Energy Information System Product Analysis

Introduction

Problem Statement

Buildings use more than 38% of all U.S. energy and 76% of U.S. electricity. And whether it be to reduce environmental harm, utility costs, or to meet legal requirements, managing building energy consumption is critical for building owners and facility managers. That being said, managing the energy consumption of a building can be extremely challenging.

Traditional methods of energy management rely on manual data collection and analysis, which can be time-consuming and error-prone. As a result, there is a need for an automated and integrated Energy Information System (EIS) that provides real-time energy consumption data, identifies areas of inefficiency, and offers actionable insights for reducing energy waste and costs.

The challenge is to develop an EIS that is affordable, user-friendly, and capable of integrating with existing building systems and equipment to optimize energy performance. The research I present in this post represents a portion of the energy-related Product Management work I did at WennSoft.

Breakdown

So how do you reduce the energy consumption of a building? What about utility costs specifically?

Well, it starts with identifying aspects of your building structure, equipment, and operations that are particularly inefficient. To give a quick summary of where these inefficiencies may lie:

  • Structure: Inefficient windows, doors, insulation, building envelope, etc. can result in significant heating/cooling losses.

  • Equipment: Inefficient, faulty, or unmaintained HVAC equipment/lighting can result in a significant amount of wasted energy.

  • Operations: Running HVAC equipment or keeping lights on while the building/room is unoccupied can result in a significant amount of wasted energy.

Additionally, the use of solar panels or small wind turbines onsite for renewable power generation can of course be used to sell power back to the grid and reduce overall energy costs.

While the exact engineering solutions are out of scope of this post, I will be sharing my thoughts on how an Energy Information System might optimally provide facility managers with the tools and insights they need to make improvements to their building. Specifically, I will be focusing on the electricity usage of HVAC equipment to limit the scope.

The 2012 Commercial Buildings Energy Consumption Survey estimates that a majority of energy consumption for commercial buildings comes from Heating, Ventilation and Air Conditioning (HVAC) equipment. The remainder of energy consumption goes to lighting and “plug loads” (things like electronics, refrigerators, etc. that are plugged into the wall).

An obvious conclusion from this is that HVAC equipment presents a major opportunity for reducing energy consumption and should be a priority for Energy Information Systems.

Data Methodology

Utility Data

It would be useful to extract data straight from the utility company since this is the data that is ultimately going to be used to calculate your energy bill. Fortunately, there are online resources for this like Utility API or Urjanet, which provide APIs that allow you to request energy data from utility companies. This data can range from high-level, monthly data from your energy bill, down to ~15 minute interval data. Additionally, this API can be a useful integration of utility bill data, which of course provides valuable information about how much you are actually being charged for your energy.

However, site-level energy data is only part of the story. Facility managers need some way of seeing energy use at the equipment-level so they can better prioritize improvements across their set of equipment.

Virtual Energy Meters

In a perfect world, we would have data on every single device that draws power in the building. Including when it is on, how much power it is drawing at that time, what it’s costing us in utilities, etc. Unfortunately, this data is not perfectly available.

For example, in order to get this data for an HVAC equipment, we would need a submeter on that equipment - a sensor that directly measures the flow of electricity across that equipment. This is expensive and not exactly practical or common from what I have seen with customers.

However, there are ways for us to “simulate” the monitoring of real-time energy use of HVAC equipment without using any power meters. Instead we have to take advantage of IoT sensors, which are very common in commercial buildings.

Skipping over the technical details, implementing this solution requires integrating with the Building Automation System (BAS) of your client. With this integration we can view the BAS inventory of IoT sensor data. Of which, the data most relevant to energy is:

  • Temperature data: The supply and return air temperature of the air in a room, as well as the thermostat setpoint.

  • Equipment data: The status of whether an equipment is on or off, what the fan speed is, the position of certain mechanical components, etc.

  • Occupancy data: Some buildings have occupancy sensors which can detect whether a room is occupied or not.

Now, for the purpose of our Virtual Energy Meters, we only really care about the sensors that determine when a piece of equipment is on or off. Because the logic we’re going to implement is:

IF the Equipment is ON

THEN Power = 100 kW

ELSE Power = 0 kW

…where the “100 kW” figure is the power that this equipment is consuming when it is on. For most equipment, you will actually have multiple Virtual Meters that simulate the different components of the equipment. For example, a VAV consumes energy when:

  • It’s supply fan is on → 10kW

  • It’s return fan is on → 8kW

  • It’s heater is on → 12kW

These values are typically listed on the nameplate of the equipment and documented by the facility management.

