Prigo X Technologies · working demonstration on public data

MD Guard

One 30-minute average sets a whole month's capacity and network charges. This page takes a month of half-hour interval readings from a mid-size Peninsular Malaysia factory, prices them under the RP4 tariff structure in force since 1 July 2025, and shows what a single avoidable afternoon cost — and how much warning there was.

Billing period · August 2026 Tariff · RP4 (1 Jul 2025 – 31 Dec 2027) Synthetic load profile Rates corroborated — see §7

What one event cost

Billed maximum demand

Set by half-hour readings

Month consumption

Act 861 status

01 — The decision

The tariff choice nobody is selling him

Under RP4 the Time-of-Use demand rate is higher per kilowatt than General — but ToU bills demand only on the weekday 14:00–22:00 window, and discounts off-peak energy. Which way a site lands is a question about its own interval data, not a rule of thumb. On this profile the answer is worth more per year than the peak-shaving below, and it needs no hardware, no site visit and no permission — only a month of readings the client already has.

Same month, same meter, both tariffs

At the current cap. Bars share one axis.

Set by maximum demand Set by consumption

ChargeBasisAmount

Why the two differ

The rule that decides it, and what it is worth here.

02 — The money

What holding a cap is worth

Move the cap and every figure on this page moves with it. The savings below come only from capacity and network charges — no energy saving is claimed, because shifting a load does not reduce the kilowatt-hours it uses.

Saved this month

Half-hours to act in

Worst single event

Annualised

If a month like this recurs. One month is not a year — see §8.

03 — The month

Every half-hour of August, and the one that mattered

1,488 readings. The bill's capacity and network charges are set entirely by the highest point on this line — not by the area under it.

Half-hour average demand — August 2026

Shaded bands are weekends and Merdeka Day. Drag the cap in §2 to move the dashed line.

Half-hour demand Maximum demand Demand cap

04 — The day

Tuesday 18 August, half-hour by half-hour

Demand profile — the day the peak was set

The tinted band is the weekday 14:00–22:00 peak window, which matters for the tariff comparison in §1.

Ordinary half-hour The excursion Peak window

05 — The half-hour

Inside 15:00–15:30, while it was still preventable

Maximum demand is an average, so for the first minutes of a window the outcome is not yet decided. A running average lags and only confirms the peak once it is too late to matter. Projecting where the average will land is what buys the time to act.

None of which is new. Maximum-demand controllers have shed load on a within-window projection for years, and several Malaysian platforms do it in software. This section is here to show the arithmetic is right, not to claim the mechanism — the scarce part is the decision layer above it: which tariff, what cap is worth holding, how much storage that implies, and what to tell a client who asks what last month actually saved them.

Minute-by-minute, 15:00–15:30

Projection holds the current instantaneous load for the balance of the window.

Instantaneous load Projected half-hour average Running average so far

06 — What survives the rate uncertainty

Which of these numbers you can quote today

The medium-voltage energy charges could not be sourced. That matters for some figures on this page and not at all for others, so here is the split rather than a blanket disclaimer.

Quotable now — priced on confirmed rates

These use only the capacity and network charges, which agree across independent sources.

    Wait for the gazetted rates

    These move with the energy charges. The rates are corroborated now, but a real bill is still the only proof.

      07 — The rates

      Every number here is an input, and every input is sourced

      Every rate below now agrees across independent published sources, bar the retail charge, which is RM200 on a bill of hundreds of thousands and moves nothing. Each is still an editable input, carries its confidence, and is checked for plausibility as you type — a decimal point in the wrong place is caught before it reaches a client.

      RP4 tariff schedule

      confirmed agrees across independent published sources · corroborated the same figure across independent published sources · assumed placeholder, not sourced

      Automatic Fuel Adjustment

      Reset monthly, which is the point: it cannot be budgeted a year ahead. Published figure for August, then the outlook.

      08 — Method

      What this is, and what it is not

      What the model does

      • Maximum demand is the highest 30-minute average kW in the month — the same definition the bill uses.
      • General bills that peak wherever it falls. Time-of-Use bills only the highest half-hour inside the weekday 14:00–22:00 window.
      • Capping is modelled as a load shift: billed demand falls, total kilowatt-hours do not. Every ringgit shown is a capacity-and-network delta, and nothing else.
      • The event is isolated by comparing each half-hour with the site's own typical load at that time of day, then taking the contiguous excursion around the peak.
      • The projection holds the current instantaneous load for the rest of the window — the simplest honest forecast, and enough to fire in time.

      What it does not do

      • The load profile is synthetic. It is shaped to be ordinary — two shifts, afternoon thermal load, a half-day Saturday, a public holiday — with one avoidable coincidence. It is not any real site's data.
      • The rates are corroborated from published sources, not read off a bill. The retail charge is still a placeholder. Check them against the gazetted schedule or a real bill before quoting a figure to a client.
      • Annualised figures multiply one month by twelve. A real year has a different peak each month, so treat the annual number as an order of magnitude, not a forecast.
      • No power factor, no KWTBB, no SST, no demand ratchet, no Solar ATAP export credit. Those change the bill and are not modelled.
      • A real deployment reads the meter's own interval data. This is arithmetic on a profile, not a metering integration.