Release History​

[1.3.0]​

Breaking Changes​

• Replaced specificAllocationInstance with allocationInstanceSelection in InterventionConfig, which allows selection of multiple allocation instances for consideration when interventions.
• InterventionConfig in the work package request has been updated to support multiple concurrent interventions.
• TARIFF_REFORM and CONTROLLED_LOAD_HOT_WATER interventions have been replaced by LOAD_RESHAPING which is configured via InterventionConfig.loadReshaping.

New Features​

• Add a StreamDeletionFailed event to track stream deletion statistics.
• Added applied_interventions table to output the applications of the following interventions:
  • Community BESS
  • LV STATCOMs
  • Distribution transformer tap optimization
  • OLTC
• Added config flags in the ModelConfig for controlling undersized LV line fixes and load-data window filling:
  • model.autoFix.fixUndersizedLvLines, to configure whether to upgrade LV lines above large customers.
  • model.windowFillMissingLoad, to configure whether missing load is filled from the last available reading at the same time of day.
• Added support for sizingLookaheadYears to Criteria CandidateGenerationConfig.

Enhancements​

• Update container lifecycle tracking on the Kubernetes runtime to better reflect the actual container lifecycle.
• Make RabbitMQ queue and stream deletion more reliable with confirmations and retries.

Fixes​

• Fix an issue with permits being temporarily lost (for the timeout period) when a container failed ungracefully.
• Fix an issue with permits becoming lost due to the Kubernetes API not timing out.
• Redact sensitive configuration values.
• Time out work packages that never start (e.g. due to a lost message), instead of leaving them stuck indefinitely.
• Updated Sqlite Input database schema to match the postgres schema.

Notes​

• The intervention configuration stored in the work_packages table is now clear (set to null) on completion of the work package.

[1.2.0]​

Breaking Changes​

• None.

New Features​

• None.

Enhancements​

• Update stream deletion to delete non-active work package streams in small batches.

Fixes​

• Update monitors to automatically restart on failure.

Notes​

• None.

[1.1.0]​

Breaking Changes​

• Updated RunWorkPackage endpoint use the new work package request format.
• PRRP must accept the new feeders field.

New Features​

• Added support for intrinsic work packages.
• Intervention work packages now pass through the list of feeders into PRRP for filtering.

Enhancements​

• Expose intrinsic permits through existing permits_available metric.

Fixes​

• None.

Notes​

• None.

• Support up to com.zepben.commons and com.zepben.ewb v1.1.0

Fixes​

• None.

Notes​

• None.

[1.0.0]​

Breaking Changes​

• Configuration changes including new required resource configs hcsDb for hosting capacity database and storageBackend for cloud blob storage, change rabbitmq to a single configuration, removed workPackageInactivityTimeout and messageConsumerTimeout.
• Dropped and recreated the qa_event_log table in the results database. The table now include a new id column in the primary key to uniquely identify events within the execution of a SYF. The second column has also been changed from Integer to Double.
• Dropped the following columns on the input.bess_instances table:
  • kw_band_discharging
  • kw_target_charging
  • kw_target_discharging
• Renamed the kw_band_charging column on the input.bess_instances table to percentage_rated_kva_deadband, to be used for the deadband on both charging and discharging targets.
• Updated to supply unified runtime configuration to model generator.
• Updated to supply unified runtime configuration to model executor.
• Updated to supply unified runtime configuration to results processor.
• Updated to supply unified runtime configuration to Post Results Results Processor (Prrp).
• Minimum required JVM version for compilation is 17
• Updated Succeeded to Completed in WorkPackageDetailsFetcher - wpState to match updated enum WorkPackageState from its dependency commons.

