Integrating multi-brand ATGs into one TMS: protocols, gateways and testing

ATGs from different brands usually speak different protocols, so integrating them into one TMS is done through an interface unit or gateway that translates level, temperature and status into one format, and each tank is then tested against a manual dip.

Many Indonesian terminals carry gauges from several suppliers, installed in different decades. The basics of automatic tank gauging (ATG) are the same for all of them, but the way each gauge reports its reading is not. This guide covers the protocols, the gateway architecture, how to map the data into one model and how to test the result before the tank management system relies on it.

A gauge is a small computer on a tank roof, and each manufacturer chose its own way for that computer to talk.

Older installations often use a proprietary field bus that only the maker's interface unit understands. Newer gauges commonly offer Modbus RTU over RS-485 or Modbus TCP over Ethernet, and some add HART or a vendor protocol alongside. Even within one brand, the model, firmware and installed interface card decide what is available. The ATG integration checklist makes the same point: experience with one device does not mean every model or configuration behaves the same way.

Modbus helps, but it is not a common language for tank data. The Modbus Organization publishes the application protocol and the serial line specification, which define how registers are read, not what each register means. One gauge may report level in millimetres as an integer, another in metres as a floating-point value, with a different word order. The integration work lies in that mapping, and a mistake there produces values that look reasonable but are wrong by a fixed factor.

The usual answer is a layered path: each gauge family talks to its own interface, and one gateway turns everything into a single stream for the TMS.

Where a brand uses a proprietary bus, its interface unit sits in the field or in the control room and exposes the readings, often over Modbus. Gauges that already speak Modbus can connect to the gateway directly.

The gateway polls every source, converts units and data types, adds a time stamp and a quality flag, and serves one consistent data set to the TMS server over Ethernet. API MPMS Chapter 3.1B treats data collection, transmission and receiving as part of the gauging system, so this path belongs in the measurement design, not only in the IT design.

Integration is accepted tank by tank, not system by system. A dashboard that shows plausible numbers for every tank can still have two tanks swapped.

The table below lists the checks that catch most mistakes, from swapped tags to alarms that never reach the screen. Acceptance limits for value comparisons should be agreed with operations and the metrology requirements before the test, so nobody argues about the result afterwards.

Once integrated, the weakest part of the system is rarely the gauge. It is the chain of power supplies, cables, converters and gateways that carry the reading.

Tank farms are exposed to lightning, and long field cables pick up surges and carry them to the devices at each end.

At the Jambi fuel terminal in 2026, Citia replaced the ATG gateway, an Advantech ECU-1051, and added surge protectors on both the RS-485 Modbus line and the Ethernet line; the project notes say such protection should be installed from the start. In the ATG power supply repair at Redeco in 2025, a surge left an Endress+Hauser ATG dead with a blown fuse and a damaged varistor. Intermittent drops on the serial side are covered in troubleshooting Modbus on RS-485.

Spares are the other half. Interface modules go out of production long before tanks do. A long-term service agreement (LTSA) for the gauging system should list the critical spares, who holds them and the migration path when a part can no longer be bought. The ATG high alarm also depends on this chain, which is one more reason to keep independent high-high alarms on a separate path.

The gateway sits on the border between the field network and the office network, which makes it a security device as much as a translator.

Keep the gauge network separate from the office network, and let the gateway or the TMS server be the only bridge. Read-only polling is the normal mode; write access to gauge configuration should be limited to engineering laptops under a work permit. Change default passwords on gateways and interface units, record who can log in, and keep a backup of every configuration file with its date.

Citia's bulletin on tank monitoring data security makes the same point from the operations side: integration with SCADA and ERP has to come with access protection, or the data loses its integrity. A tank figure that anyone can overwrite is no better than a handwritten one.

Citia's own records show where multi-brand integration succeeds and where it stalls.

Citia's technical bulletin on Tokyo Keiso and Endress+Hauser integration describes bringing gauges from both makers into one central system through compatible protocols, with the central system configured to receive and process the data in real time. INTEGRA TMS is described as working with ATGs from Honeywell, Endress+Hauser and Tokyo Keiso.

The Redeco ATG assessment in 2024 shows the harder case. Analog and digital gauges sat side by side, IFX-20000 interface modules had blown fuses and voltage drops, and those modules were discontinued. Communication was restored and readings were compared with stick measurements, but the evaluation concluded that accuracy could not be guaranteed on equipment past its life, and migration was recommended. Integration cannot fix a sensor that no longer measures well; an assessment should say so early.

The practical sequence that follows from these cases is simple. Assess the installed gauges first, decide which to keep, repair or replace, and only then design the gateway layer. Integrating a gauge that will be replaced next year doubles the testing work.