Running a shared instrument facility: booking, cost recovery and the technician who makes or breaks it

An electron microscope under a dust sheet, three years old, with a few hundred hours on the column. It is a familiar sight on Bangladeshi campuses. Booking rules, cost recovery, access tiers and maintenance budgeting that keep an instrument working.

Walk into enough university buildings in Dhaka, Rajshahi and Khulna and you will be shown a room with an expensive instrument in it, covered, powered down, with a logbook that stops eighteen months after commissioning. Nobody did anything obviously wrong. The grant was won, the tender ran, the machine was installed and it worked. What was never designed was the thing that keeps it working: an operating model. That is a management problem with known solutions, and it costs far less to solve than to replace an instrument.

Why the instrument goes quiet

The failure is nearly always one of six things, and usually several at once.

• One person knew how to run it and that person finished a PhD, took a scholarship abroad, or moved to a better paid post.
• No consumables budget. The instrument works but there are no filaments, no gold targets, no probes, no calibration standards, and no recurrent line to buy them from.
• The warranty ended and nobody renewed anything. The first fault after that becomes a capital request rather than a maintenance job, and capital requests take a year.
• No booking system, so access depends on knowing the right person. Users outside that circle stop asking and utilisation collapses to one research group.
• No cost recovery, so every hour of use is a pure cost to whoever holds the room. The rational response is to discourage use, and that is what happens.
• Fear. Nobody wants to be the person who broke the university's most expensive machine, so the safest behaviour is to leave it off.

Access tiers, written down before the first user

The instinct to protect an instrument by restricting it to two people is understandable and it is what kills utilisation. The alternative is not open access, it is graded access with recorded competence.

Tier	Who it is for	What they may do	How they qualify
Sample submission	Undergraduates, external clients, one-off samples	Submit a sample and a request form, receive data and a short report	No training, pays a service rate
Supervised user	Postgraduates early in a project	Operate with the technician present	Safety briefing and a half day induction
Independent user	Trained postgraduates and staff	Book and operate alone in working hours	A recorded number of supervised sessions plus a practical sign-off
Out of hours user	A small number of proven independent users	Evening and weekend sessions	Additional sign-off, a buddy rule, and a named emergency contact
Super user	Technician plus one or two academics	Train others, change configurations, first line diagnosis	Manufacturer training, refreshed periodically

Write the sign-off as a checklist of things the user must demonstrate, not as a certificate of attendance. Keep it in a folder. When something is damaged, and eventually something will be, the record tells you whether the problem is a person or a procedure.

Booking rules that actually change behaviour

• A shared calendar everyone can see. Visibility alone removes most of the political argument about access.
• Fixed session blocks rather than open-ended bookings, with a limit on how many hours one person can hold and how far ahead they can book.
• A no-show policy with consequences. Two unexplained no-shows and independent status drops back to supervised for a month. That sounds harsh until you have watched a machine sit idle all week against a full calendar.
• A cancellation window after which the session is charged anyway, so late cancellation is not free.
• A reserved block every week for the technician covering preventive maintenance, training and external work. If it is not in the calendar it will not happen.
• A mandatory log entry at the end of every session, recording sample type, operating conditions, consumables used and anything unusual. Lock the booking account until the entry exists. Nothing else produces a fault history, and a fault history is what lets an engineer diagnose a recurring problem remotely instead of flying in.

Cost recovery arithmetic that survives an audit

Charging for instrument time is often resisted on the grounds that everybody is on the same budget anyway. That argument loses the moment the first major repair is needed. Build the rate properly and it stops being controversial, because it becomes arithmetic rather than opinion.

Start with the annual cost of ownership. Then divide by the hours you can realistically bill, which is a much smaller number than the hours in a year.

Cost line	Why it gets left out
Service contract, or a provision equivalent to one	Feels optional while the warranty is running
Consumables	Bought from whichever project has money that month, so never totalled
Locally held spares float	Treated as capital rather than as a cost of operating
Electricity, cooling and air conditioning	Paid by the estates budget, invisible to the facility
Technician time, apportioned across instruments	Salary sits in an establishment budget
Calibration, standards and reference samples	Bought once at installation and forgotten
Software licences and analysis packages	Renewals arrive in a different financial year
UPS battery replacement	A predictable scheduled cost treated as a surprise failure
Sinking fund toward replacement or major refurbishment	Nobody plans past the current grant

The denominator is where most facilities deceive themselves. An instrument is not available for two thousand hours a year. Subtract preventive maintenance, breakdowns, training, holidays and the hours nobody wants to book. Set the rate against what you actually achieved last year plus a modest improvement, and revise it annually against the log. A rate built on a fantasy denominator produces an income that cannot pay for a single engineer visit, which then proves to everyone that cost recovery does not work.

