A data-center connection invoice needs an obligation, not just a project name
Link each connection payment to the contract and facility that establish its responsibility.
Responsibility precedes billing
Large-load growth creates an operational question before it becomes an electricity bill: who pays for which connection asset? A project name is not enough to settle that responsibility. Truthound can check a prepared obligation ledger against approved contract references. It cannot determine a regulated tariff, interpret an agreement or establish that a payment has actually been made.
The settlement's stated principle
Duke Energy announced a North Carolina settlement on October 7, 2026 involving its utilities, Public Staff and other parties. The release describes protections including upfront payments for customer-dedicated connection facilities. These are the utility's statements about the agreement. A billing team still needs the applicable contract and facility scope before turning that description into an individual charge.
Separate assets and obligations
Separate a dedicated facility from shared infrastructure, and a payment obligation from its settlement. An illustrative ledger can record contract, facility, cost class, payer, obligation identity and payment reference. Assign an owner to resolve ambiguous classifications. Copying one payer across a group of similarly named projects can hide the very cost allocation that the agreement is intended to clarify.
Check the parent obligation
For the billing and obligation reconciliation table prepared as a Polars DataFrame, Truthound Core Python SDK 3.3.4's ForeignKeyValidator can inspect supplied child and parent tables using the defined keys. A billing row whose obligation does not exist in the approved register can therefore be flagged. A valid parent link still needs the correct facility and contract revision. Reference existence is not evidence that the amount, payment status or legal allocation is correct.
A controlled pandas reconciliation
pandas 3.1.1 is a useful custom alternative for a small billing extract. A merge with an explicit cardinality policy and indicator can expose unresolved links, provided the team adds a reject condition for unmatched rows. many_to_many does not check cardinality, and null-key matches need care. Keep a tested reconciliation when it expresses the actual billing policy; a library change is not a substitute for contract review. pandas matches null keys to each other, unlike ordinary SQL. A billing row with an empty obligation key can otherwise appear to have found a real obligation. validate="many_to_many" performs no cardinality check; use the intended 1:1 or m:1 policy and a separate unmatched-key assertion.
Resolve exceptions before the batch
Before a recurring batch, have the responsible owner inspect unresolved obligations and duplicated payment references. A correction should preserve the original row and explain which agreement or asset mapping changed. Do not silently infer a missing payer from the nearest complete project. The exception needs a decision, not simply a value that makes a constraint pass.
Keep legal and financial evidence distinct
Release only the billing scope supported by the approved references and financial evidence. Reconcile actual payments through the authorized finance process, separately from the structural check. When the contract or asset ownership changes, revisit the mapping before the next batch. This turns a public cost-protection principle into a traceable operating responsibility without claiming to calculate the tariff.
Evidence
- Event checked
- Subject checked
- Modified
