I85PREP
⌘ K
CEM study materials
CEM study resources

CEM: Building Automation, Controls and Artificial Intelligence Systems

CEM · Work through your exam, one topic at a time.

All guide sections

Building Automation, Controls and Artificial Intelligence Systems

This domain tests control architecture, sequences, scheduling, fault detection, and the realistic use of AI in buildings. Questions often ask you to estimate scheduling savings or identify why a control sequence is wasting energy.

Architecture and protocols

  • Field devices include sensors, actuators, and controllers. DDC controllers run control loops; supervisory systems trend, schedule, and coordinate.
  • Common protocols include BACnet, Modbus, LonWorks, and KNX. Gateways translate protocols but can lose points or add latency.
  • PID control uses proportional, integral, and derivative terms. Poor tuning causes hunting, offset, or slow response.
  • Standard strategies include optimal start/stop, night setback, supply-air reset, static-pressure reset, demand-controlled ventilation, economizer lockout, and load shedding.

Worked example

A 100 kW supply fan runs continuously, 24 hours per day, but the space is occupied only 12 hours per day for 250 days per year. If the fan is scheduled off during unoccupied hours, annual savings = 100 kW × 12 h/day × 250 days = 300,000 kWh. At $0.12/kWh, savings = $36,000 per year. The reasoning: scheduling saves energy only for hours actually eliminated, and the fan may still need to run for pre-cooling, freeze protection, or code-required ventilation. Verify with trend data before claiming the full savings.

Fault detection and AI

Fault detection and diagnostics (FDD) compares measured behavior with expected models to find simultaneous heating and cooling, stuck dampers, sensor drift, and schedule overrides. AI and machine-learning models can forecast load, optimize setpoints, and detect anomalies, but they depend on clean trend data, metadata, and a stable baseline. Generative AI tools should not directly write control sequences without engineering review. Cybersecurity controls, network segmentation, and role-based access are part of the control system design.

Traps that cost marks

  • Claiming scheduling savings without checking actual occupancy and required off-hours operation.
  • Ignoring overrides and manual bypasses that defeat schedules.
  • Trusting a single sensor without calibration or cross-checking.
  • Assuming AI automatically saves energy without data quality, training, and M&V.
  • Mixing up protocols and physical media.
  • Forgetting cybersecurity and network segmentation in BAS projects.