Integrated project delivery and alliancing

Every method in the previous lessons draws contract lines that put the owner, the designer and the builder on different sides. Each party protects its own interests, and when something goes wrong each can gain by showing that someone else was at fault. Integrated project delivery and alliancing try to remove that structure. The key parties sign one agreement, agree a target cost together, share the profit or loss against that target, waive most claims against each other, and make decisions jointly. The idea is simple: if everyone wins or loses together, everyone has a reason to solve problems rather than assign blame.

Integrated project delivery, usually shortened to IPD, developed in North America in the 2000s, building on lean construction ideas and on the experience of a few California health systems that used it for hospitals. Alliancing developed earlier, in the North Sea oil industry and then in Australian public infrastructure in the 1990s and 2000s, and in the United Kingdom. The two are close relatives. CON-260 lesson 11 discussed collaborative contracts from the point of view of digital information; this lesson looks at how they are structured and how money moves.

The principles

Not every project labeled IPD has all of these. Practitioners distinguish pure IPD, with a multi-party agreement, shared risk and reward and liability waivers, from "IPD-ish" or "IPD-lite" projects that use collaborative tools under conventional contracts such as CMAR. Both can work, but only the multi-party agreement changes the incentives fundamentally.

The multi-party agreement and its governance

In IPD the owner, the lead designer and the builder, and often several key trade partners and consultants, sign the same agreement. A senior management team of executives from each party sets direction and resolves escalated issues. A project management team runs the project day to day, and cluster or work-stream teams, mixing designers, builders and trades, develop and deliver each part of the work. Decisions are made by consensus, with the owner usually having the final say if the team cannot agree, and with disputes escalating through defined levels before any external process.

In the United States, the main standard forms are the AIA's C191-2009 (a single multi-party agreement) and C195-2008 (a single-purpose entity owned by the parties), ConsensusDocs 300 (a tri-party collaborative agreement), and the Integrated Form of Agreement, or IFOA, developed for Sutter Health in California and widely adapted by health systems and their lawyers.

Validation and the target cost

An IPD project normally begins with a validation phase, sometimes called a conceptualization phase. The team tests whether the owner's program, budget and schedule are achievable, develops an early design and a target cost, and produces a validation report. The owner then decides whether to proceed. If the target cost is not achievable within the owner's budget, the owner can change the program, increase the budget or stop. This step matters: a target that is unrealistic from the start drives behavior toward the very blame and cost-cutting the method is meant to avoid.

The target cost, often called the expected maximum price or estimated maximum price in American IPD contracts, is the team's agreed estimate of the cost to deliver the project. It is set aggressively but realistically, often below the owner's budget, using target value design: designing to a cost rather than costing a design. The team breaks the target into clusters, each with its own target, and tracks the forecast cost of each against its target throughout design, adjusting the design when a cluster runs over.

How the parties are paid

IPD compensation has three layers for each non-owner party.

If the actual cost finishes below the target, the savings are shared between the owner and the profit pool, so the non-owner parties earn more than their planned profit. If the actual cost exceeds the target, the overrun first consumes the profit pool. Once the pool is gone, most IPD agreements have the owner pay remaining costs, so the non-owner parties lose their profit but not their own money. That limit is what makes the liability waivers acceptable to designers and builders: they put profit at risk, not their companies.

# Profit pool under an IPD agreement for a $150 million target cost (illustrative)
target = 150_000_000
pool = 9_000_000          # planned profit of designer, builder and trade partners at risk
gain_share = 0.50         # share of savings to the profit pool
pain_share = 1.00         # overrun absorbed by the pool until exhausted
def outcome(actual):
    if actual <= target:
        profit = pool + gain_share * (target - actual)
    else:
        profit = max(0, pool - pain_share * (actual - target))
    owner_pays = actual + profit
    return profit, owner_pays
for actual in (138e6, 145e6, 150e6, 154e6, 159e6, 165e6):
    p, o = outcome(actual)
    print(f"actual cost {actual / 1e6:6.1f}M: profit pool paid {p / 1e6:5.2f}M, owner pays {o / 1e6:6.2f}M")

actual cost  138.0M: profit pool paid 15.00M, owner pays 153.00M
actual cost  145.0M: profit pool paid 11.50M, owner pays 156.50M
actual cost  150.0M: profit pool paid  9.00M, owner pays 159.00M
actual cost  154.0M: profit pool paid  5.00M, owner pays 159.00M
actual cost  159.0M: profit pool paid  0.00M, owner pays 159.00M
actual cost  165.0M: profit pool paid  0.00M, owner pays 165.00M

At target, the owner pays the cost plus the planned profit of $9 million. At $138 million, the parties earn $15 million and the owner still pays $6 million less than at target. At $159 million, the pool is exhausted and the parties earn no profit; above that, the owner bears every additional dollar. Real agreements vary: some share overruns with the owner from the first dollar, some cap the gain share, and some include an owner contribution to the pool if goals are met.

Alliancing

An alliance is a similar arrangement used mainly for infrastructure. The owner and one or more non-owner participants, such as a contractor and a designer, form an alliance under a project alliance agreement. Decisions are made unanimously by an alliance leadership team, the parties commit to "no blame, no dispute" (they agree not to sue each other except for deliberate default), and commercial terms combine reimbursed costs, a fee for corporate overhead and profit, and a pain/gain share against a target outturn cost. Alliances were used extensively in Australia for roads, rail and water from the late 1990s, guided by national alliance contracting guidelines, and in the United Kingdom, where the NEC4 suite added an alliance contract in 2018.

Alliancing is less common in United States public infrastructure than in Australia or Canada, because public procurement laws often require price competition. Collaborative models in US transportation more often use CM/GC or progressive design-build. Some health systems, universities and private owners use IPD agreements that function much like alliances.

Does it work?

Case studies of IPD and alliance projects report strong results on cost, schedule, quality and relationships, and surveys of participants are generally positive. But most evidence comes from projects whose owners chose the method enthusiastically and staffed it well, so it is hard to separate the method from the people. The same studies identify recurring conditions for success: an owner willing to lead and make decisions quickly, partners chosen for capability and culture, a realistic target from validation, co-location, real training in lean and collaborative behaviors, and executives who resist the habit of turning problems into claims. Where those conditions are missing, IPD projects can stall in consensus, overrun the target and leave parties disappointed that the profit pool vanished.