New Features​

• Added new wp-details endpoint to retrieve detailed progress metrics (counts by status) and model groupings for work packages.
• Added new wp-active endpoint to retrieve the currently active (started, pending, running) work packages.
• Added configuration option to enable load fallback functionality for PQV based models.
• Added new field inputDb.forecastEquipmentCommissionMonthDay to configuration. This value is passed to the model generator as a command line argument --forecast-equipment-commission-month-day.
• Support optional envRef's on Kubernetes container config templates.
• Add a config option notificationHandler.parallelism to configure maximum notification handler paralllelism
• Added config option nominalLowVoltage to configure nominal phase-to-phase voltage for LV customers, in volts. It is 400 by default.
  • The version of PRRP must be v0.3.0 or higher to support this new config option.
• Added the following columns to the input.bess_instances table to support enhancements to the community BESS intervention:
  • percentage_idling_kw
  • phase
  • percentage_rated_kva_target_charging
  • percentage_rated_kva_target_discharging
• Added negligibleImpedanceMinHvThreshold and negligibleImpedanceMinLvThreshold to ModelConfig

Enhancements​

• Major overhaul to the internal workings to make hosting capacity service
  • Durable
  • Remove the dependency between work package configuration and the total number of queues in RabbitMQ
  • Support concurrent execution of multiple work packages
• No longer explicitly delete Kubernetes jobs when stop() is triggered, and utilise automatic cleanup. This allows logs to be collected before the job is removed.
• Allow the RabbitMQ client to prefetch any number of notification messages
• RabbitMQ client will now retry the initial connection attempt 3 times rather than exiting after one failure.
• Delete RabbitMQ streams for failed results processors when they fail
• Tag created containers with their corresponding SYF information
• Support Kubernetes 1.35 as a container runtime

Fixes​

• Fix hosting capacity database deadlock
• Fix notifications being discarded from the queue on notification handler exception
• Fix notifications becoming stuck in the queue
• Discard notifications that cannot be deserialized

Notes​

• Kubernetes jobs now automatically cleanup after 15m rather than 1h
• Updated project to Kotlin 2.3.20

[0.6.1]​

Breaking Changes​

• Dropped and recreated the qa_event_log table in the results database. The table now include a new id column in the primary key to uniquely identify events within the execution of a SYF. The second column has also been changed from Integer to Double.

New Features​

• Added new field inputDb.forecastEquipmentCommissionMonthDay to configuration. This value is passed to the model generator as a command line argument --forecast-equipment-commission-month-day.

Enhancements​

• Add configuration for requesting pqv-fallback functionality from the model processor.

• All timestamp strings being parsed to Kotlin/Java types must include seconds. This is due to the update of kotlinx-datetime to 0.6.1.

• Result queues are now assumed to be streams, so a compatible version of opendss-jvm-executor must be used.

• RabbitMQ instance must use the rabbitmq_stream plugin and the stream.advertised_host and stream.advertised_port must be configured. resources.rabbitmq.streamPortForContainers in the HC service config must match stream.advertised_port.

• Configuration change from hosting-capacity-utils

  • Added new configuration class SyfConfig to group variables moved out of WorkPackageRequest and WorkPackageConfig.
    • ForecastConfig - Functionally similar to old set up where it holds a list of feeders, years, scenarios and load time.
    • FeederConfigs - Holds a list of individual feeder settings so each feeder can run a hosting capacity study with distinct settings.
  • Added new configuration class LoadTime and LocalLoadTime which holds load specific variables.
    • FixedTime/LocalFixedTime - Fixed time setting for the study, includes an optional map of FixedTimeLoadOverrides.
    • TimePeriod/LocalTimePeriod - Time period setting for the study, includes an optional map of TimePEriodLoadOverrides.
  • Major refactoring of configuration.
    • WorkPackageRequest
      • Moved feeders, years, scenarios, timePeriod and fixedTime into SyfConfig and LocalLoadTime.
    • WorkPackageConfig
      • Moved feeders, years, scenarios, timePeriod and fixedTime into SyfConfig and LocalLoadTime.
    • CommonConfig
      • Moved timePeriod and fixedTime into LoadTime.
    • ModelConfig
      • Moved fixedTimeLoadOverride and timePeriodLoadOverride into LoadTime and LocalLoadTime
    • Introducing WorkPackageInput change and access of load time.

