LNG Compressor Starting: A Transformer Study for Voltage Dip and Acceleration

1. Start with acceleration, not installed kVA

A compressor motor can fit the continuous transformer rating and still fail to accelerate on the weakest operating network. Starting draws reactive current, depresses voltage and reduces motor torque. Longer acceleration then increases rotor heating and exposes other loads to a longer dip. Define the operating case before comparing transformer sizes: normal utility, one transformer unavailable, islanded generation, minimum network strength and the process state at start.

ABB's motor-starting application paper explains the current-versus-torque tradeoff of reduced-voltage starting. Its principles are useful, but a large LNG synchronous machine needs its own manufacturer model and excitation sequence. Do not apply an induction-motor curve to every compressor train. ABB starting methods.

2. Request the right model inputs

Obtain source short-circuit strength and X/R for each topology, transformer rated impedance and tolerance, tap position, winding connection and cable impedances. Include pre-start loads, reactive compensation, generator excitation limits and bus sectionalization. A maximum fault-current case usually does not represent the worst starting voltage.

The motor package must provide current and torque versus speed at specified voltages, inertia of the complete train, compressor resisting torque, starting method, allowable thermal starts and restart restrictions. For a VFD, request the input current-versus-time envelope, precharge sequence and harmonic behavior instead of a direct-on-line locked-rotor multiplier. The mechanical team must confirm whether the train starts unloaded and how unloading transitions during acceleration.

3. A hypothetical screening calculation

Assume a balanced direct-on-line induction motor and an initially unloaded bus for this preliminary example. Starting demand at rated voltage is 30 MVA. A 25 MVA transformer has 0.10 pu predominantly reactive impedance; the upstream source contributes 0.05 pu on the same base. Neglect resistance and cable drop. Treat locked-rotor motor impedance as constant for the first estimate. The scalar voltage divider also approximates the locked-rotor motor impedance as inductive; a model retaining resistance requires complex impedances.

The illustrative initial voltage is 84.7% and initial torque about 71.7% of its rated-voltage value. The denominator accounts approximately for current reducing as voltage falls; simply subtracting a fixed current drop would describe a different model. These figures are not acceptable-dip limits, predicted full-start performance or guaranteed OEM values.

If the compressor torque already exceeds this reduced motor torque near a critical speed, making the start slower will not solve the torque deficit. Time-stepping the speed equation with verified curves is necessary. Include voltage recovery, motor heating and synchronous pull-in where applicable.

4. Compare feasible interventions

A tap change raises the starting and running voltage together; check the entire voltage range before using it as a remedy. Parallel transformers can improve stiffness but require compatible ratios, impedances, protection and operating rules. No intervention should be chosen solely from the initial dip estimate.

5. Assess the loads that remain online

Overlay the modeled voltage-versus-time at each important bus with actual device envelopes: contactor retention, drive undervoltage response, instrument supplies and protection. A short deep dip and a shallower long dip may produce different failures. Schneider's ATV320 guidance demonstrates why the installed drive's undervoltage management belongs in this review. Drive response.

Use the refinery ride-through chain for these interfaces. Confirm model connectivity and bases with the single-line and study-model guide.

6. Make the acceptance package reproducible

Specify the selected source cases, initial loading, allowed process starting states, motor and compressor curve revisions, transformer impedance tolerance and the requested results. Results should include bus traces, speed, current, torque margin, acceleration time and thermal duty. Document sensitivities to minimum source strength and temperature-dependent starting restrictions.

At commissioning, compare an approved start with the study using synchronized electrical and mechanical records. Agree deviations and hold points before the test. A successful unloaded start does not validate a different loaded restart. Keep the final operating restrictions with the protection and control documentation.

7. Questions and primary references

Can transformer kVA determine starting capability? It is one input; source impedance and torque-speed balance are also decisive.

Is lower impedance always preferable? It improves stiffness but raises fault current and may alter protection requirements.

Can a VFD eliminate every voltage dip? No. Its supported input envelope and source behavior still need study.

Primary references: ABB motor-starting paper; Schneider undervoltage guidance.