Theory & Background

Survey all electrical loads and classify by curtailability: Critical (cannot be interrupted — servers, medical equipment, safety systems), Essential (short interruption acceptable — lighting, ventilation), Sheddable (15-30 min interruption acceptable — EV chargers, non-urgent HVAC), Flexible (timing-flexible — water heaters, refrigeration, dishwashers). Assign priority 1–4 accordingly. Map each load to control hardware (smart plug, relay, or BMS interface).

Components & Requirements

10 components required for this project.

#	Component	Purpose	Qty
1	Raspberry Pi 4 (DSM Controller)	Central optimization engine	x1
2	Smart Plugs with Power Monitoring	Load control and measurement points	x8
3	Clamp Current Meter (Main Feed)	Total building demand monitoring	x1
4	Demand Controller Relay Panel	HVAC and industrial load control	x1
5	DS18B20 Temperature Sensors	Room temperature for HVAC optimization	x4
6	Occupancy Sensors (Zigbee)	Presence-based load control	x4
7	4G Router for Utility API	Demand response program integration	x1
8	Touchscreen (7" DSM dashboard)	Real-time demand display and override	x1
9	Historical Data Server (local)	Pattern learning and forecasting	x1
10	Real-Time Pricing API Integration	Tariff-based optimization input	x1

Step-by-Step Implementation

Follow these 3 steps carefully.

1

Load Classification and Priority Assignment

Survey all electrical loads and classify by curtailability: Critical (cannot be interrupted — servers, medical equipment, safety systems), Essential (short interruption acceptable — lighting, ventilation), Sheddable (15-30 min interruption acceptable — EV chargers, non-urgent HVAC), Flexible (timing-flexible — water heaters, refrigeration, dishwashers). Assign priority 1–4 accordingly. Map each load to control hardware (smart plug, relay, or BMS interface).

2

Demand Threshold Algorithm

Set a maximum demand threshold (e.g., 80% of contract demand to avoid excess demand charges). Monitor instantaneous demand every 5 seconds. When demand approaches threshold (>90%), trigger demand response sequence: shed lowest-priority loads first (EV charger → pool pump → secondary HVAC zones → lighting dimming). Restore loads in reverse order when demand drops below 75% threshold. Implement minimum off-time (15 minutes) for HVAC equipment protection.

3

Machine Learning for Load Forecasting

Train a gradient boosting model (XGBoost) on 6 months of 15-minute interval demand data. Features: hour of day, day of week, day of year, temperature, humidity, occupancy, calendar events (public holidays, school terms). Target: 15-minute ahead demand in kW. Achieve MAPE < 8% for 1-hour ahead forecast. Use forecasts proactively: if model predicts peak demand period, pre-cool building and pre-charge batteries 30 minutes before.

Code & Implementation

Core code for dsm_optimizer.py:

dsm_optimizer.py
  Python

import numpy as np from datetime import datetime  class DemandOptimizer:     def __init__(self, threshold_kw=45):         self.threshold = threshold_kw         self.loads = {           "ev_charger": {"kw": 7.2, "priority": 4, "min_off": 15, "state": True},           "hvac_zone2": {"kw": 3.5, "priority": 3, "min_off": 15, "state": True},           "pool_pump":  {"kw": 2.5, "priority": 4, "min_off": 30, "state": True},           "hot_water":  {"kw": 4.0, "priority": 3, "min_off": 60, "state": True},         }      def optimize(self, current_demand_kw):         if current_demand_kw < self.threshold * 0.9:             self.restore_loads()             return         excess = current_demand_kw - self.threshold         # Shed loads by priority (highest number first = lowest priority)         for name, load in sorted(self.loads.items(),                                   key=lambda x: -x[1]["priority"]):             if excess <= 0: break             if load["state"]:                 print(f"Shedding {name} ({load['kw']}kW)")                 load["state"] = False                 excess -= load["kw"]      def restore_loads(self):         for name, load in sorted(self.loads.items(), key=lambda x: x[1]["priority"]):             if not load["state"]:                 print(f"Restoring {name}")                 load["state"] = True

Testing & Troubleshooting

Test Demand Side Management System by verifying each subsystem individually before full integration.

!

Troubleshooting Tips

Verify power voltages, check ground connections, use serial monitor for debug.

Real-World Applications

*Commercial building peak demand management

*Industrial plant energy cost optimization

*Campus-wide energy management systems

*Virtual power plant aggregation for demand response

*Utility demand response program management

*Data center power usage effectiveness improvement

*Hotel and hospital energy optimization

*Smart city district energy management

Extensions & Next Steps

Integrate battery storage for peak shaving with financial optimization
Implement real-time electricity market price optimization
Add distributed energy resource management (solar + battery + EV)
Build a demand response aggregator serving multiple buildings
Implement blockchain-based peer-to-peer energy trading

Interactive Playground

Coming Soon

An interactive simulator will be available here — simulate circuits and run code in-browser without hardware.

Frequently Asked Questions

How much can DSM reduce electricity bills?

DSM savings depend on tariff structure and load flexibility. For commercial consumers with demand charges (typically ₹100–400/kVA/month), reducing peak demand by 20% can cut monthly bills 15–25%. For time-of-use (ToU) tariff users, shifting flexible loads to off-peak hours (11 PM – 7 AM, typically 40–60% cheaper per kWh) can save 20–35% on energy charges. Combined with demand response incentives from utilities, total bill savings of 25–40% are achievable.

Introduction