Sizing SIMATIC Process Historian Hardware Using the PH-HWAdvisor Tool

When sizing SIMATIC Process Historian (PH) server hardware using the Siemens PH-HWAdvisor planning tool during early engineering development, the required inputs—Archive Tag Values per Second and Alarm Messages per Minute—can be calculated using standardized process automation estimation models. For tag values per second, divide the total number of cyclical archived tags by their logging cycle times, and add 15% of total change-driven analog and digital tags. For alarm rates, apply the EEMUA 191 baseline of 1 to 2 alarms per minute per 1,000 I/O points for steady-state operation, factoring a 100 to 500 alarm-per-minute flood burst during process trips to determine CPU, RAM, and RAID storage sizing.

The Role of PH-HWAdvisor in Process Historian Engineering

SIMATIC Process Historian (PH) is the centralized, real-time SQL archiving database for SIMATIC PCS 7 and WinCC. Sizing PH server hardware (such as SIMATIC IPC647D, IPC847D, or high-performance commercial enterprise servers) requires running the official PH-HWAdvisor sizing algorithm provided by Siemens.

The PH-HWAdvisor tool calculates the necessary processor cores, RAM capacity, SQL database transaction throughput, and RAID storage array dimensions (e.g., RAID 1 for OS, RAID 5 or RAID 10 for database and log files) over a planned retention lifespan (e.g., 3, 5, or 10 years). However, early in project design when software development is incomplete, engineers must utilize standardized process automation estimation heuristics to derive the two essential sizing inputs.

Calculating Archive Tag Value Rates (Values per Second)

The metric Archive Tag Values per Second represents the sustained write frequency into Microsoft SQL Server. It comprises two tag acquisition mechanisms:

Cyclic Archived Tags ($R_{\text{cyclic}}$): Tags logged at fixed time intervals (e.g., 1s, 2s, 5s, 10s, 60s). Calculate by summing the frequency of each tag group: $$R_{\text{cyclic}} = \sum \frac{N_i}{T_i}$$
(Example: 500 fast flow tags at 1s = 500 val/s; 2,000 temperature tags at 10s = 200 val/s; Total cyclic = 700 val/s).
Acyclic / Change-Driven Tags ($R_{\text{acyclic}}$): Tags logged only when their value exceeds a deadband or binary state changes. In typical continuous and batch chemical processes, between 10% and 20% of acyclic tags fluctuate in any given second during normal operation.
(Example: For 5,000 change-driven tags, estimated rate = $5000 \times 0.15 = 750$ val/s).
Total Ingestion Rate: Sum cyclic and acyclic rates ($700 + 750 = 1,450$ val/s), and apply a 25% safety margin for plant expansion.

Calculating Alarm Message Rates (Alarms per Minute)

The metric Alarm Messages per Minute sizes the alarm journal database and message queuing bandwidth:

Steady-State Baseline (EEMUA 191 / ISA-18.2): Industry alarm management standards dictate that a well-designed alarm system generates an average of less than 1 alarm per 10 minutes per operator (or approximately 1 to 2 alarms per minute per 1,000 physical I/O points).
Upset / Alarm Flood Peak: During major plant disturbances (such as a power failure or compressor trip), alarm rates surge dramatically. Process Historian must buffer and write up to 100 to 500 alarms per minute without dropping packets or causing Microsoft Message Queuing (MSMQ) overflows.
Sizing Recommendation: Input a sustained rate of 10 to 30 alarms/min for moderate plants, or 60 to 120 alarms/min for large facilities into PH-HWAdvisor.

Process Historian Sizing and Hardware Specification Matrix

The following table provides verified hardware sizing profiles derived from PH-HWAdvisor guidelines:

Plant Scale ProfileProcess Values / SecAlarms / MinRecommended Hardware PlatformRAM & Storage ArchitectureSmall (Basic)Up to 1,500 val/sUp to 20 alarms/minSIMATIC IPC647D / IPC647E16 GB RAM; 2x 1TB HDD (RAID 1)Medium (Standard)Up to 5,000 val/sUp to 60 alarms/minSIMATIC IPC847D / IPC847E32 GB RAM; RAID 1 (OS) + RAID 5 (Data)Large (High Performance)Up to 15,000 val/sUp to 120 alarms/minSIMATIC IPC847E / High-End Server64 GB RAM; SAS SSDs (RAID 10)Enterprise / Multi-OS> 20,000 val/s> 200 alarms/minEnterprise Server (e.g., HP/Dell)128 GB RAM; Dedicated SAN / High-Speed NVMe

[!NOTE]
Commercial Accuracy Guardrail: Hardware specifications generated by PH-HWAdvisor serve as technical engineering planning aids. Final server MLFBs (e.g., 6ES7660-... or 6ES7661-...), storage capacities, and Microsoft SQL Server core licenses must be validated against Siemens Catalog ST PCS 7 and quoted through an authorized distributor.

Verification

Verify sizing during commissioning:
1. On the commissioned Process Historian, open the Process Historian Management Console.
2. Navigate to the Diagnostics -> Performance dashboard.
3. Verify that the actual measured Tags/s and Alarms/min remain comfortably below the calculated hardware limits.
4. Check Windows Performance Monitor (perfmon.msc) to confirm that SQL Server average disk queue length remains < 2 and CPU utilization averages < 40% under normal operating conditions.

Limitations and Operational Precautions

Fast-Logging Traps: Configuring hundreds of analog tags with 100 ms or 500 ms cyclic logging severely skews sizing, requiring expensive enterprise SSD storage arrays. Use deadband archiving for high-frequency measurements.
Archive Segmentation: Sizing must account for daily or weekly database segment splitting. If segments become too large (> 50 GB per segment), automated database detaching and backup routines will cause performance bottlenecks.
Separate OS and SQL Disks: Operating systems and SQL database files must never share the same physical drive; dedicated RAID arrays are mandatory for database stability.

Engineer FAQ

Can Process Historian run on a virtual machine (VMware ESXi)?

Yes, Process Historian is officially certified for VMware vSphere and Microsoft Hyper-V, provided dedicated vCPUs, non-overcommitted RAM, and high-performance SAN/vSAN datastores are allocated.

How much hard disk space does one archived tag value consume?

On average, each archived process value sample consumes approximately 15 to 25 bytes of storage inside the Microsoft SQL Server database tables.

Does PH-HWAdvisor calculate backup storage requirements?

Yes, the tool calculates both the active online database volume and the secondary backup storage required for long-term detached archive storage (NAS, LTO tape, or cloud).

What happens if actual tag write rates exceed the hardware sizing?

The MSMQ message queue on the OS Servers will fill up. If queues reach their memory limit, real-time archive data is dropped and permanent gaps will appear in historical trends.