Energy Consumption Analysis: Reading Your Load Profile to Cut Waste

Most operators can quote their monthly kWh and the figure at the bottom of the TNB bill, and almost nothing about the shape of that consumption. The shape is where the money is. Energy consumption analysis reads the load profile hour by hour, day by day, and asset by asset, and it is the first move in any serious efficiency or cost-reduction effort. A single monthly total hides a base load that never drops at 3am, a chiller that ramps before anyone is in the building, and a demand peak that quietly resets your capacity and network charges for the month.

This guide walks through what a load profile is, the waste patterns it exposes, the Malaysian cost structure that makes the shape matter, and how to turn the analysis into verified savings.

What is energy consumption analysis?

Energy consumption analysis is the practice of breaking down how, when, and where electricity (and water, gas, or steam) is used, instead of treating the bill as one lump number. At its core sits the load profile: a plot of power demand (kW) sampled at short intervals, typically every 15 or 30 minutes, across the day and week. From that single curve you can read base load, peak demand, load factor, and the timing of every major draw.

Done properly, the analysis moves down the hierarchy. A site-level profile tells you the building's rhythm. An equipment-level profile, built from sub-meters or current sensors on chillers, compressors, and pumps, tells you which asset is responsible for each feature in the curve. Without that second layer you can see that something is wrong; with it you can name the offender.

How to read a load profile

Start with four numbers any operator can pull from interval data:

- Base load — the floor the curve never drops below, usually overnight or on weekends. For a five-day operation, anything above roughly 15 to 25 percent of peak running at 3am on a Sunday deserves an explanation.
- Peak demand (maximum demand) — the highest 30-minute average in the billing period. This is the number TNB bills you on, not your average.
- Load factor — average demand divided by peak demand. A low load factor (well under 0.5) means you are paying for capacity you use only in short bursts.
- Shape and timing — when the curve rises and falls relative to occupancy, production schedule, and the tariff peak window.

A flat overnight line on a daytime operation is the clearest signal of waste: equipment left running because no one scheduled it off. A base load that creeps upward month over month points to degrading assets, a fouling chiller, a failing trap, a compressor with a worsening leak. Spikes that do not line up with production usually trace back to a control fault or a startup sequence stacking loads on top of each other.

Why the shape matters under TNB's RP4 tariff

Under the RP4 tariff structure effective 1 July 2025, the shape of your load profile is no longer a back-office curiosity. The old single Maximum Demand charge has been replaced, and the bill now has five components: Energy (sen/kWh), Capacity charge (RM/kW), Network charge (RM/kW), a fixed Retail charge (RM/month), and the Automatic Fuel Adjustment (AFA), which replaced ICPT. (TNB tariff schedule)

The two per-kW charges are what your peak demand drives. For a general (non-ToU) commercial or industrial account on the C1/E1 tariff, the Capacity charge is RM29.43/kW and the Network charge is RM59.84/kW, totalling RM89.27 per kW of monthly peak demand. On a Time-of-Use account (C2/E2), the rates are RM30.19/kW capacity plus RM66.87/kW network, totalling RM97.06 per kW. (TNB)

Work the example. Suppose your profile shows a 700 kW base running comfortably, but a poorly sequenced morning startup pushes a single 30-minute interval to 900 kW. That 200 kW of avoidable peak costs, on a general tariff, about 200 x RM89.27 = RM17,854 every month, roughly RM214,000 a year, for demand you only touched for half an hour. Energy analysis is what turns that invisible spike into a line item you can attack. The fix is often pure scheduling: stagger the startup so the peak never forms. See our full breakdown in how to cut TNB maximum demand charges and the method in how to calculate maximum demand.

Time-of-Use accounts add a second dimension to the analysis. The ToU peak window is 2:00pm to 10:00pm on weekdays; off-peak covers 10pm to 2pm on weekdays plus all weekends and public holidays. Profiling lets you see how much energy you are pulling inside that expensive window and whether any of it can be shifted, pre-cooled, batch-scheduled, or served from storage. For battery-backed shifting, see BESS for peak shaving.

Waste patterns worth chasing first

The patterns repeat across Malaysian buildings and plants, and they rank roughly by payback:

- Phantom base load — HVAC, lighting, or compressed air running outside occupied hours. Usually a scheduling or control fix, near-zero cost.
- Weekend and holiday consumption on a weekday operation — the curve should collapse but doesn't.
- Demand peaks from stacked startups — chillers, AHUs, and pumps all spinning up together in one interval.
- Creeping base load — month-over-month drift that flags equipment degradation before it becomes a breakdown.
- Power factor penalties — a profile with reactive power draw eroding into your charges, correctable with capacitor banks.
- Peak-window energy on ToU — production or cooling load sitting inside the 2pm to 10pm window that could move.

Each pattern is a lead to chase, and each becomes far easier to find when you can correlate the curve against occupancy, weather, and the production schedule rather than eyeballing a chart.

Turning analysis into verified savings with CobiNeural

CobiNeural builds these profiles automatically from live, equipment-level data so your team works from evidence instead of intuition. The Insights to Energy module tracks consumption, demand, the Max Demand KPI, power factor, and EUI at both location and equipment level, which is where the base load and the startup spikes become visible by asset. Insights to Equipment ties anomalies in the curve to specific motors and chillers via condition and sub-metering data, and weather and occupancy correlation explains the rises that are legitimate.

When the analysis surfaces an avoidable peak, Alerts can notify the team by WhatsApp or email before the interval closes, and Actions together with automation can stagger startups or trim demand directly on the BMS, PLC, or SCADA layer. Because CobiNeural deploys standalone or as an overlay on existing controls, you do not rip out what you have.

The last step is proof. Use Plan & Verify to run the fix as a measurement and verification project, so the before-and-after is documented to standard rather than asserted. That same data feeds Reporting for EECA and ISO 50001 evidence, which means the profiling work that cut your bill also satisfies the regulator. Sub-metering is what makes all of this granular enough to act on; see sub-metering and the hidden problems it finds.

Energy consumption analysis is not a one-time audit. It is a continuous read on the shape of your operation, and the operators who treat it that way are the ones who keep their peaks flat and their base load honest.

If you want to see your own load profile broken down by asset, request a demo and we'll walk through where your energy actually goes.