How to Prove Energy Savings: A Simple M&V Plan | Cobler
Blog
How to Prove Energy Savings: A Simple M&V Plan
A lower bill does not prove a saving, and a higher one does not disprove it. Prove energy savings with a measurement and verification (M&V) plan: fixed periods, a measurement boundary, routine and non-routine adjustments, and a named evidence owner, worked through for an example office tower.
Tan Kok XinEnergy Management: The Economics of Saving Energy
Part 4 of 10 in Cobler's Energy Management: The Economics of Saving Energy course. New here?See the course page.
Part 3 built a baseline for our example office tower and found what pushes its consumption up and down; this part puts that baseline to work. A year after a savings project, someone from finance will look at the electricity bill and ask whether the project worked. The bill alone cannot answer, because weather, working days, tenants and the monthly fuel adjustment all change it too. You prove energy savings by comparing the energy metered after the project with the energy the building would have used without it, worked out from the baseline and adjusted for those changes. The rules for doing this are written down before the project starts, in a measurement and verification (M&V) plan.
The tower is a fictional building on the MV (medium voltage) non-domestic General tariff, and Part 2 valued its three measures at about RM158,065 and 112.5 tonnes of CO₂e a year. This part writes the M&V plan that will show whether those savings really happen.
Why a lower bill, or a higher one, proves nothing on its own
Here is what the tower's bills might show after the project. In April 2026, before the measures, the tower used 530,000 kWh. In April 2027, after them, it used 532,000 kWh. Read on its own, the bill says the project failed: consumption went up.
But April 2027 was hotter. Part 3 measured how hot a month is in cooling degree days (CDD): for each day, the degrees by which the average outdoor temperature was above 24 °C, added up over the month. April 2026 had 135 CDD; April 2027 had 160. Part 3 also found that each extra CDD adds roughly 600 kWh to the tower's month. Both Aprils had 22 working days, so only the weather needs an adjustment:
FAQ
Frequently asked questions
Ask us about this topic
Something in this article you want to dig into — or a situation in your own building it doesn't quite cover? Send us your question. We don't run public comments; the team replies to you directly by email.
Extra cooling expected from the hotter weather: 25 CDD × 600 kWh = 15,000 kWh.
What the tower would have used in April 2027 without the measures: 530,000 + 15,000 = 545,000 kWh. This figure is called the adjusted baseline.
What it actually used: 532,000 kWh.
Avoided energy: 545,000 − 532,000 = 13,000 kWh.
That is close to the 12,700 kWh a month that Part 3 expected. The project worked, even though the bill went up.
{{media:523}}
The saving is the gap between what the building would have used and what it did use, not the difference between two bills.
The same problem works in reverse. A cooler month can make a failed project look successful. The ringgit on the bill also move for reasons unrelated to the project, because the AFA (Automatic Fuel Adjustment) rate changes every month. So M&V compares kWh and kW first, and only then turns the verified savings into ringgit.
Fix three periods before the project starts
An M&V plan names three periods:
Baseline period: the "before". For the tower, October 2025 to September 2026, the twelve months of bills and meter data from Part 3.
Installation period: while the measures are fitted and tested, October to December 2026. Savings in this period are not counted, because the building is neither "before" nor "after".
Reporting period: the "after", January to December 2027. The savings are measured here.
{{media:524}}
Both measured periods are a full year, so each includes every season and holiday once.
Both measured periods cover a full year for the reason Part 3 gave: consumption changes with the seasons and the holidays. A baseline chosen from a quiet quarter would make every later month look like a saving.
Choose what the meter should see
The measurement boundary is the part of the building whose energy you measure. There are two broad choices:
The whole building, using the main meter or the TNB (Tenaga Nasional Berhad) bill. This captures every effect of the project, including side effects, but also every unrelated change.
One system, using a sub-meter on the equipment that the measure changes, such as the chiller plant. This isolates the measure from the rest of the building.
