1.0 Executive Summary

A simulation example is presented in this paper, which demonstrates how BuildingIQ’s platform can be used to optimize energy consumption in a building, while simultaneously managing the maximum demand conditions preset by policy or regulation. A stiff fine is generally imposed on building owners for exceeding maximum demand.

(In the USA, regulation sets the billing variable for demand charges which can make up 40% of the monthly electric bill.) The BuildingIQ optimization process is automated and relies upon pre-cooling the building, notably shifting the cooling load to the early morning hours, when ambient temperatures and humidity are lower and power is cheaper. It then lets the temperature of the building drift slowly upward— cooling in moderation—as outside ambient temperatures climb. Shifting and moderating the cooling load can keep power consumption below the maximum demand limitations even at the hottest time of the day. The optimization process runs counter to the traditional practice of static temperature control throughout the day, using a fixed setpoint with little to no coordination between zones/rooms.

2.0 Introduction and Background

The Cloud-based Predictive Energy Optimization™ (PEO) platform is a paradigm-shifting technology used to reduce energy consumption in commercial, academic, and government buildings. The multiobjective optimization process involves three basic steps: first learning the thermal dynamics of the building; second, refining the model through the slow convergence between parameter prediction and actual readings from the building; and finally optimization, in terms of both energy efficiency and cost.

The status quo for today’s HVAC technology includes temperature control that is static and fragmented; that is, building temperatures directed by a fixed setpoint, and HVAC scheduling that is independent for each zone/room. This means, there is no flexibility and no coordination between zones/rooms.

There remains a general failure to appreciate that setpoints and comfort bounds can be moved throughout the day without occupants experiencing discomfort, as shown in the psychometric research by CSIRO1 / ASHRAE2. Adding degrees of freedom to move temperatures around leads to enormous opportunities to optimize energy efficiency and cost.

In huge buildings, for example, a temperature shift of as little as 0.5 degree Celsius (.9F) can save massive amounts of energy.

Without the ability to constrain demand below a certain amount can lead to large fines. In some places, for example, where rates average around 10 cents/kWh, demand charges over the maximum level can soar to $10 – $15/kW. Charges also vary by Time of Use (TOU), with price per kilowatt-hour rising during peak demand periods and falling during low demand periods.

3.0 General Principles

The BuildingIQ optimization model can be programmed, based on the building’s changing max demand limit, to respect a firm upper limit on power consumption. The rate of power consumption is then kept below the max demand limit by providing “wiggle room” for the internal temperatures, notably by allowing temperatures to fluctuate within a tolerable band. These additional degrees of freedom open up the possibility of load shifting as a central strategy by varying internal temperature throughout the day but while ensuring internal occupant comfort. As will be shown in the simulation, efforts to hold temperatures constant as outside/ambient temperature rise throughout the day can drive power consumption through the roof, making demand management nearly impossible.

The recommended answer is to use pre-cooling –while power is cheap– and to do so in a more gradual manner than is typical of more rigid fixedpoint approaches. Similarly, allowing the building temperature to drift upward gradually – when power costs are high– to the maximum comfort level allows power consumption to be moderated and held below the max demand. The use of cooling energy is thus minimized at times when max demand charges and cost of cooling are at their highest.

4.0 Simulation Example

The simulation example below shows the results before and after implementation of BuildingIQ’s optimization software. Figures 1 and 2 show the results over the first 12 hours of the day under the unrestrained conditions, before BuildingIQ assumes management, while Figures 3 and 4 show the simulated conditions after BuildingIQ implementation. The key objective is to find the most efficient temperature and power consumption profile over time without exceeding the Maximum Demand Limit, set in this example at 500kW.

Managing Maximum Demand Through Optimization

Managing Maximum Demand Through Optimization

4.1 Simulating the Before Conditions

In Figure 1, the green dashed line and the red dashed line show the pre-defined temperature boundaries. This is typically defined in a contractual agreement with the building owner that says when the building is unoccupied the HVAC operator can allow the temperature to rise to 30 degrees Celsius (86 Fahrenheit,) but when occupants start to arrive at around 7:30am (0.3 day) the temperature must be brought to between 21-24 degrees Celsius (~7075 Fahrenheit). This is the “degree of freedom,” the flexibility allowed to the HVAC operator.

The red solid line with squares in Figure 1 corresponds to the set point signal sent to the building’s HVAC system. It begins to cool rapidly around 6:00am (.25 day), so that the building temperature is driven to within the contractual min/ max bounds during occupancy.

