COMPARISONS OF BUILDING SYSTEM MODELING APPROACHES FOR CONTROL SYSTEM DESIGN Donghun Kim a, Wangda Zuo cb, James E. Braun a and Michael Wetter b a Ray W. Herrick Laboratories, School of Mechanical Engineering, Purdue University b Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory c Department of Civil, Architectural and Environmental Engineering, University of Miami
OBJECTIVES Evaluate Modelica Buildings Library as a tool for design and analysis of building control systems Feedback control response Energy consumption predictions Computational time Evaluate computational improvements and accuracy associated with implementation of reduced-order model (ROM) for the building within Modelica Provide comparisons with well-developed simulation tool (TRNSYS) as a reference model 2
NOTATIONS TRN: TRNSYS V17 MOD: Modelica Buildings Library MOD+ROM: Integrated building ROM within MOD MOD: HVAC & control system model ROM: Envelope model Envelope model Replaced with ROM HVAC & Control system 3
Reduced Order Model (ROM) ROM of a building envelope system A linearized time invariant model from a detail description of a thermal network of building envelope model Balanced truncation, a model order reduction method, is performed Reference D. Kim and J.E. Braun, Reduced-order Building Modeling For Application To Model-based Predictive Control, 2012, IBPSA-USA conference 4
MODEL DESCRIPTIONS North wing of an existing building with 9 VAV boxes Separate models for TRNSYS, Modelica Buildings Library and the reduced-order model (ROM *) were developed. Significant time was spent to provide building models with consistent descriptions and data inputs 5
OPEN LOOP TEST OF ENVELOPE MODEL Open loop : Temp. response driven by weather Responses were compared to make sure their envelope models are similar Mean and RMS difference between MOD and TRN: 0.16 [ o C] and 0.42 [ o C] for one-year simulation ROM has similar differences 6
COMPUTATION TIME COMPARISON (OPEN LOOP ) MODEL TIME [SEC] SOLVER TRN MOD ROM 30.1 59.1 Successive 60min Time step Successive 30min Time step 27.8 Dassl, 10-3 tolerence 77.5 Dassl, 10-4 tolerence 0.8 Simulink, 60 min 1.4 Simulink, 30 min For the open loop test of building model, MOD is as fast as TRN (but doesn t include feedback control of equipment) ROM for building requires 1/50 th computation time 7
COMPUTATION TIME COMPARISON (CLOSED LOOP) Including feedback control of equipment with variable time steps for MOD and MOD+ROM TRN is 1.67 and 2.85 times faster than the MOD+ROM and MOD ROM application saves up to 30% computational time. Removing state event of MOD, the speed improved. 8
OVERALL PERFORMANCE (CLOSED) Maximum RMS Temp. differences are 0.5 o K. Relative energy consumption (reference: TRNSYS) MOD MOD+ROM Heating - 5.63% 0.02% Cooling - 2.43% 1.36% Differences may be due to different modeling approaches (window, radiation, pressure, etc.) 9
Feed back response of a local controller From March/17 Response of an outdoor air (OA) damper Solving pressure distributions within a building system 10
FEEDBACK RESPONSE (2) Leaving coil temp. =Entering supply fan Fluctuations within TRN results occur due to a digital controller (due to numerical algorithm) Fluctuations do not appear within MOD results 11
MODELICA TIME STEP MOD time step chosen by a stiffness solver Mean: 183 [sec], Median: 20.5 [sec] Capturing a short time scale 12
CONCLUSIONS The following advantages of the Modelica Buildings Library are confirmed Input/output consistency with the actual control system Applicable to both supervisory and local level controller design/analysis Solves pressure distribution within the system Captures short time dynamics Requires more computation time (2 to 3 times), but can be improved by Removing state events Simplifying building envelope approach Replacing the detailed building description and model with ROM 13
QUESTIONS? THANK YOU 14
CONTROL ALGORITHM Control for minimum outdoor air A rule based control for TSP of entering supply fan PI controller for heating/cooling plant to meet the setpoint ON/OFF control for heating/cooling plant based on OA PI controllers to meet a thermostat SP for VAV boxes (dampers and reheat coils) Thermostat SP: 21.11 o C * Obtained from the technical report, Building 101 TRNSYS Baseline Control Logics, United Technologies Research Center