Finding the hidden scope before approval

Every concept estimate has scope missing from it. That is not a criticism of whoever prepared it. A concept describes what the owner wants: four rooms, a new plant, a grader. It does not describe what everything around the project will require in return: the electricity network, the ground under the slab, the building surveyor, the fire engineer, the old plant that the new one must work with. Those requirements exist whether or not anyone has asked about them, and they will be paid for either way. The only choice is whether they are found while they are still lines in an estimate, or while they are holes in a site.

This chapter follows the five items that Northbank’s front end found, from the investigation that found each one to what it would have cost if found during delivery. It works through the electrical supply and the refrigeration plant in detail, sets out how to judge whether an investigation is worth commissioning before its result is known, and ends with a ground investigation brief that left out the most heavily loaded part of the site.

Why a concept misses scope

Hidden scope sits at interfaces: the places where the project meets something it does not control. Northbank’s upgrade connects to the distributor’s network, rests on ground nobody had tested, must satisfy a building surveyor and a council, shares a refrigeration system with six existing rooms, handles food, and drains to a council system. Each of those parties has requirements that do not appear on a site sketch, and each tends to reveal them only when asked a specific question with a specific design in front of it.

A front end finds hidden scope by asking those questions deliberately. The definition index in Chapter 3 is one way of making sure they are asked: elements B1, B2, B3, B5 and B7 are, in effect, a list of interfaces, and their weights reflect how much each could cost if left unanswered.

What Northbank’s front end found

This is an illustration. Five items that were not in the concept estimate were found during FEL 2 and FEL 3. For each, the table gives the investigation that found it, what that investigation cost, the item’s cost when designed in before approval, and the project engineer’s estimate of its cost and delay if it had first been discovered during delivery. All money is in Australian dollars excluding GST.

Figure 1Five items the concept missed

	Found by	Investigation	Cost found early	Cost found in delivery	Delay if found in delivery
Electrical supply upgrade	Capacity study, FEL 2	A$9k	A$380k	A$520k	14 weeks
Refrigeration plant for old and new rooms	Duty study, FEL 2 and 3	A$12k	A$270k	A$390k	3 weeks
Ground improvement, plant room pad	Geotechnical investigation, FEL 3	A$14k	A$120k	A$210k	4 weeks
Fire services upgrade	Fire engineering basis, FEL 3	A$11k	A$190k	A$260k	5 weeks
Dock levellers and seals	Design development, FEL 3	none separate	A$30k	A$45k	none
Total	| A$46k	A$990k	A$1,425k |

The A$990k is not a saving. Northbank pays it on either path: the supply has to be upgraded and the fire services installed whether they are found in January or in July. What the front end saves is the difference between finding them early and finding them late, A$435k in cost, which is 9.5 times the A$46k spent on the investigations, plus weeks of delay that, on a project with a harvest deadline, can be worth more than the cost.

Figure 2What the investigations bought

A$46kSpent on investigations18.5% of the A$248k front end

A$990kScope found before approval97% of the A$1,018k growth from Class 5 to Class 3

A$1,425kThe same scope found in deliverybefore counting delay

A$435kLate premium avoided9.5 times the cost of the investigations

The electrical supply

The concept allowed A$180k for electrical work: new circuits from the existing main switchboard to the rooms, the plant and the grader. It assumed the existing supply could carry the extra load, because nobody had any reason to think otherwise.

The A$9k capacity study in FEL 2 tested that assumption. An electrical engineer compared the site’s recorded maximum demand with the load of a new plant serving ten rooms, the grader and the extra lighting and dock equipment, and found that the total would exceed the capacity of the existing supply. The upgrade needed a larger connection from the distributor’s network, a kiosk substation on site and a new main switchboard: A$380k on top of the A$180k already allowed, so the electrical line rose to A$560k.

The cost was only half of the finding. Connecting a larger load to the distribution network runs on the distributor’s timetable, not the project’s. The distributor needed about 8 weeks to make a connection offer after an application, and about 30 weeks after the offer was accepted to complete its works. Because the study was done in FEL 2, the application went in at week 12, the offer arrived at week 20, it was accepted at approval in week 22, and the connection works were due at week 52, the week the plan needs power for commissioning.