The international reference for M&V is the International Performance Measurement and Verification Protocol (IPMVP), maintained by the Efficiency Valuation Organization. It describes four options:
Option A: one system, with the key parameter measured (for example, the running hours of new lights) and the rest estimated.
Option B: one system, with all its energy measured by a sub-meter.
Option C: the whole building, using the main meter and a model of the drivers. It works best when the savings are large compared with the building's normal swing, roughly more than 10% of the whole-building bill.
Option D: a calibrated computer simulation, used when there is no usable baseline, such as a new building.
The tower's measures save about 2.5% of its electricity, much smaller than the 14% difference between its May and December bills. A whole-building comparison could not separate a saving that small from the weather. So the plan measures each measure where it happens:
M1 (plant stops at 6pm): the chiller plant sub-meter, looking at the hours from 6pm to 8pm on working days.
M2 (VSDs on the condenser pumps): new meters on the two pump feeders, fitted with the variable speed drives (VSDs).
M3 (demand limiting): the maximum demand (MD) on the TNB bill, together with the BMS (building management system) log of the loads it held off.
The whole-building April comparison from the start of this part stays in the plan as a cross-check, not as the proof.
Adjust the baseline for the things that change every month
Some drivers change every period in a way everyone expects: the weather, the number of working days, the running hours. Allowing for them is called a routine adjustment. The rule for each adjustment is written into the plan before the reporting period, so nobody can choose a convenient method afterwards.
For M2, the routine adjustment is simple, because the old pumps ran at one fixed speed. Before the VSDs, the two pumps together drew 54.0 kW whenever they ran. So the adjusted baseline for any month is 54.0 kW × that month's actual running hours. Take a reporting month with 22 working days and the usual 11 hours a day:
Running hours: 22 × 11 = 242 hours.
Adjusted baseline: 54.0 kW × 242 h = 13,068 kWh.
Metered on the two pump feeders: 9,300 kWh.
Verified saving: 13,068 − 9,300 = 3,768 kWh.
Part 2 expected 16 kW × 242 h = 3,872 kWh for such a month. The verified figure is about 3% lower, which is well within the range Part 3 gave. The monthly report records both numbers.
For M1, the adjustment works the same way. Before the change, the plant drew about 220 kW from 6pm to 8pm. After it, the plant sub-meter should show almost nothing in those hours, apart from the evenings a tenant asked for cooling. The BMS logs each request, so the report can show the hours saved and the hours used on request.
Agree in advance how to handle one-off changes
Some changes are not part of the normal pattern: a tenant adds a server room, a floor stands empty for six months, or office hours change. Allowing for these is called a non-routine adjustment. The plan cannot predict them, but it can say how they will be handled.
For the tower, the plan says: when a load is added or removed for reasons unrelated to the project, the facility manager records the date and meters the load, and the baseline is adjusted by that metered energy. Suppose a tenant installs a server room that its own meter shows using 21,600 kWh in a 30-day month (a steady 30 kW). That month's adjusted baseline for the whole-building check rises by 21,600 kWh, so the new load is not mistaken for a failed project.
Deciding the rule before the reporting period is what makes the result credible. If the rules are chosen after the numbers are known, every adjustment looks like moving the goalposts.
Verify a demand saving month by month
M3 saves no kWh, so kWh comparisons cannot show it. What it saves is kW off the month's highest half hour, and Part 2 counted that only if the peak really falls. The plan therefore checks it every month with two records:
the MD on the TNB bill, which is the highest half hour of the month; and
the BMS log of how many kW the demand limiting was holding off during that half hour.
The demand saving for the month is the kW held off in the highest half hour. For example, suppose the July 2027 bill shows an MD of 1,510 kW, and the BMS log shows 95 kW held off during that half hour. Without the limiting, the MD would have been about 1,605 kW, which is in line with July 2026's 1,610 kW. The verified demand saving is 95 kW, worth 95 × RM89.27 = RM8,480.65 that month.
If a month is cool and the demand limiting never needs to act, the demand saving for that month is zero, and the report says so. This is why Part 2 pointed out that M3 carries most of the value and most of the uncertainty.
