Address chaos in a factory network does not arise from a single decision; each machine project brings its own habits, and they accumulate over the years. The result is predictable: conflicting addresses, devices whose owner nobody knows, and a maintenance team trying to rebuild the topology from memory at the moment of a fault. Below, we propose a standard that can be used across the plant, with a subnet plan per line and cell, address assignment rules, PROFINET device name constraints and a practical naming template.
The cost of unplanned addressing in the field
In a factory without a standard, the picture is always similar. Every machine builder arrives with its own default network (often 192.168.0.0/24 or 192.168.1.0/24), the same addresses conflict when lines are connected together, an address given manually to a service computer conflicts with another device’s, and intermittent communication dropouts begin. Intermittent dropouts are the most expensive fault. And if the address table lives in a file on someone’s laptop, every change made while that person is away goes undocumented.
The second cost is paid in security. When subnets are allocated randomly rather than by function, segmentation and firewall rules are written with individual IPs instead of address ranges; the rule list grows and becomes impossible to audit. A consistent address plan also makes it easier to apply architectural decisions such as IT/OT network separation with the Purdue model on site.
Subnet planning per line and cell
The basis of the plan is that the address structure reflects the physical structure of the plant. From the private address ranges (RFC 1918), a dedicated block that does not overlap with IT is chosen for OT. For example, if IT is within 10.0.0.0/12, a separate block such as 10.100.0.0/14 can be reserved for OT. The scheme below is an example:
| Octet | Meaning | Example |
|---|---|---|
| 2nd octet (100–103) | Plant (site) | 10.100 = Plant 1, 10.101 = Plant 2 |
| 3rd octet | Area and line | 10.100.23 = Press hall, line 3 |
| 4th octet | Device role range | .10–.19 PLC |
Allocating a /24 to each cell makes calculation easy; 254 addresses within a cell are more than enough in most cases. Because PROFINET RT and DCP run at Layer 2, an IO controller and the devices connected to it must be in the same VLAN and subnet. The “cell = VLAN = subnet” mapping is therefore the natural design unit. Whoever looks at the address knows the location.
Role-based ranges within a cell make it possible to infer the device type from the address:
| Range | Role |
|---|---|
| .1 | Gateway (L3 switch or firewall interface) |
| .2–.9 | Switches and network devices |
| .10–.19 | PLCs and controllers |
| .20–.39 | HMIs, panel PCs and industrial PCs |
| .40–.199 | Field devices (remote I/O, drives, robots, sensors) |
| .200–.229 | Service computers and temporary devices |
| .230–.254 | Reserved |
If the machine builder’s internal network cannot be changed, the machine network is left in its own segment and connected to the factory network via NAT. The details of this scenario are in the article connecting machine builder networks to the factory with 1:1 NAT.
Address assignment methods: static, DHCP and PROFINET DCP
| Method | Suitable devices | Point to watch |
|---|---|---|
| Static (on the device) | PLCs, switches, firewalls, critical HMIs | Must be entered in IPAM |
| DHCP reservation | Industrial PCs, printers, cameras | If the server is unreachable, no address is obtained at startup |
| DHCP dynamic pool | Service computers, temporary devices | Only within the reserved range |
| BOOTP/DHCP (initial setup) | EtherNet/IP devices that default to BOOTP/DHCP | Must be switched to static after the address is assigned |
| PROFINET DCP | PROFINET IO devices | The address is assigned by the controller according to the device name |
In PROFINET, a device’s identity is not its IP address but its device name (NameOfStation). During commissioning, the engineering tool writes the name to the device via DCP; during operation, the IO controller finds the device with this name on the network via DCP and assigns the IP address defined in the project. For PROFINET devices, the “IP plan” and the “name plan” are therefore managed together. If the topology is defined in the project, many IO controllers can automatically give the name, based on neighbour information, to a new factory-default (unnamed) device fitted in place of a failed one (device replacement without exchangeable medium). If this feature is to be used, port connections must exactly match the project; a technician who plugs the cable into the neighbouring port leaves the device unnamed.
PROFINET device name rules
A PROFINET device name must follow rules defined in IEC 61158 that resemble DNS names. What you need to know in practice:
- Total length at most 240 characters; each dot-separated label at most 63 characters
- Only lowercase letters (a–z), digits (0–9), hyphens, and dots as label separators
- A label cannot start or end with a hyphen
- The first label cannot be in the form
port-xyzorport-xyz-abcde(x, y, z, a–e digits) - The name cannot look like an IP address in the form
n.n.n.n
Some engineering tools accept invalid characters such as uppercase letters or underscores and produce a compliant “converted” name in the background. The name read on site and the name shown in the project then differ; at the moment of a fault, this difference can cost hours. The cleanest way is to generate compliant names from the start. The Python function below checks a list of names in bulk:
import re
LABEL = r"[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?"
