Self-Generation vs Utility Power – Which Costs Less for Large Users?
Every large power user eventually asks the same thing. The utility bill is one of the biggest operating costs on the books, natural gas is sitting in a pipeline nearby, and engine vendors are happy to quote a plant. Would it be cheaper to make your own electricity? Sometimes the answer is a clear yes, sometimes a clear no, and often it depends on a handful of numbers you can work out on one page. In the matchup of self-generation vs utility power, which one actually costs less for a large user?
For most large users, utility power costs less on a simple per-MWh basis unless natural gas is cheap, the load runs nearly around the clock, and the waste heat or reliability benefits have real value. The comparison of self-generation vs utility power comes down to three cost blocks for the on-site plant (fuel, operations and maintenance, and capital recovery) against the utility's all-in rate, including demand charges.
Self-generation also brings costs that are easy to miss, such as redundant units, standby charges from the utility and staff. It wins most often where gas is cheap, grid power is pricey or slow, or the plant can sell its heat.
What Does Self-Generated Power Cost per MWh?
Self-generated power costs the sum of fuel, operations and maintenance (O&M) and capital recovery, divided by the MWh produced. Fuel is usually the largest piece, and it depends on two numbers: the gas price and the plant's heat rate.
Heat rate is the fuel needed to make one kWh, in British thermal units (Btu). The US Energy Information Administration explains that one kWh contains 3,412 Btu, so a plant with a heat rate of 7,500 Btu/kWh is about 45% efficient. The heat rate article goes further into the details. For on-site gas engines, a heat rate around 8,500 Btu/kWh is a reasonable planning number.
The fuel cost formula is simple: heat rate in MMBtu per MWh times the gas price in dollars per million Btu (MMBtu). At 8,500 Btu/kWh, the plant burns 8.5 MMBtu per MWh.
A Worked Example: Gas Engines vs the Utility
Picture a facility with a steady 20 MW load. Every input below is an assumption you'd replace with real quotes.
Fuel
- Heat rate: 8,500 Btu/kWh, or 8.5 MMBtu/MWh
- Delivered gas price: $3.00/MMBtu
- Fuel cost: 8.5 × $3.00 = $25.50/MWh
Operations and Maintenance
The EPA's catalog of reciprocating engine technologies lists O&M costs of roughly $0.0085 to $0.025 per kWh, depending on engine size. For a multi-megawatt plant with staff and overhauls, use $15/MWh.
Capital Recovery
- Installed cost: $1,500 per kW, in line with the EPA catalog's figure of about $1,433/kW for a 9.3 MW engine plant (2013 dollars)
- Redundancy: installed capacity 1.3 times the load, so 26 MW for a 20 MW load
- Capital per kW of load: $1,500 × 1.3 = $1,950
- Capital recovery factor at 8% over 20 years: about 0.102
- Yearly capital cost per kW of load: $1,950 × 0.102 = about $199
- Energy per kW of load at a 90% load factor: 8,760 × 0.9 = 7,884 kWh
- Capital cost per MWh: $199 ÷ 7.884 MWh = about $25.20/MWh
Total
$25.50 fuel + $15.00 O&M + $25.20 capital = about $65.70/MWh. If the utility's all-in industrial rate, including demand charges, is $70/MWh, self-generation saves a little. If it's $55/MWh, the utility wins easily.
How Much Does the Gas Price Change the Answer?
The gas price changes the answer more than any other input. Each $1/MMBtu adds $8.50/MWh at this heat rate. The table below holds O&M and capital constant and varies only gas.
| Delivered gas price | Fuel ($/MWh) | O&M + capital ($/MWh) | Total ($/MWh) |
|---|---|---|---|
| $2.00/MMBtu | $17.00 | $40.20 | $57.20 |
| $3.00/MMBtu | $25.50 | $40.20 | $65.70 |
| $4.00/MMBtu | $34.00 | $40.20 | $74.20 |
| $5.00/MMBtu | $42.50 | $40.20 | $82.70 |
That spread is why gas price risk deserves as much attention as the equipment quote. A plant that pencils at $3 gas can lose money at $5, and a long-term fixed-price gas contract can be worth paying a premium for.
What Is a Spark Spread?
A spark spread is the difference between the price of electricity and the cost of the gas needed to make it. Power traders use it to judge whether a gas plant makes money, and it works just as well as a quick screen for self-generation.
