη, CCCCCC) QQ AAAA = PP iiii. pp PP oooooo PP iiiiii = PP oooooooo ηη 9
Parameterization TT eeeeee,oooooo TT eeeeee,iiii TT cccccc,iiii mm eeeeee Evaporator outlet-temperature Evaporator inlet-temperature Evaporator outlet-temperature Evaporator mass flow QQ eeeeee QQ cccccc PP eeee Evaporator Temperature Evaporator Temperature Electrical Power Demand Identify Specification Values QQ SSSSSSSS =max( QQ eeeeee ) PP SSSSSSSS =PP eeee QQ SSSSSSSS QQ EEEEEE SSSSSSSS = SSSSSSSS PP SSSSSSSS Full-Load condition data set CCCCCC 0.85 Calculate Fit Coefficients for Cap Calculate Fit Coefficients for η Part-load conditions data set Generate part load data set based on found parameters Calculate Coefficients for PLR Capacity [kw] EIR [-] R² [-] RMSE [%] Machine 1 1760 3.13 0.45 10.46 Machine 2 2649 5.48 0.97 3.36 Machine 3 2861 5.49 0.99 3.83 Machine 4 2443 5.39 0.99 2.39 10
Optimization production program goals optimization optimization algorithm scheduler search for (better) alternative solutions parameter Set hybrid simulation (prediction) monitoring data monitori ng Energy system Building forecast Machine Logistics heuristic & domain knowledge simulation results objective function evaluation of forecast optimized plan optimization results 11
Outlook: Full Implementation 12
Plots - Oven & Packaging 13
Plots - Zone Temperature & Overall Energy 14
Contact DI Benjamin Mörzinger Technische Universität Wien IFT - Institut für Fertigungstechnik und Hochleistungslasertechnik Forschungsgruppe Energie- und Ressourceneffizienz Getreidemarkt 9 / 311 1060 Wien Österreich Tel.: +43 1 58801 31118 Mobil: +43 6648966685 Email: moerzinger@ift.at www: http://www.ift.at 15