scada

Retrofit vs. rip-and-replace — why phased modernization wins for mid-market plants

Why a 5–100-person plant should retrofit observability and dashboards first, and replace PLCs only when they fail. With the real $230k–$690k SCADA TCO range and the integrator playbooks that back it up.

By Akshay Sarode· August 7, 2025· 9 min readscadaplcretrofitmodernization

Lead

Two integrator playbooks — E Tech Group's "modernizing legacy control systems with minimal disruption" and Live Automation's retrofit-upgrading-migration page — agree on something most SCADA sales decks don't say out loud: for plants under 100 people, rip-and-replace is the wrong default. The right default is phased retrofit, starting with the cheapest layer to swap.

MachineCDN's 2026 SCADA alternatives roundup pegs a fresh traditional SCADA deployment at $230,000–$690,000 in year one for a mid-market plant. That's the rip-and-replace quote. This post is about what to do instead.

The TCO that scares people away from modernizing at all

The MachineCDN range deserves to be read carefully:

$230,000 – $690,000 all-in for a fresh SCADA deployment, including software licenses, integrator hours, panel upgrades, server hardware, and first-year support.

Low end: Ignition Unlimited + a small regional integrator + reuse of existing servers. High end: Wonderware/AVEVA or FactoryTalk View SE with a Tier-1 integrator, fresh server cluster, and a full HMI panel refresh.

Both ends share two assumptions: (1) you're replacing the SCADA in one project, and (2) the integrator scopes the project around their billable hours. Both assumptions push the number up. Both are also avoidable.

Why retrofit beats rip-and-replace

1. The PLC is rarely the bottleneck

A 2010 ControlLogix or S7-300 is doing its job fine. The CPU has spare cycles, the I/O is wired and known, and the PLC has been running 24/7 for a decade. Nothing about adding a dashboard requires touching the PLC.

E Tech Group says it directly: the most expensive thing a small plant can do is rip out a working PLC to "standardize the stack" before there's a real reason to. The PLC migration is the costliest sub-project in any rip-and-replace, and most of the time it isn't justified by anything other than vendor preference.

2. The dashboard layer is the cheapest layer to swap

PLCs are wired. HMI panels are bolted. SCADA servers are licensed. The dashboard layer — the thing that reads PLC tags and displays them — is just software talking over a network. Swapping it is a config change, not a wiring change.

The corollary: start your modernization at the dashboard. It's the layer where you can prove value in a week, not a year. If the new dashboard works, you've started the modernization. If it doesn't, you've lost a week and a Pi.

3. Read-only is risk-free

A read-only OPC UA client (or Modbus client, or EtherNet/IP CIP client) cannot break the PLC. It opens a session, asks for tag values, and writes nothing. The safety system is untouched. The control logic is untouched. The line keeps running.

This is why the Live Automation retrofit playbook treats read-only observability as the first retrofit step in every project. It's the lowest-risk, highest-information move available.

The phased retrofit timeline

A typical 18-month phased retrofit, as we'd run it with a mid-market customer:

Months 1–2 — Observability layer

Drop the Sutrace edge agent on a Pi or small x86 box.
Read-only OPC UA / Modbus / EtherNet/IP from existing PLCs.
First 50 tags, alarms in Slack.
See our use-case page for the detailed week-by-week.

Cost: Sutrace subscription + a few hours of the IT/OT lead's time. Outcome: plant has cross-stack visibility for the first time.

Months 3–6 — Drop redundant SCADA seats

Identify which FactoryTalk View SE, Wonderware Client, or Ignition Vision seats are not driving an operator-touched HMI.
Park them on Sutrace dashboards.
Drop the seats at the next renewal.

Cost: zero new spend. Outcome: $14k–$50k/year recovered. See Rockwell FactoryTalk 2026 pricing decoded for the per-seat math.

Months 6–12 — Add the missing protocols

BACnet from the BMS for HVAC visibility.
J1939 from the diesel genset.
SunSpec Modbus from the inverter array.
SNMP from the network gear.
OTel from the MES; Prometheus from the cloud.

Each new protocol is a config change to the same edge agent. No new license. No new vendor.

Cost: zero incremental software. Outcome: the plant manager sees the whole plant.