The formula is: electricity price minus (heat rate × gas price). Using the example above, a $70/MWh utility rate minus $25.50/MWh of fuel leaves a spark spread of $44.50/MWh. That margin has to cover O&M and capital, which together came to $40.20/MWh. The result is a thin $4.30/MWh cushion.
A thin spread is a warning sign. It means a modest rise in gas prices, an extra overhaul or a few points of lower load factor could erase the savings. A wide spread, say $20/MWh or more after all costs, gives a project room to absorb surprises.
Does Combined Heat and Power Change the Math?
Combined heat and power (CHP) can change the math a lot, but only if the heat has a real use. Engines turn roughly 35% to 40% of fuel energy into electricity and reject most of the rest as heat in the exhaust and jacket water.
Try rough numbers. Say each MWh of electricity comes with about 3 MMBtu of recoverable heat, and that heat replaces boiler fuel that would have cost $3.00/MMBtu at 80% boiler efficiency. The avoided fuel is 3 ÷ 0.8 = 3.75 MMBtu, worth about $11.25 per MWh generated. Subtract that credit from the $65.70 total and the effective cost drops to about $54.45/MWh.
The catch is that the heat has to be used every hour the plant runs. A data center can use some of it in absorption chillers, but low-grade heat with no buyer is worth nothing.
What Costs Do People Forget About Self-Generation?
The biggest forgotten costs are redundancy, standby charges, staffing and permitting. Each can shift the answer.
- Redundancy: you can't run one engine and hope. Spare units sit idle much of the time but still cost capital.
- Standby or backup service: if you keep a grid connection for backup, the utility usually charges a monthly fee for the capacity it holds ready for you, even if you rarely use it.
- Staff: a plant needs operators, technicians and on-call coverage.
- Permits and emission controls: air permits and catalysts add cost and time.
- Fuel delivery: a pipeline lateral, meter station and gas conditioning can add millions up front.
- Cooling water: some plant designs, especially combined cycle with wet cooling towers, need water, which raises questions about data center water rights and supply.
On the utility side, watch for demand charges, ratchet clauses that bill on your highest peak for months, and riders added on top of the base rate. The posted energy rate rarely tells the whole story.
When Does Self-Generation Make Sense?
Self-generation makes sense when several conditions line up at once. Any one of them alone usually isn't enough.
- High load factor: the facility runs near full load around the clock, so capital is spread across many MWh.
- Cheap, reliable gas: a pipeline or field gas at a low, stable delivered price.
- Expensive or slow grid power: high industrial rates, big demand charges or a long wait for a new connection.
- Useful heat: combined heat and power can use engine heat for process steam or absorption chilling, raising total efficiency.
- Reliability value: the plant replaces backup generators the site would have bought anyway.
Data centers fit some of these well, especially the high load factor. Waste heat is harder, because the heat is low grade and few neighbors want it.
Is Self-Generation More Reliable Than the Grid?
Self-generation is not automatically more reliable than the grid. A well-run transmission-fed site sees very few outages a year, and the grid acts like one giant pool of backup capacity. An on-site plant has to build its own backup, usually with spare engines and storage.
The best reliability often comes from having both: a grid connection plus on-site generation that can island. That's also the most expensive setup, because you pay for two systems. My take: if reliability is the main reason for self-generation, price the hybrid honestly, including standby charges, before you call it a savings play.
How Should You Compare the Two Fairly?
Compare the two on the same basis: delivered, all-in cost per MWh at your real load profile. Use these steps:
- Pull 12 months of utility bills and calculate the true average cost per MWh, including demand charges and riders.
- Get firm quotes for equipment, installation, gas supply and O&M.
- Model fuel at low, base and high gas prices.
- Include redundancy, standby charges, staffing and emission controls.
- Run the numbers over 15 to 20 years with your real cost of capital.
Self-Generation vs Utility Power: Let the All-In Number Decide
In the self-generation vs utility power debate, neither side wins by default. On-site gas generation can beat utility rates when gas is cheap, the load is steady and grid power is expensive or late. In many places, though, the utility's scale and shared backup capacity make it the cheaper choice.
The fair way to decide is to calculate the all-in cost per MWh for both options, stress-test the gas price and count every hidden cost. Start with the cheapest step: pull last year's utility bills and work out what you really paid per MWh, demand charges included. If that number is well under the on-site estimate at $4 gas, you have your answer. If it's close, bring in an engineer and a financial advisor to firm up the quotes, since this article isn't investment advice.