New Features​

• Support intervention
  • PRRP is run at the start of a work package (before model generation) if intervention.cadidateGeneration is specified in the work package's request.
  • Intervention config is passed through to the model generator via the Job protobuf message.
• Schema migration scripts for load, result and input db.
• From hosting-capacity-utils, volumes can now be specified for docker container templates. Notably useful when combined with the config field generator.cli.outputModelsPath, by setting generator.containers.volumes.<volume name> to the same directory name e.g. /opendss-models.
• Support applying transformer tap settings from input db by specifying a transformerTapSettings in the model config when building models.
• Added new config option ewb.accessToken to allow authenticating to EWB with an EAS generated AccessToken.
• Added new table calibrated_taps to the input database to store transformer tap settings for use by the model generator.
• Create new end point for hosting capacity work package cost estimation /v1/estimate
• Added new table calibration_meter_results to the results database to store meter level voltage results from a calibration run.
• Takes in FixedTimeLoadOverride or TimePeriodLoadOverride in appConfig to allow load profile overriding for specified meters.

Enhancements​

• Supports kubernetes image pull secrets in container template configuration.
• Added support for passing a ewb authentication issuer and audience to the model generator.
• Can now specify service account name in the template for kubernetes containers.
• Now the service will clean up queues left over by work package failures.
• Models output from the generator to disk are now suffixed by the work package ID.

Fixes​

• Fixed the configured client secret for authenticating with EWB using M2M auth not getting passed to the opendss-generator.
• Fixed issue where the hosting capacity service did not support config outputs from zconf.
• Fixed issue where you had to specify both cpu and memory for kubernetes resource limits and request.
• Cluster autoscaler will no longer evict jobs created by this service when using Kubernetes

Notes​

• Changelog for upcoming load schema placed in resources/db/changelog/load2. Work still needs to be done to support generic aggregations.

[0.6.0]​

Breaking Changes​

• Renamed configuration property maxActivePipelines to maxActiveWorkPackages as part of the refactor to support running multiple work packages concurrently.

New Features​

• Support running multiple work packages concurrently. This simplifies the service's architecture and allows for more efficient use of RabbitMQ resources.
• Added new table node_level_results to the results database to support node level opendss results.
• Added useSpanLevelThreshold, simplifyPLSIThreshold, spanLevelThreshold, and emergAmpScaling to ModelConfig
  • simplifyPLSIThreshold is required to Enable ImpedanceNormalizre in NetworkSimplifier
  • spanLevelThreshold can be supplied to CommonImpedanceCollapsor to include DesignedRating of Conductor to be considered during AcLineSegment collapse.
  • if useSpanLevelThreshold is set to true, designRating will be used during current comparison during CommonImpedanceCollapsor instead of ratedCurrent
  • emergAmpScaling is used to calculate the emergAmps of LineCode. It was previously hard set to 1.5 times but now exposed as a configuration.
• Added new table inverter_control_profiles to the input database to support translating generators to PvSystems in OpenDss(Advanced PV).
• Containers started using the Kubernetes Runtime are now also labelled with the work package id

Enhancements​

• Support Super-Pom 0.45.1
• Support up to com.zepben.commons and com.zepben.ewb v1.1.0

Fixes​

• None.

Notes​

• None.

[0.5.1]​

Breaking Changes​

• Rework load schema with a new set of Liquibase migrations, starting with version 0.5.0. This new schema was previously placed in the load2 folder, which has since been removed. You will need to create a new database (possibly in the same Postgres server) and point to it for the new schema, then migrate any data from old tables (e.g. network_hierarchy).
• PQV schema has been updated to be long form with more types of values supported. This will break anything using the pqv schema.
• Existing calibration_meter_results table will be dropped and recreated with a new schema. Deleting any existing data in the table.

New Features​

• None.

Enhancements​

• Support up to Super-Pom 0.39.0

Fixes​

• None.

Notes​

• None.\n\n---\n

[0.5.0]​

Breaking Changes​

• None.

New Features​

• None.

Enhancements​

• None.

Fixes​

• Support configuration of impedance collapsing and span thresholds for ratings.

Notes​

• None.\n\n---\n