Every decision
your plant makes
should be evidence.

Heat rate deviation: For a 500 MW coal-fired unit on Singareni domestic coal, design heat rate assumptions are built around a specific GCV and combustion profile. When actual coal GCV runs significantly below the design assumption — as it does for plants dependent on a single domestic colliery — specific coal consumption rises well above normative. Every additional 100 kcal/kWh of heat rate costs roughly ₹25–32 crore per year in excess fuel on a 500 MW unit at 70% PLF. The gap between design and cumulative actual heat rate at most ageing Indian thermal stations has been widening for years. What changes the trajectory is continuous root-cause attribution — knowing which of coal quality, combustion, condenser, or auxiliary power is the dominant cause on any given shift.

Auxiliary power consumption: Indian thermal plants typically run APC between 9% and 12% of gross generation. The normative target is closer to 8.5–9.5% depending on plant configuration. Cooling water pumps alone can represent 15–20% of total auxiliary power at older 250 MW stations — making CWP speed optimisation the single highest-return APC intervention available without capital investment. The gap between normative and actual APC, annualised, typically runs to ₹8–12 crore for a 2×250 MW stage.

Demurrage and coal handling: When wagon tippler availability is constrained, the ripple effect is immediate — rakes queue, demurrage accumulates, bunker levels fluctuate, and mill loading becomes reactive rather than planned. The direct demurrage cost is the visible number; the less-visible cost is the combustion quality impact of inconsistent coal feed. Plants with Singareni coal dependency and single-tippler configurations are particularly exposed.

What a well-run thermal plant measures today: DCS historian capturing every process variable across all units. Stack CEMS — SOx, NOx, PM — reported continuously. Coal quality in lab analysis per rake. Mill fineness and combustion performance tracked per shift. Boiler water chemistry within defined limits. All of this is measured. All of it is reported. Most of it arrives in a monthly summary — retrospective, not real-time.

What is missing: A system that connects an incoming coal GCV reading to a combustion setpoint advisory within minutes of that rake arriving. A pre-trip advisory 20–30 minutes before a boiler pressure part approaches its limit — not a post-incident report. A structured shift handover that transfers every open advisory, acknowledged alarm, and pending action digitally — not verbally. A tamper-evident record that answers an audit query or a regulatory inquiry with a database query — not a document assembly exercise. A ranked corrective action list the moment PLF drops below schedule — not a discovery at month-end.

The gap is not sensors. It is not dashboards. Indian thermal plants often have extensive instrumentation. The gap is the governed decision record — connecting the measurement to the advisory, the advisory to the human decision, the human decision to the outcome. Every shift. In real time. With accountability at every step.

Why domestic coal GCV matters so much for Unit 11: Larger supercritical and near-supercritical units are typically designed around a higher GCV assumption — often blended or imported coal. When these units are operated on domestic Singareni coal, the GCV can fall significantly below the design assumption. The consequence is compounding: specific coal consumption rises, mills work harder, combustion quality degrades, and heat rate deteriorates — all at the same time. No single number captures the cumulative efficiency cost better than the gap between design heat rate and actual heat rate.

TwinCore™ recalculates the full thermodynamic impact of each incoming rake's GCV within minutes of the lab report being entered — advising on mill loading adjustment, air-fuel ratio correction, and expected generation shortfall for the shift. Today this chain runs manually, monthly, and retrospectively. PowerGuardian™ makes it continuous and real-time.

On mill liner maintenance: Mill liner replacement schedules are typically based on running hours from the last replacement. When replacements are deferred — which is common during high-demand periods — mills run with degraded grinding surfaces, producing coarser pulverised coal, higher unburned carbon in ash, and worsening combustion efficiency. TwinCore™ models current mill performance against expected performance at known running hours, flagging the maintenance window before the efficiency loss becomes significant.

Every AI system generates recommendations. Very few record what recommendation was made, who received it, what the operator decided, what the outcome was — in a form that is tamper-evident and verifiable by a third party.

GuardianLedger™ does this for every advisory across every system the OAL monitors. The record is append-only and cryptographically chained — meaning it cannot be altered retroactively. When a PCB inspector asks what the plant did in response to a specific GCV deviation or CEMS exceedance, the answer is a query result — not a memory search or a document reconstruction.

At KTPS specifically, this matters for three reasons: First, pending audit paras require evidence that specific decisions were made correctly at specific times — GuardianLedger™ produces that evidence as a by-product of normal operations, not as a separate compliance exercise. Second, SOx compliance on a large unit without FGD requires a defensible record that the plant was monitoring, advising, and responding — not just measuring. Third, boiler licence renewals require an operational compliance record for the preceding period — GuardianLedger™ Commission means that record accumulates from Day 1.

SkyEdgeAI was founded in March 2025 in Bengaluru — which means the platform was built entirely with current AI capability, current governance frameworks, and the specific design principle of operational admissibility from its first line of code. There is no legacy architecture to work around.

The founding team brings specific domain depth: Devi Prasad Vuriti's 20+ years in power and industrial operations provides the OT integration credibility that IT-only vendors lack. Kameswara Rao Tangudu's integration architecture experience underpins SkyConnect™. The platform was designed for environments exactly like KTPS — legacy OT, high-stakes operations, state-sector governance requirements — not adapted from a smart city or fintech template.

DPIIT recognition and GeM registration mean the commercial pathway is clear for Telangana Power Genco without the procurement complications that international vendors often create.

To initiate the 90-day baseline engagement, SkyEdgeAI needs four things:

1. A letter of intent to engage — not a commercial contract. A statement that KTPS intends to participate in the 90-day baseline. This allows us to resource the KTPS-specific configuration of TwinCore™ and SkyConnect™ before Day 1.

2. Read-only access credentials to the DCS historian and SCADA system for Units 9, 10, and 11. This is the same class of access your existing monitoring and reporting systems use. No control write access is requested or accepted at any stage.

3. Key design parameters for each unit — design heat rate, design GCV assumption, APC normative targets, and equipment commissioning dates. This allows TwinCore™ to be pre-seeded with KTPS-specific baselines before Day 1.

4. A nominated KTPS technical contact — one person who can confirm advisory outputs during the first 30 days and provide the Day 30 mill liner validation feedback.