PN_NAME = re.compile(rf"^{LABEL}(?:\.{LABEL})*$")
PORT_FORM = re.compile(r"^port-\d{3}(?:-\d{5})?(?:\.|$)")
IP_FORM = re.compile(r"^\d{1,3}(?:\.\d{1,3}){3}$") # n.n.n.n form not allowed
def pn_name_is_valid(name: str) -> bool:
if len(name) > 240 or not PN_NAME.match(name):
return False
if PORT_FORM.match(name) or IP_FORM.match(name):
return False
return True
for name in ["ist1-pres-h03-plc01", "PLC_1", "port-001", "10.12.3.4", "-hmi01"]:
print(name, pn_name_is_valid(name))
Device naming template
A good template is short and has a fixed order; reading it tells you where the device is. The structure we recommend:
<plant>-<area>-<line>-<role><seq>
ist1-pres-h03-plc01 Plant 1, press hall, line 3, first PLC
ist1-pres-h03-sw02 Second switch of the same line
ist1-pres-h03-hmi01 Line 3 operator panel
ist1-boya-h01-rio04 Paint shop line 1, fourth remote I/O
Template rules:
- Role codes are chosen from a fixed list:
plc,sw,hmi,ipc,rio(remote I/O),drv(drive),rob(robot),fw(firewall),ap(wireless access point). - The sequence number has two digits; the number of a removed device is not reused.
- No Turkish characters are used; the transliteration ç→c, ğ→g, ı→i, ö→o, ş→s, ü→u is applied.
- The same name is used identically in the PROFINET device name, the switch hostname, the DNS record, IPAM and the physical label on the cabinet.
The last rule is the most useful. When the cabinet label and the name in the network monitoring system are the same, the field technician and the network team are talking about the same device on the phone. Otherwise, you first have to find the device.
Keeping IPAM and documentation up to date
The value of an address plan is measured by how up to date it is. In small plants, a well-maintained spreadsheet may be enough; in multi-line plants, open-source IPAM tools such as NetBox or phpIPAM keep subnet, VLAN, device, cabinet and cable information in one place. Whatever the tool, write down these process rules:
- Addresses and names are taken only from IPAM; an address that is not in IPAM is not used on site.
- Every machine commissioning and device replacement includes an IPAM update in the closing step of the work order.
- At regular intervals, IPAM is compared with switch ARP/MAC tables and LLDP neighbours, and the list of differences is reviewed.
- Consistency between engineering projects (PLC hardware configuration) and IPAM is checked when the project is archived.
If the rules are not written down, IPAM drifts away from the field within a few months. Typical Cisco IOS commands that can be used for the comparison:
show ip arp | include 10.100.23.
show mac address-table vlan 123
show lldp neighbors detail
For the general method of documenting topology and port maps, see the network documentation, topology and IPAM guide.
New machine commissioning checklist
- A clause requiring compliance with the plant’s addressing and naming standard has been added to the machine contract
- All of the machine’s network interfaces, default addresses and protocols used have been obtained from the builder
- A VLAN and subnet for the cell have been allocated from IPAM; addresses have been assigned according to the role ranges
- PROFINET device names have been generated according to the template and naming rules and run through the check script
- If the machine’s internal network cannot be changed, a NAT solution has been designed
- Switch hostnames, DNS records and cabinet labels have been updated with the same names
- At the end of commissioning, ARP/MAC and LLDP output was compared with IPAM
- It has been verified that service computers obtain addresses only from the reserved range
The first item is the one most often skipped: if the standard does not go into the contract, the machine will still arrive with its own default network. For topology and switch mistakes in PROFINET cell design, the article common mistakes in PROFINET network design is a useful complement. For mapping the current addressing situation and preparing the standard, see our factory OT infrastructure survey service page, and for cell and backbone design, our industrial Ethernet and factory networks page. The reference for the definition of private address ranges is RFC 1918.
Frequently asked questions
Can PROFINET devices be on different subnets?
Because cyclic RT communication and DCP run at Layer 2, an IO controller and the IO devices connected to it must be in the same broadcast domain (the same VLAN and subnet). Data exchange between cells is done via routing or a firewall, with non-cyclic methods or a separate controller-to-controller connection.
Is using DHCP on an OT network risky?
Yes, for devices whose address must not change, such as controllers, switches and field devices. If the DHCP server cannot be reached, the device may fail to get an address at startup. Use static addresses or reservations on these devices; reserve the dynamic pool for service computers and temporary devices only.
What should be done if a machine builder delivers with 192.168.0.x?
First, discuss with the builder whether the machine’s internal network address can be changed. If it cannot, the machine network is left in its own segment and connected to the factory network through a router or firewall capable of 1:1 NAT; the machine appears to the outside with an address reserved from the factory plan.
Can Turkish characters be used in the naming template?
They should not be. PROFINET device names, DNS and many management tools only reliably handle a limited character set such as lowercase letters, digits, hyphens and dots. Transliterate Turkish characters as ç→c, ğ→g, ı→i, ö→o, ş→s, ü→u, and state this rule explicitly in the template document.