Engine Performance Monitoring: PMI Systems

PMI systems measure in-cylinder gas pressure against crank angle on every firing cycle to produce the indicator diagram, from which engineers calculate indicated mean effective pressure (IMEP), peak firing pressure Pmax, and compression pressure Pcomp for each cylinder. These three quantities are the primary tools for balancing cylinder loads, detecting mechanical faults, and tracking engine performance against the fuel-consumption baseline that feeds SFOC and CII calculations. Modern slow-speed two-stroke engines carry one piezoelectric transducer per cylinder, streaming data continuously; earlier practice, still used for periodic checks and comparison, employs a portable pressure analyser connected to the indicator cock one cylinder at a time.

The indicator diagram is not a modern invention. James Watt patented a steam-engine indicator in 1796 to measure the work done by the working fluid; Richard Trevithick adapted the concept for steam locomotives in the early 1800s; by the 1880s mechanical indicators based on a spring-loaded piston and a recording stylus were standard diagnostic tools on marine reciprocating steam engines. The shift to diesel propulsion after the First World War carried the same measurement logic into a new combustion environment, and mechanical indicators remained the norm on marine diesel engines well into the 1970s. Electronic PMI, with solid-state piezoelectric transducers and digital signal processing, entered service on large slow-speed two-stroke engines in the late 1980s and became routine equipment on new newbuildings by the mid-2000s. The term “PMI” itself, standing for Performance Measurement Instrument (or, in some MAN Energy Solutions documents, Performance Monitoring Instrument), entered the engine-room vocabulary with the first MAN B&W electronic indicator installations.

This article covers the full PMI measurement chain: the pressure transducer and crank-angle encoder, the indicator diagram and what each phase of the trace reveals, the derived quantities (IMEP, Pmax, Pcomp, pressure rise rate) with their defining equations, the diagnostic interpretation of trace anomalies, the offline-versus-online system comparison, the crank-angle reference and accuracy constraints, and the connection between PMI data and the shipboard SFOC and CII accounting frameworks. For the detailed taxonomy of indicator diagram types, including the power card, draw card, compression diagram, and light-spring diagram, the companion article on indicator diagram analysis should be read alongside this one. For the IMEP-to-BMEP power chain and mechanical efficiency, see engine power and BMEP relationships.

What PMI measures and why it matters

The in-cylinder gas pressure is the one quantity that directly records what thermodynamics is doing inside a firing diesel cylinder. Every other routine measurement, exhaust gas temperature, turbocharger speed, fuel rack position, lubricating oil temperature, is a downstream consequence of what happens at the piston crown. Measuring pressure directly at the source, against a time base tied to crank angle, produces a complete thermodynamic map of the cycle.

From that map, four quantities drive most of the diagnostic work.

Indicated mean effective pressure (IMEP) is the net work delivered by gas to the piston in one complete cycle, divided by the swept volume of that cylinder. It is the area enclosed by the pressure-volume loop on the P-V indicator diagram. IMEP is the only cylinder-level power quantity that can be calculated without any knowledge of friction or bearing loads; it is purely thermodynamic. Because it is normalized by swept volume, IMEP from cylinder 6 on a nine-cylinder engine can be compared directly with IMEP from cylinder 3, regardless of any small machining tolerance differences.

Peak firing pressure Pmax is the highest cylinder pressure reached during the cycle, normally occurring just after top dead centre when the combustion heat-release rate peaks. Pmax is the governing structural load on the cylinder cover, cylinder liner, piston crown, connecting rod, and crankshaft. Engine makers publish a maximum allowable Pmax at each service rating; exceeding it risks fatigue cracking. Running Pmax too low, relative to the target for that load and speed, indicates retarded injection or a defective fuel valve.

Compression pressure Pcomp is the cylinder pressure at top dead centre when the fuel injection is suppressed (a compression-only diagram) or, on continuous-monitoring systems, estimated from the pressure trace immediately before the injection event by projecting the polytropic compression line. Pcomp depends on the compression ratio, the charge-air pressure entering the cylinder, the scavenge efficiency, and the sealing integrity of the exhaust valve and piston ring pack. A falling Pcomp trend on one cylinder, relative to its neighbours, is one of the first reliable indicators of ring or liner wear, or of exhaust valve seat leakage.

Pressure rise rate ($\Delta p / \Delta \theta$, in bar per degree crank angle) governs the rate at which combustion load builds on the cylinder head and connecting-rod. Values above the maker’s limit, typically 5-8 bar per degree on modern slow-speed two-stroke engines, indicate premature ignition or excessively advanced injection timing. The pressure rise rate is calculated automatically from the digitised trace.

