# Reliability

DERs can provide backup power to critical loads. For example, in the aftermath of natural disasters when an entire region might go out of power, DERs can provide backup power to traffic lights, schools, hospitals, supermarkets, and gas stations to maintain basic functions. The scheduling of the device for this purpose can require that a minimum energy capability is maintained during normal operation so it is ready for an emergency.

Reliability is a constraint service that applies a minimum state of charge constraint on energy storage systems so that the DERs can be ready to completely cover any unforeseen outage. The minimum SOC constraint dynamically adjusts to load and PV conditions to allow for as much flexibility (and therefore economic potential) as possible while still guaranteeing coverage of all critical load during any grid outage. If the post-facto only option is selected, then no constraints will apply to the main optimization but the reliability by-products of the economic optimization will be calculated.

# Inputs

## Target

Tag | Key | Description |

Reliability | target | The reliability service will dynamically ensure there is enough power and energy capacity in the DER mix to cover critical load for this number of hours in case of an unforeseen grid power outage.
Note: this input should be evenly divisible by dt This input is ignored when post_facto_only = 1. |

## Post-Facto Only Reliability Initial SOC

Tag | Key | Description |

Reliability | post_facto_initial_soc | This input is only used under very specific circumstances when reliability is the only active service (The GUI requires at least one energy service, so this option is for the command line version only). In this case, the reliability contribution of fixed-size DERs can be calculated by assuming that storage systems start any random outage with the SOC defined by this input.
This input is ignored when there is an SOC profile from the optimization to use (this will be most cases). |

## Post-Facto Only

Tag | Key | Description |

Reliability | post_facto_only | This binary input determines whether or not reliability will be considered in the optimization.
If post_facto_only=0, then reliability will be considered in the optimization. In other words, DERs will be optimally sized and operated to achieve the reliability objective. If post_facto_only=1, then reliability will be treated as a post-optimization calculation and will have no impact on the optimization itself. In this case, the operational profile for DER will be determined in the optimization and used to calculate the reliability outcomes unless reliability is the only service. If reliability is the only active service, the post_facto_initial_soc will be used to calculate reliability. |

## Maximum Outage Duration

Tag | Key | Description |

Reliability | max_outage_duration | This input is only used to generate the load_coverage_probability.csv output file by setting an upper limit on the outage duration considered in the reliability calculations. Setting this number higher results in some extra calculation time. |

## Load Shed

Tag | Key | Description |

Reliability | load_shed_percentage | This binary input is used to determine if the load shedding schedule defined in the input file load_shed_percentage.csv is applied. If this equals 0, then the full critical load will be served for the duration of every outage. |

## Load Shed Percentage Filename

Tag | Key | Description |

Reliability | load_shed_perc_filename | This file name input points to a load shed percentage file. |

# Size Optimization Approach

Reliability is a constraint service which applies a minimum SOC time series constraint to energy storage systems. This constraint is not sufficient to calculate the optimal DER mix to meet a reliability objective if a combination of storage, intermittent generation, and traditional generation is included. So, the reliability service implements its own size optimization that runs before the core DER-VET optimization. This is a simple optimization that simulates the operation of all DERs during any possible outage simultaneously with optimizing the size of all DERs based on their capital costs. Because this optimization does not consider financial benefits, ongoing costs, or anything besides capital costs, the normal DER-VET optimization is executed after the reliability-only optimization. DERs can be oversized by the core DER-VET optimization, but the core optimization cannot reduce the size of any DER determined by the reliability-only optimization.

## Reliability-Only Optimization

The number of possible outages in a year can be calculated by [(# timesteps in the year) - (# timesteps in a single outage) + 1]. If we have 15-minute data and are considering 4-hr outages, there are 35,025 potential outages, each of which contain 16 time steps. If DER-VET were to solve a simultaneous size optimization problem considering the size and dispatch of all DERs in every potential outage, there would be (# size variables) + (# dispatch variables per time step) * 35,025 possible outages * 16 timesteps per outage. This optimization would take far too long to solve, so some simplifications are included to reduce the size of this problem.

Firstly, only the **top 10 outages** by the cumulative energy in the critical load during the outage are considered. Because of the complexity of multi-DER microgrids, the coincidence between solar generation and load, and other factors, these selected outages might not include the worst-case outage (read: the outage that takes the highest capital cost to cover).

Secondly, the detailed technology and cost models implemented in the main DER-VET optimization are reduced to only the core operational and size features - no minimum generation, no non-curtailable PV, no variable costs or power loss besides roundtrip efficiency, etc.

# Dispatch Optimization Constraints

Whether sizing equipment or not, DER-VET will run a normal optimization after determining the minimum DER size for reliability (or have the sizes provided by the user). During this normal optimization, the state of all storage systems will be constrained to be high enough to cover any outage in the year that lasts the target duration.