Estimating PLC I/O Count for a Project Bid.
How to count analog and digital I/O from engineering drawings without a full tag-by-tag walk, how to size card racks off the estimate, and the rules of thumb that hold up during detailed design.
An I/O count drives the hardware budget, the cabinet layout, the wiring schedule, and a large portion of the programming estimate. Getting it wrong on a bid means either leaving margin on the table or losing the job when the real design reveals the shortfall.
This is a practical guide to the numbers that hold up, the counting method that works for bid-stage estimates, and the pitfalls that cause overruns.
The four signal classes
Every PLC point is one of.
| Class | Abbreviation | Typical source | Typical destination |
|---|---|---|---|
| Analog input | AI | Transmitter, flow, pressure, level, temp | PLC reads a scaled value |
| Analog output | AO | PLC writes a scaled value | Control valve, variable-speed drive setpoint |
| Digital input | DI | Switch, limit, status contact | PLC reads a BOOL |
| Digital output | DO | PLC writes a BOOL | On, off valve, motor starter, solenoid |
A rough-cut estimate only needs to count these four. HART and Foundation Fieldbus instrument signals still land on AI cards in the I/O count. The difference shows up in card cost and module selection, not in the count itself.
The structured counting method
A bid-stage estimate follows five steps in sequence. Skipping a step is where estimates go wrong.
Step 1. Extract the instrument list from drawings
Walk the P&IDs with the decision tree below applied to each instrument bubble. If you are working from digital drawings, extraction tooling produces this list automatically. If you are working from paper or scanned drawings, manual tagging is the only option for the first few pages. Extrapolation fills in the rest. How that raw instrument list becomes a properly structured I/O list with signal class, area, and protocol columns is covered in the I/O list creation guide.
Decision tree per instrument bubble.
- Transmitter or indicator driven by a transmitter. 1 AI
- Local indicator or gauge with no wiring symbol. 0, no PLC point
- Control valve with modulating actuator. 1 AO, plus 1 DI for position feedback if present
- On-off valve or solenoid. 1 DO, plus 1 DI for each position limit switch
- Pressure, level, or flow switch, discrete. 1 DI
- Motor start, stop. 1 DO per command, often 2. Start and stop plus 1 DI for run status
- Analyser transmitter. 1 AI per measured variable, a gas chromatograph may carry 4-8 AI
- Safety instrumented function input. 1 AI or DI on SIS, not BPCS, count separately
A 12-page P&ID set for a mid-size process plant typically produces 300-600 I/O points. A 200-page offshore platform produces 3,000-6,000. Count the first two or three pages carefully, then extrapolate.
Step 2. Classify by signal class and system, BPCS vs. SIS
After extraction, split the list into BPCS and SIS. IEC 61511, ISA 84 requires physical separation, so these are separate hardware budgets.
A typical petrochemical plant runs 10-25 percent of its instruments on SIS. A refinery with extensive emergency shutdown scope runs higher. The SIS fraction is driven by the HAZOP and SIL classification study, which at bid stage is usually preliminary or not yet complete. Use a conservative percentage from comparable projects if the SIL study is not available.
Step 3. Apply spare ratios
Standard spare ratios by signal class and project type.
| Signal class | Greenfield spare | Brownfield spare |
|---|---|---|
| AI | 20-25% | 30-40% |
| AO | 15-20% | 25-30% |
| DI | 20-25% | 30-40% |
| DO | 10-15% | 20-25% |
Brownfield projects carry higher spares because the as-built drawings diverge from the field reality. See Brownfield P&ID digitization for the drawing-quality considerations that affect how much the count can be trusted.
Step 4. Apply IS and SIS allocation
After sparing, flag the instruments that live in hazardous areas, IEC Zone 1, Zone 2 or NEC Class I Div 1, Div 2. IS-protected instruments require IS-rated or IS-barriered input cards. Card density for IS-rated AI is typically half that of standard AI, 4 channels instead of 8.
At bid stage, the area classification drawing may not exist. Use the process description and plant area layout to estimate the IS fraction. Refinery and offshore floating production units typically run 40-60 percent of AI channels in Zone 1, Zone 2. Utility buildings, compressor buildings with natural ventilation, and tank-farm pump stations in classified service run lower.