Together, these four quantities give a complete picture of whether a cylinder is doing its share of the work and doing it correctly. The engine power and BMEP relationships article explains how IMEP connects to shaft brake power; the present article focuses on how PMI obtains IMEP in the first place.

The indicator diagram: phases and structure

The indicator diagram plots cylinder gas pressure on the vertical axis against either crank angle (a P-theta diagram, also called the draw card or pressure-angle trace) or swept volume from top dead centre (a P-V diagram, also called the power card). Both representations derive from the same pressure-versus-crank-angle signal; the P-V form converts crank angle to piston displacement through the slider-crank geometry of the specific engine.

On a uniflow-scavenged slow-speed two-stroke engine, one complete cycle spans 360 degrees of crank rotation. The trace on a P-V diagram forms a roughly lens-shaped closed loop, with the high-pressure upper limb representing the compression and expansion strokes and the low-pressure lower limb representing the scavenging phase. The enclosed area is the net work. Specifically, for a single cylinder:

where $p$ is cylinder gas pressure (Pa) and $V$ is instantaneous cylinder volume (m³). The integral is taken over one complete cycle.

Dividing the net cycle work by swept volume $V_s$ gives IMEP:

And indicated power per cylinder, summing over cycle frequency $n$ (rev/s for a two-stroke engine where one revolution is one complete cycle):

Total engine indicated power:

where $z$ is the number of cylinders, assuming equal IMEP across all cylinders. In practice the PMI system calculates each cylinder’s IMEP individually and the sum is the total.

For a modern MAN B&W G90ME-C10.5 engine at 85% MCR (approximately 52,000 kW at 79 rpm, bore 900 mm, stroke 3720 mm) the swept volume per cylinder is 2.37 m³ and the cycle frequency is 1.32 rev/s. IMEP at that rating is approximately 21.5 bar (2.15 MPa), giving indicated power per cylinder of about 6,700 kW. Summing eight cylinders gives 53,600 kW indicated; subtracting typical friction losses of 8-10% leaves the 52,000 kW shaft figure, consistent with the published rating.

Phase-by-phase description

Compression phase. Starting from the point at which the scavenge ports close (around 50-65 degrees after BDC on typical uniflow designs), the piston rises and compresses the trapped charge. On a well-maintained engine the compression line follows a polytropic relationship $pV^n = C$ with the polytropic exponent $n$ between 1.35 and 1.40, close to the adiabatic value of 1.38 for air. Deviations from this straight-line relationship on a log-log plot indicate charge leakage (ring or valve seat failure).

Combustion and heat-release phase. Fuel injection typically begins 5-25 degrees before TDC on a slow-speed two-stroke engine at full load, ignition follows after an ignition delay of 3-8 degrees, and peak heat release occurs 5-15 degrees after TDC. The pressure rise during this phase is sharp, reaching Pmax typically at 5-15 degrees after TDC. On modern electronically controlled engines (MAN ME-C series, WinGD X-B series), the injection timing is adjustable per cylinder to hit a target Pmax within narrow limits.

Expansion phase. After Pmax the gases expand as the piston descends. The expansion line also follows a roughly polytropic law, with exponents of 1.25-1.35. A compression and expansion line that are not parallel in the log-log representation indicates combustion continuing well into the expansion stroke, characteristic of late injection or a degraded fuel valve.

Exhaust valve opening and blowdown. On two-stroke engines the exhaust valve opens well before BDC (typically 90-110 degrees before BDC), dropping cylinder pressure rapidly from around 10-20 bar to close to the exhaust manifold pressure of 3-4 bar. The sharpness of this pressure drop is a check on exhaust valve operation.

Scavenging phase. From roughly 60 degrees before BDC to 65 degrees after BDC, scavenge air sweeps the cylinder at manifold pressure (3-5 bar on a well-charged engine). The light-spring diagram, a separate measurement taken at low pressure sensor range, maps this phase in detail; port blockage or turbocharger degradation shows as a lowered scavenge pressure plateau.

Derived quantities in detail

IMEP and indicated power

The numerical integration of the pressure-volume loop is performed by the PMI software from the digitised pressure and crank-angle data. With pressure sampled at 0.5-1.0 degree intervals and piston position computed from the slider-crank geometry (bore, stroke, connecting-rod-to-crank-ratio), the integral is accurate to within about 1% when the transducer calibration is current. On a nine-cylinder 60,000 kW engine at full service power, a 1% IMEP error equates to a 600 kW uncertainty in total indicated power.

