--- title: Retrofit vs. rip-and-replace — why phased modernization wins for mid-market plants description: 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. author: Akshay Sarode published: 2025-08-07 updated: 2026-04-08 cluster: c2-scada tags: [scada, plc, retrofit, modernization, ignition, rockwell, opc-ua] reading: 9 min hero: The cheapest layer to upgrade is the dashboard. Start there. --- ## Lead Two integrator playbooks — [E Tech Group's "modernizing legacy control systems with minimal disruption"](https://etechgroup.com/modernizing-legacy-control-systems-with-minimal-disruption/) and [Live Automation's retrofit-upgrading-migration page](https://liveautomation.com/automation-solutions/retrofitting-upgrading-migration/) — 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](https://www.machinecdn.com/blog/best-scada-alternatives-2026/) 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](https://etechgroup.com/modernizing-legacy-control-systems-with-minimal-disruption/): 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](https://liveautomation.com/automation-solutions/retrofitting-upgrading-migration/) 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](/use-cases/industrial-monitoring-mid-market) 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](/blog/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. > [!NOTE] > Diagram placeholder: 18-month phased retrofit timeline showing observability first, then SCADA seat drops, then protocol expansion, then PLC refreshes — vs. rip-and-replace big-bang quote in month 1. ## 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: 1. **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. 2. **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. 3. **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](https://industrialcontrolacademy.com/plc-cyber-security-secure-plcs-remote-access-and-networks/) and [CISA's Iranian-actor PLC advisory](https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-097a) 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: 1. What's the lowest-risk thing we could do in month 1 that would tell us 80% of what we need? 2. Can we run the new SCADA in parallel for 90 days before dropping the old one? 3. Which PLCs actually need to be replaced *now* vs. *whenever they fail*? 4. What's the cost of doing only the dashboard retrofit and revisiting in 12 months? 5. 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](/blog/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. ## Further reading - [Industrial monitoring for the mid-market](/use-cases/industrial-monitoring-mid-market) — the detailed playbook - [Sutrace as an Ignition SCADA alternative](/alternatives/ignition-scada) - [Sutrace as a Rockwell FactoryTalk alternative](/alternatives/rockwell-factorytalk) - [Rockwell FactoryTalk 2026 pricing decoded](/blog/rockwell-factorytalk-2026-pricing-decoded) - [No per-tag pricing — the buyer's filter](/blog/no-per-tag-pricing-the-buyers-filter-most-vendors-fail) - [PLCtalk: PLC remote monitoring](https://www.plctalk.net/forums/threads/plc-remote-monitoring.143881/) - [PLCtalk: Cloud SCADA / HMI options](https://www.plctalk.net/forums/threads/cloud-scada-hmi-for-remote-monitoring-control-options.112936/) - [FlowFuse on building HMI for equipment control](https://flowfuse.com/blog/2025/11/building-hmi-for-equipment-control/)