Skip to content

Latest commit

 

History

History
594 lines (377 loc) · 19.6 KB

CHANGELOG.md

File metadata and controls

594 lines (377 loc) · 19.6 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on: Keep a Changelog and this project adheres to: Semantic Versioning.

[1.8.0] - 2019-07-08 - RIHY

Added

  • plot_stats() method added to Multiple() class in dyn_analysis module. This can be used to present results from multiple time-stepping analysese. Refer Train Analysis example script for demo of its usage.

Changed

  • collate_stats() method of Multiple() class in dyn_analysis module now returns a Pandas DataFrame, with MultiIndex columns and index to organise data hierachically

  • dashpot argument added to TMD().__init___() method, to allow dashpot of TMDs to be defined (rather than TMD damping ratio)

  • Inputs used to define TMDs are now recorded as class attributes, to allow later retrieval (but not editing; no set methods are provided)

  • Tutorial updated, with HTML export file included for accessibility

[1.7.5] - 2019-05-14 - RIHY

Fixed

  • Commented-out sections of Train Analysis example script resurrected. This involved updating the collate_stats() method of the Multiple() class, to make use of updated internal methods for handling results from time-stepping analyses. Needs some further work to handle the case of systems comprising multiple sub-systems however, as indicated by comments in the code.

[1.7.4] - 2019-05-02 - RIHY

Fixed

  • Bug when running analysis for SDOF modal systems fixed

  • Issue with automatic forcing of ylim in freq_response_results.plot() fixed (this had been causing plots to go off-scale previously)

[1.7.3] - 2019-04-05 - RIHY

Fixed

  • Error with EigResults() plot method #4; damping ratio now correctly expressed as a percentage

Added

  • Absolute displacement output added by default for TMDs

Changed

  • Improvements to plotting in freq_response_rslts.py and eig_results.py

[1.7.2] - 2019-03-21 - RIHY

Fixed

  • Fixed bugs in modalsys.py for case of system with only 1 mode

  • Improved plotting in freq_response_rslts.py

Added

  • TMD example script added

[1.7.1] - 2019-03-21 - RIHY

Fixed

  • Hotfix to check_class() method in common.py

[1.7.0] - 2019-03-21 - RIHY

Changed

  • Previous module wind.py split across three new modules:

    • wind_response.py houses functions/classes concerned with calculation of response to wind loading

    • wind_env.py houses functions/classes concerned with defining wind environments

    • wind_misc.py houses all other functions not allocated to the above. The intention is that these will be sorted-through and relocated / deleted eventually

Added

  • New module hysteresis.py, together with related example, to provide classes to model hysteretic behaviour

  • New module eig_results.py, with class to act as container for eigenvalue results, to replace previous dict usage. Coding is intended to be
    backwards-compatible, but allows eigenproperties plot methods previously housed as DynSys() class methods to more logically be implemented as methods associated with the new Eig_Results() class.

  • Note with this change, the Eig_Results() class only holds unique data, i.e. eigenvalues and left- and right-eigenvectors. All other eigenproperties are calculated upon request. This is intended to be more robust and will ensure consistent properties are always obtained.

  • Lots of alternative getter methods defined for Eig_Results() to be more flexible with spellings / names when requesting certain attributes

  • New module freq_response_results.py, with class to act as container for frequency response results, to replace previous dict usage. Coding is backwards-compatible, except PlotFrequencyResponse() method previously implemented within dynsys.py now used as FreqResponse_Results.plot() method

  • New module mesh.py, to faciliate definition or and manipulation of meshes (i.e. interconnected nodes and elements)

  • New module nodle.py, to provide methods for importing data from NODLE Excel input files (COO and MEM data) and .res files (DIS data)

[1.6.2] - 2018-10-19 - RIHY

Added

  • Option added to LatSync_McRobie.__init__() method, to allow system matrices determined at each Np value considered to be stored for subsequent usage

  • Verbose output control added to LatSync_McRobie() class methods

  • LatSync_McRobie().calc_Np_crit(): parameters / key results at Np=Np_crit instability point extracted and held as attributes

  • New module wind_section.py added to define wind sections / resistances

[1.6.1] - 2018-10-18 - RIHY

Fixed

  • Default mp(f) curve corrected; had previously mis-interpeted Fig 9(b) of MacDonald's paper

[1.6.0] - 2018-10-18 - RIHY

Changed

  • 'LatSync_McRobie' analysis extended to allow for frequency-dependent damping and added mass effect of pedestrians, with default curves defined based largely on John MacDonalds 'Lateral excitation of bridges by balancing pedestrians' paper, which has been included in the 'references' section.

