Issue1298 linear actuators

[AntoineGautier]

This partly closes #1298 (the valves components still need to be updated with the same logic).
(Corresponding issue in IBPSA library: #1188)

• VAVBoxExponential is retired. The option for the fixed resistance has been integrated into Exponential. (A conversion script is attached.)
• The exposed parameter dp_nominal in Exponential and PressureIndependent has been replaced by dpExp_nominal and dpDam_nominal (modification also automated by the above script). The motivation is to restrict the use of dp_nominal to the nominal pressure drop at the model boundaries (which is used by PartialDamperExponential in the homotopy initialization cf. dp/dp_nominal_pos). In the master branch, it is indeed the case for Exponential, it might be the case for VAVBoxExponential if dp_nominalIncludesDamper==True and it is never the case for PressureIndependent. Introducing the new parameters ensures consistency (and appears to be the best way to merge VAVBoxExponential into Exponential properly).
• The option to linearize the characteristics of the damper plus fixed resistance has been added.
• New limiting values for parameters k0 and k1 are used in PartialDamperExponential motivated by https://antoinegautier.github.io/plots/Actuators.html#/6.
• PressureIndependent:
  • the algorithm implemented to deal with small derivatives of m_flow / dp while ensuring a bounded error (approximately) is illustrated here https://antoinegautier.github.io/plots/Actuators.html#/4. It relies on 1) an intermediary dp_lim variable limiting the linear domain where m_flow = y_actual * m_flow_nominal + cregul *..., 2) an extended domain for the upper spline interpolation and 3) a leakage flow supposed to be mainly turbulent (to limit the domain where the 2 flow functions are not inverse from each other). The resulting flow characteristics is also part of the updated example see Buildings/Fluid/Actuators/Dampers/Examples/Damper.mo case preIndCha. The impact on the test case described in Implement linearization algorithm for damper and valve models #1298 is presented in https://antoinegautier.github.io/plots/Actuators.html#/2/1.
  • the steep transition of the opening from 1 to 0 when y=0 and dp transitions from negative to positive values is handled through a threshold value (2%) of the input signal below which the opening is forced to zero.

Note:

• Some models of any library using PressureIndependent might fail to simulate without changing some parameters values to be consistent with the exponential damper characteristics that is now used by the model to compute the opening (typically models in Buildings/Examples/VAVReheat had a too high nominal pressure drop with a too low nominal velocity: those parameters have been changed manually, the script mentioned above does not handle that modification).
• The reference results have been updated for the 3 models hereafter which use Exponential or MixingBox damper models. This is unexpected but needed due to small numerical effects yielded by the modification of the model structure only (as one can see in the modified translation statistics): the governing equations are unchanged and it has been verified that all the parameters (including the protected ones) have the same value between the master and the development branch (except for dp_nominal as explained before but that should not impact the simulation after initialization).
  • Buildings_Examples_DualFanDualDuct_ClosedLoop.txt
  • Buildings_Examples_HydronicHeating_TwoRoomsWithStorage.txt
  • Buildings_Applications_DataCenters_DXCooled_Examples_DXCooledAirsideEconomizer.txt
• Buildings.Examples.DualFanDualDuct.ClosedLoop fails to simulate with JModelica for the same reason. I will notify Modelon (Agnes). In the meantime simulation is excluded from JModelica regression tests.

Conversion script:

VAVBoxExponentialRefactor.py.txt

[AntoineGautier]

Close as it is now addressed in issue1188_dampersRefactor of IBPSA.