What the pressure trend diagnostic workbook is for
A single gauge reading tells you very little. A high reading in the compressor room can coexist with a starved machine 200 metres away, and a low reading at a machine can be caused by a clogged filter, a sequencing fault, a sudden large user or a faulty transmitter. The workbook structures pressure logs so that these explanations can be separated before money is spent.
Use it for:
- Intermittent machine stoppages or quality faults that operators blame on “low air”
- Repeating header dips at shift changes, start-up or particular production steps
- Pressure oscillation between compressors or after a control change
- A slowly rising pressure drop across the dryer or filters
- Unexplained changes after maintenance, a new machine or a piping modification
The workbook is a structured analysis record. It is not a network simulator, a compressor controller or a vessel-sizing approval. Its job is to tell you where to look and what test will confirm it.
Step one: define the failure before collecting data
Write down, in plain terms, what goes wrong and when. “Line 3 filler stops with a low-air alarm two or three times a shift, usually after a changeover” is a usable definition. “Pressure is low” is not. For the affected machine, record:
- The pressure it needs at its inlet and where that point is (before or after its own regulator)
- How long a dip it can tolerate before it trips or produces scrap
- The times and dates of recent events, from the machine’s alarm log if it has one
This defines the threshold and duration that count as an event in the analysis.
Where should pressure sensors be placed?
Place sensors where they divide the system into sections that can be blamed separately. A practical minimum is:
| Location | What it tells you |
|---|---|
| Compressor discharge or common header in the compressor room | What the supply is doing and how the controls respond |
| After the dryer and filter train | Treatment loss, and whether it rises with flow or with time |
| Main header near the affected area | Distribution loss in the trunk lines |
| Inlet of the affected machine, after its hose and coupling | What the machine actually receives |
| Optional: second machine on the same branch | Whether the problem is branch-wide or local to one machine |
Log compressor state alongside pressure: load, unload, off, VSD speed or motor current. Where possible, log flow at the header or power at each compressor. Pressure without demand and supply information is half the picture.
Measurement, clocks and logging interval
Before analysing anything, check the measurement chain:
- Compare sensors on a common point at the start of the campaign. A 0.2 bar disagreement between two transmitters can invent or hide a whole section loss.
- Keep units and references consistent. Record everything in bar(g) or psig; never mix gauge and absolute values in the same trend.
- Synchronise clocks. Loggers drift. Record the offset of each logger against a reference clock at the start and end. A one-minute offset can make a downstream dip appear before the upstream one.
- Match the logging interval to the event. A slow sag over an hour can be seen at one-minute intervals; a two-second dip when a large valve opens needs a one-second interval or faster. Averaging hides short events.
- Cover enough operating states. Include start-up, normal production, peaks, breaks, weekends and at least a few of the reported events.
Keep the raw files untouched. Do any alignment, filtering or gap-filling on a copy, and note what was done.
How to calculate section losses from the trends
At each timestamp, subtract the downstream pressure from the upstream pressure for each pair of sensors. This gives simultaneous section losses: treatment loss, header loss and point-of-use loss. Plot them against time and against flow or compressor load.
Then define events explicitly. An event starts when the machine-inlet pressure falls below its minimum, ends when it recovers above that minimum, and is measured by its depth (lowest pressure), duration and recovery time. For each event, note the section losses at the worst moment and the compressor state just before.
Summarise with duration-weighted figures, not simple averages of samples, and report data coverage. If a logger was off for 20% of the campaign, say so; the missing period may contain the events that matter.
Pattern recognition: what the trend shapes usually mean
Treat each pattern as a hypothesis to be tested, not a diagnosis.
| Pattern seen | Likely explanations | Test that separates them |
|---|---|---|
| Header steady, only the machine inlet dips | Hose, coupling, regulator or drop restriction; machine-internal fault | Fit a temporary gauge before and after the regulator and hose during a cycle |
| Whole system sags slowly at peak, compressors fully loaded | Capacity shortfall, high leakage, an added user | Leak test at a non-production time; compare loaded time with the previous baseline |
| Sharp dips when a large user starts, compressors respond late | Insufficient storage near the user, slow control response, standby machine start delay | Log at one-second intervals; note start and load times of each compressor |
| Treatment loss rises over weeks at similar flow | Filter elements loading, dryer heat-exchanger fouling | Read filter differentials; change elements and repeat |
| Regular oscillation between compressors | Overlapping pressure bands, two machines fighting for trim, sequencer settings | Map each compressor’s load and unload points against the trend |
| Downstream dip appears before upstream dip | Clock offset between loggers | Re-check logger offsets; align on a known event |
| Readings flat or stepped, with no noise | Sensor fault, frozen value or logging resolution | Check the transmitter and logger settings |
Capacity, control response or availability?