Added

  • AnimateResults() method updated to plot loading as well as deformed system

  • Lateral loading model in the spirit of UK NA to BS EN 1991-2 added

Fixed

  • Minor bug fixed, relating to selection of correct S_eff for mode

[1.5.1] - 2018-09-20 - RIHY

Changed

  • AnimateResults() API updated to allow optional keywords to be passed down to methods called within, e.g. to customise plots/animations produced

  • Train analysis example updated to illustrate usage of AnimateResults() with animation customisation

[1.5.0] - 2018-09-19 - RIHY

Added

  • AnimateResults() method implemented properly (in a generalised fashion) for the first time. System-specific plot methods defined for common system types

  • PlotEnergyResults() and related calculation methods added to TStep_results class

[1.4.2] - 2018-09-17 - RIHY

Changed

  • Improvements to how event_funcs and post_event_funcs are implemented within TStep() class of tstep.py

  • Bounce sim example implemented (in conjunction with the above code updates)

  • CalcFreqResponse() method and function interface revised to give code that is better structured and more generally applicable.

  • Minor improvements to lat sync analysis:

    • Analysis is extended in case of not detecting net zero damping within range of pedestrian numbers originally specified
    • User control over verbose output added

Added

  • UKNA_BSEN1991_2_Figure_NA_11() function added to define Figure NA.11 in UK NA to BS EN 1991-2:2003.

Fixed

  • Correction to Figure NA.9 in UK NA to BS EN 1991-2, as defined in PD6688-2 has been implemented. Option added to select values according to each standard, but default behaviour is to use the PD6688-2 correction.

[1.4.1] - 2018-07-23 - RIHY

Added

  • calc_Np_crit() method added to LatSync_McRobie() class, to allow critical number of pedestrians to cause lateral instability to be considered. This method is also now called by LatSync_McRobie().run().

  • Improvements to related plotting routines; Np_crit overlaid onto plots

[1.4.0] - 2018-07-23 - RIHY

Added

  • Check added as to Scipy version when importing tstep.py module; v1.0 or greater required (an exception will be raised otherwise, as solve_ivp() method unavaliable in earlier versions.

  • scipy.linalg.null_space() method now used when possible (i.e. when Scipy v1.1 or greater is used), to save duplication of code in null_space() method with dynsys.py module. In the future the plan will be to use the Scipy method alone, to allow reduction in code.

  • LatSync_McRobie() class defined within ped_dyn.py module. Implements lateral synchorous vibration calculations per McRobie's Stockton on Tees footbridge paper (refer /references). See also validation script, which obtains excellent agreement with figures presented in McRobie's paper.

[1.3.0] - 2018-07-20 - RIHY

Added

  • New method transform_to_unconstrained() defined in dynsys.py module, to implement method for transforming a constrained system (i.e. a system with constraint equations) into an equivilent unconstrained one. [This is a key step in calculations such as complex eigensolution and evaluation of frequency transfer matrix.]. This method can be generally used - but the main motivation behind its implementation is to allow code reuse in multiple methods of the core DynSys class, e.g. DynSys.CalcEigenproperties().

  • DynSys.CalcEigenproperties() method will now work for systems with constraints. Accuracy of the implementation has been validated by extension of Warburton TMD validation script.

[1.2.2] - 2018-07-19 - RIHY

Fixed

  • Further fix associated with lambda fix at v1.2.1: when multiple modes included array of maximum modeshape ordinate now calculated and appropriate value used in calculations targeting specific mode. Previous behaviour was to obtain maximum ordinate across all modes, which is inaccurate when mode being considered does not have this ordinate.

  • Minor fixes to iron-out runtime bugs in routines associated with steady-state crowd loading analysis. Bugs relates to plotting and results reporting, i.e. no impact on the accuracy of calculations made using previous versions.

[1.2.1] - 2018-07-17 - RIHY

Added

  • damper.py module added; TMD() class defined within this module, to allow tuned mass damper systems to be conveniently defined

Fixed

  • Bug associated with calculation of lambda factor for UK NA steady-state crowd loading analysis fixed.