Based on this logic, we might expect a low-fidelity, binary behavior like this when we look at power over time for one Virtual Meter:

However, when we stack many Virtual Meters up to include our entire site, we might see something like the following behavior, with a more continuous curve and higher usage during working hours:

Cost & Emission Factors

While energy is obviously the “core” metric that we’re tracking, the real impact of energy consumption is the cost of the electricity and the equivalent emissions that the energy produced in the process of being generated.

For the most part, the amount that any customer spends on utilities comes down to some combination of the following types of rates:

  • Contracted Rates: Negotiated between a utility company and a customer for a fixed period.

  • Fixed Rates: Remain the same over a specified period, regardless of the amount of electricity consumed.

  • Consumption Rates: Based on the amount of electricity consumed by a customer over a given period.

    • Time of Use Pricing: The rate per kWh can vary depending on the time of day or season, with peak hours or summer months often having higher rates.

  • Demand Rates: Based on the highest amount of electricity used by a customer during a specific time period, regardless of how long it lasted.

    • Demand Ratcheting: A pricing strategy where, if a peak demand is hit, the customer is charged for that 80% of that peak for the next ~year.

  • Quantities: Refers to the minimum and maximum amounts of electricity that a customer must consume to be eligible for a particular rate.

  • Minimums: The minimum amount that a customer must pay each month, regardless of the amount of electricity consumed.

  • Maximums: The maximum amount that a customer can be charged for electricity usage over a given period.

Most residential customers will have a simple rate structure based on a fixed rate and a consumption rate. However, many commercial & industrial customers have much more complicated rate structures.

The highlighted rates above are particularly important for customers to know.

Luckily, this pricing data is available via API by companies like Genability, allowing for a seamless conversion from energy use to cost.

On the other hand, converting energy use into an equivalent amount of emissions is somewhat more straight forward. This conversion factor is simply based on the type of energy generation sources that the electricity is drawn from. Utilities will generate their power based on some proportion of coal, natural gas, nuclear, renewables, etc., which all contribute to carbon emissions at different rates. Based on these proportions, the equivalent greenhouse gas emissions per kWh can be calculated.

Companies like Persefoni can help automate this calculation.

Solutions

Energy KPIs

Based on the above information, other online research, and the feedback I’ve gotten from customers, I’ve compiled a list of the primary insights and metrics that facility managers might find valuable for tracking building energy consumption and taking action to reduce it:

  • Total energy use, utility cost, and equivalent emissions for a given a period (both at the site- and equipment-level)

  • Power use and cost rate at any given time

    • Peak demand in a period

  • Recent energy use compared to historical performance

    • Total energy, cost, or emissions reduced/increased since a previous date

  • Recent energy use compared to a reduction target

    • Total energy, cost, or emissions that still need to be reduced to hit your goal

Energy Views

For the remainder of this post, I will be describing the different features and views that would be valuable to include in an Energy Information System based on the above points.

  • A graph very much like the one I showed above that measures power at a short interval (~15 minutes). Along with power measurements, we should include a notification system that notifies users when the power demand reaches a certain point. This is to help them avoid demand ratchet charges.

  • Similarly, a graph that shows energy consumption over a period, where each point is a subtotal that is grouped by the day, week, month, etc. Ex., by month:

  • Of course, it would be useful to have this data in tabular form. Including timeseries data for all of these views, as well as a detailed view of energy use by equipment for any period of time:

  • Next, we have a similar view to the graph above, except we’re incorporating a projection of the historical “baseline”. The baseline is defined by a machine learning model that is trained on at least a year’s worth of energy, temperature, and occupancy data. It takes inputs of heating degree days, cooling degree days, and # of occupied hours of the building, and outputs the predicted daily energy use. In the example below, a major equipment improvement was made on 7/1/22 and a reduction in the actual energy use can be seen. By comparing actual performance to our baseline, we can assess the perceived impact of equipment/performance improvements (listed to the right):

  • Lastly, users may have an energy reduction goal - “We want to reduce our energy use by 10% by the end of the year.” By incorporating this goal into our baseline model, we can create a energy use forecast that decreases to the specified target by the end of the year. Using this target line, we can compare our current energy performance to this target to assess progress.