Found during delivery instead, the same upgrade would have cost about A$520k: the switchboard ordered to the old design would have been replaced, the design redone under a contract already let, the variation priced without competition, and temporary generation hired to commission the plant. The project engineer estimated 14 weeks of delay, because the distributor’s 38 weeks of process would have started only when the problem surfaced. On the rush path, planned to be ready at week 54, that alone would push readiness to week 68, past the week-65 deadline.

Connection to the distribution network, the electrical installation and its certification are regulated, and are matters for the distributor and for licensed electrical contractors and engineers. The front end’s job was not to do their work but to ask them early enough.

The refrigeration plant

The concept assumed a new plant for the four new rooms, with the six old rooms staying on the existing plant. The options study in FEL 2 looked at the old plant and found it near the end of its life, with a refrigerant and components the business would soon struggle to service. Replacing it within a few years would mean a second project, a second shutdown and a second set of approvals. The better answer was one new plant sized for all ten rooms. FEL 2 added A$170k for the larger plant, and the A$12k duty study and specification in FEL 3 added a further A$100k once the loads were calculated properly: A$270k in all.

A refrigeration duty study adds up the heat each room must lose: through its panels and floor, with each load of warm fruit, through the door and from the fans, lights and people inside, at the worst conditions of the season. Cold room loads in Heat Pumps and Refrigeration by the Numbers works through that calculation. Found in delivery, after the plant was ordered for the new rooms only, the change would have cost about A$390k and 3 weeks: a revised order, a larger plant room and a change to the electrical load, which itself feeds back into the supply question above.

Ground, fire and docks

The other three items were found in FEL 3, once there was a layout to investigate.

• Ground improvement under the plant room pad, A$120k. The geotechnical investigation found soft ground beneath the area where the compressors and condensers would sit. The error later in this chapter explains how close it came to being missed.
• Fire services upgrade, A$190k. The four new rooms enlarge the building and add large areas of insulated panel. The fire engineering basis, agreed with the building surveyor, set out what fire services the enlarged building would need. Found during delivery, after the building permit application had been prepared without it, the upgrade would have cost about A$260k and 5 weeks.
• Dock levellers and seals, A$30k. The concept showed the new rooms with doors but no dock equipment. Design development, with the packhouse manager’s input on how bins and pallets would move, added levellers and seals at no separate investigation cost.

Figure 3Each item's cost, found early and found in delivery

Electrical supply, early380 A$k

Electrical supply, in delivery520 A$k14 weeks late

Refrigeration plant, early270 A$k

Refrigeration plant, in delivery390 A$k3 weeks late

Ground improvement, early120 A$k

Ground improvement, in delivery210 A$k4 weeks late

Fire services, early190 A$k

Fire services, in delivery260 A$k5 weeks late

Dock equipment, early30 A$k

Dock equipment, in delivery45 A$k

Which investigations come first

The investigations were not all started at once, and their order was not arbitrary. Northbank ranked them by two things: how long the answer would take to act on, and whether one answer changed another question.

The electrical capacity study came first, in the first weeks of FEL 2, because its consequence ran on an external timetable. If the supply was inadequate, the distributor’s 38 weeks of process had to start as early as possible, and no amount of effort by the project team could shorten it. The refrigeration options came next, because the size of the plant set much of the electrical load: a study of the supply that assumed the wrong plant would have answered the wrong question. The ground investigation and the fire engineering basis waited for FEL 3, because both needed a layout. Bores drilled before the layout was settled might have missed the loaded areas altogether, and a fire engineer cannot set a basis for a building whose size is not yet known.

The general rule is to investigate first whatever has the longest external lead time and whatever other answers depend on. Investigations that need a design come once the design exists, and no earlier.

Judging an investigation before it is done

The table above is hindsight: it shows what the investigations found. The decision to commission each one had to be made before its result was known. The test is simple. An investigation is worth its cost if the chance that it finds a problem, multiplied by what finding it early saves, exceeds what it costs.