[1.2.0] - 2018-06-12 - RIHY

Changed

  • Pedestrian dynamics classes and functions moved to new module ped_dyn.py (previously these were within the dyn_analysis.py module)

  • Many method previously implemented as class methods of DynSys class now re-implemented as functions within the dynsys.py module. Class methods still exist as per previous versions, but these use the public functions to do the work

  • Various misc updates to plotting methods within tstep_results.py module

  • The majority of docstrings have been reviewed/improved, to ensure they reflect the source code and provide more references to theory, codes etc.

Added

  • tutorial.ipynb, a IPython (Jupyter) notebook which provides an introduction on how to go about using the classes/functions within the package

  • Crowd loading to UK NA to BS EN 1991-2 now implemented

  • CalcEigenproperties() method of DynSys class reimplemented to handle systems with constraints. Approach for unconstrained systems similar to previous versions except scipy.linalg.eig() method now used.

  • /tests updated: new test of CalcEigenvalues() method devised

Fixed

  • Bug as noted by BNCY in dyn_analysis.py now fixed. A check of the class name of loadtrain_obj was being made, but as a result of recent changes this was checking the derived class name.

  • trainAnalysis.py updated with minor change to reflect new stats_dict nested dict structure

  • Bug associated with output matrices fixed; was only an issue when re-defining systems

[1.1.1] - 2018-05-22 - RIHY

Hotfix responding to BNCY's 21/05/2018 email

Fixed

  • Tempororary fix implemented to address bug, by commenting-out rows. To be addressed properly in future releases.

[1.1.0] - 2018-04-23 - RIHY

Changed

  • Version number now included in html docs

Added

  • New functionality to implement UK NA to BS EN 1992-1 footbridge dynamics:
    • New class in loading.py, UKNA_BSEN1991_2_walkers_joggers_loading. Defines moving load group for walkers/joggers. Inherits most of the functionality from existing LoadTrain class.
    • Figure functions, to return parameters given by code figures:
      • UKNA_BSEN1991_2_Figure_NA_8 in loading.py
      • UKNA_BSEN1991_2_Figure_NA_9 in dyn_analysis.py
    • New class in dyn_analysis.py, UKNA_BSEN1991_2_walkers_joggers (this inherits functionality from existing MovingLoadAnalysis class)
    • New class in dyn_analysis.py, PedestrianDynamics_transientAnalyses. Inherits functionality from existing Multiple class. This can be used to carry out all the necessary walking/jogging analyses for all modes.
    • Example added to examples\Pedestrian dynamics\UK NA walkers joggers to test and illustrate the use of these number classes and functions

[1.0.0] - 2018-04-17 - RIHY

Changed

  • warburton_TMD.py validation script moved to subfolder. In future all validation scripts should be kept in subfolders to avoid confusion / cross-over of inputs

  • System matrices (e.g. M_mtrx, K_mtrx etc) now intended to be held as private attributes; renames _M_mtrx, _C_mtrx to denote this change in intent. GetSystemMatrices() member function of DynSys class re-instated (this was previously marked as deprecated) and overhauled to act as the primary get() function for obtaining system matrices

  • Significant non backwards-compatible overhaul of how constraints are defined and handled internally within the DynSys class. Constraint matrices are now held within a dict and kept local to each dynamic system. This allows for straightforward addition/removal of constraints and improves modularity.

  • Improved functionality for systems with sub-systems appended using AppendSystem() method

  • Plotting methods and response / response stats calculation methods revised to be more structured and cater for systems with multiple subsystems

  • Graphviz package used to produce call graphs for train analysis example script. This type of testing was found to be very assistive in diagnosing slow steps in the analysis and allowed overall runtime to be dramatically improved. Tests like this should be carried out periodically in the future. Dedicted subfolder made in /tests folder for this purpose.

Added

  • Validation example added to verify accuracy or revised ResponseSpectrum() function. Example computes response spectra for the classic El-Centro (1940) earthquake and makes comparison against published plots. Excellent agreement found!