Three different supply problems look similar on a header trend:
- Capacity: all available compressors are loaded and pressure still falls. Demand exceeds supply, whether from production, leakage or misuse.
- Control response: a compressor is available but unloaded, off or at low VSD speed when the dip starts, and loads too late. The fix is in pressure bands, sequencing, start times or storage, not a new machine.
- Availability: a compressor tripped, was isolated or was under maintenance. The trend shows a step change in supply.
Storage interacts with all three. A receiver close to a large intermittent user can supply a short event while the compressors catch up. The trend can show whether storage is helping; sizing it is a separate calculation in the receiver transient storage calculator.
Worked examples
Example A: a local path dominates
During events, the compressor-room header reads 6.9 bar(g) (about 100 psig) and the area header 6.7 bar(g), but the machine inlet falls to 5.2 bar(g) against a minimum of 5.8 bar(g). The point-of-use loss is 1.5 bar, while the distribution and treatment loss is only 0.2 bar. Compressors are not fully loaded. The answer lies in the hose, coupling and regulator, not in supply. Raising the setpoint would cost energy everywhere and still might not cure it.
Example B: missing time hides the answer
A week-long log shows only two minor dips, but the machine logged nine trips. Checking coverage shows the machine-inlet logger stopped for two nights when its battery ran low, and seven trips fell in that period. The conclusion is that the data are insufficient, not that the problem is minor. The logger is restarted with mains power for a second campaign.
Example C: a clock offset imitates reverse flow
The machine-inlet dip appears to start 40 seconds before the header dip, which is physically implausible. Comparing both loggers against a known event, a compressor trip that both recorded, reveals a 45-second offset. After alignment, the header dip leads by about five seconds and the event is traced to a compressor that loads late.
Turning findings into action
- State the conclusion, the evidence for it and what remains uncertain.
- Choose the cheapest test or correction that would confirm it: an element change, a hose replacement, a control-band adjustment by a competent engineer.
- Repeat the logging under similar demand and compare event counts, depth and duration.
- Only consider extra capacity once restrictions, leakage, control and storage have been addressed, and the trend still shows all compressors loaded during events.
Changes to compressor controls, safety settings or treatment bypasses must be made by a competent person with the equipment maker’s guidance; the workbook does not authorise them.
How Reddy Compressors helps
We carry out pressure profiling and logging on installations of all major brands, including ELGi, Atlas Copco, Ingersoll Rand, KAESER and Chicago Pneumatic. We are an independent multi-brand specialist, not an authorised dealer. We have tie-ups with all major and reputed brands. Our engineers place loggers at the section boundaries, record compressor state, analyse the events and explain the findings in plain terms. Where the fix is a part, a control setting or a dryer service, we can carry it out. Where extra air is genuinely needed, we can supply a new or used compressor, or rent one from our fleet while a permanent decision is made.
Our office is at Tarnaka, Secunderabad, and our service teams cover the whole of Telangana and Andhra Pradesh. Equipment, rentals and spares are supplied across India.
What to send us
- A description of the problem: which machine, what happens, how often and when
- Compressor makes, models and current pressure settings, and whether a sequencer or VSD is fitted
- Dryer and filter details and recent maintenance history
- A sketch of the header and the location of the affected machine
- Any existing trend data, with a note of the logging interval and units
Frequently asked questions
Why does my compressed-air pressure drop at certain times of day?
Regular dips usually match a demand event: a shift start, a large machine starting, cleaning with blow guns, or several users running together. They can also match compressor events, such as a machine unloading or a standby compressor starting late. Logging pressure and compressor state together, at the right interval, shows which event lines up with each dip and whether supply, storage or a restriction is responsible.
How do I know if I need a bigger air compressor?
You need more capacity only if all available compressors are fully loaded during the low-pressure events and the pressure still falls, after leakage, restrictions and control settings have been checked. If a compressor is unloaded or off when the dip begins, the problem is control response or storage. If the header stays steady while one machine dips, the cause is local to that machine’s supply.
What logging interval should I use for pressure trends?
Match it to the event you are looking for. Slow sags over minutes or hours can be seen with one-minute logging. Short dips caused by a valve opening or a cylinder bank firing may last only seconds and need one-second logging or faster. Averaged data hides short events, so check how the logger records values and keep the raw files.
Where should I fit pressure gauges in a compressed-air system?
Fit them at the compressor or common header, after the dryer and filters, on the main header near the affected area and at the inlet of the critical machine after its hose and coupling. These positions split the system into sections, so the loss in each can be measured at the same moment. Check all gauges against each other on a common point before relying on the differences.
Why do my two compressors keep loading and unloading against each other?
This usually happens when their pressure bands overlap, so both try to act as the trim machine, or when a sequencer or VSD target is set inconsistently with the fixed-speed machine’s band. The trend shows a regular oscillation in header pressure and alternating load states. A competent engineer should map each machine’s settings and adjust them so that one compressor clearly leads and the others follow.