The individual cylinder IMEP values are the primary tool for cylinder-load balancing. MAN Energy Solutions specifies that the spread (maximum minus minimum) in IMEP across cylinders should be held within ±0.5 bar of the mean at steady-state load, and that the corresponding spread in Pmax should be within ±3-5 bar of the target value; the exact tolerance appears in the engine-specific instruction manual for each model.

Peak firing pressure Pmax

Pmax is read directly as the maximum value in the pressure trace for each cycle, with no further calculation required. On modern slow-speed two-stroke engines at MCR the target Pmax typically ranges from 180 to 220 bar, depending on the engine design and the rating point selected. The MAN B&W G95ME-C10.5 has a published maximum allowable Pmax of 220 bar. WinGD’s X92DF engine targets Pmax of 195-210 bar at the R1 rating point (100% MCR). Exceeding the structural limit even briefly during a load transient can initiate fatigue cracks in the cylinder head, piston crown, or connecting-rod; the PMI system’s peak-pressure alarm is therefore set at, or slightly below, the structural limit.

Compression pressure Pcomp

On offline portable systems, Pcomp is measured by a dedicated compression-only indicator run: the fuel to that cylinder is shut off, and the pressure at TDC (actually the pressure at 358-360 degrees, where the crank is closest to TDC and pressure is at its highest in a motored cycle) is recorded. On continuous online systems, Pcomp is estimated by projecting the polytropic compression line to the TDC volume, or by reading the pressure immediately before the start of the injection-phase pressure rise.

For a slow-speed two-stroke engine with a geometric compression ratio of 18:1 to 21:1 and charge-air manifold pressure of 3.5-4.5 bar (absolute), Pcomp at full load typically falls in the range 130-175 bar. A value 10 bar or more below the cylinder-average value is the threshold at which MAN Energy Solutions and DNV recommend investigation of ring and liner condition.

Pressure rise rate and combustion timing

The maximum rate of pressure rise $\left(\frac{dp}{d\theta}\right)_{\max}$ occurs during the premixed combustion phase, a few degrees after ignition. For modern slow-speed two-stroke engines the limit stated in most engine manuals is 5-8 bar per crank-angle degree. Expressed as a time rate, at 100 rpm that 8 bar/deg limit equates to about 0.8 MPa per millisecond. Excessive pressure rise rates impose high dynamic bending loads on the connecting rod and bearing journals, and they are audible as diesel knock.

The crank angle at which Pmax occurs (often called the firing angle or pressure angle at peak) is a direct indicator of injection timing. If Pmax occurs at, say, 18 degrees after TDC when the target is 10 degrees after TDC, injection is retarded by approximately 8 degrees relative to the design point. Retarded timing reduces Pmax, pushes heat release further into the expansion stroke, raises exhaust-gas temperature, and reduces thermal efficiency. The SFOC rises in proportion. PMI timing data is the diagnostic tool that separates an injection-timing fault from a fuel-quality issue, because fuel quality changes affect the ignition delay and thus the timing of Pmax even with unchanged injection start.

Offline versus online PMI systems

The technology splits into two distinct product lines, with different use cases and different data density.

Offline portable systems

A portable PMI unit consists of a pressure transducer that screws into the indicator cock on the cylinder cover (the cock is opened to connect the transducer to the combustion space), a crank-angle signal cable connecting to the engine’s crank-angle encoder output, a handheld or laptop-based data logger, and analysis software.

The indicator cock on a two-stroke marine diesel engine is a small manually operated valve with a conical-bore plug and a flanged connection point, standardized so that any indicator unit compatible with that engine family can be connected. On older engines without permanent transducers, the indicator cock is the only in-cylinder access point. The transducer is exposed to full combustion temperatures (peak gas temperature near TDC can exceed 1,600°C) but the measurement duration is short enough, typically 30-60 seconds per cylinder, that thermal overload is not a practical problem with properly cooled transducer designs.

Kistler (Switzerland) and AVL (Austria) are the leading independent suppliers of portable PMI transducers and analyser units used in the marine sector. MAN Energy Solutions and WinGD also supply portable PMI kits matched to their own engines. Resolution of the crank-angle measurement in portable systems is typically 0.5-1.0 degree; some units achieve 0.1 degree by oversampling and interpolation.

The key limitation of the offline approach is that it provides only a snapshot. An intermittent injector fault that manifests at a specific load, an exhaust valve that leaks only when hot, or a slow ring-pack degradation that shows up over weeks rather than hours, are not reliably captured by periodic portable measurements taken once a week or once per port call.