  • New method PrintResponseStats() added to tstep_results class, to allow response statistics to be printed to text window in a nice manner

Fixed

  • Bug fix for ResponseSpectrum() function in dyn_analysis. Equation of motion corrected to be -M.a (minus sign previously omitted) Acceleration spectrum now based on absolute acceleration, as is conventional (previous versions omitted to add the input ground motion to the SDOF responses calculated)

Removed

  • GetRealWorldDOFs() method removed from DynSys class (this had been marked as deprecated since v0.1.0)

[0.5.1] - 2018-03-19 - RIHY

Changed

  • docs remade (forgot to do this before releasing v0.5.0)

[0.5.0] - 2018-03-19 - RIHY

Added

  • ResponseSpectrum function added to dyn_analysis module, to allow calculation of SDOF response spectrum (per seismic analysis) based on ground motion acceleration time series

Fixed

  • Bugs in plotting routines within tstep_results arising from change to response results internal storage structure
  • Miscellaneous fixes associated with None optional arguments

Changed

  • Statistics calculated using CalcResponseStats() within tstep_results module are now returned as a dict, rather than list of dicts

Removed

[0.4.0] - 2018-03-19 - RIHY

Added

  • tstep_results.WriteResults2File() method allows time-series results to be written to file. Linking to this function made in tstep and dyn_analysis class methods. Files exported have QA-relevant headers (date, version etc)

  • dyn_analysis.Dyn_Analysisand dyn_analysis.Multiple objects can now be saved and loaded using dill module (similar to pickle). This allows objects (incl results) to be stored as binary files and reloaded later. This is very useful when carrying out lots of analyses, as it allows all data to be saved and reloaded later e.g. for further manuipluation

  • Optional argument fLimit added to modalsys.ModalSys(), to allow maximum modal natural frequency to be specified. Modes with fn > fLimit will not be included in analysis (a warning message is printed to show this)

  • DynSys.PlotResponsePSD() added to implement plotting of PSDs of response time series

  • Git version control used now

Fixed

  • Code hide x_axis within tstep_results.PlotResponseResults() function removed as obsolete: subplots now use sharex=True

[0.3.0] - 2018-03-16 - RIHY

Added

  • In DynSys.__init__() option added to represent system matrices as sparse matrices (which they usually are), using scipy.sparse.csc_matrix format. Code updates implemented in all modules to handle sparse matrices. However current dense methods are used by default as this is regarded as a work in progress

  • New module tstep_analysis added. This modules contains classes to implement certain forms of time-history analysis

  • setup.py created, per PyPi guidance

  • Within dynsys and derived classes, output matrix and corresponding names no longer stored as a list of matrices / list of names, as this was proving unnecessarily confusing! Corrections made in all modules (I think...)

Fixed

  • When modalsys.AppendTMDs is used, this has the effect of increasing the nDOF of system. Need therefore to adjust structure of output_mtrx. This is now done

Changed

  • If J_mtrx=None is passed into DynSys() init function then J_mtrx is now initialised as an array of shape [0,nDOF]

  • Significant overhaul of tstep and tstep_results

  • Package version now in __init__.py as single point of reference

  • Treatment of response / output matrices in tstep_results, tstep and dynsys harmonised

  • Miscellaneous improvements / options added to results plotting routines

[0.2.0] - 2018-03-02 - RIHY

Added

  • /tests/tests_basic.py, to contain test functions

  • /tests/test_docs.py, to auto-generate documentation concerning tests

  • SDOF properties conversion functions added to dynsys for common use in other modules

  • tstep added to package, to provide time-stepping analysis functionality. This was previously a seperate package, but given inter-dependecies it makes more sense for this to be part of dynsys

Changed

  • Package folder structure extended:

    • validation subfolder created, to contain scripts which are more involved than simple unittest routines and serve to validate the library by comparing results obtained using dynsys routines against published results given in classical dynamics literature

    • docs, img, ref subfolders created in tests

[0.1.0] - 2018-02-28 - RIHY

Added

  • CHANGELOG.md (this file) created for the first time

  • README.md created for the first time

Changed

  • Package folder structure changed:

    • All modules in dynsys: import statements revised accordingly

    • Documentation in doc: this is generated by running make_docs.py

    • Tests to be in tests: this folder is for formal unittest routines

    • Examples are in examples: these are not formal tests, but are examples of the functionality offered by this package

    • References (e.g. PDFs of published papers, derivations) should be stored in the references folder. These should be linked to via docstrings.