Step 5. Derive card counts
Divide the spared and IS-allocated count by the channels per card for each class and vendor. Round up. This is the card count. Group cards into chassis, group chassis into cabinets.
The full card-sizing math is in PLC card sizing from a P&ID, including per-vendor channel density tables for Siemens S7-1500, Rockwell ControlLogix, and Schneider Modicon.
Rules of thumb that survive
| Metric | Value | Use |
|---|---|---|
| AI. AO ratio | 4. 1 to 6. 1 | Predicts AO count from AI count |
| DI. DO ratio | 2. 1 to 3. 1 | Predicts DO count from DI count |
| Total digital. Total analog | 1. 1 to 2. 1 | Cross-checks overall mix |
| Points per 19-inch cabinet | 800-1,200 | Cabinet count from total I/O |
| Spare capacity per signal class | 20% greenfield | Added to sized hardware |
If your estimate violates these ratios significantly, double-check. Either the process is unusual, offshore safety systems and pharma lean heavy on digital. Continuous chemical plants lean heavy on analog or the count has a gap.
Dialect-specific counting conventions
Standard ISA 5.1 P&IDs use the conventions above. Non-ISA dialects introduce counting anomalies that inflate or deflate the bid count if you apply the standard rules.
Dutch HVAC and building automation P&IDs. Tag format AB2WV05, a combination of system, area, and function codes does not follow the ISA first-letter instrument-type convention. Control valves are often implicit in the diagram rather than marked with a modulating-actuator symbol. A straight bubble count understates the AO count by 10-20 percent on these drawings.
KKS power-station diagrams. KKS tags, 10LAC01CT001 encode the plant hierarchy into the tag rather than the ISA signal class. Every KKS tag maps to one signal, but the signal class follows from the KKS function code, not from the ISA first letter. The count is accurate once the KKS function-code-to-signal-class mapping is used. On a bid where the drawings are KKS-coded and the mapping table has not been verified, add a 10 percent contingency to the count.
Instruments inside equipment skids, hidden I/O. Packaged equipment, compressors, air coolers, proprietary heat exchangers often shows a single DCS-interface bubble on the P&ID, not the full internal I/O count. The internal I/O lives in the skid vendor's scope and their local panel. At bid, count only the interface signals to the main PLC. Typically 4-8 DI, DO per skid for status and permissives, plus 2-4 AI for key process variables.
Shared HART tags. A Coriolis flowmeter measured in HART may report mass flow, PV, density, SV, and temperature, TV from one instrument tag. The P&ID shows one bubble with one tag. The I/O list carries three AI entries. If you count instrument bubbles without accounting for multi-variable HART devices, the AI count is understated. Ask the instrument supplier list which instruments are multi-variable during the bid phase.
Why bid-stage estimates differ from detail-engineering counts
A bid-stage count from real P&IDs typically lands within 10-15 percent of the final detail-engineering count. The gap exists for three reasons.
Drawing maturity. Bid drawings are preliminary. Instruments added during HAZOP resolution, operability reviews, and vendor tie-in coordination accumulate as the design matures. The typical growth from bid-stage to FEED P&IDs is 5-12 percent. From FEED to issued-for-construction is another 5-10 percent.
Packaged equipment definition. At bid stage, the skid vendor has not issued their signal interface list. The interface count is an estimate based on equipment type and size. The actual count at detail engineering is usually 15-30 percent higher than the bid estimate for complex packages like compression trains and heat-integrated column systems.
Safety study outcomes. HAZOP and SIL classification reclassify instruments from BPCS to SIS scope. Each reclassification removes a point from the BPCS count and adds one to the SIS count. At the system level the total count is unchanged. The cost and hardware architecture change because SIS-rated hardware is more expensive.
A rough-cut estimate based on process description alone, no drawings is usually within 25-40 percent, which is why the first step of any control-system proposal is "get the drawings." If the drawings do not exist, the estimate is a function of process size, equipment count, and engineering judgment, and carries a corresponding contingency.