Name who owns the evidence
A saving that nobody can document is a claim, not a result. The plan names who does what:
The evidence owner is the tower's facility manager. They collect the meter data, the BMS logs and the bills each month, and they record any non-routine changes as they happen.
The monthly report is due by the 10th of the following month. It shows, for each measure, the adjusted baseline, the metered value, the verified saving and any adjustments made.
Finance signs off the annual savings figure at the end of the reporting period.
Meters and data are checked too. The new pump meters are commissioned before the reporting period starts, and any data gap is recorded and filled by the method written in the plan, just as Part 3 filled the missing March days.
M&V also has a cost: meters, staff time and reports. For the tower, the plan budgets RM8,000 a year, which the cash-flow model in Part 8 includes.
The tower's one-page M&V plan
Everything above fits on one page. This is the plan for the running case.
Item
The tower's plan
Measures
M1 chiller plant stops at 6pm; M2 VSDs on the two condenser water pumps; M3 BMS demand limiting
Expected savings
105,600 kWh (M1) + 46,464 kWh (M2) a year; 100 kW off each month's MD (M3); about RM158,065 and 112.5 t CO₂e a year at September 2026 rates
Periods
Baseline Oct 2025 to Sep 2026; installation Oct to Dec 2026; reporting Jan to Dec 2027
Boundaries and meters
M1: chiller plant sub-meter, 6pm to 8pm on working days. M2: new meters on both pump feeders. M3: TNB bill MD plus BMS log. Whole-building check: TNB bills
Routine adjustments
M1 and M2: actual running hours. Whole-building check: working days and cooling degree days (about 600 kWh per CDD)
Non-routine adjustments
Added or removed loads: date recorded, load metered, baseline adjusted by the metered energy
Demand rule
Saving each month = kW the BMS held off in the month's highest half hour; zero if it did not act
Valuing the savings
Verified kWh at the energy rate plus that month's AFA; verified kW at RM89.27 (MV General); also shown at September 2026 rates to compare with the business case
Evidence owner and reports
Facility manager; monthly report by the 10th; annual sign-off by finance; budget RM8,000 a year
We now have a value for each measure, a baseline and a plan to prove the savings. The next question is whether the savings justify the cost. Simple Payback and ROI: The Quick Screening Tools (Part 5) starts the investment tools, beginning with the two quickest checks.
Check your understanding
In May 2027 the tower used 540,000 kWh, against 535,000 kWh in May 2026. Both months had 22 working days, but May 2027 had 10 more cooling degree days. Using about 600 kWh per CDD, how much energy did the measures avoid, and what should the facility manager do? Adjusted baseline: 535,000 + 10 × 600 = 541,000 kWh. Avoided: 541,000 − 540,000 = 1,000 kWh, far below the roughly 12,700 kWh a month expected. The whole-building check is only a cross-check, so the manager should look at the sub-meter results for M1 and M2 first, for example whether the plant's 6pm stop was overridden, and record what they find.
In a cool month the BMS demand limiting never had to act. Why does the M&V report count no demand saving that month? The saving is the kW the measure removed from the month's highest half hour. If the measure did nothing during that half hour, the MD would have been the same without it, so there is no demand saving to count.
Recap: A bill that goes down, or up, does not prove a saving, because weather, working days, tenants and AFA change it too. An M&V plan, written before the project, fixes a baseline period, an installation period and a reporting period; chooses a measurement boundary for each measure; and sets rules for routine adjustments (such as weather and running hours) and non-routine ones (such as a new server room). The saving is the adjusted baseline minus the metered energy. Demand savings are checked month by month against the highest half hour. A named evidence owner collects the records, and finance signs off the result.
Cobler builds CobiNeural, a platform that shows a facility team its building's energy, water and indoor air data as live numbers across the whole site. To see how your building performs, talk to us.
Tan Kok XinEnergy Management: The Economics of Saving Energy