Online permanent systems

A permanent PMI installation uses one piezoelectric pressure transducer per cylinder, fitted into the cylinder cover through a purpose-designed bore aligned with the combustion chamber. The transducer is pressure-cooled (using a thin water jacket supplied from the cylinder cooling-water circuit) to keep its body below 180°C despite the adjacent combustion gas reaching 1,600°C at peak. Typical sensor types are charge-mode piezoelectric (producing a charge proportional to pressure, integrated by a charge amplifier to give a voltage signal) or piezoresistive designs used in lower-temperature applications.

Crank-angle data comes from a high-resolution encoder, typically 1,024 or 4,096 pulses per revolution, on the free end of the crankshaft or on a dedicated measurement shaft. This sets the sampling density: at 1,024 pulses per revolution on a 100-rpm engine, the time between samples is about 0.58 milliseconds (0.35 degrees), which is sufficient to resolve Pmax and pressure-rise rate accurately. Some installations sample at 0.1 or 0.25 degree to improve heat-release analysis.

On MAN ME-C series engines, the electronic control system includes a permanent PMI installation as standard equipment. The PMI data feeds the Engine Control System (ECS), which compares the measured IMEP and Pmax for each cylinder against the target values and applies corrections to the fuel injection timing and quantity, and to the exhaust valve timing, to bring each cylinder onto its target. This closed-loop cylinder balancing is automatic and continuous; the engineer monitors the resulting balance quality on the engine room display but does not need to manually tune individual cylinders under normal operating conditions. MAN reports that closed-loop cylinder balancing on the ME-C series maintains IMEP spread within ±0.3 bar and Pmax spread within ±3 bar under steady-load conditions.

WinGD’s X-series engines (X52, X62, X72, X82, X92 bore classes) and X-DF dual-fuel variants carry a similar integrated PMI and cylinder-balancing system. The WinGD Integrated Automation System (IAS) interfaces with the PMI data to provide real-time cylinder condition status, trend analysis, and fleet cloud connectivity through the WinGD ALFRED analytics platform.

Crank-angle reference, TDC calibration, and measurement accuracy

Every PMI-derived quantity depends on accurate knowledge of where the piston is in the cycle when each pressure sample is taken. The critical reference point is top dead centre (TDC), the crank angle at which the piston is at its smallest distance from the cylinder head.

An error in the TDC reference translates directly into an error in combustion timing (the apparent crank angle of Pmax and the heat-release peak). A 1-degree TDC error shifts the apparent Pmax timing by 1 degree, which changes the apparent IMEP by roughly 0.1-0.2 bar (less than 1% of absolute IMEP) but shifts the apparent injection timing by a degree, potentially masking a real timing drift. On MAN B&W engines the TDC mark on the flywheel and the corresponding mark on the engine room casing are the primary reference; the PMI system’s encoder is synchronized to this mark during commissioning and checked at each calibration interval, typically every 2,000 running hours or at each port call for portable systems.

Pressure transducer drift is the second major accuracy concern. Piezoelectric charge-mode transducers do not have an absolute zero reference; they measure pressure changes, not absolute pressure. The PMI software must establish a reference pressure at some point in the cycle where the pressure is known, typically at the start of the compression phase when the cylinder is communicating with the scavenge manifold and the pressure is therefore equal to the manifold pressure (which is measured independently). This process, called the “pegging” or auto-zero correction, is performed automatically on online systems once per cycle or once per averaging window. A pegging error of 1 bar in the absolute reference shifts all derived quantities including Pcomp and Pmax by the same 1 bar, and shifts IMEP by approximately 0.2-0.5 bar depending on the engine geometry.

Kistler’s specification for their 6613C marine PMI transducer states a sensitivity of 17 pC/bar, a range of 0-300 bar, an operating temperature up to 350°C at the tip, and a linearity better than ±0.5% of full scale. At a calibration standard of 0.5% of 300 bar, the absolute pressure uncertainty is ±1.5 bar. Lloyd’s Register’s guidance on cylinder pressure monitoring recommends that operators verify transducer calibration using a dead-weight tester at least every 12 months and after any maintenance event that involves removing the transducer.

Crank-angle encoder accuracy matters too. An encoder with 1,024 pulses per revolution has a raw angular resolution of 0.352 degrees. The TDC reference (indexing the encoder to the flywheel mark) introduces an uncertainty of roughly ±0.1-0.2 degrees depending on the width of the flywheel mark and the method of detection. Total angular uncertainty is therefore about ±0.3-0.5 degrees, sufficient for most operational purposes but potentially limiting for research-grade heat-release analysis.