Converting I/O count to cards and racks
A rough worked example for 500 AI, 100 AO, 400 DI, 200 DO, all BPCS, greenfield, standard non-IS.
| Class | Count | Spare 20% | Sized | Points, card | Cards |
|---|---|---|---|---|---|
| AI | 500 | 600 | 600 | 8 | 75 |
| AO | 100 | 120 | 120 | 8 | 15 |
| DI | 400 | 480 | 480 | 16 | 30 |
| DO | 200 | 240 | 240 | 16 | 15 |
| Total | 1,200 | 1,440 | 1,440 | n, a | 135 cards |
At 13 cards per chassis and approximately 7 chassis per rack row, 135 cards fits in roughly 11 chassis across 2 rack rows. Every PLC vendor has slightly different geometry, but the order of magnitude holds.
The 135-card example is a large plant section. If this is the full plant, expect 2-3 system cabinets plus separate marshalling cabinets. See Marshalling cabinet sizing for the terminal-count math that determines how many marshalling cabinets run alongside the PLC cabinets.
Pitfalls
Double-counting. A pump start, stop from the DCS usually lists both the PLC command and the starter feedback. If you count both sides of every interface, you will over-size hardware by 20-30 percent.
Missing packaged equipment. Packaged skids, compressors, chillers, filter packages often have their own control panel with a DP interface. You carry a handful of interface signals per package, not the full internal I/O. Ask the supplier for a signal list rather than counting bubbles inside the package envelope.
Safety vs. Basic process control. SIS and BPCS I/O are usually on separate hardware. Do not mix them in your PLC hardware count. A 1,000-point plant might have 850 BPCS points and 150 SIS points on a separate logic solver.
Wired vs. Indicated. Not every bubble on the P&ID is wired. Pure local indicators, gauge faces, sight glasses should be excluded from the PLC count.
Motorized valve position switches. Each on-off valve can carry 2 DI, open limit, close limit and 1 DO, actuate. The P&ID may show only the valve symbol without the position-switch symbols, particularly on preliminary drawings. Check the valve data sheets or the on-off valve schedule before locking the DI count.
The estimate-to-budget gap
A bid-stage I/O estimate is usually within 10-15 percent of the final count if drawn from real P&IDs. The estimate carries an explicit contingency. Greenfield 10 percent, brownfield 20 percent, early-FEED or description-only 30 percent. Write the contingency into the bid letter rather than absorbing it into the base count silently, so the owner understands what the number means.
Related
- P&ID to I/O list. Produces the count directly from drawing extraction
- Convert P&ID drawings to a TIA Portal XML tag table. Take the count into Siemens S7-1200, S7-1500
- PLC tag naming conventions. What to do with the count
- Brownfield P&ID digitization. Handling scanned drawings
- Spare I/O Percentage: How Much to Carry, and Where It Goes on the List. The per-type spare calculation that avoids over- and under-provisioning
- PLC card sizing from a P&ID. The card-count math after the I/O count is locked
- PLCCreator export guide. Taking the I/O count through to a PLCCreator Device IO List
- How to count instruments on a P&ID quickly. Manual counting methods for quick bid estimates
FAQ
What is a fast way to estimate I/O count for a bid.
Count the instrument bubbles on the P&IDs. Each transmitter or indicator is typically one AI. Each control valve is one AO. Each switch is one DI. Each on-off valve is one DO. That rough pass, multiplied by an equipment-level factor, gets you within 10 percent of the real count in most projects.
What spare I/O percentage should I carry in the estimate.
20-25 percent spare on each signal class is the usual rule. Below 15 percent risks running out during late design changes. Above 30 percent starts to waste cabinet space and card slots.
How do I convert I/O count to card and cabinet requirements.
Divide by points-per-card for each signal class, typically 8 or 16, round up, add spare cards. Rough rule. Every 200 I/O points needs roughly 1 rack slot of cards plus field termination, so a 1,000-point project typically lands in a single 19-inch cabinet row.
What proportion of project cost is tied to I/O count.
For small to mid-size control system projects, I/O count drives PLC hardware cost, 5-15 percent of total, wiring cost, 10-20 percent, and programming effort, varies. A 30 percent swing in I/O count moves the total project cost materially.
What is the difference between counted I/O and wired I/O.
Counted I/O is what the drawings show. Wired I/O is what actually terminates on the PLC. Some indicators are local-only, gauge faces, sight glasses and carry zero PLC I/O despite appearing on the P&ID. Your count should exclude these.