Months 12–18 — Targeted PLC refreshes

Identify PLCs that are EOL or close to it. (Manufacturer support dates, not "vendor wants you to upgrade.")
Replace them one at a time, on planned downtime.
The new PLC speaks OPC UA natively, which Sutrace already speaks. Zero dashboard rework.

Cost: the actual PLC replacement, when it was going to be replaced anyway. Outcome: modernization completes without a single line stop that wasn't already scheduled.

The financial argument

Rough numbers for a 4-engineer / 3-line / 10,000-tag mid-market plant:

Approach	Year 1	Year 2	Year 3
Rip-and-replace (MachineCDN low end)	~$230k	~$60k	~$60k
Rip-and-replace (MachineCDN high end)	~$690k	~$120k	~$120k
Phased retrofit (Sutrace + selective SCADA drops)	< $30k	$20–25k	$20–25k

The phased retrofit also has a property the rip-and-replace doesn't: you can stop after month 2 and you've still gained 80% of the value. If the budget pulls back, the project gracefully degrades. Rip-and-replace projects that get paused mid-flight are the worst-case outcome — you're paying for the new system and still running the old one.

Where rip-and-replace is actually right

Honest section. Three cases where rip-and-replace genuinely is the right call:

The existing SCADA runs on a Windows Server 2008 box that won't pass the next security audit. Sometimes the platform is so out-of-date that retrofitting on top of it is more expensive than starting fresh.
A plant doubling production with a major capex round. If you're adding 5 lines and a new building, fresh SCADA is part of the building project. Don't retrofit a green-field expansion.
A regulated environment between validation cycles. GxP plants do major changes only at validation boundaries. If the timing aligns, do the work in one cycle.

If none of those describe you, retrofit.

The cybersecurity angle

The Industrial Control Academy's PLC cybersecurity guide and CISA's Iranian-actor PLC advisory are the 2026 backdrop. Both make the same point: legacy PLCs exposed to the internet are a critical risk, but the answer is not to rush a rip-and-replace. The answer is to put the legacy PLC behind an outbound-only edge agent on its own VLAN, expose nothing inbound, and modernize the perimeter and the dashboards.

That's exactly the phased-retrofit posture. It happens to also be the posture that maximizes leverage from an outbound-only observability tool like Sutrace.

What to ask your integrator

If your integrator is quoting a $400k rip-and-replace, ask them:

What's the lowest-risk thing we could do in month 1 that would tell us 80% of what we need?
Can we run the new SCADA in parallel for 90 days before dropping the old one?
Which PLCs actually need to be replaced now vs. whenever they fail?
What's the cost of doing only the dashboard retrofit and revisiting in 12 months?
Why isn't read-only observability the first deliverable?

A good integrator will have answers. A bad integrator will tell you "that's not how we scope projects." Listen carefully.

What success looks like at month 6

A useful exercise: imagine you're running this playbook and it's working. What does month 6 look like for your plant?

The plant manager has a single dashboard that shows every line, every utility, every cloud-side API the operation depends on. They no longer ping the maintenance guy on Slack at 02:00 to ask "is line 3 running?" — they look.
The on-call rotation gets Slack alerts when a chiller flat-lines, when a motor current trends out of band, when the order intake API has been failing for 4 minutes. Not every minor blip, just the things that matter.
Two FactoryTalk View SE seats came off the renewal at the last billing cycle, freeing $27,920/year. That money's been earmarked for the PLC refresh in month 14.
The IT/OT lead has merged the OT alarm pipeline with the existing software-side observability tooling. There's one on-call rotation now instead of two.
The next Rockwell renewal review is on the calendar, with a defensible "what to keep, what to drop" memo in hand. See Rockwell FactoryTalk 2026 pricing decoded for the per-tier walkthrough that informs the memo.
Most importantly: the line is still running. No downtime was caused by the modernization. That's the point of phased retrofit.

If month 6 doesn't look like that, something's wrong with the rollout, not with the approach.

The deeper point

Modernization is not a single decision. It's a sequence of small decisions, each of which should be reversible if it doesn't work. Rip-and-replace is the opposite — one big decision, irreversible after the kickoff invoice. For a mid-market plant, the small-decisions model is a better fit.

The dashboard is the cheapest layer to retrofit, and it's also the layer where the customer feels the most pain right now. That's where to start.