The dashed gray line is the ambient temperature outside the building. It reaches a minimum level at around 21-22 degrees Celsius (~70-71.5 Fahrenheit) around 5:00am (.2 day). It then begins to climb rapidly during the day., reaching nearly 40 degrees Celsius (100 Fahrenheit) by noon (.5 day).

The corresponding power profile is shown in Figure 2. The dashed purple line is Maximum Demand Limit, set for purposes of this simulation at 500kW.

The black line shows total power, the blue line shows the cooling component of total power, which can range from as high as 80% of total power to as low as 50%. The green line is called baseload power, which is shorthand for the predictable, nonvarying component of power load, such as that
used for computers, equipment, and lighting.

Power consumption surges around 6:00am as the HVAC operator cools the building rapidly, driving total power above the Maximum Demand Limit for roughly 20 minutes before dropping back. However, by 10:00am, the total power consumption once again exceeds the Maximum Demand Limit and does not stop climbing. This is a result of the cooling plant keeping the temperature setpoint constant while the power consumption rises as the ambient temperature increases. The economics become untenable for the building owner, likely forcing load curtailment for some tenants.

4.2 Simulating the After Conditions

Using the basic architecture of BuildingIQ’s PEO model, max demand optimization can be achieved while adhering to the requirement that total power consumption never go beyond the Maximum
Demand Limit.

The means for doing this can be seen in the slope of the red line in Figure 3. The rate of cooling begins at 6:00am as it did in the “before” simulation; however, it is done at a slower rate. This keeps the total power consumption from exceeding the Maximum Demand Limit as shown in Figure 4.

Managing Maximum Demand Through Optimization

Managing Maximum Demand Through Optimization

In addition, following the gradual pre-cool, the set points are then moderated so that the temperature in the building (the red solid squared line) is allowed to rise slowly during the morning hours until it reaches the contractual maximum (green line) around noon, when thereafter it is stabilized.

These seemingly small changes in HVAC operational dynamics have important economic consequences.
As show in Figure 4, the power consumption required for cooling is considerably restrained. The 6:00am surge in power was cut from a spike of around 450kW to around 300kW, allowing total power to remain below the Maximum Demand Limit. After the building is cooled and ready for occupancy, the cooling requirements drop off but are maintained at a moderate level (about 300 kW) for about two hours before the building temperature is allowed to coast upward. The coasting enables the building to minimize energy consumption, while ensuring

occupant comfort. In this simulation of a 12-hour scenario, total power never exceeds the Maximum Demand Limit.

5.0 Approach to the Customer

A number of large facilities, ranging from university campuses to major commercial, medical, or
government complexes are constrained in how much power they can use. They haven’t the luxury of increasing consumption by adding another power station. They are capacity constrained, which makes combining energy efficiency with the management of Maximum Demand Limits particularly important.
BuildingIQ’s algorithms can do this automatically, and still maintain occupant comfort.
The key advantage of BuildingIQ’s energy management solution is the ability to intelligently load-shift energy consumption, based on the knowledge of building thermal and power
dynamics (obtained during the learning period), as well as weather forecasts, in order to constrain power consumption below the max demand limit as imposed by the utility. Our machine-learning algorithm does this automatically.

“Learning the building” is inherent in BuildingIQ’s model. The model has the capability of learning the thermal properties and dynamics of the building based upon materials, mass, and physics. Key questions such as how fast the building cools, or conversely, how fast a cooled building converges to the ambient temperature are key. Also, what does the power response of your building look like during cooling? The building physics and power profile, in addition to max demand and time-of-use charges, are all factored within BuildingIQ’s predictive optimization algorithms, which then optimizes building operations to minimize energy costs while ensuring occupant comfort.

The value of automation is such that operators do not have to standby, fine-tuning the HVAC system as the day proceeds. Lack of automation is considerably more expensive and not as effective.
The BuildingIQ system is a learning system, which means efficiency grows as the building dynamics and occupancy patterns are better understood.

  1. Commonwealth Scientific and Industrial Research Organisation, Australia’s national science agency and one of the largest and most diverse research agencies in the world.
  2. American Society of Heating, Refrigerating and Air-Conditioning Engineers – a global society advancing human well-being through sustainable technology for the built environment.

Site Reference: https://buildingiq.com/resources/case-studies/