Cylinder balancing and fault diagnosis

The balancing problem

A diesel engine’s cylinders are mechanically identical by design, but operational life introduces differences. Fuel injectors wear at different rates, exhaust valve seats erode differently, piston rings wear to different clearances, scavenge port cleanliness varies, and small differences in charge-air distribution along the scavenge manifold produce different trapped-air masses. The result is that without active balancing, the cylinders produce different amounts of work, creating uneven torque ripple and uneven thermal loading on crankshaft, bedplate, and main bearings.

On older mechanically controlled engines, the engineer balanced the engine by manual adjustment of individual fuel-pump racks and injection-timing marks, checking the results with a portable PMI unit after each adjustment. On electronically controlled engines with permanent PMI, the engine control system performs this continuously.

IMEP spread and Pmax spread as diagnostic numbers

The engine’s overall load balance is assessed from two spread statistics computed by the PMI system:

IMEP spread: the difference between the highest and lowest individual cylinder IMEP values, expressed in bar. MAN Energy Solutions specifies that the IMEP spread at steady load should not exceed 1.0 bar for a well-balanced engine, and triggers an automated adjustment if any cylinder’s IMEP deviates more than 0.5 bar from the mean.

Pmax spread: the difference between the highest and lowest individual cylinder Pmax values. Pmax spread is controlled simultaneously with IMEP spread by adjusting injection timing per cylinder: advancing timing raises Pmax (and, if correctly balanced, also raises IMEP); retarding timing lowers Pmax. The typical target spread for Pmax is ±5 bar from the mean value, tighter on high-output modern engines.

IMEP spread and Pmax spread are reported per load step and stored as trend data. A growing Pmax spread at constant load is one of the early indicators of injector wear, because a worn injector with a larger nozzle opening delivers fuel at lower velocity and over a longer period, delaying combustion and reducing Pmax while raising exhaust temperature.

Fault diagnosis from pressure trace shape

The indicator diagram carries diagnostic information beyond the four scalar quantities. Shape anomalies visible on the trace reveal specific fault modes.

Leaking exhaust valve. A leaking exhaust valve allows combustion gas to escape back through the valve seat during the compression stroke, before the valve fully seals. The compression line shows a kink or step at the crank angle where the valve seals, because the rate of pressure rise changes as the combustion chamber suddenly becomes fully sealed. Pcomp is depressed relative to neighbouring cylinders even though the valve eventually seals before TDC; IMEP is reduced because the effective compression ratio is lower at the start of injection. The leak also deposits combustion products on the valve spindle, seat, and guide, accelerating wear. The exhaust valve actuation in two-stroke engines article covers the valve design in detail; PMI is the primary tool for identifying valve condition between planned overhauls.

Worn piston ring pack. Ring wear reduces the sealing between the piston crown and the liner, allowing combustion gas to blow past the rings into the crankcase. This shows as a suppressed Pcomp (the compression line rises less steeply than expected and reaches a lower peak) with a normal or near-normal Pmax if injection timing is unchanged. The distinction from an exhaust valve leak is that ring blow-by produces a smooth compression line with no kink; only the slope is reduced. Cross-checking with the lube oil consumption data and the piston-rod stuffing-box leakage (on crosshead engines) confirms the diagnosis. See cylinder liner design for two-stroke engines for the liner geometry context.

Late or retarded fuel injection. Retarded injection shifts the Pmax peak to a later crank angle, reduces Pmax, and raises exhaust-gas temperature because combustion occurs later in the expansion stroke when the gas has less time to do work before exhaust valve opening. The heat-release analysis from the P-theta trace shows the ignition event moved to a later angle. Exhaust temperature alone cannot distinguish retarded injection from a partially blocked injector nozzle, which produces a similar temperature effect for different mechanical reasons. PMI resolves the ambiguity: retarded injection shows a normal Pcomp but a reduced and late Pmax; a blocked nozzle shows reduced fuel flow and correspondingly lower Pmax but with a normal ignition angle once fuel does ignite.

Early or advanced injection. Advanced injection raises Pmax and advances its timing toward or slightly before TDC. The pressure-rise rate $dp/d\theta$ increases, potentially exceeding the structural limit. The trace shows the maximum pressure occurring before the nominal 5-15 degrees after TDC window. Sustained advanced injection on one cylinder risks fatigue of the cylinder cover stud threads and the piston crown. The PMI high-Pmax alarm catches this before structural damage occurs.