Turned around, each investigation has a break-even probability: its cost divided by the premium it avoids. For the electrical study, A$9k against a premium of A$140k gives about 6.4%. For the ground investigation, A$14k against A$90k gives about 15.6%; for the fire engineering basis, A$11k against A$70k, about 15.7%. If the team thought there was even a one-in-six chance of a problem with the ground, the investigation paid on cost alone, before counting the weeks of delay, which on Northbank’s calendar are often the larger part. The value of information: when is it worth finding out more? develops this reasoning in general, and Finding problems while they are cheap: the cost of change by stage in Prototypes and Simulation by the Numbers applies the same arithmetic to design problems in product development, where the cost of a fix also rises steeply with the stage at which it is found.

The front end also missed something. Element B7 in Chapter 3 was left at level 3: nobody checked whether the council would require stormwater detention for the new roof and hardstand. It did, and Chapter 10 records its cost of A$64k, met from contingency. A front end lowers the amount of hidden scope; it does not remove it, which is why contingency survives approval.

Figure 4Interfaces to investigate before approval

• Do: Electricity supplycapacity study; distributor's application, offer and lead time
• Do: Groundinvestigation under every loaded area, located on the current layout
• Do: Planning and building approvalspermit path, lead time, conditions likely to be imposed
• Do: Fire safetyfire engineering basis agreed with the building surveyor
• Do: Existing plant and servicescondition, remaining life and capacity to share
• Do: Food safety and hygienesurfaces, drainage, pest exclusion, cleaning access
• Ask: Stormwatercouncil's detention and discharge requirements for new roof and hardstand

The error: a borehole plan that missed the plant room pad

This is an illustration. The geotechnical brief, drafted early in FEL 3, asked for four boreholes under the footprint of the new cold rooms. It was written from the Gate 2 layout, on which the rooms were the obvious new structure. The plant room, with its concrete pad carrying compressors, condensers and receivers, the heaviest concentrated loads on the site, sat beside the rooms on a separate pad and had no borehole at all.

The gap was found in week 14, at a design review, when the project engineer overlaid the borehole locations on the latest layout drawing. Every bore sat under the rooms; the plant room pad sat on untested ground. The drilling rig had not yet been mobilised, so two extra bores under the pad were added to the same visit, for A$3.5k, taking the geotechnical investigation from A$10.5k to the A$14k in the FEL 3 budget.

The extra bores found soft ground under the pad. The geotechnical engineer recommended ground improvement, which was designed and priced at A$120k before Gate 3. Had the bores not been added, the soft ground would have been found when the pad was excavated: about A$210k and 4 weeks, with the plant room on the critical path to commissioning. A$3.5k of drilling avoided a premium of A$90k, about 26 times its cost.

The correction was a rule for every future investigation brief: before it is issued, the investigation locations are drawn on the current layout, and the brief names every loaded area, every connection point and every existing structure it must cover. A brief written from memory of the layout covers what the writer was thinking about, which is seldom the whole site.

Exercise: the break-even probability of a capacity study

Before FEL 2, the team did not know whether the existing supply was adequate. The capacity study cost A$9k. If it found a problem, finding it early would avoid a late premium of A$140k, ignoring the 14 weeks of delay. What probability of a problem makes the study worth its cost? How does counting the delay change the answer?

Worked answer. The study breaks even when the probability of a problem times A$140k equals A$9k: 9 ÷ 140 = 6.4%. Any belief above about one chance in sixteen justified it on cost alone. Counting the delay makes the case much stronger. Time-related delivery costs of A$14k a week for 14 weeks add A$196k, raising the avoided premium to A$336k and lowering the break-even probability to 9 ÷ 336 = 2.7%. And 14 weeks is longer than either path’s float, so a late discovery would also put the harvest at risk, a cost Chapter 8 shows can exceed A$600k on its own. With the delay counted, the study pays at a much lower probability, which is the usual result for investigations at interfaces with long external lead times.