← Back to Enerplain

ENERPLAIN / FIELD NOTES

Before you accept a new connection

A compact review checklist for one source-to-application energy-data connection, with explicit expectations, reproducible differences and visible unknowns.

The question to answer: does this specific connection preserve the meaning agreed with the source and the receiving application? A successful request or a populated dashboard alone does not answer it.

This is a proposed review method. It is not certification, an industry standard or proof that a live integration is correct.

1. Bound the comparison

2. Write the expected meaning

3. Preserve a reproducible discrepancy

For each finding, keep the approved source example, the corresponding application result, the expected transformation rule and a minimal reproduction. Separate an explained difference from an unresolved one. A report should make it possible for another engineer to repeat the comparison.

For example, the missing-zero fixture has eight expected interval starts, seven source observations and eight downstream values. The finding is the silent zero fill at 00:30 UTC; the equipment condition remains unknown.

4. Agree how findings are handled

5. Handing a PV portfolio over: what to ask for

The same questions apply when a portfolio changes hands and the records arrive as exports. Write down the expected meaning before counting missing records.

Fill in one qualification sheet

Use a private copy of this sheet. For each check, record whether it is confirmed or unknown, the source reference and who can answer the open question. This page does not collect your answers.

On small screens, scroll the sheet sideways. You can also focus it and use the arrow keys.

Six checks before comparing a PV register with an export
CheckWhat to confirmConfirmed or unknownEvidence referenceQuestion owner
ScopeSites and devices, review window, source owner, agreed cutoff and allowed sharing method.To fill inTo fill inTo fill in
IdentityStable register IDs, channel-to-device mappings and how duplicate records are identified.To fill inTo fill inTo fill in
ExportSource or portal, export time, date range, filters and any excluded devices or fields.To fill inTo fill inTo fill in
TimeTime zone and UTC offset, interval start or end, expected cadence and window boundaries.To fill inTo fill inTo fill in
ValuesMetric, unit, scale, sign and precision. Distinguish power, interval energy and cumulative energy.To fill inTo fill inTo fill in
MissingnessMeaning of absent rows, null or unavailable values, measured zero and unresolved identifiers. Keep unresolved rows visible in a separate list; do not drop them.To fill inTo fill inTo fill in

Pause an affected conclusion if its meaning is unknown. Ask the source owner one specific question. Do not fill an absent record with zero or treat an unresolved ID as a missing device.

A small fictional example

Suppose an agreed register lists I-A, I-B and I-C for one interval. The supplied export contains a mapped I-A reading of 0 kW, no row mapped to I-B, and a row labelled device-17 with no agreed mapping.

One useful question is: Which register device does device-17 refer to, and where is that mapping recorded? This is a reading example, separate from the fictional portfolio below.

See also missing records and measured zero and the hour that happens twice.

Two answers are worth writing down: a site that was never mapped into a portal does not appear in that portal's overview at all, and a site that still reports can be missing one of its expected devices. Comparing the overview with the agreed register helps check which expected sites and devices are represented. Assessing record completeness also requires the agreed mappings, review window and reporting expectations. Whether a particular portal raises an alert when a mapped device stops arriving is a property of that portal — check it rather than assume it.

What if an expected site is absent from a portal export?

Check its stable site and device IDs against the agreed register, the export's filters and date window, and a permitted portal inventory if one is available. Distinguish a site missing from the observed inventory from one present there without records in the reviewed export. If no inventory is available, absence from the export alone does not establish absence from the portal. Ask the source owner to clarify an unknown mapping or filter. None of these observations alone proves equipment failure or lost generation.

In our fictional example, S-05 has no portal assignment, while S-07 still reports but has one expected inverter marked absent. These are declared example conditions, not findings about a customer.

What Enerplain currently sells

This checklist is a technical reference, not a product. The offer Enerplain currently sells is the Portfolio Blind-Spot Snapshot: a 30-day, read-only reconciliation of a site list against supplied records from up to two monitoring portals, delivered as a written exception list. It uses the same discipline as the checklist above, applied to a portfolio rather than to one connection.

There is no live connector or upload service on this website. Describe a difficult check without credentials, customer identifiers, addresses or raw telemetry. A short, non-sensitive description is enough.

Read the checklist in Markdown.

Example: when a missing reading becomes a zero.