Fouled scavenge ports or low charge-air pressure. Fouled scavenge ports reduce the mass of fresh air trapped per cycle. This shows as a lower Pcomp (less air mass, lower compression pressure despite the same compression ratio) and a lower IMEP. The light-spring diagram, which maps the scavenging phase at low pressure resolution, shows a lower-than-normal plateau. Distinguishing a turbocharger fault from scavenge port fouling uses the turbocharger speed and charge-air temperature data alongside the PMI trace. See scavenge port geometry and timing for the design context.

Misfire. A cylinder that fails to ignite fuel produces an indicator diagram that follows the compression line to TDC and then returns along a near-identical expansion line, with no combustion pressure rise. The IMEP of a misfiring cylinder is slightly negative (the work of compression is done by the crankshaft but not recovered by combustion). Online PMI detects misfires on the first cycle they occur; offline systems miss transient misfires.

Fuel valve degradation (worn injector nozzle). A worn or eroded injector nozzle produces a distorted spray pattern and delayed atomization. The indicator diagram shows a double-humped or extended heat-release curve in the rate-of-pressure-rise analysis, with Pmax occurring later and at a lower peak than expected. A stuck-open nozzle needle produces an excessively rich mixture and a high-IMEP, high-Pmax trace. See fuel valve injector design for two-stroke engines for injector design context.

Integration with SFOC, CII, and condition-based maintenance

SFOC and the PMI power balance

Specific fuel oil consumption (SFOC) at the engine level is defined as:

where $\dot{m}_f$ is the measured fuel mass flow rate (g/h) and $P_b$ is the shaft brake power (kW). SFOC is therefore a ratio of two measured quantities. The PMI system is not the direct source of either measurement, but it provides the cylinder-level breakdown that explains why SFOC deviates from the baseline.

If one cylinder’s IMEP degrades by 0.5 bar from worn injectors while the engine control system holds total shaft power constant by increasing fuel to that cylinder (or advancing timing), the SFOC rises. The PMI trace identifies which cylinder is consuming more fuel for the same work output. Without PMI, the engineer sees elevated SFOC but has no cylinder-level resolution to pinpoint the cause. With PMI, the trajectory, one cylinder’s IMEP falling while fuel flow to that cylinder stays constant or rises, points directly to the fuel valve or injection system.

The specific fuel oil consumption curves article covers the load-SFOC relationship; PMI data is the prerequisite for diagnosing deviations from that baseline.

CII and the PMI connection

The Carbon Intensity Indicator (CII), mandated under IMO MARPOL Annex VI Regulation 28 as amended by MEPC.337(76), rates vessel carbon intensity based on total CO₂ emitted per cargo-transport work (gCO₂ per deadweight tonne-nautical mile for most vessel types). Because CO₂ emissions are calculated from fuel consumption (using the MARPOL Annex VI fuel oil conversion factors), any factor that increases fuel consumption at a given transport output degrades CII.

PMI data creates a direct mechanical link from cylinder condition to CII grade. A vessel whose main engine runs with IMEP spread exceeding 1.5 bar, or with one cylinder’s injection timing 5 degrees retarded from the target, burns more fuel than a well-balanced engine at the same shaft power. The penalty can be 0.5-2% additional fuel consumption, which on a capesize bulk carrier burning 60 tonnes per day amounts to 0.3-1.2 additional tonnes of HFO daily. Over a 365-day trading year at current HFO prices, that is both a direct cost and a CII penalty that can push a vessel from a C rating into a D rating under the current CII grading thresholds. See what is CII and slow steaming and CII for the CII regulatory and operational context.

Condition-based maintenance

Traditional planned-maintenance schedules on marine diesel engines specify fixed-interval overhauls, for example, piston overhaul every 18,000-24,000 running hours on a typical slow-speed two-stroke engine. Condition-based maintenance (CBM) replaces fixed intervals with data-triggered decisions: a piston is overhauled when the PMI data shows that compression pressure has fallen by 10 bar from baseline, or when cylinder oil consumption crosses a threshold, or when ring-gap measurements taken at the last inspection are projected (by wear rate) to exceed the class limit within the next 2,000 hours.

PMI is the primary sensor for the cylinder-side CBM data stream. CIMAC Working Group 17’s 2016 guidance document on condition monitoring of marine diesel engines identifies cylinder pressure monitoring (the CIMAC term for what MAN calls PMI) as the single highest-value data source for slow-speed two-stroke engine CBM, because it directly observes both combustion quality (IMEP, Pmax, timing) and sealing integrity (Pcomp, ring and valve condition).