Can a faulty pressure sensor cause false low-pressure alarms?
Yes. A drifting transmitter, a blocked impulse line, a loose connection or a wrongly scaled input can all report low pressure that does not exist. Signs include flat or stepped readings, values that disagree with a nearby gauge and alarms with no matching effect on production. Comparing the sensor against a calibrated gauge on the same point is a quick first check.
Related pages
Calculators and worked examples
These tools give indicative figures for screening, comparison and for preparing an enquiry. They run in your browser and send nothing. No prices or tariffs are assumed: enter your own to see costs. Efficiencies and other figures shown are example values: replace them with your own. Results are not a performance guarantee or a design certificate; final selection is confirmed by a service engineer against the machine data sheet and the site.
Section losses at the worst moment of a pressure dip
Enter the pressures read at the same moment at the four section boundaries during a low-pressure event. The tool splits the loss into treatment, distribution and point-of-use sections and points to the part of the system to test first.
The worked example below shows the method with fixed numbers. Turn on JavaScript to change the inputs.
Worked example
Inputs
- Compressor-room header (bar g): 6.9
- After dryer and filters (bar g): 6.8
- Main header near the affected area (bar g): 6.7
- Machine inlet, after its hose and coupling (bar g): 5.2
- Minimum pressure the machine needs at that point (bar g): 5.8
- Compressor state when the dip started: At least one compressor unloaded, off or below full speed
Result
- Treatment loss: 0.1 bar
- Distribution loss: 0.1 bar
- Point-of-use loss: 1.5 bar
- Total, compressor room to machine inlet: 1.7 bar
- Machine inlet against its minimum: -0.6 bar (below the minimum)
- Largest section loss: Point of use (hose, coupling, regulator, drop) 88% of the total
- Where to look first: Local path: the area header is above the machine’s minimum, so the shortfall is created between the header and the machine. Check hose, coupling, regulator and drop (gauge before and after them during a cycle) before touching supply.
Method. Section loss = upstream pressure − downstream pressure, all in bar(g) at the same timestamp: treatment = compressor-room header − after dryer/filters; distribution = after dryer/filters − area header; point of use = area header − machine inlet (hose, coupling, regulator, drop). If the area header is at or above the machine minimum while the inlet is below it, the shortfall is created locally. If the compressor-room header is above the minimum but the area header is below it, the shortfall is created in treatment or distribution. Only if the compressor-room header is itself below the minimum does the compressor state at the start of the dip separate capacity (all loaded), control response or storage (a machine unloaded, off or slow) and availability (a machine tripped or isolated). A negative section loss means the two sensors disagree: compare them on a common point.
What to take from it. Raising the compressor set-point costs energy everywhere. It is only the fix when the loss is in supply, not in one machine’s hose and regulator.
Technical content reviewed by D. Srinivas Reddy, Managing Partner and service engineer, Reddy Compressors. Last updated 17 October 2026. How our engineering content is reviewed.
Pressure event count, depth, duration and data coverage
Paste the machine-inlet pressure log at a fixed interval. The tool finds every event below the machine’s minimum and reports its depth, duration and recovery time, together with how much of the campaign the log actually covers.
The worked example below shows the method with fixed numbers. Turn on JavaScript to change the inputs.
Worked example
Inputs
- Machine-inlet pressure readings in time order (bar g): 6.1, 6.1, 6.0, 6.0, 5.9, 5.7, 5.4, 5.2, 5.3, 5.6, 5.9, 6.0, 6.1, 6.1, 6.0, 6.1, 5.8, 5.6, 5.5, 5.7, 5.9, 6.0, 6.1, 6.1
- Logging interval (s): 10
- Minimum pressure the machine needs (bar g): 5.8
- Campaign length the log should cover (min): 5
Result
- Logged time: 4 min (24 readings)
- Data coverage: 80 % of the campaign
- Events below the minimum: 2
- Time below the minimum: 80 s = 33.3% of the logged time
- Lowest reading: 5.2 bar g
- Deepest event: lowest pressure: 5.2 bar g, 0.6 bar below the minimum, starting at reading 6
- Deepest event: duration: 50 s
- Deepest event: recovery time: 30 s from the lowest reading back to the minimum
- Longest event: duration: 50 s, starting at reading 6
- Event rate: 30 events per hour of logged data
Method. An event starts at the first reading below the minimum and ends at the first reading back at or above it. Each reading is taken to represent one logging interval, so duration = number of readings below the minimum × interval. Depth = lowest reading in the event. Recovery time = time from the lowest reading to the first reading back at or above the minimum. Time below the minimum is duration-weighted over the logged time. Coverage = logged time ÷ campaign length; events in unlogged periods cannot be seen.
What to take from it. Few events in a log with poor coverage means insufficient data, not a minor problem. Report coverage with every event count.