# DN Grid Sim

A multi-component power system simulator for testing the ΔN–ΔD approach in operating mode control.

## Project Idea

This project tests the following working hypothesis:

taking into account not only the external power imbalance, but also the internal structural heterogeneity of the current operating mode, makes it possible to keep a power system in a more stable operating state, especially under disturbances and stress conditions.

Conventional control schemes mainly focus on already manifested imbalance:

- insufficient generation;

- load growth;

- renewable output drop;

- the need to redistribute generation and reserves.

This project adds one more layer: the internal tension of the operating mode itself:

- how unevenly the sources are loaded;

- how abrupt and conflicting the control actions become;

- how local imbalances accumulate across nodes.

This internal heterogeneity is represented by the parameter ΔD.

## What Is Simulated

The simulator implements a simplified but engineering-interpretable multi-component power system including:

- base generation;

- flexible generation;

- solar generation;

- wind generation;

- battery energy storage;

- nodal loads;

- a simplified network structure with local balances.

The model works in discrete time and accounts for:

- minimum and maximum power of sources;

- ramp-rate limits;

- battery power and state-of-charge constraints;

- renewable curtailment;

- balancing costs;

- critical operating conditions;

- unserved energy.

## Control Modes

Three control modes are compared.

### 1. Baseline

A simple reactive logic.

The system reacts to an already emerged deficit or surplus:

- activates flexible generation;

- uses storage;

- curtails part of renewable output in case of surplus.

This is a simplified analogue of rule-based control.

### 2. ΔN-only

Control based only on external non-equilibrium.

The system responds to:

- power imbalance;

- load variation;

- renewable fluctuations.

But it does not account for the internal heterogeneity of the operating mode.

This mode is included to test an important question: is it enough to consider only the external stress of the system?

### 3. Full DN

Control based on two factors simultaneously:

- **ΔN** — external stress of the operating mode;

- **ΔD** — internal structural heterogeneity of the operating mode.

This mode is the main target of the project.

## Meaning of ΔN and ΔD in the Power-System Context

### ΔN

External non-equilibrium of the operating mode.

In the power-system interpretation, this is related to:

- generation-load imbalance;

- load change rate;

- renewable output change rate;

- external disturbances of the operating mode.

In simple terms, ΔN answers the question:

**how strong is the external stress currently acting on the system?**

### ΔD

Internal heterogeneity of the operating mode.

In the current implementation, ΔD is linked to:

- uneven source loading;

- heterogeneity of control actions;

- local nodal imbalances.

In simple terms, ΔD answers the question:

**how internally distorted, abrupt, and tense has the operating mode itself become?**

## What Is Being Tested

The project does not aim to replace a full dynamic power-system model.

Its goal is to test whether explicitly accounting for the internal tension of the operating mode provides a practical advantage over:

- simple reactive logic;

- control based only on external imbalance.

The main metrics are:

- unserved energy;

- number of critical episodes;

- balancing cost;

- cumulative imbalance;

- number of switching actions;

- average and accumulated structural tension of the mode.

## Scenarios

The simulator supports several scenarios:

- normal

- stress\_test

- high\_res

- critical\_zone

At the current stage, the key scenario is stress\_test, because it shows the difference between the approaches most clearly.

## Main Results

At the current stage, batch runs were performed for 30 seeds for the normal and stress\_test scenarios.

### Stress Test Scenario

Average results over 30 runs:

#### Baseline

- unserved_energy_mean: 56.82

- critical_episodes_mean: 32.33

- balancing_cost_mean: 93.95

- sum_imbalance_mean: 307.20

- switches_mean: 55.07

- mean_dsdt_mean: 1.93

- integral_S_mean: 184.81

#### ΔN-only

- unserved_energy_mean: 147.85

- critical_episodes_mean: 51.23

- balancing_cost_mean: 92.02

- sum_imbalance_mean: 192.39

- switches_mean: 192.80

- mean_dsdt_mean: 2.18

- integral_S_mean: 209.34

#### Full DN

- unserved_energy_mean: 9.58

- critical_episodes_mean: 5.87

- balancing_cost_mean: 81.65

- sum_imbalance_mean: 66.40

- switches_mean: 177.73

- mean_dsdt_mean: 1.49

- integral_S_mean: 142.93

### Interpretation of Results

The obtained results show the following.

#### 1. Full DN strongly reduces unserved energy

Compared to baseline and especially compared to ΔN-only, the system covers the load much better under stress conditions.

#### 2. Full DN sharply reduces the number of critical episodes

This means that the system enters dangerous operating conditions much less often and remains inside the working region more reliably.

#### 3. Full DN reduces balancing cost

The improvement is achieved not at the price of more expensive balancing, but together with a reduction in balancing cost.

#### 4. Full DN strongly reduces cumulative imbalance

This means that the system passes through the stress scenario in a much more even and stable way.

#### 5. ΔN-only performs systematically worse

This is one of the key findings of the project.

Taking into account only the external imbalance is not enough.

If the internal heterogeneity of the operating mode is ignored, the system enters dangerous states more often and covers the load worse.

## Practical Conclusion

The main result of the project is the following:

**taking into account the internal heterogeneity of the operating mode together with the external imbalance gives a better result than simple reactive logic or control based only on external stress.**

In engineering terms, this means that a power system should account not only for the deficit or surplus itself, but also for the quality of the operating mode through which balance is being maintained.

This is exactly what distinguishes the DN approach from more traditional logic.

## Current Limitation

Despite the strong result, the current version has an important limitation:

### Increased Number of Switching Actions

In Full DN mode, especially in the calm scenario, the number of switching actions and corrective interventions is high.

This means that the controller already maintains reliability well, but still does so too actively.

The next engineering task is to reduce unnecessary switching without losing the achieved improvement in:

- unserved energy;

- critical episodes;

- balancing cost;

- cumulative imbalance.

## Project Structure

```text

dn-grid-sim/

│

├── core/

│ ├── dn_model.py

│ ├── controller.py

│ └── metrics.py

│

├── grid/

│ ├── assets.py

│ ├── network.py

│ └── simulator.py

│

├── scenarios/

│ ├── normal.py

│ ├── high_res.py

│ ├── stress_test.py

│ └── critical_zone.py

│

├── experiments/

│ ├── run_baseline.py

│ ├── run_deltaN_only.py

│ ├── run_full_dn.py

│ ├── compare_all.py

│ ├── batch_run.py

│ └── outputs/

│

├── visualizer/

│ └── plots.py

│

├── config/

│ └── base.yaml

│

├── README.md

└── requirements.txt