DNV’s notation ECO (Enhanced Condition Monitoring for Machinery) and Lloyd’s Register’s CIMS (Continuous Integrated Maintenance System) notation both require PMI or cylinder pressure monitoring data as part of the monitoring suite for the notation to be awarded. These class notations allow extended overhaul intervals, in some cases to 36,000 hours for piston overhaul rather than the standard 24,000 hours, provided that the CBM data shows no deterioration trend.

The data chain in a modern CBM implementation typically runs from the PMI hardware through the vessel’s integrated automation system (IAS) to a vessel performance monitoring system (VPMS), then via satellite to an onshore fleet operations center. MAN Energy Solutions markets this as the PMI Online system integrated with CEUS (Combustion and Engine monitoring and Utilization System); WinGD offers the equivalent through ALFRED (Algorithmic Fleet Reliability Enablement and Diagnostics). Both systems aggregate PMI data across the engine’s operating history, display trend lines for each cylinder’s IMEP, Pmax, and Pcomp, and generate automated maintenance recommendations when a trend crosses a threshold.

Portable versus permanent: what the class societies require

Neither IMO nor SOLAS mandates specific PMI equipment on ships; the requirement is for the ship to be maintained in a seaworthy condition, and the class society rules specify the survey and maintenance regime.

DNV’s Class Programme DNV-CP-0484 (“Machinery condition monitoring”) requires that vessels holding the ECO class notation have a continuous cylinder pressure monitoring system (effectively, online permanent PMI) for all main propulsion diesel engines above 2,000 kW cylinder output. Below that threshold, quarterly portable PMI measurements suffice.

Lloyd’s Register’s Rules for Ships Part 5, Chapter 12 covers engine condition monitoring and references “cylinder pressure measurement” as a mandatory item in the periodic survey programme; the frequency depends on the engine rating and the maintenance system notation held by the vessel. LR’s CIMS notation requires online PMI for the main engine.

Bureau Veritas, American Bureau of Shipping, and ClassNK have equivalent requirements within their respective machinery survey programmes. Common to all: a record of cylinder pressure measurements at defined intervals, with comparison against the engine maker’s baseline data, is required as documentary evidence at the periodical survey.

For vessels without online PMI, the chief engineer is expected to perform portable PMI measurements at intervals defined by the engine maker and the class society, typically every 500-1,000 running hours for new or recently overhauled engines and monthly (at 24-hour loads) for engines in steady-state trading. Results are logged in the planned-maintenance system and produced at class survey.

Calibration, practical operation, and data quality

A PMI system produces useful data only if the transducers are calibrated and the pegging reference is correct. In practice, three error sources dominate data quality on shipboard systems.

First, transducer sensitivity drift: piezoelectric charge-mode transducers lose sensitivity over time as the piezoelectric ceramic ages. Kistler and AVL quote aging rates of 0.5-1% per year for their marine-grade sensors; after five years of service without recalibration, the indicated pressure could be systematically 2-5% low, which would understate IMEP by the same proportion. Annual calibration with a reference dead-weight pressure tester or an in-situ shunt calibrator is the standard remedy.

Second, pegging error: if the scavenge manifold pressure sensor used as the pegging reference is itself miscalibrated, or if the indicator cock is not fully open during the reference measurement, the absolute pressure baseline shifts. A 0.5-bar pegging error shifts all pressure values by 0.5 bar; IMEP moves by roughly 0.1-0.2 bar (depending on the engine’s bore-stroke ratio), and Pmax moves by the full 0.5 bar. Regular comparison of the PMI-indicated scavenge pressure against the engine’s own manifold pressure gauge provides a quick check.

Third, cable and connector integrity: the charge cable between the transducer and the charge amplifier is a low-capacitance coaxial cable that must be kept dry and away from high-voltage ignition sources. Even a small cable defect introduces electrical noise that can produce false spikes in the pressure trace, corrupting single-cycle data. The practical fix on ships is to use the cycle-averaging function (averaging over 30-100 consecutive cycles) which suppresses random electrical noise; single-cycle analysis for misfire detection or knock detection requires a clean signal path.

Limitations

PMI measures combustion-chamber thermodynamics precisely but has no direct visibility into the mechanical failure modes that occur outside the combustion chamber. It does not detect bearing shell fatigue, crankshaft fatigue, crosshead pin wear, or turbocharger bearing wear. A complete condition-monitoring programme combines PMI with vibration analysis (for bearing and structural faults), lube-oil debris monitoring (for metal wear particles), crankcase oil mist detection, and exhaust-gas composition monitoring.