Publish a rate card with three bands: internal users, other academic and research institutions, and industry. Publishing it is the point. A published rate is what allows the facility manager to say no to a senior colleague asking for free weekend time without it becoming a personal conflict.

A facility with no sinking fund ends where it started: a dead instrument, a locked room and a fresh grant application eight years later.: Observation from equipment installations across Bangladeshi universities

The technician makes or breaks it

The largest single determinant of whether a research instrument is used is one person, and universities routinely grade that post as junior support staff, pay it accordingly, then act surprised when the holder leaves within two years. A technician who knows the instrument, knows the users and knows when to say no is worth more to a department's research output than another lecturer, and should be treated that way.

• Grade and pay the post so it can be filled by someone with a good masters degree and retained past the point where they become genuinely expert, which is roughly year three.
• Let them co-author. Where the technician did the measurement and the interpretation, authorship costs nothing and retains people better than a small raise.
• Cross-train a second person. One trained operator with a passport is not a staffing plan. The second person can be part time, but they must run the instrument regularly, not in theory.
• Give them the authority to refuse a session on safety, sample or condition grounds, and back them publicly the first time they use it. That single act determines whether the rules mean anything.
• Let the facility keep a share of external income. A technician who can see that industry work funds the spares cupboard behaves very differently from one whose earnings disappear into central accounts.

Maintenance budgeting under Bangladeshi conditions

The service contract question deserves an honest answer rather than a sales one. A contract is a predictable annual sum covering planned visits and usually priority response. Pay as you go looks cheaper and stays cheaper until the first fault needing an engineer to travel and a part imported under a fresh letter of credit. What tips the balance in Bangladesh is not the repair cost but the elapsed time: weeks of customs and banking on top of the part lead time, during which a research group loses a semester. Decide before the warranty lapses, because re-entry after a gap costs more and is sometimes conditional on an inspection first.

• Hold a local spares float of the items that actually fail: filaments and apertures, O-rings and seals, pump oil, probes, detector windows, filters and fuses.
• Put preventive maintenance in the calendar with a named owner, and treat a missed visit as an incident. Preventive visits are also when an engineer spots the fault you have not yet noticed.
• Treat air conditioning as instrument maintenance, not building comfort: filter cleaning, gas top-up, a service schedule, and a second unit where a single failure would stop the instrument.
• Schedule UPS battery replacement rather than waiting for failure. Batteries in a warm room age faster than the datasheet suggests, and a UPS with dead batteries is worse than none because everyone believes they are protected.
• Keep a written fault history. Date, symptom, what was tried, what fixed it. The third occurrence of a fault should take an hour to diagnose, not a week.

Governance that survives a change of department head

1. Write a facility charter. What the facility is for, which instruments it holds, who employs the staff, who sets rates, who arbitrates disputes and how the charter itself is amended.
2. Form a user committee drawn from more than one department, meeting a few times a year, with the log and the accounts in front of it.
3. Publish the rate card and revise it annually against real costs and real hours.
4. Report usage and income to the university every year, in writing, whether or not anyone asks.
5. Sign formal arrangements with other institutions so that external users are routine rather than a personal favour. This is also where the strongest argument for the next grant comes from.
6. State plainly that the instrument belongs to the facility, not to the principal investigator who won the grant that bought it.

That last point is the hardest in practice and the most consequential. An instrument treated as a research group's private property will be used by that group alone, will not generate service income, will not build the case for the next purchase, and will fall silent when the group's interests move on. Better to cross that line at the planning stage than five years in.

The numbers worth publishing every year

• Operating hours, split between internal, external academic and industry use.
• Unique users, and how many came from outside the host department.
• Income received against the cost of ownership, with the sinking fund shown separately.
• Downtime hours and mean time to repair, which is the number that justifies a service contract more effectively than any argument.
• Publications, theses and industry reports that acknowledge the facility.
• Training sessions delivered, and how many users moved up a tier.

A facility that can produce those figures on one page gets its next instrument funded. A facility that cannot will be asked, quite reasonably, why the university should buy another one.

Back to all Insights