Pcomp as estimated by online PMI from the continuous trace has a systematic uncertainty of 2-5 bar compared to the dedicated compression-only diagram taken with fuel shut off. The continuous-monitoring estimate relies on projecting the polytropic line to TDC volume, but any combustion that starts before TDC interferes with the projection. For diagnostic decisions where Pcomp precision is critical, a dedicated compression diagram with fuel shut off remains the gold standard.

Cycle-to-cycle pressure variation is real and can be substantial: cycle-to-cycle variation in Pmax on a well-running slow-speed two-stroke engine is typically 3-8 bar (1-4% of mean Pmax). Averaging over 30-100 cycles, as most commercial PMI systems do for the display value, suppresses this variation and reveals the trend clearly, but it also masks short-duration faults like occasional injector sticking or intermittent ring flutter. The engineer should know whether the displayed values are averages or single-cycle readings before interpreting small deviations.

The indicator cock introduces its own complication for offline portable systems. The small bore of the indicator cock (typically 4-8 mm on large slow-speed engines) acts as an orifice between the combustion space and the transducer; the pressure at the transducer lags the combustion-space pressure by a crank-angle amount that depends on the cock bore, the connecting-pipe volume, and the rate of pressure change. This dynamic damping is negligible for the slowly varying compression and expansion phases but can distort the measurement of fast events near Pmax. Makers of portable PMI units account for this by specifying a maximum connecting-pipe length (typically 100-150 mm from the cock to the transducer) and by providing a frequency-response correction in the analysis software.

Temperature compensation of the transducer is a further practical issue. The piezoelectric constant of the sensing element changes with temperature, typically by 0.1-0.3% per 10°C change in tip temperature. A transducer that heats up during a long measurement session will drift in sensitivity. Modern marine transducers incorporate a temperature sensor at the tip and apply a correction; older or budget units do not.

Finally, PMI data tells you what the in-cylinder conditions are at the time of measurement. It does not predict how conditions will evolve unless trend data is available. A single offline measurement taken after a fuel-system clean and before a port call gives a clean-condition snapshot; the trajectory between cleanings is invisible. Continuous online PMI closes this gap but requires capital investment and ongoing maintenance of the permanent transducer installation.

Practical workflow on an oceangoing vessel

A typical onboard PMI measurement cycle on a vessel with online PMI runs as follows. At each watch change (typically every 4 hours), the engineer checks the PMI console for active alarms: IMEP spread above 1.0 bar, any individual cylinder’s Pmax above the high-Pmax alarm threshold, or any Pcomp below the low-Pcomp alarm threshold. These trigger immediate investigation. At the weekly engineering review the engineer pulls 24-hour trend data for all cylinders, checks for gradual drift in Pcomp or Pmax on any cylinder, and documents the findings in the planned-maintenance log. At each arrival and departure, when engine load changes significantly, the engineer confirms that cylinder balance within the new load band meets the maker’s specification.

For vessels with offline portable systems only, the standard workflow is quarterly full-engine PMI measurements at steady-state sea speed, with the results compared against the previous quarter’s values and against the engine maker’s baseline data sheet supplied with the engine. Class surveyors inspect these records at the annual survey. The CIMAC WG17 guidance recommends that at least two sets of back-to-back measurements are taken per session, at the same load, to confirm that the values are stable rather than transient.

The comparison against baseline is as important as the absolute values. A Pcomp of 155 bar on a cylinder whose baseline is 160 bar is more informative than an isolated 155-bar reading; the 5-bar drop over a known service period points to a specific degradation rate that can be compared against the ring-wear model to project the next intervention. Class society guidance from Lloyd’s Register recommends maintaining a cylinder history card for each cylinder, recording Pcomp, Pmax, and IMEP at each measurement, dating each entry, and including a note of any maintenance performed.

See also

• Indicator Diagram Analysis on Marine Diesel Engines
• Engine Power and BMEP Relationships
• Cylinder Peak Pressure (Pmax) Analysis on Marine Engines
• Cylinder Compression Pressure (Pcomp) Analysis on Marine Engines
• Specific Fuel Oil Consumption Curves
• Exhaust Valve Actuation in Two-Stroke Engines
• Cylinder Liner Design for Two-Stroke Engines
• Fuel Valve and Injector Design for Two-Stroke Engines
• What Is CII
• Slow Steaming and CII
• MAN B&W ME-C Electronic Control Overview
• Scavenge Port Geometry and Timing

Related calculators:

• Indicated Mean Effective Pressure (IMEP)
• Engine BMEP
• Engine Pmax to Compression Pressure Ratio
• Engine Cylinder Balance
• Mechanical Efficiency